diff options
Diffstat (limited to '')
| -rw-r--r-- | src/Makefile.am | 19 | ||||
| -rw-r--r-- | src/Makefile.in | 340 | ||||
| -rw-r--r-- | src/buffer.c | 308 | ||||
| -rw-r--r-- | src/buffer.h | 62 | ||||
| -rw-r--r-- | src/config.h | 101 | ||||
| -rw-r--r-- | src/conns.c | 187 | ||||
| -rw-r--r-- | src/conns.h | 59 | ||||
| -rw-r--r-- | src/defines.h.in | 89 | ||||
| -rw-r--r-- | src/dnscache.c | 198 | ||||
| -rw-r--r-- | src/dnscache.h | 38 | ||||
| -rw-r--r-- | src/filter.c | 150 | ||||
| -rw-r--r-- | src/filter.h | 25 | ||||
| -rw-r--r-- | src/gnuregex.c | 5880 | ||||
| -rw-r--r-- | src/gnuregex.h | 542 | ||||
| -rw-r--r-- | src/log.c | 90 | ||||
| -rw-r--r-- | src/log.h | 24 | ||||
| -rw-r--r-- | src/regexp.h | 24 | ||||
| -rw-r--r-- | src/reqs.c | 832 | ||||
| -rw-r--r-- | src/reqs.h | 24 | ||||
| -rw-r--r-- | src/sock.c | 362 | ||||
| -rw-r--r-- | src/sock.h | 40 | ||||
| -rw-r--r-- | src/stamp-h.in | 1 | ||||
| -rw-r--r-- | src/tinyproxy.c | 478 | ||||
| -rw-r--r-- | src/tinyproxy.h | 87 | ||||
| -rw-r--r-- | src/uri.c | 124 | ||||
| -rw-r--r-- | src/uri.h | 32 | ||||
| -rw-r--r-- | src/utils.c | 264 | ||||
| -rw-r--r-- | src/utils.h | 38 |
28 files changed, 10418 insertions, 0 deletions
diff --git a/src/Makefile.am b/src/Makefile.am new file mode 100644 index 0000000..63ffb6b --- /dev/null +++ b/src/Makefile.am @@ -0,0 +1,19 @@ +bin_PROGRAMS = tinyproxy +tinyproxy_SOURCES = config.h \ + buffer.c buffer.h \ + conns.c conns.h \ + dnscache.c dnscache.h \ + log.c log.h \ + reqs.c reqs.h \ + sock.c sock.h \ + tinyproxy.c tinyproxy.h \ + uri.c uri.h \ + utils.c utils.h \ + regexp.h + +tinyproxy_LDADD = @LIBOBJS@ + +EXTRA_DIST = gnuregex.c gnuregex.h \ + filter.c filter.h + +LDADD = @LIBOBJS diff --git a/src/Makefile.in b/src/Makefile.in new file mode 100644 index 0000000..0bc6f5c --- /dev/null +++ b/src/Makefile.in @@ -0,0 +1,340 @@ +# Makefile.in generated automatically by automake 1.4 from Makefile.am + +# Copyright (C) 1994, 1995-8, 1999 Free Software Foundation, Inc. +# This Makefile.in is free software; the Free Software Foundation +# gives unlimited permission to copy and/or distribute it, +# with or without modifications, as long as this notice is preserved. + +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY, to the extent permitted by law; without +# even the implied warranty of MERCHANTABILITY or FITNESS FOR A +# PARTICULAR PURPOSE. + + +SHELL = @SHELL@ + +srcdir = @srcdir@ +top_srcdir = @top_srcdir@ +VPATH = @srcdir@ +prefix = @prefix@ +exec_prefix = @exec_prefix@ + +bindir = @bindir@ +sbindir = @sbindir@ +libexecdir = @libexecdir@ +datadir = @datadir@ +sysconfdir = @sysconfdir@ +sharedstatedir = @sharedstatedir@ +localstatedir = @localstatedir@ +libdir = @libdir@ +infodir = @infodir@ +mandir = @mandir@ +includedir = @includedir@ +oldincludedir = /usr/include + +DESTDIR = + +pkgdatadir = $(datadir)/@PACKAGE@ +pkglibdir = $(libdir)/@PACKAGE@ +pkgincludedir = $(includedir)/@PACKAGE@ + +top_builddir = .. + +ACLOCAL = @ACLOCAL@ +AUTOCONF = @AUTOCONF@ +AUTOMAKE = @AUTOMAKE@ +AUTOHEADER = @AUTOHEADER@ + +INSTALL = @INSTALL@ +INSTALL_PROGRAM = @INSTALL_PROGRAM@ $(AM_INSTALL_PROGRAM_FLAGS) +INSTALL_DATA = @INSTALL_DATA@ +INSTALL_SCRIPT = @INSTALL_SCRIPT@ +transform = @program_transform_name@ + +NORMAL_INSTALL = : +PRE_INSTALL = : +POST_INSTALL = : +NORMAL_UNINSTALL = : +PRE_UNINSTALL = : +POST_UNINSTALL = : +CC = @CC@ +CFLAGS = @CFLAGS@ -I. -DHAVE_CONFIG_H +LIBOBJS = @LIBOBJS@ +LIBS = @LIBS@ +MAKEINFO = @MAKEINFO@ +PACKAGE = @PACKAGE@ +VERSION = @VERSION@ + +bin_PROGRAMS = tinyproxy +tinyproxy_SOURCES = config.h buffer.c buffer.h conns.c conns.h dnscache.c dnscache.h log.c log.h reqs.c reqs.h sock.c sock.h tinyproxy.c tinyproxy.h uri.c uri.h utils.c utils.h regexp.h + + +tinyproxy_LDADD = @LIBOBJS@ + +EXTRA_DIST = gnuregex.c gnuregex.h filter.c filter.h + + +LDADD = @LIBOBJS +mkinstalldirs = $(SHELL) $(top_srcdir)/mkinstalldirs +CONFIG_HEADER = defines.h +CONFIG_CLEAN_FILES = +PROGRAMS = $(bin_PROGRAMS) + + +DEFS = @DEFS@ -I. -I$(srcdir) -I. +CPPFLAGS = @CPPFLAGS@ +LDFLAGS = @LDFLAGS@ +tinyproxy_OBJECTS = buffer.o conns.o dnscache.o log.o reqs.o sock.o \ +tinyproxy.o uri.o utils.o +tinyproxy_DEPENDENCIES = @LIBOBJS@ +tinyproxy_LDFLAGS = +COMPILE = $(CC) $(DEFS) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS) +CCLD = $(CC) +LINK = $(CCLD) $(AM_CFLAGS) $(CFLAGS) $(LDFLAGS) -o $@ +DIST_COMMON = ./stamp-h.in Makefile.am Makefile.in defines.h.in \ +filter.c gnuregex.c + + +DISTFILES = $(DIST_COMMON) $(SOURCES) $(HEADERS) $(TEXINFOS) $(EXTRA_DIST) + +TAR = tar +GZIP_ENV = --best +DEP_FILES = .deps/buffer.P .deps/conns.P .deps/dnscache.P \ +.deps/filter.P .deps/gnuregex.P .deps/log.P .deps/reqs.P .deps/sock.P \ +.deps/tinyproxy.P .deps/uri.P .deps/utils.P +SOURCES = $(tinyproxy_SOURCES) +OBJECTS = $(tinyproxy_OBJECTS) + +all: all-redirect +.SUFFIXES: +.SUFFIXES: .S .c .o .s +$(srcdir)/Makefile.in: Makefile.am $(top_srcdir)/configure.in $(ACLOCAL_M4) + cd $(top_srcdir) && $(AUTOMAKE) --gnu src/Makefile + +Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status $(BUILT_SOURCES) + cd $(top_builddir) \ + && CONFIG_FILES=$(subdir)/$@ CONFIG_HEADERS= $(SHELL) ./config.status + + +defines.h: stamp-h + @if test ! -f $@; then \ + rm -f stamp-h; \ + $(MAKE) stamp-h; \ + else :; fi +stamp-h: $(srcdir)/defines.h.in $(top_builddir)/config.status + cd $(top_builddir) \ + && CONFIG_FILES= CONFIG_HEADERS=src/defines.h \ + $(SHELL) ./config.status + @echo timestamp > stamp-h 2> /dev/null +$(srcdir)/defines.h.in: $(srcdir)/stamp-h.in + @if test ! -f $@; then \ + rm -f $(srcdir)/stamp-h.in; \ + $(MAKE) $(srcdir)/stamp-h.in; \ + else :; fi +$(srcdir)/stamp-h.in: $(top_srcdir)/configure.in $(ACLOCAL_M4) + cd $(top_srcdir) && $(AUTOHEADER) + @echo timestamp > $(srcdir)/stamp-h.in 2> /dev/null + +mostlyclean-hdr: + +clean-hdr: + +distclean-hdr: + -rm -f defines.h + +maintainer-clean-hdr: + +mostlyclean-binPROGRAMS: + +clean-binPROGRAMS: + -test -z "$(bin_PROGRAMS)" || rm -f $(bin_PROGRAMS) + +distclean-binPROGRAMS: + +maintainer-clean-binPROGRAMS: + +install-binPROGRAMS: $(bin_PROGRAMS) + @$(NORMAL_INSTALL) + $(mkinstalldirs) $(DESTDIR)$(bindir) + @list='$(bin_PROGRAMS)'; for p in $$list; do \ + if test -f $$p; then \ + echo " $(INSTALL_PROGRAM) $$p $(DESTDIR)$(bindir)/`echo $$p|sed 's/$(EXEEXT)$$//'|sed '$(transform)'|sed 's/$$/$(EXEEXT)/'`"; \ + $(INSTALL_PROGRAM) $$p $(DESTDIR)$(bindir)/`echo $$p|sed 's/$(EXEEXT)$$//'|sed '$(transform)'|sed 's/$$/$(EXEEXT)/'`; \ + else :; fi; \ + done + +uninstall-binPROGRAMS: + @$(NORMAL_UNINSTALL) + list='$(bin_PROGRAMS)'; for p in $$list; do \ + rm -f $(DESTDIR)$(bindir)/`echo $$p|sed 's/$(EXEEXT)$$//'|sed '$(transform)'|sed 's/$$/$(EXEEXT)/'`; \ + done + +.s.o: + $(COMPILE) -c $< + +.S.o: + $(COMPILE) -c $< + +mostlyclean-compile: + -rm -f *.o core *.core + +clean-compile: + +distclean-compile: + -rm -f *.tab.c + +maintainer-clean-compile: + +tinyproxy: $(tinyproxy_OBJECTS) $(tinyproxy_DEPENDENCIES) + @rm -f tinyproxy + $(LINK) $(tinyproxy_LDFLAGS) $(tinyproxy_OBJECTS) $(tinyproxy_LDADD) $(LIBS) + +tags: TAGS + +ID: $(HEADERS) $(SOURCES) $(LISP) + list='$(SOURCES) $(HEADERS)'; \ + unique=`for i in $$list; do echo $$i; done | \ + awk ' { files[$$0] = 1; } \ + END { for (i in files) print i; }'`; \ + here=`pwd` && cd $(srcdir) \ + && mkid -f$$here/ID $$unique $(LISP) + +TAGS: $(HEADERS) $(SOURCES) defines.h.in $(TAGS_DEPENDENCIES) $(LISP) + tags=; \ + here=`pwd`; \ + list='$(SOURCES) $(HEADERS)'; \ + unique=`for i in $$list; do echo $$i; done | \ + awk ' { files[$$0] = 1; } \ + END { for (i in files) print i; }'`; \ + test -z "$(ETAGS_ARGS)defines.h.in$$unique$(LISP)$$tags" \ + || (cd $(srcdir) && etags $(ETAGS_ARGS) $$tags defines.h.in $$unique $(LISP) -o $$here/TAGS) + +mostlyclean-tags: + +clean-tags: + +distclean-tags: + -rm -f TAGS ID + +maintainer-clean-tags: + +distdir = $(top_builddir)/$(PACKAGE)-$(VERSION)/$(subdir) + +subdir = src + +distdir: $(DISTFILES) + here=`cd $(top_builddir) && pwd`; \ + top_distdir=`cd $(top_distdir) && pwd`; \ + distdir=`cd $(distdir) && pwd`; \ + cd $(top_srcdir) \ + && $(AUTOMAKE) --include-deps --build-dir=$$here --srcdir-name=$(top_srcdir) --output-dir=$$top_distdir --gnu doc/Makefile + @for file in $(DISTFILES); do \ + d=$(srcdir); \ + if test -d $$d/$$file; then \ + cp -pr $$/$$file $(distdir)/$$file; \ + else \ + test -f $(distdir)/$$file \ + || ln $$d/$$file $(distdir)/$$file 2> /dev/null \ + || cp -p $$d/$$file $(distdir)/$$file || :; \ + fi; \ + done +buffer.o: buffer.c defines.h utils.h conns.h tinyproxy.h config.h log.h \ + buffer.h +conns.o: conns.c defines.h config.h log.h utils.h conns.h tinyproxy.h \ + buffer.h dnscache.h +dnscache.o: dnscache.c defines.h utils.h conns.h tinyproxy.h config.h \ + dnscache.h +log.o: log.c defines.h tinyproxy.h config.h log.h +reqs.o: reqs.c defines.h config.h tinyproxy.h sock.h buffer.h utils.h \ + conns.h log.h reqs.h filter.h uri.h regexp.h +sock.o: sock.c defines.h tinyproxy.h config.h sock.h buffer.h log.h \ + utils.h conns.h dnscache.h +tinyproxy.o: tinyproxy.c defines.h config.h tinyproxy.h utils.h conns.h \ + log.h sock.h buffer.h reqs.h filter.h +uri.o: uri.c defines.h uri.h utils.h conns.h tinyproxy.h config.h log.h \ + regexp.h +utils.o: utils.c defines.h config.h tinyproxy.h utils.h conns.h log.h \ + buffer.h + +info-am: +info: info-am +dvi-am: +dvi: dvi-am +check-am: all-am +check: check-am +installcheck-am: +installcheck: installcheck-am +all-recursive-am: defines.h + $(MAKE) $(AM_MAKEFLAGS) all-recursive + +install-exec-am: install-binPROGRAMS +install-exec: install-exec-am + +install-data-am: +install-data: install-data-am + +install-am: all-am + @$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am +install: install-am +uninstall-am: uninstall-binPROGRAMS +uninstall: uninstall-am +all-am: Makefile $(PROGRAMS) defines.h +all-redirect: all-am +install-strip: + $(MAKE) $(AM_MAKEFLAGS) AM_INSTALL_PROGRAM_FLAGS=-s install +installdirs: + $(mkinstalldirs) $(DESTDIR)$(bindir) + + +mostlyclean-generic: + +clean-generic: + +distclean-generic: + -rm -f Makefile $(CONFIG_CLEAN_FILES) + -rm -f config.cache config.log stamp-h stamp-h[0-9]* + +maintainer-clean-generic: +mostlyclean-am: mostlyclean-hdr mostlyclean-binPROGRAMS \ + mostlyclean-compile mostlyclean-tags mostlyclean-depend \ + mostlyclean-generic + +mostlyclean: mostlyclean-am + +clean-am: clean-hdr clean-binPROGRAMS clean-compile clean-tags \ + clean-depend clean-generic mostlyclean-am + +clean: clean-am + +distclean-am: distclean-hdr distclean-binPROGRAMS distclean-compile \ + distclean-tags distclean-depend distclean-generic \ + clean-am + +distclean: distclean-am + +maintainer-clean-am: maintainer-clean-hdr maintainer-clean-binPROGRAMS \ + maintainer-clean-compile maintainer-clean-tags \ + maintainer-clean-depend maintainer-clean-generic \ + distclean-am + @echo "This command is intended for maintainers to use;" + @echo "it deletes files that may require special tools to rebuild." + +maintainer-clean: maintainer-clean-am + +.PHONY: mostlyclean-hdr distclean-hdr clean-hdr maintainer-clean-hdr \ +mostlyclean-binPROGRAMS distclean-binPROGRAMS clean-binPROGRAMS \ +maintainer-clean-binPROGRAMS uninstall-binPROGRAMS install-binPROGRAMS \ +mostlyclean-compile distclean-compile clean-compile \ +maintainer-clean-compile tags mostlyclean-tags distclean-tags \ +clean-tags maintainer-clean-tags distdir mostlyclean-depend \ +distclean-depend clean-depend maintainer-clean-depend info-am info \ +dvi-am dvi check check-am installcheck-am installcheck all-recursive-am \ +install-exec-am install-exec install-data-am install-data install-am \ +install uninstall-am uninstall all-redirect all-am all installdirs \ +mostlyclean-generic distclean-generic clean-generic \ +maintainer-clean-generic clean mostlyclean distclean maintainer-clean + + +# Tell versions [3.59,3.63) of GNU make to not export all variables. +# Otherwise a system limit (for SysV at least) may be exceeded. +.NOEXPORT: diff --git a/src/buffer.c b/src/buffer.c new file mode 100644 index 0000000..8e76e26 --- /dev/null +++ b/src/buffer.c @@ -0,0 +1,308 @@ +/* $Id: buffer.c,v 1.1.1.1 2000-02-16 17:32:22 sdyoung Exp $ + * + * The buffer used in each connection is a linked list of lines. As the lines + * are read in and written out the buffer expands and contracts. Basically, + * by using this method we can increase the buffer size dynamicly. However, + * we have a hard limit of 64 KB for the size of the buffer. The buffer can be + * thought of as a queue were we act on both the head and tail. The various + * functions act on each end (the names are taken from what Perl uses to act on + * the ends of an array. :) + * + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <stdlib.h> +#include <errno.h> +#include <string.h> +#include <sys/types.h> +#include <sys/socket.h> +#include <assert.h> + +#include "utils.h" +#include "log.h" +#include "tinyproxy.h" +#include "buffer.h" + +/* + * Take a string of data and a length and make a new line which can be added + * to the buffer + */ +static struct bufline_s *makenewline(unsigned char *data, unsigned int length) +{ + struct bufline_s *newline; + + assert(data); + assert(length > 0); + + if (!(newline = xmalloc(sizeof(struct bufline_s)))) + return NULL; + + newline->string = data; + newline->next = NULL; + newline->length = length; + newline->pos = 0; + + return newline; +} + +/* + * Create a new buffer + */ +struct buffer_s *new_buffer(void) +{ + struct buffer_s *buffptr; + + if (!(buffptr = xmalloc(sizeof(struct buffer_s)))) + return NULL; + + buffptr->head = buffptr->tail = NULL; + buffptr->size = 0; + + return buffptr; +} + +/* + * Delete all the lines in the buffer and the buffer itself + */ +void delete_buffer(struct buffer_s *buffptr) +{ + struct bufline_s *next; + + assert(buffptr); + + while (buffer_head(buffptr)) { + next = buffer_head(buffptr)->next; + free_line(buffer_head(buffptr)); + buffer_head(buffptr) = next; + } + buffer_head(buffptr) = NULL; + buffer_tail(buffptr) = NULL; + + safefree(buffptr); +} + +/* + * Free the allocated buffer line + */ +void free_line(struct bufline_s *line) +{ + if (!line) + return; + + if (line->string) { + safefree(line->string); + } + + safefree(line); +} + +/* + * Push a new line on to the end of the buffer + */ +struct bufline_s *push_buffer(struct buffer_s *buffptr, unsigned char *data, + unsigned int length) +{ + struct bufline_s *newline; + + assert(buffptr); + assert(data); + assert(length > 0); + + if (!(newline = makenewline(data, length))) + return NULL; + + if (!buffer_head(buffptr) && !buffer_tail(buffptr)) + buffer_head(buffptr) = buffer_tail(buffptr) = newline; + else + buffer_tail(buffptr) = (buffer_tail(buffptr)->next = newline); + + buffptr->size += length; + + return newline; +} + +/* + * Pop a buffer line off the end of the buffer + */ +struct bufline_s *pop_buffer(struct buffer_s *buffptr) +{ + struct bufline_s *line, *newend; + + assert(buffptr); + + if (buffer_head(buffptr) == buffer_tail(buffptr)) { + line = buffer_head(buffptr); + buffer_head(buffptr) = buffer_tail(buffptr) = NULL; + buffptr->size = 0; + return line; + } + + line = buffer_tail(buffptr); + newend = buffer_head(buffptr); + + while (newend->next != line && newend->next) + newend = newend->next; + + buffer_tail(buffptr) = newend; + buffptr->size -= line->length; + + return line; +} + +/* + * Unshift a buffer line from the top of the buffer (meaning add a new line + * to the top of the buffer) + */ +struct bufline_s *unshift_buffer(struct buffer_s *buffptr, unsigned char *data, + unsigned int length) +{ + struct bufline_s *newline; + + assert(buffptr); + assert(data); + assert(length > 0); + + if (!(newline = makenewline(data, length))) + return NULL; + + if (!buffer_head(buffptr) && buffer_tail(buffptr)) { + buffer_head(buffptr) = buffer_tail(buffptr) = newline; + } else { + newline->next = buffer_head(buffptr); + buffer_head(buffptr) = newline; + if (!buffer_tail(buffptr)) + buffer_tail(buffptr) = newline; + } + + buffptr->size += length; + + return newline; +} + +/* + * Shift a line off the top of the buffer (remove the line from the top of + * the buffer) + */ +struct bufline_s *shift_buffer(struct buffer_s *buffptr) +{ + struct bufline_s *line; + + assert(buffptr); + + if (!buffer_head(buffptr) && !buffer_tail(buffptr)) { + line = buffer_head(buffptr); + buffer_head(buffptr) = buffer_tail(buffptr) = NULL; + buffptr->size = 0; + return line; + } + + line = buffer_head(buffptr); + buffer_head(buffptr) = line->next; + + if (!buffer_head(buffptr)) + buffer_tail(buffptr) = NULL; + + buffptr->size -= line->length; + + return line; +} + +/* + * Reads the bytes from the socket, and adds them to the buffer. + * Takes a connection and returns the number of bytes read. + */ +int readbuff(int fd, struct buffer_s *buffptr) +{ + int bytesin; + unsigned char inbuf[BUFFER]; + unsigned char *buffer; + + assert(fd >= 0); + assert(buffptr); + + bytesin = recv(fd, inbuf, BUFFER, 0); + + if (bytesin > 0) { + if (!(buffer = xmalloc(bytesin))) + return 0; + + memcpy(buffer, inbuf, bytesin); + push_buffer(buffptr, buffer, bytesin); + return bytesin; + } else if (bytesin == 0) { + /* connection was closed by client */ + return -1; + } else { + switch (errno) { +#ifdef EWOULDBLOCK + case EWOULDBLOCK: +#else +# ifdef EAGAIN + case EAGAIN: +# endif +#endif + case EINTR: + return 0; + case ECONNRESET: + return -1; + default: + log("ERROR readbuff: recv (%s)", strerror(errno)); + return -1; + } + } +} + +/* + * Write the bytes in the buffer to the socket. + * Takes a connection and returns the number of bytes written. + */ +int writebuff(int fd, struct buffer_s *buffptr) +{ + int bytessent; + struct bufline_s *line = buffer_head(buffptr); + + assert(fd >= 0); + assert(buffptr); + + bytessent = send(fd, line->string + line->pos, + (size_t) (line->length - line->pos), 0); + + if (bytessent >= 0) { + /* bytes sent, adjust buffer */ + line->pos += bytessent; + if (line->pos == line->length) + free_line(shift_buffer(buffptr)); + return bytessent; + } else { + switch (errno) { +#ifdef EWOULDBLOCK + case EWOULDBLOCK: +#else +# ifdef EAGAIN + case EAGAIN: +# endif +#endif + case ENOBUFS: + case EINTR: + case ENOMEM: + return 0; + default: + log("ERROR writebuff: send (%s)", strerror(errno)); + return -1; + } + } +} diff --git a/src/buffer.h b/src/buffer.h new file mode 100644 index 0000000..a1ee86c --- /dev/null +++ b/src/buffer.h @@ -0,0 +1,62 @@ +/* $Id: buffer.h,v 1.1.1.1 2000-02-16 17:32:22 sdyoung Exp $ + * + * See 'buffer.c' for a detailed description. + * + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef __BUFFER_H_ +#define __BUFFER_H_ 1 + +#define MAXBUFFSIZE 20480 + +struct bufline_s { + unsigned char *string; /* the actual string of data */ + struct bufline_s *next; /* pointer to next in linked list */ + unsigned int length; /* length of the string of data */ + unsigned int pos; /* start sending from this offset */ +}; + +struct buffer_s { + struct bufline_s *head; /* top of the buffer */ + struct bufline_s *tail; /* bottom of the buffer */ + unsigned long int size; /* total size of the buffer */ +}; + +#define buffer_head(x) (((struct buffer_s *)(x))->head) +#define buffer_tail(x) (((struct buffer_s *)(x))->tail) +#define buffer_size(x) (((struct buffer_s *)(x))->size) + +/* Create and delete buffers */ +extern struct buffer_s *new_buffer(void); +extern void delete_buffer(struct buffer_s *buffptr); + +/* Operate on the end of the list */ +extern struct bufline_s *push_buffer(struct buffer_s *buffptr, + unsigned char *data, unsigned int length); +extern struct bufline_s *pop_buffer(struct buffer_s *buffptr); + +/* Operate on the head of the list */ +extern struct bufline_s *unshift_buffer(struct buffer_s *buffptr, + unsigned char *data, + unsigned int length); +extern struct bufline_s *shift_buffer(struct buffer_s *buffptr); + +/* Free a buffer line */ +extern void free_line(struct bufline_s *line); + +/* Read/Write the buffer to a socket */ +extern int readbuff(int fd, struct buffer_s *buffptr); +extern int writebuff(int fd, struct buffer_s *buffptr); + +#endif /* __BUFFER_H_ */ diff --git a/src/config.h b/src/config.h new file mode 100644 index 0000000..f5f7dda --- /dev/null +++ b/src/config.h @@ -0,0 +1,101 @@ +/* $Id: config.h,v 1.1.1.1 2000-02-16 17:32:22 sdyoung Exp $ + * + * Contains all the tune-able variables which are used by tinyproxy. + * Modifications made to these variables WILL change the default behaviour + * of tinyproxy. Please read the comments to better understand what each + * variable does. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef _CONFIG_H_ +#define _CONFIG_H_ 1 + +/* Change these if you want */ + +/* Default log file */ +#ifndef DEFAULT_LOG +#define DEFAULT_LOG "/usr/local/var/log/tinyproxy.log" +#endif + +/* Default port tinyproxy listens on */ +#ifndef DEFAULT_PORT +#define DEFAULT_PORT 8080 +#endif + +/* Default user to change to after */ +#ifndef DEFAULT_USER +#define DEFAULT_USER "nobody" +#endif + +/* + * Define this if you want to have all requests sent to a particular machine. + * This is useful if you have another proxy further upstream which you must + * send your HTTP requests through. + */ +#undef UPSTREAM + +#ifdef UPSTREAM +#define UPSTREAM_PROXY_NAME "name.of.upstream.machine" +#define UPSTREAM_PROXY_PORT 80 +#endif + +/* + * Define this if you want tinyproxy to use /proc/loadavg to determine + * system load (Linux only, I think) + */ +#define USE_PROC + +#ifndef USE_PROC +/* + * Path to uptime to determin system load. This path doesn't have to be + * valid if DEFAULT_CUTOFFLOAD is 0 + */ +#define UPTIME_PATH "/usr/bin/uptime" +#endif /* !USE_PROC */ + +/* + * The default load at which tinyproxy will start refusing connections. + * 0 == disabled by default + */ +#define DEFAULT_CUTOFFLOAD 0 + +/* + * NOTE: for DEFAULT_STATHOST: this controls remote proxy stats display. + * for example, the default DEFAULT_STATHOST of "tinyproxy.stats" will + * mean that when you use the proxy to access http://tinyproxy.stats/", + * you will be shown the proxy stats. Set this to something obscure + * if you don't want random people to be able to see them, or set it to + * "" to disable. In the future, I figure maybe some sort of auth + * might be desirable, but that would involve a major simplicity + * sacrifice. + */ + +/* The "hostname" for getting tinyproxy stats. "" = disabled by default */ +#define DEFAULT_STATHOST "tinyproxy.stats" + +/* + * NOTE: change these if you know what you're doing + */ +#define SOCK_TIMEOUT 5 +/* Recalculate load avery 30 seconds */ +#define LOAD_RECALCTIMER 30 +/* Default HTTP port */ +#define HTTP_PORT 80 +/* Every time conncoll is called, nuke connections idle for > 60 seconds */ +#define STALECONN_TIME (60 * 15) +/* Every 100 times through run the garbage collection */ +#define GARBCOLL_INTERVAL 10 + +#endif diff --git a/src/conns.c b/src/conns.c new file mode 100644 index 0000000..a96067f --- /dev/null +++ b/src/conns.c @@ -0,0 +1,187 @@ +/* $Id: conns.c,v 1.1.1.1 2000-02-16 17:32:22 sdyoung Exp $ + * + * These functions handle the various stages a connection will go through in + * the course of its life in tinyproxy. New connections are initialized and + * added to the linked list of active connections. As these connections are + * completed, they are closed, and the a garbage collection process removes + * them from the list. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <stdio.h> +#include <time.h> +#include <stdlib.h> +#include <unistd.h> +#include <string.h> +#include <sysexits.h> +#include <sys/types.h> +#include <sys/socket.h> +#include <errno.h> +#include <assert.h> + +#include <adns.h> + +extern adns_state adns; + +#include "config.h" +#include "log.h" +#include "utils.h" +#include "conns.h" +#include "buffer.h" +#include "dnscache.h" + +struct conn_s *connections = NULL; + +/* + * Add a new connection to the linked list + */ +struct conn_s *new_conn(int fd) +{ + struct conn_s **rpConnptr = &connections; + struct conn_s *connptr = connections; + struct conn_s *newconn; + + assert(fd >= 0); + + while (connptr) { + rpConnptr = &connptr->next; + connptr = connptr->next; + } + + if (!(newconn = xmalloc(sizeof(struct conn_s)))) { + log("ERROR new_conn: could not allocate memory for conn"); + return NULL; + } + + /* Allocate the new buffer */ + newconn->cbuffer = NULL; + newconn->sbuffer = NULL; + if (!(newconn->cbuffer = new_buffer()) + || !(newconn->sbuffer = new_buffer())) { + log("ERROR new_conn: could not allocate memory for buffer"); + safefree(newconn->cbuffer); + safefree(newconn->sbuffer); + + newconn->next = NULL; + safefree(newconn); + return NULL; + } + + newconn->client_fd = fd; + newconn->server_fd = -1; + newconn->type = NEWCONN; + newconn->inittime = newconn->actiontime = time(NULL); + + newconn->clientheader = newconn->serverheader = FALSE; + newconn->simple_req = FALSE; + + *rpConnptr = newconn; + newconn->next = connptr; + + stats.num_cons++; + + return newconn; +} + +/* + * Delete a connection from the linked list + */ +int del_conn(struct conn_s *delconn) +{ + struct conn_s **rpConnptr = &connections; + struct conn_s *connptr = connections; + + assert(delconn); + + if (delconn->cbuffer) { + delete_buffer(delconn->cbuffer); + delconn->cbuffer = NULL; + } + if (delconn->sbuffer) { + delete_buffer(delconn->sbuffer); + delconn->sbuffer = NULL; + } + + close(delconn->client_fd); + close(delconn->server_fd); + + while (connptr && (connptr != delconn)) { + rpConnptr = &connptr->next; + connptr = connptr->next; + } + + if (connptr == delconn) { + *rpConnptr = delconn->next; + safefree(delconn); + } + + return 0; +} + +/* + * Check for connections that have been idle too long + */ +void conncoll(void) +{ + struct conn_s *connptr = connections; + + while (connptr) { + if ( + (difftime(time(NULL), connptr->actiontime) > + STALECONN_TIME) && connptr->type != FINISHCONN) { + connptr->type = FINISHCONN; + stats.num_idles++; + } + connptr = connptr->next; + } +} + +/* + * Actually remove all entries in the linked list that have been marked for + * deletion. + */ +void garbcoll(void) +{ + struct conn_s *connptr = connections; + struct conn_s *tmp; + + static unsigned int dnscount = 0; + +#ifdef __DEBUG__ + log("Garbage collecting (%lu)", stats.num_garbage); +#endif + + stats.num_garbage++; + + while (connptr) { + tmp = connptr->next; + if (connptr->type == FINISHCONN) { +#ifdef __DEBUG__ + log("Deleting connection: %p", connptr); +#endif + del_conn(connptr); + } + connptr = tmp; + } + + if (dnscount++ > DNS_GARBAGE_COL) { + dnscount = 0; + dnsclean(); + } +} diff --git a/src/conns.h b/src/conns.h new file mode 100644 index 0000000..05cfdb5 --- /dev/null +++ b/src/conns.h @@ -0,0 +1,59 @@ +/* $Id: conns.h,v 1.1.1.1 2000-02-16 17:32:22 sdyoung Exp $ + * + * See 'conns.c' for a detailed description. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef _CONN_H_ +#define _CONN_H_ 1 + +#include "tinyproxy.h" + +#include <adns.h> + +/* Different connection types */ +enum conn_type { + NEWCONN, + WAITCONN, + DNS_WAITCONN, + RELAYCONN, + CLOSINGCONN, + FINISHCONN +}; + +struct conn_s { + struct conn_s *next; + int client_fd, server_fd; + enum conn_type type; + struct buffer_s *cbuffer; + struct buffer_s *sbuffer; + time_t inittime, actiontime; + flag clientheader, serverheader; + flag simple_req; + + adns_query adns_qu; + char *domain; + int port_no; +}; + +extern struct conn_s *connections; + +/* Handle the creation and deletion of connections */ +extern struct conn_s *new_conn(int fd); +extern int del_conn(struct conn_s *connptr); +extern void conncoll(void); +extern void garbcoll(void); + +#endif diff --git a/src/defines.h.in b/src/defines.h.in new file mode 100644 index 0000000..b7202fe --- /dev/null +++ b/src/defines.h.in @@ -0,0 +1,89 @@ +/* src/defines.h.in. Generated automatically from configure.in by autoheader. */ + +/* Define to empty if the keyword does not work. */ +#undef const + +/* Define if you don't have vprintf but do have _doprnt. */ +#undef HAVE_DOPRNT + +/* Define if you have the strftime function. */ +#undef HAVE_STRFTIME + +/* Define if you have the vprintf function. */ +#undef HAVE_VPRINTF + +/* Define as __inline if that's what the C compiler calls it. */ +#undef inline + +/* Define as the return type of signal handlers (int or void). */ +#undef RETSIGTYPE + +/* Define to `unsigned' if <sys/types.h> doesn't define. */ +#undef size_t + +/* Define if you have the ANSI C header files. */ +#undef STDC_HEADERS + +/* Define if you can safely include both <sys/time.h> and <time.h>. */ +#undef TIME_WITH_SYS_TIME + +/* + * Define if you want to have the peer's IP address to be included in a + * XTinyproxy header sent to the server. + */ +#undef XTINYPROXY + +/* chris - undef'd here so that they can be set up in the configure script */ +#undef DEFAULT_LOG +#undef DEFAULT_PORT +#undef DEFAULT_USER + +/* + * Define if you would like to include filtering code. + */ +#undef FILTER_ENABLE + +/* + * Define if you want to use the included GNU regex routine + */ +#undef USE_GNU_REGEX + +/* Define if you have the regcomp function. */ +#undef HAVE_REGCOMP + +/* Define if you have the select function. */ +#undef HAVE_SELECT + +/* Define if you have the socket function. */ +#undef HAVE_SOCKET + +/* Define if you have the strdup function. */ +#undef HAVE_STRDUP + +/* Define if you have the strerror function. */ +#undef HAVE_STRERROR + +/* Define if you have the vsnprintf function. */ +#undef HAVE_VSNPRINTF + +/* Define if you have the vsyslog function. */ +#undef HAVE_VSYSLOG + +/* Define if you have the <fcntl.h> header file. */ +#undef HAVE_FCNTL_H + +/* Define if you have the <sys/time.h> header file. */ +#undef HAVE_SYS_TIME_H + +/* Define if you have the <syslog.h> header file. */ +#undef HAVE_SYSLOG_H + +/* Define if you have the <unistd.h> header file. */ +#undef HAVE_UNISTD_H + +/* Name of package */ +#undef PACKAGE + +/* Version number of package */ +#undef VERSION + diff --git a/src/dnscache.c b/src/dnscache.c new file mode 100644 index 0000000..ecd7cee --- /dev/null +++ b/src/dnscache.c @@ -0,0 +1,198 @@ +/* $Id: dnscache.c,v 1.1.1.1 2000-02-16 17:32:22 sdyoung Exp $ + * + * This is a caching DNS system. When a host name is needed we look it up here + * and see if there is already an answer for it. The domains are placed in a + * hashed linked list. If the name is not here, then we need to look it up and + * add it to the system. This really speeds up the connection to servers since + * the DNS name does not need to be looked up each time. It's kind of cool. :) + * + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * Copyright (C) 2000 Chris Lightfoot (chris@ex-parrot.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <stdlib.h> +#include <string.h> +#include <netdb.h> +#include <time.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <arpa/inet.h> +#include <assert.h> +#include <ctype.h> + +#include "utils.h" +#include "dnscache.h" + +#define HASH_BOX 25 +#define DNSEXPIRE (5 * 60) + +struct dnscache_s { + struct in_addr ipaddr; + char *domain; + time_t expire; + struct dnscache_s *next; +}; + +struct dnscache_s *cache[HASH_BOX]; + +static unsigned int hash(unsigned char *keystr, unsigned int box) +{ + unsigned long hashc = 0; + unsigned char ch; + + assert(keystr); + assert(box > 0); + + while ((ch = *keystr++)) + hashc += tolower(ch); + + return hashc % box; +} + +int lookup(struct in_addr *addr, char *domain) +{ + unsigned int box = hash(domain, HASH_BOX); + struct dnscache_s **rptr = &cache[box]; + struct dnscache_s *ptr = cache[box]; + + assert(addr); + assert(domain); + + while (ptr && strcasecmp(ptr->domain, domain)) { + rptr = &ptr->next; + ptr = ptr->next; + } + + if (ptr && !strcasecmp(ptr->domain, domain)) { + /* Woohoo... found it. Make sure it hasn't expired */ + if (difftime(time(NULL), ptr->expire) > DNSEXPIRE) { + /* Oops... expired */ + *rptr = ptr->next; + safefree(ptr->domain); + safefree(ptr); + return -1; + } + + /* chris - added this so that the routine can be used to just + * look stuff up. + */ + if (addr) *addr = ptr->ipaddr; + return 0; + } + + return -1; +} + +int insert(struct in_addr *addr, char *domain) +{ + unsigned int box = hash(domain, HASH_BOX); + struct dnscache_s **rptr = &cache[box]; + struct dnscache_s *ptr = cache[box]; + struct dnscache_s *newptr; + + assert(addr); + assert(domain); + + while (ptr) { + rptr = &ptr->next; + ptr = ptr->next; + } + + if (!(newptr = xmalloc(sizeof(struct dnscache_s)))) { + return -1; + } + + if (!(newptr->domain = xstrdup(domain))) { + safefree(newptr); + return -1; + } + + newptr->ipaddr = *addr; + + newptr->expire = time(NULL); + + *rptr = newptr; + newptr->next = ptr; + + return 0; +} + +int dnscache(struct in_addr *addr, char *domain) +{ + struct hostent *resolv; + + assert(addr); + assert(domain); + + if (inet_aton(domain, (struct in_addr *) addr) != 0) + return 0; + + /* Well, we're not dotted-decimal so we need to look it up */ + if (lookup(addr, domain) == 0) + return 0; + + /* Okay, so not in the list... need to actually look it up. */ + if (!(resolv = gethostbyname(domain))) + return -1; + + memcpy(addr, resolv->h_addr_list[0], resolv->h_length); + insert(addr, domain); + + return 0; +} + +static void dnsdelete(unsigned int c, struct dnscache_s *del) +{ + struct dnscache_s **rptr; + struct dnscache_s *ptr; + + assert(c > 0); + assert(del); + + rptr = &cache[c]; + ptr = cache[c]; + + while (ptr && (ptr != del)) { + rptr = &ptr->next; + ptr = ptr->next; + } + + if (ptr == del) { + *rptr = ptr->next; + safefree(ptr->domain); + safefree(ptr); + } +} + +void dnsclean(void) +{ + unsigned int c; + struct dnscache_s *ptr, *tmp; + + for (c = 0; c < HASH_BOX; c++) { + ptr = cache[c]; + + while (ptr) { + tmp = ptr->next; + + if (difftime(time(NULL), ptr->expire) > DNSEXPIRE) + dnsdelete(c, ptr); + + ptr = tmp; + } + } +} diff --git a/src/dnscache.h b/src/dnscache.h new file mode 100644 index 0000000..ecb3cd6 --- /dev/null +++ b/src/dnscache.h @@ -0,0 +1,38 @@ +/* $Id: dnscache.h,v 1.1.1.1 2000-02-16 17:32:22 sdyoung Exp $ + * + * See 'dnscache.c' for a detailed description. + * + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef _DNSCACHE_H_ +#define _DNSCACHE_H_ 1 + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <sys/socket.h> +#include <netinet/in.h> +#include <arpa/inet.h> + +#define DNS_GARBAGE_COL 100 + +extern int dnscache(struct in_addr *addr, char *domain); +extern void dnsclean(void); + +/* chris - Access these from reqs.c because of ADNS. Ugly. */ +extern int lookup(struct in_addr *addr, char *domain); +extern int insert(struct in_addr *addr, char *domain); + +#endif diff --git a/src/filter.c b/src/filter.c new file mode 100644 index 0000000..4571b8a --- /dev/null +++ b/src/filter.c @@ -0,0 +1,150 @@ +/* $Id: filter.c,v 1.1.1.1 2000-02-16 17:32:18 sdyoung Exp $ + * + * Copyright (c) 1999 George Talusan (gstalusan@uwaterloo.ca) + * + * A substring of the domain to be filtered goes into the file + * pointed at by DEFAULT_FILTER. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <sys/types.h> +#include <sys/stat.h> +#include <fcntl.h> +#include <unistd.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <ctype.h> +#include <assert.h> + +#include "config.h" +#include "utils.h" +#include "tinyproxy.h" +#include "filter.h" + +#include "regexp.h" +#include <sysexits.h> + +static int err; + +struct filter_list { + struct filter_list *next; + char *pat; + regex_t *cpat; +}; + + +static struct filter_list *fl = NULL; +static int already_init = 0; + +/* initializes a linked list of strings containing hosts to be filtered */ +void filter_init(void) +{ + FILE *fd; + struct filter_list *p; + char buf[255]; + char *s; + + if (!fl && !already_init) { + fd = fopen(config.filter, "r"); + if (fd) { + p = NULL; + + while (fgets(buf, 255, (FILE *) fd)) { + s = buf; + if (!p) /* head of list */ + fl = p = (struct filter_list *) + xmalloc(sizeof + (struct filter_list)); + else { /* next entry */ + p->next = (struct filter_list *) + xmalloc(sizeof + (struct filter_list)); + p = p->next; + } + /* initialize space to zero */ + memset(p, 0, sizeof(struct filter_list)); + + /* replace first whitespace with \0 */ + while (*s++) + if (isspace((int) *s)) + *s = '\0'; + + p->pat = xstrdup(buf); + p->cpat = xmalloc(sizeof(regex_t)); + if ( + (err = + regcomp(p->cpat, p->pat, + REG_NEWLINE | REG_NOSUB)) != 0) { + fprintf(stderr, + "Bad regex in %s: %s\n", + config.filter, p->pat); + exit(EX_DATAERR); + } + } + already_init = 1; + fclose(fd); + } + } +} + +/* unlink the list */ +void filter_destroy(void) +{ + struct filter_list *p, *q; + + if (already_init) { + for (p = q = fl; p; p = q) { + regfree(p->cpat); + safefree(p->cpat); + safefree(p->pat); + q = p->next; + safefree(p); + } + fl = NULL; + already_init = 0; + } +} + +/* returns 0 if host is not an element of filter list, non-zero otherwise */ +int filter_host(char *host) +{ + struct filter_list *p; + char *s, *port; + int result; + + assert(host); + + if (!fl || !already_init) + return (0); + + /* strip off the port number */ + s = xstrdup(host); + port = strchr(s, ':'); + if (port) + *port = '\0'; + + result = 0; + + for (p = fl; p; p = p->next) { + result = !regexec(p->cpat, s, (size_t) 0, (regmatch_t *) 0, 0); + + if (result) + break; + } + safefree(s); + return (result); +} diff --git a/src/filter.h b/src/filter.h new file mode 100644 index 0000000..fb08f2f --- /dev/null +++ b/src/filter.h @@ -0,0 +1,25 @@ +/* $Id: filter.h,v 1.1.1.1 2000-02-16 17:32:24 sdyoung Exp $ + * + * See 'filter.c' for a detailed description. + * + * Copyright (c) 1999 George Talusan (gstalusan@uwaterloo.ca) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef __FILTER_H_ +#define __FILTER_H_ 1 + +extern void filter_init(void); +extern void filter_destroy(void); +extern int filter_host(char *host); + +#endif diff --git a/src/gnuregex.c b/src/gnuregex.c new file mode 100644 index 0000000..785d419 --- /dev/null +++ b/src/gnuregex.c @@ -0,0 +1,5880 @@ +/* Extended regular expression matching and search library, + version 0.12. + (Implements POSIX draft P1003.2/D11.2, except for some of the + internationalization features.) + Copyright (C) 1993, 94, 95, 96, 97, 98, 99 Free Software Foundation, Inc. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +/* AIX requires this to be the first thing in the file. */ +#if defined _AIX && !defined REGEX_MALLOC + #pragma alloca +#endif + +#undef _GNU_SOURCE +#define _GNU_SOURCE + +#ifdef HAVE_CONFIG_H +# include <defines.h> /* Tinyproxy uses defines.h for autoconf defs */ +#endif + +#ifndef PARAMS +# if defined __GNUC__ || (defined __STDC__ && __STDC__) +# define PARAMS(args) args +# else +# define PARAMS(args) () +# endif /* GCC. */ +#endif /* Not PARAMS. */ + +#if defined STDC_HEADERS && !defined emacs +# include <stddef.h> +#else +/* We need this for `regex.h', and perhaps for the Emacs include files. */ +# include <sys/types.h> +#endif + +#define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC) + +/* For platform which support the ISO C amendement 1 functionality we + support user defined character classes. */ +#if defined _LIBC || WIDE_CHAR_SUPPORT +/* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */ +# include <wchar.h> +# include <wctype.h> +#endif + +#ifdef _LIBC +/* We have to keep the namespace clean. */ +# define regfree(preg) __regfree (preg) +# define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef) +# define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags) +# define regerror(errcode, preg, errbuf, errbuf_size) \ + __regerror(errcode, preg, errbuf, errbuf_size) +# define re_set_registers(bu, re, nu, st, en) \ + __re_set_registers (bu, re, nu, st, en) +# define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \ + __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop) +# define re_match(bufp, string, size, pos, regs) \ + __re_match (bufp, string, size, pos, regs) +# define re_search(bufp, string, size, startpos, range, regs) \ + __re_search (bufp, string, size, startpos, range, regs) +# define re_compile_pattern(pattern, length, bufp) \ + __re_compile_pattern (pattern, length, bufp) +# define re_set_syntax(syntax) __re_set_syntax (syntax) +# define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \ + __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop) +# define re_compile_fastmap(bufp) __re_compile_fastmap (bufp) + +#define btowc __btowc +#endif + +/* This is for other GNU distributions with internationalized messages. */ +#if HAVE_LIBINTL_H || defined _LIBC +# include <libintl.h> +#else +# define gettext(msgid) (msgid) +#endif + +#ifndef gettext_noop +/* This define is so xgettext can find the internationalizable + strings. */ +# define gettext_noop(String) String +#endif + +/* The `emacs' switch turns on certain matching commands + that make sense only in Emacs. */ +#ifdef emacs + +# include "lisp.h" +# include "buffer.h" +# include "syntax.h" + +#else /* not emacs */ + +/* If we are not linking with Emacs proper, + we can't use the relocating allocator + even if config.h says that we can. */ +# undef REL_ALLOC + +# if defined STDC_HEADERS || defined _LIBC +# include <stdlib.h> +# else +char *malloc (); +char *realloc (); +# endif + +/* When used in Emacs's lib-src, we need to get bzero and bcopy somehow. + If nothing else has been done, use the method below. */ +# ifdef INHIBIT_STRING_HEADER +# if !(defined HAVE_BZERO && defined HAVE_BCOPY) +# if !defined bzero && !defined bcopy +# undef INHIBIT_STRING_HEADER +# endif +# endif +# endif + +/* This is the normal way of making sure we have a bcopy and a bzero. + This is used in most programs--a few other programs avoid this + by defining INHIBIT_STRING_HEADER. */ +# ifndef INHIBIT_STRING_HEADER +# if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC +# include <string.h> +# ifndef bzero +# ifndef _LIBC +# define bzero(s, n) (memset (s, '\0', n), (s)) +# else +# define bzero(s, n) __bzero (s, n) +# endif +# endif +# else +# include <strings.h> +# ifndef memcmp +# define memcmp(s1, s2, n) bcmp (s1, s2, n) +# endif +# ifndef memcpy +# define memcpy(d, s, n) (bcopy (s, d, n), (d)) +# endif +# endif +# endif + +/* Define the syntax stuff for \<, \>, etc. */ + +/* This must be nonzero for the wordchar and notwordchar pattern + commands in re_match_2. */ +# ifndef Sword +# define Sword 1 +# endif + +# ifdef SWITCH_ENUM_BUG +# define SWITCH_ENUM_CAST(x) ((int)(x)) +# else +# define SWITCH_ENUM_CAST(x) (x) +# endif + +/* How many characters in the character set. */ +# define CHAR_SET_SIZE 256 + +# ifdef SYNTAX_TABLE + +extern char *re_syntax_table; + +# else /* not SYNTAX_TABLE */ + +static char re_syntax_table[CHAR_SET_SIZE]; + +static void +init_syntax_once () +{ + register int c; + static int done; + + if (done) + return; + + bzero (re_syntax_table, sizeof re_syntax_table); + + for (c = 'a'; c <= 'z'; c++) + re_syntax_table[c] = Sword; + + for (c = 'A'; c <= 'Z'; c++) + re_syntax_table[c] = Sword; + + for (c = '0'; c <= '9'; c++) + re_syntax_table[c] = Sword; + + re_syntax_table['_'] = Sword; + + done = 1; +} + +# endif /* not SYNTAX_TABLE */ + +# define SYNTAX(c) re_syntax_table[c] + +#endif /* not emacs */ + +/* Get the interface, including the syntax bits. */ +#include "gnuregex.h" + +/* isalpha etc. are used for the character classes. */ +#include <ctype.h> + +/* Jim Meyering writes: + + "... Some ctype macros are valid only for character codes that + isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when + using /bin/cc or gcc but without giving an ansi option). So, all + ctype uses should be through macros like ISPRINT... If + STDC_HEADERS is defined, then autoconf has verified that the ctype + macros don't need to be guarded with references to isascii. ... + Defining isascii to 1 should let any compiler worth its salt + eliminate the && through constant folding." + Solaris defines some of these symbols so we must undefine them first. */ + +#undef ISASCII +#if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII) +# define ISASCII(c) 1 +#else +# define ISASCII(c) isascii(c) +#endif + +#ifdef isblank +# define ISBLANK(c) (ISASCII (c) && isblank (c)) +#else +# define ISBLANK(c) ((c) == ' ' || (c) == '\t') +#endif +#ifdef isgraph +# define ISGRAPH(c) (ISASCII (c) && isgraph (c)) +#else +# define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c)) +#endif + +#undef ISPRINT +#define ISPRINT(c) (ISASCII (c) && isprint (c)) +#define ISDIGIT(c) (ISASCII (c) && isdigit (c)) +#define ISALNUM(c) (ISASCII (c) && isalnum (c)) +#define ISALPHA(c) (ISASCII (c) && isalpha (c)) +#define ISCNTRL(c) (ISASCII (c) && iscntrl (c)) +#define ISLOWER(c) (ISASCII (c) && islower (c)) +#define ISPUNCT(c) (ISASCII (c) && ispunct (c)) +#define ISSPACE(c) (ISASCII (c) && isspace (c)) +#define ISUPPER(c) (ISASCII (c) && isupper (c)) +#define ISXDIGIT(c) (ISASCII (c) && isxdigit (c)) + +#ifdef _tolower +# define TOLOWER(c) _tolower(c) +#else +# define TOLOWER(c) tolower(c) +#endif + +#ifndef NULL +# define NULL (void *)0 +#endif + +/* We remove any previous definition of `SIGN_EXTEND_CHAR', + since ours (we hope) works properly with all combinations of + machines, compilers, `char' and `unsigned char' argument types. + (Per Bothner suggested the basic approach.) */ +#undef SIGN_EXTEND_CHAR +#if __STDC__ +# define SIGN_EXTEND_CHAR(c) ((signed char) (c)) +#else /* not __STDC__ */ +/* As in Harbison and Steele. */ +# define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128) +#endif + +/* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we + use `alloca' instead of `malloc'. This is because using malloc in + re_search* or re_match* could cause memory leaks when C-g is used in + Emacs; also, malloc is slower and causes storage fragmentation. On + the other hand, malloc is more portable, and easier to debug. + + Because we sometimes use alloca, some routines have to be macros, + not functions -- `alloca'-allocated space disappears at the end of the + function it is called in. */ + +#ifdef REGEX_MALLOC + +# define REGEX_ALLOCATE malloc +# define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize) +# define REGEX_FREE free + +#else /* not REGEX_MALLOC */ + +/* Emacs already defines alloca, sometimes. */ +# ifndef alloca + +/* Make alloca work the best possible way. */ +# ifdef __GNUC__ +# define alloca __builtin_alloca +# else /* not __GNUC__ */ +# if HAVE_ALLOCA_H +# include <alloca.h> +# endif /* HAVE_ALLOCA_H */ +# endif /* not __GNUC__ */ + +# endif /* not alloca */ + +# define REGEX_ALLOCATE alloca + +/* Assumes a `char *destination' variable. */ +# define REGEX_REALLOCATE(source, osize, nsize) \ + (destination = (char *) alloca (nsize), \ + memcpy (destination, source, osize)) + +/* No need to do anything to free, after alloca. */ +# define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */ + +#endif /* not REGEX_MALLOC */ + +/* Define how to allocate the failure stack. */ + +#if defined REL_ALLOC && defined REGEX_MALLOC + +# define REGEX_ALLOCATE_STACK(size) \ + r_alloc (&failure_stack_ptr, (size)) +# define REGEX_REALLOCATE_STACK(source, osize, nsize) \ + r_re_alloc (&failure_stack_ptr, (nsize)) +# define REGEX_FREE_STACK(ptr) \ + r_alloc_free (&failure_stack_ptr) + +#else /* not using relocating allocator */ + +# ifdef REGEX_MALLOC + +# define REGEX_ALLOCATE_STACK malloc +# define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize) +# define REGEX_FREE_STACK free + +# else /* not REGEX_MALLOC */ + +# define REGEX_ALLOCATE_STACK alloca + +# define REGEX_REALLOCATE_STACK(source, osize, nsize) \ + REGEX_REALLOCATE (source, osize, nsize) +/* No need to explicitly free anything. */ +# define REGEX_FREE_STACK(arg) + +# endif /* not REGEX_MALLOC */ +#endif /* not using relocating allocator */ + + +/* True if `size1' is non-NULL and PTR is pointing anywhere inside + `string1' or just past its end. This works if PTR is NULL, which is + a good thing. */ +#define FIRST_STRING_P(ptr) \ + (size1 && string1 <= (ptr) && (ptr) <= string1 + size1) + +/* (Re)Allocate N items of type T using malloc, or fail. */ +#define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t))) +#define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t))) +#define RETALLOC_IF(addr, n, t) \ + if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t) +#define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t))) + +#define BYTEWIDTH 8 /* In bits. */ + +#define STREQ(s1, s2) ((strcmp (s1, s2) == 0)) + +#undef MAX +#undef MIN +#define MAX(a, b) ((a) > (b) ? (a) : (b)) +#define MIN(a, b) ((a) < (b) ? (a) : (b)) + +typedef char boolean; +#define false 0 +#define true 1 + +static int re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp, + const char *string1, int size1, + const char *string2, int size2, + int pos, + struct re_registers *regs, + int stop)); + +/* These are the command codes that appear in compiled regular + expressions. Some opcodes are followed by argument bytes. A + command code can specify any interpretation whatsoever for its + arguments. Zero bytes may appear in the compiled regular expression. */ + +typedef enum +{ + no_op = 0, + + /* Succeed right away--no more backtracking. */ + succeed, + + /* Followed by one byte giving n, then by n literal bytes. */ + exactn, + + /* Matches any (more or less) character. */ + anychar, + + /* Matches any one char belonging to specified set. First + following byte is number of bitmap bytes. Then come bytes + for a bitmap saying which chars are in. Bits in each byte + are ordered low-bit-first. A character is in the set if its + bit is 1. A character too large to have a bit in the map is + automatically not in the set. */ + charset, + + /* Same parameters as charset, but match any character that is + not one of those specified. */ + charset_not, + + /* Start remembering the text that is matched, for storing in a + register. Followed by one byte with the register number, in + the range 0 to one less than the pattern buffer's re_nsub + field. Then followed by one byte with the number of groups + inner to this one. (This last has to be part of the + start_memory only because we need it in the on_failure_jump + of re_match_2.) */ + start_memory, + + /* Stop remembering the text that is matched and store it in a + memory register. Followed by one byte with the register + number, in the range 0 to one less than `re_nsub' in the + pattern buffer, and one byte with the number of inner groups, + just like `start_memory'. (We need the number of inner + groups here because we don't have any easy way of finding the + corresponding start_memory when we're at a stop_memory.) */ + stop_memory, + + /* Match a duplicate of something remembered. Followed by one + byte containing the register number. */ + duplicate, + + /* Fail unless at beginning of line. */ + begline, + + /* Fail unless at end of line. */ + endline, + + /* Succeeds if at beginning of buffer (if emacs) or at beginning + of string to be matched (if not). */ + begbuf, + + /* Analogously, for end of buffer/string. */ + endbuf, + + /* Followed by two byte relative address to which to jump. */ + jump, + + /* Same as jump, but marks the end of an alternative. */ + jump_past_alt, + + /* Followed by two-byte relative address of place to resume at + in case of failure. */ + on_failure_jump, + + /* Like on_failure_jump, but pushes a placeholder instead of the + current string position when executed. */ + on_failure_keep_string_jump, + + /* Throw away latest failure point and then jump to following + two-byte relative address. */ + pop_failure_jump, + + /* Change to pop_failure_jump if know won't have to backtrack to + match; otherwise change to jump. This is used to jump + back to the beginning of a repeat. If what follows this jump + clearly won't match what the repeat does, such that we can be + sure that there is no use backtracking out of repetitions + already matched, then we change it to a pop_failure_jump. + Followed by two-byte address. */ + maybe_pop_jump, + + /* Jump to following two-byte address, and push a dummy failure + point. This failure point will be thrown away if an attempt + is made to use it for a failure. A `+' construct makes this + before the first repeat. Also used as an intermediary kind + of jump when compiling an alternative. */ + dummy_failure_jump, + + /* Push a dummy failure point and continue. Used at the end of + alternatives. */ + push_dummy_failure, + + /* Followed by two-byte relative address and two-byte number n. + After matching N times, jump to the address upon failure. */ + succeed_n, + + /* Followed by two-byte relative address, and two-byte number n. + Jump to the address N times, then fail. */ + jump_n, + + /* Set the following two-byte relative address to the + subsequent two-byte number. The address *includes* the two + bytes of number. */ + set_number_at, + + wordchar, /* Matches any word-constituent character. */ + notwordchar, /* Matches any char that is not a word-constituent. */ + + wordbeg, /* Succeeds if at word beginning. */ + wordend, /* Succeeds if at word end. */ + + wordbound, /* Succeeds if at a word boundary. */ + notwordbound /* Succeeds if not at a word boundary. */ + +#ifdef emacs + ,before_dot, /* Succeeds if before point. */ + at_dot, /* Succeeds if at point. */ + after_dot, /* Succeeds if after point. */ + + /* Matches any character whose syntax is specified. Followed by + a byte which contains a syntax code, e.g., Sword. */ + syntaxspec, + + /* Matches any character whose syntax is not that specified. */ + notsyntaxspec +#endif /* emacs */ +} re_opcode_t; + +/* Common operations on the compiled pattern. */ + +/* Store NUMBER in two contiguous bytes starting at DESTINATION. */ + +#define STORE_NUMBER(destination, number) \ + do { \ + (destination)[0] = (number) & 0377; \ + (destination)[1] = (number) >> 8; \ + } while (0) + +/* Same as STORE_NUMBER, except increment DESTINATION to + the byte after where the number is stored. Therefore, DESTINATION + must be an lvalue. */ + +#define STORE_NUMBER_AND_INCR(destination, number) \ + do { \ + STORE_NUMBER (destination, number); \ + (destination) += 2; \ + } while (0) + +/* Put into DESTINATION a number stored in two contiguous bytes starting + at SOURCE. */ + +#define EXTRACT_NUMBER(destination, source) \ + do { \ + (destination) = *(source) & 0377; \ + (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \ + } while (0) + +#ifdef DEBUG +static void extract_number _RE_ARGS ((int *dest, unsigned char *source)); +static void +extract_number (dest, source) + int *dest; + unsigned char *source; +{ + int temp = SIGN_EXTEND_CHAR (*(source + 1)); + *dest = *source & 0377; + *dest += temp << 8; +} + +# ifndef EXTRACT_MACROS /* To debug the macros. */ +# undef EXTRACT_NUMBER +# define EXTRACT_NUMBER(dest, src) extract_number (&dest, src) +# endif /* not EXTRACT_MACROS */ + +#endif /* DEBUG */ + +/* Same as EXTRACT_NUMBER, except increment SOURCE to after the number. + SOURCE must be an lvalue. */ + +#define EXTRACT_NUMBER_AND_INCR(destination, source) \ + do { \ + EXTRACT_NUMBER (destination, source); \ + (source) += 2; \ + } while (0) + +#ifdef DEBUG +static void extract_number_and_incr _RE_ARGS ((int *destination, + unsigned char **source)); +static void +extract_number_and_incr (destination, source) + int *destination; + unsigned char **source; +{ + extract_number (destination, *source); + *source += 2; +} + +# ifndef EXTRACT_MACROS +# undef EXTRACT_NUMBER_AND_INCR +# define EXTRACT_NUMBER_AND_INCR(dest, src) \ + extract_number_and_incr (&dest, &src) +# endif /* not EXTRACT_MACROS */ + +#endif /* DEBUG */ + +/* If DEBUG is defined, Regex prints many voluminous messages about what + it is doing (if the variable `debug' is nonzero). If linked with the + main program in `iregex.c', you can enter patterns and strings + interactively. And if linked with the main program in `main.c' and + the other test files, you can run the already-written tests. */ + +#ifdef DEBUG + +/* We use standard I/O for debugging. */ +# include <stdio.h> + +/* It is useful to test things that ``must'' be true when debugging. */ +# include <assert.h> + +static int debug; + +# define DEBUG_STATEMENT(e) e +# define DEBUG_PRINT1(x) if (debug) printf (x) +# define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2) +# define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3) +# define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4) +# define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \ + if (debug) print_partial_compiled_pattern (s, e) +# define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ + if (debug) print_double_string (w, s1, sz1, s2, sz2) + + +/* Print the fastmap in human-readable form. */ + +void +print_fastmap (fastmap) + char *fastmap; +{ + unsigned was_a_range = 0; + unsigned i = 0; + + while (i < (1 << BYTEWIDTH)) + { + if (fastmap[i++]) + { + was_a_range = 0; + putchar (i - 1); + while (i < (1 << BYTEWIDTH) && fastmap[i]) + { + was_a_range = 1; + i++; + } + if (was_a_range) + { + printf ("-"); + putchar (i - 1); + } + } + } + putchar ('\n'); +} + + +/* Print a compiled pattern string in human-readable form, starting at + the START pointer into it and ending just before the pointer END. */ + +void +print_partial_compiled_pattern (start, end) + unsigned char *start; + unsigned char *end; +{ + int mcnt, mcnt2; + unsigned char *p1; + unsigned char *p = start; + unsigned char *pend = end; + + if (start == NULL) + { + printf ("(null)\n"); + return; + } + + /* Loop over pattern commands. */ + while (p < pend) + { + printf ("%d:\t", p - start); + + switch ((re_opcode_t) *p++) + { + case no_op: + printf ("/no_op"); + break; + + case exactn: + mcnt = *p++; + printf ("/exactn/%d", mcnt); + do + { + putchar ('/'); + putchar (*p++); + } + while (--mcnt); + break; + + case start_memory: + mcnt = *p++; + printf ("/start_memory/%d/%d", mcnt, *p++); + break; + + case stop_memory: + mcnt = *p++; + printf ("/stop_memory/%d/%d", mcnt, *p++); + break; + + case duplicate: + printf ("/duplicate/%d", *p++); + break; + + case anychar: + printf ("/anychar"); + break; + + case charset: + case charset_not: + { + register int c, last = -100; + register int in_range = 0; + + printf ("/charset [%s", + (re_opcode_t) *(p - 1) == charset_not ? "^" : ""); + + assert (p + *p < pend); + + for (c = 0; c < 256; c++) + if (c / 8 < *p + && (p[1 + (c/8)] & (1 << (c % 8)))) + { + /* Are we starting a range? */ + if (last + 1 == c && ! in_range) + { + putchar ('-'); + in_range = 1; + } + /* Have we broken a range? */ + else if (last + 1 != c && in_range) + { + putchar (last); + in_range = 0; + } + + if (! in_range) + putchar (c); + + last = c; + } + + if (in_range) + putchar (last); + + putchar (']'); + + p += 1 + *p; + } + break; + + case begline: + printf ("/begline"); + break; + + case endline: + printf ("/endline"); + break; + + case on_failure_jump: + extract_number_and_incr (&mcnt, &p); + printf ("/on_failure_jump to %d", p + mcnt - start); + break; + + case on_failure_keep_string_jump: + extract_number_and_incr (&mcnt, &p); + printf ("/on_failure_keep_string_jump to %d", p + mcnt - start); + break; + + case dummy_failure_jump: + extract_number_and_incr (&mcnt, &p); + printf ("/dummy_failure_jump to %d", p + mcnt - start); + break; + + case push_dummy_failure: + printf ("/push_dummy_failure"); + break; + + case maybe_pop_jump: + extract_number_and_incr (&mcnt, &p); + printf ("/maybe_pop_jump to %d", p + mcnt - start); + break; + + case pop_failure_jump: + extract_number_and_incr (&mcnt, &p); + printf ("/pop_failure_jump to %d", p + mcnt - start); + break; + + case jump_past_alt: + extract_number_and_incr (&mcnt, &p); + printf ("/jump_past_alt to %d", p + mcnt - start); + break; + + case jump: + extract_number_and_incr (&mcnt, &p); + printf ("/jump to %d", p + mcnt - start); + break; + + case succeed_n: + extract_number_and_incr (&mcnt, &p); + p1 = p + mcnt; + extract_number_and_incr (&mcnt2, &p); + printf ("/succeed_n to %d, %d times", p1 - start, mcnt2); + break; + + case jump_n: + extract_number_and_incr (&mcnt, &p); + p1 = p + mcnt; + extract_number_and_incr (&mcnt2, &p); + printf ("/jump_n to %d, %d times", p1 - start, mcnt2); + break; + + case set_number_at: + extract_number_and_incr (&mcnt, &p); + p1 = p + mcnt; + extract_number_and_incr (&mcnt2, &p); + printf ("/set_number_at location %d to %d", p1 - start, mcnt2); + break; + + case wordbound: + printf ("/wordbound"); + break; + + case notwordbound: + printf ("/notwordbound"); + break; + + case wordbeg: + printf ("/wordbeg"); + break; + + case wordend: + printf ("/wordend"); + +# ifdef emacs + case before_dot: + printf ("/before_dot"); + break; + + case at_dot: + printf ("/at_dot"); + break; + + case after_dot: + printf ("/after_dot"); + break; + + case syntaxspec: + printf ("/syntaxspec"); + mcnt = *p++; + printf ("/%d", mcnt); + break; + + case notsyntaxspec: + printf ("/notsyntaxspec"); + mcnt = *p++; + printf ("/%d", mcnt); + break; +# endif /* emacs */ + + case wordchar: + printf ("/wordchar"); + break; + + case notwordchar: + printf ("/notwordchar"); + break; + + case begbuf: + printf ("/begbuf"); + break; + + case endbuf: + printf ("/endbuf"); + break; + + default: + printf ("?%d", *(p-1)); + } + + putchar ('\n'); + } + + printf ("%d:\tend of pattern.\n", p - start); +} + + +void +print_compiled_pattern (bufp) + struct re_pattern_buffer *bufp; +{ + unsigned char *buffer = bufp->buffer; + + print_partial_compiled_pattern (buffer, buffer + bufp->used); + printf ("%ld bytes used/%ld bytes allocated.\n", + bufp->used, bufp->allocated); + + if (bufp->fastmap_accurate && bufp->fastmap) + { + printf ("fastmap: "); + print_fastmap (bufp->fastmap); + } + + printf ("re_nsub: %d\t", bufp->re_nsub); + printf ("regs_alloc: %d\t", bufp->regs_allocated); + printf ("can_be_null: %d\t", bufp->can_be_null); + printf ("newline_anchor: %d\n", bufp->newline_anchor); + printf ("no_sub: %d\t", bufp->no_sub); + printf ("not_bol: %d\t", bufp->not_bol); + printf ("not_eol: %d\t", bufp->not_eol); + printf ("syntax: %lx\n", bufp->syntax); + /* Perhaps we should print the translate table? */ +} + + +void +print_double_string (where, string1, size1, string2, size2) + const char *where; + const char *string1; + const char *string2; + int size1; + int size2; +{ + int this_char; + + if (where == NULL) + printf ("(null)"); + else + { + if (FIRST_STRING_P (where)) + { + for (this_char = where - string1; this_char < size1; this_char++) + putchar (string1[this_char]); + + where = string2; + } + + for (this_char = where - string2; this_char < size2; this_char++) + putchar (string2[this_char]); + } +} + +void +printchar (c) + int c; +{ + putc (c, stderr); +} + +#else /* not DEBUG */ + +# undef assert +# define assert(e) + +# define DEBUG_STATEMENT(e) +# define DEBUG_PRINT1(x) +# define DEBUG_PRINT2(x1, x2) +# define DEBUG_PRINT3(x1, x2, x3) +# define DEBUG_PRINT4(x1, x2, x3, x4) +# define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) +# define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) + +#endif /* not DEBUG */ + +/* Set by `re_set_syntax' to the current regexp syntax to recognize. Can + also be assigned to arbitrarily: each pattern buffer stores its own + syntax, so it can be changed between regex compilations. */ +/* This has no initializer because initialized variables in Emacs + become read-only after dumping. */ +reg_syntax_t re_syntax_options; + + +/* Specify the precise syntax of regexps for compilation. This provides + for compatibility for various utilities which historically have + different, incompatible syntaxes. + + The argument SYNTAX is a bit mask comprised of the various bits + defined in regex.h. We return the old syntax. */ + +reg_syntax_t +re_set_syntax (syntax) + reg_syntax_t syntax; +{ + reg_syntax_t ret = re_syntax_options; + + re_syntax_options = syntax; +#ifdef DEBUG + if (syntax & RE_DEBUG) + debug = 1; + else if (debug) /* was on but now is not */ + debug = 0; +#endif /* DEBUG */ + return ret; +} +#ifdef _LIBC +weak_alias (__re_set_syntax, re_set_syntax) +#endif + +/* This table gives an error message for each of the error codes listed + in regex.h. Obviously the order here has to be same as there. + POSIX doesn't require that we do anything for REG_NOERROR, + but why not be nice? */ + +static const char re_error_msgid[] = + { +#define REG_NOERROR_IDX 0 + gettext_noop ("Success") /* REG_NOERROR */ + "\0" +#define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success") + gettext_noop ("No match") /* REG_NOMATCH */ + "\0" +#define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match") + gettext_noop ("Invalid regular expression") /* REG_BADPAT */ + "\0" +#define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression") + gettext_noop ("Invalid collation character") /* REG_ECOLLATE */ + "\0" +#define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character") + gettext_noop ("Invalid character class name") /* REG_ECTYPE */ + "\0" +#define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name") + gettext_noop ("Trailing backslash") /* REG_EESCAPE */ + "\0" +#define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash") + gettext_noop ("Invalid back reference") /* REG_ESUBREG */ + "\0" +#define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference") + gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */ + "\0" +#define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^") + gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */ + "\0" +#define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(") + gettext_noop ("Unmatched \\{") /* REG_EBRACE */ + "\0" +#define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{") + gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */ + "\0" +#define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}") + gettext_noop ("Invalid range end") /* REG_ERANGE */ + "\0" +#define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end") + gettext_noop ("Memory exhausted") /* REG_ESPACE */ + "\0" +#define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted") + gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */ + "\0" +#define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression") + gettext_noop ("Premature end of regular expression") /* REG_EEND */ + "\0" +#define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression") + gettext_noop ("Regular expression too big") /* REG_ESIZE */ + "\0" +#define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big") + gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */ + }; + +static const size_t re_error_msgid_idx[] = + { + REG_NOERROR_IDX, + REG_NOMATCH_IDX, + REG_BADPAT_IDX, + REG_ECOLLATE_IDX, + REG_ECTYPE_IDX, + REG_EESCAPE_IDX, + REG_ESUBREG_IDX, + REG_EBRACK_IDX, + REG_EPAREN_IDX, + REG_EBRACE_IDX, + REG_BADBR_IDX, + REG_ERANGE_IDX, + REG_ESPACE_IDX, + REG_BADRPT_IDX, + REG_EEND_IDX, + REG_ESIZE_IDX, + REG_ERPAREN_IDX + }; + +/* Avoiding alloca during matching, to placate r_alloc. */ + +/* Define MATCH_MAY_ALLOCATE unless we need to make sure that the + searching and matching functions should not call alloca. On some + systems, alloca is implemented in terms of malloc, and if we're + using the relocating allocator routines, then malloc could cause a + relocation, which might (if the strings being searched are in the + ralloc heap) shift the data out from underneath the regexp + routines. + + Here's another reason to avoid allocation: Emacs + processes input from X in a signal handler; processing X input may + call malloc; if input arrives while a matching routine is calling + malloc, then we're scrod. But Emacs can't just block input while + calling matching routines; then we don't notice interrupts when + they come in. So, Emacs blocks input around all regexp calls + except the matching calls, which it leaves unprotected, in the + faith that they will not malloc. */ + +/* Normally, this is fine. */ +#define MATCH_MAY_ALLOCATE + +/* When using GNU C, we are not REALLY using the C alloca, no matter + what config.h may say. So don't take precautions for it. */ +#ifdef __GNUC__ +# undef C_ALLOCA +#endif + +/* The match routines may not allocate if (1) they would do it with malloc + and (2) it's not safe for them to use malloc. + Note that if REL_ALLOC is defined, matching would not use malloc for the + failure stack, but we would still use it for the register vectors; + so REL_ALLOC should not affect this. */ +#if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs +# undef MATCH_MAY_ALLOCATE +#endif + + +/* Failure stack declarations and macros; both re_compile_fastmap and + re_match_2 use a failure stack. These have to be macros because of + REGEX_ALLOCATE_STACK. */ + + +/* Number of failure points for which to initially allocate space + when matching. If this number is exceeded, we allocate more + space, so it is not a hard limit. */ +#ifndef INIT_FAILURE_ALLOC +# define INIT_FAILURE_ALLOC 5 +#endif + +/* Roughly the maximum number of failure points on the stack. Would be + exactly that if always used MAX_FAILURE_ITEMS items each time we failed. + This is a variable only so users of regex can assign to it; we never + change it ourselves. */ + +#ifdef INT_IS_16BIT + +# if defined MATCH_MAY_ALLOCATE +/* 4400 was enough to cause a crash on Alpha OSF/1, + whose default stack limit is 2mb. */ +long int re_max_failures = 4000; +# else +long int re_max_failures = 2000; +# endif + +union fail_stack_elt +{ + unsigned char *pointer; + long int integer; +}; + +typedef union fail_stack_elt fail_stack_elt_t; + +typedef struct +{ + fail_stack_elt_t *stack; + unsigned long int size; + unsigned long int avail; /* Offset of next open position. */ +} fail_stack_type; + +#else /* not INT_IS_16BIT */ + +# if defined MATCH_MAY_ALLOCATE +/* 4400 was enough to cause a crash on Alpha OSF/1, + whose default stack limit is 2mb. */ +int re_max_failures = 20000; +# else +int re_max_failures = 2000; +# endif + +union fail_stack_elt +{ + unsigned char *pointer; + int integer; +}; + +typedef union fail_stack_elt fail_stack_elt_t; + +typedef struct +{ + fail_stack_elt_t *stack; + unsigned size; + unsigned avail; /* Offset of next open position. */ +} fail_stack_type; + +#endif /* INT_IS_16BIT */ + +#define FAIL_STACK_EMPTY() (fail_stack.avail == 0) +#define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0) +#define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size) + + +/* Define macros to initialize and free the failure stack. + Do `return -2' if the alloc fails. */ + +#ifdef MATCH_MAY_ALLOCATE +# define INIT_FAIL_STACK() \ + do { \ + fail_stack.stack = (fail_stack_elt_t *) \ + REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \ + \ + if (fail_stack.stack == NULL) \ + return -2; \ + \ + fail_stack.size = INIT_FAILURE_ALLOC; \ + fail_stack.avail = 0; \ + } while (0) + +# define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack) +#else +# define INIT_FAIL_STACK() \ + do { \ + fail_stack.avail = 0; \ + } while (0) + +# define RESET_FAIL_STACK() +#endif + + +/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. + + Return 1 if succeeds, and 0 if either ran out of memory + allocating space for it or it was already too large. + + REGEX_REALLOCATE_STACK requires `destination' be declared. */ + +#define DOUBLE_FAIL_STACK(fail_stack) \ + ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \ + ? 0 \ + : ((fail_stack).stack = (fail_stack_elt_t *) \ + REGEX_REALLOCATE_STACK ((fail_stack).stack, \ + (fail_stack).size * sizeof (fail_stack_elt_t), \ + ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \ + \ + (fail_stack).stack == NULL \ + ? 0 \ + : ((fail_stack).size <<= 1, \ + 1))) + + +/* Push pointer POINTER on FAIL_STACK. + Return 1 if was able to do so and 0 if ran out of memory allocating + space to do so. */ +#define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \ + ((FAIL_STACK_FULL () \ + && !DOUBLE_FAIL_STACK (FAIL_STACK)) \ + ? 0 \ + : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \ + 1)) + +/* Push a pointer value onto the failure stack. + Assumes the variable `fail_stack'. Probably should only + be called from within `PUSH_FAILURE_POINT'. */ +#define PUSH_FAILURE_POINTER(item) \ + fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item) + +/* This pushes an integer-valued item onto the failure stack. + Assumes the variable `fail_stack'. Probably should only + be called from within `PUSH_FAILURE_POINT'. */ +#define PUSH_FAILURE_INT(item) \ + fail_stack.stack[fail_stack.avail++].integer = (item) + +/* Push a fail_stack_elt_t value onto the failure stack. + Assumes the variable `fail_stack'. Probably should only + be called from within `PUSH_FAILURE_POINT'. */ +#define PUSH_FAILURE_ELT(item) \ + fail_stack.stack[fail_stack.avail++] = (item) + +/* These three POP... operations complement the three PUSH... operations. + All assume that `fail_stack' is nonempty. */ +#define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer +#define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer +#define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail] + +/* Used to omit pushing failure point id's when we're not debugging. */ +#ifdef DEBUG +# define DEBUG_PUSH PUSH_FAILURE_INT +# define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT () +#else +# define DEBUG_PUSH(item) +# define DEBUG_POP(item_addr) +#endif + + +/* Push the information about the state we will need + if we ever fail back to it. + + Requires variables fail_stack, regstart, regend, reg_info, and + num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination' + be declared. + + Does `return FAILURE_CODE' if runs out of memory. */ + +#define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \ + do { \ + char *destination; \ + /* Must be int, so when we don't save any registers, the arithmetic \ + of 0 + -1 isn't done as unsigned. */ \ + /* Can't be int, since there is not a shred of a guarantee that int \ + is wide enough to hold a value of something to which pointer can \ + be assigned */ \ + active_reg_t this_reg; \ + \ + DEBUG_STATEMENT (failure_id++); \ + DEBUG_STATEMENT (nfailure_points_pushed++); \ + DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \ + DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\ + DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\ + \ + DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \ + DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \ + \ + /* Ensure we have enough space allocated for what we will push. */ \ + while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \ + { \ + if (!DOUBLE_FAIL_STACK (fail_stack)) \ + return failure_code; \ + \ + DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \ + (fail_stack).size); \ + DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\ + } \ + \ + /* Push the info, starting with the registers. */ \ + DEBUG_PRINT1 ("\n"); \ + \ + if (1) \ + for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \ + this_reg++) \ + { \ + DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \ + DEBUG_STATEMENT (num_regs_pushed++); \ + \ + DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \ + PUSH_FAILURE_POINTER (regstart[this_reg]); \ + \ + DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \ + PUSH_FAILURE_POINTER (regend[this_reg]); \ + \ + DEBUG_PRINT2 (" info: %p\n ", \ + reg_info[this_reg].word.pointer); \ + DEBUG_PRINT2 (" match_null=%d", \ + REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \ + DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \ + DEBUG_PRINT2 (" matched_something=%d", \ + MATCHED_SOMETHING (reg_info[this_reg])); \ + DEBUG_PRINT2 (" ever_matched=%d", \ + EVER_MATCHED_SOMETHING (reg_info[this_reg])); \ + DEBUG_PRINT1 ("\n"); \ + PUSH_FAILURE_ELT (reg_info[this_reg].word); \ + } \ + \ + DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\ + PUSH_FAILURE_INT (lowest_active_reg); \ + \ + DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\ + PUSH_FAILURE_INT (highest_active_reg); \ + \ + DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \ + DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \ + PUSH_FAILURE_POINTER (pattern_place); \ + \ + DEBUG_PRINT2 (" Pushing string %p: `", string_place); \ + DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \ + size2); \ + DEBUG_PRINT1 ("'\n"); \ + PUSH_FAILURE_POINTER (string_place); \ + \ + DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \ + DEBUG_PUSH (failure_id); \ + } while (0) + +/* This is the number of items that are pushed and popped on the stack + for each register. */ +#define NUM_REG_ITEMS 3 + +/* Individual items aside from the registers. */ +#ifdef DEBUG +# define NUM_NONREG_ITEMS 5 /* Includes failure point id. */ +#else +# define NUM_NONREG_ITEMS 4 +#endif + +/* We push at most this many items on the stack. */ +/* We used to use (num_regs - 1), which is the number of registers + this regexp will save; but that was changed to 5 + to avoid stack overflow for a regexp with lots of parens. */ +#define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS) + +/* We actually push this many items. */ +#define NUM_FAILURE_ITEMS \ + (((0 \ + ? 0 : highest_active_reg - lowest_active_reg + 1) \ + * NUM_REG_ITEMS) \ + + NUM_NONREG_ITEMS) + +/* How many items can still be added to the stack without overflowing it. */ +#define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail) + + +/* Pops what PUSH_FAIL_STACK pushes. + + We restore into the parameters, all of which should be lvalues: + STR -- the saved data position. + PAT -- the saved pattern position. + LOW_REG, HIGH_REG -- the highest and lowest active registers. + REGSTART, REGEND -- arrays of string positions. + REG_INFO -- array of information about each subexpression. + + Also assumes the variables `fail_stack' and (if debugging), `bufp', + `pend', `string1', `size1', `string2', and `size2'. */ + +#define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\ +{ \ + DEBUG_STATEMENT (unsigned failure_id;) \ + active_reg_t this_reg; \ + const unsigned char *string_temp; \ + \ + assert (!FAIL_STACK_EMPTY ()); \ + \ + /* Remove failure points and point to how many regs pushed. */ \ + DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \ + DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \ + DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \ + \ + assert (fail_stack.avail >= NUM_NONREG_ITEMS); \ + \ + DEBUG_POP (&failure_id); \ + DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \ + \ + /* If the saved string location is NULL, it came from an \ + on_failure_keep_string_jump opcode, and we want to throw away the \ + saved NULL, thus retaining our current position in the string. */ \ + string_temp = POP_FAILURE_POINTER (); \ + if (string_temp != NULL) \ + str = (const char *) string_temp; \ + \ + DEBUG_PRINT2 (" Popping string %p: `", str); \ + DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \ + DEBUG_PRINT1 ("'\n"); \ + \ + pat = (unsigned char *) POP_FAILURE_POINTER (); \ + DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \ + DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \ + \ + /* Restore register info. */ \ + high_reg = (active_reg_t) POP_FAILURE_INT (); \ + DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \ + \ + low_reg = (active_reg_t) POP_FAILURE_INT (); \ + DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \ + \ + if (1) \ + for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \ + { \ + DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \ + \ + reg_info[this_reg].word = POP_FAILURE_ELT (); \ + DEBUG_PRINT2 (" info: %p\n", \ + reg_info[this_reg].word.pointer); \ + \ + regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \ + DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \ + \ + regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \ + DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \ + } \ + else \ + { \ + for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \ + { \ + reg_info[this_reg].word.integer = 0; \ + regend[this_reg] = 0; \ + regstart[this_reg] = 0; \ + } \ + highest_active_reg = high_reg; \ + } \ + \ + set_regs_matched_done = 0; \ + DEBUG_STATEMENT (nfailure_points_popped++); \ +} /* POP_FAILURE_POINT */ + + + +/* Structure for per-register (a.k.a. per-group) information. + Other register information, such as the + starting and ending positions (which are addresses), and the list of + inner groups (which is a bits list) are maintained in separate + variables. + + We are making a (strictly speaking) nonportable assumption here: that + the compiler will pack our bit fields into something that fits into + the type of `word', i.e., is something that fits into one item on the + failure stack. */ + + +/* Declarations and macros for re_match_2. */ + +typedef union +{ + fail_stack_elt_t word; + struct + { + /* This field is one if this group can match the empty string, + zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */ +#define MATCH_NULL_UNSET_VALUE 3 + unsigned match_null_string_p : 2; + unsigned is_active : 1; + unsigned matched_something : 1; + unsigned ever_matched_something : 1; + } bits; +} register_info_type; + +#define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p) +#define IS_ACTIVE(R) ((R).bits.is_active) +#define MATCHED_SOMETHING(R) ((R).bits.matched_something) +#define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something) + + +/* Call this when have matched a real character; it sets `matched' flags + for the subexpressions which we are currently inside. Also records + that those subexprs have matched. */ +#define SET_REGS_MATCHED() \ + do \ + { \ + if (!set_regs_matched_done) \ + { \ + active_reg_t r; \ + set_regs_matched_done = 1; \ + for (r = lowest_active_reg; r <= highest_active_reg; r++) \ + { \ + MATCHED_SOMETHING (reg_info[r]) \ + = EVER_MATCHED_SOMETHING (reg_info[r]) \ + = 1; \ + } \ + } \ + } \ + while (0) + +/* Registers are set to a sentinel when they haven't yet matched. */ +static char reg_unset_dummy; +#define REG_UNSET_VALUE (®_unset_dummy) +#define REG_UNSET(e) ((e) == REG_UNSET_VALUE) + +/* Subroutine declarations and macros for regex_compile. */ + +static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size, + reg_syntax_t syntax, + struct re_pattern_buffer *bufp)); +static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg)); +static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc, + int arg1, int arg2)); +static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, + int arg, unsigned char *end)); +static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc, + int arg1, int arg2, unsigned char *end)); +static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p, + reg_syntax_t syntax)); +static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend, + reg_syntax_t syntax)); +static reg_errcode_t compile_range _RE_ARGS ((const char **p_ptr, + const char *pend, + char *translate, + reg_syntax_t syntax, + unsigned char *b)); + +/* Fetch the next character in the uncompiled pattern---translating it + if necessary. Also cast from a signed character in the constant + string passed to us by the user to an unsigned char that we can use + as an array index (in, e.g., `translate'). */ +#ifndef PATFETCH +# define PATFETCH(c) \ + do {if (p == pend) return REG_EEND; \ + c = (unsigned char) *p++; \ + if (translate) c = (unsigned char) translate[c]; \ + } while (0) +#endif + +/* Fetch the next character in the uncompiled pattern, with no + translation. */ +#define PATFETCH_RAW(c) \ + do {if (p == pend) return REG_EEND; \ + c = (unsigned char) *p++; \ + } while (0) + +/* Go backwards one character in the pattern. */ +#define PATUNFETCH p-- + + +/* If `translate' is non-null, return translate[D], else just D. We + cast the subscript to translate because some data is declared as + `char *', to avoid warnings when a string constant is passed. But + when we use a character as a subscript we must make it unsigned. */ +#ifndef TRANSLATE +# define TRANSLATE(d) \ + (translate ? (char) translate[(unsigned char) (d)] : (d)) +#endif + + +/* Macros for outputting the compiled pattern into `buffer'. */ + +/* If the buffer isn't allocated when it comes in, use this. */ +#define INIT_BUF_SIZE 32 + +/* Make sure we have at least N more bytes of space in buffer. */ +#define GET_BUFFER_SPACE(n) \ + while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \ + EXTEND_BUFFER () + +/* Make sure we have one more byte of buffer space and then add C to it. */ +#define BUF_PUSH(c) \ + do { \ + GET_BUFFER_SPACE (1); \ + *b++ = (unsigned char) (c); \ + } while (0) + + +/* Ensure we have two more bytes of buffer space and then append C1 and C2. */ +#define BUF_PUSH_2(c1, c2) \ + do { \ + GET_BUFFER_SPACE (2); \ + *b++ = (unsigned char) (c1); \ + *b++ = (unsigned char) (c2); \ + } while (0) + + +/* As with BUF_PUSH_2, except for three bytes. */ +#define BUF_PUSH_3(c1, c2, c3) \ + do { \ + GET_BUFFER_SPACE (3); \ + *b++ = (unsigned char) (c1); \ + *b++ = (unsigned char) (c2); \ + *b++ = (unsigned char) (c3); \ + } while (0) + + +/* Store a jump with opcode OP at LOC to location TO. We store a + relative address offset by the three bytes the jump itself occupies. */ +#define STORE_JUMP(op, loc, to) \ + store_op1 (op, loc, (int) ((to) - (loc) - 3)) + +/* Likewise, for a two-argument jump. */ +#define STORE_JUMP2(op, loc, to, arg) \ + store_op2 (op, loc, (int) ((to) - (loc) - 3), arg) + +/* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */ +#define INSERT_JUMP(op, loc, to) \ + insert_op1 (op, loc, (int) ((to) - (loc) - 3), b) + +/* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */ +#define INSERT_JUMP2(op, loc, to, arg) \ + insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b) + + +/* This is not an arbitrary limit: the arguments which represent offsets + into the pattern are two bytes long. So if 2^16 bytes turns out to + be too small, many things would have to change. */ +/* Any other compiler which, like MSC, has allocation limit below 2^16 + bytes will have to use approach similar to what was done below for + MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up + reallocating to 0 bytes. Such thing is not going to work too well. + You have been warned!! */ +#if defined _MSC_VER && !defined WIN32 +/* Microsoft C 16-bit versions limit malloc to approx 65512 bytes. + The REALLOC define eliminates a flurry of conversion warnings, + but is not required. */ +# define MAX_BUF_SIZE 65500L +# define REALLOC(p,s) realloc ((p), (size_t) (s)) +#else +# define MAX_BUF_SIZE (1L << 16) +# define REALLOC(p,s) realloc ((p), (s)) +#endif + +/* Extend the buffer by twice its current size via realloc and + reset the pointers that pointed into the old block to point to the + correct places in the new one. If extending the buffer results in it + being larger than MAX_BUF_SIZE, then flag memory exhausted. */ +#define EXTEND_BUFFER() \ + do { \ + unsigned char *old_buffer = bufp->buffer; \ + if (bufp->allocated == MAX_BUF_SIZE) \ + return REG_ESIZE; \ + bufp->allocated <<= 1; \ + if (bufp->allocated > MAX_BUF_SIZE) \ + bufp->allocated = MAX_BUF_SIZE; \ + bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\ + if (bufp->buffer == NULL) \ + return REG_ESPACE; \ + /* If the buffer moved, move all the pointers into it. */ \ + if (old_buffer != bufp->buffer) \ + { \ + b = (b - old_buffer) + bufp->buffer; \ + begalt = (begalt - old_buffer) + bufp->buffer; \ + if (fixup_alt_jump) \ + fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\ + if (laststart) \ + laststart = (laststart - old_buffer) + bufp->buffer; \ + if (pending_exact) \ + pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ + } \ + } while (0) + + +/* Since we have one byte reserved for the register number argument to + {start,stop}_memory, the maximum number of groups we can report + things about is what fits in that byte. */ +#define MAX_REGNUM 255 + +/* But patterns can have more than `MAX_REGNUM' registers. We just + ignore the excess. */ +typedef unsigned regnum_t; + + +/* Macros for the compile stack. */ + +/* Since offsets can go either forwards or backwards, this type needs to + be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ +/* int may be not enough when sizeof(int) == 2. */ +typedef long pattern_offset_t; + +typedef struct +{ + pattern_offset_t begalt_offset; + pattern_offset_t fixup_alt_jump; + pattern_offset_t inner_group_offset; + pattern_offset_t laststart_offset; + regnum_t regnum; +} compile_stack_elt_t; + + +typedef struct +{ + compile_stack_elt_t *stack; + unsigned size; + unsigned avail; /* Offset of next open position. */ +} compile_stack_type; + + +#define INIT_COMPILE_STACK_SIZE 32 + +#define COMPILE_STACK_EMPTY (compile_stack.avail == 0) +#define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) + +/* The next available element. */ +#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) + + +/* Set the bit for character C in a list. */ +#define SET_LIST_BIT(c) \ + (b[((unsigned char) (c)) / BYTEWIDTH] \ + |= 1 << (((unsigned char) c) % BYTEWIDTH)) + + +/* Get the next unsigned number in the uncompiled pattern. */ +#define GET_UNSIGNED_NUMBER(num) \ + { if (p != pend) \ + { \ + PATFETCH (c); \ + while (ISDIGIT (c)) \ + { \ + if (num < 0) \ + num = 0; \ + num = num * 10 + c - '0'; \ + if (p == pend) \ + break; \ + PATFETCH (c); \ + } \ + } \ + } + +#if defined _LIBC || WIDE_CHAR_SUPPORT +/* The GNU C library provides support for user-defined character classes + and the functions from ISO C amendement 1. */ +# ifdef CHARCLASS_NAME_MAX +# define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX +# else +/* This shouldn't happen but some implementation might still have this + problem. Use a reasonable default value. */ +# define CHAR_CLASS_MAX_LENGTH 256 +# endif + +# ifdef _LIBC +# define IS_CHAR_CLASS(string) __wctype (string) +# else +# define IS_CHAR_CLASS(string) wctype (string) +# endif +#else +# define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ + +# define IS_CHAR_CLASS(string) \ + (STREQ (string, "alpha") || STREQ (string, "upper") \ + || STREQ (string, "lower") || STREQ (string, "digit") \ + || STREQ (string, "alnum") || STREQ (string, "xdigit") \ + || STREQ (string, "space") || STREQ (string, "print") \ + || STREQ (string, "punct") || STREQ (string, "graph") \ + || STREQ (string, "cntrl") || STREQ (string, "blank")) +#endif + +#ifndef MATCH_MAY_ALLOCATE + +/* If we cannot allocate large objects within re_match_2_internal, + we make the fail stack and register vectors global. + The fail stack, we grow to the maximum size when a regexp + is compiled. + The register vectors, we adjust in size each time we + compile a regexp, according to the number of registers it needs. */ + +static fail_stack_type fail_stack; + +/* Size with which the following vectors are currently allocated. + That is so we can make them bigger as needed, + but never make them smaller. */ +static int regs_allocated_size; + +static const char ** regstart, ** regend; +static const char ** old_regstart, ** old_regend; +static const char **best_regstart, **best_regend; +static register_info_type *reg_info; +static const char **reg_dummy; +static register_info_type *reg_info_dummy; + +/* Make the register vectors big enough for NUM_REGS registers, + but don't make them smaller. */ + +static +regex_grow_registers (num_regs) + int num_regs; +{ + if (num_regs > regs_allocated_size) + { + RETALLOC_IF (regstart, num_regs, const char *); + RETALLOC_IF (regend, num_regs, const char *); + RETALLOC_IF (old_regstart, num_regs, const char *); + RETALLOC_IF (old_regend, num_regs, const char *); + RETALLOC_IF (best_regstart, num_regs, const char *); + RETALLOC_IF (best_regend, num_regs, const char *); + RETALLOC_IF (reg_info, num_regs, register_info_type); + RETALLOC_IF (reg_dummy, num_regs, const char *); + RETALLOC_IF (reg_info_dummy, num_regs, register_info_type); + + regs_allocated_size = num_regs; + } +} + +#endif /* not MATCH_MAY_ALLOCATE */ + +static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type + compile_stack, + regnum_t regnum)); + +/* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX. + Returns one of error codes defined in `regex.h', or zero for success. + + Assumes the `allocated' (and perhaps `buffer') and `translate' + fields are set in BUFP on entry. + + If it succeeds, results are put in BUFP (if it returns an error, the + contents of BUFP are undefined): + `buffer' is the compiled pattern; + `syntax' is set to SYNTAX; + `used' is set to the length of the compiled pattern; + `fastmap_accurate' is zero; + `re_nsub' is the number of subexpressions in PATTERN; + `not_bol' and `not_eol' are zero; + + The `fastmap' and `newline_anchor' fields are neither + examined nor set. */ + +/* Return, freeing storage we allocated. */ +#define FREE_STACK_RETURN(value) \ + return (free (compile_stack.stack), value) + +static reg_errcode_t +regex_compile (pattern, size, syntax, bufp) + const char *pattern; + size_t size; + reg_syntax_t syntax; + struct re_pattern_buffer *bufp; +{ + /* We fetch characters from PATTERN here. Even though PATTERN is + `char *' (i.e., signed), we declare these variables as unsigned, so + they can be reliably used as array indices. */ + register unsigned char c, c1; + + /* A random temporary spot in PATTERN. */ + const char *p1; + + /* Points to the end of the buffer, where we should append. */ + register unsigned char *b; + + /* Keeps track of unclosed groups. */ + compile_stack_type compile_stack; + + /* Points to the current (ending) position in the pattern. */ + const char *p = pattern; + const char *pend = pattern + size; + + /* How to translate the characters in the pattern. */ + RE_TRANSLATE_TYPE translate = bufp->translate; + + /* Address of the count-byte of the most recently inserted `exactn' + command. This makes it possible to tell if a new exact-match + character can be added to that command or if the character requires + a new `exactn' command. */ + unsigned char *pending_exact = 0; + + /* Address of start of the most recently finished expression. + This tells, e.g., postfix * where to find the start of its + operand. Reset at the beginning of groups and alternatives. */ + unsigned char *laststart = 0; + + /* Address of beginning of regexp, or inside of last group. */ + unsigned char *begalt; + + /* Place in the uncompiled pattern (i.e., the {) to + which to go back if the interval is invalid. */ + const char *beg_interval; + + /* Address of the place where a forward jump should go to the end of + the containing expression. Each alternative of an `or' -- except the + last -- ends with a forward jump of this sort. */ + unsigned char *fixup_alt_jump = 0; + + /* Counts open-groups as they are encountered. Remembered for the + matching close-group on the compile stack, so the same register + number is put in the stop_memory as the start_memory. */ + regnum_t regnum = 0; + +#ifdef DEBUG + DEBUG_PRINT1 ("\nCompiling pattern: "); + if (debug) + { + unsigned debug_count; + + for (debug_count = 0; debug_count < size; debug_count++) + putchar (pattern[debug_count]); + putchar ('\n'); + } +#endif /* DEBUG */ + + /* Initialize the compile stack. */ + compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t); + if (compile_stack.stack == NULL) + return REG_ESPACE; + + compile_stack.size = INIT_COMPILE_STACK_SIZE; + compile_stack.avail = 0; + + /* Initialize the pattern buffer. */ + bufp->syntax = syntax; + bufp->fastmap_accurate = 0; + bufp->not_bol = bufp->not_eol = 0; + + /* Set `used' to zero, so that if we return an error, the pattern + printer (for debugging) will think there's no pattern. We reset it + at the end. */ + bufp->used = 0; + + /* Always count groups, whether or not bufp->no_sub is set. */ + bufp->re_nsub = 0; + +#if !defined emacs && !defined SYNTAX_TABLE + /* Initialize the syntax table. */ + init_syntax_once (); +#endif + + if (bufp->allocated == 0) + { + if (bufp->buffer) + { /* If zero allocated, but buffer is non-null, try to realloc + enough space. This loses if buffer's address is bogus, but + that is the user's responsibility. */ + RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char); + } + else + { /* Caller did not allocate a buffer. Do it for them. */ + bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char); + } + if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE); + + bufp->allocated = INIT_BUF_SIZE; + } + + begalt = b = bufp->buffer; + + /* Loop through the uncompiled pattern until we're at the end. */ + while (p != pend) + { + PATFETCH (c); + + switch (c) + { + case '^': + { + if ( /* If at start of pattern, it's an operator. */ + p == pattern + 1 + /* If context independent, it's an operator. */ + || syntax & RE_CONTEXT_INDEP_ANCHORS + /* Otherwise, depends on what's come before. */ + || at_begline_loc_p (pattern, p, syntax)) + BUF_PUSH (begline); + else + goto normal_char; + } + break; + + + case '$': + { + if ( /* If at end of pattern, it's an operator. */ + p == pend + /* If context independent, it's an operator. */ + || syntax & RE_CONTEXT_INDEP_ANCHORS + /* Otherwise, depends on what's next. */ + || at_endline_loc_p (p, pend, syntax)) + BUF_PUSH (endline); + else + goto normal_char; + } + break; + + + case '+': + case '?': + if ((syntax & RE_BK_PLUS_QM) + || (syntax & RE_LIMITED_OPS)) + goto normal_char; + handle_plus: + case '*': + /* If there is no previous pattern... */ + if (!laststart) + { + if (syntax & RE_CONTEXT_INVALID_OPS) + FREE_STACK_RETURN (REG_BADRPT); + else if (!(syntax & RE_CONTEXT_INDEP_OPS)) + goto normal_char; + } + + { + /* Are we optimizing this jump? */ + boolean keep_string_p = false; + + /* 1 means zero (many) matches is allowed. */ + char zero_times_ok = 0, many_times_ok = 0; + + /* If there is a sequence of repetition chars, collapse it + down to just one (the right one). We can't combine + interval operators with these because of, e.g., `a{2}*', + which should only match an even number of `a's. */ + + for (;;) + { + zero_times_ok |= c != '+'; + many_times_ok |= c != '?'; + + if (p == pend) + break; + + PATFETCH (c); + + if (c == '*' + || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?'))) + ; + + else if (syntax & RE_BK_PLUS_QM && c == '\\') + { + if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); + + PATFETCH (c1); + if (!(c1 == '+' || c1 == '?')) + { + PATUNFETCH; + PATUNFETCH; + break; + } + + c = c1; + } + else + { + PATUNFETCH; + break; + } + + /* If we get here, we found another repeat character. */ + } + + /* Star, etc. applied to an empty pattern is equivalent + to an empty pattern. */ + if (!laststart) + break; + + /* Now we know whether or not zero matches is allowed + and also whether or not two or more matches is allowed. */ + if (many_times_ok) + { /* More than one repetition is allowed, so put in at the + end a backward relative jump from `b' to before the next + jump we're going to put in below (which jumps from + laststart to after this jump). + + But if we are at the `*' in the exact sequence `.*\n', + insert an unconditional jump backwards to the ., + instead of the beginning of the loop. This way we only + push a failure point once, instead of every time + through the loop. */ + assert (p - 1 > pattern); + + /* Allocate the space for the jump. */ + GET_BUFFER_SPACE (3); + + /* We know we are not at the first character of the pattern, + because laststart was nonzero. And we've already + incremented `p', by the way, to be the character after + the `*'. Do we have to do something analogous here + for null bytes, because of RE_DOT_NOT_NULL? */ + if (TRANSLATE (*(p - 2)) == TRANSLATE ('.') + && zero_times_ok + && p < pend && TRANSLATE (*p) == TRANSLATE ('\n') + && !(syntax & RE_DOT_NEWLINE)) + { /* We have .*\n. */ + STORE_JUMP (jump, b, laststart); + keep_string_p = true; + } + else + /* Anything else. */ + STORE_JUMP (maybe_pop_jump, b, laststart - 3); + + /* We've added more stuff to the buffer. */ + b += 3; + } + + /* On failure, jump from laststart to b + 3, which will be the + end of the buffer after this jump is inserted. */ + GET_BUFFER_SPACE (3); + INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump + : on_failure_jump, + laststart, b + 3); + pending_exact = 0; + b += 3; + + if (!zero_times_ok) + { + /* At least one repetition is required, so insert a + `dummy_failure_jump' before the initial + `on_failure_jump' instruction of the loop. This + effects a skip over that instruction the first time + we hit that loop. */ + GET_BUFFER_SPACE (3); + INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6); + b += 3; + } + } + break; + + + case '.': + laststart = b; + BUF_PUSH (anychar); + break; + + + case '[': + { + boolean had_char_class = false; + + if (p == pend) FREE_STACK_RETURN (REG_EBRACK); + + /* Ensure that we have enough space to push a charset: the + opcode, the length count, and the bitset; 34 bytes in all. */ + GET_BUFFER_SPACE (34); + + laststart = b; + + /* We test `*p == '^' twice, instead of using an if + statement, so we only need one BUF_PUSH. */ + BUF_PUSH (*p == '^' ? charset_not : charset); + if (*p == '^') + p++; + + /* Remember the first position in the bracket expression. */ + p1 = p; + + /* Push the number of bytes in the bitmap. */ + BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH); + + /* Clear the whole map. */ + bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH); + + /* charset_not matches newline according to a syntax bit. */ + if ((re_opcode_t) b[-2] == charset_not + && (syntax & RE_HAT_LISTS_NOT_NEWLINE)) + SET_LIST_BIT ('\n'); + + /* Read in characters and ranges, setting map bits. */ + for (;;) + { + if (p == pend) FREE_STACK_RETURN (REG_EBRACK); + + PATFETCH (c); + + /* \ might escape characters inside [...] and [^...]. */ + if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') + { + if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); + + PATFETCH (c1); + SET_LIST_BIT (c1); + continue; + } + + /* Could be the end of the bracket expression. If it's + not (i.e., when the bracket expression is `[]' so + far), the ']' character bit gets set way below. */ + if (c == ']' && p != p1 + 1) + break; + + /* Look ahead to see if it's a range when the last thing + was a character class. */ + if (had_char_class && c == '-' && *p != ']') + FREE_STACK_RETURN (REG_ERANGE); + + /* Look ahead to see if it's a range when the last thing + was a character: if this is a hyphen not at the + beginning or the end of a list, then it's the range + operator. */ + if (c == '-' + && !(p - 2 >= pattern && p[-2] == '[') + && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') + && *p != ']') + { + reg_errcode_t ret + = compile_range (&p, pend, translate, syntax, b); + if (ret != REG_NOERROR) FREE_STACK_RETURN (ret); + } + + else if (p[0] == '-' && p[1] != ']') + { /* This handles ranges made up of characters only. */ + reg_errcode_t ret; + + /* Move past the `-'. */ + PATFETCH (c1); + + ret = compile_range (&p, pend, translate, syntax, b); + if (ret != REG_NOERROR) FREE_STACK_RETURN (ret); + } + + /* See if we're at the beginning of a possible character + class. */ + + else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':') + { /* Leave room for the null. */ + char str[CHAR_CLASS_MAX_LENGTH + 1]; + + PATFETCH (c); + c1 = 0; + + /* If pattern is `[[:'. */ + if (p == pend) FREE_STACK_RETURN (REG_EBRACK); + + for (;;) + { + PATFETCH (c); + if ((c == ':' && *p == ']') || p == pend) + break; + if (c1 < CHAR_CLASS_MAX_LENGTH) + str[c1++] = c; + else + /* This is in any case an invalid class name. */ + str[0] = '\0'; + } + str[c1] = '\0'; + + /* If isn't a word bracketed by `[:' and `:]': + undo the ending character, the letters, and leave + the leading `:' and `[' (but set bits for them). */ + if (c == ':' && *p == ']') + { +#if defined _LIBC || WIDE_CHAR_SUPPORT + boolean is_lower = STREQ (str, "lower"); + boolean is_upper = STREQ (str, "upper"); + wctype_t wt; + int ch; + + wt = IS_CHAR_CLASS (str); + if (wt == 0) + FREE_STACK_RETURN (REG_ECTYPE); + + /* Throw away the ] at the end of the character + class. */ + PATFETCH (c); + + if (p == pend) FREE_STACK_RETURN (REG_EBRACK); + + for (ch = 0; ch < 1 << BYTEWIDTH; ++ch) + { +# ifdef _LIBC + if (__iswctype (__btowc (ch), wt)) + SET_LIST_BIT (ch); +# else + if (iswctype (btowc (ch), wt)) + SET_LIST_BIT (ch); +# endif + + if (translate && (is_upper || is_lower) + && (ISUPPER (ch) || ISLOWER (ch))) + SET_LIST_BIT (ch); + } + + had_char_class = true; +#else + int ch; + boolean is_alnum = STREQ (str, "alnum"); + boolean is_alpha = STREQ (str, "alpha"); + boolean is_blank = STREQ (str, "blank"); + boolean is_cntrl = STREQ (str, "cntrl"); + boolean is_digit = STREQ (str, "digit"); + boolean is_graph = STREQ (str, "graph"); + boolean is_lower = STREQ (str, "lower"); + boolean is_print = STREQ (str, "print"); + boolean is_punct = STREQ (str, "punct"); + boolean is_space = STREQ (str, "space"); + boolean is_upper = STREQ (str, "upper"); + boolean is_xdigit = STREQ (str, "xdigit"); + + if (!IS_CHAR_CLASS (str)) + FREE_STACK_RETURN (REG_ECTYPE); + + /* Throw away the ] at the end of the character + class. */ + PATFETCH (c); + + if (p == pend) FREE_STACK_RETURN (REG_EBRACK); + + for (ch = 0; ch < 1 << BYTEWIDTH; ch++) + { + /* This was split into 3 if's to + avoid an arbitrary limit in some compiler. */ + if ( (is_alnum && ISALNUM (ch)) + || (is_alpha && ISALPHA (ch)) + || (is_blank && ISBLANK (ch)) + || (is_cntrl && ISCNTRL (ch))) + SET_LIST_BIT (ch); + if ( (is_digit && ISDIGIT (ch)) + || (is_graph && ISGRAPH (ch)) + || (is_lower && ISLOWER (ch)) + || (is_print && ISPRINT (ch))) + SET_LIST_BIT (ch); + if ( (is_punct && ISPUNCT (ch)) + || (is_space && ISSPACE (ch)) + || (is_upper && ISUPPER (ch)) + || (is_xdigit && ISXDIGIT (ch))) + SET_LIST_BIT (ch); + if ( translate && (is_upper || is_lower) + && (ISUPPER (ch) || ISLOWER (ch))) + SET_LIST_BIT (ch); + } + had_char_class = true; +#endif /* libc || wctype.h */ + } + else + { + c1++; + while (c1--) + PATUNFETCH; + SET_LIST_BIT ('['); + SET_LIST_BIT (':'); + had_char_class = false; + } + } + else + { + had_char_class = false; + SET_LIST_BIT (c); + } + } + + /* Discard any (non)matching list bytes that are all 0 at the + end of the map. Decrease the map-length byte too. */ + while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) + b[-1]--; + b += b[-1]; + } + break; + + + case '(': + if (syntax & RE_NO_BK_PARENS) + goto handle_open; + else + goto normal_char; + + + case ')': + if (syntax & RE_NO_BK_PARENS) + goto handle_close; + else + goto normal_char; + + + case '\n': + if (syntax & RE_NEWLINE_ALT) + goto handle_alt; + else + goto normal_char; + + + case '|': + if (syntax & RE_NO_BK_VBAR) + goto handle_alt; + else + goto normal_char; + + + case '{': + if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES) + goto handle_interval; + else + goto normal_char; + + + case '\\': + if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); + + /* Do not translate the character after the \, so that we can + distinguish, e.g., \B from \b, even if we normally would + translate, e.g., B to b. */ + PATFETCH_RAW (c); + + switch (c) + { + case '(': + if (syntax & RE_NO_BK_PARENS) + goto normal_backslash; + + handle_open: + bufp->re_nsub++; + regnum++; + + if (COMPILE_STACK_FULL) + { + RETALLOC (compile_stack.stack, compile_stack.size << 1, + compile_stack_elt_t); + if (compile_stack.stack == NULL) return REG_ESPACE; + + compile_stack.size <<= 1; + } + + /* These are the values to restore when we hit end of this + group. They are all relative offsets, so that if the + whole pattern moves because of realloc, they will still + be valid. */ + COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; + COMPILE_STACK_TOP.fixup_alt_jump + = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; + COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer; + COMPILE_STACK_TOP.regnum = regnum; + + /* We will eventually replace the 0 with the number of + groups inner to this one. But do not push a + start_memory for groups beyond the last one we can + represent in the compiled pattern. */ + if (regnum <= MAX_REGNUM) + { + COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2; + BUF_PUSH_3 (start_memory, regnum, 0); + } + + compile_stack.avail++; + + fixup_alt_jump = 0; + laststart = 0; + begalt = b; + /* If we've reached MAX_REGNUM groups, then this open + won't actually generate any code, so we'll have to + clear pending_exact explicitly. */ + pending_exact = 0; + break; + + + case ')': + if (syntax & RE_NO_BK_PARENS) goto normal_backslash; + + if (COMPILE_STACK_EMPTY) + { + if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) + goto normal_backslash; + else + FREE_STACK_RETURN (REG_ERPAREN); + } + + handle_close: + if (fixup_alt_jump) + { /* Push a dummy failure point at the end of the + alternative for a possible future + `pop_failure_jump' to pop. See comments at + `push_dummy_failure' in `re_match_2'. */ + BUF_PUSH (push_dummy_failure); + + /* We allocated space for this jump when we assigned + to `fixup_alt_jump', in the `handle_alt' case below. */ + STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1); + } + + /* See similar code for backslashed left paren above. */ + if (COMPILE_STACK_EMPTY) + { + if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) + goto normal_char; + else + FREE_STACK_RETURN (REG_ERPAREN); + } + + /* Since we just checked for an empty stack above, this + ``can't happen''. */ + assert (compile_stack.avail != 0); + { + /* We don't just want to restore into `regnum', because + later groups should continue to be numbered higher, + as in `(ab)c(de)' -- the second group is #2. */ + regnum_t this_group_regnum; + + compile_stack.avail--; + begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; + fixup_alt_jump + = COMPILE_STACK_TOP.fixup_alt_jump + ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 + : 0; + laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; + this_group_regnum = COMPILE_STACK_TOP.regnum; + /* If we've reached MAX_REGNUM groups, then this open + won't actually generate any code, so we'll have to + clear pending_exact explicitly. */ + pending_exact = 0; + + /* We're at the end of the group, so now we know how many + groups were inside this one. */ + if (this_group_regnum <= MAX_REGNUM) + { + unsigned char *inner_group_loc + = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; + + *inner_group_loc = regnum - this_group_regnum; + BUF_PUSH_3 (stop_memory, this_group_regnum, + regnum - this_group_regnum); + } + } + break; + + + case '|': /* `\|'. */ + if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR) + goto normal_backslash; + handle_alt: + if (syntax & RE_LIMITED_OPS) + goto normal_char; + + /* Insert before the previous alternative a jump which + jumps to this alternative if the former fails. */ + GET_BUFFER_SPACE (3); + INSERT_JUMP (on_failure_jump, begalt, b + 6); + pending_exact = 0; + b += 3; + + /* The alternative before this one has a jump after it + which gets executed if it gets matched. Adjust that + jump so it will jump to this alternative's analogous + jump (put in below, which in turn will jump to the next + (if any) alternative's such jump, etc.). The last such + jump jumps to the correct final destination. A picture: + _____ _____ + | | | | + | v | v + a | b | c + + If we are at `b', then fixup_alt_jump right now points to a + three-byte space after `a'. We'll put in the jump, set + fixup_alt_jump to right after `b', and leave behind three + bytes which we'll fill in when we get to after `c'. */ + + if (fixup_alt_jump) + STORE_JUMP (jump_past_alt, fixup_alt_jump, b); + + /* Mark and leave space for a jump after this alternative, + to be filled in later either by next alternative or + when know we're at the end of a series of alternatives. */ + fixup_alt_jump = b; + GET_BUFFER_SPACE (3); + b += 3; + + laststart = 0; + begalt = b; + break; + + + case '{': + /* If \{ is a literal. */ + if (!(syntax & RE_INTERVALS) + /* If we're at `\{' and it's not the open-interval + operator. */ + || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) + || (p - 2 == pattern && p == pend)) + goto normal_backslash; + + handle_interval: + { + /* If got here, then the syntax allows intervals. */ + + /* At least (most) this many matches must be made. */ + int lower_bound = -1, upper_bound = -1; + + beg_interval = p - 1; + + if (p == pend) + { + if (syntax & RE_NO_BK_BRACES) + goto unfetch_interval; + else + FREE_STACK_RETURN (REG_EBRACE); + } + + GET_UNSIGNED_NUMBER (lower_bound); + + if (c == ',') + { + GET_UNSIGNED_NUMBER (upper_bound); + if (upper_bound < 0) upper_bound = RE_DUP_MAX; + } + else + /* Interval such as `{1}' => match exactly once. */ + upper_bound = lower_bound; + + if (lower_bound < 0 || upper_bound > RE_DUP_MAX + || lower_bound > upper_bound) + { + if (syntax & RE_NO_BK_BRACES) + goto unfetch_interval; + else + FREE_STACK_RETURN (REG_BADBR); + } + + if (!(syntax & RE_NO_BK_BRACES)) + { + if (c != '\\') FREE_STACK_RETURN (REG_EBRACE); + + PATFETCH (c); + } + + if (c != '}') + { + if (syntax & RE_NO_BK_BRACES) + goto unfetch_interval; + else + FREE_STACK_RETURN (REG_BADBR); + } + + /* We just parsed a valid interval. */ + + /* If it's invalid to have no preceding re. */ + if (!laststart) + { + if (syntax & RE_CONTEXT_INVALID_OPS) + FREE_STACK_RETURN (REG_BADRPT); + else if (syntax & RE_CONTEXT_INDEP_OPS) + laststart = b; + else + goto unfetch_interval; + } + + /* If the upper bound is zero, don't want to succeed at + all; jump from `laststart' to `b + 3', which will be + the end of the buffer after we insert the jump. */ + if (upper_bound == 0) + { + GET_BUFFER_SPACE (3); + INSERT_JUMP (jump, laststart, b + 3); + b += 3; + } + + /* Otherwise, we have a nontrivial interval. When + we're all done, the pattern will look like: + set_number_at <jump count> <upper bound> + set_number_at <succeed_n count> <lower bound> + succeed_n <after jump addr> <succeed_n count> + <body of loop> + jump_n <succeed_n addr> <jump count> + (The upper bound and `jump_n' are omitted if + `upper_bound' is 1, though.) */ + else + { /* If the upper bound is > 1, we need to insert + more at the end of the loop. */ + unsigned nbytes = 10 + (upper_bound > 1) * 10; + + GET_BUFFER_SPACE (nbytes); + + /* Initialize lower bound of the `succeed_n', even + though it will be set during matching by its + attendant `set_number_at' (inserted next), + because `re_compile_fastmap' needs to know. + Jump to the `jump_n' we might insert below. */ + INSERT_JUMP2 (succeed_n, laststart, + b + 5 + (upper_bound > 1) * 5, + lower_bound); + b += 5; + + /* Code to initialize the lower bound. Insert + before the `succeed_n'. The `5' is the last two + bytes of this `set_number_at', plus 3 bytes of + the following `succeed_n'. */ + insert_op2 (set_number_at, laststart, 5, lower_bound, b); + b += 5; + + if (upper_bound > 1) + { /* More than one repetition is allowed, so + append a backward jump to the `succeed_n' + that starts this interval. + + When we've reached this during matching, + we'll have matched the interval once, so + jump back only `upper_bound - 1' times. */ + STORE_JUMP2 (jump_n, b, laststart + 5, + upper_bound - 1); + b += 5; + + /* The location we want to set is the second + parameter of the `jump_n'; that is `b-2' as + an absolute address. `laststart' will be + the `set_number_at' we're about to insert; + `laststart+3' the number to set, the source + for the relative address. But we are + inserting into the middle of the pattern -- + so everything is getting moved up by 5. + Conclusion: (b - 2) - (laststart + 3) + 5, + i.e., b - laststart. + + We insert this at the beginning of the loop + so that if we fail during matching, we'll + reinitialize the bounds. */ + insert_op2 (set_number_at, laststart, b - laststart, + upper_bound - 1, b); + b += 5; + } + } + pending_exact = 0; + beg_interval = NULL; + } + break; + + unfetch_interval: + /* If an invalid interval, match the characters as literals. */ + assert (beg_interval); + p = beg_interval; + beg_interval = NULL; + + /* normal_char and normal_backslash need `c'. */ + PATFETCH (c); + + if (!(syntax & RE_NO_BK_BRACES)) + { + if (p > pattern && p[-1] == '\\') + goto normal_backslash; + } + goto normal_char; + +#ifdef emacs + /* There is no way to specify the before_dot and after_dot + operators. rms says this is ok. --karl */ + case '=': + BUF_PUSH (at_dot); + break; + + case 's': + laststart = b; + PATFETCH (c); + BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); + break; + + case 'S': + laststart = b; + PATFETCH (c); + BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]); + break; +#endif /* emacs */ + + + case 'w': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + laststart = b; + BUF_PUSH (wordchar); + break; + + + case 'W': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + laststart = b; + BUF_PUSH (notwordchar); + break; + + + case '<': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + BUF_PUSH (wordbeg); + break; + + case '>': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + BUF_PUSH (wordend); + break; + + case 'b': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + BUF_PUSH (wordbound); + break; + + case 'B': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + BUF_PUSH (notwordbound); + break; + + case '`': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + BUF_PUSH (begbuf); + break; + + case '\'': + if (syntax & RE_NO_GNU_OPS) + goto normal_char; + BUF_PUSH (endbuf); + break; + + case '1': case '2': case '3': case '4': case '5': + case '6': case '7': case '8': case '9': + if (syntax & RE_NO_BK_REFS) + goto normal_char; + + c1 = c - '0'; + + if (c1 > regnum) + FREE_STACK_RETURN (REG_ESUBREG); + + /* Can't back reference to a subexpression if inside of it. */ + if (group_in_compile_stack (compile_stack, (regnum_t) c1)) + goto normal_char; + + laststart = b; + BUF_PUSH_2 (duplicate, c1); + break; + + + case '+': + case '?': + if (syntax & RE_BK_PLUS_QM) + goto handle_plus; + else + goto normal_backslash; + + default: + normal_backslash: + /* You might think it would be useful for \ to mean + not to translate; but if we don't translate it + it will never match anything. */ + c = TRANSLATE (c); + goto normal_char; + } + break; + + + default: + /* Expects the character in `c'. */ + normal_char: + /* If no exactn currently being built. */ + if (!pending_exact + + /* If last exactn not at current position. */ + || pending_exact + *pending_exact + 1 != b + + /* We have only one byte following the exactn for the count. */ + || *pending_exact == (1 << BYTEWIDTH) - 1 + + /* If followed by a repetition operator. */ + || *p == '*' || *p == '^' + || ((syntax & RE_BK_PLUS_QM) + ? *p == '\\' && (p[1] == '+' || p[1] == '?') + : (*p == '+' || *p == '?')) + || ((syntax & RE_INTERVALS) + && ((syntax & RE_NO_BK_BRACES) + ? *p == '{' + : (p[0] == '\\' && p[1] == '{')))) + { + /* Start building a new exactn. */ + + laststart = b; + + BUF_PUSH_2 (exactn, 0); + pending_exact = b - 1; + } + + BUF_PUSH (c); + (*pending_exact)++; + break; + } /* switch (c) */ + } /* while p != pend */ + + + /* Through the pattern now. */ + + if (fixup_alt_jump) + STORE_JUMP (jump_past_alt, fixup_alt_jump, b); + + if (!COMPILE_STACK_EMPTY) + FREE_STACK_RETURN (REG_EPAREN); + + /* If we don't want backtracking, force success + the first time we reach the end of the compiled pattern. */ + if (syntax & RE_NO_POSIX_BACKTRACKING) + BUF_PUSH (succeed); + + free (compile_stack.stack); + + /* We have succeeded; set the length of the buffer. */ + bufp->used = b - bufp->buffer; + +#ifdef DEBUG + if (debug) + { + DEBUG_PRINT1 ("\nCompiled pattern: \n"); + print_compiled_pattern (bufp); + } +#endif /* DEBUG */ + +#ifndef MATCH_MAY_ALLOCATE + /* Initialize the failure stack to the largest possible stack. This + isn't necessary unless we're trying to avoid calling alloca in + the search and match routines. */ + { + int num_regs = bufp->re_nsub + 1; + + /* Since DOUBLE_FAIL_STACK refuses to double only if the current size + is strictly greater than re_max_failures, the largest possible stack + is 2 * re_max_failures failure points. */ + if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS)) + { + fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS); + +# ifdef emacs + if (! fail_stack.stack) + fail_stack.stack + = (fail_stack_elt_t *) xmalloc (fail_stack.size + * sizeof (fail_stack_elt_t)); + else + fail_stack.stack + = (fail_stack_elt_t *) xrealloc (fail_stack.stack, + (fail_stack.size + * sizeof (fail_stack_elt_t))); +# else /* not emacs */ + if (! fail_stack.stack) + fail_stack.stack + = (fail_stack_elt_t *) malloc (fail_stack.size + * sizeof (fail_stack_elt_t)); + else + fail_stack.stack + = (fail_stack_elt_t *) realloc (fail_stack.stack, + (fail_stack.size + * sizeof (fail_stack_elt_t))); +# endif /* not emacs */ + } + + regex_grow_registers (num_regs); + } +#endif /* not MATCH_MAY_ALLOCATE */ + + return REG_NOERROR; +} /* regex_compile */ + +/* Subroutines for `regex_compile'. */ + +/* Store OP at LOC followed by two-byte integer parameter ARG. */ + +static void +store_op1 (op, loc, arg) + re_opcode_t op; + unsigned char *loc; + int arg; +{ + *loc = (unsigned char) op; + STORE_NUMBER (loc + 1, arg); +} + + +/* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */ + +static void +store_op2 (op, loc, arg1, arg2) + re_opcode_t op; + unsigned char *loc; + int arg1, arg2; +{ + *loc = (unsigned char) op; + STORE_NUMBER (loc + 1, arg1); + STORE_NUMBER (loc + 3, arg2); +} + + +/* Copy the bytes from LOC to END to open up three bytes of space at LOC + for OP followed by two-byte integer parameter ARG. */ + +static void +insert_op1 (op, loc, arg, end) + re_opcode_t op; + unsigned char *loc; + int arg; + unsigned char *end; +{ + register unsigned char *pfrom = end; + register unsigned char *pto = end + 3; + + while (pfrom != loc) + *--pto = *--pfrom; + + store_op1 (op, loc, arg); +} + + +/* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */ + +static void +insert_op2 (op, loc, arg1, arg2, end) + re_opcode_t op; + unsigned char *loc; + int arg1, arg2; + unsigned char *end; +{ + register unsigned char *pfrom = end; + register unsigned char *pto = end + 5; + + while (pfrom != loc) + *--pto = *--pfrom; + + store_op2 (op, loc, arg1, arg2); +} + + +/* P points to just after a ^ in PATTERN. Return true if that ^ comes + after an alternative or a begin-subexpression. We assume there is at + least one character before the ^. */ + +static boolean +at_begline_loc_p (pattern, p, syntax) + const char *pattern, *p; + reg_syntax_t syntax; +{ + const char *prev = p - 2; + boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\'; + + return + /* After a subexpression? */ + (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) + /* After an alternative? */ + || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash)); +} + + +/* The dual of at_begline_loc_p. This one is for $. We assume there is + at least one character after the $, i.e., `P < PEND'. */ + +static boolean +at_endline_loc_p (p, pend, syntax) + const char *p, *pend; + reg_syntax_t syntax; +{ + const char *next = p; + boolean next_backslash = *next == '\\'; + const char *next_next = p + 1 < pend ? p + 1 : 0; + + return + /* Before a subexpression? */ + (syntax & RE_NO_BK_PARENS ? *next == ')' + : next_backslash && next_next && *next_next == ')') + /* Before an alternative? */ + || (syntax & RE_NO_BK_VBAR ? *next == '|' + : next_backslash && next_next && *next_next == '|'); +} + + +/* Returns true if REGNUM is in one of COMPILE_STACK's elements and + false if it's not. */ + +static boolean +group_in_compile_stack (compile_stack, regnum) + compile_stack_type compile_stack; + regnum_t regnum; +{ + int this_element; + + for (this_element = compile_stack.avail - 1; + this_element >= 0; + this_element--) + if (compile_stack.stack[this_element].regnum == regnum) + return true; + + return false; +} + + +/* Read the ending character of a range (in a bracket expression) from the + uncompiled pattern *P_PTR (which ends at PEND). We assume the + starting character is in `P[-2]'. (`P[-1]' is the character `-'.) + Then we set the translation of all bits between the starting and + ending characters (inclusive) in the compiled pattern B. + + Return an error code. + + We use these short variable names so we can use the same macros as + `regex_compile' itself. */ + +static reg_errcode_t +compile_range (p_ptr, pend, translate, syntax, b) + const char **p_ptr, *pend; + RE_TRANSLATE_TYPE translate; + reg_syntax_t syntax; + unsigned char *b; +{ + unsigned this_char; + + const char *p = *p_ptr; + unsigned int range_start, range_end; + + if (p == pend) + return REG_ERANGE; + + /* Even though the pattern is a signed `char *', we need to fetch + with unsigned char *'s; if the high bit of the pattern character + is set, the range endpoints will be negative if we fetch using a + signed char *. + + We also want to fetch the endpoints without translating them; the + appropriate translation is done in the bit-setting loop below. */ + /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */ + range_start = ((const unsigned char *) p)[-2]; + range_end = ((const unsigned char *) p)[0]; + + /* Have to increment the pointer into the pattern string, so the + caller isn't still at the ending character. */ + (*p_ptr)++; + + /* If the start is after the end, the range is empty. */ + if (range_start > range_end) + return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; + + /* Here we see why `this_char' has to be larger than an `unsigned + char' -- the range is inclusive, so if `range_end' == 0xff + (assuming 8-bit characters), we would otherwise go into an infinite + loop, since all characters <= 0xff. */ + for (this_char = range_start; this_char <= range_end; this_char++) + { + SET_LIST_BIT (TRANSLATE (this_char)); + } + + return REG_NOERROR; +} + +/* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in + BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible + characters can start a string that matches the pattern. This fastmap + is used by re_search to skip quickly over impossible starting points. + + The caller must supply the address of a (1 << BYTEWIDTH)-byte data + area as BUFP->fastmap. + + We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in + the pattern buffer. + + Returns 0 if we succeed, -2 if an internal error. */ + +int +re_compile_fastmap (bufp) + struct re_pattern_buffer *bufp; +{ + int j, k; +#ifdef MATCH_MAY_ALLOCATE + fail_stack_type fail_stack; +#endif +#ifndef REGEX_MALLOC + char *destination; +#endif + + register char *fastmap = bufp->fastmap; + unsigned char *pattern = bufp->buffer; + unsigned char *p = pattern; + register unsigned char *pend = pattern + bufp->used; + +#ifdef REL_ALLOC + /* This holds the pointer to the failure stack, when + it is allocated relocatably. */ + fail_stack_elt_t *failure_stack_ptr; +#endif + + /* Assume that each path through the pattern can be null until + proven otherwise. We set this false at the bottom of switch + statement, to which we get only if a particular path doesn't + match the empty string. */ + boolean path_can_be_null = true; + + /* We aren't doing a `succeed_n' to begin with. */ + boolean succeed_n_p = false; + + assert (fastmap != NULL && p != NULL); + + INIT_FAIL_STACK (); + bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */ + bufp->fastmap_accurate = 1; /* It will be when we're done. */ + bufp->can_be_null = 0; + + while (1) + { + if (p == pend || *p == succeed) + { + /* We have reached the (effective) end of pattern. */ + if (!FAIL_STACK_EMPTY ()) + { + bufp->can_be_null |= path_can_be_null; + + /* Reset for next path. */ + path_can_be_null = true; + + p = fail_stack.stack[--fail_stack.avail].pointer; + + continue; + } + else + break; + } + + /* We should never be about to go beyond the end of the pattern. */ + assert (p < pend); + + switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++)) + { + + /* I guess the idea here is to simply not bother with a fastmap + if a backreference is used, since it's too hard to figure out + the fastmap for the corresponding group. Setting + `can_be_null' stops `re_search_2' from using the fastmap, so + that is all we do. */ + case duplicate: + bufp->can_be_null = 1; + goto done; + + + /* Following are the cases which match a character. These end + with `break'. */ + + case exactn: + fastmap[p[1]] = 1; + break; + + + case charset: + for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) + if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) + fastmap[j] = 1; + break; + + + case charset_not: + /* Chars beyond end of map must be allowed. */ + for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) + fastmap[j] = 1; + + for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) + if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) + fastmap[j] = 1; + break; + + + case wordchar: + for (j = 0; j < (1 << BYTEWIDTH); j++) + if (SYNTAX (j) == Sword) + fastmap[j] = 1; + break; + + + case notwordchar: + for (j = 0; j < (1 << BYTEWIDTH); j++) + if (SYNTAX (j) != Sword) + fastmap[j] = 1; + break; + + + case anychar: + { + int fastmap_newline = fastmap['\n']; + + /* `.' matches anything ... */ + for (j = 0; j < (1 << BYTEWIDTH); j++) + fastmap[j] = 1; + + /* ... except perhaps newline. */ + if (!(bufp->syntax & RE_DOT_NEWLINE)) + fastmap['\n'] = fastmap_newline; + + /* Return if we have already set `can_be_null'; if we have, + then the fastmap is irrelevant. Something's wrong here. */ + else if (bufp->can_be_null) + goto done; + + /* Otherwise, have to check alternative paths. */ + break; + } + +#ifdef emacs + case syntaxspec: + k = *p++; + for (j = 0; j < (1 << BYTEWIDTH); j++) + if (SYNTAX (j) == (enum syntaxcode) k) + fastmap[j] = 1; + break; + + + case notsyntaxspec: + k = *p++; + for (j = 0; j < (1 << BYTEWIDTH); j++) + if (SYNTAX (j) != (enum syntaxcode) k) + fastmap[j] = 1; + break; + + + /* All cases after this match the empty string. These end with + `continue'. */ + + + case before_dot: + case at_dot: + case after_dot: + continue; +#endif /* emacs */ + + + case no_op: + case begline: + case endline: + case begbuf: + case endbuf: + case wordbound: + case notwordbound: + case wordbeg: + case wordend: + case push_dummy_failure: + continue; + + + case jump_n: + case pop_failure_jump: + case maybe_pop_jump: + case jump: + case jump_past_alt: + case dummy_failure_jump: + EXTRACT_NUMBER_AND_INCR (j, p); + p += j; + if (j > 0) + continue; + + /* Jump backward implies we just went through the body of a + loop and matched nothing. Opcode jumped to should be + `on_failure_jump' or `succeed_n'. Just treat it like an + ordinary jump. For a * loop, it has pushed its failure + point already; if so, discard that as redundant. */ + if ((re_opcode_t) *p != on_failure_jump + && (re_opcode_t) *p != succeed_n) + continue; + + p++; + EXTRACT_NUMBER_AND_INCR (j, p); + p += j; + + /* If what's on the stack is where we are now, pop it. */ + if (!FAIL_STACK_EMPTY () + && fail_stack.stack[fail_stack.avail - 1].pointer == p) + fail_stack.avail--; + + continue; + + + case on_failure_jump: + case on_failure_keep_string_jump: + handle_on_failure_jump: + EXTRACT_NUMBER_AND_INCR (j, p); + + /* For some patterns, e.g., `(a?)?', `p+j' here points to the + end of the pattern. We don't want to push such a point, + since when we restore it above, entering the switch will + increment `p' past the end of the pattern. We don't need + to push such a point since we obviously won't find any more + fastmap entries beyond `pend'. Such a pattern can match + the null string, though. */ + if (p + j < pend) + { + if (!PUSH_PATTERN_OP (p + j, fail_stack)) + { + RESET_FAIL_STACK (); + return -2; + } + } + else + bufp->can_be_null = 1; + + if (succeed_n_p) + { + EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */ + succeed_n_p = false; + } + + continue; + + + case succeed_n: + /* Get to the number of times to succeed. */ + p += 2; + + /* Increment p past the n for when k != 0. */ + EXTRACT_NUMBER_AND_INCR (k, p); + if (k == 0) + { + p -= 4; + succeed_n_p = true; /* Spaghetti code alert. */ + goto handle_on_failure_jump; + } + continue; + + + case set_number_at: + p += 4; + continue; + + + case start_memory: + case stop_memory: + p += 2; + continue; + + + default: + abort (); /* We have listed all the cases. */ + } /* switch *p++ */ + + /* Getting here means we have found the possible starting + characters for one path of the pattern -- and that the empty + string does not match. We need not follow this path further. + Instead, look at the next alternative (remembered on the + stack), or quit if no more. The test at the top of the loop + does these things. */ + path_can_be_null = false; + p = pend; + } /* while p */ + + /* Set `can_be_null' for the last path (also the first path, if the + pattern is empty). */ + bufp->can_be_null |= path_can_be_null; + + done: + RESET_FAIL_STACK (); + return 0; +} /* re_compile_fastmap */ +#ifdef _LIBC +weak_alias (__re_compile_fastmap, re_compile_fastmap) +#endif + +/* Set REGS to hold NUM_REGS registers, storing them in STARTS and + ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use + this memory for recording register information. STARTS and ENDS + must be allocated using the malloc library routine, and must each + be at least NUM_REGS * sizeof (regoff_t) bytes long. + + If NUM_REGS == 0, then subsequent matches should allocate their own + register data. + + Unless this function is called, the first search or match using + PATTERN_BUFFER will allocate its own register data, without + freeing the old data. */ + +void +re_set_registers (bufp, regs, num_regs, starts, ends) + struct re_pattern_buffer *bufp; + struct re_registers *regs; + unsigned num_regs; + regoff_t *starts, *ends; +{ + if (num_regs) + { + bufp->regs_allocated = REGS_REALLOCATE; + regs->num_regs = num_regs; + regs->start = starts; + regs->end = ends; + } + else + { + bufp->regs_allocated = REGS_UNALLOCATED; + regs->num_regs = 0; + regs->start = regs->end = (regoff_t *) 0; + } +} +#ifdef _LIBC +weak_alias (__re_set_registers, re_set_registers) +#endif + +/* Searching routines. */ + +/* Like re_search_2, below, but only one string is specified, and + doesn't let you say where to stop matching. */ + +int +re_search (bufp, string, size, startpos, range, regs) + struct re_pattern_buffer *bufp; + const char *string; + int size, startpos, range; + struct re_registers *regs; +{ + return re_search_2 (bufp, NULL, 0, string, size, startpos, range, + regs, size); +} +#ifdef _LIBC +weak_alias (__re_search, re_search) +#endif + + +/* Using the compiled pattern in BUFP->buffer, first tries to match the + virtual concatenation of STRING1 and STRING2, starting first at index + STARTPOS, then at STARTPOS + 1, and so on. + + STRING1 and STRING2 have length SIZE1 and SIZE2, respectively. + + RANGE is how far to scan while trying to match. RANGE = 0 means try + only at STARTPOS; in general, the last start tried is STARTPOS + + RANGE. + + In REGS, return the indices of the virtual concatenation of STRING1 + and STRING2 that matched the entire BUFP->buffer and its contained + subexpressions. + + Do not consider matching one past the index STOP in the virtual + concatenation of STRING1 and STRING2. + + We return either the position in the strings at which the match was + found, -1 if no match, or -2 if error (such as failure + stack overflow). */ + +int +re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop) + struct re_pattern_buffer *bufp; + const char *string1, *string2; + int size1, size2; + int startpos; + int range; + struct re_registers *regs; + int stop; +{ + int val; + register char *fastmap = bufp->fastmap; + register RE_TRANSLATE_TYPE translate = bufp->translate; + int total_size = size1 + size2; + int endpos = startpos + range; + + /* Check for out-of-range STARTPOS. */ + if (startpos < 0 || startpos > total_size) + return -1; + + /* Fix up RANGE if it might eventually take us outside + the virtual concatenation of STRING1 and STRING2. + Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */ + if (endpos < 0) + range = 0 - startpos; + else if (endpos > total_size) + range = total_size - startpos; + + /* If the search isn't to be a backwards one, don't waste time in a + search for a pattern that must be anchored. */ + if (bufp->used > 0 && range > 0 + && ((re_opcode_t) bufp->buffer[0] == begbuf + /* `begline' is like `begbuf' if it cannot match at newlines. */ + || ((re_opcode_t) bufp->buffer[0] == begline + && !bufp->newline_anchor))) + { + if (startpos > 0) + return -1; + else + range = 1; + } + +#ifdef emacs + /* In a forward search for something that starts with \=. + don't keep searching past point. */ + if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0) + { + range = PT - startpos; + if (range <= 0) + return -1; + } +#endif /* emacs */ + + /* Update the fastmap now if not correct already. */ + if (fastmap && !bufp->fastmap_accurate) + if (re_compile_fastmap (bufp) == -2) + return -2; + + /* Loop through the string, looking for a place to start matching. */ + for (;;) + { + /* If a fastmap is supplied, skip quickly over characters that + cannot be the start of a match. If the pattern can match the + null string, however, we don't need to skip characters; we want + the first null string. */ + if (fastmap && startpos < total_size && !bufp->can_be_null) + { + if (range > 0) /* Searching forwards. */ + { + register const char *d; + register int lim = 0; + int irange = range; + + if (startpos < size1 && startpos + range >= size1) + lim = range - (size1 - startpos); + + d = (startpos >= size1 ? string2 - size1 : string1) + startpos; + + /* Written out as an if-else to avoid testing `translate' + inside the loop. */ + if (translate) + while (range > lim + && !fastmap[(unsigned char) + translate[(unsigned char) *d++]]) + range--; + else + while (range > lim && !fastmap[(unsigned char) *d++]) + range--; + + startpos += irange - range; + } + else /* Searching backwards. */ + { + register char c = (size1 == 0 || startpos >= size1 + ? string2[startpos - size1] + : string1[startpos]); + + if (!fastmap[(unsigned char) TRANSLATE (c)]) + goto advance; + } + } + + /* If can't match the null string, and that's all we have left, fail. */ + if (range >= 0 && startpos == total_size && fastmap + && !bufp->can_be_null) + return -1; + + val = re_match_2_internal (bufp, string1, size1, string2, size2, + startpos, regs, stop); +#ifndef REGEX_MALLOC +# ifdef C_ALLOCA + alloca (0); +# endif +#endif + + if (val >= 0) + return startpos; + + if (val == -2) + return -2; + + advance: + if (!range) + break; + else if (range > 0) + { + range--; + startpos++; + } + else + { + range++; + startpos--; + } + } + return -1; +} /* re_search_2 */ +#ifdef _LIBC +weak_alias (__re_search_2, re_search_2) +#endif + +/* This converts PTR, a pointer into one of the search strings `string1' + and `string2' into an offset from the beginning of that string. */ +#define POINTER_TO_OFFSET(ptr) \ + (FIRST_STRING_P (ptr) \ + ? ((regoff_t) ((ptr) - string1)) \ + : ((regoff_t) ((ptr) - string2 + size1))) + +/* Macros for dealing with the split strings in re_match_2. */ + +#define MATCHING_IN_FIRST_STRING (dend == end_match_1) + +/* Call before fetching a character with *d. This switches over to + string2 if necessary. */ +#define PREFETCH() \ + while (d == dend) \ + { \ + /* End of string2 => fail. */ \ + if (dend == end_match_2) \ + goto fail; \ + /* End of string1 => advance to string2. */ \ + d = string2; \ + dend = end_match_2; \ + } + + +/* Test if at very beginning or at very end of the virtual concatenation + of `string1' and `string2'. If only one string, it's `string2'. */ +#define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) +#define AT_STRINGS_END(d) ((d) == end2) + + +/* Test if D points to a character which is word-constituent. We have + two special cases to check for: if past the end of string1, look at + the first character in string2; and if before the beginning of + string2, look at the last character in string1. */ +#define WORDCHAR_P(d) \ + (SYNTAX ((d) == end1 ? *string2 \ + : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \ + == Sword) + +/* Disabled due to a compiler bug -- see comment at case wordbound */ +#if 0 +/* Test if the character before D and the one at D differ with respect + to being word-constituent. */ +#define AT_WORD_BOUNDARY(d) \ + (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \ + || WORDCHAR_P (d - 1) != WORDCHAR_P (d)) +#endif + +/* Free everything we malloc. */ +#ifdef MATCH_MAY_ALLOCATE +# define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL +# define FREE_VARIABLES() \ + do { \ + REGEX_FREE_STACK (fail_stack.stack); \ + FREE_VAR (regstart); \ + FREE_VAR (regend); \ + FREE_VAR (old_regstart); \ + FREE_VAR (old_regend); \ + FREE_VAR (best_regstart); \ + FREE_VAR (best_regend); \ + FREE_VAR (reg_info); \ + FREE_VAR (reg_dummy); \ + FREE_VAR (reg_info_dummy); \ + } while (0) +#else +# define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */ +#endif /* not MATCH_MAY_ALLOCATE */ + +/* These values must meet several constraints. They must not be valid + register values; since we have a limit of 255 registers (because + we use only one byte in the pattern for the register number), we can + use numbers larger than 255. They must differ by 1, because of + NUM_FAILURE_ITEMS above. And the value for the lowest register must + be larger than the value for the highest register, so we do not try + to actually save any registers when none are active. */ +#define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH) +#define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1) + +/* Matching routines. */ + +#ifndef emacs /* Emacs never uses this. */ +/* re_match is like re_match_2 except it takes only a single string. */ + +int +re_match (bufp, string, size, pos, regs) + struct re_pattern_buffer *bufp; + const char *string; + int size, pos; + struct re_registers *regs; +{ + int result = re_match_2_internal (bufp, NULL, 0, string, size, + pos, regs, size); +# ifndef REGEX_MALLOC +# ifdef C_ALLOCA + alloca (0); +# endif +# endif + return result; +} +# ifdef _LIBC +weak_alias (__re_match, re_match) +# endif +#endif /* not emacs */ + +static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p, + unsigned char *end, + register_info_type *reg_info)); +static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p, + unsigned char *end, + register_info_type *reg_info)); +static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p, + unsigned char *end, + register_info_type *reg_info)); +static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2, + int len, char *translate)); + +/* re_match_2 matches the compiled pattern in BUFP against the + the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 + and SIZE2, respectively). We start matching at POS, and stop + matching at STOP. + + If REGS is non-null and the `no_sub' field of BUFP is nonzero, we + store offsets for the substring each group matched in REGS. See the + documentation for exactly how many groups we fill. + + We return -1 if no match, -2 if an internal error (such as the + failure stack overflowing). Otherwise, we return the length of the + matched substring. */ + +int +re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop) + struct re_pattern_buffer *bufp; + const char *string1, *string2; + int size1, size2; + int pos; + struct re_registers *regs; + int stop; +{ + int result = re_match_2_internal (bufp, string1, size1, string2, size2, + pos, regs, stop); +#ifndef REGEX_MALLOC +# ifdef C_ALLOCA + alloca (0); +# endif +#endif + return result; +} +#ifdef _LIBC +weak_alias (__re_match_2, re_match_2) +#endif + +/* This is a separate function so that we can force an alloca cleanup + afterwards. */ +static int +re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop) + struct re_pattern_buffer *bufp; + const char *string1, *string2; + int size1, size2; + int pos; + struct re_registers *regs; + int stop; +{ + /* General temporaries. */ + int mcnt; + unsigned char *p1; + + /* Just past the end of the corresponding string. */ + const char *end1, *end2; + + /* Pointers into string1 and string2, just past the last characters in + each to consider matching. */ + const char *end_match_1, *end_match_2; + + /* Where we are in the data, and the end of the current string. */ + const char *d, *dend; + + /* Where we are in the pattern, and the end of the pattern. */ + unsigned char *p = bufp->buffer; + register unsigned char *pend = p + bufp->used; + + /* Mark the opcode just after a start_memory, so we can test for an + empty subpattern when we get to the stop_memory. */ + unsigned char *just_past_start_mem = 0; + + /* We use this to map every character in the string. */ + RE_TRANSLATE_TYPE translate = bufp->translate; + + /* Failure point stack. Each place that can handle a failure further + down the line pushes a failure point on this stack. It consists of + restart, regend, and reg_info for all registers corresponding to + the subexpressions we're currently inside, plus the number of such + registers, and, finally, two char *'s. The first char * is where + to resume scanning the pattern; the second one is where to resume + scanning the strings. If the latter is zero, the failure point is + a ``dummy''; if a failure happens and the failure point is a dummy, + it gets discarded and the next next one is tried. */ +#ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ + fail_stack_type fail_stack; +#endif +#ifdef DEBUG + static unsigned failure_id; + unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0; +#endif + +#ifdef REL_ALLOC + /* This holds the pointer to the failure stack, when + it is allocated relocatably. */ + fail_stack_elt_t *failure_stack_ptr; +#endif + + /* We fill all the registers internally, independent of what we + return, for use in backreferences. The number here includes + an element for register zero. */ + size_t num_regs = bufp->re_nsub + 1; + + /* The currently active registers. */ + active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG; + active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG; + + /* Information on the contents of registers. These are pointers into + the input strings; they record just what was matched (on this + attempt) by a subexpression part of the pattern, that is, the + regnum-th regstart pointer points to where in the pattern we began + matching and the regnum-th regend points to right after where we + stopped matching the regnum-th subexpression. (The zeroth register + keeps track of what the whole pattern matches.) */ +#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ + const char **regstart, **regend; +#endif + + /* If a group that's operated upon by a repetition operator fails to + match anything, then the register for its start will need to be + restored because it will have been set to wherever in the string we + are when we last see its open-group operator. Similarly for a + register's end. */ +#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ + const char **old_regstart, **old_regend; +#endif + + /* The is_active field of reg_info helps us keep track of which (possibly + nested) subexpressions we are currently in. The matched_something + field of reg_info[reg_num] helps us tell whether or not we have + matched any of the pattern so far this time through the reg_num-th + subexpression. These two fields get reset each time through any + loop their register is in. */ +#ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ + register_info_type *reg_info; +#endif + + /* The following record the register info as found in the above + variables when we find a match better than any we've seen before. + This happens as we backtrack through the failure points, which in + turn happens only if we have not yet matched the entire string. */ + unsigned best_regs_set = false; +#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ + const char **best_regstart, **best_regend; +#endif + + /* Logically, this is `best_regend[0]'. But we don't want to have to + allocate space for that if we're not allocating space for anything + else (see below). Also, we never need info about register 0 for + any of the other register vectors, and it seems rather a kludge to + treat `best_regend' differently than the rest. So we keep track of + the end of the best match so far in a separate variable. We + initialize this to NULL so that when we backtrack the first time + and need to test it, it's not garbage. */ + const char *match_end = NULL; + + /* This helps SET_REGS_MATCHED avoid doing redundant work. */ + int set_regs_matched_done = 0; + + /* Used when we pop values we don't care about. */ +#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ + const char **reg_dummy; + register_info_type *reg_info_dummy; +#endif + +#ifdef DEBUG + /* Counts the total number of registers pushed. */ + unsigned num_regs_pushed = 0; +#endif + + DEBUG_PRINT1 ("\n\nEntering re_match_2.\n"); + + INIT_FAIL_STACK (); + +#ifdef MATCH_MAY_ALLOCATE + /* Do not bother to initialize all the register variables if there are + no groups in the pattern, as it takes a fair amount of time. If + there are groups, we include space for register 0 (the whole + pattern), even though we never use it, since it simplifies the + array indexing. We should fix this. */ + if (bufp->re_nsub) + { + regstart = REGEX_TALLOC (num_regs, const char *); + regend = REGEX_TALLOC (num_regs, const char *); + old_regstart = REGEX_TALLOC (num_regs, const char *); + old_regend = REGEX_TALLOC (num_regs, const char *); + best_regstart = REGEX_TALLOC (num_regs, const char *); + best_regend = REGEX_TALLOC (num_regs, const char *); + reg_info = REGEX_TALLOC (num_regs, register_info_type); + reg_dummy = REGEX_TALLOC (num_regs, const char *); + reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); + + if (!(regstart && regend && old_regstart && old_regend && reg_info + && best_regstart && best_regend && reg_dummy && reg_info_dummy)) + { + FREE_VARIABLES (); + return -2; + } + } + else + { + /* We must initialize all our variables to NULL, so that + `FREE_VARIABLES' doesn't try to free them. */ + regstart = regend = old_regstart = old_regend = best_regstart + = best_regend = reg_dummy = NULL; + reg_info = reg_info_dummy = (register_info_type *) NULL; + } +#endif /* MATCH_MAY_ALLOCATE */ + + /* The starting position is bogus. */ + if (pos < 0 || pos > size1 + size2) + { + FREE_VARIABLES (); + return -1; + } + + /* Initialize subexpression text positions to -1 to mark ones that no + start_memory/stop_memory has been seen for. Also initialize the + register information struct. */ + for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++) + { + regstart[mcnt] = regend[mcnt] + = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; + + REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; + IS_ACTIVE (reg_info[mcnt]) = 0; + MATCHED_SOMETHING (reg_info[mcnt]) = 0; + EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; + } + + /* We move `string1' into `string2' if the latter's empty -- but not if + `string1' is null. */ + if (size2 == 0 && string1 != NULL) + { + string2 = string1; + size2 = size1; + string1 = 0; + size1 = 0; + } + end1 = string1 + size1; + end2 = string2 + size2; + + /* Compute where to stop matching, within the two strings. */ + if (stop <= size1) + { + end_match_1 = string1 + stop; + end_match_2 = string2; + } + else + { + end_match_1 = end1; + end_match_2 = string2 + stop - size1; + } + + /* `p' scans through the pattern as `d' scans through the data. + `dend' is the end of the input string that `d' points within. `d' + is advanced into the following input string whenever necessary, but + this happens before fetching; therefore, at the beginning of the + loop, `d' can be pointing at the end of a string, but it cannot + equal `string2'. */ + if (size1 > 0 && pos <= size1) + { + d = string1 + pos; + dend = end_match_1; + } + else + { + d = string2 + pos - size1; + dend = end_match_2; + } + + DEBUG_PRINT1 ("The compiled pattern is:\n"); + DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend); + DEBUG_PRINT1 ("The string to match is: `"); + DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); + DEBUG_PRINT1 ("'\n"); + + /* This loops over pattern commands. It exits by returning from the + function if the match is complete, or it drops through if the match + fails at this starting point in the input data. */ + for (;;) + { +#ifdef _LIBC + DEBUG_PRINT2 ("\n%p: ", p); +#else + DEBUG_PRINT2 ("\n0x%x: ", p); +#endif + + if (p == pend) + { /* End of pattern means we might have succeeded. */ + DEBUG_PRINT1 ("end of pattern ... "); + + /* If we haven't matched the entire string, and we want the + longest match, try backtracking. */ + if (d != end_match_2) + { + /* 1 if this match ends in the same string (string1 or string2) + as the best previous match. */ + boolean same_str_p = (FIRST_STRING_P (match_end) + == MATCHING_IN_FIRST_STRING); + /* 1 if this match is the best seen so far. */ + boolean best_match_p; + + /* AIX compiler got confused when this was combined + with the previous declaration. */ + if (same_str_p) + best_match_p = d > match_end; + else + best_match_p = !MATCHING_IN_FIRST_STRING; + + DEBUG_PRINT1 ("backtracking.\n"); + + if (!FAIL_STACK_EMPTY ()) + { /* More failure points to try. */ + + /* If exceeds best match so far, save it. */ + if (!best_regs_set || best_match_p) + { + best_regs_set = true; + match_end = d; + + DEBUG_PRINT1 ("\nSAVING match as best so far.\n"); + + for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++) + { + best_regstart[mcnt] = regstart[mcnt]; + best_regend[mcnt] = regend[mcnt]; + } + } + goto fail; + } + + /* If no failure points, don't restore garbage. And if + last match is real best match, don't restore second + best one. */ + else if (best_regs_set && !best_match_p) + { + restore_best_regs: + /* Restore best match. It may happen that `dend == + end_match_1' while the restored d is in string2. + For example, the pattern `x.*y.*z' against the + strings `x-' and `y-z-', if the two strings are + not consecutive in memory. */ + DEBUG_PRINT1 ("Restoring best registers.\n"); + + d = match_end; + dend = ((d >= string1 && d <= end1) + ? end_match_1 : end_match_2); + + for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++) + { + regstart[mcnt] = best_regstart[mcnt]; + regend[mcnt] = best_regend[mcnt]; + } + } + } /* d != end_match_2 */ + + succeed_label: + DEBUG_PRINT1 ("Accepting match.\n"); + + /* If caller wants register contents data back, do it. */ + if (regs && !bufp->no_sub) + { + /* Have the register data arrays been allocated? */ + if (bufp->regs_allocated == REGS_UNALLOCATED) + { /* No. So allocate them with malloc. We need one + extra element beyond `num_regs' for the `-1' marker + GNU code uses. */ + regs->num_regs = MAX (RE_NREGS, num_regs + 1); + regs->start = TALLOC (regs->num_regs, regoff_t); + regs->end = TALLOC (regs->num_regs, regoff_t); + if (regs->start == NULL || regs->end == NULL) + { + FREE_VARIABLES (); + return -2; + } + bufp->regs_allocated = REGS_REALLOCATE; + } + else if (bufp->regs_allocated == REGS_REALLOCATE) + { /* Yes. If we need more elements than were already + allocated, reallocate them. If we need fewer, just + leave it alone. */ + if (regs->num_regs < num_regs + 1) + { + regs->num_regs = num_regs + 1; + RETALLOC (regs->start, regs->num_regs, regoff_t); + RETALLOC (regs->end, regs->num_regs, regoff_t); + if (regs->start == NULL || regs->end == NULL) + { + FREE_VARIABLES (); + return -2; + } + } + } + else + { + /* These braces fend off a "empty body in an else-statement" + warning under GCC when assert expands to nothing. */ + assert (bufp->regs_allocated == REGS_FIXED); + } + + /* Convert the pointer data in `regstart' and `regend' to + indices. Register zero has to be set differently, + since we haven't kept track of any info for it. */ + if (regs->num_regs > 0) + { + regs->start[0] = pos; + regs->end[0] = (MATCHING_IN_FIRST_STRING + ? ((regoff_t) (d - string1)) + : ((regoff_t) (d - string2 + size1))); + } + + /* Go through the first `min (num_regs, regs->num_regs)' + registers, since that is all we initialized. */ + for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs); + mcnt++) + { + if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt])) + regs->start[mcnt] = regs->end[mcnt] = -1; + else + { + regs->start[mcnt] + = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]); + regs->end[mcnt] + = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]); + } + } + + /* If the regs structure we return has more elements than + were in the pattern, set the extra elements to -1. If + we (re)allocated the registers, this is the case, + because we always allocate enough to have at least one + -1 at the end. */ + for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++) + regs->start[mcnt] = regs->end[mcnt] = -1; + } /* regs && !bufp->no_sub */ + + DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n", + nfailure_points_pushed, nfailure_points_popped, + nfailure_points_pushed - nfailure_points_popped); + DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed); + + mcnt = d - pos - (MATCHING_IN_FIRST_STRING + ? string1 + : string2 - size1); + + DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt); + + FREE_VARIABLES (); + return mcnt; + } + + /* Otherwise match next pattern command. */ + switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++)) + { + /* Ignore these. Used to ignore the n of succeed_n's which + currently have n == 0. */ + case no_op: + DEBUG_PRINT1 ("EXECUTING no_op.\n"); + break; + + case succeed: + DEBUG_PRINT1 ("EXECUTING succeed.\n"); + goto succeed_label; + + /* Match the next n pattern characters exactly. The following + byte in the pattern defines n, and the n bytes after that + are the characters to match. */ + case exactn: + mcnt = *p++; + DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt); + + /* This is written out as an if-else so we don't waste time + testing `translate' inside the loop. */ + if (translate) + { + do + { + PREFETCH (); + if ((unsigned char) translate[(unsigned char) *d++] + != (unsigned char) *p++) + goto fail; + } + while (--mcnt); + } + else + { + do + { + PREFETCH (); + if (*d++ != (char) *p++) goto fail; + } + while (--mcnt); + } + SET_REGS_MATCHED (); + break; + + + /* Match any character except possibly a newline or a null. */ + case anychar: + DEBUG_PRINT1 ("EXECUTING anychar.\n"); + + PREFETCH (); + + if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n') + || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000')) + goto fail; + + SET_REGS_MATCHED (); + DEBUG_PRINT2 (" Matched `%d'.\n", *d); + d++; + break; + + + case charset: + case charset_not: + { + register unsigned char c; + boolean not = (re_opcode_t) *(p - 1) == charset_not; + + DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : ""); + + PREFETCH (); + c = TRANSLATE (*d); /* The character to match. */ + + /* Cast to `unsigned' instead of `unsigned char' in case the + bit list is a full 32 bytes long. */ + if (c < (unsigned) (*p * BYTEWIDTH) + && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) + not = !not; + + p += 1 + *p; + + if (!not) goto fail; + + SET_REGS_MATCHED (); + d++; + break; + } + + + /* The beginning of a group is represented by start_memory. + The arguments are the register number in the next byte, and the + number of groups inner to this one in the next. The text + matched within the group is recorded (in the internal + registers data structure) under the register number. */ + case start_memory: + DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]); + + /* Find out if this group can match the empty string. */ + p1 = p; /* To send to group_match_null_string_p. */ + + if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) + REG_MATCH_NULL_STRING_P (reg_info[*p]) + = group_match_null_string_p (&p1, pend, reg_info); + + /* Save the position in the string where we were the last time + we were at this open-group operator in case the group is + operated upon by a repetition operator, e.g., with `(a*)*b' + against `ab'; then we want to ignore where we are now in + the string in case this attempt to match fails. */ + old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) + ? REG_UNSET (regstart[*p]) ? d : regstart[*p] + : regstart[*p]; + DEBUG_PRINT2 (" old_regstart: %d\n", + POINTER_TO_OFFSET (old_regstart[*p])); + + regstart[*p] = d; + DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p])); + + IS_ACTIVE (reg_info[*p]) = 1; + MATCHED_SOMETHING (reg_info[*p]) = 0; + + /* Clear this whenever we change the register activity status. */ + set_regs_matched_done = 0; + + /* This is the new highest active register. */ + highest_active_reg = *p; + + /* If nothing was active before, this is the new lowest active + register. */ + if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) + lowest_active_reg = *p; + + /* Move past the register number and inner group count. */ + p += 2; + just_past_start_mem = p; + + break; + + + /* The stop_memory opcode represents the end of a group. Its + arguments are the same as start_memory's: the register + number, and the number of inner groups. */ + case stop_memory: + DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); + + /* We need to save the string position the last time we were at + this close-group operator in case the group is operated + upon by a repetition operator, e.g., with `((a*)*(b*)*)*' + against `aba'; then we want to ignore where we are now in + the string in case this attempt to match fails. */ + old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) + ? REG_UNSET (regend[*p]) ? d : regend[*p] + : regend[*p]; + DEBUG_PRINT2 (" old_regend: %d\n", + POINTER_TO_OFFSET (old_regend[*p])); + + regend[*p] = d; + DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p])); + + /* This register isn't active anymore. */ + IS_ACTIVE (reg_info[*p]) = 0; + + /* Clear this whenever we change the register activity status. */ + set_regs_matched_done = 0; + + /* If this was the only register active, nothing is active + anymore. */ + if (lowest_active_reg == highest_active_reg) + { + lowest_active_reg = NO_LOWEST_ACTIVE_REG; + highest_active_reg = NO_HIGHEST_ACTIVE_REG; + } + else + { /* We must scan for the new highest active register, since + it isn't necessarily one less than now: consider + (a(b)c(d(e)f)g). When group 3 ends, after the f), the + new highest active register is 1. */ + unsigned char r = *p - 1; + while (r > 0 && !IS_ACTIVE (reg_info[r])) + r--; + + /* If we end up at register zero, that means that we saved + the registers as the result of an `on_failure_jump', not + a `start_memory', and we jumped to past the innermost + `stop_memory'. For example, in ((.)*) we save + registers 1 and 2 as a result of the *, but when we pop + back to the second ), we are at the stop_memory 1. + Thus, nothing is active. */ + if (r == 0) + { + lowest_active_reg = NO_LOWEST_ACTIVE_REG; + highest_active_reg = NO_HIGHEST_ACTIVE_REG; + } + else + highest_active_reg = r; + } + + /* If just failed to match something this time around with a + group that's operated on by a repetition operator, try to + force exit from the ``loop'', and restore the register + information for this group that we had before trying this + last match. */ + if ((!MATCHED_SOMETHING (reg_info[*p]) + || just_past_start_mem == p - 1) + && (p + 2) < pend) + { + boolean is_a_jump_n = false; + + p1 = p + 2; + mcnt = 0; + switch ((re_opcode_t) *p1++) + { + case jump_n: + is_a_jump_n = true; + case pop_failure_jump: + case maybe_pop_jump: + case jump: + case dummy_failure_jump: + EXTRACT_NUMBER_AND_INCR (mcnt, p1); + if (is_a_jump_n) + p1 += 2; + break; + + default: + /* do nothing */ ; + } + p1 += mcnt; + + /* If the next operation is a jump backwards in the pattern + to an on_failure_jump right before the start_memory + corresponding to this stop_memory, exit from the loop + by forcing a failure after pushing on the stack the + on_failure_jump's jump in the pattern, and d. */ + if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump + && (re_opcode_t) p1[3] == start_memory && p1[4] == *p) + { + /* If this group ever matched anything, then restore + what its registers were before trying this last + failed match, e.g., with `(a*)*b' against `ab' for + regstart[1], and, e.g., with `((a*)*(b*)*)*' + against `aba' for regend[3]. + + Also restore the registers for inner groups for, + e.g., `((a*)(b*))*' against `aba' (register 3 would + otherwise get trashed). */ + + if (EVER_MATCHED_SOMETHING (reg_info[*p])) + { + unsigned r; + + EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; + + /* Restore this and inner groups' (if any) registers. */ + for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1); + r++) + { + regstart[r] = old_regstart[r]; + + /* xx why this test? */ + if (old_regend[r] >= regstart[r]) + regend[r] = old_regend[r]; + } + } + p1++; + EXTRACT_NUMBER_AND_INCR (mcnt, p1); + PUSH_FAILURE_POINT (p1 + mcnt, d, -2); + + goto fail; + } + } + + /* Move past the register number and the inner group count. */ + p += 2; + break; + + + /* \<digit> has been turned into a `duplicate' command which is + followed by the numeric value of <digit> as the register number. */ + case duplicate: + { + register const char *d2, *dend2; + int regno = *p++; /* Get which register to match against. */ + DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno); + + /* Can't back reference a group which we've never matched. */ + if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) + goto fail; + + /* Where in input to try to start matching. */ + d2 = regstart[regno]; + + /* Where to stop matching; if both the place to start and + the place to stop matching are in the same string, then + set to the place to stop, otherwise, for now have to use + the end of the first string. */ + + dend2 = ((FIRST_STRING_P (regstart[regno]) + == FIRST_STRING_P (regend[regno])) + ? regend[regno] : end_match_1); + for (;;) + { + /* If necessary, advance to next segment in register + contents. */ + while (d2 == dend2) + { + if (dend2 == end_match_2) break; + if (dend2 == regend[regno]) break; + + /* End of string1 => advance to string2. */ + d2 = string2; + dend2 = regend[regno]; + } + /* At end of register contents => success */ + if (d2 == dend2) break; + + /* If necessary, advance to next segment in data. */ + PREFETCH (); + + /* How many characters left in this segment to match. */ + mcnt = dend - d; + + /* Want how many consecutive characters we can match in + one shot, so, if necessary, adjust the count. */ + if (mcnt > dend2 - d2) + mcnt = dend2 - d2; + + /* Compare that many; failure if mismatch, else move + past them. */ + if (translate + ? bcmp_translate (d, d2, mcnt, translate) + : memcmp (d, d2, mcnt)) + goto fail; + d += mcnt, d2 += mcnt; + + /* Do this because we've match some characters. */ + SET_REGS_MATCHED (); + } + } + break; + + + /* begline matches the empty string at the beginning of the string + (unless `not_bol' is set in `bufp'), and, if + `newline_anchor' is set, after newlines. */ + case begline: + DEBUG_PRINT1 ("EXECUTING begline.\n"); + + if (AT_STRINGS_BEG (d)) + { + if (!bufp->not_bol) break; + } + else if (d[-1] == '\n' && bufp->newline_anchor) + { + break; + } + /* In all other cases, we fail. */ + goto fail; + + + /* endline is the dual of begline. */ + case endline: + DEBUG_PRINT1 ("EXECUTING endline.\n"); + + if (AT_STRINGS_END (d)) + { + if (!bufp->not_eol) break; + } + + /* We have to ``prefetch'' the next character. */ + else if ((d == end1 ? *string2 : *d) == '\n' + && bufp->newline_anchor) + { + break; + } + goto fail; + + + /* Match at the very beginning of the data. */ + case begbuf: + DEBUG_PRINT1 ("EXECUTING begbuf.\n"); + if (AT_STRINGS_BEG (d)) + break; + goto fail; + + + /* Match at the very end of the data. */ + case endbuf: + DEBUG_PRINT1 ("EXECUTING endbuf.\n"); + if (AT_STRINGS_END (d)) + break; + goto fail; + + + /* on_failure_keep_string_jump is used to optimize `.*\n'. It + pushes NULL as the value for the string on the stack. Then + `pop_failure_point' will keep the current value for the + string, instead of restoring it. To see why, consider + matching `foo\nbar' against `.*\n'. The .* matches the foo; + then the . fails against the \n. But the next thing we want + to do is match the \n against the \n; if we restored the + string value, we would be back at the foo. + + Because this is used only in specific cases, we don't need to + check all the things that `on_failure_jump' does, to make + sure the right things get saved on the stack. Hence we don't + share its code. The only reason to push anything on the + stack at all is that otherwise we would have to change + `anychar's code to do something besides goto fail in this + case; that seems worse than this. */ + case on_failure_keep_string_jump: + DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump"); + + EXTRACT_NUMBER_AND_INCR (mcnt, p); +#ifdef _LIBC + DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt); +#else + DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt); +#endif + + PUSH_FAILURE_POINT (p + mcnt, NULL, -2); + break; + + + /* Uses of on_failure_jump: + + Each alternative starts with an on_failure_jump that points + to the beginning of the next alternative. Each alternative + except the last ends with a jump that in effect jumps past + the rest of the alternatives. (They really jump to the + ending jump of the following alternative, because tensioning + these jumps is a hassle.) + + Repeats start with an on_failure_jump that points past both + the repetition text and either the following jump or + pop_failure_jump back to this on_failure_jump. */ + case on_failure_jump: + on_failure: + DEBUG_PRINT1 ("EXECUTING on_failure_jump"); + + EXTRACT_NUMBER_AND_INCR (mcnt, p); +#ifdef _LIBC + DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt); +#else + DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt); +#endif + + /* If this on_failure_jump comes right before a group (i.e., + the original * applied to a group), save the information + for that group and all inner ones, so that if we fail back + to this point, the group's information will be correct. + For example, in \(a*\)*\1, we need the preceding group, + and in \(zz\(a*\)b*\)\2, we need the inner group. */ + + /* We can't use `p' to check ahead because we push + a failure point to `p + mcnt' after we do this. */ + p1 = p; + + /* We need to skip no_op's before we look for the + start_memory in case this on_failure_jump is happening as + the result of a completed succeed_n, as in \(a\)\{1,3\}b\1 + against aba. */ + while (p1 < pend && (re_opcode_t) *p1 == no_op) + p1++; + + if (p1 < pend && (re_opcode_t) *p1 == start_memory) + { + /* We have a new highest active register now. This will + get reset at the start_memory we are about to get to, + but we will have saved all the registers relevant to + this repetition op, as described above. */ + highest_active_reg = *(p1 + 1) + *(p1 + 2); + if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) + lowest_active_reg = *(p1 + 1); + } + + DEBUG_PRINT1 (":\n"); + PUSH_FAILURE_POINT (p + mcnt, d, -2); + break; + + + /* A smart repeat ends with `maybe_pop_jump'. + We change it to either `pop_failure_jump' or `jump'. */ + case maybe_pop_jump: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt); + { + register unsigned char *p2 = p; + + /* Compare the beginning of the repeat with what in the + pattern follows its end. If we can establish that there + is nothing that they would both match, i.e., that we + would have to backtrack because of (as in, e.g., `a*a') + then we can change to pop_failure_jump, because we'll + never have to backtrack. + + This is not true in the case of alternatives: in + `(a|ab)*' we do need to backtrack to the `ab' alternative + (e.g., if the string was `ab'). But instead of trying to + detect that here, the alternative has put on a dummy + failure point which is what we will end up popping. */ + + /* Skip over open/close-group commands. + If what follows this loop is a ...+ construct, + look at what begins its body, since we will have to + match at least one of that. */ + while (1) + { + if (p2 + 2 < pend + && ((re_opcode_t) *p2 == stop_memory + || (re_opcode_t) *p2 == start_memory)) + p2 += 3; + else if (p2 + 6 < pend + && (re_opcode_t) *p2 == dummy_failure_jump) + p2 += 6; + else + break; + } + + p1 = p + mcnt; + /* p1[0] ... p1[2] are the `on_failure_jump' corresponding + to the `maybe_finalize_jump' of this case. Examine what + follows. */ + + /* If we're at the end of the pattern, we can change. */ + if (p2 == pend) + { + /* Consider what happens when matching ":\(.*\)" + against ":/". I don't really understand this code + yet. */ + p[-3] = (unsigned char) pop_failure_jump; + DEBUG_PRINT1 + (" End of pattern: change to `pop_failure_jump'.\n"); + } + + else if ((re_opcode_t) *p2 == exactn + || (bufp->newline_anchor && (re_opcode_t) *p2 == endline)) + { + register unsigned char c + = *p2 == (unsigned char) endline ? '\n' : p2[2]; + + if ((re_opcode_t) p1[3] == exactn && p1[5] != c) + { + p[-3] = (unsigned char) pop_failure_jump; + DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", + c, p1[5]); + } + + else if ((re_opcode_t) p1[3] == charset + || (re_opcode_t) p1[3] == charset_not) + { + int not = (re_opcode_t) p1[3] == charset_not; + + if (c < (unsigned char) (p1[4] * BYTEWIDTH) + && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) + not = !not; + + /* `not' is equal to 1 if c would match, which means + that we can't change to pop_failure_jump. */ + if (!not) + { + p[-3] = (unsigned char) pop_failure_jump; + DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); + } + } + } + else if ((re_opcode_t) *p2 == charset) + { + /* We win if the first character of the loop is not part + of the charset. */ + if ((re_opcode_t) p1[3] == exactn + && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5] + && (p2[2 + p1[5] / BYTEWIDTH] + & (1 << (p1[5] % BYTEWIDTH))))) + { + p[-3] = (unsigned char) pop_failure_jump; + DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); + } + + else if ((re_opcode_t) p1[3] == charset_not) + { + int idx; + /* We win if the charset_not inside the loop + lists every character listed in the charset after. */ + for (idx = 0; idx < (int) p2[1]; idx++) + if (! (p2[2 + idx] == 0 + || (idx < (int) p1[4] + && ((p2[2 + idx] & ~ p1[5 + idx]) == 0)))) + break; + + if (idx == p2[1]) + { + p[-3] = (unsigned char) pop_failure_jump; + DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); + } + } + else if ((re_opcode_t) p1[3] == charset) + { + int idx; + /* We win if the charset inside the loop + has no overlap with the one after the loop. */ + for (idx = 0; + idx < (int) p2[1] && idx < (int) p1[4]; + idx++) + if ((p2[2 + idx] & p1[5 + idx]) != 0) + break; + + if (idx == p2[1] || idx == p1[4]) + { + p[-3] = (unsigned char) pop_failure_jump; + DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); + } + } + } + } + p -= 2; /* Point at relative address again. */ + if ((re_opcode_t) p[-1] != pop_failure_jump) + { + p[-1] = (unsigned char) jump; + DEBUG_PRINT1 (" Match => jump.\n"); + goto unconditional_jump; + } + /* Note fall through. */ + + + /* The end of a simple repeat has a pop_failure_jump back to + its matching on_failure_jump, where the latter will push a + failure point. The pop_failure_jump takes off failure + points put on by this pop_failure_jump's matching + on_failure_jump; we got through the pattern to here from the + matching on_failure_jump, so didn't fail. */ + case pop_failure_jump: + { + /* We need to pass separate storage for the lowest and + highest registers, even though we don't care about the + actual values. Otherwise, we will restore only one + register from the stack, since lowest will == highest in + `pop_failure_point'. */ + active_reg_t dummy_low_reg, dummy_high_reg; + unsigned char *pdummy; + const char *sdummy; + + DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n"); + POP_FAILURE_POINT (sdummy, pdummy, + dummy_low_reg, dummy_high_reg, + reg_dummy, reg_dummy, reg_info_dummy); + } + /* Note fall through. */ + + unconditional_jump: +#ifdef _LIBC + DEBUG_PRINT2 ("\n%p: ", p); +#else + DEBUG_PRINT2 ("\n0x%x: ", p); +#endif + /* Note fall through. */ + + /* Unconditionally jump (without popping any failure points). */ + case jump: + EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */ + DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt); + p += mcnt; /* Do the jump. */ +#ifdef _LIBC + DEBUG_PRINT2 ("(to %p).\n", p); +#else + DEBUG_PRINT2 ("(to 0x%x).\n", p); +#endif + break; + + + /* We need this opcode so we can detect where alternatives end + in `group_match_null_string_p' et al. */ + case jump_past_alt: + DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n"); + goto unconditional_jump; + + + /* Normally, the on_failure_jump pushes a failure point, which + then gets popped at pop_failure_jump. We will end up at + pop_failure_jump, also, and with a pattern of, say, `a+', we + are skipping over the on_failure_jump, so we have to push + something meaningless for pop_failure_jump to pop. */ + case dummy_failure_jump: + DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n"); + /* It doesn't matter what we push for the string here. What + the code at `fail' tests is the value for the pattern. */ + PUSH_FAILURE_POINT (NULL, NULL, -2); + goto unconditional_jump; + + + /* At the end of an alternative, we need to push a dummy failure + point in case we are followed by a `pop_failure_jump', because + we don't want the failure point for the alternative to be + popped. For example, matching `(a|ab)*' against `aab' + requires that we match the `ab' alternative. */ + case push_dummy_failure: + DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n"); + /* See comments just above at `dummy_failure_jump' about the + two zeroes. */ + PUSH_FAILURE_POINT (NULL, NULL, -2); + break; + + /* Have to succeed matching what follows at least n times. + After that, handle like `on_failure_jump'. */ + case succeed_n: + EXTRACT_NUMBER (mcnt, p + 2); + DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); + + assert (mcnt >= 0); + /* Originally, this is how many times we HAVE to succeed. */ + if (mcnt > 0) + { + mcnt--; + p += 2; + STORE_NUMBER_AND_INCR (p, mcnt); +#ifdef _LIBC + DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt); +#else + DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt); +#endif + } + else if (mcnt == 0) + { +#ifdef _LIBC + DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2); +#else + DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2); +#endif + p[2] = (unsigned char) no_op; + p[3] = (unsigned char) no_op; + goto on_failure; + } + break; + + case jump_n: + EXTRACT_NUMBER (mcnt, p + 2); + DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); + + /* Originally, this is how many times we CAN jump. */ + if (mcnt) + { + mcnt--; + STORE_NUMBER (p + 2, mcnt); +#ifdef _LIBC + DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt); +#else + DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt); +#endif + goto unconditional_jump; + } + /* If don't have to jump any more, skip over the rest of command. */ + else + p += 4; + break; + + case set_number_at: + { + DEBUG_PRINT1 ("EXECUTING set_number_at.\n"); + + EXTRACT_NUMBER_AND_INCR (mcnt, p); + p1 = p + mcnt; + EXTRACT_NUMBER_AND_INCR (mcnt, p); +#ifdef _LIBC + DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt); +#else + DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt); +#endif + STORE_NUMBER (p1, mcnt); + break; + } + +#if 0 + /* The DEC Alpha C compiler 3.x generates incorrect code for the + test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of + AT_WORD_BOUNDARY, so this code is disabled. Expanding the + macro and introducing temporary variables works around the bug. */ + + case wordbound: + DEBUG_PRINT1 ("EXECUTING wordbound.\n"); + if (AT_WORD_BOUNDARY (d)) + break; + goto fail; + + case notwordbound: + DEBUG_PRINT1 ("EXECUTING notwordbound.\n"); + if (AT_WORD_BOUNDARY (d)) + goto fail; + break; +#else + case wordbound: + { + boolean prevchar, thischar; + + DEBUG_PRINT1 ("EXECUTING wordbound.\n"); + if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d)) + break; + + prevchar = WORDCHAR_P (d - 1); + thischar = WORDCHAR_P (d); + if (prevchar != thischar) + break; + goto fail; + } + + case notwordbound: + { + boolean prevchar, thischar; + + DEBUG_PRINT1 ("EXECUTING notwordbound.\n"); + if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d)) + goto fail; + + prevchar = WORDCHAR_P (d - 1); + thischar = WORDCHAR_P (d); + if (prevchar != thischar) + goto fail; + break; + } +#endif + + case wordbeg: + DEBUG_PRINT1 ("EXECUTING wordbeg.\n"); + if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1))) + break; + goto fail; + + case wordend: + DEBUG_PRINT1 ("EXECUTING wordend.\n"); + if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1) + && (!WORDCHAR_P (d) || AT_STRINGS_END (d))) + break; + goto fail; + +#ifdef emacs + case before_dot: + DEBUG_PRINT1 ("EXECUTING before_dot.\n"); + if (PTR_CHAR_POS ((unsigned char *) d) >= point) + goto fail; + break; + + case at_dot: + DEBUG_PRINT1 ("EXECUTING at_dot.\n"); + if (PTR_CHAR_POS ((unsigned char *) d) != point) + goto fail; + break; + + case after_dot: + DEBUG_PRINT1 ("EXECUTING after_dot.\n"); + if (PTR_CHAR_POS ((unsigned char *) d) <= point) + goto fail; + break; + + case syntaxspec: + DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt); + mcnt = *p++; + goto matchsyntax; + + case wordchar: + DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n"); + mcnt = (int) Sword; + matchsyntax: + PREFETCH (); + /* Can't use *d++ here; SYNTAX may be an unsafe macro. */ + d++; + if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt) + goto fail; + SET_REGS_MATCHED (); + break; + + case notsyntaxspec: + DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt); + mcnt = *p++; + goto matchnotsyntax; + + case notwordchar: + DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n"); + mcnt = (int) Sword; + matchnotsyntax: + PREFETCH (); + /* Can't use *d++ here; SYNTAX may be an unsafe macro. */ + d++; + if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt) + goto fail; + SET_REGS_MATCHED (); + break; + +#else /* not emacs */ + case wordchar: + DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n"); + PREFETCH (); + if (!WORDCHAR_P (d)) + goto fail; + SET_REGS_MATCHED (); + d++; + break; + + case notwordchar: + DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n"); + PREFETCH (); + if (WORDCHAR_P (d)) + goto fail; + SET_REGS_MATCHED (); + d++; + break; +#endif /* not emacs */ + + default: + abort (); + } + continue; /* Successfully executed one pattern command; keep going. */ + + + /* We goto here if a matching operation fails. */ + fail: + if (!FAIL_STACK_EMPTY ()) + { /* A restart point is known. Restore to that state. */ + DEBUG_PRINT1 ("\nFAIL:\n"); + POP_FAILURE_POINT (d, p, + lowest_active_reg, highest_active_reg, + regstart, regend, reg_info); + + /* If this failure point is a dummy, try the next one. */ + if (!p) + goto fail; + + /* If we failed to the end of the pattern, don't examine *p. */ + assert (p <= pend); + if (p < pend) + { + boolean is_a_jump_n = false; + + /* If failed to a backwards jump that's part of a repetition + loop, need to pop this failure point and use the next one. */ + switch ((re_opcode_t) *p) + { + case jump_n: + is_a_jump_n = true; + case maybe_pop_jump: + case pop_failure_jump: + case jump: + p1 = p + 1; + EXTRACT_NUMBER_AND_INCR (mcnt, p1); + p1 += mcnt; + + if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n) + || (!is_a_jump_n + && (re_opcode_t) *p1 == on_failure_jump)) + goto fail; + break; + default: + /* do nothing */ ; + } + } + + if (d >= string1 && d <= end1) + dend = end_match_1; + } + else + break; /* Matching at this starting point really fails. */ + } /* for (;;) */ + + if (best_regs_set) + goto restore_best_regs; + + FREE_VARIABLES (); + + return -1; /* Failure to match. */ +} /* re_match_2 */ + +/* Subroutine definitions for re_match_2. */ + + +/* We are passed P pointing to a register number after a start_memory. + + Return true if the pattern up to the corresponding stop_memory can + match the empty string, and false otherwise. + + If we find the matching stop_memory, sets P to point to one past its number. + Otherwise, sets P to an undefined byte less than or equal to END. + + We don't handle duplicates properly (yet). */ + +static boolean +group_match_null_string_p (p, end, reg_info) + unsigned char **p, *end; + register_info_type *reg_info; +{ + int mcnt; + /* Point to after the args to the start_memory. */ + unsigned char *p1 = *p + 2; + + while (p1 < end) + { + /* Skip over opcodes that can match nothing, and return true or + false, as appropriate, when we get to one that can't, or to the + matching stop_memory. */ + + switch ((re_opcode_t) *p1) + { + /* Could be either a loop or a series of alternatives. */ + case on_failure_jump: + p1++; + EXTRACT_NUMBER_AND_INCR (mcnt, p1); + + /* If the next operation is not a jump backwards in the + pattern. */ + + if (mcnt >= 0) + { + /* Go through the on_failure_jumps of the alternatives, + seeing if any of the alternatives cannot match nothing. + The last alternative starts with only a jump, + whereas the rest start with on_failure_jump and end + with a jump, e.g., here is the pattern for `a|b|c': + + /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6 + /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3 + /exactn/1/c + + So, we have to first go through the first (n-1) + alternatives and then deal with the last one separately. */ + + + /* Deal with the first (n-1) alternatives, which start + with an on_failure_jump (see above) that jumps to right + past a jump_past_alt. */ + + while ((re_opcode_t) p1[mcnt-3] == jump_past_alt) + { + /* `mcnt' holds how many bytes long the alternative + is, including the ending `jump_past_alt' and + its number. */ + + if (!alt_match_null_string_p (p1, p1 + mcnt - 3, + reg_info)) + return false; + + /* Move to right after this alternative, including the + jump_past_alt. */ + p1 += mcnt; + + /* Break if it's the beginning of an n-th alternative + that doesn't begin with an on_failure_jump. */ + if ((re_opcode_t) *p1 != on_failure_jump) + break; + + /* Still have to check that it's not an n-th + alternative that starts with an on_failure_jump. */ + p1++; + EXTRACT_NUMBER_AND_INCR (mcnt, p1); + if ((re_opcode_t) p1[mcnt-3] != jump_past_alt) + { + /* Get to the beginning of the n-th alternative. */ + p1 -= 3; + break; + } + } + + /* Deal with the last alternative: go back and get number + of the `jump_past_alt' just before it. `mcnt' contains + the length of the alternative. */ + EXTRACT_NUMBER (mcnt, p1 - 2); + + if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info)) + return false; + + p1 += mcnt; /* Get past the n-th alternative. */ + } /* if mcnt > 0 */ + break; + + + case stop_memory: + assert (p1[1] == **p); + *p = p1 + 2; + return true; + + + default: + if (!common_op_match_null_string_p (&p1, end, reg_info)) + return false; + } + } /* while p1 < end */ + + return false; +} /* group_match_null_string_p */ + + +/* Similar to group_match_null_string_p, but doesn't deal with alternatives: + It expects P to be the first byte of a single alternative and END one + byte past the last. The alternative can contain groups. */ + +static boolean +alt_match_null_string_p (p, end, reg_info) + unsigned char *p, *end; + register_info_type *reg_info; +{ + int mcnt; + unsigned char *p1 = p; + + while (p1 < end) + { + /* Skip over opcodes that can match nothing, and break when we get + to one that can't. */ + + switch ((re_opcode_t) *p1) + { + /* It's a loop. */ + case on_failure_jump: + p1++; + EXTRACT_NUMBER_AND_INCR (mcnt, p1); + p1 += mcnt; + break; + + default: + if (!common_op_match_null_string_p (&p1, end, reg_info)) + return false; + } + } /* while p1 < end */ + + return true; +} /* alt_match_null_string_p */ + + +/* Deals with the ops common to group_match_null_string_p and + alt_match_null_string_p. + + Sets P to one after the op and its arguments, if any. */ + +static boolean +common_op_match_null_string_p (p, end, reg_info) + unsigned char **p, *end; + register_info_type *reg_info; +{ + int mcnt; + boolean ret; + int reg_no; + unsigned char *p1 = *p; + + switch ((re_opcode_t) *p1++) + { + case no_op: + case begline: + case endline: + case begbuf: + case endbuf: + case wordbeg: + case wordend: + case wordbound: + case notwordbound: +#ifdef emacs + case before_dot: + case at_dot: + case after_dot: +#endif + break; + + case start_memory: + reg_no = *p1; + assert (reg_no > 0 && reg_no <= MAX_REGNUM); + ret = group_match_null_string_p (&p1, end, reg_info); + + /* Have to set this here in case we're checking a group which + contains a group and a back reference to it. */ + + if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE) + REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret; + + if (!ret) + return false; + break; + + /* If this is an optimized succeed_n for zero times, make the jump. */ + case jump: + EXTRACT_NUMBER_AND_INCR (mcnt, p1); + if (mcnt >= 0) + p1 += mcnt; + else + return false; + break; + + case succeed_n: + /* Get to the number of times to succeed. */ + p1 += 2; + EXTRACT_NUMBER_AND_INCR (mcnt, p1); + + if (mcnt == 0) + { + p1 -= 4; + EXTRACT_NUMBER_AND_INCR (mcnt, p1); + p1 += mcnt; + } + else + return false; + break; + + case duplicate: + if (!REG_MATCH_NULL_STRING_P (reg_info[*p1])) + return false; + break; + + case set_number_at: + p1 += 4; + + default: + /* All other opcodes mean we cannot match the empty string. */ + return false; + } + + *p = p1; + return true; +} /* common_op_match_null_string_p */ + + +/* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN + bytes; nonzero otherwise. */ + +static int +bcmp_translate (s1, s2, len, translate) + const char *s1, *s2; + register int len; + RE_TRANSLATE_TYPE translate; +{ + register const unsigned char *p1 = (const unsigned char *) s1; + register const unsigned char *p2 = (const unsigned char *) s2; + while (len) + { + if (translate[*p1++] != translate[*p2++]) return 1; + len--; + } + return 0; +} + +/* Entry points for GNU code. */ + +/* re_compile_pattern is the GNU regular expression compiler: it + compiles PATTERN (of length SIZE) and puts the result in BUFP. + Returns 0 if the pattern was valid, otherwise an error string. + + Assumes the `allocated' (and perhaps `buffer') and `translate' fields + are set in BUFP on entry. + + We call regex_compile to do the actual compilation. */ + +const char * +re_compile_pattern (pattern, length, bufp) + const char *pattern; + size_t length; + struct re_pattern_buffer *bufp; +{ + reg_errcode_t ret; + + /* GNU code is written to assume at least RE_NREGS registers will be set + (and at least one extra will be -1). */ + bufp->regs_allocated = REGS_UNALLOCATED; + + /* And GNU code determines whether or not to get register information + by passing null for the REGS argument to re_match, etc., not by + setting no_sub. */ + bufp->no_sub = 0; + + /* Match anchors at newline. */ + bufp->newline_anchor = 1; + + ret = regex_compile (pattern, length, re_syntax_options, bufp); + + if (!ret) + return NULL; + return gettext (re_error_msgid + re_error_msgid_idx[(int) ret]); +} +#ifdef _LIBC +weak_alias (__re_compile_pattern, re_compile_pattern) +#endif + +/* Entry points compatible with 4.2 BSD regex library. We don't define + them unless specifically requested. */ + +#if defined _REGEX_RE_COMP || defined _LIBC + +/* BSD has one and only one pattern buffer. */ +static struct re_pattern_buffer re_comp_buf; + +char * +#ifdef _LIBC +/* Make these definitions weak in libc, so POSIX programs can redefine + these names if they don't use our functions, and still use + regcomp/regexec below without link errors. */ +weak_function +#endif +re_comp (s) + const char *s; +{ + reg_errcode_t ret; + + if (!s) + { + if (!re_comp_buf.buffer) + return gettext ("No previous regular expression"); + return 0; + } + + if (!re_comp_buf.buffer) + { + re_comp_buf.buffer = (unsigned char *) malloc (200); + if (re_comp_buf.buffer == NULL) + return (char *) gettext (re_error_msgid + + re_error_msgid_idx[(int) REG_ESPACE]); + re_comp_buf.allocated = 200; + + re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH); + if (re_comp_buf.fastmap == NULL) + return (char *) gettext (re_error_msgid + + re_error_msgid_idx[(int) REG_ESPACE]); + } + + /* Since `re_exec' always passes NULL for the `regs' argument, we + don't need to initialize the pattern buffer fields which affect it. */ + + /* Match anchors at newlines. */ + re_comp_buf.newline_anchor = 1; + + ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf); + + if (!ret) + return NULL; + + /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */ + return (char *) gettext (re_error_msgid + re_error_msgid_idx[(int) ret]); +} + + +int +#ifdef _LIBC +weak_function +#endif +re_exec (s) + const char *s; +{ + const int len = strlen (s); + return + 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0); +} + +#endif /* _REGEX_RE_COMP */ + +/* POSIX.2 functions. Don't define these for Emacs. */ + +#ifndef emacs + +/* regcomp takes a regular expression as a string and compiles it. + + PREG is a regex_t *. We do not expect any fields to be initialized, + since POSIX says we shouldn't. Thus, we set + + `buffer' to the compiled pattern; + `used' to the length of the compiled pattern; + `syntax' to RE_SYNTAX_POSIX_EXTENDED if the + REG_EXTENDED bit in CFLAGS is set; otherwise, to + RE_SYNTAX_POSIX_BASIC; + `newline_anchor' to REG_NEWLINE being set in CFLAGS; + `fastmap' to an allocated space for the fastmap; + `fastmap_accurate' to zero; + `re_nsub' to the number of subexpressions in PATTERN. + + PATTERN is the address of the pattern string. + + CFLAGS is a series of bits which affect compilation. + + If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we + use POSIX basic syntax. + + If REG_NEWLINE is set, then . and [^...] don't match newline. + Also, regexec will try a match beginning after every newline. + + If REG_ICASE is set, then we considers upper- and lowercase + versions of letters to be equivalent when matching. + + If REG_NOSUB is set, then when PREG is passed to regexec, that + routine will report only success or failure, and nothing about the + registers. + + It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for + the return codes and their meanings.) */ + +int +regcomp (preg, pattern, cflags) + regex_t *preg; + const char *pattern; + int cflags; +{ + reg_errcode_t ret; + reg_syntax_t syntax + = (cflags & REG_EXTENDED) ? + RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC; + + /* regex_compile will allocate the space for the compiled pattern. */ + preg->buffer = 0; + preg->allocated = 0; + preg->used = 0; + + /* Try to allocate space for the fastmap. */ + preg->fastmap = (char *) malloc (1 << BYTEWIDTH); + + if (cflags & REG_ICASE) + { + unsigned i; + + preg->translate + = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE + * sizeof (*(RE_TRANSLATE_TYPE)0)); + if (preg->translate == NULL) + return (int) REG_ESPACE; + + /* Map uppercase characters to corresponding lowercase ones. */ + for (i = 0; i < CHAR_SET_SIZE; i++) + preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i; + } + else + preg->translate = NULL; + + /* If REG_NEWLINE is set, newlines are treated differently. */ + if (cflags & REG_NEWLINE) + { /* REG_NEWLINE implies neither . nor [^...] match newline. */ + syntax &= ~RE_DOT_NEWLINE; + syntax |= RE_HAT_LISTS_NOT_NEWLINE; + /* It also changes the matching behavior. */ + preg->newline_anchor = 1; + } + else + preg->newline_anchor = 0; + + preg->no_sub = !!(cflags & REG_NOSUB); + + /* POSIX says a null character in the pattern terminates it, so we + can use strlen here in compiling the pattern. */ + ret = regex_compile (pattern, strlen (pattern), syntax, preg); + + /* POSIX doesn't distinguish between an unmatched open-group and an + unmatched close-group: both are REG_EPAREN. */ + if (ret == REG_ERPAREN) ret = REG_EPAREN; + + if (ret == REG_NOERROR && preg->fastmap) + { + /* Compute the fastmap now, since regexec cannot modify the pattern + buffer. */ + if (re_compile_fastmap (preg) == -2) + { + /* Some error occured while computing the fastmap, just forget + about it. */ + free (preg->fastmap); + preg->fastmap = NULL; + } + } + + return (int) ret; +} +#ifdef _LIBC +weak_alias (__regcomp, regcomp) +#endif + + +/* regexec searches for a given pattern, specified by PREG, in the + string STRING. + + If NMATCH is zero or REG_NOSUB was set in the cflags argument to + `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at + least NMATCH elements, and we set them to the offsets of the + corresponding matched substrings. + + EFLAGS specifies `execution flags' which affect matching: if + REG_NOTBOL is set, then ^ does not match at the beginning of the + string; if REG_NOTEOL is set, then $ does not match at the end. + + We return 0 if we find a match and REG_NOMATCH if not. */ + +int +regexec (preg, string, nmatch, pmatch, eflags) + const regex_t *preg; + const char *string; + size_t nmatch; + regmatch_t pmatch[]; + int eflags; +{ + int ret; + struct re_registers regs; + regex_t private_preg; + int len = strlen (string); + boolean want_reg_info = !preg->no_sub && nmatch > 0; + + private_preg = *preg; + + private_preg.not_bol = !!(eflags & REG_NOTBOL); + private_preg.not_eol = !!(eflags & REG_NOTEOL); + + /* The user has told us exactly how many registers to return + information about, via `nmatch'. We have to pass that on to the + matching routines. */ + private_preg.regs_allocated = REGS_FIXED; + + if (want_reg_info) + { + regs.num_regs = nmatch; + regs.start = TALLOC (nmatch * 2, regoff_t); + if (regs.start == NULL) + return (int) REG_NOMATCH; + regs.end = regs.start + nmatch; + } + + /* Perform the searching operation. */ + ret = re_search (&private_preg, string, len, + /* start: */ 0, /* range: */ len, + want_reg_info ? ®s : (struct re_registers *) 0); + + /* Copy the register information to the POSIX structure. */ + if (want_reg_info) + { + if (ret >= 0) + { + unsigned r; + + for (r = 0; r < nmatch; r++) + { + pmatch[r].rm_so = regs.start[r]; + pmatch[r].rm_eo = regs.end[r]; + } + } + + /* If we needed the temporary register info, free the space now. */ + free (regs.start); + } + + /* We want zero return to mean success, unlike `re_search'. */ + return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH; +} +#ifdef _LIBC +weak_alias (__regexec, regexec) +#endif + + +/* Returns a message corresponding to an error code, ERRCODE, returned + from either regcomp or regexec. We don't use PREG here. */ + +size_t +regerror (errcode, preg, errbuf, errbuf_size) + int errcode; + const regex_t *preg; + char *errbuf; + size_t errbuf_size; +{ + const char *msg; + size_t msg_size; + + if (errcode < 0 + || errcode >= (int) (sizeof (re_error_msgid_idx) + / sizeof (re_error_msgid_idx[0]))) + /* Only error codes returned by the rest of the code should be passed + to this routine. If we are given anything else, or if other regex + code generates an invalid error code, then the program has a bug. + Dump core so we can fix it. */ + abort (); + + msg = gettext (re_error_msgid + re_error_msgid_idx[errcode]); + + msg_size = strlen (msg) + 1; /* Includes the null. */ + + if (errbuf_size != 0) + { + if (msg_size > errbuf_size) + { +#if defined HAVE_MEMPCPY || defined _LIBC + *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0'; +#else + memcpy (errbuf, msg, errbuf_size - 1); + errbuf[errbuf_size - 1] = 0; +#endif + } + else + memcpy (errbuf, msg, msg_size); + } + + return msg_size; +} +#ifdef _LIBC +weak_alias (__regerror, regerror) +#endif + + +/* Free dynamically allocated space used by PREG. */ + +void +regfree (preg) + regex_t *preg; +{ + if (preg->buffer != NULL) + free (preg->buffer); + preg->buffer = NULL; + + preg->allocated = 0; + preg->used = 0; + + if (preg->fastmap != NULL) + free (preg->fastmap); + preg->fastmap = NULL; + preg->fastmap_accurate = 0; + + if (preg->translate != NULL) + free (preg->translate); + preg->translate = NULL; +} +#ifdef _LIBC +weak_alias (__regfree, regfree) +#endif + +#endif /* not emacs */ diff --git a/src/gnuregex.h b/src/gnuregex.h new file mode 100644 index 0000000..d88ab92 --- /dev/null +++ b/src/gnuregex.h @@ -0,0 +1,542 @@ +/* Definitions for data structures and routines for the regular + expression library, version 0.12. + Copyright (C) 1985,89,90,91,92,93,95,96,97,98 Free Software Foundation, Inc. + + This file is part of the GNU C Library. Its master source is NOT part of + the C library, however. The master source lives in /gd/gnu/lib. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 of the + License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +#ifndef _REGEX_H +#define _REGEX_H 1 + +/* Allow the use in C++ code. */ +#ifdef __cplusplus +extern "C" { +#endif + +/* POSIX says that <sys/types.h> must be included (by the caller) before + <regex.h>. */ + +#if !defined _POSIX_C_SOURCE && !defined _POSIX_SOURCE && defined VMS +/* VMS doesn't have `size_t' in <sys/types.h>, even though POSIX says it + should be there. */ +# include <stddef.h> +#endif + +/* The following two types have to be signed and unsigned integer type + wide enough to hold a value of a pointer. For most ANSI compilers + ptrdiff_t and size_t should be likely OK. Still size of these two + types is 2 for Microsoft C. Ugh... */ +typedef long int s_reg_t; +typedef unsigned long int active_reg_t; + +/* The following bits are used to determine the regexp syntax we + recognize. The set/not-set meanings are chosen so that Emacs syntax + remains the value 0. The bits are given in alphabetical order, and + the definitions shifted by one from the previous bit; thus, when we + add or remove a bit, only one other definition need change. */ +typedef unsigned long int reg_syntax_t; + +/* If this bit is not set, then \ inside a bracket expression is literal. + If set, then such a \ quotes the following character. */ +#define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1) + +/* If this bit is not set, then + and ? are operators, and \+ and \? are + literals. + If set, then \+ and \? are operators and + and ? are literals. */ +#define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1) + +/* If this bit is set, then character classes are supported. They are: + [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:], + [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:]. + If not set, then character classes are not supported. */ +#define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1) + +/* If this bit is set, then ^ and $ are always anchors (outside bracket + expressions, of course). + If this bit is not set, then it depends: + ^ is an anchor if it is at the beginning of a regular + expression or after an open-group or an alternation operator; + $ is an anchor if it is at the end of a regular expression, or + before a close-group or an alternation operator. + + This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because + POSIX draft 11.2 says that * etc. in leading positions is undefined. + We already implemented a previous draft which made those constructs + invalid, though, so we haven't changed the code back. */ +#define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1) + +/* If this bit is set, then special characters are always special + regardless of where they are in the pattern. + If this bit is not set, then special characters are special only in + some contexts; otherwise they are ordinary. Specifically, + * + ? and intervals are only special when not after the beginning, + open-group, or alternation operator. */ +#define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1) + +/* If this bit is set, then *, +, ?, and { cannot be first in an re or + immediately after an alternation or begin-group operator. */ +#define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1) + +/* If this bit is set, then . matches newline. + If not set, then it doesn't. */ +#define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1) + +/* If this bit is set, then . doesn't match NUL. + If not set, then it does. */ +#define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1) + +/* If this bit is set, nonmatching lists [^...] do not match newline. + If not set, they do. */ +#define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1) + +/* If this bit is set, either \{...\} or {...} defines an + interval, depending on RE_NO_BK_BRACES. + If not set, \{, \}, {, and } are literals. */ +#define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1) + +/* If this bit is set, +, ? and | aren't recognized as operators. + If not set, they are. */ +#define RE_LIMITED_OPS (RE_INTERVALS << 1) + +/* If this bit is set, newline is an alternation operator. + If not set, newline is literal. */ +#define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1) + +/* If this bit is set, then `{...}' defines an interval, and \{ and \} + are literals. + If not set, then `\{...\}' defines an interval. */ +#define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1) + +/* If this bit is set, (...) defines a group, and \( and \) are literals. + If not set, \(...\) defines a group, and ( and ) are literals. */ +#define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1) + +/* If this bit is set, then \<digit> matches <digit>. + If not set, then \<digit> is a back-reference. */ +#define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1) + +/* If this bit is set, then | is an alternation operator, and \| is literal. + If not set, then \| is an alternation operator, and | is literal. */ +#define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1) + +/* If this bit is set, then an ending range point collating higher + than the starting range point, as in [z-a], is invalid. + If not set, then when ending range point collates higher than the + starting range point, the range is ignored. */ +#define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1) + +/* If this bit is set, then an unmatched ) is ordinary. + If not set, then an unmatched ) is invalid. */ +#define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1) + +/* If this bit is set, succeed as soon as we match the whole pattern, + without further backtracking. */ +#define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1) + +/* If this bit is set, do not process the GNU regex operators. + If not set, then the GNU regex operators are recognized. */ +#define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1) + +/* If this bit is set, turn on internal regex debugging. + If not set, and debugging was on, turn it off. + This only works if regex.c is compiled -DDEBUG. + We define this bit always, so that all that's needed to turn on + debugging is to recompile regex.c; the calling code can always have + this bit set, and it won't affect anything in the normal case. */ +#define RE_DEBUG (RE_NO_GNU_OPS << 1) + +/* This global variable defines the particular regexp syntax to use (for + some interfaces). When a regexp is compiled, the syntax used is + stored in the pattern buffer, so changing this does not affect + already-compiled regexps. */ +extern reg_syntax_t re_syntax_options; + +/* Define combinations of the above bits for the standard possibilities. + (The [[[ comments delimit what gets put into the Texinfo file, so + don't delete them!) */ +/* [[[begin syntaxes]]] */ +#define RE_SYNTAX_EMACS 0 + +#define RE_SYNTAX_AWK \ + (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \ + | RE_NO_BK_PARENS | RE_NO_BK_REFS \ + | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \ + | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \ + | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS) + +#define RE_SYNTAX_GNU_AWK \ + ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \ + & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS)) + +#define RE_SYNTAX_POSIX_AWK \ + (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \ + | RE_INTERVALS | RE_NO_GNU_OPS) + +#define RE_SYNTAX_GREP \ + (RE_BK_PLUS_QM | RE_CHAR_CLASSES \ + | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \ + | RE_NEWLINE_ALT) + +#define RE_SYNTAX_EGREP \ + (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \ + | RE_NEWLINE_ALT | RE_NO_BK_PARENS \ + | RE_NO_BK_VBAR) + +#define RE_SYNTAX_POSIX_EGREP \ + (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES) + +/* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */ +#define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC + +#define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC + +/* Syntax bits common to both basic and extended POSIX regex syntax. */ +#define _RE_SYNTAX_POSIX_COMMON \ + (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \ + | RE_INTERVALS | RE_NO_EMPTY_RANGES) + +#define RE_SYNTAX_POSIX_BASIC \ + (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM) + +/* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes + RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this + isn't minimal, since other operators, such as \`, aren't disabled. */ +#define RE_SYNTAX_POSIX_MINIMAL_BASIC \ + (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS) + +#define RE_SYNTAX_POSIX_EXTENDED \ + (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \ + | RE_NO_BK_PARENS | RE_NO_BK_VBAR \ + | RE_UNMATCHED_RIGHT_PAREN_ORD) + +/* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INVALID_OPS + replaces RE_CONTEXT_INDEP_OPS and RE_NO_BK_REFS is added. */ +#define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \ + (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \ + | RE_NO_BK_PARENS | RE_NO_BK_REFS \ + | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD) +/* [[[end syntaxes]]] */ + +/* Maximum number of duplicates an interval can allow. Some systems + (erroneously) define this in other header files, but we want our + value, so remove any previous define. */ +#ifdef RE_DUP_MAX +# undef RE_DUP_MAX +#endif +/* If sizeof(int) == 2, then ((1 << 15) - 1) overflows. */ +#define RE_DUP_MAX (0x7fff) + + +/* POSIX `cflags' bits (i.e., information for `regcomp'). */ + +/* If this bit is set, then use extended regular expression syntax. + If not set, then use basic regular expression syntax. */ +#define REG_EXTENDED 1 + +/* If this bit is set, then ignore case when matching. + If not set, then case is significant. */ +#define REG_ICASE (REG_EXTENDED << 1) + +/* If this bit is set, then anchors do not match at newline + characters in the string. + If not set, then anchors do match at newlines. */ +#define REG_NEWLINE (REG_ICASE << 1) + +/* If this bit is set, then report only success or fail in regexec. + If not set, then returns differ between not matching and errors. */ +#define REG_NOSUB (REG_NEWLINE << 1) + + +/* POSIX `eflags' bits (i.e., information for regexec). */ + +/* If this bit is set, then the beginning-of-line operator doesn't match + the beginning of the string (presumably because it's not the + beginning of a line). + If not set, then the beginning-of-line operator does match the + beginning of the string. */ +#define REG_NOTBOL 1 + +/* Like REG_NOTBOL, except for the end-of-line. */ +#define REG_NOTEOL (1 << 1) + + +/* If any error codes are removed, changed, or added, update the + `re_error_msg' table in regex.c. */ +typedef enum +{ +#ifdef _XOPEN_SOURCE + REG_ENOSYS = -1, /* This will never happen for this implementation. */ +#endif + + REG_NOERROR = 0, /* Success. */ + REG_NOMATCH, /* Didn't find a match (for regexec). */ + + /* POSIX regcomp return error codes. (In the order listed in the + standard.) */ + REG_BADPAT, /* Invalid pattern. */ + REG_ECOLLATE, /* Not implemented. */ + REG_ECTYPE, /* Invalid character class name. */ + REG_EESCAPE, /* Trailing backslash. */ + REG_ESUBREG, /* Invalid back reference. */ + REG_EBRACK, /* Unmatched left bracket. */ + REG_EPAREN, /* Parenthesis imbalance. */ + REG_EBRACE, /* Unmatched \{. */ + REG_BADBR, /* Invalid contents of \{\}. */ + REG_ERANGE, /* Invalid range end. */ + REG_ESPACE, /* Ran out of memory. */ + REG_BADRPT, /* No preceding re for repetition op. */ + + /* Error codes we've added. */ + REG_EEND, /* Premature end. */ + REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */ + REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */ +} reg_errcode_t; + +/* This data structure represents a compiled pattern. Before calling + the pattern compiler, the fields `buffer', `allocated', `fastmap', + `translate', and `no_sub' can be set. After the pattern has been + compiled, the `re_nsub' field is available. All other fields are + private to the regex routines. */ + +#ifndef RE_TRANSLATE_TYPE +# define RE_TRANSLATE_TYPE char * +#endif + +struct re_pattern_buffer +{ +/* [[[begin pattern_buffer]]] */ + /* Space that holds the compiled pattern. It is declared as + `unsigned char *' because its elements are + sometimes used as array indexes. */ + unsigned char *buffer; + + /* Number of bytes to which `buffer' points. */ + unsigned long int allocated; + + /* Number of bytes actually used in `buffer'. */ + unsigned long int used; + + /* Syntax setting with which the pattern was compiled. */ + reg_syntax_t syntax; + + /* Pointer to a fastmap, if any, otherwise zero. re_search uses + the fastmap, if there is one, to skip over impossible + starting points for matches. */ + char *fastmap; + + /* Either a translate table to apply to all characters before + comparing them, or zero for no translation. The translation + is applied to a pattern when it is compiled and to a string + when it is matched. */ + RE_TRANSLATE_TYPE translate; + + /* Number of subexpressions found by the compiler. */ + size_t re_nsub; + + /* Zero if this pattern cannot match the empty string, one else. + Well, in truth it's used only in `re_search_2', to see + whether or not we should use the fastmap, so we don't set + this absolutely perfectly; see `re_compile_fastmap' (the + `duplicate' case). */ + unsigned can_be_null : 1; + + /* If REGS_UNALLOCATED, allocate space in the `regs' structure + for `max (RE_NREGS, re_nsub + 1)' groups. + If REGS_REALLOCATE, reallocate space if necessary. + If REGS_FIXED, use what's there. */ +#define REGS_UNALLOCATED 0 +#define REGS_REALLOCATE 1 +#define REGS_FIXED 2 + unsigned regs_allocated : 2; + + /* Set to zero when `regex_compile' compiles a pattern; set to one + by `re_compile_fastmap' if it updates the fastmap. */ + unsigned fastmap_accurate : 1; + + /* If set, `re_match_2' does not return information about + subexpressions. */ + unsigned no_sub : 1; + + /* If set, a beginning-of-line anchor doesn't match at the + beginning of the string. */ + unsigned not_bol : 1; + + /* Similarly for an end-of-line anchor. */ + unsigned not_eol : 1; + + /* If true, an anchor at a newline matches. */ + unsigned newline_anchor : 1; + +/* [[[end pattern_buffer]]] */ +}; + +typedef struct re_pattern_buffer regex_t; + +/* Type for byte offsets within the string. POSIX mandates this. */ +typedef int regoff_t; + + +/* This is the structure we store register match data in. See + regex.texinfo for a full description of what registers match. */ +struct re_registers +{ + unsigned num_regs; + regoff_t *start; + regoff_t *end; +}; + + +/* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer, + `re_match_2' returns information about at least this many registers + the first time a `regs' structure is passed. */ +#ifndef RE_NREGS +# define RE_NREGS 30 +#endif + + +/* POSIX specification for registers. Aside from the different names than + `re_registers', POSIX uses an array of structures, instead of a + structure of arrays. */ +typedef struct +{ + regoff_t rm_so; /* Byte offset from string's start to substring's start. */ + regoff_t rm_eo; /* Byte offset from string's start to substring's end. */ +} regmatch_t; + +/* Declarations for routines. */ + +/* To avoid duplicating every routine declaration -- once with a + prototype (if we are ANSI), and once without (if we aren't) -- we + use the following macro to declare argument types. This + unfortunately clutters up the declarations a bit, but I think it's + worth it. */ + +#if __STDC__ + +# define _RE_ARGS(args) args + +#else /* not __STDC__ */ + +# define _RE_ARGS(args) () + +#endif /* not __STDC__ */ + +/* Sets the current default syntax to SYNTAX, and return the old syntax. + You can also simply assign to the `re_syntax_options' variable. */ +extern reg_syntax_t re_set_syntax _RE_ARGS ((reg_syntax_t syntax)); + +/* Compile the regular expression PATTERN, with length LENGTH + and syntax given by the global `re_syntax_options', into the buffer + BUFFER. Return NULL if successful, and an error string if not. */ +extern const char *re_compile_pattern + _RE_ARGS ((const char *pattern, size_t length, + struct re_pattern_buffer *buffer)); + + +/* Compile a fastmap for the compiled pattern in BUFFER; used to + accelerate searches. Return 0 if successful and -2 if was an + internal error. */ +extern int re_compile_fastmap _RE_ARGS ((struct re_pattern_buffer *buffer)); + + +/* Search in the string STRING (with length LENGTH) for the pattern + compiled into BUFFER. Start searching at position START, for RANGE + characters. Return the starting position of the match, -1 for no + match, or -2 for an internal error. Also return register + information in REGS (if REGS and BUFFER->no_sub are nonzero). */ +extern int re_search + _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string, + int length, int start, int range, struct re_registers *regs)); + + +/* Like `re_search', but search in the concatenation of STRING1 and + STRING2. Also, stop searching at index START + STOP. */ +extern int re_search_2 + _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1, + int length1, const char *string2, int length2, + int start, int range, struct re_registers *regs, int stop)); + + +/* Like `re_search', but return how many characters in STRING the regexp + in BUFFER matched, starting at position START. */ +extern int re_match + _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string, + int length, int start, struct re_registers *regs)); + + +/* Relates to `re_match' as `re_search_2' relates to `re_search'. */ +extern int re_match_2 + _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1, + int length1, const char *string2, int length2, + int start, struct re_registers *regs, int stop)); + + +/* Set REGS to hold NUM_REGS registers, storing them in STARTS and + ENDS. Subsequent matches using BUFFER and REGS will use this memory + for recording register information. STARTS and ENDS must be + allocated with malloc, and must each be at least `NUM_REGS * sizeof + (regoff_t)' bytes long. + + If NUM_REGS == 0, then subsequent matches should allocate their own + register data. + + Unless this function is called, the first search or match using + PATTERN_BUFFER will allocate its own register data, without + freeing the old data. */ +extern void re_set_registers + _RE_ARGS ((struct re_pattern_buffer *buffer, struct re_registers *regs, + unsigned num_regs, regoff_t *starts, regoff_t *ends)); + +#if defined _REGEX_RE_COMP || defined _LIBC +# ifndef _CRAY +/* 4.2 bsd compatibility. */ +extern char *re_comp _RE_ARGS ((const char *)); +extern int re_exec _RE_ARGS ((const char *)); +# endif +#endif + +/* POSIX compatibility. */ +extern int regcomp _RE_ARGS ((regex_t *__preg, const char *__pattern, + int __cflags)); + +extern int regexec _RE_ARGS ((const regex_t *__preg, + const char *__string, size_t __nmatch, + regmatch_t __pmatch[], int __eflags)); + +extern size_t regerror _RE_ARGS ((int __errcode, const regex_t *__preg, + char *__errbuf, size_t __errbuf_size)); + +extern void regfree _RE_ARGS ((regex_t *__preg)); + + +#ifdef __cplusplus +} +#endif /* C++ */ + +#endif /* regex.h */ + +/* +Local variables: +make-backup-files: t +version-control: t +trim-versions-without-asking: nil +End: +*/ diff --git a/src/log.c b/src/log.c new file mode 100644 index 0000000..ca64f8e --- /dev/null +++ b/src/log.c @@ -0,0 +1,90 @@ +/* $Id: log.c,v 1.1.1.1 2000-02-16 17:32:22 sdyoung Exp $ + * + * Logs the various messages which tinyproxy produces to either a log file or + * the syslog daemon. Not much to it... + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * log.c - For the manipulation of log files. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <stdio.h> +#include <stdarg.h> +#include <time.h> +#include <syslog.h> +#include <stdlib.h> +#include <unistd.h> +#include <assert.h> + +#include "tinyproxy.h" +#include "log.h" + +#define LENGTH 16 + +/* + * This routine logs messages to either the log file or the syslog function. + */ +void log(char *fmt, ...) +{ + va_list args; + time_t nowtime; + FILE *cf; + +#if defined(HAVE_SYSLOG_H) && !defined(HAVE_VSYSLOG_H) + static char str[800]; +#endif + static char time_string[LENGTH]; + + assert(fmt); + + va_start(args, fmt); + +#ifdef HAVE_SYSLOG_H + if (config.syslog == FALSE) { +#endif + nowtime = time(NULL); + /* Format is month day hour:minute:second (24 time) */ + strftime(time_string, LENGTH, "%b %d %H:%M:%S", + localtime(&nowtime)); + + if (!(cf = config.logf)) + cf = stderr; + + fprintf(cf, "%s [%d]: ", time_string, getpid()); + vfprintf(cf, fmt, args); + fprintf(cf, "\n"); + fflush(cf); +#ifdef HAVE_SYSLOG_H + } else { +# ifdef HAVE_VSYSLOG_H + vsyslog(LOG_INFO, fmt, args); +# else +# ifdef HAVE_VSNPRINTF + vsnprintf(str, 800, fmt, args); +# else +# ifdef HAVE_VPRINTF + vsprintf(str, fmt, args); +# endif +# endif + syslog(LOG_INFO, str); +# endif + } +#endif + + va_end(args); +} diff --git a/src/log.h b/src/log.h new file mode 100644 index 0000000..e928b4c --- /dev/null +++ b/src/log.h @@ -0,0 +1,24 @@ +/* $Id: log.h,v 1.1.1.1 2000-02-16 17:32:22 sdyoung Exp $ + * + * See 'log.c' for a detailed description. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef _LOG_H_ +#define _LOG_H_ 1 + +extern void log(char *fmt, ...); + +#endif diff --git a/src/regexp.h b/src/regexp.h new file mode 100644 index 0000000..1c40332 --- /dev/null +++ b/src/regexp.h @@ -0,0 +1,24 @@ +/* $Id: regexp.h,v 1.1.1.1 2000-02-16 17:32:24 sdyoung Exp $ + * + * We need this little header to help distinguish whether to use the REGEX + * library installed in the system, or to include our own version (the GNU + * version to be exact.) + * + * Copyright (C) 2000 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef USE_GNU_REGEX +# include "gnuregex.h" +#else +# include <regex.h> +#endif diff --git a/src/reqs.c b/src/reqs.c new file mode 100644 index 0000000..7d6aa9f --- /dev/null +++ b/src/reqs.c @@ -0,0 +1,832 @@ +/* $Id: reqs.c,v 1.1.1.1 2000-02-16 17:32:23 sdyoung Exp $ + * + * This is where all the work in tinyproxy is actually done. Incoming + * connections are added to the active list of connections and then the header + * is processed to determine what is being requested. tinyproxy then connects + * to the remote server and begins to relay the bytes back and forth between + * the client and the remote server. Basically, we sit there and sling bytes + * back and forth. Optionally, we can weed out headers we do not want to send, + * and also add a header of our own. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * Copyright (C) 2000 Chris Lightfoot (chris@ex-parrot.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <stdio.h> +#include <stdlib.h> +#include <time.h> +#include <sys/time.h> +#include <sys/types.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <arpa/inet.h> +#include <netdb.h> +#include <sys/uio.h> +#include <unistd.h> +#include <string.h> +#include <errno.h> +#include <sysexits.h> +#include <assert.h> + +#include "config.h" +#include "tinyproxy.h" +#include "sock.h" +#include "utils.h" +#include "conns.h" +#include "log.h" +#include "reqs.h" +#include "buffer.h" +#include "filter.h" +#include "uri.h" +#include "regexp.h" + +/* chris - for asynchronous DNS */ +#include "dnscache.h" +#include <adns.h> +extern adns_state adns; + +#ifdef XTINYPROXY + +static void Add_XTinyproxy_Header(struct conn_s *connptr) +{ + char *header_line; + char ipaddr[PEER_IP_LENGTH]; + int length; + + assert(connptr); + + if (!(header_line = xmalloc(sizeof(char) * 32))) + return; + + length = sprintf(header_line, "X-Tinyproxy: %s\r\n", + getpeer_ip(connptr->client_fd, ipaddr)); + + unshift_buffer(connptr->cbuffer, header_line, length); +} + +#endif + +#define HTTPPATTERN "^([a-z]+)[ \t]+([^ \t]+)([ \t]+(HTTP/[0-9]+\\.[0-9]+))?" +#define NMATCH 4 +#define METHOD_IND 1 +#define URI_IND 2 +#define VERSION_MARK 3 +#define VERSION_IND 4 + +#define HTTP400ERROR "Unrecognizable request. Only HTTP is allowed." +#define HTTP500ERROR "Unable to connect to remote server." +#define HTTP503ERROR "Internal server error." + +/* + * Parse a client HTTP request and then establish connection. + */ +static int clientreq(struct conn_s *connptr) +{ + URI *uri = NULL; + char *inbuf, *buffer, *request, *port; + char *inbuf_ptr; + + regex_t preg; + regmatch_t pmatch[NMATCH]; + + long len; + int fd, port_no; + + char peer_ipaddr[PEER_IP_LENGTH]; + + assert(connptr); + + getpeer_ip(connptr->client_fd, peer_ipaddr); + /* chris - getpeer_string could block, so for the moment take it out */ + + if ((len + = readline(connptr->client_fd, connptr->cbuffer, &inbuf)) <= 0) { + return len; + } + + inbuf_ptr = inbuf + len - 1; + while (*inbuf_ptr == '\r' || *inbuf_ptr == '\n') + *inbuf_ptr-- = '\0'; + + /* Log the incoming connection */ + if (!config.restricted) { + log("Connect: %s", peer_ipaddr); + log("Request: %s", inbuf); + } + + if (regcomp(&preg, HTTPPATTERN, REG_EXTENDED | REG_ICASE) != 0) { + log("ERROR clientreq: regcomp"); + return 0; + } + if (regexec(&preg, inbuf, NMATCH, pmatch, 0) != 0) { + log("ERROR clientreq: regexec"); + regfree(&preg); + return -1; + } + regfree(&preg); + + if (pmatch[VERSION_MARK].rm_so == -1) + connptr->simple_req = TRUE; + + if (pmatch[METHOD_IND].rm_so == -1 || pmatch[URI_IND].rm_so == -1) { + log("ERROR clientreq: Incomplete line from %s (%s)", + peer_ipaddr, inbuf); + httperr(connptr, 400, HTTP400ERROR); + goto COMMON_EXIT; + } + + len = pmatch[URI_IND].rm_eo - pmatch[URI_IND].rm_so; + if (!(buffer = xmalloc(len + 1))) { + log("ERROR clientreq: Cannot allocate buffer for request from %s", + peer_ipaddr); + httperr(connptr, 503, HTTP503ERROR); + goto COMMON_EXIT; + } + memcpy(buffer, inbuf + pmatch[URI_IND].rm_so, len); + buffer[len] = '\0'; + if (!(uri = explode_uri(buffer))) { + safefree(buffer); + log("ERROR clientreq: Problem with explode_uri"); + httperr(connptr, 503, HTTP503ERROR); + goto COMMON_EXIT; + } + safefree(buffer); + + if (strcasecmp(uri->scheme, "http") != 0) { + char *error_string = xmalloc(strlen(uri->scheme) + 64); + sprintf(error_string, "Invalid scheme (%s). Only HTTP is allowed.", + uri->scheme); + httperr(connptr, 400, error_string); + safefree(error_string); + goto COMMON_EXIT; + } + + if ((strlen(config.stathost) > 0) && + strcasecmp(uri->authority, config.stathost) == 0) { + showstats(connptr); + goto COMMON_EXIT; + } + + port_no = 80; + if ((port = strchr(uri->authority, ':'))) { + *port++ = '\0'; + if (strlen(port) > 0) + port_no = atoi(port); + } + + /* chris - so this can be passed on to clientreq_dnscomplete. */ + connptr->port_no = port_no; + +#ifdef FILTER_ENABLE + /* Filter domains out */ + if (config.filter) { + if (filter_host(uri->authority)) { + log("ERROR clientreq: Filtered connection (%s)", + peer_ipaddr); + httperr(connptr, 404, + "Unable to connect to filtered host."); + goto COMMON_EXIT; + } + } +#endif /* FILTER_ENABLE */ + + /* Build a new request from the first line of the header */ + if (!(request = xmalloc(strlen(inbuf) + 1))) { + log("ERROR clientreq: cannot allocate buffer for request from %s", + peer_ipaddr); + httperr(connptr, 503, HTTP503ERROR); + goto COMMON_EXIT; + } + + /* We need to set the version number WE support */ + memcpy(request, inbuf, pmatch[METHOD_IND].rm_eo); + request[pmatch[METHOD_IND].rm_eo] = '\0'; + strcat(request, " "); + strcat(request, uri->path); + if (uri->query) { + strcat(request, "?"); + strcat(request, uri->query); + } + strcat(request, " HTTP/1.0\r\n"); + + /* chris - If domain is in dotted-quad format or is already in the + * DNS cache, then go straight to WAITCONN. + */ + if (inet_aton(uri->authority, NULL) || lookup(NULL, uri->authority) == 0) { + if ((fd = opensock(uri->authority, port_no)) < 0) { + safefree(request); + httperr(connptr, 500, + "Unable to connect to host (cannot create sock)"); + stats.num_badcons++; + goto COMMON_EXIT; + } + + connptr->server_fd = fd; + connptr->type = WAITCONN; + } + /* Otherwise submit a DNS request and hope for the best. */ + else { + if (adns_submit(adns, uri->authority, adns_r_a, adns_qf_quoteok_cname | adns_qf_cname_loose, connptr, &(connptr->adns_qu))) { + safefree(request); + httperr(connptr, 500, "Resolver error connecting to host"); + stats.num_badcons++; + goto COMMON_EXIT; + } else { +#ifdef __DEBUG__ + log("DNS request submitted"); +#endif + /* copy domain for later caching */ + connptr->domain = strdup(uri->authority); + connptr->type = DNS_WAITCONN; + } + } + +#ifdef XTINYPROXY + /* Add a X-Tinyproxy header which contains our IP address */ + if (config.my_domain + && xstrstr(uri->authority, config.my_domain, + strlen(uri->authority), FALSE)) { + Add_XTinyproxy_Header(connptr); + } +#endif + + /* Add the rewritten request to the buffer */ + unshift_buffer(connptr->cbuffer, request, strlen(request)); + + /* + * HACK HACK HACK: When we're sending a POST there is no restriction + * on the length of the header. If we don't let all the header lines + * through, the POST will not work. This _definitely_ needs to be + * fixed. - rjkaes + */ + if (!xstrstr(inbuf, "POST ", 5, FALSE)) { + connptr->clientheader = TRUE; + } + + COMMON_EXIT: + safefree(inbuf); + free_uri(uri); + return 0; +} + +/* chris - added this to move a connection from the DNS_WAITCONN state + * to the WAITCONN state by connecting it to the server, once the name + * has been resolved. + */ +static int clientreq_dnscomplete(struct conn_s *connptr, struct in_addr *inaddr) { + int fd; + + fd = opensock_inaddr(inaddr, connptr->port_no); + + if (fd < 0) { +#ifdef __DEBUG__ + log("Failed to open connection to server"); +#endif + httperr(connptr, 500, + "Unable to connect to host (cannot create sock)"); + stats.num_badcons++; + + return 0; + } else { +#ifdef __DEBUG__ + log("Connected to server"); +#endif + connptr->server_fd = fd; + connptr->type = WAITCONN; + } + + return 0; +} + +/* + * Finish the client request + */ +static int clientreq_finish(struct conn_s *connptr) +{ + int sockopt, len = sizeof(sockopt); + + assert(connptr); + + if (getsockopt + (connptr->server_fd, SOL_SOCKET, SO_ERROR, &sockopt, &len) < 0) { + log("ERROR clientreq_finish: getsockopt error (%s)", + strerror(errno)); + return -1; + } + + if (sockopt != 0) { + if (sockopt == EINPROGRESS) + return 0; + else if (sockopt == ETIMEDOUT) { + httperr(connptr, 408, "Connect Timed Out"); + return 0; + } else { + log("ERROR clientreq_finish: could not create connection (%s)", + strerror(sockopt)); + return -1; + } + } + + connptr->type = RELAYCONN; + stats.num_opens++; + return 0; +} + +/* + * Check to see if the line is allowed or not depending on the anonymous + * headers which are to be allowed. + */ +static int anonheader(char *line) +{ + struct allowedhdr_s *allowedptr = allowedhdrs; + + assert(line); + assert(allowedhdrs); + + if (!xstrstr(line, "GET ", 4, FALSE) + || !xstrstr(line, "POST ", 5, FALSE) + || !xstrstr(line, "HEAD ", 5, FALSE)) + return 1; + + for (allowedptr = allowedhdrs; allowedptr; + allowedptr = allowedptr->next) { + if (!strncasecmp + (line, allowedptr->hdrname, strlen(allowedptr->hdrname))) { + return 1; + } + } + return 0; +} + +/* + * Used to read in the lines from the header (client side) when we're doing + * the anonymous header reduction. + */ +static int readanonconn(struct conn_s *connptr) +{ + char *line = NULL; + int retv; + + assert(connptr); + + if ((retv = readline(connptr->client_fd, connptr->cbuffer, &line)) <= + 0) { + return retv; + } + + if ((line[0] == '\n') || (strncmp(line, "\r\n", 2) == 0)) { + connptr->clientheader = TRUE; + } else if (!anonheader(line)) { + safefree(line); + return 0; + } + + push_buffer(connptr->cbuffer, line, strlen(line)); + return 0; +} + +/* + * Read in the bytes from the socket + */ +static int readconn(int fd, struct buffer_s *buffptr) +{ + int bytesin; + + assert(fd >= 0); + assert(buffptr); + + if ((bytesin = readbuff(fd, buffptr)) < 0) { + return -1; + } +#ifdef __DEBUG__ + log("readconn [%d]: %d", fd, bytesin); +#endif + + stats.num_rx += bytesin; + return bytesin; +} + +/* + * Write the bytes from the buffer to the socket + */ +static int writeconn(int fd, struct buffer_s *buffptr) +{ + int bytessent; + + assert(fd >= 0); + assert(buffptr); + + if ((bytessent = writebuff(fd, buffptr)) < 0) { + return -1; + } + + stats.num_tx += bytessent; + return bytessent; +} + +/* + * Factored out the common function to read from the client. It was used in + * two different places with no change, so no point in having the same code + * twice. + */ +static int read_from_client(struct conn_s *connptr, fd_set * readfds) +{ + assert(connptr); + assert(readfds); + + if (FD_ISSET(connptr->client_fd, readfds)) { + if (config.anonymous && !connptr->clientheader) { + if (readanonconn(connptr) < 0) { + shutdown(connptr->client_fd, 2); + shutdown(connptr->server_fd, 2); + connptr->type = FINISHCONN; + return -1; + } + } else if (readconn(connptr->client_fd, connptr->cbuffer) < 0) { + shutdown(connptr->client_fd, 2); + shutdown(connptr->server_fd, 2); + connptr->type = FINISHCONN; + return -1; + } + connptr->actiontime = time(NULL); + } + + return 0; +} + +/* + * Factored out the common write to client function since, again, it was used + * in two different places with no changes. + */ +static int write_to_client(struct conn_s *connptr, fd_set * writefds) +{ + assert(connptr); + assert(writefds); + + if (FD_ISSET(connptr->client_fd, writefds)) { + if (writeconn(connptr->client_fd, connptr->sbuffer) < 0) { + shutdown(connptr->client_fd, 2); + shutdown(connptr->server_fd, 2); + connptr->type = FINISHCONN; + return -1; + } + connptr->actiontime = time(NULL); + } + + return 0; +} + +/* + * All of the *_req functions handle the various stages a connection can go + * through. I moved them out from getreqs because they handle a lot of error + * code and it was indenting too far in. As you can see they are very simple, + * and are only called once from getreqs, hence the inlining. + */ + +inline static void newconn_req(struct conn_s *connptr, fd_set * readfds) +{ + assert(connptr); + assert(readfds); + + if (FD_ISSET(connptr->client_fd, readfds)) { + if (clientreq(connptr) < 0) { + shutdown(connptr->client_fd, 2); + connptr->type = FINISHCONN; + return; + } + + if (!connptr) + abort(); + + connptr->actiontime = time(NULL); + } +} + +inline static void waitconn_req(struct conn_s *connptr, fd_set * readfds, + fd_set * writefds) +{ + assert(connptr); + assert(readfds); + assert(writefds); + + if (read_from_client(connptr, readfds) < 0) + return; + + if (FD_ISSET(connptr->server_fd, readfds) + || FD_ISSET(connptr->server_fd, writefds)) { + if (clientreq_finish(connptr) < 0) { + shutdown(connptr->server_fd, 2); + shutdown(connptr->client_fd, 2); + connptr->type = FINISHCONN; + return; + } + connptr->actiontime = time(NULL); + } +} + +inline static void relayconn_req(struct conn_s *connptr, fd_set * readfds, + fd_set * writefds) +{ + assert(connptr); + assert(readfds); + assert(writefds); + + if (read_from_client(connptr, readfds) < 0) + return; + + if (FD_ISSET(connptr->server_fd, readfds)) { + if (connptr->serverheader) { + if (readconn(connptr->server_fd, connptr->sbuffer) < 0) { + shutdown(connptr->server_fd, 2); + connptr->type = CLOSINGCONN; + return; + } + } else { + /* + * We need to read in the first line to rewrite the + * version back down to HTTP/1.0 (if needed) + */ + char *line = NULL, *ptr, *newline; + int retv; + + if ( + (retv = + readline(connptr->server_fd, connptr->sbuffer, + &line)) < 0) { + shutdown(connptr->server_fd, 2); + httperr(connptr, 500, "Server Closed Early"); + return; + } else if (retv == 0) + return; + + connptr->serverheader = TRUE; + + if (strncasecmp(line, "HTTP/1.0", 8)) { + /* Okay, we need to rewrite it then */ + if (!(ptr = strchr(line, ' '))) { + shutdown(connptr->server_fd, 2); + httperr(connptr, 500, + "There was Server Error"); + return; + } + ptr++; + + if (!(newline = xmalloc(strlen(line) + 1))) { + shutdown(connptr->server_fd, 2); + httperr(connptr, 503, + "No Memory Available"); + return; + } + + sprintf(newline, "HTTP/1.0 %s", ptr); + safefree(line); + line = newline; + } + + push_buffer(connptr->sbuffer, line, strlen(line)); + } + connptr->actiontime = time(NULL); + } + + if (write_to_client(connptr, writefds) < 0) + return; + + if (FD_ISSET(connptr->server_fd, writefds)) { + if (writeconn(connptr->server_fd, connptr->cbuffer) < 0) { + shutdown(connptr->server_fd, 2); + connptr->type = CLOSINGCONN; + return; + } + connptr->actiontime = time(NULL); + } +} + +inline static void closingconn_req(struct conn_s *connptr, fd_set * writefds) +{ + assert(connptr); + assert(writefds); + + write_to_client(connptr, writefds); +} + +/* + * Check against the valid subnet to see if we should allow the access + */ +static int validuser(int fd) +{ + char ipaddr[PEER_IP_LENGTH]; + + assert(fd >= 0); + + if (config.subnet == NULL) + return 1; + + if (!strncmp(config.subnet, getpeer_ip(fd, ipaddr), + strlen(config.subnet))) { + return 1; + } else { + return 0; + } +} + +/* + * Loop that checks for new connections, dispatches to the correct + * routines if bytes are pending, checks to see if it's time for a + * garbage collect. + */ +int getreqs(void) +{ + static unsigned int garbc = 0; + fd_set readfds, writefds, exceptfds; /* chris - ADNS expects exceptfds */ + struct conn_s *connptr; + int fd; + struct timeval tv, now; /* chris - for ADNS timeouts */ + + char peer_ipaddr[PEER_IP_LENGTH]; + + if (setup_fd < 0) { + log("ERROR getreqs: setup_fd not a socket"); + return -1; + } + + /* Garbage collect the dead connections and close any idle ones */ + if (garbc++ >= GARBCOLL_INTERVAL) { + garbcoll(); + garbc = 0; + } + conncoll(); + + FD_ZERO(&readfds); + FD_ZERO(&writefds); + FD_SET(setup_fd, &readfds); + + for (connptr = connections; connptr; connptr = connptr->next) { +#ifdef __DEBUG__ + log("Connptr: %p - %d / client %d server %d", connptr, + connptr->type, connptr->client_fd, connptr->server_fd); +#endif + switch (connptr->type) { + case NEWCONN: + if (buffer_size(connptr->cbuffer) < MAXBUFFSIZE) + FD_SET(connptr->client_fd, &readfds); + else { + httperr(connptr, 414, + "Your Request is way too long."); + } + break; + + /* no case here for DNS_WAITCONN */ + + case WAITCONN: + FD_SET(connptr->server_fd, &readfds); + FD_SET(connptr->server_fd, &writefds); + + if (buffer_size(connptr->cbuffer) < MAXBUFFSIZE) + FD_SET(connptr->client_fd, &readfds); + break; + + case RELAYCONN: + if (buffer_size(connptr->sbuffer) > 0) + FD_SET(connptr->client_fd, &writefds); + if (buffer_size(connptr->sbuffer) < MAXBUFFSIZE) + FD_SET(connptr->server_fd, &readfds); + + if (buffer_size(connptr->cbuffer) > 0) + FD_SET(connptr->server_fd, &writefds); + if (buffer_size(connptr->cbuffer) < MAXBUFFSIZE) + FD_SET(connptr->client_fd, &readfds); + + break; + + case CLOSINGCONN: + if (buffer_size(connptr->sbuffer) > 0) + FD_SET(connptr->client_fd, &writefds); + else { + shutdown(connptr->client_fd, 2); + shutdown(connptr->server_fd, 2); + connptr->type = FINISHCONN; + } + break; + + default: + break; + } + } + + /* Set a 60 second time out */ + tv.tv_sec = 1;//60; + tv.tv_usec = 0; + + /* chris - Make ADNS do its stuff, too. */ + { + struct timeval *tv_mod = &tv; + int foo = FD_SETSIZE; + gettimeofday(&now, NULL); + FD_ZERO(&exceptfds); + adns_beforeselect(adns, &foo, &readfds, &writefds, &exceptfds, &tv_mod, NULL, &now); + } + + if (select(FD_SETSIZE, &readfds, &writefds, &exceptfds, &tv) < 0) { +#ifdef __DEBUG__ + log("select error: %s", strerror(errno)); +#endif + return 0; + } + + /* chris - see whether any ADNS lookups have completed */ + gettimeofday(&now, NULL); + adns_afterselect(adns, FD_SETSIZE, &readfds, &writefds, &exceptfds, &now); + + for (connptr = connections; connptr; connptr = connptr->next) { + adns_answer *ans; + + if (connptr->type == DNS_WAITCONN && + adns_check(adns, &(connptr->adns_qu), &ans, (void**)&connptr) == 0) { + + if (ans->status == adns_s_ok) { + if (connptr->domain) { + insert(ans->rrs.inaddr, connptr->domain); + free(connptr->domain); + } + + clientreq_dnscomplete(connptr, ans->rrs.inaddr); + free(ans); + + /* hack */ + FD_SET(connptr->server_fd, &readfds); +#ifdef __DEBUG__ + log("DNS resolution successful"); +#endif + } else { + free(ans); + + httperr(connptr, 500, "Unable to resolve hostname in URL"); +#ifdef __DEBUG__ + log("DNS resolution failed"); +#endif + } + } + } + + /* Check to see if there are new connections pending */ + if (FD_ISSET(setup_fd, &readfds) && (fd = listen_sock()) >= 0) { + new_conn(fd); /* make a connection from the FD */ + + if (validuser(fd)) { + if (config.cutoffload && (load > config.cutoffload)) { + stats.num_refused++; + httperr(connptr, 503, + "tinyproxy is not accepting connections due to high system load"); + } + } else { + httperr(connptr, 403, + "You are not authorized to use the service."); + log("AUTH Rejected connection from %s", + getpeer_ip(fd, peer_ipaddr)); + } + } + + /* + * Loop through the connections and dispatch them to the appropriate + * handler + */ + for (connptr = connections; connptr; connptr = connptr->next) { + switch (connptr->type) { + case NEWCONN: + newconn_req(connptr, &readfds); + break; + + case WAITCONN: + waitconn_req(connptr, &readfds, &writefds); + break; + + case RELAYCONN: + relayconn_req(connptr, &readfds, &writefds); + break; + + case CLOSINGCONN: + closingconn_req(connptr, &writefds); + break; + + default: + break; + } + } + + return 0; +} diff --git a/src/reqs.h b/src/reqs.h new file mode 100644 index 0000000..182fc20 --- /dev/null +++ b/src/reqs.h @@ -0,0 +1,24 @@ +/* $Id: reqs.h,v 1.1.1.1 2000-02-16 17:32:23 sdyoung Exp $ + * + * See 'reqs.c' for a detailed description. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef _REQS_H_ +#define _REQS_H_ 1 + +extern int getreqs(void); + +#endif diff --git a/src/sock.c b/src/sock.c new file mode 100644 index 0000000..dc415e5 --- /dev/null +++ b/src/sock.c @@ -0,0 +1,362 @@ +/* $Id: sock.c,v 1.1.1.1 2000-02-16 17:32:23 sdyoung Exp $ + * + * Sockets are created and destroyed here. When a new connection comes in from + * a client, we need to copy the socket and the create a second socket to the + * remote server the client is trying to connect to. Also, the listening + * socket is created and destroyed here. Sounds more impressive than it + * actually is. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * Copyright (C) 2000 Chris Lightfoot (chris@ex-parrot.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <stdio.h> +#include <stdarg.h> +#include <stdlib.h> +#include <sys/types.h> +#include <sys/socket.h> +#include <netinet/in.h> +#include <errno.h> +#include <netdb.h> +#include <string.h> +#include <unistd.h> +#include <arpa/inet.h> +#include <fcntl.h> +#include <assert.h> + +#include "tinyproxy.h" +#include "sock.h" +#include "log.h" +#include "utils.h" +#include "dnscache.h" + +/* This routine is so old I can't even remember writing it. But I do + * remember that it was an .h file because I didn't know putting code in a + * header was bad magic yet. anyway, this routine opens a connection to a + * system and returns the fd. + */ + +/* + * Cleaned up some of the code to use memory routines which are now the + * default. Also, the routine first checks to see if the address is in + * dotted-decimal form before it does a name lookup. Finally, the newly + * created socket is made non-blocking. + * - rjkaes + */ +int opensock(char *ip_addr, int port) +{ + int sock_fd, flags; + struct sockaddr_in port_info; + + assert(ip_addr); + assert(port > 0); + + memset((struct sockaddr *) &port_info, 0, sizeof(port_info)); + + port_info.sin_family = AF_INET; + + /* chris - Could block; neet to ensure that this is never called + * before a non-blocking DNS query happens for this address. Not + * relevant in the code as it stands. + */ + if (dnscache(&port_info.sin_addr, ip_addr) < 0) { + log("ERROR opensock: Could not lookup address: %s", ip_addr); + return -1; + } + + port_info.sin_port = htons(port); + + if ((sock_fd = socket(AF_INET, SOCK_STREAM, 0)) == -1) { + log("ERROR opensock: socket (%s)", strerror(errno)); + return -1; + } + + flags = fcntl(sock_fd, F_GETFL, 0); + fcntl(sock_fd, F_SETFL, flags | O_NONBLOCK); + + if (connect + (sock_fd, (struct sockaddr *) &port_info, sizeof(port_info)) < 0) { + if (errno != EINPROGRESS) { + log("ERROR opensock: connect (%s)", strerror(errno)); + return -1; + } + } + + return sock_fd; +} + +/* chris - added this to open a socket given a struct in_addr */ +int opensock_inaddr(struct in_addr *inaddr, int port) +{ + int sock_fd, flags; + struct sockaddr_in port_info; + + assert(inaddr); + assert(port > 0); + + memset((struct sockaddr *) &port_info, 0, sizeof(port_info)); + + port_info.sin_family = AF_INET; + + memcpy(&port_info.sin_addr, inaddr, sizeof(struct in_addr)); + + port_info.sin_port = htons(port); + + if ((sock_fd = socket(AF_INET, SOCK_STREAM, 0)) == -1) { + log("ERROR opensock_inaddr: socket (%s)", strerror(errno)); + return -1; + } + + flags = fcntl(sock_fd, F_GETFL, 0); + fcntl(sock_fd, F_SETFL, flags | O_NONBLOCK); + + if (connect + (sock_fd, (struct sockaddr *) &port_info, sizeof(port_info)) < 0) { + if (errno != EINPROGRESS) { + log("ERROR opensock: connect (%s)", strerror(errno)); + return -1; + } + } + + return sock_fd; +} + +int setup_fd; +static struct sockaddr listen_sock_addr; + +/* + * Start listening to a socket. + */ +int init_listen_sock(int port) +{ + struct sockaddr_in laddr; + int i = 1; + + assert(port > 0); + + if ((setup_fd = socket(AF_INET, SOCK_STREAM, 0)) < 0) { + log("ERROR init_listen_sock: socket (%s)", strerror(errno)); + return -1; + } + + if (setsockopt(setup_fd, SOL_SOCKET, SO_REUSEADDR, &i, sizeof(i)) < 0) { + log("ERROR init_listen_sock: setsockopt (%s)", + strerror(errno)); + return -1; + } + + memset(&listen_sock_addr, 0, sizeof(listen_sock_addr)); + memset(&laddr, 0, sizeof(laddr)); + laddr.sin_family = AF_INET; + laddr.sin_port = htons(port); + + if (config.ipAddr) { + laddr.sin_addr.s_addr = inet_addr(config.ipAddr); + } else { + laddr.sin_addr.s_addr = inet_addr("0.0.0.0"); + } + if (bind(setup_fd, (struct sockaddr *) &laddr, sizeof(laddr)) < 0) { + log("ERROR init_listen_sock: bind (%s)", strerror(errno)); + return -1; + } + if ((listen(setup_fd, MAXLISTEN)) != 0) { + log("ERROR init_listen_sock: listen (%s)", strerror(errno)); + return -1; + } + + return 0; +} + +/* + * Grab a pending connection + */ +int listen_sock(void) +{ + static int sock; + int sz = sizeof(listen_sock_addr); + + if ((sock = accept(setup_fd, &listen_sock_addr, &sz)) < 0) { + if (errno != ECONNABORTED +#ifdef EPROTO + && errno != EPROTO +#endif +#ifdef EWOULDBLOCK + && errno != EWOULDBLOCK +#endif + && errno != EINTR) + log("ERROR listen_sock: accept (%s)", strerror(errno)); + return -1; + } + stats.num_listens++; + + return sock; +} + +/* + * Stop listening on a socket. + */ +void de_init_listen_sock(void) +{ + close(setup_fd); +} + +/* + * Takes a socket descriptor and returns the string contain the peer's + * IP address. + */ +char *getpeer_ip(int fd, char *ipaddr) +{ + struct sockaddr_in name; + int namelen = sizeof(name); + + assert(fd >= 0); + assert(ipaddr); + + memset(ipaddr, '\0', PEER_IP_LENGTH); + + if (getpeername(fd, (struct sockaddr *) &name, &namelen) != 0) { + log("ERROR Connect: 'could not get peer name'"); + } else { + strncpy(ipaddr, + inet_ntoa(*(struct in_addr *) &name.sin_addr.s_addr), + PEER_IP_LENGTH - 1); + } + + return ipaddr; +} + +/* + * Takes a socket descriptor and returns the string containing the peer's + * address. + */ +char *getpeer_string(int fd, char *string) +{ + struct sockaddr_in name; + int namelen = sizeof(name); + struct hostent *peername; + + assert(fd >= 0); + assert(string); + + memset(string, '\0', PEER_STRING_LENGTH); + + if (getpeername(fd, (struct sockaddr *) &name, &namelen) != 0) { + log("ERROR Connect: 'could not get peer name'"); + } else { + if ( + (peername = + gethostbyaddr((char *) &name.sin_addr.s_addr, + sizeof(name.sin_addr.s_addr), + AF_INET)) != NULL) { + strncpy(string, peername->h_name, + PEER_STRING_LENGTH - 1); + } + } + + return string; +} + + +/* + * Read one line of the header + */ +int readline(int fd, struct buffer_s *buffer, char **line) +{ + char inbuf[BUFFER]; + int bytesin; + char *endline = NULL; + char *newline; + struct bufline_s *oldline; + unsigned long len = 0, length; + + assert(fd >= 0); + assert(buffer); + assert(line); + + *line = NULL; + + if ((bytesin = recv(fd, inbuf, BUFFER - 1, MSG_PEEK)) <= 0) { + goto CONN_ERROR; + } + + endline = xstrstr(inbuf, "\n", bytesin, FALSE); + + if (endline) { + endline++; + *endline = '\0'; + + length = strlen(inbuf); + memset(inbuf, '\0', BUFFER); + + /* Actually pull it off the queue */ + if ((bytesin = recv(fd, inbuf, length, 0)) <= 0) { + goto CONN_ERROR; + } + + *line = xstrdup(inbuf); + + if (buffer_size(buffer) > 0) { + if (!(newline = xmalloc(buffer_size(buffer) + + length + 1))) { + return -1; + } + + newline[0] = '\0'; + + while ((oldline = shift_buffer(buffer))) { + memcpy(newline + len, oldline->string, + oldline->length); + + len += oldline->length; + free_line(oldline); + } + + memcpy(newline + len, *line, strlen(*line) + 1); + safefree(*line); + + *line = newline; + } + + return strlen(*line); + } + + /* + * We didn't find a '\r\n', but we've filled the buffer.. pull it off + * and try again later... + */ + if ((bytesin = recv(fd, inbuf, bytesin, 0)) <= 0) { + goto CONN_ERROR; + } + + if (!(*line = xmalloc(bytesin + 1))) { + return -1; + } + + memcpy(*line, inbuf, bytesin); + (*line)[bytesin] = '\0'; + + push_buffer(buffer, *line, strlen(*line)); + + return 0; + + CONN_ERROR: + if (bytesin == 0 || (errno != EAGAIN && errno != EINTR)) { + return -1; + } + return 0; +} diff --git a/src/sock.h b/src/sock.h new file mode 100644 index 0000000..952a673 --- /dev/null +++ b/src/sock.h @@ -0,0 +1,40 @@ +/* $Id: sock.h,v 1.1.1.1 2000-02-16 17:32:23 sdyoung Exp $ + * + * See 'sock.c' for a detailed description. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef _SOCK_H_ +#define _SOCK_H_ 1 + +#include "buffer.h" + +#define PEER_IP_LENGTH 16 +#define PEER_STRING_LENGTH 256 + +extern int setup_fd; + +extern int opensock(char *ip_addr, int port); +extern int opensock_inaddr(struct in_addr *inaddr, int port); +extern int init_listen_sock(int port); +extern int listen_sock(void); +extern void de_init_listen_sock(void); +extern int setsocketopt(int *sock_fd, int options, int flip); + +extern char *getpeer_ip(int fd, char *ipaddr); +extern char *getpeer_string(int fd, char *string); +extern int readline(int fd, struct buffer_s *buffer, char **line); + +#endif diff --git a/src/stamp-h.in b/src/stamp-h.in new file mode 100644 index 0000000..9788f70 --- /dev/null +++ b/src/stamp-h.in @@ -0,0 +1 @@ +timestamp diff --git a/src/tinyproxy.c b/src/tinyproxy.c new file mode 100644 index 0000000..1425492 --- /dev/null +++ b/src/tinyproxy.c @@ -0,0 +1,478 @@ +/* $Id: tinyproxy.c,v 1.1.1.1 2000-02-16 17:32:23 sdyoung Exp $ + * + * The initialize routine. Basically sets up all the initial stuff (logfile, + * listening socket, config options, etc.) and then sits there and loops + * over the new connections until the daemon is closed. Also has additional + * functions to handle the "user friendly" aspects of a program (usage, + * stats, etc.) Like any good program, most of the work is actually done + * elsewhere. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * Copyright (C) 2000 Chris Lightfoot (chris@ex-parrot.com> + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <stdio.h> +#include <sys/types.h> +#include <sys/socket.h> +#include <signal.h> +#include <stdlib.h> +#include <sysexits.h> +#include <pwd.h> +#include <string.h> +#include <unistd.h> +#include <errno.h> +#include <time.h> +#include <sys/time.h> +#include <syslog.h> +#include <unistd.h> +#include <fcntl.h> + +/* chris - need this for asynchronous DNS resolution */ +#include <adns.h> + +adns_state adns; + +#include "config.h" +#include "tinyproxy.h" +#include "utils.h" +#include "log.h" +#include "sock.h" +#include "conns.h" +#include "reqs.h" +#include "buffer.h" +#include "filter.h" + +void takesig(int sig); + +/* + * Global Structures + */ +struct config_s config = { + NULL, /* Log file handle */ + DEFAULT_LOG, /* Logfile name */ + FALSE, /* Use syslog instead? */ + DEFAULT_CUTOFFLOAD, /* Cut off load */ + DEFAULT_PORT, /* Listen on this port */ + DEFAULT_STATHOST, /* URL of stats host */ + FALSE, /* Quit? */ + DEFAULT_USER, /* Name of user to change to */ + FALSE, /* Run anonymous by default? */ + NULL, /* String containing the subnet allowed */ + NULL, /* IP address to listen on */ +#ifdef FILTER_ENABLE + NULL, /* Location of filter file */ +#endif /* FILTER_ENABLE */ + FALSE, /* Restrict the log to only errors */ +#ifdef XTINYPROXY + NULL /* The name of this domain */ +#endif +}; + +struct stat_s stats; +float load = 0.00; +struct allowedhdr_s *allowedhdrs = NULL; + +/* + * Dump info to the logfile + */ +static void dumpdebug(void) +{ + struct conn_s *connptr = connections; + long clients = 0, waiting = 0, relaying = 0, closing = 0, finished = 0; + + log("SIGUSR1 received, debug dump follows."); + + while (connptr) { + switch (connptr->type) { + case NEWCONN: + clients++; + break; + case WAITCONN: + waiting++; + break; + case RELAYCONN: + relaying++; + break; + case CLOSINGCONN: + closing++; + break; + case FINISHCONN: + finished++; + break; + default: + break; + } + connptr = connptr->next; + } + log("clients: %d, waiting: %d, relaying: %d," \ + "closing: %d, finished: %d", + clients, waiting, relaying, closing, finished); + log("total requests handled: %lu", stats.num_reqs); + log("total connections handled: %lu", stats.num_cons); + log("total sockets listened: %lu", stats.num_listens); + log("total sockets opened: %lu", stats.num_opens); + log("total bad opens: %lu", stats.num_badcons); + log("total bytes tx: %lu", stats.num_tx); + log("total bytes rx: %lu", stats.num_rx); + log("connections refused due to load: %lu", stats.num_refused); + log("garbage collections: %lu", stats.num_garbage); + log("idle connections killed: %lu", stats.num_idles); + log("end debug dump."); +} + +/* + * Handle a signal + */ +void takesig(int sig) +{ + switch (sig) { + case SIGUSR1: + dumpdebug(); + break; + case SIGHUP: + if (config.logf) + ftruncate(fileno(config.logf), 0); + + log("SIGHUP received, cleaning up..."); + conncoll(); + garbcoll(); + +#ifdef FILTER_ENABLE + if (config.filter) { + filter_destroy(); + filter_init(); + } + log("Re-reading filter file."); +#endif /* FILTER_ENABLE */ + log("Finished cleaning memory/connections."); + break; + case SIGTERM: +#ifdef FILTER_ENABLE + if (config.filter) + filter_destroy(); +#endif /* FILTER_ENABLE */ + config.quit = TRUE; + break; + case SIGALRM: + calcload(); + alarm(LOAD_RECALCTIMER); + break; + } + if (sig != SIGTERM) + signal(sig, takesig); + signal(SIGPIPE, SIG_IGN); +} + +/* + * Display usage to the user on stderr. + */ +static void usagedisp(void) +{ + printf("tinyproxy version " VERSION "\n"); + printf("Copyright 1998 Steven Young (sdyoung@well.com)\n"); + printf + ("Copyright 1998-1999 Robert James Kaes (rjkaes@flarenet.com)\n\n"); + printf("Copyright 2000 Chris Lightfoot (chris@ex-parrot.com)\n"); + + printf + ("This software is licensed under the GNU General Public License (GPL).\n"); + printf("See the file 'COPYING' included with tinyproxy source.\n\n"); + + printf("Compiled with Ian Jackson's adns:\n"); + printf(" http://www.chiark.greenend.org.uk/~ian/adns/\n\n"); + + printf("Usage: tinyproxy [args]\n"); + printf("Options:\n"); + printf("\t-v\t\tdisplay version number\n"); + printf("\t-h\t\tdisplay usage\n"); + printf("\t-d\t\tdo not daemonize\n"); + printf + ("\t-n ip_address\tallow access from only this subnet. (i.e. 192.168.0.)\n"); + printf("\t-i ip_address\tonly listen on this address\n"); + printf("\t-p port\t\tlisten on 'port'\n"); + printf("\t-l logfile\tlog to 'logfile'\n"); +#ifdef HAVE_SYSLOG_H + printf("\t-S\t\tlog using the syslog instead\n"); +#endif + printf("\t-r\t\trestrict the log to only errors\n"); + printf + ("\t-w load\t\tstop accepting new connections at 'load'. 0 disables\n"); + printf("\t-s stathost\tset stathost to 'stathost'\n"); + printf("\t-u user\t\tchange to user after startup. \"\" disables\n"); + printf("\t-a header\tallow 'header' through the anon block\n"); +#ifdef FILTER_ENABLE + printf("\t-f filterfile\tblock sites specified in filterfile\n"); +#endif /* FILTER_ENABLE */ +#ifdef XTINYPROXY + printf + ("\t-x domain\tAdd a XTinyproxy header with the peer's IP address\n"); +#endif + + /* Display the modes compiled into tinyproxy */ + printf("\nFeatures Compiled In:\n"); +#ifdef XTINYPROXY + printf(" XTinyproxy Header\n"); +#endif /* XTINYPROXY */ +#ifdef FILTER_ENABLE + printf(" Filtering\n"); + printf(" * with Regular Expression support\n"); +#endif /* FILTER_ENABLE */ +#ifndef NDEBUG + printf(" Debuggin code\n"); +#endif /* NDEBUG */ +} + +int main(int argc, char **argv) +{ + int optch; + flag usage = FALSE, godaemon = TRUE, changeid = FALSE; + struct passwd *thisuser = NULL; + + struct allowedhdr_s **rpallowedptr = &allowedhdrs; + struct allowedhdr_s *allowedptr = allowedhdrs, *newallowed; + + while ((optch = getopt(argc, argv, "vh?dp:l:Sa:w:s:u:n:i:rx:f:")) != + EOF) { + switch (optch) { + case 'v': + fprintf(stderr, "tinyproxy version " VERSION "\n"); + exit(EX_OK); + break; + case 'p': + if (!(config.port = atoi(optarg))) { + log + ("bad port on commandline, defaulting to %d", + DEFAULT_PORT); + config.port = DEFAULT_PORT; + } + break; + case 'l': + if (!(config.logf_name = xstrdup(optarg))) { + log("bad log file, defaulting to %s", + DEFAULT_LOG); + config.logf_name = DEFAULT_LOG; + } + break; +#ifdef HAVE_SYSLOG_H + case 'S': /* Use the syslog function to handle logging */ + config.syslog = TRUE; + break; +#endif + case 'd': + godaemon = FALSE; + break; + case 'w': + sscanf(optarg, "%f", &config.cutoffload); + break; + case 's': + if (!(config.stathost = xstrdup(optarg))) { + log("bad stathost, defaulting to %s", + DEFAULT_STATHOST); + config.stathost = DEFAULT_STATHOST; + } + break; + case 'u': + if (!(config.changeuser = xstrdup(optarg))) { + log("bad user name, defaulting to %s", + DEFAULT_USER); + config.changeuser = DEFAULT_USER; + } + break; + case 'a': + config.anonymous = TRUE; + + while (allowedptr) { + rpallowedptr = &allowedptr->next; + allowedptr = allowedptr->next; + } + + if (! + (newallowed = + xmalloc(sizeof(struct allowedhdr_s)))) { + log("tinyproxy: cannot allocate headers"); + exit(EX_SOFTWARE); + } + + if (!(newallowed->hdrname = xstrdup(optarg))) { + log("tinyproxy: cannot duplicate string"); + exit(EX_SOFTWARE); + } + + *rpallowedptr = newallowed; + newallowed->next = allowedptr; + + break; + case 'n': + if (!(config.subnet = xstrdup(optarg))) { + log("tinyproxy: could not allocate memory"); + exit(EX_SOFTWARE); + } + break; + case 'i': + if (!(config.ipAddr = xstrdup(optarg))) { + log("tinyproxy: could not allocate memory"); + exit(EX_SOFTWARE); + } + break; +#ifdef FILTER_ENABLE + case 'f': + if (!(config.filter = xstrdup(optarg))) { + log("tinyproxy: could not allocate memory"); + } + break; +#endif /* FILTER_ENABLE */ + case 'r': + config.restricted = TRUE; + break; +#ifdef XTINYPROXY + case 'x': + if (!(config.my_domain = xstrdup(optarg))) { + log("tinyproxy: could not allocate memory"); + exit(EX_SOFTWARE); + } + break; +#endif + case '?': + case 'h': + default: + usage = TRUE; + break; + } + } + + if (usage == TRUE) { + usagedisp(); + exit(EX_OK); + } + + /* chris - Initialise asynchronous DNS */ + if (adns_init(&adns, 0, 0)) { + log("tinyproxy: could not initialise ADNS"); + exit(EX_SOFTWARE); + } + + /* Open the log file if not using syslog */ + if (config.syslog == FALSE) { + if (!(config.logf = fopen(config.logf_name, "a"))) { + fprintf(stderr, + "Unable to open logfile %s for appending!\n", + config.logf_name); + exit(EX_CANTCREAT); + } + } else { + if (godaemon == TRUE) + openlog("tinyproxy", LOG_PID, LOG_DAEMON); + else + openlog("tinyproxy", LOG_PID, LOG_USER); + } + + log(PACKAGE " " VERSION " starting..."); + + if (strlen(config.changeuser)) { + if ((getuid() != 0) && (geteuid() != 0)) { + log + ("not running as root, therefore not changing uid/gid."); + } else { + changeid = TRUE; + if (!(thisuser = getpwnam(config.changeuser))) { + log("unable to find user \"%s\"!", + config.changeuser); + exit(EX_NOUSER); + } + log("changing to user \"%s\" (%d/%d).", + config.changeuser, thisuser->pw_uid, + thisuser->pw_gid); + } + } +#ifdef NDEBUG + if (godaemon == TRUE) + makedaemon(); +#else + printf("Debugging is enabled, so you can not go daemon.\n"); +#endif + + if (signal(SIGPIPE, SIG_IGN) == SIG_ERR) { + fprintf(stderr, "Could not set SIGPIPE\n"); + exit(EX_OSERR); + } + if (signal(SIGUSR1, takesig) == SIG_ERR) { + fprintf(stderr, "Could not set SIGUSR1\n"); + exit(EX_OSERR); + } + if (signal(SIGTERM, takesig) == SIG_ERR) { + fprintf(stderr, "Could not set SIGTERM\n"); + exit(EX_OSERR); + } + if (signal(SIGHUP, takesig) == SIG_ERR) { + fprintf(stderr, "Could not set SIGHUP\n"); + exit(EX_OSERR); + } + if (signal(SIGALRM, takesig) == SIG_ERR) { + fprintf(stderr, "Could not set SIGALRM\n"); + exit(EX_OSERR); + } + alarm(LOAD_RECALCTIMER); + calcload(); + + if (init_listen_sock(config.port) < 0) { + log("unable to bind port %d!", config.port); + exit(EX_UNAVAILABLE); + } + if (changeid == TRUE) { + setuid(thisuser->pw_uid); + setgid(thisuser->pw_gid); + } + log("now accepting connections."); + +#ifdef FILTER_ENABLE + if (config.filter) + filter_init(); +#endif /* FILTER_ENABLE */ + + while (config.quit == FALSE) { + if (getreqs() < 0) + break; + } + +#ifdef FILTER_ENABLE + if (config.filter) + filter_destroy(); +#endif /* FILTER_ENABLE */ + + log("shutting down."); + de_init_listen_sock(); + + if (config.syslog == FALSE) + fclose(config.logf); + else + closelog(); + + allowedptr = allowedhdrs; + while (allowedptr) { + struct allowedhdr_s *delptr = NULL; + delptr = allowedptr; + safefree(delptr->hdrname); + allowedptr = delptr->next; + safefree(delptr); + } + + /* finsih up ADNS */ + adns_finish(adns); + + exit(EX_OK); +} diff --git a/src/tinyproxy.h b/src/tinyproxy.h new file mode 100644 index 0000000..218d729 --- /dev/null +++ b/src/tinyproxy.h @@ -0,0 +1,87 @@ +/* $Id: tinyproxy.h,v 1.1.1.1 2000-02-16 17:32:23 sdyoung Exp $ + * + * See 'tinyproxy.c' for a detailed description. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef _TINYPROXY_H_ +#define _TINYPROXY_H_ 1 + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <stdio.h> +#include <time.h> + +#include "config.h" + +/* Global variables for the main controls of the program */ +#define BUFFER (1024 * 2) /* Size of buffer for reading */ +#define MAXLISTEN 128 /* Max number of connections to listen for */ + +/* Make a new type: flag */ +typedef char flag; + +/* Other stuff */ +#define FALSE (0) +#define TRUE (!FALSE) + +struct config_s { + FILE *logf; + char *logf_name; + flag syslog; + float cutoffload; + int port; + char *stathost; + flag quit; + char *changeuser; + flag anonymous; + char *subnet; + char *ipAddr; +#ifdef FILTER_ENABLE + char *filter; +#endif /* FILTER_ENABLE */ + flag restricted; +#ifdef XTINYPROXY + char *my_domain; +#endif +}; + +struct stat_s { + unsigned long int num_reqs; + unsigned long int num_cons; + unsigned long int num_badcons; + unsigned long int num_opens; + unsigned long int num_listens; + unsigned long int num_tx; + unsigned long int num_rx; + unsigned long int num_garbage; + unsigned long int num_idles; + unsigned long int num_refused; +}; + +struct allowedhdr_s { + char *hdrname; + struct allowedhdr_s *next; +}; + +/* Global Structures used in the program */ +extern struct config_s config; +extern struct stat_s stats; +extern struct allowedhdr_s *allowedhdrs; +extern float load; + +#endif diff --git a/src/uri.c b/src/uri.c new file mode 100644 index 0000000..8a2b744 --- /dev/null +++ b/src/uri.c @@ -0,0 +1,124 @@ +/* $Id: uri.c,v 1.1.1.1 2000-02-16 17:32:23 sdyoung Exp $ + * + * This borrows the REGEX from RFC2396 to split a URI string into the five + * primary components. The components are: + * scheme the uri method (like "http", "ftp", "gopher") + * authority the domain and optional ":" port + * path path to the document/resource + * query an optional query (separated with a "?") + * fragment an optional fragement (separated with a "#") + * + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <string.h> +#include <stdlib.h> +#include <assert.h> + +#include "uri.h" +#include "utils.h" +#include "log.h" +#include "regexp.h" + +#define NMATCH 10 + +#define URIPATTERN "^(([^:/?#]+):)?(//([^/?#]*))?([^?#]*)(\\?([^#]*))?(#(.*))?" + +#define SCHEME 2 +#define AUTHORITY 4 +#define PATH 5 +#define QUERY_MARK 6 +#define QUERY 7 +#define FRAGMENT_MARK 8 +#define FRAGMENT 9 + +static int extract_uri(regmatch_t pmatch[], const char *buffer, char **section, + int substring) +{ + int len = pmatch[substring].rm_eo - pmatch[substring].rm_so; + if ((*section = xmalloc(len + 1)) == NULL) + return -1; + + memset(*section, '\0', len + 1); + memcpy(*section, buffer + pmatch[substring].rm_so, len); + + return 0; +} + +void free_uri(URI * uri) +{ + safefree(uri->scheme); + safefree(uri->authority); + safefree(uri->path); + safefree(uri->query); + safefree(uri->fragment); + safefree(uri); +} + +URI *explode_uri(const char *string) +{ + URI *uri; + regmatch_t pmatch[NMATCH]; + regex_t preg; + + if (!(uri = xmalloc(sizeof(URI)))) + return NULL; + memset(uri, 0, sizeof(URI)); + + if (regcomp(&preg, URIPATTERN, REG_EXTENDED) != 0) { + log("ERROR explode_uri: regcomp"); + goto ERROR_EXIT; + } + + if (regexec(&preg, string, NMATCH, pmatch, 0) != 0) { + log("ERROR explode_uri: regexec"); + goto ERROR_EXIT; + } + + regfree(&preg); + + if (pmatch[SCHEME].rm_so != -1) { + if (extract_uri(pmatch, string, &uri->scheme, SCHEME) < 0) + goto ERROR_EXIT; + } + + if (pmatch[AUTHORITY].rm_so != -1) { + if (extract_uri(pmatch, string, &uri->authority, AUTHORITY) < + 0) goto ERROR_EXIT; + } + + if (pmatch[PATH].rm_so != -1) { + if (extract_uri(pmatch, string, &uri->path, PATH) < 0) + goto ERROR_EXIT; + } + + if (pmatch[QUERY_MARK].rm_so != -1) { + if (extract_uri(pmatch, string, &uri->query, QUERY) < 0) + goto ERROR_EXIT; + } + + if (pmatch[FRAGMENT_MARK].rm_so != -1) { + if (extract_uri(pmatch, string, &uri->fragment, FRAGMENT) < 0) + goto ERROR_EXIT; + } + + return uri; + + ERROR_EXIT: + free_uri(uri); + return NULL; +} diff --git a/src/uri.h b/src/uri.h new file mode 100644 index 0000000..bb18d15 --- /dev/null +++ b/src/uri.h @@ -0,0 +1,32 @@ +/* $Id: uri.h,v 1.1.1.1 2000-02-16 17:32:24 sdyoung Exp $ + * + * See 'uri.c' for a detailed description. + * + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef __URI_H_ +#define __URI_H_ + +typedef struct { + char *scheme; + char *authority; + char *path; + char *query; + char *fragment; +} URI; + +extern URI *explode_uri(const char *string); +extern void free_uri(URI * uri); + +#endif diff --git a/src/utils.c b/src/utils.c new file mode 100644 index 0000000..1a91d7a --- /dev/null +++ b/src/utils.c @@ -0,0 +1,264 @@ +/* $Id: utils.c,v 1.1.1.1 2000-02-16 17:32:24 sdyoung Exp $ + * + * Misc. routines which are used by the various functions to handle strings + * and memory allocation and pretty much anything else we can think of. Also, + * the load cutoff routine is in here, along with the HTML show stats + * function. Could not think of a better place for it, so it's in here. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifdef HAVE_CONFIG_H +#include <defines.h> +#endif + +#include <stdio.h> +#include <string.h> +#include <stdlib.h> +#include <unistd.h> +#include <time.h> +#include <ctype.h> +#include <sysexits.h> +#include <sys/types.h> +#include <sys/socket.h> +#include <errno.h> +#include <signal.h> +#include <sys/stat.h> +#include <assert.h> + +#include "config.h" +#include "tinyproxy.h" +#include "utils.h" +#include "log.h" +#include "conns.h" +#include "buffer.h" + +char *xstrdup(char *st) +{ + char *p; + + assert(st); + + if (!(p = strdup(st))) { + log("ERROR xstrdup: out of memory (%s)", strerror(errno)); + return NULL; + } else { + return p; + } +} + +/* + * Find the start of the needle in the haystack. Limits the search to less + * than "length" characters. Returns NULL if the needle is not found. + */ +char *xstrstr(char *haystack, char *needle, unsigned int length, + int case_sensitive) +{ + unsigned int i; + /* Used to specify which function to use... need the decl. */ + int (*fn) (const char *s1, const char *s2, unsigned int n); + + assert(haystack); + assert(needle); + assert(length > 0); + assert(case_sensitive == FALSE || case_sensitive == TRUE); + + if (case_sensitive) + fn = strncmp; + else + fn = strncasecmp; + + if (strlen(needle) > length) + return NULL; + + for (i = 0; i <= length - strlen(needle); i++) { + if ((*fn) (haystack + i, needle, strlen(needle)) == 0) + return haystack + i; + + } + + return NULL; +} + +/* + * for-sure malloc + */ +void *xmalloc(unsigned long int sz) +{ + void *p; + + assert(sz > 0); + + if (!(p = malloc((size_t) sz))) { + log("ERROR xmalloc: out of memory (%s)", strerror(errno)); + return NULL; + } + return p; +} + +#ifdef USE_PROC +int calcload(void) +{ + char buf[BUFFER], *p; + FILE *f; + + if (!config.cutoffload) { + return -1; + } + + if (!(f = fopen("/proc/loadavg", "rt"))) { + log("unable to read /proc/loadavg"); + config.cutoffload = 0.0; + return -1; + } + fgets(buf, BUFFER, f); + p = strchr(buf, ' '); + *p = '\0'; + load = atof(buf); + fclose(f); + return 0; +} + +#else +int calcload(void) +{ + FILE *f; + char buf[BUFFER]; + char *p, *y; + + if (!config.cutoffload) { + return -1; + } + + if (!(f = popen(UPTIME_PATH, "r"))) { + log("calcload: unable to exec uptime"); + config.cutoffload = 0.0; + return -1; + } + fgets(buf, BUFFER, f); + p = strrchr(buf, ':'); + p += 2; + y = strchr(p, ','); + *y = '\0'; + load = atof(p); + pclose(f); + return 0; +} + +#endif + +/* + * Delete the server's buffer and replace it with a premade message which will + * be sent to the client. + */ +static void update_output_buffer(struct conn_s *connptr, char *outbuf) +{ + assert(connptr); + assert(outbuf); + + delete_buffer(connptr->sbuffer); + connptr->sbuffer = new_buffer(); + + push_buffer(connptr->sbuffer, outbuf, strlen(outbuf)); + shutdown(connptr->server_fd, 2); + connptr->type = CLOSINGCONN; +} + +/* + * Display the statics of the tinyproxy server. + */ +int showstats(struct conn_s *connptr) +{ + char *outbuf; + static char *msg = "HTTP/1.0 200 OK\r\n" \ + "Content-type: text/html\r\n\r\n" \ + "<html><head><title>%s stats</title></head>\r\n" \ + "<body>\r\n" \ + "<center><h2>%s run-time statistics</h2></center><hr>\r\n" \ + "<blockquote>\r\n" \ + "Number of requests: %lu<br>\r\n" \ + "Number of connections: %lu<br>\r\n" \ + "Number of bad connections: %lu<br>\r\n" \ + "Number of opens: %lu<br>\r\n" \ + "Number of listens: %lu<br>\r\n" \ + "Number of bytes (tx): %lu<br>\r\n" \ + "Number of bytes (rx): %lu<br>\r\n" \ + "Number of garbage collects:%lu<br>\r\n" \ + "Number of idle connection kills:%lu<br>\r\n" \ + "Number of refused connections due to high load:%lu<br>\r\n" \ + "Current system load average:%.2f" \ + "(recalculated every % lu seconds)<br>\r\n" \ + "</blockquote>\r\n</body></html>\r\n"; + + assert(connptr); + + outbuf = xmalloc(BUFFER); + + sprintf(outbuf, msg, VERSION, VERSION, stats.num_reqs, + stats.num_cons, stats.num_badcons, stats.num_opens, + stats.num_listens, stats.num_tx, stats.num_rx, + stats.num_garbage, stats.num_idles, stats.num_refused, load, + LOAD_RECALCTIMER); + + update_output_buffer(connptr, outbuf); + + return 0; +} + +/* + * Display an error to the client. + */ +int httperr(struct conn_s *connptr, int err, char *msg) +{ + char *outbuf; + static char *premsg = "HTTP/1.0 %d %s\r\n" \ + "Content-type: text/html\r\n\r\n" \ + "<html><head><title>%s</title></head>\r\n" \ + "<body>\r\n" \ + "<font size=\"+2\">Cache Error!</font><br>\r\n" \ + "An error of type %d occurred: %s\r\n" \ + "<hr>\r\n" \ + "<font size=\"-1\"><em>Generated by %s</em></font>\r\n" \ + "</body></html>\r\n"; + + assert(connptr); + assert(err > 0); + assert(msg); + + outbuf = xmalloc(BUFFER); + sprintf(outbuf, premsg, err, msg, msg, err, msg, VERSION); + + update_output_buffer(connptr, outbuf); + + return 0; +} + +void makedaemon(void) +{ + if (fork() != 0) + exit(0); + + setsid(); + signal(SIGHUP, SIG_IGN); + + if (fork() != 0) + exit(0); + + chdir("/"); + umask(0); + + close(0); + close(1); + close(2); +} diff --git a/src/utils.h b/src/utils.h new file mode 100644 index 0000000..76a3565 --- /dev/null +++ b/src/utils.h @@ -0,0 +1,38 @@ +/* $Id: utils.h,v 1.1.1.1 2000-02-16 17:32:24 sdyoung Exp $ + * + * See 'utils.h' for a detailed description. + * + * Copyright (C) 1998 Steven Young + * Copyright (C) 1999 Robert James Kaes (rjkaes@flarenet.com) + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + */ + +#ifndef __UTILS_H_ +#define __UTILS_H_ 1 + +#include "conns.h" + +#define safefree(x) free(x); x = NULL + +extern char *xstrdup(char *st); +extern void *xmalloc(unsigned long int sz); +extern char *xstrstr(char *haystack, char *needle, unsigned int length, + int case_sensitive); + +extern int showstats(struct conn_s *connptr); +extern int httperr(struct conn_s *connptr, int err, char *msg); + +extern int calcload(void); + +extern void makedaemon(void); + +#endif |