/[pcre]/code/trunk/pcre_compile.c
ViewVC logotype

Diff of /code/trunk/pcre_compile.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 904 by ph10, Mon Jan 23 17:30:49 2012 UTC revision 1363 by ph10, Tue Oct 1 16:54:40 2013 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2012 University of Cambridge             Copyright (c) 1997-2013 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  is also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. We do not need to select pcre16_printint.c specially, because the  library. We do not need to select pcre16_printint.c specially, because the
59  COMPILE_PCREx macro will already be appropriately set. */  COMPILE_PCREx macro will already be appropriately set. */
# Line 68  COMPILE_PCREx macro will already be appr Line 68  COMPILE_PCREx macro will already be appr
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 77  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,
88        pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *,
89        compile_data *, int *);
90    
91    
92    
93  /*************************************************  /*************************************************
94  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 103  kicks in at the same number of forward r Line 115  kicks in at the same number of forward r
115  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)
116  #define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE)  #define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE)
117    
118    /* This value determines the size of the initial vector that is used for
119    remembering named groups during the pre-compile. It is allocated on the stack,
120    but if it is too small, it is expanded using malloc(), in a similar way to the
121    workspace. The value is the number of slots in the list. */
122    
123    #define NAMED_GROUP_LIST_SIZE  20
124    
125  /* The overrun tests check for a slightly smaller size so that they detect the  /* The overrun tests check for a slightly smaller size so that they detect the
126  overrun before it actually does run off the end of the data block. */  overrun before it actually does run off the end of the data block. */
127    
# Line 110  overrun before it actually does run off Line 129  overrun before it actually does run off
129    
130  /* Private flags added to firstchar and reqchar. */  /* Private flags added to firstchar and reqchar. */
131    
132  #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */  #define REQ_CASELESS    (1 << 0)        /* Indicates caselessness */
133  #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */  #define REQ_VARY        (1 << 1)        /* Reqchar followed non-literal item */
134    /* Negative values for the firstchar and reqchar flags */
135    #define REQ_UNSET       (-2)
136    #define REQ_NONE        (-1)
137    
138  /* Repeated character flags. */  /* Repeated character flags. */
139    
# Line 472  static const char error_texts[] = Line 494  static const char error_texts[] =
494    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\0"
495    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"
496    /* 60 */    /* 60 */
497    "(*VERB) not recognized\0"    "(*VERB) not recognized or malformed\0"
498    "number is too big\0"    "number is too big\0"
499    "subpattern name expected\0"    "subpattern name expected\0"
500    "digit expected after (?+\0"    "digit expected after (?+\0"
# Line 489  static const char error_texts[] = Line 511  static const char error_texts[] =
511    "too many forward references\0"    "too many forward references\0"
512    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"    "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
513    "invalid UTF-16 string\0"    "invalid UTF-16 string\0"
514      /* 75 */
515      "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
516      "character value in \\u.... sequence is too large\0"
517      "invalid UTF-32 string\0"
518      "setting UTF is disabled by the application\0"
519    ;    ;
520    
521  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 628  static const pcre_uint8 ebcdic_chartab[] Line 655  static const pcre_uint8 ebcdic_chartab[]
655  #endif  #endif
656    
657    
658  /* Definition to allow mutual recursion */  /* This table is used to check whether auto-possessification is possible
659    between adjacent character-type opcodes. The left-hand (repeated) opcode is
660    used to select the row, and the right-hand opcode is use to select the column.
661    A value of 1 means that auto-possessification is OK. For example, the second
662    value in the first row means that \D+\d can be turned into \D++\d.
663    
664    The Unicode property types (\P and \p) have to be present to fill out the table
665    because of what their opcode values are, but the table values should always be
666    zero because property types are handled separately in the code. The last four
667    columns apply to items that cannot be repeated, so there is no need to have
668    rows for them. Note that OP_DIGIT etc. are generated only when PCRE_UCP is
669    *not* set. When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
670    
671    #define APTROWS (LAST_AUTOTAB_LEFT_OP - FIRST_AUTOTAB_OP + 1)
672    #define APTCOLS (LAST_AUTOTAB_RIGHT_OP - FIRST_AUTOTAB_OP + 1)
673    
674    static const pcre_uint8 autoposstab[APTROWS][APTCOLS] = {
675    /* \D \d \S \s \W \w  . .+ \C \P \p \R \H \h \V \v \X \Z \z  $ $M */
676      { 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* \D */
677      { 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 },  /* \d */
678      { 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 },  /* \S */
679      { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* \s */
680      { 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* \W */
681      { 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 },  /* \w */
682      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* .  */
683      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* .+ */
684      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* \C */
685      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },  /* \P */
686      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },  /* \p */
687      { 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 },  /* \R */
688      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 },  /* \H */
689      { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0 },  /* \h */
690      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0 },  /* \V */
691      { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0 },  /* \v */
692      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }   /* \X */
693    };
694    
695  static BOOL  
696    compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,  /* This table is used to check whether auto-possessification is possible
697      int *, int *, branch_chain *, compile_data *, int *);  between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP). The
698    left-hand (repeated) opcode is used to select the row, and the right-hand
699    opcode is used to select the column. The values are as follows:
700    
701      0   Always return FALSE (never auto-possessify)
702      1   Character groups are distinct (possessify if both are OP_PROP)
703      2   Check character categories in the same group (general or particular)
704      3   TRUE if the two opcodes are not the same (PROP vs NOTPROP)
705    
706      4   Check left general category vs right particular category
707      5   Check right general category vs left particular category
708    
709      6   Left alphanum vs right general category
710      7   Left space vs right general category
711      8   Left word vs right general category
712    
713      9   Right alphanum vs left general category
714     10   Right space vs left general category
715     11   Right word vs left general category
716    
717     12   Left alphanum vs right particular category
718     13   Left space vs right particular category
719     14   Left word vs right particular category
720    
721     15   Right alphanum vs left particular category
722     16   Right space vs left particular category
723     17   Right word vs left particular category
724    */
725    
726    static const pcre_uint8 propposstab[PT_TABSIZE][PT_TABSIZE] = {
727    /* ANY LAMP GC  PC  SC ALNUM SPACE PXSPACE WORD CLIST UCNC */
728      { 0,  0,  0,  0,  0,    0,    0,      0,   0,    0,   0 },  /* PT_ANY */
729      { 0,  3,  0,  0,  0,    3,    1,      1,   0,    0,   0 },  /* PT_LAMP */
730      { 0,  0,  2,  4,  0,    9,   10,     10,  11,    0,   0 },  /* PT_GC */
731      { 0,  0,  5,  2,  0,   15,   16,     16,  17,    0,   0 },  /* PT_PC */
732      { 0,  0,  0,  0,  2,    0,    0,      0,   0,    0,   0 },  /* PT_SC */
733      { 0,  3,  6, 12,  0,    3,    1,      1,   0,    0,   0 },  /* PT_ALNUM */
734      { 0,  1,  7, 13,  0,    1,    3,      3,   1,    0,   0 },  /* PT_SPACE */
735      { 0,  1,  7, 13,  0,    1,    3,      3,   1,    0,   0 },  /* PT_PXSPACE */
736      { 0,  0,  8, 14,  0,    0,    1,      1,   3,    0,   0 },  /* PT_WORD */
737      { 0,  0,  0,  0,  0,    0,    0,      0,   0,    0,   0 },  /* PT_CLIST */
738      { 0,  0,  0,  0,  0,    0,    0,      0,   0,    0,   3 }   /* PT_UCNC */
739    };
740    
741    /* This table is used to check whether auto-possessification is possible
742    between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP) when one
743    specifies a general category and the other specifies a particular category. The
744    row is selected by the general category and the column by the particular
745    category. The value is 1 if the particular category is not part of the general
746    category. */
747    
748    static const pcre_uint8 catposstab[7][30] = {
749    /* Cc Cf Cn Co Cs Ll Lm Lo Lt Lu Mc Me Mn Nd Nl No Pc Pd Pe Pf Pi Po Ps Sc Sk Sm So Zl Zp Zs */
750      { 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 },  /* C */
751      { 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 },  /* L */
752      { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 },  /* M */
753      { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 },  /* N */
754      { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1 },  /* P */
755      { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1 },  /* S */
756      { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0 }   /* Z */
757    };
758    
759    /* This table is used when checking ALNUM, (PX)SPACE, SPACE, and WORD against
760    a general or particular category. The properties in each row are those
761    that apply to the character set in question. Duplication means that a little
762    unnecessary work is done when checking, but this keeps things much simpler
763    because they can all use the same code. For more details see the comment where
764    this table is used.
765    
766    Note: SPACE and PXSPACE used to be different because Perl excluded VT from
767    "space", but from Perl 5.18 it's included, so both categories are treated the
768    same here. */
769    
770    static const pcre_uint8 posspropstab[3][4] = {
771      { ucp_L, ucp_N, ucp_N, ucp_Nl },  /* ALNUM, 3rd and 4th values redundant */
772      { ucp_Z, ucp_Z, ucp_C, ucp_Cc },  /* SPACE and PXSPACE, 2nd value redundant */
773      { ucp_L, ucp_N, ucp_P, ucp_Po }   /* WORD */
774    };
775    
776    
777    
# Line 655  find_error_text(int n) Line 794  find_error_text(int n)
794  const char *s = error_texts;  const char *s = error_texts;
795  for (; n > 0; n--)  for (; n > 0; n--)
796    {    {
797    while (*s++ != 0) {};    while (*s++ != CHAR_NULL) {};
798    if (*s == 0) return "Error text not found (please report)";    if (*s == CHAR_NULL) return "Error text not found (please report)";
799    }    }
800  return s;  return s;
801  }  }
802    
803    
804    
805  /*************************************************  /*************************************************
806  *           Expand the workspace                 *  *           Expand the workspace                 *
807  *************************************************/  *************************************************/
# Line 739  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 879  return (*p == CHAR_RIGHT_CURLY_BRACKET);
879  *************************************************/  *************************************************/
880    
881  /* This function is called when a \ has been encountered. It either returns a  /* This function is called when a \ has been encountered. It either returns a
882  positive value for a simple escape such as \n, or a negative value which  positive value for a simple escape such as \n, or 0 for a data character
883  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group n is returned as
884  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  negative n. When UTF-8 is enabled, a positive value greater than 255 may
885  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
886  ptr is pointing at the \. On exit, it is on the final character of the escape  On entry,ptr is pointing at the \. On exit, it is on the final character of the
887  sequence.  escape sequence.
888    
889  Arguments:  Arguments:
890    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
891      chptr          points to the data character
892    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
893    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
894    options        the options bits    options        the options bits
895    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
896    
897  Returns:         zero or positive => a data character  Returns:         zero => a data character
898                   negative => a special escape sequence                   positive => a special escape sequence
899                     negative => a back reference
900                   on error, errorcodeptr is set                   on error, errorcodeptr is set
901  */  */
902    
903  static int  static int
904  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
905    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
906  {  {
907  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
908  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
909  const pcre_uchar *ptr = *ptrptr + 1;  const pcre_uchar *ptr = *ptrptr + 1;
910  pcre_int32 c;  pcre_uint32 c;
911    int escape = 0;
912  int i;  int i;
913    
914  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
# Line 773  ptr--;                            /* Set Line 916  ptr--;                            /* Set
916    
917  /* If backslash is at the end of the pattern, it's an error. */  /* If backslash is at the end of the pattern, it's an error. */
918    
919  if (c == 0) *errorcodeptr = ERR1;  if (c == CHAR_NULL) *errorcodeptr = ERR1;
920    
921  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
922  in a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
# Line 782  Otherwise further processing may be requ Line 925  Otherwise further processing may be requ
925  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
926  /* Not alphanumeric */  /* Not alphanumeric */
927  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
928  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0)
929      { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
930    
931  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
932  /* Not alphanumeric */  /* Not alphanumeric */
933  else if (c < 'a' || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
934  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
935  #endif  #endif
936    
937  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
# Line 795  else if ((i = escapes[c - 0x48]) != 0) Line 939  else if ((i = escapes[c - 0x48]) != 0)
939  else  else
940    {    {
941    const pcre_uchar *oldptr;    const pcre_uchar *oldptr;
942    BOOL braced, negated;    BOOL braced, negated, overflow;
943      int s;
944    
945    switch (c)    switch (c)
946      {      {
# Line 820  else Line 965  else
965          c = 0;          c = 0;
966          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
967            {            {
968            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
969  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
970            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
971            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 829  else Line 974  else
974            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
975  #endif  #endif
976            }            }
977    
978    #if defined COMPILE_PCRE8
979            if (c > (utf ? 0x10ffffU : 0xffU))
980    #elif defined COMPILE_PCRE16
981            if (c > (utf ? 0x10ffffU : 0xffffU))
982    #elif defined COMPILE_PCRE32
983            if (utf && c > 0x10ffffU)
984    #endif
985              {
986              *errorcodeptr = ERR76;
987              }
988            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
989          }          }
990        }        }
991      else      else
# Line 855  else Line 1012  else
1012      (3) For Oniguruma compatibility we also support \g followed by a name or a      (3) For Oniguruma compatibility we also support \g followed by a name or a
1013      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
1014      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
1015      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
1016    
1017      case CHAR_g:      case CHAR_g:
1018      if (isclass) break;      if (isclass) break;
1019      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
1020        {        {
1021        c = -ESC_g;        escape = ESC_g;
1022        break;        break;
1023        }        }
1024    
# Line 870  else Line 1027  else
1027      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1028        {        {
1029        const pcre_uchar *p;        const pcre_uchar *p;
1030        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
1031          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
1032        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
1033          {          {
1034          c = -ESC_k;          escape = ESC_k;
1035          break;          break;
1036          }          }
1037        braced = TRUE;        braced = TRUE;
# Line 890  else Line 1047  else
1047      else negated = FALSE;      else negated = FALSE;
1048    
1049      /* The integer range is limited by the machine's int representation. */      /* The integer range is limited by the machine's int representation. */
1050      c = 0;      s = 0;
1051        overflow = FALSE;
1052      while (IS_DIGIT(ptr[1]))      while (IS_DIGIT(ptr[1]))
1053        {        {
1054        if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */        if (s > INT_MAX / 10 - 1) /* Integer overflow */
1055          {          {
1056          c = -1;          overflow = TRUE;
1057          break;          break;
1058          }          }
1059        c = c * 10 + *(++ptr) - CHAR_0;        s = s * 10 + (int)(*(++ptr) - CHAR_0);
1060        }        }
1061      if (((unsigned int)c) > INT_MAX) /* Integer overflow */      if (overflow) /* Integer overflow */
1062        {        {
1063        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
1064          ptr++;          ptr++;
# Line 914  else Line 1072  else
1072        break;        break;
1073        }        }
1074    
1075      if (c == 0)      if (s == 0)
1076        {        {
1077        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
1078        break;        break;
# Line 922  else Line 1080  else
1080    
1081      if (negated)      if (negated)
1082        {        {
1083        if (c > bracount)        if (s > bracount)
1084          {          {
1085          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
1086          break;          break;
1087          }          }
1088        c = bracount - (c - 1);        s = bracount - (s - 1);
1089        }        }
1090    
1091      c = -(ESC_REF + c);      escape = -s;
1092      break;      break;
1093    
1094      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 952  else Line 1110  else
1110        {        {
1111        oldptr = ptr;        oldptr = ptr;
1112        /* The integer range is limited by the machine's int representation. */        /* The integer range is limited by the machine's int representation. */
1113        c -= CHAR_0;        s = (int)(c -CHAR_0);
1114          overflow = FALSE;
1115        while (IS_DIGIT(ptr[1]))        while (IS_DIGIT(ptr[1]))
1116          {          {
1117          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */          if (s > INT_MAX / 10 - 1) /* Integer overflow */
1118            {            {
1119            c = -1;            overflow = TRUE;
1120            break;            break;
1121            }            }
1122          c = c * 10 + *(++ptr) - CHAR_0;          s = s * 10 + (int)(*(++ptr) - CHAR_0);
1123          }          }
1124        if (((unsigned int)c) > INT_MAX) /* Integer overflow */        if (overflow) /* Integer overflow */
1125          {          {
1126          while (IS_DIGIT(ptr[1]))          while (IS_DIGIT(ptr[1]))
1127            ptr++;            ptr++;
1128          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1129          break;          break;
1130          }          }
1131        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1132          {          {
1133          c = -(ESC_REF + c);          escape = -s;
1134          break;          break;
1135          }          }
1136        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 1018  else Line 1177  else
1177          c = 0;          c = 0;
1178          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1179            {            {
1180            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1181  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1182            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1183            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 1036  else Line 1195  else
1195        const pcre_uchar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
1196    
1197        c = 0;        c = 0;
1198          overflow = FALSE;
1199        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1200          {          {
1201          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1202          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1203    
1204    #ifdef COMPILE_PCRE32
1205            if (c >= 0x10000000l) { overflow = TRUE; break; }
1206    #endif
1207    
1208  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1209          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1210          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 1049  else Line 1213  else
1213          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1214  #endif  #endif
1215    
1216  #ifdef COMPILE_PCRE8  #if defined COMPILE_PCRE8
1217          if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }          if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1218  #else  #elif defined COMPILE_PCRE16
1219  #ifdef COMPILE_PCRE16          if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1220          if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }  #elif defined COMPILE_PCRE32
1221  #endif          if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1222  #endif  #endif
1223          }          }
1224    
1225        if (c < 0)        if (overflow)
1226          {          {
1227          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1228          *errorcodeptr = ERR34;          *errorcodeptr = ERR34;
# Line 1080  else Line 1244  else
1244      c = 0;      c = 0;
1245      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1246        {        {
1247        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1248        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1249  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1250        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 1099  else Line 1263  else
1263    
1264      case CHAR_c:      case CHAR_c:
1265      c = *(++ptr);      c = *(++ptr);
1266      if (c == 0)      if (c == CHAR_NULL)
1267        {        {
1268        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1269        break;        break;
# Line 1139  else Line 1303  else
1303  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1304  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1305    
1306  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1307       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1308    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1309    
1310  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1311    
1312  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1313    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1314    
1315  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1316    
1317  *ptrptr = ptr;  *ptrptr = ptr;
1318  return c;  *chptr = c;
1319    return escape;
1320  }  }
1321    
1322    
# Line 1169  escape sequence. Line 1334  escape sequence.
1334  Argument:  Argument:
1335    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1336    negptr         points to a boolean that is set TRUE for negation else FALSE    negptr         points to a boolean that is set TRUE for negation else FALSE
1337    dptr           points to an int that is set to the detailed property value    ptypeptr       points to an unsigned int that is set to the type value
1338      pdataptr       points to an unsigned int that is set to the detailed property value
1339    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1340    
1341  Returns:         type value from ucp_type_table, or -1 for an invalid type  Returns:         TRUE if the type value was found, or FALSE for an invalid type
1342  */  */
1343    
1344  static int  static BOOL
1345  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1346      unsigned int *pdataptr, int *errorcodeptr)
1347  {  {
1348  int c, i, bot, top;  pcre_uchar c;
1349    int i, bot, top;
1350  const pcre_uchar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1351  pcre_uchar name[32];  pcre_uchar name[32];
1352    
1353  c = *(++ptr);  c = *(++ptr);
1354  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1355    
1356  *negptr = FALSE;  *negptr = FALSE;
1357    
# Line 1200  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1368  if (c == CHAR_LEFT_CURLY_BRACKET)
1368    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1369      {      {
1370      c = *(++ptr);      c = *(++ptr);
1371      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1372      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1373      name[i] = c;      name[i] = c;
1374      }      }
# Line 1225  top = PRIV(utt_size); Line 1393  top = PRIV(utt_size);
1393    
1394  while (bot < top)  while (bot < top)
1395    {    {
1396      int r;
1397    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1398    c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1399    if (c == 0)    if (r == 0)
1400      {      {
1401      *dptr = PRIV(utt)[i].value;      *ptypeptr = PRIV(utt)[i].type;
1402      return PRIV(utt)[i].type;      *pdataptr = PRIV(utt)[i].value;
1403        return TRUE;
1404      }      }
1405    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1406    }    }
1407    
1408  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1409  *ptrptr = ptr;  *ptrptr = ptr;
1410  return -1;  return FALSE;
1411    
1412  ERROR_RETURN:  ERROR_RETURN:
1413  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1414  *ptrptr = ptr;  *ptrptr = ptr;
1415  return -1;  return FALSE;
1416  }  }
1417  #endif  #endif
1418    
1419    
1420    
   
1421  /*************************************************  /*************************************************
1422  *         Read repeat counts                     *  *         Read repeat counts                     *
1423  *************************************************/  *************************************************/
# Line 1277  int max = -1; Line 1446  int max = -1;
1446  /* Read the minimum value and do a paranoid check: a negative value indicates  /* Read the minimum value and do a paranoid check: a negative value indicates
1447  an integer overflow. */  an integer overflow. */
1448    
1449  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1450  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1451    {    {
1452    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1292  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1461  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1461    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1462      {      {
1463      max = 0;      max = 0;
1464      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1465      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1466        {        {
1467        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1317  return p; Line 1486  return p;
1486    
1487    
1488  /*************************************************  /*************************************************
 *  Subroutine for finding forward reference      *  
 *************************************************/  
   
 /* This recursive function is called only from find_parens() below. The  
 top-level call starts at the beginning of the pattern. All other calls must  
 start at a parenthesis. It scans along a pattern's text looking for capturing  
 subpatterns, and counting them. If it finds a named pattern that matches the  
 name it is given, it returns its number. Alternatively, if the name is NULL, it  
 returns when it reaches a given numbered subpattern. Recursion is used to keep  
 track of subpatterns that reset the capturing group numbers - the (?| feature.  
   
 This function was originally called only from the second pass, in which we know  
 that if (?< or (?' or (?P< is encountered, the name will be correctly  
 terminated because that is checked in the first pass. There is now one call to  
 this function in the first pass, to check for a recursive back reference by  
 name (so that we can make the whole group atomic). In this case, we need check  
 only up to the current position in the pattern, and that is still OK because  
 and previous occurrences will have been checked. To make this work, the test  
 for "end of pattern" is a check against cd->end_pattern in the main loop,  
 instead of looking for a binary zero. This means that the special first-pass  
 call can adjust cd->end_pattern temporarily. (Checks for binary zero while  
 processing items within the loop are OK, because afterwards the main loop will  
 terminate.)  
   
 Arguments:  
   ptrptr       address of the current character pointer (updated)  
   cd           compile background data  
   name         name to seek, or NULL if seeking a numbered subpattern  
   lorn         name length, or subpattern number if name is NULL  
   xmode        TRUE if we are in /x mode  
   utf          TRUE if we are in UTF-8 / UTF-16 mode  
   count        pointer to the current capturing subpattern number (updated)  
   
 Returns:       the number of the named subpattern, or -1 if not found  
 */  
   
 static int  
 find_parens_sub(pcre_uchar **ptrptr, compile_data *cd, const pcre_uchar *name, int lorn,  
   BOOL xmode, BOOL utf, int *count)  
 {  
 pcre_uchar *ptr = *ptrptr;  
 int start_count = *count;  
 int hwm_count = start_count;  
 BOOL dup_parens = FALSE;  
   
 /* If the first character is a parenthesis, check on the type of group we are  
 dealing with. The very first call may not start with a parenthesis. */  
   
 if (ptr[0] == CHAR_LEFT_PARENTHESIS)  
   {  
   /* Handle specials such as (*SKIP) or (*UTF8) etc. */  
   
   if (ptr[1] == CHAR_ASTERISK) ptr += 2;  
   
   /* Handle a normal, unnamed capturing parenthesis. */  
   
   else if (ptr[1] != CHAR_QUESTION_MARK)  
     {  
     *count += 1;  
     if (name == NULL && *count == lorn) return *count;  
     ptr++;  
     }  
   
   /* All cases now have (? at the start. Remember when we are in a group  
   where the parenthesis numbers are duplicated. */  
   
   else if (ptr[2] == CHAR_VERTICAL_LINE)  
     {  
     ptr += 3;  
     dup_parens = TRUE;  
     }  
   
   /* Handle comments; all characters are allowed until a ket is reached. */  
   
   else if (ptr[2] == CHAR_NUMBER_SIGN)  
     {  
     for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;  
     goto FAIL_EXIT;  
     }  
   
   /* Handle a condition. If it is an assertion, just carry on so that it  
   is processed as normal. If not, skip to the closing parenthesis of the  
   condition (there can't be any nested parens). */  
   
   else if (ptr[2] == CHAR_LEFT_PARENTHESIS)  
     {  
     ptr += 2;  
     if (ptr[1] != CHAR_QUESTION_MARK)  
       {  
       while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;  
       if (*ptr != 0) ptr++;  
       }  
     }  
   
   /* Start with (? but not a condition. */  
   
   else  
     {  
     ptr += 2;  
     if (*ptr == CHAR_P) ptr++;                      /* Allow optional P */  
   
     /* We have to disambiguate (?<! and (?<= from (?<name> for named groups */  
   
     if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&  
         ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)  
       {  
       int term;  
       const pcre_uchar *thisname;  
       *count += 1;  
       if (name == NULL && *count == lorn) return *count;  
       term = *ptr++;  
       if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;  
       thisname = ptr;  
       while (*ptr != term) ptr++;  
       if (name != NULL && lorn == ptr - thisname &&  
           STRNCMP_UC_UC(name, thisname, lorn) == 0)  
         return *count;  
       term++;  
       }  
     }  
   }  
   
 /* Past any initial parenthesis handling, scan for parentheses or vertical  
 bars. Stop if we get to cd->end_pattern. Note that this is important for the  
 first-pass call when this value is temporarily adjusted to stop at the current  
 position. So DO NOT change this to a test for binary zero. */  
   
 for (; ptr < cd->end_pattern; ptr++)  
   {  
   /* Skip over backslashed characters and also entire \Q...\E */  
   
   if (*ptr == CHAR_BACKSLASH)  
     {  
     if (*(++ptr) == 0) goto FAIL_EXIT;  
     if (*ptr == CHAR_Q) for (;;)  
       {  
       while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};  
       if (*ptr == 0) goto FAIL_EXIT;  
       if (*(++ptr) == CHAR_E) break;  
       }  
     continue;  
     }  
   
   /* Skip over character classes; this logic must be similar to the way they  
   are handled for real. If the first character is '^', skip it. Also, if the  
   first few characters (either before or after ^) are \Q\E or \E we skip them  
   too. This makes for compatibility with Perl. Note the use of STR macros to  
   encode "Q\\E" so that it works in UTF-8 on EBCDIC platforms. */  
   
   if (*ptr == CHAR_LEFT_SQUARE_BRACKET)  
     {  
     BOOL negate_class = FALSE;  
     for (;;)  
       {  
       if (ptr[1] == CHAR_BACKSLASH)  
         {  
         if (ptr[2] == CHAR_E)  
           ptr+= 2;  
         else if (STRNCMP_UC_C8(ptr + 2,  
                  STR_Q STR_BACKSLASH STR_E, 3) == 0)  
           ptr += 4;  
         else  
           break;  
         }  
       else if (!negate_class && ptr[1] == CHAR_CIRCUMFLEX_ACCENT)  
         {  
         negate_class = TRUE;  
         ptr++;  
         }  
       else break;  
       }  
   
     /* If the next character is ']', it is a data character that must be  
     skipped, except in JavaScript compatibility mode. */  
   
     if (ptr[1] == CHAR_RIGHT_SQUARE_BRACKET &&  
         (cd->external_options & PCRE_JAVASCRIPT_COMPAT) == 0)  
       ptr++;  
   
     while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)  
       {  
       if (*ptr == 0) return -1;  
       if (*ptr == CHAR_BACKSLASH)  
         {  
         if (*(++ptr) == 0) goto FAIL_EXIT;  
         if (*ptr == CHAR_Q) for (;;)  
           {  
           while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};  
           if (*ptr == 0) goto FAIL_EXIT;  
           if (*(++ptr) == CHAR_E) break;  
           }  
         continue;  
         }  
       }  
     continue;  
     }  
   
   /* Skip comments in /x mode */  
   
   if (xmode && *ptr == CHAR_NUMBER_SIGN)  
     {  
     ptr++;  
     while (*ptr != 0)  
       {  
       if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }  
       ptr++;  
 #ifdef SUPPORT_UTF  
       if (utf) FORWARDCHAR(ptr);  
 #endif  
       }  
     if (*ptr == 0) goto FAIL_EXIT;  
     continue;  
     }  
   
   /* Check for the special metacharacters */  
   
   if (*ptr == CHAR_LEFT_PARENTHESIS)  
     {  
     int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);  
     if (rc > 0) return rc;  
     if (*ptr == 0) goto FAIL_EXIT;  
     }  
   
   else if (*ptr == CHAR_RIGHT_PARENTHESIS)  
     {  
     if (dup_parens && *count < hwm_count) *count = hwm_count;  
     goto FAIL_EXIT;  
     }  
   
   else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)  
     {  
     if (*count > hwm_count) hwm_count = *count;  
     *count = start_count;  
     }  
   }  
   
 FAIL_EXIT:  
 *ptrptr = ptr;  
 return -1;  
 }  
   
   
   
   
 /*************************************************  
 *       Find forward referenced subpattern       *  
 *************************************************/  
   
 /* This function scans along a pattern's text looking for capturing  
 subpatterns, and counting them. If it finds a named pattern that matches the  
 name it is given, it returns its number. Alternatively, if the name is NULL, it  
 returns when it reaches a given numbered subpattern. This is used for forward  
 references to subpatterns. We used to be able to start this scan from the  
 current compiling point, using the current count value from cd->bracount, and  
 do it all in a single loop, but the addition of the possibility of duplicate  
 subpattern numbers means that we have to scan from the very start, in order to  
 take account of such duplicates, and to use a recursive function to keep track  
 of the different types of group.  
   
 Arguments:  
   cd           compile background data  
   name         name to seek, or NULL if seeking a numbered subpattern  
   lorn         name length, or subpattern number if name is NULL  
   xmode        TRUE if we are in /x mode  
   utf          TRUE if we are in UTF-8 / UTF-16 mode  
   
 Returns:       the number of the found subpattern, or -1 if not found  
 */  
   
 static int  
 find_parens(compile_data *cd, const pcre_uchar *name, int lorn, BOOL xmode,  
   BOOL utf)  
 {  
 pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;  
 int count = 0;  
 int rc;  
   
 /* If the pattern does not start with an opening parenthesis, the first call  
 to find_parens_sub() will scan right to the end (if necessary). However, if it  
 does start with a parenthesis, find_parens_sub() will return when it hits the  
 matching closing parens. That is why we have to have a loop. */  
   
 for (;;)  
   {  
   rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);  
   if (rc > 0 || *ptr++ == 0) break;  
   }  
   
 return rc;  
 }  
   
   
   
   
 /*************************************************  
1489  *      Find first significant op code            *  *      Find first significant op code            *
1490  *************************************************/  *************************************************/
1491    
# Line 1666  for (;;) Line 1540  for (;;)
1540    
1541    
1542    
   
1543  /*************************************************  /*************************************************
1544  *        Find the fixed length of a branch       *  *        Find the fixed length of a branch       *
1545  *************************************************/  *************************************************/
# Line 1684  and doing the check at the end; a flag s Line 1557  and doing the check at the end; a flag s
1557    
1558  Arguments:  Arguments:
1559    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1560    utf      TRUE in UTF-8 / UTF-16 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1561    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1562    cd       the "compile data" structure    cd       the "compile data" structure
1563    
# Line 1710  for (;;) Line 1583  for (;;)
1583    {    {
1584    int d;    int d;
1585    pcre_uchar *ce, *cs;    pcre_uchar *ce, *cs;
1586    register int op = *cc;    register pcre_uchar op = *cc;
1587    
1588    switch (op)    switch (op)
1589      {      {
# Line 1830  for (;;) Line 1703  for (;;)
1703      case OP_EXACTI:      case OP_EXACTI:
1704      case OP_NOTEXACT:      case OP_NOTEXACT:
1705      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1706      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1707      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1708  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
1709      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
# Line 1839  for (;;) Line 1712  for (;;)
1712    
1713      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1714      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1715      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1716          cc += 2;
1717      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1718      break;      break;
1719    
# Line 1874  for (;;) Line 1748  for (;;)
1748    
1749      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1750    
 #if defined SUPPORT_UTF || defined COMPILE_PCRE16  
     case OP_XCLASS:  
     cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];  
     /* Fall through */  
 #endif  
   
1751      case OP_CLASS:      case OP_CLASS:
1752      case OP_NCLASS:      case OP_NCLASS:
1753    #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1754        case OP_XCLASS:
1755        /* The original code caused an unsigned overflow in 64 bit systems,
1756        so now we use a conditional statement. */
1757        if (op == OP_XCLASS)
1758          cc += GET(cc, 1);
1759        else
1760          cc += PRIV(OP_lengths)[OP_CLASS];
1761    #else
1762      cc += PRIV(OP_lengths)[OP_CLASS];      cc += PRIV(OP_lengths)[OP_CLASS];
1763    #endif
1764    
1765      switch (*cc)      switch (*cc)
1766        {        {
# Line 1897  for (;;) Line 1775  for (;;)
1775        case OP_CRRANGE:        case OP_CRRANGE:
1776        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1777        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1778        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1779        cc += 1 + 2 * IMM2_SIZE;        cc += 1 + 2 * IMM2_SIZE;
1780        break;        break;
1781    
# Line 1964  for (;;) Line 1842  for (;;)
1842      case OP_QUERYI:      case OP_QUERYI:
1843      case OP_REF:      case OP_REF:
1844      case OP_REFI:      case OP_REFI:
1845        case OP_DNREF:
1846        case OP_DNREFI:
1847      case OP_SBRA:      case OP_SBRA:
1848      case OP_SBRAPOS:      case OP_SBRAPOS:
1849      case OP_SCBRA:      case OP_SCBRA:
# Line 2000  for (;;) Line 1880  for (;;)
1880    
1881    
1882    
   
1883  /*************************************************  /*************************************************
1884  *    Scan compiled regex for specific bracket    *  *    Scan compiled regex for specific bracket    *
1885  *************************************************/  *************************************************/
# Line 2013  length. Line 1892  length.
1892    
1893  Arguments:  Arguments:
1894    code        points to start of expression    code        points to start of expression
1895    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
1896    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
1897    
1898  Returns:      pointer to the opcode for the bracket, or NULL if not found  Returns:      pointer to the opcode for the bracket, or NULL if not found
# Line 2024  PRIV(find_bracket)(const pcre_uchar *cod Line 1903  PRIV(find_bracket)(const pcre_uchar *cod
1903  {  {
1904  for (;;)  for (;;)
1905    {    {
1906    register int c = *code;    register pcre_uchar c = *code;
1907    
1908    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
1909    
# Line 2047  for (;;) Line 1926  for (;;)
1926    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
1927             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
1928      {      {
1929      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
1930      if (n == number) return (pcre_uchar *)code;      if (n == number) return (pcre_uchar *)code;
1931      code += PRIV(OP_lengths)[c];      code += PRIV(OP_lengths)[c];
1932      }      }
# Line 2077  for (;;) Line 1956  for (;;)
1956        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
1957        case OP_TYPEEXACT:        case OP_TYPEEXACT:
1958        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
1959        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
1960          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
1961        break;        break;
1962    
1963        case OP_MARK:        case OP_MARK:
1964        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
1965        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
1966        case OP_THEN_ARG:        case OP_THEN_ARG:
1967        code += code[1];        code += code[1];
1968        break;        break;
# Line 2100  for (;;) Line 1976  for (;;)
1976    a multi-byte character. The length in the table is a minimum, so we have to    a multi-byte character. The length in the table is a minimum, so we have to
1977    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
1978    
1979  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1980      if (utf) switch(c)      if (utf) switch(c)
1981        {        {
1982        case OP_CHAR:        case OP_CHAR:
# Line 2152  instance of OP_RECURSE. Line 2028  instance of OP_RECURSE.
2028    
2029  Arguments:  Arguments:
2030    code        points to start of expression    code        points to start of expression
2031    utf         TRUE in UTF-8 / UTF-16 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2032    
2033  Returns:      pointer to the opcode for OP_RECURSE, or NULL if not found  Returns:      pointer to the opcode for OP_RECURSE, or NULL if not found
2034  */  */
# Line 2162  find_recurse(const pcre_uchar *code, BOO Line 2038  find_recurse(const pcre_uchar *code, BOO
2038  {  {
2039  for (;;)  for (;;)
2040    {    {
2041    register int c = *code;    register pcre_uchar c = *code;
2042    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2043    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2044    
# Line 2197  for (;;) Line 2073  for (;;)
2073        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2074        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2075        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2076        if (code[1 + IMM2_SIZE] == OP_PROP        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2077          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;          code += 2;
2078        break;        break;
2079    
2080        case OP_MARK:        case OP_MARK:
2081        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2082        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2083        case OP_THEN_ARG:        case OP_THEN_ARG:
2084        code += code[1];        code += code[1];
2085        break;        break;
# Line 2220  for (;;) Line 2093  for (;;)
2093      by a multi-byte character. The length in the table is a minimum, so we have      by a multi-byte character. The length in the table is a minimum, so we have
2094      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2095    
2096  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2097      if (utf) switch(c)      if (utf) switch(c)
2098        {        {
2099        case OP_CHAR:        case OP_CHAR:
2100        case OP_CHARI:        case OP_CHARI:
2101          case OP_NOT:
2102          case OP_NOTI:
2103        case OP_EXACT:        case OP_EXACT:
2104        case OP_EXACTI:        case OP_EXACTI:
2105          case OP_NOTEXACT:
2106          case OP_NOTEXACTI:
2107        case OP_UPTO:        case OP_UPTO:
2108        case OP_UPTOI:        case OP_UPTOI:
2109          case OP_NOTUPTO:
2110          case OP_NOTUPTOI:
2111        case OP_MINUPTO:        case OP_MINUPTO:
2112        case OP_MINUPTOI:        case OP_MINUPTOI:
2113          case OP_NOTMINUPTO:
2114          case OP_NOTMINUPTOI:
2115        case OP_POSUPTO:        case OP_POSUPTO:
2116        case OP_POSUPTOI:        case OP_POSUPTOI:
2117          case OP_NOTPOSUPTO:
2118          case OP_NOTPOSUPTOI:
2119        case OP_STAR:        case OP_STAR:
2120        case OP_STARI:        case OP_STARI:
2121          case OP_NOTSTAR:
2122          case OP_NOTSTARI:
2123        case OP_MINSTAR:        case OP_MINSTAR:
2124        case OP_MINSTARI:        case OP_MINSTARI:
2125          case OP_NOTMINSTAR:
2126          case OP_NOTMINSTARI:
2127        case OP_POSSTAR:        case OP_POSSTAR:
2128        case OP_POSSTARI:        case OP_POSSTARI:
2129          case OP_NOTPOSSTAR:
2130          case OP_NOTPOSSTARI:
2131        case OP_PLUS:        case OP_PLUS:
2132        case OP_PLUSI:        case OP_PLUSI:
2133          case OP_NOTPLUS:
2134          case OP_NOTPLUSI:
2135        case OP_MINPLUS:        case OP_MINPLUS:
2136        case OP_MINPLUSI:        case OP_MINPLUSI:
2137          case OP_NOTMINPLUS:
2138          case OP_NOTMINPLUSI:
2139        case OP_POSPLUS:        case OP_POSPLUS:
2140        case OP_POSPLUSI:        case OP_POSPLUSI:
2141          case OP_NOTPOSPLUS:
2142          case OP_NOTPOSPLUSI:
2143        case OP_QUERY:        case OP_QUERY:
2144        case OP_QUERYI:        case OP_QUERYI:
2145          case OP_NOTQUERY:
2146          case OP_NOTQUERYI:
2147        case OP_MINQUERY:        case OP_MINQUERY:
2148        case OP_MINQUERYI:        case OP_MINQUERYI:
2149          case OP_NOTMINQUERY:
2150          case OP_NOTMINQUERYI:
2151        case OP_POSQUERY:        case OP_POSQUERY:
2152        case OP_POSQUERYI:        case OP_POSQUERYI:
2153          case OP_NOTPOSQUERY:
2154          case OP_NOTPOSQUERYI:
2155        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2156        break;        break;
2157        }        }
# Line 2278  bracket whose current branch will alread Line 2179  bracket whose current branch will alread
2179  Arguments:  Arguments:
2180    code        points to start of search    code        points to start of search
2181    endcode     points to where to stop    endcode     points to where to stop
2182    utf         TRUE if in UTF-8 / UTF-16 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2183    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2184      recurses    chain of recurse_check to catch mutual recursion
2185    
2186  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2187  */  */
2188    
2189    typedef struct recurse_check {
2190      struct recurse_check *prev;
2191      const pcre_uchar *group;
2192    } recurse_check;
2193    
2194  static BOOL  static BOOL
2195  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2196    BOOL utf, compile_data *cd)    BOOL utf, compile_data *cd, recurse_check *recurses)
2197  {  {
2198  register int c;  register pcre_uchar c;
2199    recurse_check this_recurse;
2200    
2201  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2202       code < endcode;       code < endcode;
2203       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
# Line 2316  for (code = first_significant_code(code Line 2225  for (code = first_significant_code(code
2225    
2226    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2227      {      {
2228      const pcre_uchar *scode;      const pcre_uchar *scode = cd->start_code + GET(code, 1);
2229      BOOL empty_branch;      BOOL empty_branch;
2230    
2231      /* Test for forward reference */      /* Test for forward reference or uncompleted reference. This is disabled
2232        when called to scan a completed pattern by setting cd->start_workspace to
2233        NULL. */
2234    
2235        if (cd->start_workspace != NULL)
2236          {
2237          const pcre_uchar *tcode;
2238          for (tcode = cd->start_workspace; tcode < cd->hwm; tcode += LINK_SIZE)
2239            if ((int)GET(tcode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2240          if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2241          }
2242    
2243        /* If we are scanning a completed pattern, there are no forward references
2244        and all groups are complete. We need to detect whether this is a recursive
2245        call, as otherwise there will be an infinite loop. If it is a recursion,
2246        just skip over it. Simple recursions are easily detected. For mutual
2247        recursions we keep a chain on the stack. */
2248    
2249      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      else
2250        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        {
2251          recurse_check *r = recurses;
2252          const pcre_uchar *endgroup = scode;
2253    
2254      /* Not a forward reference, test for completed backward reference */        do endgroup += GET(endgroup, 1); while (*endgroup == OP_ALT);
2255          if (code >= scode && code <= endgroup) continue;  /* Simple recursion */
2256    
2257      empty_branch = FALSE;        for (r = recurses; r != NULL; r = r->prev)
2258      scode = cd->start_code + GET(code, 1);          if (r->group == scode) break;
2259      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */        if (r != NULL) continue;   /* Mutual recursion */
2260          }
2261    
2262        /* Completed reference; scan the referenced group, remembering it on the
2263        stack chain to detect mutual recursions. */
2264    
2265      /* Completed backwards reference */      empty_branch = FALSE;
2266        this_recurse.prev = recurses;
2267        this_recurse.group = scode;
2268    
2269      do      do
2270        {        {
2271        if (could_be_empty_branch(scode, endcode, utf, cd))        if (could_be_empty_branch(scode, endcode, utf, cd, &this_recurse))
2272          {          {
2273          empty_branch = TRUE;          empty_branch = TRUE;
2274          break;          break;
# Line 2390  for (code = first_significant_code(code Line 2324  for (code = first_significant_code(code
2324        empty_branch = FALSE;        empty_branch = FALSE;
2325        do        do
2326          {          {
2327          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd, NULL))
2328            empty_branch = TRUE;            empty_branch = TRUE;
2329          code += GET(code, 1);          code += GET(code, 1);
2330          }          }
# Line 2448  for (code = first_significant_code(code Line 2382  for (code = first_significant_code(code
2382    
2383      /* Opcodes that must match a character */      /* Opcodes that must match a character */
2384    
2385        case OP_ANY:
2386        case OP_ALLANY:
2387        case OP_ANYBYTE:
2388    
2389      case OP_PROP:      case OP_PROP:
2390      case OP_NOTPROP:      case OP_NOTPROP:
2391        case OP_ANYNL:
2392    
2393        case OP_NOT_HSPACE:
2394        case OP_HSPACE:
2395        case OP_NOT_VSPACE:
2396        case OP_VSPACE:
2397      case OP_EXTUNI:      case OP_EXTUNI:
2398    
2399      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2400      case OP_DIGIT:      case OP_DIGIT:
2401      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2402      case OP_WHITESPACE:      case OP_WHITESPACE:
2403      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2404      case OP_WORDCHAR:      case OP_WORDCHAR:
2405      case OP_ANY:  
     case OP_ALLANY:  
     case OP_ANYBYTE:  
2406      case OP_CHAR:      case OP_CHAR:
2407      case OP_CHARI:      case OP_CHARI:
2408      case OP_NOT:      case OP_NOT:
2409      case OP_NOTI:      case OP_NOTI:
2410    
2411      case OP_PLUS:      case OP_PLUS:
2412        case OP_PLUSI:
2413      case OP_MINPLUS:      case OP_MINPLUS:
2414      case OP_POSPLUS:      case OP_MINPLUSI:
2415      case OP_EXACT:  
2416      case OP_NOTPLUS:      case OP_NOTPLUS:
2417        case OP_NOTPLUSI:
2418      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
2419        case OP_NOTMINPLUSI:
2420    
2421        case OP_POSPLUS:
2422        case OP_POSPLUSI:
2423      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
2424        case OP_NOTPOSPLUSI:
2425    
2426        case OP_EXACT:
2427        case OP_EXACTI:
2428      case OP_NOTEXACT:      case OP_NOTEXACT:
2429        case OP_NOTEXACTI:
2430    
2431      case OP_TYPEPLUS:      case OP_TYPEPLUS:
2432      case OP_TYPEMINPLUS:      case OP_TYPEMINPLUS:
2433      case OP_TYPEPOSPLUS:      case OP_TYPEPOSPLUS:
2434      case OP_TYPEEXACT:      case OP_TYPEEXACT:
2435    
2436      return FALSE;      return FALSE;
2437    
2438      /* These are going to continue, as they may be empty, but we have to      /* These are going to continue, as they may be empty, but we have to
# Line 2495  for (code = first_significant_code(code Line 2452  for (code = first_significant_code(code
2452      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2453      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2454      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2455      if (code[1 + IMM2_SIZE] == OP_PROP      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2456        || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;        code += 2;
2457      break;      break;
2458    
2459      /* End of branch */      /* End of branch */
# Line 2509  for (code = first_significant_code(code Line 2466  for (code = first_significant_code(code
2466      return TRUE;      return TRUE;
2467    
2468      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2469      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO and their caseless and negative versions may be
2470        followed by a multibyte character. */
2471    
2472  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2473      case OP_STAR:      case OP_STAR:
2474      case OP_STARI:      case OP_STARI:
2475        case OP_NOTSTAR:
2476        case OP_NOTSTARI:
2477    
2478      case OP_MINSTAR:      case OP_MINSTAR:
2479      case OP_MINSTARI:      case OP_MINSTARI:
2480        case OP_NOTMINSTAR:
2481        case OP_NOTMINSTARI:
2482    
2483      case OP_POSSTAR:      case OP_POSSTAR:
2484      case OP_POSSTARI:      case OP_POSSTARI:
2485        case OP_NOTPOSSTAR:
2486        case OP_NOTPOSSTARI:
2487    
2488      case OP_QUERY:      case OP_QUERY:
2489      case OP_QUERYI:      case OP_QUERYI:
2490        case OP_NOTQUERY:
2491        case OP_NOTQUERYI:
2492    
2493      case OP_MINQUERY:      case OP_MINQUERY:
2494      case OP_MINQUERYI:      case OP_MINQUERYI:
2495        case OP_NOTMINQUERY:
2496        case OP_NOTMINQUERYI:
2497    
2498      case OP_POSQUERY:      case OP_POSQUERY:
2499      case OP_POSQUERYI:      case OP_POSQUERYI:
2500        case OP_NOTPOSQUERY:
2501        case OP_NOTPOSQUERYI:
2502    
2503      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2504      break;      break;
2505    
2506      case OP_UPTO:      case OP_UPTO:
2507      case OP_UPTOI:      case OP_UPTOI:
2508        case OP_NOTUPTO:
2509        case OP_NOTUPTOI:
2510    
2511      case OP_MINUPTO:      case OP_MINUPTO:
2512      case OP_MINUPTOI:      case OP_MINUPTOI:
2513      case OP_POSUPTO:      case OP_NOTMINUPTO:
2514      case OP_POSUPTOI:      case OP_NOTMINUPTOI:
2515    
2516        case OP_POSUPTO:
2517        case OP_POSUPTOI:
2518        case OP_NOTPOSUPTO:
2519        case OP_NOTPOSUPTOI:
2520    
2521      if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);      if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
2522      break;      break;
2523  #endif  #endif
# Line 2543  for (code = first_significant_code(code Line 2528  for (code = first_significant_code(code
2528      case OP_MARK:      case OP_MARK:
2529      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
2530      case OP_SKIP_ARG:      case OP_SKIP_ARG:
     code += code[1];  
     break;  
   
2531      case OP_THEN_ARG:      case OP_THEN_ARG:
2532      code += code[1];      code += code[1];
2533      break;      break;
2534    
2535      /* None of the remaining opcodes are required to match a character. */      /* None of the remaining opcodes are required to match a character. */
2536    
2537      default:      default:
2538      break;      break;
2539        }
2540      }
2541    
2542    return TRUE;
2543    }
2544    
2545    
2546    
2547    /*************************************************
2548    *    Scan compiled regex for non-emptiness       *
2549    *************************************************/
2550    
2551    /* This function is called to check for left recursive calls. We want to check
2552    the current branch of the current pattern to see if it could match the empty
2553    string. If it could, we must look outwards for branches at other levels,
2554    stopping when we pass beyond the bracket which is the subject of the recursion.
2555    This function is called only during the real compile, not during the
2556    pre-compile.
2557    
2558    Arguments:
2559      code        points to start of the recursion
2560      endcode     points to where to stop (current RECURSE item)
2561      bcptr       points to the chain of current (unclosed) branch starts
2562      utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2563      cd          pointers to tables etc
2564    
2565    Returns:      TRUE if what is matched could be empty
2566    */
2567    
2568    static BOOL
2569    could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2570      branch_chain *bcptr, BOOL utf, compile_data *cd)
2571    {
2572    while (bcptr != NULL && bcptr->current_branch >= code)
2573      {
2574      if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd, NULL))
2575        return FALSE;
2576      bcptr = bcptr->outer;
2577      }
2578    return TRUE;
2579    }
2580    
2581    
2582    
2583    /*************************************************
2584    *        Base opcode of repeated opcodes         *
2585    *************************************************/
2586    
2587    /* Returns the base opcode for repeated single character type opcodes. If the
2588    opcode is not a repeated character type, it returns with the original value.
2589    
2590    Arguments:  c opcode
2591    Returns:    base opcode for the type
2592    */
2593    
2594    static pcre_uchar
2595    get_repeat_base(pcre_uchar c)
2596    {
2597    return (c > OP_TYPEPOSUPTO)? c :
2598           (c >= OP_TYPESTAR)?   OP_TYPESTAR :
2599           (c >= OP_NOTSTARI)?   OP_NOTSTARI :
2600           (c >= OP_NOTSTAR)?    OP_NOTSTAR :
2601           (c >= OP_STARI)?      OP_STARI :
2602                                 OP_STAR;
2603    }
2604    
2605    
2606    
2607    #ifdef SUPPORT_UCP
2608    /*************************************************
2609    *        Check a character and a property        *
2610    *************************************************/
2611    
2612    /* This function is called by check_auto_possessive() when a property item
2613    is adjacent to a fixed character.
2614    
2615    Arguments:
2616      c            the character
2617      ptype        the property type
2618      pdata        the data for the type
2619      negated      TRUE if it's a negated property (\P or \p{^)
2620    
2621    Returns:       TRUE if auto-possessifying is OK
2622    */
2623    
2624    static BOOL
2625    check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata,
2626      BOOL negated)
2627    {
2628    const pcre_uint32 *p;
2629    const ucd_record *prop = GET_UCD(c);
2630    
2631    switch(ptype)
2632      {
2633      case PT_LAMP:
2634      return (prop->chartype == ucp_Lu ||
2635              prop->chartype == ucp_Ll ||
2636              prop->chartype == ucp_Lt) == negated;
2637    
2638      case PT_GC:
2639      return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2640    
2641      case PT_PC:
2642      return (pdata == prop->chartype) == negated;
2643    
2644      case PT_SC:
2645      return (pdata == prop->script) == negated;
2646    
2647      /* These are specials */
2648    
2649      case PT_ALNUM:
2650      return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2651              PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
2652    
2653      case PT_SPACE:    /* Perl space */
2654      return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2655              c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2656              == negated;
2657    
2658      case PT_PXSPACE:  /* POSIX space */
2659      return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2660              c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2661              c == CHAR_FF || c == CHAR_CR)
2662              == negated;
2663    
2664      case PT_WORD:
2665      return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2666              PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2667              c == CHAR_UNDERSCORE) == negated;
2668    
2669      case PT_CLIST:
2670      p = PRIV(ucd_caseless_sets) + prop->caseset;
2671      for (;;)
2672        {
2673        if (c < *p) return !negated;
2674        if (c == *p++) return negated;
2675        }
2676      break;  /* Control never reaches here */
2677      }
2678    
2679    return FALSE;
2680    }
2681    #endif  /* SUPPORT_UCP */
2682    
2683    
2684    
2685    /*************************************************
2686    *        Fill the character property list        *
2687    *************************************************/
2688    
2689    /* Checks whether the code points to an opcode that can take part in auto-
2690    possessification, and if so, fills a list with its properties.
2691    
2692    Arguments:
2693      code        points to start of expression
2694      utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2695      fcc         points to case-flipping table
2696      list        points to output list
2697                  list[0] will be filled with the opcode
2698                  list[1] will be non-zero if this opcode
2699                    can match an empty character string
2700                  list[2..7] depends on the opcode
2701    
2702    Returns:      points to the start of the next opcode if *code is accepted
2703                  NULL if *code is not accepted
2704    */
2705    
2706    static const pcre_uchar *
2707    get_chr_property_list(const pcre_uchar *code, BOOL utf,
2708      const pcre_uint8 *fcc, pcre_uint32 *list)
2709    {
2710    pcre_uchar c = *code;
2711    const pcre_uchar *end;
2712    const pcre_uint32 *clist_src;
2713    pcre_uint32 *clist_dest;
2714    pcre_uint32 chr;
2715    pcre_uchar base;
2716    
2717    list[0] = c;
2718    list[1] = FALSE;
2719    code++;
2720    
2721    if (c >= OP_STAR && c <= OP_TYPEPOSUPTO)
2722      {
2723      base = get_repeat_base(c);
2724      c -= (base - OP_STAR);
2725    
2726      if (c == OP_UPTO || c == OP_MINUPTO || c == OP_EXACT || c == OP_POSUPTO)
2727        code += IMM2_SIZE;
2728    
2729      list[1] = (c != OP_PLUS && c != OP_MINPLUS && c != OP_EXACT && c != OP_POSPLUS);
2730    
2731      switch(base)
2732        {
2733        case OP_STAR:
2734        list[0] = OP_CHAR;
2735        break;
2736    
2737        case OP_STARI:
2738        list[0] = OP_CHARI;
2739        break;
2740    
2741        case OP_NOTSTAR:
2742        list[0] = OP_NOT;
2743        break;
2744    
2745        case OP_NOTSTARI:
2746        list[0] = OP_NOTI;
2747        break;
2748    
2749        case OP_TYPESTAR:
2750        list[0] = *code;
2751        code++;
2752        break;
2753        }
2754      c = list[0];
2755      }
2756    
2757    switch(c)
2758      {
2759      case OP_NOT_DIGIT:
2760      case OP_DIGIT:
2761      case OP_NOT_WHITESPACE:
2762      case OP_WHITESPACE:
2763      case OP_NOT_WORDCHAR:
2764      case OP_WORDCHAR:
2765      case OP_ANY:
2766      case OP_ALLANY:
2767      case OP_ANYNL:
2768      case OP_NOT_HSPACE:
2769      case OP_HSPACE:
2770      case OP_NOT_VSPACE:
2771      case OP_VSPACE:
2772      case OP_EXTUNI:
2773      case OP_EODN:
2774      case OP_EOD:
2775      case OP_DOLL:
2776      case OP_DOLLM:
2777      return code;
2778    
2779      case OP_CHAR:
2780      case OP_NOT:
2781      GETCHARINCTEST(chr, code);
2782      list[2] = chr;
2783      list[3] = NOTACHAR;
2784      return code;
2785    
2786      case OP_CHARI:
2787      case OP_NOTI:
2788      list[0] = (c == OP_CHARI) ? OP_CHAR : OP_NOT;
2789      GETCHARINCTEST(chr, code);
2790      list[2] = chr;
2791    
2792    #ifdef SUPPORT_UCP
2793      if (chr < 128 || (chr < 256 && !utf))
2794        list[3] = fcc[chr];
2795      else
2796        list[3] = UCD_OTHERCASE(chr);
2797    #elif defined SUPPORT_UTF || !defined COMPILE_PCRE8
2798      list[3] = (chr < 256) ? fcc[chr] : chr;
2799    #else
2800      list[3] = fcc[chr];
2801    #endif
2802    
2803      /* The othercase might be the same value. */
2804    
2805      if (chr == list[3])
2806        list[3] = NOTACHAR;
2807      else
2808        list[4] = NOTACHAR;
2809      return code;
2810    
2811    #ifdef SUPPORT_UCP
2812      case OP_PROP:
2813      case OP_NOTPROP:
2814      if (code[0] != PT_CLIST)
2815        {
2816        list[2] = code[0];
2817        list[3] = code[1];
2818        return code + 2;
2819        }
2820    
2821      /* Convert only if we have anough space. */
2822    
2823      clist_src = PRIV(ucd_caseless_sets) + code[1];
2824      clist_dest = list + 2;
2825      code += 2;
2826    
2827      do {
2828         /* Early return if there is not enough space. */
2829         if (clist_dest >= list + 8)
2830           {
2831           list[2] = code[0];
2832           list[3] = code[1];
2833           return code;
2834           }
2835         *clist_dest++ = *clist_src;
2836         }
2837       while(*clist_src++ != NOTACHAR);
2838    
2839      /* Enough space to store all characters. */
2840    
2841      list[0] = (c == OP_PROP) ? OP_CHAR : OP_NOT;
2842      return code;
2843    #endif
2844    
2845      case OP_NCLASS:
2846      case OP_CLASS:
2847    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2848      case OP_XCLASS:
2849    
2850      if (c == OP_XCLASS)
2851        end = code + GET(code, 0);
2852      else
2853    #endif
2854        end = code + 32 / sizeof(pcre_uchar);
2855    
2856      switch(*end)
2857        {
2858        case OP_CRSTAR:
2859        case OP_CRMINSTAR:
2860        case OP_CRQUERY:
2861        case OP_CRMINQUERY:
2862        list[1] = TRUE;
2863        end++;
2864        break;
2865    
2866        case OP_CRRANGE:
2867        case OP_CRMINRANGE:
2868        list[1] = (GET2(end, 1) == 0);
2869        end += 1 + 2 * IMM2_SIZE;
2870        break;
2871        }
2872      list[2] = end - code;
2873      return end;
2874      }
2875    return NULL;    /* Opcode not accepted */
2876    }
2877    
2878    
2879    
2880    /*************************************************
2881    *    Scan further character sets for match       *
2882    *************************************************/
2883    
2884    /* Checks whether the base and the current opcode have a common character, in
2885    which case the base cannot be possessified.
2886    
2887    Arguments:
2888      code        points to the byte code
2889      utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2890      cd          static compile data
2891      base_list   the data list of the base opcode
2892    
2893    Returns:      TRUE if the auto-possessification is possible
2894    */
2895    
2896    static BOOL
2897    compare_opcodes(const pcre_uchar *code, BOOL utf, const compile_data *cd,
2898      const pcre_uint32* base_list)
2899    {
2900    pcre_uchar c;
2901    pcre_uint32 list[8];
2902    const pcre_uint32* chr_ptr;
2903    const pcre_uint32* ochr_ptr;
2904    const pcre_uint32* list_ptr;
2905    pcre_uint32 chr;
2906    
2907    for(;;)
2908      {
2909      c = *code;
2910    
2911      /* Skip over callouts */
2912    
2913      if (c == OP_CALLOUT)
2914        {
2915        code += PRIV(OP_lengths)[c];
2916        continue;
2917        }
2918    
2919      if (c == OP_ALT)
2920        {
2921        do code += GET(code, 1); while (*code == OP_ALT);
2922        c = *code;
2923        }
2924    
2925      switch(c)
2926        {
2927        case OP_END:
2928        /* TRUE only in greedy case. The non-greedy case could be replaced by an
2929        OP_EXACT, but it is probably not worth it. (And note that OP_EXACT uses
2930        more memory, which we cannot get at this stage.) */
2931    
2932        return base_list[1] != 0;
2933    
2934        case OP_KET:
2935        /* If the bracket is capturing, and referenced by an OP_RECURSE, the
2936        non-greedy case cannot be converted to a possessive form. We do not test
2937        the bracket type at the moment, but we might do it in the future to improve
2938        this condition. (But note that recursive calls are always atomic.) */
2939    
2940        if (base_list[1] == 0) return FALSE;
2941        code += PRIV(OP_lengths)[c];
2942        continue;
2943        }
2944    
2945      /* Check for a supported opcode, and load its properties. */
2946    
2947      code = get_chr_property_list(code, utf, cd->fcc, list);
2948      if (code == NULL) return FALSE;    /* Unsupported */
2949    
2950      /* If either opcode is a small character list, set pointers for comparing
2951      characters from that list with another list, or with a property. */
2952    
2953      if (base_list[0] == OP_CHAR)
2954        {
2955        chr_ptr = base_list + 2;
2956        list_ptr = list;
2957        }
2958      else if (list[0] == OP_CHAR)
2959        {
2960        chr_ptr = list + 2;
2961        list_ptr = base_list;
2962        }
2963    
2964      /* Some property combinations also acceptable. Unicode property opcodes are
2965      processed specially; the rest can be handled with a lookup table. */
2966    
2967      else
2968        {
2969        pcre_uint32 leftop, rightop;
2970    
2971        if (list[1] != 0) return FALSE;   /* Must match at least one character */
2972        leftop = base_list[0];
2973        rightop = list[0];
2974    
2975    #ifdef SUPPORT_UCP
2976        if (leftop == OP_PROP || leftop == OP_NOTPROP)
2977          {
2978          if (rightop == OP_EOD) return TRUE;
2979          if (rightop == OP_PROP || rightop == OP_NOTPROP)
2980            {
2981            int n;
2982            const pcre_uint8 *p;
2983            BOOL same = leftop == rightop;
2984            BOOL lisprop = leftop == OP_PROP;
2985            BOOL risprop = rightop == OP_PROP;
2986            BOOL bothprop = lisprop && risprop;
2987    
2988            /* There's a table that specifies how each combination is to be
2989            processed:
2990              0   Always return FALSE (never auto-possessify)
2991              1   Character groups are distinct (possessify if both are OP_PROP)
2992              2   Check character categories in the same group (general or particular)
2993              3   Return TRUE if the two opcodes are not the same
2994              ... see comments below
2995            */
2996    
2997            n = propposstab[base_list[2]][list[2]];
2998            switch(n)
2999              {
3000              case 0: return FALSE;
3001              case 1: return bothprop;
3002              case 2: return (base_list[3] == list[3]) != same;
3003              case 3: return !same;
3004    
3005              case 4:  /* Left general category, right particular category */
3006              return risprop && catposstab[base_list[3]][list[3]] == same;
3007    
3008              case 5:  /* Right general category, left particular category */
3009              return lisprop && catposstab[list[3]][base_list[3]] == same;
3010    
3011              /* This code is logically tricky. Think hard before fiddling with it.
3012              The posspropstab table has four entries per row. Each row relates to
3013              one of PCRE's special properties such as ALNUM or SPACE or WORD.
3014              Only WORD actually needs all four entries, but using repeats for the
3015              others means they can all use the same code below.
3016    
3017              The first two entries in each row are Unicode general categories, and
3018              apply always, because all the characters they include are part of the
3019              PCRE character set. The third and fourth entries are a general and a
3020              particular category, respectively, that include one or more relevant
3021              characters. One or the other is used, depending on whether the check
3022              is for a general or a particular category. However, in both cases the
3023              category contains more characters than the specials that are defined
3024              for the property being tested against. Therefore, it cannot be used
3025              in a NOTPROP case.
3026    
3027              Example: the row for WORD contains ucp_L, ucp_N, ucp_P, ucp_Po.
3028              Underscore is covered by ucp_P or ucp_Po. */
3029    
3030              case 6:  /* Left alphanum vs right general category */
3031              case 7:  /* Left space vs right general category */
3032              case 8:  /* Left word vs right general category */
3033              p = posspropstab[n-6];
3034              return risprop && lisprop ==
3035                (list[3] != p[0] &&
3036                 list[3] != p[1] &&
3037                (list[3] != p[2] || !lisprop));
3038    
3039              case 9:   /* Right alphanum vs left general category */
3040              case 10:  /* Right space vs left general category */
3041              case 11:  /* Right word vs left general category */
3042              p = posspropstab[n-9];
3043              return lisprop && risprop ==
3044                (base_list[3] != p[0] &&
3045                 base_list[3] != p[1] &&
3046                (base_list[3] != p[2] || !risprop));
3047    
3048              case 12:  /* Left alphanum vs right particular category */
3049              case 13:  /* Left space vs right particular category */
3050              case 14:  /* Left word vs right particular category */
3051              p = posspropstab[n-12];
3052              return risprop && lisprop ==
3053                (catposstab[p[0]][list[3]] &&
3054                 catposstab[p[1]][list[3]] &&
3055                (list[3] != p[3] || !lisprop));
3056    
3057              case 15:  /* Right alphanum vs left particular category */
3058              case 16:  /* Right space vs left particular category */
3059              case 17:  /* Right word vs left particular category */
3060              p = posspropstab[n-15];
3061              return lisprop && risprop ==
3062                (catposstab[p[0]][base_list[3]] &&
3063                 catposstab[p[1]][base_list[3]] &&
3064                (base_list[3] != p[3] || !risprop));
3065              }
3066            }
3067          return FALSE;
3068          }
3069    
3070        else
3071    #endif  /* SUPPORT_UCP */
3072    
3073        return leftop >= FIRST_AUTOTAB_OP && leftop <= LAST_AUTOTAB_LEFT_OP &&
3074               rightop >= FIRST_AUTOTAB_OP && rightop <= LAST_AUTOTAB_RIGHT_OP &&
3075               autoposstab[leftop - FIRST_AUTOTAB_OP][rightop - FIRST_AUTOTAB_OP];
3076        }
3077    
3078      /* Control reaches here only if one of the items is a small character list.
3079      All characters are checked against the other side. */
3080    
3081      do
3082        {
3083        chr = *chr_ptr;
3084    
3085        switch(list_ptr[0])
3086          {
3087          case OP_CHAR:
3088          ochr_ptr = list_ptr + 2;
3089          do
3090            {
3091            if (chr == *ochr_ptr) return FALSE;
3092            ochr_ptr++;
3093            }
3094          while(*ochr_ptr != NOTACHAR);
3095          break;
3096    
3097          case OP_NOT:
3098          ochr_ptr = list_ptr + 2;
3099          do
3100            {
3101            if (chr == *ochr_ptr)
3102              break;
3103            ochr_ptr++;
3104            }
3105          while(*ochr_ptr != NOTACHAR);
3106          if (*ochr_ptr == NOTACHAR) return FALSE;   /* Not found */
3107          break;
3108    
3109          /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not*
3110          set. When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
3111    
3112          case OP_DIGIT:
3113          if (chr < 256 && (cd->ctypes[chr] & ctype_digit) != 0) return FALSE;
3114          break;
3115    
3116          case OP_NOT_DIGIT:
3117          if (chr > 255 || (cd->ctypes[chr] & ctype_digit) == 0) return FALSE;
3118          break;
3119    
3120          case OP_WHITESPACE:
3121          if (chr < 256 && (cd->ctypes[chr] & ctype_space) != 0) return FALSE;
3122          break;
3123    
3124          case OP_NOT_WHITESPACE:
3125          if (chr > 255 || (cd->ctypes[chr] & ctype_space) == 0) return FALSE;
3126          break;
3127    
3128          case OP_WORDCHAR:
3129          if (chr < 255 && (cd->ctypes[chr] & ctype_word) != 0) return FALSE;
3130          break;
3131    
3132          case OP_NOT_WORDCHAR:
3133          if (chr > 255 || (cd->ctypes[chr] & ctype_word) == 0) return FALSE;
3134          break;
3135    
3136          case OP_HSPACE:
3137          switch(chr)
3138            {
3139            HSPACE_CASES: return FALSE;
3140            default: break;
3141            }
3142          break;
3143    
3144          case OP_NOT_HSPACE:
3145          switch(chr)
3146            {
3147            HSPACE_CASES: break;
3148            default: return FALSE;
3149            }
3150          break;
3151    
3152          case OP_ANYNL:
3153          case OP_VSPACE:
3154          switch(chr)
3155            {
3156            VSPACE_CASES: return FALSE;
3157            default: break;
3158            }
3159          break;
3160    
3161          case OP_NOT_VSPACE:
3162          switch(chr)
3163            {
3164            VSPACE_CASES: break;
3165            default: return FALSE;
3166            }
3167          break;
3168    
3169          case OP_DOLL:
3170          case OP_EODN:
3171          switch (chr)
3172            {
3173            case CHAR_CR:
3174            case CHAR_LF:
3175            case CHAR_VT:
3176            case CHAR_FF:
3177            case CHAR_NEL:
3178    #ifndef EBCDIC
3179            case 0x2028:
3180            case 0x2029:
3181    #endif  /* Not EBCDIC */
3182            return FALSE;
3183            }
3184          break;
3185    
3186          case OP_EOD:    /* Can always possessify before \z */
3187          break;
3188    
3189          case OP_PROP:
3190          case OP_NOTPROP:
3191          if (!check_char_prop(chr, list_ptr[2], list_ptr[3],
3192                list_ptr[0] == OP_NOTPROP))
3193            return FALSE;
3194          break;
3195    
3196          /* The class comparisons work only when the class is the second item
3197          of the pair, because there are at present no possessive forms of the
3198          class opcodes. Note also that the "code" variable that is used below
3199          points after the second item, and that the pointer for the first item
3200          is not available, so even if there were possessive forms of the class
3201          opcodes, the correct comparison could not be done. */
3202    
3203          case OP_NCLASS:
3204          if (chr > 255) return FALSE;
3205          /* Fall through */
3206    
3207          case OP_CLASS:
3208          if (list_ptr != list) return FALSE;   /* Class is first opcode */
3209          if (chr > 255) break;
3210          if ((((pcre_uint8 *)(code - list_ptr[2] + 1))[chr >> 3] & (1 << (chr & 7))) != 0)
3211            return FALSE;
3212          break;
3213    
3214    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3215          case OP_XCLASS:
3216          if (list_ptr != list) return FALSE;   /* Class is first opcode */
3217          if (PRIV(xclass)(chr, code - list_ptr[2] + 1 + LINK_SIZE, utf))
3218            return FALSE;
3219          break;
3220    #endif
3221    
3222          default:
3223          return FALSE;
3224          }
3225    
3226        chr_ptr++;
3227        }
3228      while(*chr_ptr != NOTACHAR);
3229    
3230      /* At least one character must be matched from this opcode. */
3231    
3232      if (list[1] == 0) return TRUE;
3233      }
3234    
3235    return FALSE;
3236    }
3237    
3238    
3239    
3240    /*************************************************
3241    *    Scan compiled regex for auto-possession     *
3242    *************************************************/
3243    
3244    /* Replaces single character iterations with their possessive alternatives
3245    if appropriate. This function modifies the compiled opcode!
3246    
3247    Arguments:
3248      code        points to start of the byte code
3249      utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
3250      cd          static compile data
3251    
3252    Returns:      nothing
3253    */
3254    
3255    static void
3256    auto_possessify(pcre_uchar *code, BOOL utf, const compile_data *cd)
3257    {
3258    register pcre_uchar c;
3259    const pcre_uchar *end;
3260    pcre_uint32 list[8];
3261    
3262    for (;;)
3263      {
3264      c = *code;
3265    
3266      if (c >= OP_STAR && c <= OP_TYPEPOSUPTO)
3267        {
3268        c -= get_repeat_base(c) - OP_STAR;
3269        end = (c <= OP_MINUPTO) ?
3270          get_chr_property_list(code, utf, cd->fcc, list) : NULL;
3271        list[1] = c == OP_STAR || c == OP_PLUS || c == OP_QUERY || c == OP_UPTO;
3272    
3273        if (end != NULL && compare_opcodes(end, utf, cd, list))
3274          {
3275          switch(c)
3276            {
3277            case OP_STAR:
3278            *code += OP_POSSTAR - OP_STAR;
3279            break;
3280    
3281            case OP_MINSTAR:
3282            *code += OP_POSSTAR - OP_MINSTAR;
3283            break;
3284    
3285            case OP_PLUS:
3286            *code += OP_POSPLUS - OP_PLUS;
3287            break;
3288    
3289            case OP_MINPLUS:
3290            *code += OP_POSPLUS - OP_MINPLUS;
3291            break;
3292    
3293            case OP_QUERY:
3294            *code += OP_POSQUERY - OP_QUERY;
3295            break;
3296    
3297            case OP_MINQUERY:
3298            *code += OP_POSQUERY - OP_MINQUERY;
3299            break;
3300    
3301            case OP_UPTO:
3302            *code += OP_POSUPTO - OP_UPTO;
3303            break;
3304    
3305            case OP_MINUPTO:
3306            *code += OP_MINUPTO - OP_UPTO;
3307            break;
3308            }
3309          }
3310        c = *code;
3311        }
3312    
3313      switch(c)
3314        {
3315        case OP_END:
3316        return;
3317    
3318        case OP_TYPESTAR:
3319        case OP_TYPEMINSTAR:
3320        case OP_TYPEPLUS:
3321        case OP_TYPEMINPLUS:
3322        case OP_TYPEQUERY:
3323        case OP_TYPEMINQUERY:
3324        case OP_TYPEPOSSTAR:
3325        case OP_TYPEPOSPLUS:
3326        case OP_TYPEPOSQUERY:
3327        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
3328        break;
3329    
3330        case OP_TYPEUPTO:
3331        case OP_TYPEMINUPTO:
3332        case OP_TYPEEXACT:
3333        case OP_TYPEPOSUPTO:
3334        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
3335          code += 2;
3336        break;
3337    
3338        case OP_XCLASS:
3339        code += GET(code, 1);
3340        break;
3341    
3342        case OP_MARK:
3343        case OP_PRUNE_ARG:
3344        case OP_SKIP_ARG:
3345        case OP_THEN_ARG:
3346        code += code[1];
3347        break;
3348      }      }
   }  
   
 return TRUE;  
 }  
   
   
3349    
3350  /*************************************************    /* Add in the fixed length from the table */
 *    Scan compiled regex for non-emptiness       *  
 *************************************************/  
   
 /* This function is called to check for left recursive calls. We want to check  
 the current branch of the current pattern to see if it could match the empty  
 string. If it could, we must look outwards for branches at other levels,  
 stopping when we pass beyond the bracket which is the subject of the recursion.  
 This function is called only during the real compile, not during the  
 pre-compile.  
3351    
3352  Arguments:    code += PRIV(OP_lengths)[c];
   code        points to start of the recursion  
   endcode     points to where to stop (current RECURSE item)  
   bcptr       points to the chain of current (unclosed) branch starts  
   utf         TRUE if in UTF-8 / UTF-16 mode  
   cd          pointers to tables etc  
3353    
3354  Returns:      TRUE if what is matched could be empty    /* In UTF-8 mode, opcodes that are followed by a character may be followed by
3355  */    a multi-byte character. The length in the table is a minimum, so we have to
3356      arrange to skip the extra bytes. */
3357    
3358  static BOOL  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3359  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,    if (utf) switch(c)
3360    branch_chain *bcptr, BOOL utf, compile_data *cd)      {
3361  {      case OP_CHAR:
3362  while (bcptr != NULL && bcptr->current_branch >= code)      case OP_CHARI:
3363    {      case OP_NOT:
3364    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))      case OP_NOTI:
3365      return FALSE;      case OP_STAR:
3366    bcptr = bcptr->outer;      case OP_MINSTAR:
3367        case OP_PLUS:
3368        case OP_MINPLUS:
3369        case OP_QUERY:
3370        case OP_MINQUERY:
3371        case OP_UPTO:
3372        case OP_MINUPTO:
3373        case OP_EXACT:
3374        case OP_POSSTAR:
3375        case OP_POSPLUS:
3376        case OP_POSQUERY:
3377        case OP_POSUPTO:
3378        case OP_STARI:
3379        case OP_MINSTARI:
3380        case OP_PLUSI:
3381        case OP_MINPLUSI:
3382        case OP_QUERYI:
3383        case OP_MINQUERYI:
3384        case OP_UPTOI:
3385        case OP_MINUPTOI:
3386        case OP_EXACTI:
3387        case OP_POSSTARI:
3388        case OP_POSPLUSI:
3389        case OP_POSQUERYI:
3390        case OP_POSUPTOI:
3391        case OP_NOTSTAR:
3392        case OP_NOTMINSTAR:
3393        case OP_NOTPLUS:
3394        case OP_NOTMINPLUS:
3395        case OP_NOTQUERY:
3396        case OP_NOTMINQUERY:
3397        case OP_NOTUPTO:
3398        case OP_NOTMINUPTO:
3399        case OP_NOTEXACT:
3400        case OP_NOTPOSSTAR:
3401        case OP_NOTPOSPLUS:
3402        case OP_NOTPOSQUERY:
3403        case OP_NOTPOSUPTO:
3404        case OP_NOTSTARI:
3405        case OP_NOTMINSTARI:
3406        case OP_NOTPLUSI:
3407        case OP_NOTMINPLUSI:
3408        case OP_NOTQUERYI:
3409        case OP_NOTMINQUERYI:
3410        case OP_NOTUPTOI:
3411        case OP_NOTMINUPTOI:
3412        case OP_NOTEXACTI:
3413        case OP_NOTPOSSTARI:
3414        case OP_NOTPOSPLUSI:
3415        case OP_NOTPOSQUERYI:
3416        case OP_NOTPOSUPTOI:
3417        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
3418        break;
3419        }
3420    #else
3421      (void)(utf);  /* Keep compiler happy by referencing function argument */
3422    #endif
3423    }    }
 return TRUE;  
3424  }  }
3425    
3426    
# Line 2643  Returns:   TRUE or FALSE Line 3470  Returns:   TRUE or FALSE
3470  static BOOL  static BOOL
3471  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
3472  {  {
3473  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
3474  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
3475  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != CHAR_NULL; ptr++)
3476    {    {
3477    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
3478      ptr++;      ptr++;
# Line 2692  register int yield = 0; Line 3519  register int yield = 0;
3519  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
3520    {    {
3521    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
3522      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
3523    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
3524    yield++;    yield++;
3525    }    }
# Line 2724  value in the reference (which is a group Line 3551  value in the reference (which is a group
3551  Arguments:  Arguments:
3552    group      points to the start of the group    group      points to the start of the group
3553    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
3554    utf        TRUE in UTF-8 / UTF-16 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
3555    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
3556    save_hwm   the hwm forward reference pointer at the start of the group    save_hwm   the hwm forward reference pointer at the start of the group
3557    
# Line 2747  while ((ptr = (pcre_uchar *)find_recurse Line 3574  while ((ptr = (pcre_uchar *)find_recurse
3574    
3575    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
3576      {      {
3577      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
3578      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
3579        {        {
3580        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2760  while ((ptr = (pcre_uchar *)find_recurse Line 3587  while ((ptr = (pcre_uchar *)find_recurse
3587    
3588    if (hc >= cd->hwm)    if (hc >= cd->hwm)
3589      {      {
3590      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
3591      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
3592      }      }
3593    
# Line 2828  PUT(previous_callout, 2 + LINK_SIZE, len Line 3655  PUT(previous_callout, 2 + LINK_SIZE, len
3655  *************************************************/  *************************************************/
3656    
3657  /* This function is passed the start and end of a class range, in UTF-8 mode  /* This function is passed the start and end of a class range, in UTF-8 mode
3658  with UCP support. It searches up the characters, looking for internal ranges of  with UCP support. It searches up the characters, looking for ranges of
3659  characters in the "other" case. Each call returns the next one, updating the  characters in the "other" case. Each call returns the next one, updating the
3660  start address.  start address. A character with multiple other cases is returned on its own
3661    with a special return value.
3662    
3663  Arguments:  Arguments:
3664    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2838  Arguments: Line 3666  Arguments:
3666    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
3667    odptr       where to put end of othercase range    odptr       where to put end of othercase range
3668    
3669  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
3670                   0 when a range is returned
3671                  >0 the CASESET offset for char with multiple other cases
3672                    in this case, ocptr contains the original
3673  */  */
3674    
3675  static BOOL  static int
3676  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
3677    unsigned int *odptr)    pcre_uint32 *odptr)
3678  {  {
3679  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
3680    unsigned int co;
3681    
3682    /* Find the first character that has an other case. If it has multiple other
3683    cases, return its case offset value. */
3684    
3685  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
3686    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
3687      if ((co = UCD_CASESET(c)) != 0)
3688        {
3689        *ocptr = c++;   /* Character that has the set */
3690        *cptr = c;      /* Rest of input range */
3691        return (int)co;
3692        }
3693      if ((othercase = UCD_OTHERCASE(c)) != c) break;
3694      }
3695    
3696  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
3697    
3698  *ocptr = othercase;  *ocptr = othercase;
3699  next = othercase + 1;  next = othercase + 1;
# Line 2861  for (++c; c <= d; c++) Line 3704  for (++c; c <= d; c++)
3704    next++;    next++;
3705    }    }
3706    
3707  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
3708  *cptr = c;  *cptr = c;             /* Rest of input range */
3709    return 0;
 return TRUE;  
 }  
   
   
   
 /*************************************************  
 *        Check a character and a property        *  
 *************************************************/  
   
 /* This function is called by check_auto_possessive() when a property item  
 is adjacent to a fixed character.  
   
 Arguments:  
   c            the character  
   ptype        the property type  
   pdata        the data for the type  
   negated      TRUE if it's a negated property (\P or \p{^)  
   
 Returns:       TRUE if auto-possessifying is OK  
 */  
   
 static BOOL  
 check_char_prop(int c, int ptype, int pdata, BOOL negated)  
 {  
 const ucd_record *prop = GET_UCD(c);  
 switch(ptype)  
   {  
   case PT_LAMP:  
   return (prop->chartype == ucp_Lu ||  
           prop->chartype == ucp_Ll ||  
           prop->chartype == ucp_Lt) == negated;  
   
   case PT_GC:  
   return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;  
   
   case PT_PC:  
   return (pdata == prop->chartype) == negated;  
   
   case PT_SC:  
   return (pdata == prop->script) == negated;  
   
   /* These are specials */  
   
   case PT_ALNUM:  
   return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||  
           PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;  
   
   case PT_SPACE:    /* Perl space */  
   return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||  
           c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)  
           == negated;  
   
   case PT_PXSPACE:  /* POSIX space */  
   return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||  
           c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||  
           c == CHAR_FF || c == CHAR_CR)  
           == negated;  
   
   case PT_WORD:  
   return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||  
           PRIV(ucp_gentype)[prop->chartype] == ucp_N ||  
           c == CHAR_UNDERSCORE) == negated;  
   }  
 return FALSE;  
3710  }  }
3711  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
3712    
3713    
3714    
3715  /*************************************************  /*************************************************
3716  *     Check if auto-possessifying is possible    *  *        Add a character or range to a class     *
3717  *************************************************/  *************************************************/
3718    
3719  /* This function is called for unlimited repeats of certain items, to see  /* This function packages up the logic of adding a character or range of
3720  whether the next thing could possibly match the repeated item. If not, it makes  characters to a class. The character values in the arguments will be within the
3721  sense to automatically possessify the repeated item.  valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3722    mutually recursive with the function immediately below.
3723    
3724  Arguments:  Arguments:
3725    previous      pointer to the repeated opcode    classbits     the bit map for characters < 256
3726    utf           TRUE in UTF-8 / UTF-16 mode    uchardptr     points to the pointer for extra data
3727    ptr           next character in pattern    options       the options word
   options       options bits  
3728    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
3729      start         start of range character
3730      end           end of range character
3731    
3732  Returns:        TRUE if possessifying is wanted  Returns:        the number of < 256 characters added
3733                    the pointer to extra data is updated
3734  */  */
3735    
3736  static BOOL  static int
3737  check_auto_possessive(const pcre_uchar *previous, BOOL utf,  add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3738    const pcre_uchar *ptr, int options, compile_data *cd)    compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3739  {  {
3740  pcre_int32 c, next;  pcre_uint32 c;
3741  int op_code = *previous++;  int n8 = 0;
   
 /* Skip whitespace and comments in extended mode */  
   
 if ((options & PCRE_EXTENDED) != 0)  
   {  
   for (;;)  
     {  
     while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;  
     if (*ptr == CHAR_NUMBER_SIGN)  
       {  
       ptr++;  
       while (*ptr != 0)  
         {  
         if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }  
         ptr++;  
 #ifdef SUPPORT_UTF  
         if (utf) FORWARDCHAR(ptr);  
 #endif  
         }  
       }  
     else break;  
     }  
   }  
   
 /* If the next item is one that we can handle, get its value. A non-negative  
 value is a character, a negative value is an escape value. */  
   
 if (*ptr == CHAR_BACKSLASH)  
   {  
   int temperrorcode = 0;  
   next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);  
   if (temperrorcode != 0) return FALSE;  
   ptr++;    /* Point after the escape sequence */  
   }  
 else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  
   {  
 #ifdef SUPPORT_UTF  
   if (utf) { GETCHARINC(next, ptr); } else  
 #endif  
   next = *ptr++;  
   }  
 else return FALSE;  
   
 /* Skip whitespace and comments in extended mode */  
   
 if ((options & PCRE_EXTENDED) != 0)  
   {  
   for (;;)  
     {  
     while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;  
     if (*ptr == CHAR_NUMBER_SIGN)  
       {  
       ptr++;  
       while (*ptr != 0)  
         {  
         if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }  
         ptr++;  
 #ifdef SUPPORT_UTF  
         if (utf) FORWARDCHAR(ptr);  
 #endif  
         }  
       }  
     else break;  
     }  
   }  
3742    
3743  /* If the next thing is itself optional, we have to give up. */  /* If caseless matching is required, scan the range and process alternate
3744    cases. In Unicode, there are 8-bit characters that have alternate cases that
3745    are greater than 255 and vice-versa. Sometimes we can just extend the original
3746    range. */
3747    
3748  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if ((options & PCRE_CASELESS) != 0)
   STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)  
     return FALSE;  
   
 /* Now compare the next item with the previous opcode. First, handle cases when  
 the next item is a character. */  
   
 if (next >= 0) switch(op_code)  
3749    {    {
   case OP_CHAR:  
 #ifdef SUPPORT_UTF  
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   return c != next;  
   
   /* For CHARI (caseless character) we must check the other case. If we have  
   Unicode property support, we can use it to test the other case of  
   high-valued characters. */  
   
   case OP_CHARI:  
 #ifdef SUPPORT_UTF  
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   if (c == next) return FALSE;  
 #ifdef SUPPORT_UTF  
   if (utf)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3750  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3751      othercase = UCD_OTHERCASE((unsigned int)next);    if ((options & PCRE_UTF8) != 0)
 #else  
     othercase = NOTACHAR;  
 #endif  
     return (unsigned int)c != othercase;  
     }  
   else  
 #endif  /* SUPPORT_UTF */  
   return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Non-UTF-8 mode */  
   
   /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These  
   opcodes are not used for multi-byte characters, because they are coded using  
   an XCLASS instead. */  
   
   case OP_NOT:  
   return (c = *previous) == next;  
   
   case OP_NOTI:  
   if ((c = *previous) == next) return TRUE;  
 #ifdef SUPPORT_UTF  
   if (utf)  
3752      {      {
3753      unsigned int othercase;      int rc;
3754      if (next < 128) othercase = cd->fcc[next]; else      pcre_uint32 oc, od;
 #ifdef SUPPORT_UCP  
     othercase = UCD_OTHERCASE(next);  
 #else  
     othercase = NOTACHAR;  
 #endif  
     return (unsigned int)c == othercase;  
     }  
   else  
 #endif  /* SUPPORT_UTF */  
   return (c == (int)(TABLE_GET((unsigned int)next, cd->fcc, next)));  /* Non-UTF-8 mode */  
3755    
3756    /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.      options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3757    When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */      c = start;
3758    
3759    case OP_DIGIT:      while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3760    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;        {
3761          /* Handle a single character that has more than one other case. */
   case OP_NOT_DIGIT:  
   return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;  
3762    
3763    case OP_WHITESPACE:        if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3764    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;          PRIV(ucd_caseless_sets) + rc, oc);
3765    
3766    case OP_NOT_WHITESPACE:        /* Do nothing if the other case range is within the original range. */
   return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;  
3767    
3768    case OP_WORDCHAR:        else if (oc >= start && od <= end) continue;
   return next > 127 || (cd->ctypes[next] & ctype_word) == 0;  
3769    
3770    case OP_NOT_WORDCHAR:        /* Extend the original range if there is overlap, noting that if oc < c, we
3771    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;        can't have od > end because a subrange is always shorter than the basic
3772          range. Otherwise, use a recursive call to add the additional range. */
3773    
3774    case OP_HSPACE:        else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3775    case OP_NOT_HSPACE:        else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3776    switch(next)        else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3777      {        }
     case 0x09:  
     case 0x20:  
     case 0xa0:  
     case 0x1680:  
     case 0x180e:  
     case 0x2000:  
     case 0x2001:  
     case 0x2002:  
     case 0x2003:  
     case 0x2004:  
     case 0x2005:  
     case 0x2006:  
     case 0x2007:  
     case 0x2008:  
     case 0x2009:  
     case 0x200A:  
     case 0x202f:  
     case 0x205f:  
     case 0x3000:  
     return op_code == OP_NOT_HSPACE;  
     default:  
     return op_code != OP_NOT_HSPACE;  
3778      }      }
3779      else
3780    #endif  /* SUPPORT_UCP */
3781    
3782    case OP_ANYNL:    /* Not UTF-mode, or no UCP */
3783    case OP_VSPACE:  
3784    case OP_NOT_VSPACE:    for (c = start; c <= end && c < 256; c++)
   switch(next)  
3785      {      {
3786      case 0x0a:      SETBIT(classbits, cd->fcc[c]);
3787      case 0x0b:      n8++;
     case 0x0c:  
     case 0x0d:  
     case 0x85:  
     case 0x2028:  
     case 0x2029:  
     return op_code == OP_NOT_VSPACE;  
     default:  
     return op_code != OP_NOT_VSPACE;  
3788      }      }
   
 #ifdef SUPPORT_UCP  
   case OP_PROP:  
   return check_char_prop(next, previous[0], previous[1], FALSE);  
   
   case OP_NOTPROP:  
   return check_char_prop(next, previous[0], previous[1], TRUE);  
 #endif  
   
   default:  
   return FALSE;  
3789    }    }
3790    
3791    /* Now handle the original range. Adjust the final value according to the bit
3792    length - this means that the same lists of (e.g.) horizontal spaces can be used
3793    in all cases. */
3794    
3795  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP  #if defined COMPILE_PCRE8
 is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are  
 generated only when PCRE_UCP is *not* set, that is, when only ASCII  
 characteristics are recognized. Similarly, the opcodes OP_DIGIT etc. are  
 replaced by OP_PROP codes when PCRE_UCP is set. */  
   
 switch(op_code)  
   {  
   case OP_CHAR:  
   case OP_CHARI:  
3796  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3797    GETCHARTEST(c, previous);    if ((options & PCRE_UTF8) == 0)
 #else  
   c = *previous;  
3798  #endif  #endif
3799    switch(-next)    if (end > 0xff) end = 0xff;
     {  
     case ESC_d:  
     return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;  
3800    
3801      case ESC_D:  #elif defined COMPILE_PCRE16
3802      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;  #ifdef SUPPORT_UTF
3803      if ((options & PCRE_UTF16) == 0)
3804    #endif
3805      if (end > 0xffff) end = 0xffff;
3806    
3807      case ESC_s:  #endif /* COMPILE_PCRE[8|16] */
     return c > 127 || (cd->ctypes[c] & ctype_space) == 0;  
3808    
3809      case ESC_S:  /* If all characters are less than 256, use the bit map. Otherwise use extra
3810      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;  data. */
3811    
3812      case ESC_w:  if (end < 0x100)
3813      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;    {
3814      for (c = start; c <= end; c++)
3815        {
3816        n8++;
3817        SETBIT(classbits, c);
3818        }
3819      }
3820    
3821      case ESC_W:  else
3822      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;    {
3823      pcre_uchar *uchardata = *uchardptr;
3824    
3825      case ESC_h:  #ifdef SUPPORT_UTF
3826      case ESC_H:    if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3827      switch(c)      {
3828        if (start < end)
3829        {        {
3830        case 0x09:        *uchardata++ = XCL_RANGE;
3831        case 0x20:        uchardata += PRIV(ord2utf)(start, uchardata);
3832        case 0xa0:        uchardata += PRIV(ord2utf)(end, uchardata);
       case 0x1680:  
       case 0x180e:  
       case 0x2000:  
       case 0x2001:  
       case 0x2002:  
       case 0x2003:  
       case 0x2004:  
       case 0x2005:  
       case 0x2006:  
       case 0x2007:  
       case 0x2008:  
       case 0x2009:  
       case 0x200A:  
       case 0x202f:  
       case 0x205f:  
       case 0x3000:  
       return -next != ESC_h;  
       default:  
       return -next == ESC_h;  
3833        }        }
3834        else if (start == end)
     case ESC_v:  
     case ESC_V:  
     switch(c)  
3835        {        {
3836        case 0x0a:        *uchardata++ = XCL_SINGLE;
3837        case 0x0b:        uchardata += PRIV(ord2utf)(start, uchardata);
       case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
       default:  
       return -next == ESC_v;  
       }  
   
     /* When PCRE_UCP is set, these values get generated for \d etc. Find  
     their substitutions and process them. The result will always be either  
     -ESC_p or -ESC_P. Then fall through to process those values. */  
   
 #ifdef SUPPORT_UCP  
     case ESC_du:  
     case ESC_DU:  
     case ESC_wu:  
     case ESC_WU:  
     case ESC_su:  
     case ESC_SU:  
       {  
       int temperrorcode = 0;  
       ptr = substitutes[-next - ESC_DU];  
       next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);  
       if (temperrorcode != 0) return FALSE;  
       ptr++;    /* For compatibility */  
3838        }        }
3839      /* Fall through */      }
3840      else
3841    #endif  /* SUPPORT_UTF */
3842    
3843      case ESC_p:    /* Without UTF support, character values are constrained by the bit length,
3844      case ESC_P:    and can only be > 256 for 16-bit and 32-bit libraries. */
       {  
       int ptype, pdata, errorcodeptr;  
       BOOL negated;  
3845    
3846        ptr--;      /* Make ptr point at the p or P */  #ifdef COMPILE_PCRE8
3847        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);      {}
3848        if (ptype < 0) return FALSE;  #else
3849        ptr++;      /* Point past the final curly ket */    if (start < end)
3850        {
3851        *uchardata++ = XCL_RANGE;
3852        *uchardata++ = start;
3853        *uchardata++ = end;
3854        }
3855      else if (start == end)
3856        {
3857        *uchardata++ = XCL_SINGLE;
3858        *uchardata++ = start;
3859        }
3860    #endif
3861    
3862        /* If the property item is optional, we have to give up. (When generated    *uchardptr = uchardata;   /* Updata extra data pointer */
3863        from \d etc by PCRE_UCP, this test will have been applied much earlier,    }
       to the original \d etc. At this point, ptr will point to a zero byte. */  
3864    
3865        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  return n8;    /* Number of 8-bit characters */
3866          STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)  }
           return FALSE;  
3867    
       /* Do the property check. */  
3868    
       return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);  
       }  
 #endif  
3869    
     default:  
     return FALSE;  
     }  
3870    
3871    /* In principle, support for Unicode properties should be integrated here as  /*************************************************
3872    well. It means re-organizing the above code so as to get hold of the property  *        Add a list of characters to a class     *
3873    values before switching on the op-code. However, I wonder how many patterns  *************************************************/
   combine ASCII \d etc with Unicode properties? (Note that if PCRE_UCP is set,  
   these op-codes are never generated.) */  
3874    
3875    case OP_DIGIT:  /* This function is used for adding a list of case-equivalent characters to a
3876    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||  class, and also for adding a list of horizontal or vertical whitespace. If the
3877           next == -ESC_h || next == -ESC_v || next == -ESC_R;  list is in order (which it should be), ranges of characters are detected and
3878    handled appropriately. This function is mutually recursive with the function
3879    above.
3880    
3881    case OP_NOT_DIGIT:  Arguments:
3882    return next == -ESC_d;    classbits     the bit map for characters < 256
3883      uchardptr     points to the pointer for extra data
3884      options       the options word
3885      cd            contains pointers to tables etc.
3886      p             points to row of 32-bit values, terminated by NOTACHAR
3887      except        character to omit; this is used when adding lists of
3888                      case-equivalent characters to avoid including the one we
3889                      already know about
3890    
3891    case OP_WHITESPACE:  Returns:        the number of < 256 characters added
3892    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;                  the pointer to extra data is updated
3893    */
3894    
3895    case OP_NOT_WHITESPACE:  static int
3896    return next == -ESC_s || next == -ESC_h || next == -ESC_v;  add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3897      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3898    {
3899    int n8 = 0;
3900    while (p[0] < NOTACHAR)
3901      {
3902      int n = 0;
3903      if (p[0] != except)
3904        {
3905        while(p[n+1] == p[0] + n + 1) n++;
3906        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3907        }
3908      p += n + 1;
3909      }
3910    return n8;
3911    }
3912    
   case OP_HSPACE:  
   return next == -ESC_S || next == -ESC_H || next == -ESC_d ||  
          next == -ESC_w || next == -ESC_v || next == -ESC_R;  
3913    
   case OP_NOT_HSPACE:  
   return next == -ESC_h;  
3914    
3915    /* Can't have \S in here because VT matches \S (Perl anomaly) */  /*************************************************
3916    case OP_ANYNL:  *    Add characters not in a list to a class     *
3917    case OP_VSPACE:  *************************************************/
   return next == -ESC_V || next == -ESC_d || next == -ESC_w;  
3918    
3919    case OP_NOT_VSPACE:  /* This function is used for adding the complement of a list of horizontal or
3920    return next == -ESC_v || next == -ESC_R;  vertical whitespace to a class. The list must be in order.
3921    
3922    case OP_WORDCHAR:  Arguments:
3923    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    classbits     the bit map for characters < 256
3924           next == -ESC_v || next == -ESC_R;    uchardptr     points to the pointer for extra data
3925      options       the options word
3926      cd            contains pointers to tables etc.
3927      p             points to row of 32-bit values, terminated by NOTACHAR
3928    
3929    case OP_NOT_WORDCHAR:  Returns:        the number of < 256 characters added
3930    return next == -ESC_w || next == -ESC_d;                  the pointer to extra data is updated
3931    */
3932    
3933    default:  static int
3934    return FALSE;  add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3935      int options, compile_data *cd, const pcre_uint32 *p)
3936    {
3937    BOOL utf = (options & PCRE_UTF8) != 0;
3938    int n8 = 0;
3939    if (p[0] > 0)
3940      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3941    while (p[0] < NOTACHAR)
3942      {
3943      while (p[1] == p[0] + 1) p++;
3944      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3945        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3946      p++;
3947    }    }
3948    return n8;
 /* Control does not reach here */  
3949  }  }
3950    
3951    
# Line 3362  to find out the amount of memory needed, Line 3961  to find out the amount of memory needed,
3961  phase. The value of lengthptr distinguishes the two phases.  phase. The value of lengthptr distinguishes the two phases.
3962    
3963  Arguments:  Arguments:
3964    optionsptr     pointer to the option bits    optionsptr        pointer to the option bits
3965    codeptr        points to the pointer to the current code point    codeptr           points to the pointer to the current code point
3966    ptrptr         points to the current pattern pointer    ptrptr            points to the current pattern pointer
3967    errorcodeptr   points to error code variable    errorcodeptr      points to error code variable
3968    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr      place to put the first required character
3969    reqcharptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3970    bcptr          points to current branch chain    reqcharptr        place to put the last required character
3971    cond_depth     conditional nesting depth    reqcharflagsptr   place to put the last required character flags, or a negative number
3972    cd             contains pointers to tables etc.    bcptr             points to current branch chain
3973    lengthptr      NULL during the real compile phase    cond_depth        conditional nesting depth
3974                   points to length accumulator during pre-compile phase    cd                contains pointers to tables etc.
3975      lengthptr         NULL during the real compile phase
3976                        points to length accumulator during pre-compile phase
3977    
3978  Returns:         TRUE on success  Returns:            TRUE on success
3979                   FALSE, with *errorcodeptr set non-zero on error                      FALSE, with *errorcodeptr set non-zero on error
3980  */  */
3981    
3982  static BOOL  static BOOL
3983  compile_branch(int *optionsptr, pcre_uchar **codeptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3984    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3985    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3986      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3987      branch_chain *bcptr, int cond_depth,
3988    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3989  {  {
3990  int repeat_type, op_type;  int repeat_type, op_type;
3991  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3992  int bravalue = 0;  int bravalue = 0;
3993  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3994  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
3995  pcre_int32 zeroreqchar, zerofirstchar;  pcre_int32 firstcharflags, reqcharflags;
3996    pcre_uint32 zeroreqchar, zerofirstchar;
3997    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3998  pcre_int32 req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3999  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
4000  int after_manual_callout = 0;  int after_manual_callout = 0;
4001  int length_prevgroup = 0;  int length_prevgroup = 0;
4002  register int c;  register pcre_uint32 c;
4003    int escape;
4004  register pcre_uchar *code = *codeptr;  register pcre_uchar *code = *codeptr;
4005  pcre_uchar *last_code = code;  pcre_uchar *last_code = code;
4006  pcre_uchar *orig_code = code;  pcre_uchar *orig_code = code;
# Line 3414  must not do this for other options (e.g. Line 4020  must not do this for other options (e.g.
4020  dynamically as we process the pattern. */  dynamically as we process the pattern. */
4021    
4022  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4023  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
4024  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
4025    #ifndef COMPILE_PCRE32
4026  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
4027    #endif
4028  #else  #else
4029  BOOL utf = FALSE;  BOOL utf = FALSE;
4030  #endif  #endif
4031    
4032  /* Helper variables for OP_XCLASS opcode (for characters > 255). */  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
4033    class_uchardata always so that it can be passed to add_to_class() always,
4034    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
4035    alternative calls for the different cases. */
4036    
4037    pcre_uchar *class_uchardata;
4038  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4039  BOOL xclass;  BOOL xclass;
 pcre_uchar *class_uchardata;  
4040  pcre_uchar *class_uchardata_base;  pcre_uchar *class_uchardata_base;
4041  #endif  #endif
4042    
# Line 3448  to take the zero repeat into account. Th Line 4059  to take the zero repeat into account. Th
4059  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
4060  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
4061    
4062  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
4063    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
4064    
4065  /* The variable req_caseopt contains either the REQ_CASELESS value  /* The variable req_caseopt contains either the REQ_CASELESS value
4066  or zero, according to the current setting of the caseless flag. The  or zero, according to the current setting of the caseless flag. The
# Line 3469  for (;; ptr++) Line 4081  for (;; ptr++)
4081    BOOL is_recurse;    BOOL is_recurse;
4082    BOOL reset_bracount;    BOOL reset_bracount;
4083    int class_has_8bitchar;    int class_has_8bitchar;
4084    int class_single_char;    int class_one_char;
4085    int newoptions;    int newoptions;
4086    int recno;    int recno;
4087    int refsign;    int refsign;
4088    int skipbytes;    int skipbytes;
4089    int subreqchar;    pcre_uint32 subreqchar, subfirstchar;
4090    int subfirstchar;    pcre_int32 subreqcharflags, subfirstcharflags;
4091    int terminator;    int terminator;
4092    int mclength;    unsigned int mclength;
4093    int tempbracount;    unsigned int tempbracount;
4094      pcre_uint32 ec;
4095    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
4096    
4097    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3488  for (;; ptr++) Line 4101  for (;; ptr++)
4101    /* If we are at the end of a nested substitution, revert to the outer level    /* If we are at the end of a nested substitution, revert to the outer level
4102    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
4103    
4104    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
4105      {      {
4106      ptr = nestptr;      ptr = nestptr;
4107      nestptr = NULL;      nestptr = NULL;
# Line 3563  for (;; ptr++) Line 4176  for (;; ptr++)
4176    
4177    /* 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 */
4178    
4179    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
4180      {      {
4181      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
4182        {        {
# Line 3611  for (;; ptr++) Line 4224  for (;; ptr++)
4224      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
4225        {        {
4226        ptr++;        ptr++;
4227        while (*ptr != 0)        while (*ptr != CHAR_NULL)
4228          {          {
4229          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
4230          ptr++;          ptr++;
# Line 3619  for (;; ptr++) Line 4232  for (;; ptr++)
4232          if (utf) FORWARDCHAR(ptr);          if (utf) FORWARDCHAR(ptr);
4233  #endif  #endif
4234          }          }
4235        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
4236    
4237        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
4238        c = 0;        c = 0;
# Line 3641  for (;; ptr++) Line 4254  for (;; ptr++)
4254      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
4255      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
4256      *firstcharptr = firstchar;      *firstcharptr = firstchar;
4257        *firstcharflagsptr = firstcharflags;
4258      *reqcharptr = reqchar;      *reqcharptr = reqchar;
4259        *reqcharflagsptr = reqcharflags;
4260      *codeptr = code;      *codeptr = code;
4261      *ptrptr = ptr;      *ptrptr = ptr;
4262      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3665  for (;; ptr++) Line 4280  for (;; ptr++)
4280      previous = NULL;      previous = NULL;
4281      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
4282        {        {
4283        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4284        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
4285        }        }
4286      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3680  for (;; ptr++) Line 4295  for (;; ptr++)
4295      repeats. The value of reqchar doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4296    
4297      case CHAR_DOT:      case CHAR_DOT:
4298      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4299      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4300        zerofirstcharflags = firstcharflags;
4301      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4302        zeroreqcharflags = reqcharflags;
4303      previous = code;      previous = code;
4304      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4305      break;      break;
# Line 3756  for (;; ptr++) Line 4373  for (;; ptr++)
4373          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4374        {        {
4375        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4376        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4377        zerofirstchar = firstchar;        zerofirstchar = firstchar;
4378          zerofirstcharflags = firstcharflags;
4379        break;        break;
4380        }        }
4381    
# Line 3767  for (;; ptr++) Line 4385  for (;; ptr++)
4385    
4386      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4387    
4388      /* For optimization purposes, we track some properties of the class.      /* For optimization purposes, we track some properties of the class:
4389      class_has_8bitchar will be non-zero, if the class contains at least one      class_has_8bitchar will be non-zero if the class contains at least one <
4390      < 256 character. class_single_char will be 1 if the class contains only      256 character; class_one_char will be 1 if the class contains just one
4391      a single character. */      character. */
4392    
4393      class_has_8bitchar = 0;      class_has_8bitchar = 0;
4394      class_single_char = 0;      class_one_char = 0;
4395    
4396      /* Initialize the 32-char bit map to all zeros. We build the map in a      /* Initialize the 32-char bit map to all zeros. We build the map in a
4397      temporary bit of memory, in case the class contains only 1 character (less      temporary bit of memory, in case the class contains fewer than two
4398      than 256), because in that case the compiled code doesn't use the bit map.      8-bit characters because in that case the compiled code doesn't use the bit
4399      */      map. */
4400    
4401      memset(classbits, 0, 32 * sizeof(pcre_uint8));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4402    
4403  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4404      xclass = FALSE;                           /* No chars >= 256 */      xclass = FALSE;
4405      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4406      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4407  #endif  #endif
4408    
4409      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4410      means that an initial ] is taken as a data character. At the start of the      means that an initial ] is taken as a data character. At the start of the
4411      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4412    
4413      if (c != 0) do      if (c != CHAR_NULL) do
4414        {        {
4415        const pcre_uchar *oldptr;        const pcre_uchar *oldptr;
4416    
# Line 3807  for (;; ptr++) Line 4425  for (;; ptr++)
4425        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4426        data and reset the pointer. This is so that very large classes that        data and reset the pointer. This is so that very large classes that
4427        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4428        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4429          however. */
4430    
4431        if (lengthptr != NULL)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4432          {          {
4433            xclass = TRUE;
4434          *lengthptr += class_uchardata - class_uchardata_base;          *lengthptr += class_uchardata - class_uchardata_base;
4435          class_uchardata = class_uchardata_base;          class_uchardata = class_uchardata_base;
4436          }          }
# Line 3912  for (;; ptr++) Line 4532  for (;; ptr++)
4532              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4533            }            }
4534    
4535          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4536          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4537    
4538          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3928  for (;; ptr++) Line 4548  for (;; ptr++)
4548            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4549    
4550          ptr = tempptr + 1;          ptr = tempptr + 1;
4551          /* Every class contains at least one < 256 characters. */          /* Every class contains at least one < 256 character. */
4552          class_has_8bitchar = 1;          class_has_8bitchar = 1;
4553          /* Every class contains at least two characters. */          /* Every class contains at least two characters. */
4554          class_single_char = 2;          class_one_char = 2;
4555          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4556          }          }
4557    
# Line 3939  for (;; ptr++) Line 4559  for (;; ptr++)
4559        of the specials, which just set a flag. The sequence \b is a special        of the specials, which just set a flag. The sequence \b is a special
4560        case. Inside a class (and only there) it is treated as backspace. We        case. Inside a class (and only there) it is treated as backspace. We
4561        assume that other escapes have more than one character in them, so        assume that other escapes have more than one character in them, so
4562        speculatively set both class_has_8bitchar and class_single_char bigger        speculatively set both class_has_8bitchar and class_one_char bigger
4563        than one. Unrecognized escapes fall through and are either treated        than one. Unrecognized escapes fall through and are either treated
4564        as literal characters (by default), or are faulted if        as literal characters (by default), or are faulted if
4565        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4566    
4567        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4568          {          {
4569          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options,
4570              TRUE);
4571          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4572            if (escape == 0) c = ec;
4573          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4574          else if (-c == ESC_N)            /* \N is not supported in a class */          else if (escape == ESC_N)          /* \N is not supported in a class */
4575            {            {
4576            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4577            goto FAILED;            goto FAILED;
4578            }            }
4579          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4580            {            {
4581            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4582              {              {
# Line 3964  for (;; ptr++) Line 4585  for (;; ptr++)
4585            else inescq = TRUE;            else inescq = TRUE;
4586            continue;            continue;
4587            }            }
4588          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4589    
4590          if (c < 0)          else
4591            {            {
4592            register const pcre_uint8 *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4593            /* Every class contains at least two < 256 characters. */            /* Every class contains at least two < 256 characters. */
4594            class_has_8bitchar++;            class_has_8bitchar++;
4595            /* Every class contains at least two characters. */            /* Every class contains at least two characters. */
4596            class_single_char += 2;            class_one_char += 2;
4597    
4598            switch (-c)            switch (escape)
4599              {              {
4600  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4601              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3984  for (;; ptr++) Line 4605  for (;; ptr++)
4605              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4606              case ESC_SU:              case ESC_SU:
4607              nestptr = ptr;              nestptr = ptr;
4608              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4609              class_has_8bitchar--;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4610              continue;              continue;
4611  #endif  #endif
# Line 4008  for (;; ptr++) Line 4629  for (;; ptr++)
4629    
4630              /* Perl 5.004 onwards omits VT from \s, but we must preserve it              /* Perl 5.004 onwards omits VT from \s, but we must preserve it
4631              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4632              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4633                EBCDIC, so we lazily just adjust the appropriate bit. */
4634    
4635              case ESC_s:              case ESC_s:
4636              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 4022  for (;; ptr++) Line 4644  for (;; ptr++)
4644              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4645              continue;              continue;
4646    
4647                /* The rest apply in both UCP and non-UCP cases. */
4648    
4649              case ESC_h:              case ESC_h:
4650              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4651              SETBIT(classbits, 0x20); /* SPACE */                PRIV(hspace_list), NOTACHAR);
             SETBIT(classbits, 0xa0); /* NSBP */  
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x1680;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x180e;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2000;  
             *class_uchardata++ = 0x200a;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x202f;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x205f;  
             *class_uchardata++ = XCL_SINGLE;  
             *class_uchardata++ = 0x3000;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x1680, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x180e, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2000, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x200a, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x202f, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x205f, class_uchardata);  
               *class_uchardata++ = XCL_SINGLE;  
               class_uchardata += PRIV(ord2utf)(0x3000, class_uchardata);  
               }  
 #endif  
4652              continue;              continue;
4653    
4654              case ESC_H:              case ESC_H:
4655              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4656                {                cd, PRIV(hspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x09/8: x ^= 1 << (0x09%8); break;  
                 case 0x20/8: x ^= 1 << (0x20%8); break;  
                 case 0xa0/8: x ^= 1 << (0xa0%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x0100;  
             *class_uchardata++ = 0x167f;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x1681;  
             *class_uchardata++ = 0x180d;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x180f;  
             *class_uchardata++ = 0x1fff;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x200b;  
             *class_uchardata++ = 0x202e;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2030;  
             *class_uchardata++ = 0x205e;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2060;  
             *class_uchardata++ = 0x2fff;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x3001;  
 #ifdef SUPPORT_UTF  
             if (utf)  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
             else  
 #endif  
               *class_uchardata++ = 0xffff;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x167f, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x1681, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x180d, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x180f, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x1fff, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x200b, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x202e, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2030, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x205e, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2060, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x2fff, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x3001, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
               }  
 #endif  
4657              continue;              continue;
4658    
4659              case ESC_v:              case ESC_v:
4660              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4661              SETBIT(classbits, 0x0b); /* VT */                PRIV(vspace_list), NOTACHAR);
             SETBIT(classbits, 0x0c); /* FF */  
             SETBIT(classbits, 0x0d); /* CR */  
             SETBIT(classbits, 0x85); /* NEL */  
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x2028;  
             *class_uchardata++ = 0x2029;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x2028, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x2029, class_uchardata);  
               }  
 #endif  
4662              continue;              continue;
4663    
4664              case ESC_V:              case ESC_V:
4665              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4666                {                cd, PRIV(vspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x0a/8: x ^= 1 << (0x0a%8);  
                              x ^= 1 << (0x0b%8);  
                              x ^= 1 << (0x0c%8);  
                              x ^= 1 << (0x0d%8);  
                              break;  
                 case 0x85/8: x ^= 1 << (0x85%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
   
 #ifndef COMPILE_PCRE8  
             xclass = TRUE;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x0100;  
             *class_uchardata++ = 0x2027;  
             *class_uchardata++ = XCL_RANGE;  
             *class_uchardata++ = 0x202a;  
 #ifdef SUPPORT_UTF  
             if (utf)  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
             else  
 #endif  
               *class_uchardata++ = 0xffff;  
 #elif defined SUPPORT_UTF  
             if (utf)  
               {  
               xclass = TRUE;  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x2027, class_uchardata);  
               *class_uchardata++ = XCL_RANGE;  
               class_uchardata += PRIV(ord2utf)(0x202a, class_uchardata);  
               class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);  
               }  
 #endif  
4667              continue;              continue;
4668    
4669  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4203  for (;; ptr++) Line 4671  for (;; ptr++)
4671              case ESC_P:              case ESC_P:
4672                {                {
4673                BOOL negated;                BOOL negated;
4674                int pdata;                unsigned int ptype = 0, pdata = 0;
4675                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4676                if (ptype < 0) goto FAILED;                  goto FAILED;
4677                xclass = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_uchardata++ = ((-c == ESC_p) != negated)?  
4678                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4679                *class_uchardata++ = ptype;                *class_uchardata++ = ptype;
4680                *class_uchardata++ = pdata;                *class_uchardata++ = pdata;
# Line 4226  for (;; ptr++) Line 4693  for (;; ptr++)
4693                goto FAILED;                goto FAILED;
4694                }                }
4695              class_has_8bitchar--;    /* Undo the speculative increase. */              class_has_8bitchar--;    /* Undo the speculative increase. */
4696              class_single_char -= 2;  /* Undo the speculative increase. */              class_one_char -= 2;     /* Undo the speculative increase. */
4697              c = *ptr;                /* Get the final character and fall through */              c = *ptr;                /* Get the final character and fall through */
4698              break;              break;
4699              }              }
4700            }            }
4701    
4702          /* Fall through if we have a single character (c >= 0). This may be          /* Fall through if the escape just defined a single character (c >= 0).
4703          greater than 256. */          This may be greater than 256. */
4704    
4705            escape = 0;
4706    
4707          }   /* End of backslash handling */          }   /* End of backslash handling */
4708    
4709        /* A single character may be followed by '-' to form a range. However,        /* A character may be followed by '-' to form a range. However, Perl does
4710        Perl does not permit ']' to be the end of the range. A '-' character        not permit ']' to be the end of the range. A '-' character at the end is
4711        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4712        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4713    
4714        CHECK_RANGE:        CHECK_RANGE:
4715        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4248  for (;; ptr++) Line 4717  for (;; ptr++)
4717          inescq = FALSE;          inescq = FALSE;
4718          ptr += 2;          ptr += 2;
4719          }          }
   
4720        oldptr = ptr;        oldptr = ptr;
4721    
4722        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4723    
4724        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4725    
# Line 4259  for (;; ptr++) Line 4727  for (;; ptr++)
4727    
4728        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4729          {          {
4730          int d;          pcre_uint32 d;
4731          ptr += 2;          ptr += 2;
4732          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4733    
# Line 4275  for (;; ptr++) Line 4743  for (;; ptr++)
4743            break;            break;
4744            }            }
4745    
4746          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          /* Minus (hyphen) at the end of a class is treated as a literal, so put
4747            back the pointer and jump to handle the character that preceded it. */
4748    
4749            if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4750            {            {
4751            ptr = oldptr;            ptr = oldptr;
4752            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4753            }            }
4754    
4755            /* Otherwise, we have a potential range; pick up the next character */
4756    
4757  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4758          if (utf)          if (utf)
4759            {                           /* Braces are required because the */            {                           /* Braces are required because the */
# Line 4296  for (;; ptr++) Line 4769  for (;; ptr++)
4769    
4770          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4771            {            {
4772            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4773              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4774            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4775    
4776            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4777    
4778            if (d < 0)            if (descape != 0)
4779              {              {
4780              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4781                {                {
4782                ptr = oldptr;                ptr = oldptr;
4783                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4784                }                }
4785              }              }
4786            }            }
4787    
4788          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4789          one-character ranges */          one-character ranges. */
4790    
4791          if (d < c)          if (d < c)
4792            {            {
4793            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4794            goto FAILED;            goto FAILED;
4795            }            }
4796            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4797    
4798            /* We have found a character range, so single character optimizations
4799            cannot be done anymore. Any value greater than 1 indicates that there
4800            is more than one character. */
4801    
4802          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          class_one_char = 2;
4803    
4804          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4805    
4806          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4807    
4808          /* Since we found a character range, single character optimizations          class_has_8bitchar +=
4809          cannot be done anymore. */            add_to_class(classbits, &class_uchardata, options, cd, c, d);
         class_single_char = 2;  
   
         /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless  
         matching, we have to use an XCLASS with extra data items. Caseless  
         matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
         if ((d > 255) || (utf && ((options & PCRE_CASELESS) != 0 && d > 127)))  
 #elif defined  SUPPORT_UTF  
         if (utf && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
 #elif !(defined COMPILE_PCRE8)  
         if (d > 255)  
 #endif  
 #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)  
           {  
           xclass = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
4810    
4811  #ifdef SUPPORT_UCP          continue;   /* Go get the next char in the class */
4812  #ifndef COMPILE_PCRE8          }
           if (utf && (options & PCRE_CASELESS) != 0)  
 #else  
           if ((options & PCRE_CASELESS) != 0)  
 #endif  
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
4813    
4814                if (occ == ocd)        /* Handle a single character - we can get here for a normal non-escape
4815                  {        char, or after \ that introduces a single character or for an apparent
4816                  *class_uchardata++ = XCL_SINGLE;        range that isn't. Only the value 1 matters for class_one_char, so don't
4817                  }        increase it if it is already 2 or more ... just in case there's a class
4818                else        with a zillion characters in it. */
4819                  {  
4820                  *class_uchardata++ = XCL_RANGE;        CLASS_SINGLE_CHARACTER:
4821                  class_uchardata += PRIV(ord2utf)(occ, class_uchardata);        if (class_one_char < 2) class_one_char++;
4822                  }  
4823                class_uchardata += PRIV(ord2utf)(ocd, class_uchardata);        /* If class_one_char is 1, we have the first single character in the
4824                }        class, and there have been no prior ranges, or XCLASS items generated by
4825              }        escapes. If this is the final character in the class, we can optimize by
4826  #endif  /* SUPPORT_UCP */        turning the item into a 1-character OP_CHAR[I] if it's positive, or
4827          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4828          to be set. Otherwise, there can be no first char if this item is first,
4829          whatever repeat count may follow. In the case of reqchar, save the
4830          previous value for reinstating. */
4831    
4832            /* Now record the original range, possibly modified for UCP caseless        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
4833            overlapping ranges. */          {
4834            ptr++;
4835            zeroreqchar = reqchar;
4836            zeroreqcharflags = reqcharflags;
4837    
4838            *class_uchardata++ = XCL_RANGE;          if (negate_class)
4839  #ifdef SUPPORT_UTF            {
4840  #ifndef COMPILE_PCRE8  #ifdef SUPPORT_UCP
4841            if (utf)            int d;
             {  
             class_uchardata += PRIV(ord2utf)(c, class_uchardata);  
             class_uchardata += PRIV(ord2utf)(d, class_uchardata);  
             }  
           else  
             {  
             *class_uchardata++ = c;  
             *class_uchardata++ = d;  
             }  
 #else  
           class_uchardata += PRIV(ord2utf)(c, class_uchardata);  
           class_uchardata += PRIV(ord2utf)(d, class_uchardata);  
4842  #endif  #endif
4843  #else /* SUPPORT_UTF */            if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4844            *class_uchardata++ = c;            zerofirstchar = firstchar;
4845            *class_uchardata++ = d;            zerofirstcharflags = firstcharflags;
 #endif /* SUPPORT_UTF */  
4846    
4847            /* With UCP support, we are done. Without UCP support, there is no            /* For caseless UTF-8 mode when UCP support is available, check
4848            caseless matching for UTF characters > 127; we can use the bit map            whether this character has more than one other case. If so, generate
4849            for the smaller ones. As for 16 bit characters without UTF, we            a special OP_NOTPROP item instead of OP_NOTI. */
           can still use  */  
4850    
4851  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4852  #ifndef COMPILE_PCRE8            if (utf && (options & PCRE_CASELESS) != 0 &&
4853            if (utf)                (d = UCD_CASESET(c)) != 0)
 #endif  
             continue;    /* With next character in the class */  
 #endif  /* SUPPORT_UCP */  
   
 #if defined SUPPORT_UTF && !defined(SUPPORT_UCP) && !(defined COMPILE_PCRE8)  
           if (utf)  
4854              {              {
4855              if ((options & PCRE_CASELESS) == 0 || c > 127) continue;              *code++ = OP_NOTPROP;
4856              /* Adjust upper limit and fall through to set up the map */              *code++ = PT_CLIST;
4857              d = 127;              *code++ = d;
4858              }              }
4859            else            else
4860              {  #endif
4861              if (c > 255) continue;            /* Char has only one other case, or UCP not available */
             /* Adjust upper limit and fall through to set up the map */  
             d = 255;  
             }  
 #elif defined SUPPORT_UTF && !defined(SUPPORT_UCP)  
           if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  
           /* Adjust upper limit and fall through to set up the map */  
           d = 127;  
 #else  
           if (c > 255) continue;  
           /* Adjust upper limit and fall through to set up the map */  
           d = 255;  
 #endif  /* SUPPORT_UTF && !SUPPORT_UCP && !COMPILE_PCRE8 */  
           }  
 #endif  /* SUPPORT_UTF || !COMPILE_PCRE8 */  
   
         /* We use the bit map for 8 bit mode, or when the characters fall  
         partially or entirely to [0-255] ([0-127] for UCP) ranges. */  
   
         class_has_8bitchar = 1;  
   
         /* We can save a bit of time by skipping this in the pre-compile. */  
4862    
         if (lengthptr == NULL) for (; c <= d; c++)  
           {  
           classbits[c/8] |= (1 << (c&7));  
           if ((options & PCRE_CASELESS) != 0)  
4863              {              {
4864              int uc = cd->fcc[c]; /* flip case */              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4865              classbits[uc/8] |= (1 << (uc&7));  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4866              }              if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4867            }                code += PRIV(ord2utf)(c, code);
4868                else
         continue;   /* Go get the next char in the class */  
         }  
   
       /* Handle a lone single character - we can get here for a normal  
       non-escape char, or after \ that introduces a single character or for an  
       apparent range that isn't. */  
   
       LONE_SINGLE_CHARACTER:  
   
       /* Only the value of 1 matters for class_single_char. */  
       if (class_single_char < 2) class_single_char++;  
   
       /* If class_charcount is 1, we saw precisely one character. As long as  
       there were no negated characters >= 128 and there was no use of \p or \P,  
       in other words, no use of any XCLASS features, we can optimize.  
   
       In UTF-8 mode, we can optimize the negative case only if there were no  
       characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR  
       operate on single-bytes characters only. This is an historical hangover.  
       Maybe one day we can tidy these opcodes to handle multi-byte characters.  
   
       The optimization throws away the bit map. We turn the item into a  
       1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
       Note that OP_NOT[I] does not support multibyte characters. In the positive  
       case, it can cause firstchar to be set. Otherwise, there can be no first  
       char if this item is first, whatever repeat count may follow. In the case  
       of reqchar, save the previous value for reinstating. */  
   
 #ifdef SUPPORT_UTF  
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET  
         && (!utf || !negate_class || c < (MAX_VALUE_FOR_SINGLE_CHAR + 1)))  
 #else  
       if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)  
4869  #endif  #endif
4870          {                *code++ = c;
4871          ptr++;              }
         zeroreqchar = reqchar;  
4872    
4873          /* The OP_NOT[I] opcodes work on single characters only. */            /* We are finished with this character class */
4874    
4875          if (negate_class)            goto END_CLASS;
           {  
           if (firstchar == REQ_UNSET) firstchar = REQ_NONE;  
           zerofirstchar = firstchar;  
           *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
           *code++ = c;  
           goto NOT_CHAR;  
4876            }            }
4877    
4878          /* For a single, positive character, get the value into mcbuffer, and          /* For a single, positive character, get the value into mcbuffer, and
4879          then we can handle this with the normal one-character code. */          then we can handle this with the normal one-character code. */
4880    
4881  #ifdef SUPPORT_UTF  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4882          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4883            mclength = PRIV(ord2utf)(c, mcbuffer);            mclength = PRIV(ord2utf)(c, mcbuffer);
4884          else          else
# Line 4536  for (;; ptr++) Line 4890  for (;; ptr++)
4890          goto ONE_CHAR;          goto ONE_CHAR;
4891          }       /* End of 1-char optimization */          }       /* End of 1-char optimization */
4892    
4893        /* Handle a character that cannot go in the bit map. */        /* There is more than one character in the class, or an XCLASS item
4894          has been generated. Add this character to the class. */
 #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)  
       if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))  
 #elif defined SUPPORT_UTF  
       if (utf && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
 #elif !(defined COMPILE_PCRE8)  
       if (c > 255)  
 #endif  
   
 #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)  
         {