-const char pcrs_rcs[] = "$Id: pcrs.c,v 1.42 2012/05/24 15:04:34 fabiankeil Exp $";
/*********************************************************************
*
* File : $Source: /cvsroot/ijbswa/current/pcrs.c,v $
* Copyright (C) 2006, 2007 Fabian Keil <fk@fabiankeil.de>
*
* This program is free software; you can redistribute it
- * and/or modify it under the terms of the GNU Lesser
- * General Public License (LGPL), version 2.1, which should
- * be included in this distribution (see LICENSE.txt), with
- * the exception that the permission to replace that license
- * with the GNU General Public License (GPL) given in section
- * 3 is restricted to version 2 of the GPL.
+ * and/or modify it under the terms of the GNU General
+ * Public License as published by the Free Software
+ * Foundation; either version 2 of the License, 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 license for more details.
+ * PARTICULAR PURPOSE. See the GNU General Public
+ * License for more details.
*
- * The GNU Lesser General Public License should be included
- * with this file. If not, you can view it at
- * http://www.gnu.org/licenses/lgpl.html
+ * The GNU General Public License should be included with
+ * this file. If not, you can view it at
+ * http://www.gnu.org/copyleft/gpl.html
* or write to the Free Software Foundation, Inc., 59
* Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*
#include "pcrs.h"
-const char pcrs_h_rcs[] = PCRS_H_VERSION;
-
/*
* Internal prototypes
*/
-static int pcrs_parse_perl_options(const char *optstring, int *flags);
+static int pcrs_parse_perl_options(const char *optstring, unsigned int *flags);
static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag,
int capturecount, int *errptr);
static int is_hex_sequence(const char *sequence);
*********************************************************************/
const char *pcrs_strerror(const int error)
{
+ static char buf[100];
+
if (error != 0)
{
switch (error)
{
/* Passed-through PCRE error: */
- case PCRE_ERROR_NOMEMORY: return "(pcre:) No memory";
+ case PCREn(ERROR_NOMEMORY): return "(pcre:) No memory";
/* Shouldn't happen unless PCRE or PCRS bug, or user messed with compiled job: */
- case PCRE_ERROR_NULL: return "(pcre:) NULL code or subject or ovector";
- case PCRE_ERROR_BADOPTION: return "(pcre:) Unrecognized option bit";
- case PCRE_ERROR_BADMAGIC: return "(pcre:) Bad magic number in code";
+ case PCREn(ERROR_NULL): return "(pcre:) NULL code or subject or ovector";
+ case PCREn(ERROR_BADOPTION): return "(pcre:) Unrecognized option bit";
+ case PCREn(ERROR_BADMAGIC): return "(pcre:) Bad magic number in code";
+#if defined(PCRE_ERROR_UNKNOWN_NODE)
case PCRE_ERROR_UNKNOWN_NODE: return "(pcre:) Bad node in pattern";
-
+#endif
/* Can't happen / not passed: */
- case PCRE_ERROR_NOSUBSTRING: return "(pcre:) Fire in power supply";
- case PCRE_ERROR_NOMATCH: return "(pcre:) Water in power supply";
+ case PCREn(ERROR_NOSUBSTRING): return "(pcre:) Fire in power supply";
+ case PCREn(ERROR_NOMATCH): return "(pcre:) Water in power supply";
#ifdef PCRE_ERROR_MATCHLIMIT
/*
* Only reported by PCRE versions newer than our own.
*/
- case PCRE_ERROR_MATCHLIMIT: return "(pcre:) Match limit reached";
+ case PCREn(ERROR_MATCHLIMIT): return "(pcre:) Match limit reached";
#endif /* def PCRE_ERROR_MATCHLIMIT */
-
/* PCRS errors: */
case PCRS_ERR_NOMEM: return "(pcrs:) No memory";
case PCRS_ERR_CMDSYNTAX: return "(pcrs:) Syntax error while parsing command";
case PCRS_WARN_TRUNCATION:
return "(pcrs:) At least one variable was too big and has been truncated before compilation";
- /*
- * XXX: With the exception of PCRE_ERROR_MATCHLIMIT we
- * only catch PCRE errors that can happen with our internal
- * version. If Privoxy is linked against a newer
- * PCRE version all bets are off ...
- */
- default: return "Unknown error. Privoxy out of sync with PCRE?";
+ default:
+#ifdef HAVE_PCRE2
+ pcre2_get_error_message(error, (PCRE2_UCHAR8*)buf, sizeof(buf));
+#else
+ snprintf(buf, sizeof(buf),
+ "Error code %d. For details, check the pcre documentation.",
+ error);
+#endif
+ return buf;
}
}
/* error >= 0: No error */
* Returns : option integer suitable for pcre
*
*********************************************************************/
-static int pcrs_parse_perl_options(const char *optstring, int *flags)
+static int pcrs_parse_perl_options(const char *optstring, unsigned int *flags)
{
size_t i;
int rc = 0;
{
case 'e': break; /* ToDo ;-) */
case 'g': *flags |= PCRS_GLOBAL; break;
- case 'i': rc |= PCRE_CASELESS; break;
- case 'm': rc |= PCRE_MULTILINE; break;
+ case 'i': rc |= PCREn(CASELESS); break;
+ case 'm': rc |= PCREn(MULTILINE); break;
case 'o': break;
- case 's': rc |= PCRE_DOTALL; break;
- case 'x': rc |= PCRE_EXTENDED; break;
- case 'U': rc |= PCRE_UNGREEDY; break;
+ case 's': rc |= PCREn(DOTALL); break;
+ case 'x': rc |= PCREn(EXTENDED); break;
+ case 'D': *flags |= PCRS_DYNAMIC; break;
+ case 'U': rc |= PCREn(UNGREEDY); break;
case 'T': *flags |= PCRS_TRIVIAL; break;
default: break;
}
}
+#ifdef FUZZ
+/*********************************************************************
+ *
+ * Function : pcrs_compile_fuzzed_replacement
+ *
+ * Description : Wrapper around pcrs_compile_replacement() for
+ * fuzzing purposes.
+ *
+ * Parameters :
+ * 1 : replacement = replacement part of s/// operator
+ * in perl syntax
+ * 2 : errptr = pointer to an integer in which error
+ * conditions can be returned.
+ *
+ * Returns : pcrs_substitute data structure, or NULL if an
+ * error is encountered. In that case, *errptr has
+ * the reason.
+ *
+ *********************************************************************/
+extern pcrs_substitute *pcrs_compile_fuzzed_replacement(const char *replacement, int *errptr)
+{
+ int capturecount = PCRS_MAX_SUBMATCHES; /* XXX: fuzzworthy? */
+ int trivial_flag = 0; /* We don't want to fuzz strncpy() */
+
+ *errptr = 0; /* XXX: Should pcrs_compile_replacement() do this? */
+
+ return pcrs_compile_replacement(replacement, trivial_flag, capturecount, errptr);
+
+}
+#endif
+
+
/*********************************************************************
*
* Function : pcrs_compile_replacement
static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag, int capturecount, int *errptr)
{
int i, k, l, quoted;
- size_t length;
char *text;
pcrs_substitute *r;
-
+#ifndef FUZZ
+ size_t length;
+#else
+ static size_t length;
+#endif
i = k = l = quoted = 0;
/*
*/
if (trivialflag)
{
- text = strncpy(text, replacement, length + 1);
+ strlcpy(text, replacement, length + 1);
k = (int)length;
}
if (replacement[i] == '$' && !quoted && i < (int)(length - 1))
{
char *symbol, symbols[] = "'`+&";
+ if (l >= PCRS_MAX_SUBMATCHES)
+ {
+ freez(text);
+ freez(r);
+ *errptr = PCRS_WARN_BADREF;
+ return NULL;
+ }
r->block_length[l] = (size_t)(k - r->block_offset[l]);
/* Numerical backreferences */
}
if (r->backref[l] > capturecount)
{
+ freez(text);
+ freez(r);
*errptr = PCRS_WARN_BADREF;
+ return NULL;
}
}
goto plainchar;
}
+ assert(r->backref[l] < PCRS_MAX_SUBMATCHES + 2);
/* Valid and in range? -> record */
- if (r->backref[l] < PCRS_MAX_SUBMATCHES + 2)
+ if ((0 <= r->backref[l]) &&
+ (r->backref[l] < PCRS_MAX_SUBMATCHES + 2) &&
+ (l < PCRS_MAX_SUBMATCHES - 1))
{
r->backref_count[r->backref[l]] += 1;
r->block_offset[++l] = k;
}
else
{
+ freez(text);
+ freez(r);
*errptr = PCRS_WARN_BADREF;
+ return NULL;
}
continue;
}
else
{
next = job->next;
- if (job->pattern != NULL) free(job->pattern);
- if (job->hints != NULL) free(job->hints);
+ if (job->pattern != NULL)
+ {
+#ifdef HAVE_PCRE2
+ pcre2_code_free(job->pattern);
+#else
+ free(job->pattern);
+#endif
+ }
+#ifndef HAVE_PCRE2
+ if (job->hints != NULL)
+ {
+#ifdef PCRE_CONFIG_JIT
+ pcre_free_study(job->hints);
+#else
+ free(job->hints);
+#endif
+ }
+#endif
if (job->substitute != NULL)
{
if (job->substitute->text != NULL) free(job->substitute->text);
pcrs_job *pcrs_compile(const char *pattern, const char *substitute, const char *options, int *errptr)
{
pcrs_job *newjob;
- int flags;
+ unsigned int flags;
int capturecount;
+#ifdef HAVE_PCRE2
+ int ret;
+#else
+ int pcre_study_options = 0;
const char *error;
+#endif
*errptr = 0;
/*
* Compile the pattern
*/
+#ifdef HAVE_PCRE2
+ PCRE2_SIZE error_offset;
+ newjob->pattern = pcre2_compile((const unsigned char *)pattern,
+ PCRE2_ZERO_TERMINATED, (unsigned)newjob->options, errptr,
+ &error_offset, NULL);
+#else
newjob->pattern = pcre_compile(pattern, newjob->options, &error, errptr, NULL);
+#endif
if (newjob->pattern == NULL)
{
pcrs_free_job(newjob);
return NULL;
}
+#if defined(PCRE_STUDY_JIT_COMPILE) || defined(HAVE_PCRE2)
+#ifdef DISABLE_PCRE_JIT_COMPILATION
+#warning PCRE_STUDY_JIT_COMPILE is supported but Privoxy has been configured not to use it
+#else
+ if (!(flags & PCRS_DYNAMIC))
+ {
+#ifdef HAVE_PCRE2
+ /* Try to enable JIT compilation but continue if it's unsupported. */
+ if ((ret = pcre2_jit_compile(newjob->pattern, PCRE2_JIT_COMPLETE)) &&
+ (ret != PCRE2_ERROR_JIT_BADOPTION))
+ {
+ *errptr = ret;
+ pcrs_free_job(newjob);
+ return NULL;
+ }
+#else
+ pcre_study_options = PCRE_STUDY_JIT_COMPILE;
+#endif
+ }
+#endif
+#endif
+#ifndef HAVE_PCRE2
/*
* Generate hints. This has little overhead, since the
* hints will be NULL for a boring pattern anyway.
*/
- newjob->hints = pcre_study(newjob->pattern, 0, &error);
+ newjob->hints = pcre_study(newjob->pattern, pcre_study_options, &error);
if (error != NULL)
{
*errptr = PCRS_ERR_STUDY;
pcrs_free_job(newjob);
return NULL;
}
-
+#endif
/*
* Determine the number of capturing subpatterns.
* This is needed for handling $+ in the substitute.
*/
+#ifdef HAVE_PCRE2
+ if (0 > (*errptr = pcre2_pattern_info(newjob->pattern, PCRE2_INFO_CAPTURECOUNT, &capturecount)))
+#else
if (0 > (*errptr = pcre_fullinfo(newjob->pattern, newjob->hints, PCRE_INFO_CAPTURECOUNT, &capturecount)))
+#endif
{
pcrs_free_job(newjob);
return NULL;
* 1 : job = the pcrs_job to be executed
* 2 : subject = the subject (== original) string
* 3 : subject_length = the subject's length
- * 4 : result = char** for returning the result
+ * 4 : result = char** for returning the result (NULL on error)
* 5 : result_length = size_t* for returning the result's length
*
* Returns : On success, the number of substitutions that were made.
*********************************************************************/
int pcrs_execute(pcrs_job *job, const char *subject, size_t subject_length, char **result, size_t *result_length)
{
- int offsets[3 * PCRS_MAX_SUBMATCHES],
- offset,
+ int offset,
i, k,
matches_found,
submatches,
max_matches = PCRS_MAX_MATCH_INIT;
size_t newsize;
+#ifdef HAVE_PCRE2
+ pcrs_match *matches, *dummy;
+ pcre2_match_data *pcre2_matches;
+ size_t *offsets;
+#else
pcrs_match *matches, *dummy;
+ int offsets[3 * PCRS_MAX_SUBMATCHES];
+#endif
char *result_offset;
offset = i = 0;
+ *result = NULL;
/*
* Sanity check & memory allocation
*/
if (job == NULL || job->pattern == NULL || job->substitute == NULL || NULL == subject)
{
- *result = NULL;
return(PCRS_ERR_BADJOB);
}
+#ifdef HAVE_PCRE2
+ if (NULL == (pcre2_matches = pcre2_match_data_create_from_pattern(job->pattern, NULL)))
+ {
+ return(PCRS_ERR_NOMEM);
+ }
+ offsets = pcre2_get_ovector_pointer(pcre2_matches);
+#endif
if (NULL == (matches = (pcrs_match *)malloc((size_t)max_matches * sizeof(pcrs_match))))
{
- *result = NULL;
return(PCRS_ERR_NOMEM);
}
memset(matches, '\0', (size_t)max_matches * sizeof(pcrs_match));
-
/*
* Find the pattern and calculate the space
* requirements for the result
*/
newsize = subject_length;
+#ifdef HAVE_PCRE2
+ while ((submatches = pcre2_match(job->pattern, (const unsigned char *)subject,
+ subject_length, (size_t)offset, 0, pcre2_matches, NULL)) > 0)
+#else
while ((submatches = pcre_exec(job->pattern, job->hints, subject, (int)subject_length, offset, 0, offsets, 3 * PCRS_MAX_SUBMATCHES)) > 0)
+#endif
{
job->flags |= PCRS_SUCCESS;
matches[i].submatches = submatches;
for (k = 0; k < submatches; k++)
{
- matches[i].submatch_offset[k] = offsets[2 * k];
+ matches[i].submatch_offset[k] = (int)offsets[2 * k];
/* Note: Non-found optional submatches have length -1-(-1)==0 */
matches[i].submatch_length[k] = (size_t)(offsets[2 * k + 1] - offsets[2 * k]);
newsize += (size_t)offsets[0] * (size_t)job->substitute->backref_count[PCRS_MAX_SUBMATCHES];
/* chunk after match */
- matches[i].submatch_offset[PCRS_MAX_SUBMATCHES + 1] = offsets[1];
+ matches[i].submatch_offset[PCRS_MAX_SUBMATCHES + 1] = (int)offsets[1];
matches[i].submatch_length[PCRS_MAX_SUBMATCHES + 1] = subject_length - (size_t)offsets[1] - 1;
newsize += (subject_length - (size_t)offsets[1]) * (size_t)job->substitute->backref_count[PCRS_MAX_SUBMATCHES + 1];
if (NULL == (dummy = (pcrs_match *)realloc(matches, (size_t)max_matches * sizeof(pcrs_match))))
{
free(matches);
- *result = NULL;
return(PCRS_ERR_NOMEM);
}
matches = dummy;
break;
/* Go find the next one */
else
- offset = offsets[1];
+ offset = (int)offsets[1];
}
- /* Pass pcre error through if (bad) failiure */
+ /* Pass pcre error through if (bad) failure */
+#ifdef HAVE_PCRE2
+ if (submatches < PCRE2_ERROR_NOMATCH)
+#else
if (submatches < PCRE_ERROR_NOMATCH)
+#endif
{
free(matches);
+#ifdef HAVE_PCRE2
+ pcre2_match_data_free(pcre2_matches);
+#endif
return submatches;
}
matches_found = i;
* Get memory for the result (must be freed by caller!)
* and append terminating null byte.
*/
- if ((*result = (char *)malloc(newsize + 1)) == NULL)
+ if ((*result = (char *)malloc(newsize + 1
+#ifdef HAVE_PCRE2
+ /*
+ * Work around to prevent invalid reads in the jit code.
+ */
+ + 16
+#endif
+ )) == NULL)
{
free(matches);
+#ifdef HAVE_PCRE2
+ pcre2_match_data_free(pcre2_matches);
+#endif
return PCRS_ERR_NOMEM;
}
else
memcpy(result_offset, subject + offset, subject_length - (size_t)offset);
*result_length = newsize;
+#ifdef HAVE_PCRE2
+ pcre2_match_data_free(pcre2_matches);
+#endif
free(matches);
return matches_found;
* FALSE
*
*********************************************************************/
-int pcrs_job_is_dynamic (char *job)
+int pcrs_job_is_dynamic(char *job)
{
const char delimiter = job[1];
const size_t length = strlen(job);
*********************************************************************/
char *pcrs_execute_single_command(const char *subject, const char *pcrs_command, int *hits)
{
- size_t size;
+ size_t buffer_size, new_size;
char *result = NULL;
pcrs_job *job;
assert(pcrs_command);
*hits = 0;
- size = strlen(subject);
+ buffer_size = strlen(subject);
job = pcrs_compile_command(pcrs_command, hits);
if (NULL != job)
{
- *hits = pcrs_execute(job, subject, size, &result, &size);
+ *hits = pcrs_execute(job, subject, buffer_size, &result, &new_size);
if (*hits < 0)
{
freez(result);
}
-static const char warning[] = "... [too long, truncated]";
/*********************************************************************
*
* Function : pcrs_compile_dynamic_command
{
/* No proper delimiter found */
*error = PCRS_ERR_CMDSYNTAX;
+ freez(pcrs_command_tmp);
return NULL;
}
*/
assert(NULL == strchr(v->name, d));
- ret = snprintf(buf, sizeof(buf), "s%c\\$%s%c%s%cgT", d, v->name, d, v->value, d);
+ ret = snprintf(buf, sizeof(buf), "s%c\\$%s%c%s%cDgT", d, v->name, d, v->value, d);
assert(ret >= 0);
if (ret >= sizeof(buf))
{
* with a truncation message and close the pattern
* properly.
*/
- const size_t trailer_size = sizeof(warning) + 3; /* 3 for d + "gT" */
+ static const char warning[] = "... [too long, truncated]";
+ const size_t trailer_size = sizeof(warning) + 4; /* 4 for d + "DgT" */
char *trailer_start = buf + sizeof(buf) - trailer_size;
- ret = snprintf(trailer_start, trailer_size, "%s%cgT", warning, d);
+ ret = snprintf(trailer_start, trailer_size, "%s%cDgT", warning, d);
assert(ret == trailer_size - 1);
assert(sizeof(buf) == strlen(buf) + 1);
truncation = 1;
projects / privoxy.git / blobdiff