summaryrefslogtreecommitdiff
path: root/regexec.c
diff options
context:
space:
mode:
Diffstat (limited to 'regexec.c')
-rw-r--r--regexec.c97
1 files changed, 48 insertions, 49 deletions
diff --git a/regexec.c b/regexec.c
index b0cc6647a0..260505901d 100644
--- a/regexec.c
+++ b/regexec.c
@@ -977,25 +977,24 @@ stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end,
}\
} while(0)
-#define STRING_CMP_IC(case_fold_flag,s1,ps2,len) do {\
- if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len) == 0) \
+#define STRING_CMP_IC(case_fold_flag,s1,ps2,len,text_end) do {\
+ if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len, text_end) == 0) \
goto fail; \
} while(0)
static int string_cmp_ic(OnigEncoding enc, int case_fold_flag,
- UChar* s1, UChar** ps2, int mblen)
+ UChar* s1, UChar** ps2, int mblen, const UChar* text_end)
{
UChar buf1[ONIGENC_MBC_CASE_FOLD_MAXLEN];
UChar buf2[ONIGENC_MBC_CASE_FOLD_MAXLEN];
- UChar *p1, *p2, *end1, *s2, *end2;
+ UChar *p1, *p2, *end1, *s2;
int len1, len2;
s2 = *ps2;
end1 = s1 + mblen;
- end2 = s2 + mblen;
while (s1 < end1) {
- len1 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s1, end1, buf1);
- len2 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s2, end2, buf2);
+ len1 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s1, text_end, buf1);
+ len2 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s2, text_end, buf2);
if (len1 != len2) return 0;
p1 = buf1;
p2 = buf2;
@@ -1019,8 +1018,8 @@ static int string_cmp_ic(OnigEncoding enc, int case_fold_flag,
}\
} while(0)
-#define STRING_CMP_VALUE_IC(case_fold_flag,s1,ps2,len,is_fail) do {\
- if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len) == 0) \
+#define STRING_CMP_VALUE_IC(case_fold_flag,s1,ps2,len,text_end,is_fail) do {\
+ if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len, text_end) == 0) \
is_fail = 1; \
else \
is_fail = 0; \
@@ -1126,7 +1125,7 @@ static int backref_match_at_nested_level(regex_t* reg
if (ignore_case != 0) {
if (string_cmp_ic(reg->enc, case_fold_flag,
- pstart, &ss, (int )(pend - pstart)) == 0)
+ pstart, &ss, (int )(pend - pstart), send) == 0)
return 0; /* or goto next_mem; */
}
else {
@@ -2199,7 +2198,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
n = pend - pstart;
DATA_ENSURE(n);
sprev = s;
- STRING_CMP_IC(case_fold_flag, pstart, &s, n);
+ STRING_CMP_IC(case_fold_flag, pstart, &s, n, end);
while (sprev + (len = enclen(encode, sprev, end)) < s)
sprev += len;
@@ -2271,7 +2270,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
DATA_ENSURE(n);
sprev = s;
swork = s;
- STRING_CMP_VALUE_IC(case_fold_flag, pstart, &swork, n, is_fail);
+ STRING_CMP_VALUE_IC(case_fold_flag, pstart, &swork, n, end, is_fail);
if (is_fail) continue;
s = swork;
while (sprev + (len = enclen(encode, sprev, end)) < s)
@@ -2647,9 +2646,9 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
case OP_LOOK_BEHIND: MOP_IN(OP_LOOK_BEHIND);
GET_LENGTH_INC(tlen, p);
- s = (UChar* )ONIGENC_STEP_BACK(encode, str, s, (int )tlen);
+ s = (UChar* )ONIGENC_STEP_BACK(encode, str, s, end, (int )tlen);
if (IS_NULL(s)) goto fail;
- sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s);
+ sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s, end);
MOP_OUT;
continue;
break;
@@ -2657,7 +2656,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
case OP_PUSH_LOOK_BEHIND_NOT: MOP_IN(OP_PUSH_LOOK_BEHIND_NOT);
GET_RELADDR_INC(addr, p);
GET_LENGTH_INC(tlen, p);
- q = (UChar* )ONIGENC_STEP_BACK(encode, str, s, (int )tlen);
+ q = (UChar* )ONIGENC_STEP_BACK(encode, str, s, end, (int )tlen);
if (IS_NULL(q)) {
/* too short case -> success. ex. /(?<!XXX)a/.match("a")
If you want to change to fail, replace following line. */
@@ -2667,7 +2666,7 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
else {
STACK_PUSH_LOOK_BEHIND_NOT(p + addr, s, sprev);
s = q;
- sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s);
+ sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s, end);
}
MOP_OUT;
continue;
@@ -2780,7 +2779,7 @@ slow_search(OnigEncoding enc, UChar* target, UChar* target_end,
if (target_end == t || memcmp(t, p, target_end - t) == 0)
return s;
}
- s += enclen(enc, s, end);
+ s += enclen(enc, s, text_end);
}
return (UChar* )NULL;
@@ -2843,7 +2842,7 @@ slow_search_backward(OnigEncoding enc, UChar* target, UChar* target_end,
if (s > text_start)
s = (UChar* )text_start;
else
- s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s);
+ s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s, text_end);
while (s >= text) {
if (*s == *target) {
@@ -2857,7 +2856,7 @@ slow_search_backward(OnigEncoding enc, UChar* target, UChar* target_end,
if (t == target_end)
return s;
}
- s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s);
+ s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s, text_end);
}
return (UChar* )NULL;
@@ -2876,14 +2875,14 @@ slow_search_backward_ic(OnigEncoding enc, int case_fold_flag,
if (s > text_start)
s = (UChar* )text_start;
else
- s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s);
+ s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s, text_end);
while (s >= text) {
if (str_lower_case_match(enc, case_fold_flag,
target, target_end, s, text_end))
return s;
- s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s);
+ s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s, text_end);
}
return (UChar* )NULL;
@@ -3018,7 +3017,7 @@ bm_search_backward(regex_t* reg, const UChar* target, const UChar* target_end,
if (text_start < s)
s = text_start;
else
- s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s);
+ s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s, text_end);
while (s >= text) {
p = s;
@@ -3030,7 +3029,7 @@ bm_search_backward(regex_t* reg, const UChar* target, const UChar* target_end,
return (UChar* )s;
s -= reg->int_map_backward[*s];
- s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s);
+ s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s, text_end);
}
return (UChar* )NULL;
@@ -3038,14 +3037,14 @@ bm_search_backward(regex_t* reg, const UChar* target, const UChar* target_end,
static UChar*
map_search(OnigEncoding enc, UChar map[],
- const UChar* text, const UChar* text_range)
+ const UChar* text, const UChar* text_range, const UChar* text_end)
{
const UChar *s = text;
while (s < text_range) {
if (map[*s]) return (UChar* )s;
- s += enclen(enc, s, text_range);
+ s += enclen(enc, s, text_end);
}
return (UChar* )NULL;
}
@@ -3053,14 +3052,14 @@ map_search(OnigEncoding enc, UChar map[],
static UChar*
map_search_backward(OnigEncoding enc, UChar map[],
const UChar* text, const UChar* adjust_text,
- const UChar* text_start)
+ const UChar* text_start, const UChar* text_end)
{
const UChar *s = text_start;
while (s >= text) {
if (map[*s]) return (UChar* )s;
- s = onigenc_get_prev_char_head(enc, adjust_text, s);
+ s = onigenc_get_prev_char_head(enc, adjust_text, s, text_end);
}
return (UChar* )NULL;
}
@@ -3117,7 +3116,7 @@ onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at, On
r = 0;
if (r == 0) {
- prev = (UChar* )onigenc_get_prev_char_head(reg->enc, str, at);
+ prev = (UChar* )onigenc_get_prev_char_head(reg->enc, str, at, end);
r = match_at(reg, str, end,
#ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
end,
@@ -3171,7 +3170,7 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
break;
case ONIG_OPTIMIZE_MAP:
- p = map_search(reg->enc, reg->map, p, range);
+ p = map_search(reg->enc, reg->map, p, range, end);
break;
}
@@ -3190,7 +3189,7 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
case ANCHOR_BEGIN_LINE:
if (!ON_STR_BEGIN(p)) {
prev = onigenc_get_prev_char_head(reg->enc,
- (pprev ? pprev : str), p);
+ (pprev ? pprev : str), p, end);
if (!ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end))
goto retry_gate;
}
@@ -3219,10 +3218,10 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
*low = p;
if (low_prev) {
if (*low > s)
- *low_prev = onigenc_get_prev_char_head(reg->enc, s, p);
+ *low_prev = onigenc_get_prev_char_head(reg->enc, s, p, end);
else
*low_prev = onigenc_get_prev_char_head(reg->enc,
- (pprev ? pprev : str), p);
+ (pprev ? pprev : str), p, end);
}
}
else {
@@ -3230,15 +3229,15 @@ forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
*low = p - reg->dmax;
if (*low > s) {
*low = onigenc_get_right_adjust_char_head_with_prev(reg->enc, s,
- *low, (const UChar** )low_prev);
+ *low, end, (const UChar** )low_prev);
if (low_prev && IS_NULL(*low_prev))
*low_prev = onigenc_get_prev_char_head(reg->enc,
- (pprev ? pprev : s), *low);
+ (pprev ? pprev : s), *low, end);
}
else {
if (low_prev)
*low_prev = onigenc_get_prev_char_head(reg->enc,
- (pprev ? pprev : str), *low);
+ (pprev ? pprev : str), *low, end);
}
}
}
@@ -3301,7 +3300,7 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
break;
case ONIG_OPTIMIZE_MAP:
- p = map_search_backward(reg->enc, reg->map, range, adjrange, p);
+ p = map_search_backward(reg->enc, reg->map, range, adjrange, p, end);
break;
}
@@ -3312,7 +3311,7 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
switch (reg->sub_anchor) {
case ANCHOR_BEGIN_LINE:
if (!ON_STR_BEGIN(p)) {
- prev = onigenc_get_prev_char_head(reg->enc, str, p);
+ prev = onigenc_get_prev_char_head(reg->enc, str, p, end);
if (!ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end)) {
p = prev;
goto retry;
@@ -3336,7 +3335,7 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
&& ! ONIGENC_IS_MBC_CRNL(reg->enc, p, end)
#endif
) {
- p = onigenc_get_prev_char_head(reg->enc, adjrange, p);
+ p = onigenc_get_prev_char_head(reg->enc, adjrange, p, end);
if (IS_NULL(p)) goto fail;
goto retry;
}
@@ -3348,7 +3347,7 @@ backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
if (reg->dmax != ONIG_INFINITE_DISTANCE) {
*low = p - reg->dmax;
*high = p - reg->dmin;
- *high = onigenc_get_right_adjust_char_head(reg->enc, adjrange, *high);
+ *high = onigenc_get_right_adjust_char_head(reg->enc, adjrange, *high, end);
}
#ifdef ONIG_DEBUG_SEARCH
@@ -3506,9 +3505,9 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
if ((OnigDistance )(min_semi_end - start) > reg->anchor_dmax) {
start = min_semi_end - reg->anchor_dmax;
if (start < end)
- start = onigenc_get_right_adjust_char_head(reg->enc, str, start);
+ start = onigenc_get_right_adjust_char_head(reg->enc, str, start, end);
else { /* match with empty at end */
- start = onigenc_get_prev_char_head(reg->enc, str, end);
+ start = onigenc_get_prev_char_head(reg->enc, str, end, end);
}
}
if ((OnigDistance )(max_semi_end - (range - 1)) < reg->anchor_dmin) {
@@ -3523,20 +3522,20 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
}
if ((OnigDistance )(max_semi_end - start) < reg->anchor_dmin) {
start = max_semi_end - reg->anchor_dmin;
- start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, start);
+ start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, start, end);
}
if (range > start) goto mismatch_no_msa;
}
}
else if (reg->anchor & ANCHOR_SEMI_END_BUF) {
- UChar* pre_end = ONIGENC_STEP_BACK(reg->enc, str, end, 1);
+ UChar* pre_end = ONIGENC_STEP_BACK(reg->enc, str, end, end, 1);
max_semi_end = (UChar* )end;
if (ONIGENC_IS_MBC_NEWLINE(reg->enc, pre_end, end)) {
min_semi_end = pre_end;
#ifdef USE_CRNL_AS_LINE_TERMINATOR
- pre_end = ONIGENC_STEP_BACK(reg->enc, str, pre_end, 1);
+ pre_end = ONIGENC_STEP_BACK(reg->enc, str, pre_end, end, 1);
if (IS_NOT_NULL(pre_end) &&
ONIGENC_IS_MBC_CRNL(reg->enc, pre_end, end)) {
min_semi_end = pre_end;
@@ -3594,7 +3593,7 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
s = (UChar* )start;
if (range > start) { /* forward search */
if (s > str)
- prev = onigenc_get_prev_char_head(reg->enc, str, s);
+ prev = onigenc_get_prev_char_head(reg->enc, str, s, end);
else
prev = (UChar* )NULL;
@@ -3670,7 +3669,7 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
UChar *low, *high, *adjrange, *sch_start;
if (range < end)
- adjrange = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, range);
+ adjrange = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, range, end);
else
adjrange = (UChar* )end;
@@ -3687,7 +3686,7 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
s = high;
while (s >= low) {
- prev = onigenc_get_prev_char_head(reg->enc, str, s);
+ prev = onigenc_get_prev_char_head(reg->enc, str, s, end);
MATCH_AND_RETURN_CHECK(orig_start);
s = prev;
}
@@ -3706,7 +3705,7 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
if (sch_start > end) sch_start = (UChar* )end;
else
sch_start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc,
- start, sch_start);
+ start, sch_start, end);
}
}
if (backward_search_range(reg, str, end, sch_start, range, adjrange,
@@ -3715,7 +3714,7 @@ onig_search(regex_t* reg, const UChar* str, const UChar* end,
}
do {
- prev = onigenc_get_prev_char_head(reg->enc, str, s);
+ prev = onigenc_get_prev_char_head(reg->enc, str, s, end);
MATCH_AND_RETURN_CHECK(orig_start);
s = prev;
} while (s >= range);
View Lorry Controller Status