"lookbehind assertion is not fixed length\0"
"malformed number or name after (?(\0"
"conditional group contains more than two branches\0"
- "assertion expected after (?(\0"
+ "assertion expected after (?( or (?(?C)\0"
"(?R or (?[+-]digits must be followed by )\0"
/* 30 */
"unknown POSIX class name\0"
pcre_uint32 ec;
pcre_uchar mcbuffer[8];
+ /* Come here to restart the loop without advancing the pointer. */
+
+ REDO_LOOP:
+
/* Get next character in the pattern */
c = *ptr;
goto FAILED;
}
- /* If in \Q...\E, check for the end; if not, we have a literal */
+ /* If in \Q...\E, check for the end; if not, we have a literal. Otherwise an
+ isolated \E is ignored. */
- if (inescq && c != CHAR_NULL)
+ if (c != CHAR_NULL)
{
if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
{
ptr++;
continue;
}
- else
+ else if (inescq)
{
if (previous_callout != NULL)
{
}
goto NORMAL_CHAR;
}
- /* Control does not reach here. */
+
+ /* Check for the start of a \Q...\E sequence. We must do this here rather
+ than later in case it is immediately followed by \E, which turns it into a
+ "do nothing" sequence. */
+
+ if (c == CHAR_BACKSLASH && ptr[1] == CHAR_Q)
+ {
+ inescq = TRUE;
+ ptr++;
+ continue;
+ }
}
- /* In extended mode, skip white space and comments. We need a loop in order
- to check for more white space and more comments after a comment. */
+ /* In extended mode, skip white space and comments. */
if ((options & PCRE_EXTENDED) != 0)
{
- for (;;)
+ const pcre_uchar *wscptr = ptr;
+ while (MAX_255(c) && (cd->ctypes[c] & ctype_space) != 0) c = *(++ptr);
+ if (c == CHAR_NUMBER_SIGN)
{
- while (MAX_255(c) && (cd->ctypes[c] & ctype_space) != 0) c = *(++ptr);
- if (c != CHAR_NUMBER_SIGN) break;
ptr++;
while (*ptr != CHAR_NULL)
{
if (utf) FORWARDCHAR(ptr);
#endif
}
- c = *ptr; /* Either NULL or the char after a newline */
}
+
+ /* If we skipped any characters, restart the loop. Otherwise, we didn't see
+ a comment. */
+
+ if (ptr > wscptr) goto REDO_LOOP;
+ }
+
+ /* Skip over (?# comments. We need to do this here because we want to know if
+ the next thing is a quantifier, and these comments may come between an item
+ and its quantifier. */
+
+ if (c == CHAR_LEFT_PARENTHESIS && ptr[1] == CHAR_QUESTION_MARK &&
+ ptr[2] == CHAR_NUMBER_SIGN)
+ {
+ ptr += 3;
+ while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
+ if (*ptr == CHAR_NULL)
+ {
+ *errorcodeptr = ERR18;
+ goto FAILED;
+ }
+ continue;
}
/* See if the next thing is a quantifier. */
if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_STARTWORD, 6) == 0)
{
nestptr = ptr + 7;
- ptr = sub_start_of_word - 1;
- continue;
+ ptr = sub_start_of_word;
+ goto REDO_LOOP;
}
if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_ENDWORD, 6) == 0)
{
nestptr = ptr + 7;
- ptr = sub_end_of_word - 1;
- continue;
+ ptr = sub_end_of_word;
+ goto REDO_LOOP;
}
/* Handle a real character class. */
ptr = tempptr + 1;
continue;
- /* For the other POSIX classes (ascii, xdigit) we are going to fall
- through to the non-UCP case and build a bit map for characters with
- code points less than 256. If we are in a negated POSIX class
- within a non-negated overall class, characters with code points
- greater than 255 must all match. In the special case where we have
- not yet generated any xclass data, and this is the final item in
- the overall class, we need do nothing: later on, the opcode
+ /* For the other POSIX classes (ascii, cntrl, xdigit) we are going
+ to fall through to the non-UCP case and build a bit map for
+ characters with code points less than 256. If we are in a negated
+ POSIX class, characters with code points greater than 255 must
+ either all match or all not match. In the special case where we
+ have not yet generated any xclass data, and this is the final item
+ in the overall class, we need do nothing: later on, the opcode
OP_NCLASS will be used to indicate that characters greater than 255
are acceptable. If we have already seen an xclass item or one may
follow (we have to assume that it might if this is not the end of
- the class), explicitly match all wide codepoints. */
+ the class), explicitly list all wide codepoints, which will then
+ either not match or match, depending on whether the class is or is
+ not negated. */
default:
- if (!negate_class && local_negate &&
+ if (local_negate &&
(xclass || tempptr[2] != CHAR_RIGHT_SQUARE_BRACKET))
{
*class_uchardata++ = XCL_RANGE;
case CHAR_LEFT_PARENTHESIS:
ptr++;
- /* First deal with comments. Putting this code right at the start ensures
- that comments have no bad side effects. */
-
- if (ptr[0] == CHAR_QUESTION_MARK && ptr[1] == CHAR_NUMBER_SIGN)
- {
- ptr += 2;
- while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
- if (*ptr == CHAR_NULL)
- {
- *errorcodeptr = ERR18;
- goto FAILED;
- }
- continue;
- }
-
/* Now deal with various "verbs" that can be introduced by '*'. */
if (ptr[0] == CHAR_ASTERISK && (ptr[1] == ':'
for (i = 3;; i++) if (!IS_DIGIT(ptr[i])) break;
if (ptr[i] == CHAR_RIGHT_PARENTHESIS)
tempptr += i + 1;
+
+ /* tempptr should now be pointing to the opening parenthesis of the
+ assertion condition. */
+
+ if (*tempptr != CHAR_LEFT_PARENTHESIS)
+ {
+ *errorcodeptr = ERR28;
+ goto FAILED;
+ }
}
/* For conditions that are assertions, check the syntax, and then exit
issue is fixed "properly" in PCRE2. As PCRE1 is now in maintenance
only mode, we finesse the bug by allowing more memory always. */
- *lengthptr += 2 + 2*LINK_SIZE;
+ *lengthptr += 4 + 4*LINK_SIZE;
/* It is even worse than that. The current reference may be to an
existing named group with a different number (so apparently not
so far in order to get the number. If the name is not found, leave
the value of recno as 0 for a forward reference. */
- else
+ /* This patch (removing "else") fixes a problem when a reference is
+ to multiple identically named nested groups from within the nest.
+ Once again, it is not the "proper" fix, and it results in an
+ over-allocation of memory. */
+
+ /* else */
{
ng = cd->named_groups;
for (i = 0; i < cd->names_found; i++, ng++)
newoptions = (options | set) & (~unset);
/* If the options ended with ')' this is not the start of a nested
- group with option changes, so the options change at this level. If this
- item is right at the start of the pattern, the options can be
- abstracted and made external in the pre-compile phase, and ignored in
- the compile phase. This can be helpful when matching -- for instance in
- caseless checking of required bytes.
-
- If the code pointer is not (cd->start_code + 1 + LINK_SIZE), we are
- definitely *not* at the start of the pattern because something has been
- compiled. In the pre-compile phase, however, the code pointer can have
- that value after the start, because it gets reset as code is discarded
- during the pre-compile. However, this can happen only at top level - if
- we are within parentheses, the starting BRA will still be present. At
- any parenthesis level, the length value can be used to test if anything
- has been compiled at that level. Thus, a test for both these conditions
- is necessary to ensure we correctly detect the start of the pattern in
- both phases.
-
+ group with option changes, so the options change at this level.
If we are not at the pattern start, reset the greedy defaults and the
case value for firstchar and reqchar. */
if (*ptr == CHAR_RIGHT_PARENTHESIS)
{
- if (code == cd->start_code + 1 + LINK_SIZE &&
- (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE))
- {
- cd->external_options = newoptions;
- }
- else
- {
- greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);
- greedy_non_default = greedy_default ^ 1;
- req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
- }
+ greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);
+ greedy_non_default = greedy_default ^ 1;
+ req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
/* Change options at this level, and pass them back for use
in subsequent branches. */
c = ec;
else
{
- if (escape == ESC_Q) /* Handle start of quoted string */
- {
- if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
- ptr += 2; /* avoid empty string */
- else inescq = TRUE;
- continue;
- }
-
- if (escape == ESC_E) continue; /* Perl ignores an orphan \E */
-
/* For metasequences that actually match a character, we disable the
setting of a first character if it hasn't already been set. */
~ian / pcre3.git / blobdiff