/*
* textregexreplace()
* Return a string matched by a regular expression, with replacement.
*/
datum_t textregexreplace(PG_FUNC_ARGS)
{
text *s = ARG_TEXT_PP(0);
text *p = ARG_TEXT_PP(1);
text *r = ARG_TEXT_PP(2);
text *opt = ARG_TEXT_PP(3);
regex_t *re;
pg_re_flags flags;
parse_re_flags(&flags, opt);
re = RE_compile_and_cache(p, flags.cflags, PG_COLLATION());
RET_TEXT_P(replace_text_regexp(s, (void *)re, r, flags.glob));
}
/*
* textregexreplace()
* Return a string matched by a regular expression, with replacement.
*/
Datum
textregexreplace(PG_FUNCTION_ARGS)
{
text *s = PG_GETARG_TEXT_PP(0);
text *p = PG_GETARG_TEXT_PP(1);
text *r = PG_GETARG_TEXT_PP(2);
text *opt = PG_GETARG_TEXT_PP(3);
regex_t *re;
pg_re_flags flags;
parse_re_flags(&flags, opt);
re = RE_compile_and_cache(p, flags.cflags);
PG_RETURN_TEXT_P(replace_text_regexp(s, (void *) re, r, flags.glob));
}
/*
* setup_regexp_matches --- do the initial matching for regexp_matches()
* or regexp_split()
*
* To avoid having to re-find the compiled pattern on each call, we do
* all the matching in one swoop. The returned regexp_matches_ctx contains
* the locations of all the substrings matching the pattern.
*
* The three bool parameters have only two patterns (one for each caller)
* but it seems clearer to distinguish the functionality this way than to
* key it all off one "is_split" flag.
*/
static regexp_matches_ctx *
setup_regexp_matches(text *orig_str, text *pattern, text *flags,
Oid collation,
bool force_glob, bool use_subpatterns,
bool ignore_degenerate)
{
regexp_matches_ctx *matchctx = palloc0(sizeof(regexp_matches_ctx));
int orig_len;
pg_wchar *wide_str;
int wide_len;
pg_re_flags re_flags;
regex_t *cpattern;
regmatch_t *pmatch;
int pmatch_len;
int array_len;
int array_idx;
int prev_match_end;
int start_search;
/* save original string --- we'll extract result substrings from it */
matchctx->orig_str = orig_str;
/* convert string to pg_wchar form for matching */
orig_len = VARSIZE_ANY_EXHDR(orig_str);
wide_str = (pg_wchar *) palloc(sizeof(pg_wchar) * (orig_len + 1));
wide_len = pg_mb2wchar_with_len(VARDATA_ANY(orig_str), wide_str, orig_len);
/* determine options */
parse_re_flags(&re_flags, flags);
if (force_glob)
{
/* user mustn't specify 'g' for regexp_split */
if (re_flags.glob)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("regexp_split does not support the global option")));
/* but we find all the matches anyway */
re_flags.glob = true;
}
/* set up the compiled pattern */
cpattern = RE_compile_and_cache(pattern, re_flags.cflags, collation);
/* do we want to remember subpatterns? */
if (use_subpatterns && cpattern->re_nsub > 0)
{
matchctx->npatterns = cpattern->re_nsub;
pmatch_len = cpattern->re_nsub + 1;
}
else
{
use_subpatterns = false;
matchctx->npatterns = 1;
pmatch_len = 1;
}
/* temporary output space for RE package */
pmatch = palloc(sizeof(regmatch_t) * pmatch_len);
/* the real output space (grown dynamically if needed) */
array_len = re_flags.glob ? 256 : 32;
matchctx->match_locs = (int *) palloc(sizeof(int) * array_len);
array_idx = 0;
/* search for the pattern, perhaps repeatedly */
prev_match_end = 0;
start_search = 0;
while (RE_wchar_execute(cpattern, wide_str, wide_len, start_search,
pmatch_len, pmatch))
{
/*
* If requested, ignore degenerate matches, which are zero-length
* matches occurring at the start or end of a string or just after a
* previous match.
*/
if (!ignore_degenerate ||
(pmatch[0].rm_so < wide_len &&
pmatch[0].rm_eo > prev_match_end))
{
/* enlarge output space if needed */
while (array_idx + matchctx->npatterns * 2 > array_len)
{
array_len *= 2;
matchctx->match_locs = (int *) repalloc(matchctx->match_locs,
sizeof(int) * array_len);
}
/* save this match's locations */
if (use_subpatterns)
{
int i;
for (i = 1; i <= matchctx->npatterns; i++)
{
matchctx->match_locs[array_idx++] = pmatch[i].rm_so;
matchctx->match_locs[array_idx++] = pmatch[i].rm_eo;
}
}
else
{
matchctx->match_locs[array_idx++] = pmatch[0].rm_so;
matchctx->match_locs[array_idx++] = pmatch[0].rm_eo;
}
matchctx->nmatches++;
}
prev_match_end = pmatch[0].rm_eo;
/* if not glob, stop after one match */
if (!re_flags.glob)
break;
/*
* Advance search position. Normally we start the next search at the
* end of the previous match; but if the match was of zero length, we
* have to advance by one character, or we'd just find the same match
* again.
*/
start_search = prev_match_end;
if (pmatch[0].rm_so == pmatch[0].rm_eo)
start_search++;
if (start_search > wide_len)
break;
}
/* Clean up temp storage */
pfree(wide_str);
pfree(pmatch);
return matchctx;
}
/*
* setup_regexp_matches --- do the initial matching for regexp_matches()
* or regexp_split()
*
* To avoid having to re-find the compiled pattern on each call, we do
* all the matching in one swoop. The returned regexp_matches_ctx contains
* the locations of all the substrings matching the pattern.
*
* The four bool parameters have only two patterns (one for matching, one for
* splitting) but it seems clearer to distinguish the functionality this way
* than to key it all off one "is_split" flag. We don't currently assume that
* fetching_unmatched is exclusive of fetching the matched text too; if it's
* set, the conversion buffer is large enough to fetch any single matched or
* unmatched string, but not any larger substring. (In practice, when splitting
* the matches are usually small anyway, and it didn't seem worth complicating
* the code further.)
*/
static regexp_matches_ctx *
setup_regexp_matches(text *orig_str, text *pattern, text *flags,
Oid collation,
bool force_glob,
bool use_subpatterns,
bool ignore_degenerate,
bool fetching_unmatched)
{
regexp_matches_ctx *matchctx = palloc0(sizeof(regexp_matches_ctx));
int eml = pg_database_encoding_max_length();
int orig_len;
pg_wchar *wide_str;
int wide_len;
pg_re_flags re_flags;
regex_t *cpattern;
regmatch_t *pmatch;
int pmatch_len;
int array_len;
int array_idx;
int prev_match_end;
int prev_valid_match_end;
int start_search;
int maxlen = 0; /* largest fetch length in characters */
/* save original string --- we'll extract result substrings from it */
matchctx->orig_str = orig_str;
/* convert string to pg_wchar form for matching */
orig_len = VARSIZE_ANY_EXHDR(orig_str);
wide_str = (pg_wchar *) palloc(sizeof(pg_wchar) * (orig_len + 1));
wide_len = pg_mb2wchar_with_len(VARDATA_ANY(orig_str), wide_str, orig_len);
/* determine options */
parse_re_flags(&re_flags, flags);
if (force_glob)
{
/* user mustn't specify 'g' for regexp_split */
if (re_flags.glob)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("regexp_split does not support the global option")));
/* but we find all the matches anyway */
re_flags.glob = true;
}
/* set up the compiled pattern */
cpattern = RE_compile_and_cache(pattern, re_flags.cflags, collation);
/* do we want to remember subpatterns? */
if (use_subpatterns && cpattern->re_nsub > 0)
{
matchctx->npatterns = cpattern->re_nsub;
pmatch_len = cpattern->re_nsub + 1;
}
else
{
use_subpatterns = false;
matchctx->npatterns = 1;
pmatch_len = 1;
}
/* temporary output space for RE package */
pmatch = palloc(sizeof(regmatch_t) * pmatch_len);
/*
* the real output space (grown dynamically if needed)
*
* use values 2^n-1, not 2^n, so that we hit the limit at 2^28-1 rather
* than at 2^27
*/
array_len = re_flags.glob ? 255 : 31;
matchctx->match_locs = (int *) palloc(sizeof(int) * array_len);
array_idx = 0;
/* search for the pattern, perhaps repeatedly */
prev_match_end = 0;
prev_valid_match_end = 0;
start_search = 0;
while (RE_wchar_execute(cpattern, wide_str, wide_len, start_search,
pmatch_len, pmatch))
{
/*
* If requested, ignore degenerate matches, which are zero-length
* matches occurring at the start or end of a string or just after a
* previous match.
*/
if (!ignore_degenerate ||
(pmatch[0].rm_so < wide_len &&
pmatch[0].rm_eo > prev_match_end))
{
/* enlarge output space if needed */
while (array_idx + matchctx->npatterns * 2 + 1 > array_len)
{
array_len += array_len + 1; /* 2^n-1 => 2^(n+1)-1 */
if (array_len > MaxAllocSize/sizeof(int))
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("too many regular expression matches")));
matchctx->match_locs = (int *) repalloc(matchctx->match_locs,
sizeof(int) * array_len);
}
/* save this match's locations */
if (use_subpatterns)
{
int i;
for (i = 1; i <= matchctx->npatterns; i++)
{
int so = pmatch[i].rm_so;
int eo = pmatch[i].rm_eo;
matchctx->match_locs[array_idx++] = so;
matchctx->match_locs[array_idx++] = eo;
if (so >= 0 && eo >= 0 && (eo - so) > maxlen)
maxlen = (eo - so);
}
}
else
{
int so = pmatch[0].rm_so;
int eo = pmatch[0].rm_eo;
matchctx->match_locs[array_idx++] = so;
matchctx->match_locs[array_idx++] = eo;
if (so >= 0 && eo >= 0 && (eo - so) > maxlen)
maxlen = (eo - so);
}
matchctx->nmatches++;
/*
* check length of unmatched portion between end of previous valid
* (nondegenerate, or degenerate but not ignored) match and start
* of current one
*/
if (fetching_unmatched &&
pmatch[0].rm_so >= 0 &&
(pmatch[0].rm_so - prev_valid_match_end) > maxlen)
maxlen = (pmatch[0].rm_so - prev_valid_match_end);
prev_valid_match_end = pmatch[0].rm_eo;
}
prev_match_end = pmatch[0].rm_eo;
/* if not glob, stop after one match */
if (!re_flags.glob)
break;
/*
* Advance search position. Normally we start the next search at the
* end of the previous match; but if the match was of zero length, we
* have to advance by one character, or we'd just find the same match
* again.
*/
start_search = prev_match_end;
if (pmatch[0].rm_so == pmatch[0].rm_eo)
start_search++;
if (start_search > wide_len)
break;
}
/*
* check length of unmatched portion between end of last match and end of
* input string
*/
if (fetching_unmatched &&
(wide_len - prev_valid_match_end) > maxlen)
maxlen = (wide_len - prev_valid_match_end);
/*
* Keep a note of the end position of the string for the benefit of
* splitting code.
*/
matchctx->match_locs[array_idx] = wide_len;
if (eml > 1)
{
int64 maxsiz = eml * (int64) maxlen;
int conv_bufsiz;
/*
* Make the conversion buffer large enough for any substring of
* interest.
*
* Worst case: assume we need the maximum size (maxlen*eml), but take
* advantage of the fact that the original string length in bytes is an
* upper bound on the byte length of any fetched substring (and we know
* that len+1 is safe to allocate because the varlena header is longer
* than 1 byte).
*/
if (maxsiz > orig_len)
conv_bufsiz = orig_len + 1;
else
conv_bufsiz = maxsiz + 1; /* safe since maxsiz < 2^30 */
matchctx->conv_buf = palloc(conv_bufsiz);
matchctx->conv_bufsiz = conv_bufsiz;
matchctx->wide_str = wide_str;
}
else
{
/* No need to keep the wide string if we're in a single-byte charset. */
pfree(wide_str);
matchctx->wide_str = NULL;
matchctx->conv_buf = NULL;
matchctx->conv_bufsiz = 0;
}
/* Clean up temp storage */
pfree(pmatch);
return matchctx;
}
