-const char pcrs_rcs[] = "$Id: pcrs.c,v 1.12 2001/08/18 11:35:00 oes Exp $";
+const char pcrs_rcs[] = "$Id: pcrs.c,v 1.19.2.4 2005/05/07 21:50:55 david__schmidt Exp $";
/*********************************************************************
*
- * File : $Source: /cvsroot/ijbswa/current/pcrs.c,v $
+ * File : $Source: /cvsroot/ijbswa/current/Attic/pcrs.c,v $
*
* Purpose : pcrs is a supplement to the pcre library by Philip Hazel
* <ph10@cam.ac.uk> and adds Perl-style substitution. That
*
* Revisions :
* $Log: pcrs.c,v $
+ * 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
+ *
+ * Revision 1.15 2001/09/20 16:11:06 steudten
+ *
+ * Add casting for some string functions.
+ *
+ * Revision 1.14 2001/09/09 21:41:57 oes
+ * Fixing yet another silly bug
+ *
+ * Revision 1.13 2001/09/06 14:05:59 oes
+ * Fixed silly bug
+ *
* Revision 1.12 2001/08/18 11:35:00 oes
* - Introduced pcrs_strerror()
* - made some NULL arguments non-fatal
*********************************************************************/
\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>
const char pcrs_h_rcs[] = PCRS_H_VERSION;
+/*
+ * Internal prototypes
+ */
+
+static int pcrs_parse_perl_options(const char *optstring, int *flags);
+static pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag,
+ int capturecount, int *errptr);
+
/*********************************************************************
*
* Function : pcrs_strerror
* Returns : option integer suitable for pcre
*
*********************************************************************/
-int pcrs_parse_perl_options(const char *optstring, int *flags)
+static int pcrs_parse_perl_options(const char *optstring, int *flags)
{
size_t i;
int rc = 0;
if (NULL == optstring) return 0;
- for (i=0; i < strlen(optstring); i++)
+ for (i = 0; i < strlen(optstring); i++)
{
switch(optstring[i])
{
case 's': rc |= PCRE_DOTALL; break;
case 'x': rc |= PCRE_EXTENDED; break;
case 'U': rc |= PCRE_UNGREEDY; break;
- case 'T': *flags |= PCRS_TRIVIAL; break;
+ case 'T': *flags |= PCRS_TRIVIAL; break;
default: break;
}
}
* the reason.
*
*********************************************************************/
-pcrs_substitute *pcrs_compile_replacement(const char *replacement, int trivialflag, int capturecount, int *errptr)
+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;
*errptr = PCRS_ERR_NOMEM;
return NULL;
}
- memset(r, '\0', length + 1);
+ memset(text, '\0', length + 1);
/*
*/
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';
text[k++] = '\f';
break;
case 'a':
- text[k++] = '\a';
+ text[k++] = 7;
break;
case 'e':
text[k++] = 27;
}
/* 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];
/* Numerical backreferences */
- if (isdigit(replacement[i + 1]))
+ if (isdigit((int)replacement[i + 1]))
{
- while (i < length && isdigit(replacement[++i]))
+ while (i < (int)length && isdigit((int)replacement[++i]))
{
r->backref[l] = r->backref[l] * 10 + replacement[i] - 48;
}
*********************************************************************/
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
*
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
*
int pcrs_execute(pcrs_job *job, char *subject, size_t subject_length, char **result, size_t *result_length)
{
int offsets[3 * PCRS_MAX_SUBMATCHES],
- offset, i, k,
+ offset,
+ i, k,
matches_found,
- newsize,
submatches,
max_matches = PCRS_MAX_MATCH_INIT;
+ size_t newsize;
pcrs_match *matches, *dummy;
char *result_offset;
* 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];
matches[i].submatch_length[k] = 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] * job->substitute->backref_count[k];
}
/* plus replacement text size minus match text size */
newsize += strlen(job->substitute->text) - matches[i].submatch_length[0];
/* Storage for matches exhausted? -> Extend! */
if (++i >= max_matches)
{
- max_matches *= PCRS_MAX_MATCH_GROW;
+ max_matches = (int)(max_matches * PCRS_MAX_MATCH_GROW);
if (NULL == (dummy = (pcrs_match *)realloc(matches, max_matches * sizeof(pcrs_match))))
{
free(matches);
/* Don't loop on empty matches */
if (offsets[1] == offset)
- if (offset < subject_length)
+ if ((size_t)offset < subject_length)
offset++;
else
break;
/*
- * 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. */