-const char pcrs_rcs[] = "$Id: pcrs.c,v 1.16 2001/11/30 21:32:14 jongfoster Exp $";
+const char pcrs_rcs[] = "$Id: pcrs.c,v 1.22 2006/12/24 17:34:20 fabiankeil Exp $";
/*********************************************************************
*
*
* Revisions :
* $Log: pcrs.c,v $
+ * Revision 1.22 2006/12/24 17:34:20 fabiankeil
+ * Add pcrs_strerror() message for PCRE_ERROR_MATCHLIMIT
+ * and give a hint why an error code might be unknown.
+ *
+ * Catch NULL subjects early in pcrs_execute().
+ *
+ * Revision 1.21 2006/07/18 14:48:47 david__schmidt
+ * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
+ * with what was really the latest development (the v_3_0_branch branch)
+ *
+ * Revision 1.19.2.4 2005/05/07 21:50:55 david__schmidt
+ * A few memory leaks plugged (mostly on error paths)
+ *
+ * Revision 1.19.2.3 2003/12/04 12:32:45 oes
+ * Append a trailing nullbyte to result to facilitate string processing
+ *
+ * Revision 1.19.2.2 2002/10/08 16:22:28 oes
+ * Bugfix: Need to check validity of backreferences explicitly,
+ * because when max_matches are reached and matches is expanded,
+ * realloc() does not zero the memory. Fixes Bug # 606227
+ *
+ * Revision 1.19.2.1 2002/08/10 11:23:40 oes
+ * Include prce.h via project.h, where the appropriate
+ * source will have been selected
+ *
+ * Revision 1.19 2002/03/08 14:47:48 oes
+ * Cosmetics
+ *
+ * Revision 1.18 2002/03/08 14:17:14 oes
+ * Fixing -Wconversion warnings
+ *
+ * Revision 1.17 2002/03/08 13:45:48 oes
+ * Hiding internal functions
+ *
* Revision 1.16 2001/11/30 21:32:14 jongfoster
* Fixing signed/unsigned comparison (Andreas please check this!)
* One tab->space
*********************************************************************/
\f
-#include <pcre.h>
+/*
+ * Include project.h just so that the right pcre.h gets
+ * included from there
+ */
+#include "project.h"
+
#include <string.h>
#include <ctype.h>
case PCRE_ERROR_NOSUBSTRING: return "(pcre:) Fire in power supply";
case PCRE_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";
+#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_ERR_BADJOB: return "(pcrs:) Bad job - NULL job, pattern or substitute";
case PCRS_WARN_BADREF: return "(pcrs:) Backreference out of range";
- /* What's that? */
- default: return "Unknown error";
+ /*
+ * 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?";
}
}
/* error >= 0: No error */
if (NULL == optstring) return 0;
- for (i=0; i < strlen(optstring); i++)
+ for (i = 0; i < strlen(optstring); i++)
{
switch(optstring[i])
{
*********************************************************************/
static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag, int capturecount, int *errptr)
{
- int length, i, k, l, quoted;
+ int i, k, l, quoted;
+ size_t length;
char *text;
pcrs_substitute *r;
if (trivialflag)
{
text = strncpy(text, replacement, length + 1);
- k = length;
+ k = (int)length;
}
/*
*/
else
{
- while(i < length)
+ while (i < (int)length)
{
/* Quoting */
if (replacement[i] == '\\')
{
if (replacement[i+1] && strchr("tnrfae0", replacement[i+1]))
{
- switch(replacement[++i])
+ switch (replacement[++i])
{
case 't':
text[k++] = '\t';
}
/* Backreferences */
- if (replacement[i] == '$' && !quoted && i < length - 1)
+ if (replacement[i] == '$' && !quoted && i < (int)(length - 1))
{
char *symbol, symbols[] = "'`+&";
- r->block_length[l] = k - r->block_offset[l];
+ r->block_length[l] = (size_t)(k - r->block_offset[l]);
/* Numerical backreferences */
- if (isdigit((int) replacement[i + 1]))
+ if (isdigit((int)replacement[i + 1]))
{
- while (i < length && isdigit((int) replacement[++i]))
+ while (i < (int)length && isdigit((int)replacement[++i]))
{
r->backref[l] = r->backref[l] * 10 + replacement[i] - 48;
}
*/
r->text = text;
r->backrefs = l;
- r->block_length[l] = k - r->block_offset[l];
+ r->block_length[l] = (size_t)(k - r->block_offset[l]);
return r;
*********************************************************************/
pcrs_job *pcrs_compile_command(const char *command, int *errptr)
{
- int i, k, l, limit, quoted = FALSE;
+ int i, k, l, quoted = FALSE;
+ size_t limit;
char delimiter;
char *tokens[4];
pcrs_job *newjob;
tokens[l] = (char *) malloc(limit + 1);
- for (i=0; i <= limit; i++)
+ for (i = 0; i <= (int)limit; i++)
{
if (command[i] == delimiter && !quoted)
* the joblist to the subject.
* The subject itself is left untouched, memory for the result
* is malloc()ed and it is the caller's responsibility to free
- * the result when it's no longer needed.
+ * the result when it's no longer needed.
+ *
+ * Note: For convenient string handling, a null byte is
+ * appended to the result. It does not count towards the
+ * result_length, though.
+ *
*
* Parameters :
* 1 : joblist = the chained list of pcrs_jobs to be executed
* 2 : subject = the subject string
* 3 : subject_length = the subject's length
- * INCLUDING the terminating zero, if string!
* 4 : result = char** for returning the result
* 5 : result_length = size_t* for returning the result's length
*
* Returns : On success, the number of substitutions that were made.
* May be > 1 if job->flags contained PCRS_GLOBAL
- * On failiure, the (negative) pcre error code describing the
- * failiure, which may be translated to text using pcrs_strerror().
+ * On failure, the (negative) pcre error code describing the
+ * failure, which may be translated to text using pcrs_strerror().
*
*********************************************************************/
int pcrs_execute_list(pcrs_job *joblist, char *subject, size_t subject_length, char **result, size_t *result_length)
{
pcrs_job *job;
- char *old, *new;
+ char *old, *new = NULL;
int hits, total_hits;
old = subject;
* is malloc()ed and it is the caller's responsibility to free
* the result when it's no longer needed.
*
+ * Note: For convenient string handling, a null byte is
+ * appended to the result. It does not count towards the
+ * result_length, though.
+ *
* Parameters :
* 1 : job = the pcrs_job to be executed
* 2 : subject = the subject (== original) string
* 3 : subject_length = the subject's length
- * INCLUDING the terminating zero, if string!
* 4 : result = char** for returning the result
* 5 : result_length = size_t* for returning the result's length
*
* Returns : On success, the number of substitutions that were made.
* May be > 1 if job->flags contained PCRS_GLOBAL
- * On failiure, the (negative) pcre error code describing the
- * failiure, which may be translated to text using pcrs_strerror().
+ * On failure, the (negative) pcre error code describing the
+ * failure, which may be translated to text using pcrs_strerror().
*
*********************************************************************/
int pcrs_execute(pcrs_job *job, char *subject, size_t subject_length, char **result, size_t *result_length)
offset,
i, k,
matches_found,
- newsize,
submatches,
max_matches = PCRS_MAX_MATCH_INIT;
+ size_t newsize;
pcrs_match *matches, *dummy;
char *result_offset;
/*
* Sanity check & memory allocation
*/
- if (job == NULL || job->pattern == NULL || job->substitute == NULL)
+ if (job == NULL || job->pattern == NULL || job->substitute == NULL || NULL == subject)
{
*result = NULL;
return(PCRS_ERR_BADJOB);
}
- if (NULL == (matches = (pcrs_match *)malloc(max_matches * sizeof(pcrs_match))))
+ if (NULL == (matches = (pcrs_match *)malloc((size_t)max_matches * sizeof(pcrs_match))))
{
*result = NULL;
return(PCRS_ERR_NOMEM);
}
- memset(matches, '\0', max_matches * sizeof(pcrs_match));
+ memset(matches, '\0', (size_t)max_matches * sizeof(pcrs_match));
/*
* Find the pattern and calculate the space
* requirements for the result
*/
- newsize=subject_length;
+ newsize = subject_length;
- while ((submatches = pcre_exec(job->pattern, job->hints, subject, subject_length, offset, 0, offsets, 3 * PCRS_MAX_SUBMATCHES)) > 0)
+ while ((submatches = pcre_exec(job->pattern, job->hints, subject, (int)subject_length, offset, 0, offsets, 3 * PCRS_MAX_SUBMATCHES)) > 0)
{
job->flags |= PCRS_SUCCESS;
matches[i].submatches = submatches;
- for (k=0; k < submatches; k++)
+ for (k = 0; k < submatches; k++)
{
matches[i].submatch_offset[k] = offsets[2 * k];
/* Note: Non-found optional submatches have length -1-(-1)==0 */
- matches[i].submatch_length[k] = offsets[2 * k + 1] - offsets[2 * k];
+ matches[i].submatch_length[k] = (size_t)(offsets[2 * k + 1] - offsets[2 * k]);
/* reserve mem for each submatch as often as it is ref'd */
- newsize += matches[i].submatch_length[k] * job->substitute->backref_count[k];
+ newsize += matches[i].submatch_length[k] * (size_t)job->substitute->backref_count[k];
}
/* plus replacement text size minus match text size */
newsize += strlen(job->substitute->text) - matches[i].submatch_length[0];
/* chunk before match */
matches[i].submatch_offset[PCRS_MAX_SUBMATCHES] = 0;
- matches[i].submatch_length[PCRS_MAX_SUBMATCHES] = offsets[0];
- newsize += offsets[0] * job->substitute->backref_count[PCRS_MAX_SUBMATCHES];
+ matches[i].submatch_length[PCRS_MAX_SUBMATCHES] = (size_t)offsets[0];
+ 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_length[PCRS_MAX_SUBMATCHES + 1] = subject_length - offsets[1] - 1;
- newsize += (subject_length - offsets[1]) * job->substitute->backref_count[PCRS_MAX_SUBMATCHES + 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];
/* Storage for matches exhausted? -> Extend! */
if (++i >= max_matches)
{
- max_matches = (int) (max_matches * PCRS_MAX_MATCH_GROW);
- if (NULL == (dummy = (pcrs_match *)realloc(matches, max_matches * sizeof(pcrs_match))))
+ max_matches = (int)(max_matches * PCRS_MAX_MATCH_GROW);
+ if (NULL == (dummy = (pcrs_match *)realloc(matches, (size_t)max_matches * sizeof(pcrs_match))))
{
free(matches);
*result = NULL;
/* Don't loop on empty matches */
if (offsets[1] == offset)
- /* FIXME: is offset an int or a size_t? Previous line compares
- * against int, the next one compares against size_t.
- */
if ((size_t)offset < subject_length)
offset++;
else
/*
- * Get memory for the result
+ * Get memory for the result (must be freed by caller!)
+ * and append terminating null byte.
*/
- if ((*result = (char *)malloc(newsize)) == NULL) /* must be free()d by caller */
+ if ((*result = (char *)malloc(newsize + 1)) == NULL)
{
free(matches);
return PCRS_ERR_NOMEM;
}
+ else
+ {
+ (*result)[newsize] = '\0';
+ }
/*
offset = 0;
result_offset = *result;
- for (i=0; i < matches_found; i++)
+ for (i = 0; i < matches_found; i++)
{
/* copy the chunk preceding the match */
- memcpy(result_offset, subject + offset, matches[i].submatch_offset[0] - offset);
+ memcpy(result_offset, subject + offset, (size_t)(matches[i].submatch_offset[0] - offset));
result_offset += matches[i].submatch_offset[0] - offset;
/* For every segment of the substitute.. */
- for (k=0; k <= job->substitute->backrefs; k++)
+ for (k = 0; k <= job->substitute->backrefs; k++)
{
/* ...copy its text.. */
memcpy(result_offset, job->substitute->text + job->substitute->block_offset[k], job->substitute->block_length[k]);
if (k != job->substitute->backrefs
/* ..in legal range.. */
&& job->substitute->backref[k] < PCRS_MAX_SUBMATCHES + 2
- /* ..and referencing a nonempty match.. */
+ /* ..and referencing a real submatch.. */
+ && job->substitute->backref[k] < matches[i].submatches
+ /* ..that is nonempty.. */
&& matches[i].submatch_length[job->substitute->backref[k]] > 0)
{
/* ..copy the submatch that is ref'd. */
result_offset += matches[i].submatch_length[job->substitute->backref[k]];
}
}
- offset = matches[i].submatch_offset[0] + matches[i].submatch_length[0];
+ offset = matches[i].submatch_offset[0] + (int)matches[i].submatch_length[0];
}
/* Copy the rest. */
- memcpy(result_offset, subject + offset, subject_length - offset);
+ memcpy(result_offset, subject + offset, subject_length - (size_t)offset);
*result_length = newsize;
free(matches);