/[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 1072 by chpe, Tue Oct 16 15:54:40 2012 UTC revision 1389 by ph10, Tue Nov 5 18:05:29 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 80  to check them every time. */ Line 80  to check them every time. */
80  /* Definitions to allow mutual recursion */  /* Definitions to allow mutual recursion */
81    
82  static int  static int
83    add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,    add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84      const pcre_uint32 *, unsigned int);      const pcre_uint32 *, unsigned int);
85    
86  static BOOL  static BOOL
87    compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL,    compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,
88      int, int, int *, int *, branch_chain *, compile_data *, int *);      pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *,
89        compile_data *, int *);
90    
91    
92    
# Line 114  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 121  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  #define REQ_MASK       (REQ_CASELESS | REQ_VARY)  /* 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 254  static const int verbcount = sizeof(verb Line 264  static const int verbcount = sizeof(verb
264  now all in a single string, to reduce the number of relocations when a shared  now all in a single string, to reduce the number of relocations when a shared
265  library is dynamically loaded. The list of lengths is terminated by a zero  library is dynamically loaded. The list of lengths is terminated by a zero
266  length entry. The first three must be alpha, lower, upper, as this is assumed  length entry. The first three must be alpha, lower, upper, as this is assumed
267  for handling case independence. */  for handling case independence. The indices for graph, print, and punct are
268    needed, so identify them. */
269    
270  static const char posix_names[] =  static const char posix_names[] =
271    STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0    STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0
# Line 265  static const char posix_names[] = Line 276  static const char posix_names[] =
276  static const pcre_uint8 posix_name_lengths[] = {  static const pcre_uint8 posix_name_lengths[] = {
277    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };
278    
279    #define PC_GRAPH  8
280    #define PC_PRINT  9
281    #define PC_PUNCT 10
282    
283    
284  /* Table of class bit maps for each POSIX class. Each class is formed from a  /* Table of class bit maps for each POSIX class. Each class is formed from a
285  base map, with an optional addition or removal of another map. Then, for some  base map, with an optional addition or removal of another map. Then, for some
286  classes, there is some additional tweaking: for [:blank:] the vertical space  classes, there is some additional tweaking: for [:blank:] the vertical space
# Line 292  static const int posix_class_maps[] = { Line 308  static const int posix_class_maps[] = {
308    cbit_xdigit,-1,          0              /* xdigit */    cbit_xdigit,-1,          0              /* xdigit */
309  };  };
310    
311  /* Table of substitutes for \d etc when PCRE_UCP is set. The POSIX class  /* Table of substitutes for \d etc when PCRE_UCP is set. They are replaced by
312  substitutes must be in the order of the names, defined above, and there are  Unicode property escapes. */
 both positive and negative cases. NULL means no substitute. */  
313    
314  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
315  static const pcre_uchar string_PNd[]  = {  static const pcre_uchar string_PNd[]  = {
# Line 319  static const pcre_uchar string_pXwd[] = Line 334  static const pcre_uchar string_pXwd[] =
334  static const pcre_uchar *substitutes[] = {  static const pcre_uchar *substitutes[] = {
335    string_PNd,           /* \D */    string_PNd,           /* \D */
336    string_pNd,           /* \d */    string_pNd,           /* \d */
337    string_PXsp,          /* \S */       /* NOTE: Xsp is Perl space */    string_PXsp,          /* \S */   /* Xsp is Perl space, but from 8.34, Perl */
338    string_pXsp,          /* \s */    string_pXsp,          /* \s */   /* space and POSIX space are the same. */
339    string_PXwd,          /* \W */    string_PXwd,          /* \W */
340    string_pXwd           /* \w */    string_pXwd           /* \w */
341  };  };
342    
343    /* The POSIX class substitutes must be in the order of the POSIX class names,
344    defined above, and there are both positive and negative cases. NULL means no
345    general substitute of a Unicode property escape (\p or \P). However, for some
346    POSIX classes (e.g. graph, print, punct) a special property code is compiled
347    directly. */
348    
349  static const pcre_uchar string_pL[] =   {  static const pcre_uchar string_pL[] =   {
350    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
351    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
# Line 372  static const pcre_uchar *posix_substitut Line 393  static const pcre_uchar *posix_substitut
393    NULL,                 /* graph */    NULL,                 /* graph */
394    NULL,                 /* print */    NULL,                 /* print */
395    NULL,                 /* punct */    NULL,                 /* punct */
396    string_pXps,          /* space */    /* NOTE: Xps is POSIX space */    string_pXps,          /* space */   /* Xps is POSIX space, but from 8.34 */
397    string_pXwd,          /* word */    string_pXwd,          /* word  */   /* Perl and POSIX space are the same */
398    NULL,                 /* xdigit */    NULL,                 /* xdigit */
399    /* Negated cases */    /* Negated cases */
400    string_PL,            /* ^alpha */    string_PL,            /* ^alpha */
# Line 387  static const pcre_uchar *posix_substitut Line 408  static const pcre_uchar *posix_substitut
408    NULL,                 /* ^graph */    NULL,                 /* ^graph */
409    NULL,                 /* ^print */    NULL,                 /* ^print */
410    NULL,                 /* ^punct */    NULL,                 /* ^punct */
411    string_PXps,          /* ^space */   /* NOTE: Xps is POSIX space */    string_PXps,          /* ^space */  /* Xps is POSIX space, but from 8.34 */
412    string_PXwd,          /* ^word */    string_PXwd,          /* ^word */   /* Perl and POSIX space are the same */
413    NULL                  /* ^xdigit */    NULL                  /* ^xdigit */
414  };  };
415  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))
# Line 452  static const char error_texts[] = Line 473  static const char error_texts[] =
473    "POSIX collating elements are not supported\0"    "POSIX collating elements are not supported\0"
474    "this version of PCRE is compiled without UTF support\0"    "this version of PCRE is compiled without UTF support\0"
475    "spare error\0"  /** DEAD **/    "spare error\0"  /** DEAD **/
476    "character value in \\x{...} sequence is too large\0"    "character value in \\x{} or \\o{} is too large\0"
477    /* 35 */    /* 35 */
478    "invalid condition (?(0)\0"    "invalid condition (?(0)\0"
479    "\\C not allowed in lookbehind assertion\0"    "\\C not allowed in lookbehind assertion\0"
# Line 484  static const char error_texts[] = Line 505  static const char error_texts[] =
505    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\0"
506    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"
507    /* 60 */    /* 60 */
508    "(*VERB) not recognized\0"    "(*VERB) not recognized or malformed\0"
509    "number is too big\0"    "number is too big\0"
510    "subpattern name expected\0"    "subpattern name expected\0"
511    "digit expected after (?+\0"    "digit expected after (?+\0"
# Line 505  static const char error_texts[] = Line 526  static const char error_texts[] =
526    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"    "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
527    "character value in \\u.... sequence is too large\0"    "character value in \\u.... sequence is too large\0"
528    "invalid UTF-32 string\0"    "invalid UTF-32 string\0"
529      "setting UTF is disabled by the application\0"
530      "non-hex character in \\x{} (closing brace missing?)\0"
531      /* 80 */
532      "non-octal character in \\o{} (closing brace missing?)\0"
533      "missing opening brace after \\o\0"
534      "parentheses are too deeply nested\0"
535    ;    ;
536    
537  /* 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 644  static const pcre_uint8 ebcdic_chartab[] Line 671  static const pcre_uint8 ebcdic_chartab[]
671  #endif  #endif
672    
673    
674    /* This table is used to check whether auto-possessification is possible
675    between adjacent character-type opcodes. The left-hand (repeated) opcode is
676    used to select the row, and the right-hand opcode is use to select the column.
677    A value of 1 means that auto-possessification is OK. For example, the second
678    value in the first row means that \D+\d can be turned into \D++\d.
679    
680    The Unicode property types (\P and \p) have to be present to fill out the table
681    because of what their opcode values are, but the table values should always be
682    zero because property types are handled separately in the code. The last four
683    columns apply to items that cannot be repeated, so there is no need to have
684    rows for them. Note that OP_DIGIT etc. are generated only when PCRE_UCP is
685    *not* set. When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
686    
687    #define APTROWS (LAST_AUTOTAB_LEFT_OP - FIRST_AUTOTAB_OP + 1)
688    #define APTCOLS (LAST_AUTOTAB_RIGHT_OP - FIRST_AUTOTAB_OP + 1)
689    
690    static const pcre_uint8 autoposstab[APTROWS][APTCOLS] = {
691    /* \D \d \S \s \W \w  . .+ \C \P \p \R \H \h \V \v \X \Z \z  $ $M */
692      { 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* \D */
693      { 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 },  /* \d */
694      { 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 },  /* \S */
695      { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* \s */
696      { 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* \W */
697      { 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1 },  /* \w */
698      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* .  */
699      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* .+ */
700      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 },  /* \C */
701      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },  /* \P */
702      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },  /* \p */
703      { 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 },  /* \R */
704      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0 },  /* \H */
705      { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 1, 0, 0 },  /* \h */
706      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0 },  /* \V */
707      { 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0 },  /* \v */
708      { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0 }   /* \X */
709    };
710    
711    
712    /* This table is used to check whether auto-possessification is possible
713    between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP). The
714    left-hand (repeated) opcode is used to select the row, and the right-hand
715    opcode is used to select the column. The values are as follows:
716    
717      0   Always return FALSE (never auto-possessify)
718      1   Character groups are distinct (possessify if both are OP_PROP)
719      2   Check character categories in the same group (general or particular)
720      3   TRUE if the two opcodes are not the same (PROP vs NOTPROP)
721    
722      4   Check left general category vs right particular category
723      5   Check right general category vs left particular category
724    
725      6   Left alphanum vs right general category
726      7   Left space vs right general category
727      8   Left word vs right general category
728    
729      9   Right alphanum vs left general category
730     10   Right space vs left general category
731     11   Right word vs left general category
732    
733     12   Left alphanum vs right particular category
734     13   Left space vs right particular category
735     14   Left word vs right particular category
736    
737     15   Right alphanum vs left particular category
738     16   Right space vs left particular category
739     17   Right word vs left particular category
740    */
741    
742    static const pcre_uint8 propposstab[PT_TABSIZE][PT_TABSIZE] = {
743    /* ANY LAMP GC  PC  SC ALNUM SPACE PXSPACE WORD CLIST UCNC */
744      { 0,  0,  0,  0,  0,    0,    0,      0,   0,    0,   0 },  /* PT_ANY */
745      { 0,  3,  0,  0,  0,    3,    1,      1,   0,    0,   0 },  /* PT_LAMP */
746      { 0,  0,  2,  4,  0,    9,   10,     10,  11,    0,   0 },  /* PT_GC */
747      { 0,  0,  5,  2,  0,   15,   16,     16,  17,    0,   0 },  /* PT_PC */
748      { 0,  0,  0,  0,  2,    0,    0,      0,   0,    0,   0 },  /* PT_SC */
749      { 0,  3,  6, 12,  0,    3,    1,      1,   0,    0,   0 },  /* PT_ALNUM */
750      { 0,  1,  7, 13,  0,    1,    3,      3,   1,    0,   0 },  /* PT_SPACE */
751      { 0,  1,  7, 13,  0,    1,    3,      3,   1,    0,   0 },  /* PT_PXSPACE */
752      { 0,  0,  8, 14,  0,    0,    1,      1,   3,    0,   0 },  /* PT_WORD */
753      { 0,  0,  0,  0,  0,    0,    0,      0,   0,    0,   0 },  /* PT_CLIST */
754      { 0,  0,  0,  0,  0,    0,    0,      0,   0,    0,   3 }   /* PT_UCNC */
755    };
756    
757    /* This table is used to check whether auto-possessification is possible
758    between adjacent Unicode property opcodes (OP_PROP and OP_NOTPROP) when one
759    specifies a general category and the other specifies a particular category. The
760    row is selected by the general category and the column by the particular
761    category. The value is 1 if the particular category is not part of the general
762    category. */
763    
764    static const pcre_uint8 catposstab[7][30] = {
765    /* 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 */
766      { 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 */
767      { 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 */
768      { 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 */
769      { 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 */
770      { 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 */
771      { 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 */
772      { 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 */
773    };
774    
775    /* This table is used when checking ALNUM, (PX)SPACE, SPACE, and WORD against
776    a general or particular category. The properties in each row are those
777    that apply to the character set in question. Duplication means that a little
778    unnecessary work is done when checking, but this keeps things much simpler
779    because they can all use the same code. For more details see the comment where
780    this table is used.
781    
782    Note: SPACE and PXSPACE used to be different because Perl excluded VT from
783    "space", but from Perl 5.18 it's included, so both categories are treated the
784    same here. */
785    
786    static const pcre_uint8 posspropstab[3][4] = {
787      { ucp_L, ucp_N, ucp_N, ucp_Nl },  /* ALNUM, 3rd and 4th values redundant */
788      { ucp_Z, ucp_Z, ucp_C, ucp_Cc },  /* SPACE and PXSPACE, 2nd value redundant */
789      { ucp_L, ucp_N, ucp_P, ucp_Po }   /* WORD */
790    };
791    
792    /* This table is used when converting repeating opcodes into possessified
793    versions as a result of an explicit possessive quantifier such as ++. A zero
794    value means there is no possessified version - in those cases the item in
795    question must be wrapped in ONCE brackets. The table is truncated at OP_CALLOUT
796    because all relevant opcodes are less than that. */
797    
798    static const pcre_uint8 opcode_possessify[] = {
799      0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,   /* 0 - 15  */
800      0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,   /* 16 - 31 */
801    
802      0,                       /* NOTI */
803      OP_POSSTAR, 0,           /* STAR, MINSTAR */
804      OP_POSPLUS, 0,           /* PLUS, MINPLUS */
805      OP_POSQUERY, 0,          /* QUERY, MINQUERY */
806      OP_POSUPTO, 0,           /* UPTO, MINUPTO */
807      0,                       /* EXACT */
808      0, 0, 0, 0,              /* POS{STAR,PLUS,QUERY,UPTO} */
809    
810      OP_POSSTARI, 0,          /* STARI, MINSTARI */
811      OP_POSPLUSI, 0,          /* PLUSI, MINPLUSI */
812      OP_POSQUERYI, 0,         /* QUERYI, MINQUERYI */
813      OP_POSUPTOI, 0,          /* UPTOI, MINUPTOI */
814      0,                       /* EXACTI */
815      0, 0, 0, 0,              /* POS{STARI,PLUSI,QUERYI,UPTOI} */
816    
817      OP_NOTPOSSTAR, 0,        /* NOTSTAR, NOTMINSTAR */
818      OP_NOTPOSPLUS, 0,        /* NOTPLUS, NOTMINPLUS */
819      OP_NOTPOSQUERY, 0,       /* NOTQUERY, NOTMINQUERY */
820      OP_NOTPOSUPTO, 0,        /* NOTUPTO, NOTMINUPTO */
821      0,                       /* NOTEXACT */
822      0, 0, 0, 0,              /* NOTPOS{STAR,PLUS,QUERY,UPTO} */
823    
824      OP_NOTPOSSTARI, 0,       /* NOTSTARI, NOTMINSTARI */
825      OP_NOTPOSPLUSI, 0,       /* NOTPLUSI, NOTMINPLUSI */
826      OP_NOTPOSQUERYI, 0,      /* NOTQUERYI, NOTMINQUERYI */
827      OP_NOTPOSUPTOI, 0,       /* NOTUPTOI, NOTMINUPTOI */
828      0,                       /* NOTEXACTI */
829      0, 0, 0, 0,              /* NOTPOS{STARI,PLUSI,QUERYI,UPTOI} */
830    
831      OP_TYPEPOSSTAR, 0,       /* TYPESTAR, TYPEMINSTAR */
832      OP_TYPEPOSPLUS, 0,       /* TYPEPLUS, TYPEMINPLUS */
833      OP_TYPEPOSQUERY, 0,      /* TYPEQUERY, TYPEMINQUERY */
834      OP_TYPEPOSUPTO, 0,       /* TYPEUPTO, TYPEMINUPTO */
835      0,                       /* TYPEEXACT */
836      0, 0, 0, 0,              /* TYPEPOS{STAR,PLUS,QUERY,UPTO} */
837    
838      OP_CRPOSSTAR, 0,         /* CRSTAR, CRMINSTAR */
839      OP_CRPOSPLUS, 0,         /* CRPLUS, CRMINPLUS */
840      OP_CRPOSQUERY, 0,        /* CRQUERY, CRMINQUERY */
841      OP_CRPOSRANGE, 0,        /* CRRANGE, CRMINRANGE */
842      0, 0, 0, 0,              /* CRPOS{STAR,PLUS,QUERY,RANGE} */
843    
844      0, 0, 0,                 /* CLASS, NCLASS, XCLASS */
845      0, 0,                    /* REF, REFI */
846      0, 0,                    /* DNREF, DNREFI */
847      0, 0                     /* RECURSE, CALLOUT */
848    };
849    
850    
851    
852  /*************************************************  /*************************************************
# Line 665  find_error_text(int n) Line 868  find_error_text(int n)
868  const char *s = error_texts;  const char *s = error_texts;
869  for (; n > 0; n--)  for (; n > 0; n--)
870    {    {
871    while (*s++ != 0) {};    while (*s++ != CHAR_NULL) {};
872    if (*s == 0) return "Error text not found (please report)";    if (*s == CHAR_NULL) return "Error text not found (please report)";
873    }    }
874  return s;  return s;
875  }  }
876    
877    
878    
879  /*************************************************  /*************************************************
880  *           Expand the workspace                 *  *           Expand the workspace                 *
881  *************************************************/  *************************************************/
# Line 749  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 953  return (*p == CHAR_RIGHT_CURLY_BRACKET);
953  *************************************************/  *************************************************/
954    
955  /* 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
956  positive value for a simple escape such as \n, or 0 for a data character  positive value for a simple escape such as \n, or 0 for a data character which
957  which will be placed in chptr. A backreference to group n is returned as  will be placed in chptr. A backreference to group n is returned as negative n.
958  negative n. When UTF-8 is enabled, a positive value greater than 255 may  When UTF-8 is enabled, a positive value greater than 255 may be returned in
959  be returned in chptr.  chptr. On entry, ptr is pointing at the \. On exit, it is on the final
960  On entry,ptr is pointing at the \. On exit, it is on the final character of the  character of the escape sequence.
 escape sequence.  
961    
962  Arguments:  Arguments:
963    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
964    chptr          points to the data character    chptr          points to a returned data character
965    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
966    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
967    options        the options bits    options        the options bits
# Line 771  Returns:         zero => a data characte Line 974  Returns:         zero => a data characte
974  */  */
975    
976  static int  static int
977  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
978    int bracount, int options, BOOL isclass)    int bracount, int options, BOOL isclass)
979  {  {
980  /* PCRE_UTF16 has the same value as PCRE_UTF8. */  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
# Line 786  ptr--;                            /* Set Line 989  ptr--;                            /* Set
989    
990  /* 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. */
991    
992  if (c == 0) *errorcodeptr = ERR1;  if (c == CHAR_NULL) *errorcodeptr = ERR1;
993    
994  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
995  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 795  Otherwise further processing may be requ Line 998  Otherwise further processing may be requ
998  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
999  /* Not alphanumeric */  /* Not alphanumeric */
1000  else if (c < CHAR_0 || c > CHAR_z) {}  else if (c < CHAR_0 || c > CHAR_z) {}
1001  else if ((i = escapes[c - CHAR_0]) != 0) { if (i > 0) c = (pcre_uint32)i; else escape = -i; }  else if ((i = escapes[c - CHAR_0]) != 0)
1002      { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
1003    
1004  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
1005  /* Not alphanumeric */  /* Not alphanumeric */
# Line 845  else Line 1049  else
1049            }            }
1050    
1051  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
1052          if (c > (utf ? 0x10ffff : 0xff))          if (c > (utf ? 0x10ffffU : 0xffU))
1053  #elif defined COMPILE_PCRE16  #elif defined COMPILE_PCRE16
1054          if (c > (utf ? 0x10ffff : 0xffff))          if (c > (utf ? 0x10ffffU : 0xffffU))
1055  #elif defined COMPILE_PCRE32  #elif defined COMPILE_PCRE32
1056          if (utf && c > 0x10ffff)          if (utf && c > 0x10ffffU)
1057  #endif  #endif
1058            {            {
1059            *errorcodeptr = ERR76;            *errorcodeptr = ERR76;
# Line 896  else Line 1100  else
1100      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1101        {        {
1102        const pcre_uchar *p;        const pcre_uchar *p;
1103        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++)
1104          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
1105        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
1106          {          {
1107          escape = ESC_k;          escape = ESC_k;
1108          break;          break;
# Line 961  else Line 1165  else
1165      break;      break;
1166    
1167      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
1168      starting with one that is not zero is not straightforward. By experiment,      starting with one that is not zero is not straightforward. Perl has changed
1169      the way Perl works seems to be as follows:      over the years. Nowadays \g{} for backreferences and \o{} for octal are
1170        recommended to avoid the ambiguities in the old syntax.
1171    
1172      Outside a character class, the digits are read as a decimal number. If the      Outside a character class, the digits are read as a decimal number. If the
1173      number is less than 10, or if there are that many previous extracting      number is less than 8 (used to be 10), or if there are that many previous
1174      left brackets, then it is a back reference. Otherwise, up to three octal      extracting left brackets, then it is a back reference. Otherwise, up to
1175      digits are read to form an escaped byte. Thus \123 is likely to be octal      three octal digits are read to form an escaped byte. Thus \123 is likely to
1176      123 (cf \0123, which is octal 012 followed by the literal 3). If the octal      be octal 123 (cf \0123, which is octal 012 followed by the literal 3). If
1177      value is greater than 377, the least significant 8 bits are taken. Inside a      the octal value is greater than 377, the least significant 8 bits are
1178      character class, \ followed by a digit is always an octal number. */      taken. \8 and \9 are treated as the literal characters 8 and 9.
1179    
1180        Inside a character class, \ followed by a digit is always either a literal
1181        8 or 9 or an octal number. */
1182    
1183      case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5:      case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5:
1184      case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:      case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
# Line 997  else Line 1205  else
1205          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1206          break;          break;
1207          }          }
1208        if (s < 10 || s <= bracount)        if (s < 8 || s <= bracount)  /* Check for back reference */
1209          {          {
1210          escape = -s;          escape = -s;
1211          break;          break;
# Line 1005  else Line 1213  else
1213        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
1214        }        }
1215    
1216      /* Handle an octal number following \. If the first digit is 8 or 9, Perl      /* Handle a digit following \ when the number is not a back reference. If
1217      generates a binary zero byte and treats the digit as a following literal.      the first digit is 8 or 9, Perl used to generate a binary zero byte and
1218      Thus we have to pull back the pointer by one. */      then treat the digit as a following literal. At least by Perl 5.18 this
1219        changed so as not to insert the binary zero. */
1220    
1221      if ((c = *ptr) >= CHAR_8)      if ((c = *ptr) >= CHAR_8) break;
1222        {  
1223        ptr--;      /* Fall through with a digit less than 8 */
       c = 0;  
       break;  
       }  
1224    
1225      /* \0 always starts an octal number, but we may drop through to here with a      /* \0 always starts an octal number, but we may drop through to here with a
1226      larger first octal digit. The original code used just to take the least      larger first octal digit. The original code used just to take the least
# Line 1031  else Line 1237  else
1237  #endif  #endif
1238      break;      break;
1239    
1240      /* \x is complicated. \x{ddd} is a character number which can be greater      /* \o is a relatively new Perl feature, supporting a more general way of
1241      than 0xff in utf or non-8bit mode, but only if the ddd are hex digits.      specifying character codes in octal. The only supported form is \o{ddd}. */
1242      If not, { is treated as a data character. */  
1243        case CHAR_o:
1244        if (ptr[1] != CHAR_LEFT_CURLY_BRACKET) *errorcodeptr = ERR81; else
1245          {
1246          ptr += 2;
1247          c = 0;
1248          overflow = FALSE;
1249          while (*ptr >= CHAR_0 && *ptr <= CHAR_7)
1250            {
1251            register pcre_uint32 cc = *ptr++;
1252            if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1253    #ifdef COMPILE_PCRE32
1254            if (c >= 0x20000000l) { overflow = TRUE; break; }
1255    #endif
1256            c = (c << 3) + cc - CHAR_0 ;
1257    #if defined COMPILE_PCRE8
1258            if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1259    #elif defined COMPILE_PCRE16
1260            if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1261    #elif defined COMPILE_PCRE32
1262            if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1263    #endif
1264            }
1265          if (overflow)
1266            {
1267            while (*ptr >= CHAR_0 && *ptr <= CHAR_7) ptr++;
1268            *errorcodeptr = ERR34;
1269            }
1270          else if (*ptr == CHAR_RIGHT_CURLY_BRACKET)
1271            {
1272            if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1273            }
1274          else *errorcodeptr = ERR80;
1275          }
1276        break;
1277    
1278        /* \x is complicated. In JavaScript, \x must be followed by two hexadecimal
1279        numbers. Otherwise it is a lowercase x letter. */
1280    
1281      case CHAR_x:      case CHAR_x:
1282      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1283        {        {
       /* In JavaScript, \x must be followed by two hexadecimal numbers.  
       Otherwise it is a lowercase x letter. */  
1284        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1285          && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)          && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1286          {          {
# Line 1056  else Line 1297  else
1297  #endif  #endif
1298            }            }
1299          }          }
1300        break;        }    /* End JavaScript handling */
       }  
1301    
1302      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      /* Handle \x in Perl's style. \x{ddd} is a character number which can be
1303        {      greater than 0xff in utf or non-8bit mode, but only if the ddd are hex
1304        const pcre_uchar *pt = ptr + 2;      digits. If not, { used to be treated as a data character. However, Perl
1305        seems to read hex digits up to the first non-such, and ignore the rest, so
1306        that, for example \x{zz} matches a binary zero. This seems crazy, so PCRE
1307        now gives an error. */
1308    
1309        c = 0;      else
1310        overflow = FALSE;        {
1311        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)        if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1312          {          {
1313          register pcre_uint32 cc = *pt++;          ptr += 2;
1314          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          c = 0;
1315            overflow = FALSE;
1316            while (MAX_255(*ptr) && (digitab[*ptr] & ctype_xdigit) != 0)
1317              {
1318              register pcre_uint32 cc = *ptr++;
1319              if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1320    
1321  #ifdef COMPILE_PCRE32  #ifdef COMPILE_PCRE32
1322          if (c >= 0x10000000l) { overflow = TRUE; break; }            if (c >= 0x10000000l) { overflow = TRUE; break; }
1323  #endif  #endif
1324    
1325  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1326          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1327          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
1328  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
1329          if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */            if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
1330          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1331  #endif  #endif
1332    
1333  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
1334          if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }            if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1335  #elif defined COMPILE_PCRE16  #elif defined COMPILE_PCRE16
1336          if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }            if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1337  #elif defined COMPILE_PCRE32  #elif defined COMPILE_PCRE32
1338          if (utf && c > 0x10ffff) { overflow = TRUE; break; }            if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1339  #endif  #endif
1340          }            }
1341    
1342        if (overflow)          if (overflow)
1343          {            {
1344          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;            while (MAX_255(*ptr) && (digitab[*ptr] & ctype_xdigit) != 0) ptr++;
1345          *errorcodeptr = ERR34;            *errorcodeptr = ERR34;
1346          }            }
1347    
1348        if (*pt == CHAR_RIGHT_CURLY_BRACKET)          else if (*ptr == CHAR_RIGHT_CURLY_BRACKET)
1349          {            {
1350          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;            if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1351          ptr = pt;            }
         break;  
         }  
1352    
1353        /* If the sequence of hex digits does not end with '}', then we don't          /* If the sequence of hex digits does not end with '}', give an error.
1354        recognize this construct; fall through to the normal \x handling. */          We used just to recognize this construct and fall through to the normal
1355        }          \x handling, but nowadays Perl gives an error, which seems much more
1356            sensible, so we do too. */
1357    
1358      /* Read just a single-byte hex-defined char */          else *errorcodeptr = ERR79;
1359            }   /* End of \x{} processing */
1360    
1361      c = 0;        /* Read a single-byte hex-defined char (up to two hex digits after \x) */
1362      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)  
1363        {        else
1364        pcre_uint32 cc;                          /* Some compilers don't like */          {
1365        cc = *(++ptr);                           /* ++ in initializers */          c = 0;
1366            while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1367              {
1368              pcre_uint32 cc;                          /* Some compilers don't like */
1369              cc = *(++ptr);                           /* ++ in initializers */
1370  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1371        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
1372        c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
1373  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
1374        if (cc <= CHAR_z) cc += 64;              /* Convert to upper case */            if (cc <= CHAR_z) cc += 64;              /* Convert to upper case */
1375        c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1376  #endif  #endif
1377        }            }
1378            }     /* End of \xdd handling */
1379          }       /* End of Perl-style \x handling */
1380      break;      break;
1381    
1382      /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped.      /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped.
# Line 1132  else Line 1386  else
1386    
1387      case CHAR_c:      case CHAR_c:
1388      c = *(++ptr);      c = *(++ptr);
1389      if (c == 0)      if (c == CHAR_NULL)
1390        {        {
1391        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1392        break;        break;
# Line 1188  if ((options & PCRE_UCP) != 0 && escape Line 1442  if ((options & PCRE_UCP) != 0 && escape
1442  return escape;  return escape;
1443  }  }
1444    
1445    
1446    
1447  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1448  /*************************************************  /*************************************************
1449  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1201  escape sequence. Line 1457  escape sequence.
1457  Argument:  Argument:
1458    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1459    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
1460    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
1461      pdataptr       points to an unsigned int that is set to the detailed property value
1462    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1463    
1464  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
1465  */  */
1466    
1467  static int  static BOOL
1468  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1469      unsigned int *pdataptr, int *errorcodeptr)
1470  {  {
1471  pcre_uchar c;  pcre_uchar c;
1472  int i, bot, top;  int i, bot, top;
# Line 1216  const pcre_uchar *ptr = *ptrptr; Line 1474  const pcre_uchar *ptr = *ptrptr;
1474  pcre_uchar name[32];  pcre_uchar name[32];
1475    
1476  c = *(++ptr);  c = *(++ptr);
1477  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1478    
1479  *negptr = FALSE;  *negptr = FALSE;
1480    
# Line 1233  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1491  if (c == CHAR_LEFT_CURLY_BRACKET)
1491    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1492      {      {
1493      c = *(++ptr);      c = *(++ptr);
1494      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1495      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1496      name[i] = c;      name[i] = c;
1497      }      }
# Line 1263  while (bot < top) Line 1521  while (bot < top)
1521    r = 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);
1522    if (r == 0)    if (r == 0)
1523      {      {
1524      *dptr = PRIV(utt)[i].value;      *ptypeptr = PRIV(utt)[i].type;
1525      return PRIV(utt)[i].type;      *pdataptr = PRIV(utt)[i].value;
1526        return TRUE;
1527      }      }
1528    if (r > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1529    }    }
1530    
1531  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1532  *ptrptr = ptr;  *ptrptr = ptr;
1533  return -1;  return FALSE;
1534    
1535  ERROR_RETURN:  ERROR_RETURN:
1536  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1537  *ptrptr = ptr;  *ptrptr = ptr;
1538  return -1;  return FALSE;
1539  }  }
1540  #endif  #endif
1541    
1542    
1543    
   
1544  /*************************************************  /*************************************************
1545  *         Read repeat counts                     *  *         Read repeat counts                     *
1546  *************************************************/  *************************************************/
# Line 1351  return p; Line 1609  return p;
1609    
1610    
1611  /*************************************************  /*************************************************
 *  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 / UTF-32 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)  
       {  
       pcre_uchar 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 == (int)(ptr - thisname) &&  
           STRNCMP_UC_UC(name, thisname, (unsigned int)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 / UTF-32 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;  
 }  
   
   
   
   
 /*************************************************  
1612  *      Find first significant op code            *  *      Find first significant op code            *
1613  *************************************************/  *************************************************/
1614    
# Line 1684  for (;;) Line 1647  for (;;)
1647    
1648      case OP_CALLOUT:      case OP_CALLOUT:
1649      case OP_CREF:      case OP_CREF:
1650      case OP_NCREF:      case OP_DNCREF:
1651      case OP_RREF:      case OP_RREF:
1652      case OP_NRREF:      case OP_DNRREF:
1653      case OP_DEF:      case OP_DEF:
1654      code += PRIV(OP_lengths)[*code];      code += PRIV(OP_lengths)[*code];
1655      break;      break;
# Line 1700  for (;;) Line 1663  for (;;)
1663    
1664    
1665    
   
1666  /*************************************************  /*************************************************
1667  *        Find the fixed length of a branch       *  *        Find the fixed length of a branch       *
1668  *************************************************/  *************************************************/
# Line 1824  for (;;) Line 1786  for (;;)
1786      case OP_COMMIT:      case OP_COMMIT:
1787      case OP_CREF:      case OP_CREF:
1788      case OP_DEF:      case OP_DEF:
1789        case OP_DNCREF:
1790        case OP_DNRREF:
1791      case OP_DOLL:      case OP_DOLL:
1792      case OP_DOLLM:      case OP_DOLLM:
1793      case OP_EOD:      case OP_EOD:
1794      case OP_EODN:      case OP_EODN:
1795      case OP_FAIL:      case OP_FAIL:
     case OP_NCREF:  
     case OP_NRREF:  
1796      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1797      case OP_PRUNE:      case OP_PRUNE:
1798      case OP_REVERSE:      case OP_REVERSE:
# Line 1852  for (;;) Line 1814  for (;;)
1814      case OP_NOTI:      case OP_NOTI:
1815      branchlength++;      branchlength++;
1816      cc += 2;      cc += 2;
1817  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #ifdef SUPPORT_UTF
1818      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1819  #endif  #endif
1820      break;      break;
# Line 1866  for (;;) Line 1828  for (;;)
1828      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1829      branchlength += (int)GET2(cc,1);      branchlength += (int)GET2(cc,1);
1830      cc += 2 + IMM2_SIZE;      cc += 2 + IMM2_SIZE;
1831  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #ifdef SUPPORT_UTF
1832      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1833  #endif  #endif
1834      break;      break;
1835    
1836      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1837      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1838      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1839        cc += 2;        cc += 2;
1840      cc += 1 + IMM2_SIZE + 1;      cc += 1 + IMM2_SIZE + 1;
1841      break;      break;
# Line 1909  for (;;) Line 1871  for (;;)
1871    
1872      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1873    
 #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32  
     case OP_XCLASS:  
     cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];  
     /* Fall through */  
 #endif  
   
1874      case OP_CLASS:      case OP_CLASS:
1875      case OP_NCLASS:      case OP_NCLASS:
1876    #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1877        case OP_XCLASS:
1878        /* The original code caused an unsigned overflow in 64 bit systems,
1879        so now we use a conditional statement. */
1880        if (op == OP_XCLASS)
1881          cc += GET(cc, 1);
1882        else
1883          cc += PRIV(OP_lengths)[OP_CLASS];
1884    #else
1885      cc += PRIV(OP_lengths)[OP_CLASS];      cc += PRIV(OP_lengths)[OP_CLASS];
1886    #endif
1887    
1888      switch (*cc)      switch (*cc)
1889        {        {
       case OP_CRPLUS:  
       case OP_CRMINPLUS:  
1890        case OP_CRSTAR:        case OP_CRSTAR:
1891        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1892          case OP_CRPLUS:
1893          case OP_CRMINPLUS:
1894        case OP_CRQUERY:        case OP_CRQUERY:
1895        case OP_CRMINQUERY:        case OP_CRMINQUERY:
1896          case OP_CRPOSSTAR:
1897          case OP_CRPOSPLUS:
1898          case OP_CRPOSQUERY:
1899        return -1;        return -1;
1900    
1901        case OP_CRRANGE:        case OP_CRRANGE:
1902        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1903          case OP_CRPOSRANGE:
1904        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1905        branchlength += (int)GET2(cc,1);        branchlength += (int)GET2(cc,1);
1906        cc += 1 + 2 * IMM2_SIZE;        cc += 1 + 2 * IMM2_SIZE;
# Line 1999  for (;;) Line 1969  for (;;)
1969      case OP_QUERYI:      case OP_QUERYI:
1970      case OP_REF:      case OP_REF:
1971      case OP_REFI:      case OP_REFI:
1972        case OP_DNREF:
1973        case OP_DNREFI:
1974      case OP_SBRA:      case OP_SBRA:
1975      case OP_SBRAPOS:      case OP_SBRAPOS:
1976      case OP_SCBRA:      case OP_SCBRA:
# Line 2035  for (;;) Line 2007  for (;;)
2007    
2008    
2009    
   
2010  /*************************************************  /*************************************************
2011  *    Scan compiled regex for specific bracket    *  *    Scan compiled regex for specific bracket    *
2012  *************************************************/  *************************************************/
# Line 2112  for (;;) Line 2083  for (;;)
2083        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2084        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2085        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2086        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2087          code += 2;          code += 2;
2088        break;        break;
2089    
2090        case OP_MARK:        case OP_MARK:
2091        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2092        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2093        case OP_THEN_ARG:        case OP_THEN_ARG:
2094        code += code[1];        code += code[1];
2095        break;        break;
# Line 2232  for (;;) Line 2200  for (;;)
2200        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2201        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2202        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2203        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2204          code += 2;          code += 2;
2205        break;        break;
2206    
2207        case OP_MARK:        case OP_MARK:
2208        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2209        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2210        case OP_THEN_ARG:        case OP_THEN_ARG:
2211        code += code[1];        code += code[1];
2212        break;        break;
# Line 2343  Arguments: Line 2308  Arguments:
2308    endcode     points to where to stop    endcode     points to where to stop
2309    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2310    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2311      recurses    chain of recurse_check to catch mutual recursion
2312    
2313  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2314  */  */
2315    
2316    typedef struct recurse_check {
2317      struct recurse_check *prev;
2318      const pcre_uchar *group;
2319    } recurse_check;
2320    
2321  static BOOL  static BOOL
2322  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2323    BOOL utf, compile_data *cd)    BOOL utf, compile_data *cd, recurse_check *recurses)
2324  {  {
2325  register pcre_uchar c;  register pcre_uchar c;
2326    recurse_check this_recurse;
2327    
2328  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2329       code < endcode;       code < endcode;
2330       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
# Line 2379  for (code = first_significant_code(code Line 2352  for (code = first_significant_code(code
2352    
2353    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2354      {      {
2355      const pcre_uchar *scode;      const pcre_uchar *scode = cd->start_code + GET(code, 1);
2356      BOOL empty_branch;      BOOL empty_branch;
2357    
2358      /* Test for forward reference */      /* Test for forward reference or uncompleted reference. This is disabled
2359        when called to scan a completed pattern by setting cd->start_workspace to
2360        NULL. */
2361    
2362        if (cd->start_workspace != NULL)
2363          {
2364          const pcre_uchar *tcode;
2365          for (tcode = cd->start_workspace; tcode < cd->hwm; tcode += LINK_SIZE)
2366            if ((int)GET(tcode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2367          if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2368          }
2369    
2370        /* If we are scanning a completed pattern, there are no forward references
2371        and all groups are complete. We need to detect whether this is a recursive
2372        call, as otherwise there will be an infinite loop. If it is a recursion,
2373        just skip over it. Simple recursions are easily detected. For mutual
2374        recursions we keep a chain on the stack. */
2375    
2376      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      else
2377        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;        {
2378          recurse_check *r = recurses;
2379          const pcre_uchar *endgroup = scode;
2380    
2381      /* Not a forward reference, test for completed backward reference */        do endgroup += GET(endgroup, 1); while (*endgroup == OP_ALT);
2382          if (code >= scode && code <= endgroup) continue;  /* Simple recursion */
2383    
2384      empty_branch = FALSE;        for (r = recurses; r != NULL; r = r->prev)
2385      scode = cd->start_code + GET(code, 1);          if (r->group == scode) break;
2386      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */        if (r != NULL) continue;   /* Mutual recursion */
2387          }
2388    
2389      /* Completed backwards reference */      /* Completed reference; scan the referenced group, remembering it on the
2390        stack chain to detect mutual recursions. */
2391    
2392        empty_branch = FALSE;
2393        this_recurse.prev = recurses;
2394        this_recurse.group = scode;
2395    
2396      do      do
2397        {        {
2398        if (could_be_empty_branch(scode, endcode, utf, cd))        if (could_be_empty_branch(scode, endcode, utf, cd, &this_recurse))
2399          {          {
2400          empty_branch = TRUE;          empty_branch = TRUE;
2401          break;          break;
# Line 2453  for (code = first_significant_code(code Line 2451  for (code = first_significant_code(code
2451        empty_branch = FALSE;        empty_branch = FALSE;
2452        do        do
2453          {          {
2454          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd, NULL))
2455            empty_branch = TRUE;            empty_branch = TRUE;
2456          code += GET(code, 1);          code += GET(code, 1);
2457          }          }
# Line 2495  for (code = first_significant_code(code Line 2493  for (code = first_significant_code(code
2493        case OP_CRMINSTAR:        case OP_CRMINSTAR:
2494        case OP_CRQUERY:        case OP_CRQUERY:
2495        case OP_CRMINQUERY:        case OP_CRMINQUERY:
2496          case OP_CRPOSSTAR:
2497          case OP_CRPOSQUERY:
2498        break;        break;
2499    
2500        default:                   /* Non-repeat => class must match */        default:                   /* Non-repeat => class must match */
2501        case OP_CRPLUS:            /* These repeats aren't empty */        case OP_CRPLUS:            /* These repeats aren't empty */
2502        case OP_CRMINPLUS:        case OP_CRMINPLUS:
2503          case OP_CRPOSPLUS:
2504        return FALSE;        return FALSE;
2505    
2506        case OP_CRRANGE:        case OP_CRRANGE:
2507        case OP_CRMINRANGE:        case OP_CRMINRANGE:
2508          case OP_CRPOSRANGE:
2509        if (GET2(ccode, 1) > 0) return FALSE;  /* Minimum > 0 */        if (GET2(ccode, 1) > 0) return FALSE;  /* Minimum > 0 */
2510        break;        break;
2511        }        }
# Line 2511  for (code = first_significant_code(code Line 2513  for (code = first_significant_code(code
2513    
2514      /* Opcodes that must match a character */      /* Opcodes that must match a character */
2515    
2516        case OP_ANY:
2517        case OP_ALLANY:
2518        case OP_ANYBYTE:
2519    
2520      case OP_PROP:      case OP_PROP:
2521      case OP_NOTPROP:      case OP_NOTPROP:
2522        case OP_ANYNL:
2523    
2524        case OP_NOT_HSPACE:
2525        case OP_HSPACE:
2526        case OP_NOT_VSPACE:
2527        case OP_VSPACE:
2528      case OP_EXTUNI:      case OP_EXTUNI:
2529    
2530      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2531      case OP_DIGIT:      case OP_DIGIT:
2532      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2533      case OP_WHITESPACE:      case OP_WHITESPACE:
2534      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2535      case OP_WORDCHAR:      case OP_WORDCHAR:
2536      case OP_ANY:  
     case OP_ALLANY:  
     case OP_ANYBYTE:  
2537      case OP_CHAR:      case OP_CHAR:
2538      case OP_CHARI:      case OP_CHARI:
2539      case OP_NOT:      case OP_NOT:
2540      case OP_NOTI:      case OP_NOTI:
2541    
2542      case OP_PLUS:      case OP_PLUS:
2543        case OP_PLUSI:
2544      case OP_MINPLUS:      case OP_MINPLUS:
2545      case OP_POSPLUS:      case OP_MINPLUSI:
2546      case OP_EXACT:  
2547      case OP_NOTPLUS:      case OP_NOTPLUS:
2548        case OP_NOTPLUSI:
2549      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
2550        case OP_NOTMINPLUSI:
2551    
2552        case OP_POSPLUS:
2553        case OP_POSPLUSI:
2554      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
2555        case OP_NOTPOSPLUSI:
2556    
2557        case OP_EXACT:
2558        case OP_EXACTI:
2559      case OP_NOTEXACT:      case OP_NOTEXACT:
2560        case OP_NOTEXACTI:
2561    
2562      case OP_TYPEPLUS:      case OP_TYPEPLUS:
2563      case OP_TYPEMINPLUS:      case OP_TYPEMINPLUS:
2564      case OP_TYPEPOSPLUS:      case OP_TYPEPOSPLUS:
2565      case OP_TYPEEXACT:      case OP_TYPEEXACT:
2566    
2567      return FALSE;      return FALSE;
2568    
2569      /* 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 2558  for (code = first_significant_code(code Line 2583  for (code = first_significant_code(code
2583      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2584      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2585      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2586      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2587        code += 2;        code += 2;
2588      break;      break;
2589    
# Line 2572  for (code = first_significant_code(code Line 2597  for (code = first_significant_code(code
2597      return TRUE;      return TRUE;
2598    
2599      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2600      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO and their caseless and negative versions may be
2601        followed by a multibyte character. */
2602    
2603  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2604      case OP_STAR:      case OP_STAR:
2605      case OP_STARI:      case OP_STARI:
2606        case OP_NOTSTAR:
2607        case OP_NOTSTARI:
2608    
2609      case OP_MINSTAR:      case OP_MINSTAR:
2610      case OP_MINSTARI:      case OP_MINSTARI:
2611        case OP_NOTMINSTAR:
2612        case OP_NOTMINSTARI:
2613    
2614      case OP_POSSTAR:      case OP_POSSTAR:
2615      case OP_POSSTARI:      case OP_POSSTARI:
2616        case OP_NOTPOSSTAR:
2617        case OP_NOTPOSSTARI:
2618    
2619      case OP_QUERY:      case OP_QUERY:
2620      case OP_QUERYI:      case OP_QUERYI:
2621        case OP_NOTQUERY:
2622        case OP_NOTQUERYI:
2623    
2624      case OP_MINQUERY:      case OP_MINQUERY:
2625      case OP_MINQUERYI:      case OP_MINQUERYI:
2626        case OP_NOTMINQUERY:
2627        case OP_NOTMINQUERYI:
2628    
2629      case OP_POSQUERY:      case OP_POSQUERY:
2630      case OP_POSQUERYI:      case OP_POSQUERYI:
2631        case OP_NOTPOSQUERY:
2632        case OP_NOTPOSQUERYI:
2633    
2634      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2635      break;      break;
2636    
2637      case OP_UPTO:      case OP_UPTO:
2638      case OP_UPTOI:      case OP_UPTOI:
2639        case OP_NOTUPTO:
2640        case OP_NOTUPTOI:
2641    
2642      case OP_MINUPTO:      case OP_MINUPTO:
2643      case OP_MINUPTOI:      case OP_MINUPTOI:
2644        case OP_NOTMINUPTO:
2645        case OP_NOTMINUPTOI:
2646    
2647      case OP_POSUPTO:      case OP_POSUPTO:
2648      case OP_POSUPTOI:      case OP_POSUPTOI:
2649        case OP_NOTPOSUPTO:
2650        case OP_NOTPOSUPTOI:
2651    
2652      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]);
2653      break;      break;
2654  #endif  #endif
# Line 2606  for (code = first_significant_code(code Line 2659  for (code = first_significant_code(code
2659      case OP_MARK:      case OP_MARK:
2660      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
2661      case OP_SKIP_ARG:      case OP_SKIP_ARG:
     code += code[1];  
     break;  
   
2662      case OP_THEN_ARG:      case OP_THEN_ARG:
2663      code += code[1];      code += code[1];
2664      break;      break;
# Line 2652  could_be_empty(const pcre_uchar *code, c Line 2702  could_be_empty(const pcre_uchar *code, c
2702  {  {
2703  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2704    {    {
2705    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd, NULL))
2706      return FALSE;      return FALSE;
2707    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2708    }    }
# Line 2662  return TRUE; Line 2712  return TRUE;
2712    
2713    
2714  /*************************************************  /*************************************************
2715  *           Check for POSIX class syntax         *  *        Base opcode of repeated opcodes         *
2716  *************************************************/  *************************************************/
2717    
2718  /* This function is called when the sequence "[:" or "[." or "[=" is  /* Returns the base opcode for repeated single character type opcodes. If the
2719  encountered in a character class. It checks whether this is followed by a  opcode is not a repeated character type, it returns with the original value.
 sequence of characters terminated by a matching ":]" or ".]" or "=]". If we  
 reach an unescaped ']' without the special preceding character, return FALSE.  
   
 Originally, this function only recognized a sequence of letters between the  
 terminators, but it seems that Perl recognizes any sequence of characters,  
 though of course unknown POSIX names are subsequently rejected. Perl gives an  
 "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE  
 didn't consider this to be a POSIX class. Likewise for [:1234:].  
   
 The problem in trying to be exactly like Perl is in the handling of escapes. We  
 have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX  
 class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code  
 below handles the special case of \], but does not try to do any other escape  
 processing. This makes it different from Perl for cases such as [:l\ower:]  
 where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize  
 "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,  
 I think.  
   
 A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.  
 It seems that the appearance of a nested POSIX class supersedes an apparent  
 external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or  
 a digit.  
   
 In Perl, unescaped square brackets may also appear as part of class names. For  
 example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for  
 [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not  
 seem right at all. PCRE does not allow closing square brackets in POSIX class  
 names.  
   
 Arguments:  
   ptr      pointer to the initial [  
   endptr   where to return the end pointer  
2720    
2721  Returns:   TRUE or FALSE  Arguments:  c opcode
2722    Returns:    base opcode for the type
2723  */  */
2724    
2725  static BOOL  static pcre_uchar
2726  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)  get_repeat_base(pcre_uchar c)
2727  {  {
2728  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */  return (c > OP_TYPEPOSUPTO)? c :
2729  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */         (c >= OP_TYPESTAR)?   OP_TYPESTAR :
2730  for (++ptr; *ptr != 0; ptr++)         (c >= OP_NOTSTARI)?   OP_NOTSTARI :
2731    {         (c >= OP_NOTSTAR)?    OP_NOTSTAR :
2732    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)         (c >= OP_STARI)?      OP_STARI :
2733      ptr++;                               OP_STAR;
   else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;  
   else  
     {  
     if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)  
       {  
       *endptr = ptr;  
       return TRUE;  
       }  
     if (*ptr == CHAR_LEFT_SQUARE_BRACKET &&  
          (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||  
           ptr[1] == CHAR_EQUALS_SIGN) &&  
         check_posix_syntax(ptr, endptr))  
       return FALSE;  
     }  
   }  
 return FALSE;  
2734  }  }
2735    
2736    
2737    
2738    #ifdef SUPPORT_UCP
2739  /*************************************************  /*************************************************
2740  *          Check POSIX class name                *  *        Check a character and a property        *
2741  *************************************************/  *************************************************/
2742    
2743  /* This function is called to check the name given in a POSIX-style class entry  /* This function is called by check_auto_possessive() when a property item
2744  such as [:alnum:].  is adjacent to a fixed character.
2745    
2746  Arguments:  Arguments:
2747    ptr        points to the first letter    c            the character
2748    len        the length of the name    ptype        the property type
2749      pdata        the data for the type
2750      negated      TRUE if it's a negated property (\P or \p{^)
2751    
2752  Returns:     a value representing the name, or -1 if unknown  Returns:       TRUE if auto-possessifying is OK
2753  */  */
2754    
2755  static int  static BOOL
2756  check_posix_name(const pcre_uchar *ptr, int len)  check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata,
2757      BOOL negated)
2758  {  {
2759  const char *pn = posix_names;  const pcre_uint32 *p;
2760  register int yield = 0;  const ucd_record *prop = GET_UCD(c);
 while (posix_name_lengths[yield] != 0)  
   {  
   if (len == posix_name_lengths[yield] &&  
     STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;  
   pn += posix_name_lengths[yield] + 1;  
   yield++;  
   }  
 return -1;  
 }  
   
   
 /*************************************************  
 *    Adjust OP_RECURSE items in repeated group   *  
 *************************************************/  
2761    
2762  /* OP_RECURSE items contain an offset from the start of the regex to the group  switch(ptype)
2763  that is referenced. This means that groups can be replicated for fixed    {
2764  repetition simply by copying (because the recursion is allowed to refer to    case PT_LAMP:
2765  earlier groups that are outside the current group). However, when a group is    return (prop->chartype == ucp_Lu ||
2766  optional (i.e. the minimum quantifier is zero), OP_BRAZERO or OP_SKIPZERO is            prop->chartype == ucp_Ll ||
2767  inserted before it, after it has been compiled. This means that any OP_RECURSE            prop->chartype == ucp_Lt) == negated;
 items within it that refer to the group itself or any contained groups have to  
 have their offsets adjusted. That one of the jobs of this function. Before it  
 is called, the partially compiled regex must be temporarily terminated with  
 OP_END.  
2768    
2769  This function has been extended with the possibility of forward references for    case PT_GC:
2770  recursions and subroutine calls. It must also check the list of such references    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
 for the group we are dealing with. If it finds that one of the recursions in  
 the current group is on this list, it adjusts the offset in the list, not the  
 value in the reference (which is a group number).  
2771    
2772  Arguments:    case PT_PC:
2773    group      points to the start of the group    return (pdata == prop->chartype) == negated;
   adjust     the amount by which the group is to be moved  
   utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode  
   cd         contains pointers to tables etc.  
   save_hwm   the hwm forward reference pointer at the start of the group  
2774    
2775  Returns:     nothing    case PT_SC:
2776  */    return (pdata == prop->script) == negated;
2777    
2778  static void    /* These are specials */
 adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,  
   pcre_uchar *save_hwm)  
 {  
 pcre_uchar *ptr = group;  
2779    
2780  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)    case PT_ALNUM:
2781    {    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2782    int offset;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
   pcre_uchar *hc;  
2783    
2784    /* See if this recursion is on the forward reference list. If so, adjust the    /* Perl space used to exclude VT, but from Perl 5.18 it is included, which
2785    reference. */    means that Perl space and POSIX space are now identical. PCRE was changed
2786      at release 8.34. */
2787    
2788    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    case PT_SPACE:    /* Perl space */
2789      case PT_PXSPACE:  /* POSIX space */
2790      switch(c)
2791      {      {
2792      offset = GET(hc, 0);      HSPACE_CASES:
2793      if (cd->start_code + offset == ptr + 1)      VSPACE_CASES:
2794        {      return negated;
2795        PUT(hc, 0, offset + adjust);  
2796        break;      default:
2797        }      return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z) == negated;
2798      }      }
2799      break;  /* Control never reaches here */
2800    
2801    /* Otherwise, adjust the recursion offset if it's after the start of this    case PT_WORD:
2802    group. */    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2803              PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2804              c == CHAR_UNDERSCORE) == negated;
2805    
2806    if (hc >= cd->hwm)    case PT_CLIST:
2807      p = PRIV(ucd_caseless_sets) + prop->caseset;
2808      for (;;)
2809      {      {
2810      offset = GET(ptr, 1);      if (c < *p) return !negated;
2811      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (c == *p++) return negated;
2812      }      }
2813      break;  /* Control never reaches here */
   ptr += 1 + LINK_SIZE;  
2814    }    }
2815    
2816    return FALSE;
2817  }  }
2818    #endif  /* SUPPORT_UCP */
2819    
2820    
2821    
2822  /*************************************************  /*************************************************
2823  *        Insert an automatic callout point       *  *        Fill the character property list        *
2824  *************************************************/  *************************************************/
2825    
2826  /* This function is called when the PCRE_AUTO_CALLOUT option is set, to insert  /* Checks whether the code points to an opcode that can take part in auto-
2827  callout points before each pattern item.  possessification, and if so, fills a list with its properties.
2828    
2829  Arguments:  Arguments:
2830    code           current code pointer    code        points to start of expression
2831    ptr            current pattern pointer    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2832    cd             pointers to tables etc    fcc         points to case-flipping table
2833      list        points to output list
2834                  list[0] will be filled with the opcode
2835                  list[1] will be non-zero if this opcode
2836                    can match an empty character string
2837                  list[2..7] depends on the opcode
2838    
2839  Returns:         new code pointer  Returns:      points to the start of the next opcode if *code is accepted
2840                  NULL if *code is not accepted
2841  */  */
2842    
2843  static pcre_uchar *  static const pcre_uchar *
2844  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)  get_chr_property_list(const pcre_uchar *code, BOOL utf,
2845      const pcre_uint8 *fcc, pcre_uint32 *list)
2846  {  {
2847  *code++ = OP_CALLOUT;  pcre_uchar c = *code;
2848  *code++ = 255;  const pcre_uchar *end;
2849  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  const pcre_uint32 *clist_src;
2850  PUT(code, LINK_SIZE, 0);                       /* Default length */  pcre_uint32 *clist_dest;
2851  return code + 2 * LINK_SIZE;  pcre_uint32 chr;
2852  }  pcre_uchar base;
2853    
2854    list[0] = c;
2855    list[1] = FALSE;
2856    code++;
2857    
2858    if (c >= OP_STAR && c <= OP_TYPEPOSUPTO)
2859      {
2860      base = get_repeat_base(c);
2861      c -= (base - OP_STAR);
2862    
2863  /*************************************************    if (c == OP_UPTO || c == OP_MINUPTO || c == OP_EXACT || c == OP_POSUPTO)
2864  *         Complete a callout item                *      code += IMM2_SIZE;
 *************************************************/  
2865    
2866  /* A callout item contains the length of the next item in the pattern, which    list[1] = (c != OP_PLUS && c != OP_MINPLUS && c != OP_EXACT && c != OP_POSPLUS);
 we can't fill in till after we have reached the relevant point. This is used  
 for both automatic and manual callouts.  
2867    
2868  Arguments:    switch(base)
2869    previous_callout   points to previous callout item      {
2870    ptr                current pattern pointer      case OP_STAR:
2871    cd                 pointers to tables etc      list[0] = OP_CHAR;
2872        break;
2873    
2874  Returns:             nothing      case OP_STARI:
2875  */      list[0] = OP_CHARI;
2876        break;
2877    
2878  static void      case OP_NOTSTAR:
2879  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)      list[0] = OP_NOT;
2880  {      break;
 int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  
 PUT(previous_callout, 2 + LINK_SIZE, length);  
 }  
2881    
2882        case OP_NOTSTARI:
2883        list[0] = OP_NOTI;
2884        break;
2885    
2886        case OP_TYPESTAR:
2887        list[0] = *code;
2888        code++;
2889        break;
2890        }
2891      c = list[0];
2892      }
2893    
2894  #ifdef SUPPORT_UCP  switch(c)
2895  /*************************************************    {
2896  *           Get othercase range                  *    case OP_NOT_DIGIT:
2897  *************************************************/    case OP_DIGIT:
2898      case OP_NOT_WHITESPACE:
2899      case OP_WHITESPACE:
2900      case OP_NOT_WORDCHAR:
2901      case OP_WORDCHAR:
2902      case OP_ANY:
2903      case OP_ALLANY:
2904      case OP_ANYNL:
2905      case OP_NOT_HSPACE:
2906      case OP_HSPACE:
2907      case OP_NOT_VSPACE:
2908      case OP_VSPACE:
2909      case OP_EXTUNI:
2910      case OP_EODN:
2911      case OP_EOD:
2912      case OP_DOLL:
2913      case OP_DOLLM:
2914      return code;
2915    
2916  /* This function is passed the start and end of a class range, in UTF-8 mode    case OP_CHAR:
2917  with UCP support. It searches up the characters, looking for ranges of    case OP_NOT:
2918  characters in the "other" case. Each call returns the next one, updating the    GETCHARINCTEST(chr, code);
2919  start address. A character with multiple other cases is returned on its own    list[2] = chr;
2920  with a special return value.    list[3] = NOTACHAR;
2921      return code;
2922    
2923  Arguments:    case OP_CHARI:
2924    cptr        points to starting character value; updated    case OP_NOTI:
2925    d           end value    list[0] = (c == OP_CHARI) ? OP_CHAR : OP_NOT;
2926    ocptr       where to put start of othercase range    GETCHARINCTEST(chr, code);
2927    odptr       where to put end of othercase range    list[2] = chr;
2928    
2929  Yield:        -1 when no more  #ifdef SUPPORT_UCP
2930                 0 when a range is returned    if (chr < 128 || (chr < 256 && !utf))
2931                >0 the CASESET offset for char with multiple other cases      list[3] = fcc[chr];
2932                  in this case, ocptr contains the original    else
2933  */      list[3] = UCD_OTHERCASE(chr);
2934    #elif defined SUPPORT_UTF || !defined COMPILE_PCRE8
2935      list[3] = (chr < 256) ? fcc[chr] : chr;
2936    #else
2937      list[3] = fcc[chr];
2938    #endif
2939    
2940  static int    /* The othercase might be the same value. */
 get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,  
   pcre_uint32 *odptr)  
 {  
 pcre_uint32 c, othercase, next;  
 int co;  
2941    
2942  /* Find the first character that has an other case. If it has multiple other    if (chr == list[3])
2943  cases, return its case offset value. */      list[3] = NOTACHAR;
2944      else
2945        list[4] = NOTACHAR;
2946      return code;
2947    
2948  for (c = *cptr; c <= d; c++)  #ifdef SUPPORT_UCP
2949    {    case OP_PROP:
2950    if ((co = UCD_CASESET(c)) != 0)    case OP_NOTPROP:
2951      if (code[0] != PT_CLIST)
2952      {      {
2953      *ocptr = c++;   /* Character that has the set */      list[2] = code[0];
2954      *cptr = c;      /* Rest of input range */      list[3] = code[1];
2955      return co;      return code + 2;
2956      }      }
   if ((othercase = UCD_OTHERCASE(c)) != c) break;  
   }  
2957    
2958  if (c > d) return -1;  /* Reached end of range */    /* Convert only if we have enough space. */
2959    
2960  *ocptr = othercase;    clist_src = PRIV(ucd_caseless_sets) + code[1];
2961  next = othercase + 1;    clist_dest = list + 2;
2962      code += 2;
2963    
2964  for (++c; c <= d; c++)    do {
2965    {       if (clist_dest >= list + 8)
2966    if (UCD_OTHERCASE(c) != next) break;         {
2967    next++;         /* Early return if there is not enough space. This should never
2968    }         happen, since all clists are shorter than 5 character now. */
2969           list[2] = code[0];
2970           list[3] = code[1];
2971           return code;
2972           }
2973         *clist_dest++ = *clist_src;
2974         }
2975      while(*clist_src++ != NOTACHAR);
2976    
2977  *odptr = next - 1;     /* End of othercase range */    /* All characters are stored. The terminating NOTACHAR
2978  *cptr = c;             /* Rest of input range */    is copied form the clist itself. */
2979  return 0;  
2980      list[0] = (c == OP_PROP) ? OP_CHAR : OP_NOT;
2981      return code;
2982    #endif
2983    
2984      case OP_NCLASS:
2985      case OP_CLASS:
2986    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2987      case OP_XCLASS:
2988      if (c == OP_XCLASS)
2989        end = code + GET(code, 0) - 1;
2990      else
2991    #endif
2992        end = code + 32 / sizeof(pcre_uchar);
2993    
2994      switch(*end)
2995        {
2996        case OP_CRSTAR:
2997        case OP_CRMINSTAR:
2998        case OP_CRQUERY:
2999        case OP_CRMINQUERY:
3000        case OP_CRPOSSTAR:
3001        case OP_CRPOSQUERY:
3002        list[1] = TRUE;
3003        end++;
3004        break;
3005    
3006        case OP_CRPLUS:
3007        case OP_CRMINPLUS:
3008        case OP_CRPOSPLUS:
3009        end++;
3010        break;
3011    
3012        case OP_CRRANGE:
3013        case OP_CRMINRANGE:
3014        case OP_CRPOSRANGE:
3015        list[1] = (GET2(end, 1) == 0);
3016        end += 1 + 2 * IMM2_SIZE;
3017        break;
3018        }
3019      list[2] = end - code;
3020      return end;
3021      }
3022    return NULL;    /* Opcode not accepted */
3023  }  }
3024    
3025    
3026    
3027  /*************************************************  /*************************************************
3028  *        Check a character and a property        *  *    Scan further character sets for match       *
3029  *************************************************/  *************************************************/
3030    
3031  /* This function is called by check_auto_possessive() when a property item  /* Checks whether the base and the current opcode have a common character, in
3032  is adjacent to a fixed character.  which case the base cannot be possessified.
3033    
3034  Arguments:  Arguments:
3035    c            the character    code        points to the byte code
3036    ptype        the property type    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
3037    pdata        the data for the type    cd          static compile data
3038    negated      TRUE if it's a negated property (\P or \p{^)    base_list   the data list of the base opcode
3039    
3040  Returns:       TRUE if auto-possessifying is OK  Returns:      TRUE if the auto-possessification is possible
3041  */  */
3042    
3043  static BOOL  static BOOL
3044  check_char_prop(pcre_uint32 c, int ptype, int pdata, BOOL negated)  compare_opcodes(const pcre_uchar *code, BOOL utf, const compile_data *cd,
3045      const pcre_uint32 *base_list, const pcre_uchar *base_end)
3046  {  {
3047  #ifdef SUPPORT_UCP  pcre_uchar c;
3048  const pcre_uint32 *p;  pcre_uint32 list[8];
3049  #endif  const pcre_uint32 *chr_ptr;
3050    const pcre_uint32 *ochr_ptr;
3051  const ucd_record *prop = GET_UCD(c);  const pcre_uint32 *list_ptr;
3052    const pcre_uchar *next_code;
3053    const pcre_uint8 *class_bitset;
3054    const pcre_uint32 *set1, *set2, *set_end;
3055    pcre_uint32 chr;
3056    BOOL accepted, invert_bits;
3057    
3058    /* Note: the base_list[1] contains whether the current opcode has greedy
3059    (represented by a non-zero value) quantifier. This is a different from
3060    other character type lists, which stores here that the character iterator
3061    matches to an empty string (also represented by a non-zero value). */
3062    
3063  switch(ptype)  for(;;)
3064    {    {
3065    case PT_LAMP:    /* All operations move the code pointer forward.
3066    return (prop->chartype == ucp_Lu ||    Therefore infinite recursions are not possible. */
           prop->chartype == ucp_Ll ||  
           prop->chartype == ucp_Lt) == negated;  
   
   case PT_GC:  
   return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;  
3067    
3068    case PT_PC:    c = *code;
   return (pdata == prop->chartype) == negated;  
3069    
3070    case PT_SC:    /* Skip over callouts */
   return (pdata == prop->script) == negated;  
3071    
3072    /* These are specials */    if (c == OP_CALLOUT)
3073        {
3074        code += PRIV(OP_lengths)[c];
3075        continue;
3076        }
3077    
3078    case PT_ALNUM:    if (c == OP_ALT)
3079    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||      {
3080            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;      do code += GET(code, 1); while (*code == OP_ALT);
3081        c = *code;
3082        }
3083    
3084    case PT_SPACE:    /* Perl space */    switch(c)
3085    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||      {
3086            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)      case OP_END:
3087            == negated;      case OP_KETRPOS:
3088        /* TRUE only in greedy case. The non-greedy case could be replaced by
3089        an OP_EXACT, but it is probably not worth it. (And note that OP_EXACT
3090        uses more memory, which we cannot get at this stage.) */
3091    
3092    case PT_PXSPACE:  /* POSIX space */      return base_list[1] != 0;
   return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||  
           c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||  
           c == CHAR_FF || c == CHAR_CR)  
           == negated;  
3093    
3094    case PT_WORD:      case OP_KET:
3095    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||      /* If the bracket is capturing, and referenced by an OP_RECURSE, or
3096            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||      it is an atomic sub-pattern (assert, once, etc.) the non-greedy case
3097            c == CHAR_UNDERSCORE) == negated;      cannot be converted to a possessive form. */
3098    
3099  #ifdef SUPPORT_UCP      if (base_list[1] == 0) return FALSE;
3100    case PT_CLIST:  
3101    p = PRIV(ucd_caseless_sets) + prop->caseset;      switch(*(code - GET(code, 1)))
3102    for (;;)        {
3103      {        case OP_ASSERT:
3104      if ((unsigned int)c < *p) return !negated;        case OP_ASSERT_NOT:
3105      if ((unsigned int)c == *p++) return negated;        case OP_ASSERTBACK:
3106      }        case OP_ASSERTBACK_NOT:
3107    break;  /* Control never reaches here */        case OP_ONCE:
3108  #endif        case OP_ONCE_NC:
3109    }        /* Atomic sub-patterns and assertions can always auto-possessify their
3110          last iterator. */
3111          return TRUE;
3112          }
3113    
3114  return FALSE;      code += PRIV(OP_lengths)[c];
3115  }      continue;
 #endif  /* SUPPORT_UCP */  
3116    
3117        case OP_ONCE:
3118        case OP_ONCE_NC:
3119        case OP_BRA:
3120        case OP_CBRA:
3121        next_code = code + GET(code, 1);
3122        code += PRIV(OP_lengths)[c];
3123    
3124        while (*next_code == OP_ALT)
3125          {
3126          if (!compare_opcodes(code, utf, cd, base_list, base_end)) return FALSE;
3127          code = next_code + 1 + LINK_SIZE;
3128          next_code += GET(next_code, 1);
3129          }
3130        continue;
3131    
3132  /*************************************************      case OP_BRAZERO:
3133  *     Check if auto-possessifying is possible    *      case OP_BRAMINZERO:
 *************************************************/  
3134    
3135  /* This function is called for unlimited repeats of certain items, to see      next_code = code + 1;
3136  whether the next thing could possibly match the repeated item. If not, it makes      if (*next_code != OP_BRA && *next_code != OP_CBRA
3137  sense to automatically possessify the repeated item.          && *next_code != OP_ONCE && *next_code != OP_ONCE_NC) return FALSE;
3138    
3139        do next_code += GET(next_code, 1); while (*next_code == OP_ALT);
3140    
3141        /* The bracket content will be checked by the
3142        OP_BRA/OP_CBRA case above. */
3143        next_code += 1 + LINK_SIZE;
3144        if (!compare_opcodes(next_code, utf, cd, base_list, base_end))
3145          return FALSE;
3146    
3147  Arguments:      code += PRIV(OP_lengths)[c];
3148    previous      pointer to the repeated opcode      continue;
3149    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode      }
   ptr           next character in pattern  
   options       options bits  
   cd            contains pointers to tables etc.  
3150    
3151  Returns:        TRUE if possessifying is wanted    /* Check for a supported opcode, and load its properties. */
 */  
3152    
3153  static BOOL    code = get_chr_property_list(code, utf, cd->fcc, list);
3154  check_auto_possessive(const pcre_uchar *previous, BOOL utf,    if (code == NULL) return FALSE;    /* Unsupported */
   const pcre_uchar *ptr, int options, compile_data *cd)  
 {  
 pcre_uint32 c = NOTACHAR;  
 pcre_uint32 next;  
 int escape;  
 int op_code = *previous++;  
3155    
3156  /* Skip whitespace and comments in extended mode */    /* If either opcode is a small character list, set pointers for comparing
3157      characters from that list with another list, or with a property. */
3158    
3159  if ((options & PCRE_EXTENDED) != 0)    if (base_list[0] == OP_CHAR)
   {  
   for (;;)  
3160      {      {
3161      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      chr_ptr = base_list + 2;
3162      if (*ptr == CHAR_NUMBER_SIGN)      list_ptr = list;
3163        {      }
3164        ptr++;    else if (list[0] == OP_CHAR)
3165        while (*ptr != 0)      {
3166          {      chr_ptr = list + 2;
3167          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }      list_ptr = base_list;
         ptr++;  
 #ifdef SUPPORT_UTF  
         if (utf) FORWARDCHAR(ptr);  
 #endif  
         }  
       }  
     else break;  
3168      }      }
   }  
3169    
3170  /* If the next item is one that we can handle, get its value. A non-negative    /* Character bitsets can also be compared to certain opcodes. */
 value is a character, a negative value is an escape value. */  
3171    
3172  if (*ptr == CHAR_BACKSLASH)    else if (base_list[0] == OP_CLASS || list[0] == OP_CLASS
3173    {  #ifdef COMPILE_PCRE8
3174    int temperrorcode = 0;        /* In 8 bit, non-UTF mode, OP_CLASS and OP_NCLASS are the same. */
3175    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);        || (!utf && (base_list[0] == OP_NCLASS || list[0] == OP_NCLASS))
   if (temperrorcode != 0) return FALSE;  
   ptr++;    /* Point after the escape sequence */  
   }  
 else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)  
   {  
   escape = 0;  
 #ifdef SUPPORT_UTF  
   if (utf) { GETCHARINC(next, ptr); } else  
3176  #endif  #endif
3177    next = *ptr++;        )
   }  
 else return FALSE;  
   
 /* Skip whitespace and comments in extended mode */  
   
 if ((options & PCRE_EXTENDED) != 0)  
   {  
   for (;;)  
3178      {      {
3179      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;  #ifdef COMPILE_PCRE8
3180      if (*ptr == CHAR_NUMBER_SIGN)      if (base_list[0] == OP_CLASS || (!utf && base_list[0] == OP_NCLASS))
3181        {  #else
3182        ptr++;      if (base_list[0] == OP_CLASS)
       while (*ptr != 0)  
         {  
         if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }  
         ptr++;  
 #ifdef SUPPORT_UTF  
         if (utf) FORWARDCHAR(ptr);  
3183  #endif  #endif
3184          }        {
3185          set1 = (pcre_uint32 *)(base_end - base_list[2]);
3186          list_ptr = list;
3187          }
3188        else
3189          {
3190          set1 = (pcre_uint32 *)(code - list[2]);
3191          list_ptr = base_list;
3192        }        }
     else break;  
     }  
   }  
3193    
3194  /* If the next thing is itself optional, we have to give up. */      invert_bits = FALSE;
3195        switch(list_ptr[0])
3196          {
3197          case OP_CLASS:
3198          case OP_NCLASS:
3199          set2 = (pcre_uint32 *)
3200            ((list_ptr == list ? code : base_end) - list_ptr[2]);
3201          break;
3202    
3203  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        /* OP_XCLASS cannot be supported here, because its bitset
3204    STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)        is not necessarily complete. E.g: [a-\0x{200}] is stored
3205      return FALSE;        as a character range, and the appropriate bits are not set. */
3206    
3207  /* If the previous item is a character, get its value. */        case OP_NOT_DIGIT:
3208            invert_bits = TRUE;
3209            /* Fall through */
3210          case OP_DIGIT:
3211            set2 = (pcre_uint32 *)(cd->cbits + cbit_digit);
3212            break;
3213    
3214  if (op_code == OP_CHAR || op_code == OP_CHARI ||        case OP_NOT_WHITESPACE:
3215      op_code == OP_NOT || op_code == OP_NOTI)          invert_bits = TRUE;
3216    //if (escape == 0) switch(op_code)          /* Fall through */
3217    {        case OP_WHITESPACE:
3218  #ifdef SUPPORT_UTF          set2 = (pcre_uint32 *)(cd->cbits + cbit_space);
3219    GETCHARTEST(c, previous);          break;
 #else  
   c = *previous;  
 #endif  
   }  
3220    
3221  /* Now compare the next item with the previous opcode. First, handle cases when        case OP_NOT_WORDCHAR:
3222  the next item is a character. */          invert_bits = TRUE;
3223            /* Fall through */
3224          case OP_WORDCHAR:
3225            set2 = (pcre_uint32 *)(cd->cbits + cbit_word);
3226            break;
3227    
3228  if (escape == 0)        default:
3229    {        return FALSE;
3230    /* For a caseless UTF match, the next character may have more than one other        }
3231    case, which maps to the special PT_CLIST property. Check this first. */  
3232        /* Compare 4 bytes to improve speed. */
3233  #ifdef SUPPORT_UCP      set_end = set1 + (32 / 4);
3234    if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)      if (invert_bits)
3235      {        {
3236      int ocs = UCD_CASESET(next);        do
3237      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);          {
3238            if ((*set1++ & ~(*set2++)) != 0) return FALSE;
3239            }
3240          while (set1 < set_end);
3241          }
3242        else
3243          {
3244          do
3245            {
3246            if ((*set1++ & *set2++) != 0) return FALSE;
3247            }
3248          while (set1 < set_end);
3249          }
3250    
3251        if (list[1] == 0) return TRUE;
3252        /* Might be an empty repeat. */
3253        continue;
3254      }      }
 #endif  
3255    
3256    switch(op_code)    /* Some property combinations also acceptable. Unicode property opcodes are
3257      processed specially; the rest can be handled with a lookup table. */
3258    
3259      else
3260      {      {
3261      case OP_CHAR:      pcre_uint32 leftop, rightop;
     return c != next;  
3262    
3263      /* For CHARI (caseless character) we must check the other case. If we have      leftop = base_list[0];
3264      Unicode property support, we can use it to test the other case of      rightop = list[0];
     high-valued characters. We know that next can have only one other case,  
     because multi-other-case characters are dealt with above. */  
3265    
3266      case OP_CHARI:  #ifdef SUPPORT_UCP
3267      if (c == next) return FALSE;      accepted = FALSE; /* Always set in non-unicode case. */
3268  #ifdef SUPPORT_UTF      if (leftop == OP_PROP || leftop == OP_NOTPROP)
     if (utf)  
3269        {        {
3270        pcre_uint32 othercase;        if (rightop == OP_EOD)
3271        if (next < 128) othercase = cd->fcc[next]; else          accepted = TRUE;
3272          else if (rightop == OP_PROP || rightop == OP_NOTPROP)
3273            {
3274            int n;
3275            const pcre_uint8 *p;
3276            BOOL same = leftop == rightop;
3277            BOOL lisprop = leftop == OP_PROP;
3278            BOOL risprop = rightop == OP_PROP;
3279            BOOL bothprop = lisprop && risprop;
3280    
3281            /* There's a table that specifies how each combination is to be
3282            processed:
3283              0   Always return FALSE (never auto-possessify)
3284              1   Character groups are distinct (possessify if both are OP_PROP)
3285              2   Check character categories in the same group (general or particular)
3286              3   Return TRUE if the two opcodes are not the same
3287              ... see comments below
3288            */
3289    
3290            n = propposstab[base_list[2]][list[2]];
3291            switch(n)
3292              {
3293              case 0: break;
3294              case 1: accepted = bothprop; break;
3295              case 2: accepted = (base_list[3] == list[3]) != same; break;
3296              case 3: accepted = !same; break;
3297    
3298              case 4:  /* Left general category, right particular category */
3299              accepted = risprop && catposstab[base_list[3]][list[3]] == same;
3300              break;
3301    
3302              case 5:  /* Right general category, left particular category */
3303              accepted = lisprop && catposstab[list[3]][base_list[3]] == same;
3304              break;
3305    
3306              /* This code is logically tricky. Think hard before fiddling with it.
3307              The posspropstab table has four entries per row. Each row relates to
3308              one of PCRE's special properties such as ALNUM or SPACE or WORD.
3309              Only WORD actually needs all four entries, but using repeats for the
3310              others means they can all use the same code below.
3311    
3312              The first two entries in each row are Unicode general categories, and
3313              apply always, because all the characters they include are part of the
3314              PCRE character set. The third and fourth entries are a general and a
3315              particular category, respectively, that include one or more relevant
3316              characters. One or the other is used, depending on whether the check
3317              is for a general or a particular category. However, in both cases the
3318              category contains more characters than the specials that are defined
3319              for the property being tested against. Therefore, it cannot be used
3320              in a NOTPROP case.
3321    
3322              Example: the row for WORD contains ucp_L, ucp_N, ucp_P, ucp_Po.
3323              Underscore is covered by ucp_P or ucp_Po. */
3324    
3325              case 6:  /* Left alphanum vs right general category */
3326              case 7:  /* Left space vs right general category */
3327              case 8:  /* Left word vs right general category */
3328              p = posspropstab[n-6];
3329              accepted = risprop && lisprop ==
3330                (list[3] != p[0] &&
3331                 list[3] != p[1] &&
3332                (list[3] != p[2] || !lisprop));
3333              break;
3334    
3335              case 9:   /* Right alphanum vs left general category */
3336              case 10:  /* Right space vs left general category */
3337              case 11:  /* Right word vs left general category */
3338              p = posspropstab[n-9];
3339              accepted = lisprop && risprop ==
3340                (base_list[3] != p[0] &&
3341                 base_list[3] != p[1] &&
3342                (base_list[3] != p[2] || !risprop));
3343              break;
3344    
3345              case 12:  /* Left alphanum vs right particular category */
3346              case 13:  /* Left space vs right particular category */
3347              case 14:  /* Left word vs right particular category */
3348              p = posspropstab[n-12];
3349              accepted = risprop && lisprop ==
3350                (catposstab[p[0]][list[3]] &&
3351                 catposstab[p[1]][list[3]] &&
3352                (list[3] != p[3] || !lisprop));
3353              break;
3354    
3355              case 15:  /* Right alphanum vs left particular category */
3356              case 16:  /* Right space vs left particular category */
3357              case 17:  /* Right word vs left particular category */
3358              p = posspropstab[n-15];
3359              accepted = lisprop && risprop ==
3360                (catposstab[p[0]][base_list[3]] &&
3361                 catposstab[p[1]][base_list[3]] &&
3362                (base_list[3] != p[3] || !risprop));
3363              break;
3364              }
3365            }
3366          }
3367    
3368        else
3369    #endif  /* SUPPORT_UCP */
3370    
3371        accepted = leftop >= FIRST_AUTOTAB_OP && leftop <= LAST_AUTOTAB_LEFT_OP &&
3372               rightop >= FIRST_AUTOTAB_OP && rightop <= LAST_AUTOTAB_RIGHT_OP &&
3373               autoposstab[leftop - FIRST_AUTOTAB_OP][rightop - FIRST_AUTOTAB_OP];
3374    
3375        if (!accepted)
3376          return FALSE;
3377    
3378        if (list[1] == 0) return TRUE;
3379        /* Might be an empty repeat. */
3380        continue;
3381        }
3382    
3383      /* Control reaches here only if one of the items is a small character list.
3384      All characters are checked against the other side. */
3385    
3386      do
3387        {
3388        chr = *chr_ptr;
3389    
3390        switch(list_ptr[0])
3391          {
3392          case OP_CHAR:
3393          ochr_ptr = list_ptr + 2;
3394          do
3395            {
3396            if (chr == *ochr_ptr) return FALSE;
3397            ochr_ptr++;
3398            }
3399          while(*ochr_ptr != NOTACHAR);
3400          break;
3401    
3402          case OP_NOT:
3403          ochr_ptr = list_ptr + 2;
3404          do
3405            {
3406            if (chr == *ochr_ptr)
3407              break;
3408            ochr_ptr++;
3409            }
3410          while(*ochr_ptr != NOTACHAR);
3411          if (*ochr_ptr == NOTACHAR) return FALSE;   /* Not found */
3412          break;
3413    
3414          /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not*
3415          set. When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
3416    
3417          case OP_DIGIT:
3418          if (chr < 256 && (cd->ctypes[chr] & ctype_digit) != 0) return FALSE;
3419          break;
3420    
3421          case OP_NOT_DIGIT:
3422          if (chr > 255 || (cd->ctypes[chr] & ctype_digit) == 0) return FALSE;
3423          break;
3424    
3425          case OP_WHITESPACE:
3426          if (chr < 256 && (cd->ctypes[chr] & ctype_space) != 0) return FALSE;
3427          break;
3428    
3429          case OP_NOT_WHITESPACE:
3430          if (chr > 255 || (cd->ctypes[chr] & ctype_space) == 0) return FALSE;
3431          break;
3432    
3433          case OP_WORDCHAR:
3434          if (chr < 255 && (cd->ctypes[chr] & ctype_word) != 0) return FALSE;
3435          break;
3436    
3437          case OP_NOT_WORDCHAR:
3438          if (chr > 255 || (cd->ctypes[chr] & ctype_word) == 0) return FALSE;
3439          break;
3440    
3441          case OP_HSPACE:
3442          switch(chr)
3443            {
3444            HSPACE_CASES: return FALSE;
3445            default: break;
3446            }
3447          break;
3448    
3449          case OP_NOT_HSPACE:
3450          switch(chr)
3451            {
3452            HSPACE_CASES: break;
3453            default: return FALSE;
3454            }
3455          break;
3456    
3457          case OP_ANYNL:
3458          case OP_VSPACE:
3459          switch(chr)
3460            {
3461            VSPACE_CASES: return FALSE;
3462            default: break;
3463            }
3464          break;
3465    
3466          case OP_NOT_VSPACE:
3467          switch(chr)
3468            {
3469            VSPACE_CASES: break;
3470            default: return FALSE;
3471            }
3472          break;
3473    
3474          case OP_DOLL:
3475          case OP_EODN:
3476          switch (chr)
3477            {
3478            case CHAR_CR:
3479            case CHAR_LF:
3480            case CHAR_VT:
3481            case CHAR_FF:
3482            case CHAR_NEL:
3483    #ifndef EBCDIC
3484            case 0x2028:
3485            case 0x2029:
3486    #endif  /* Not EBCDIC */
3487            return FALSE;
3488            }
3489          break;
3490    
3491          case OP_EOD:    /* Can always possessify before \z */
3492          break;
3493    
3494  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3495        othercase = UCD_OTHERCASE(next);        case OP_PROP:
3496  #else        case OP_NOTPROP:
3497        othercase = NOTACHAR;        if (!check_char_prop(chr, list_ptr[2], list_ptr[3],
3498                list_ptr[0] == OP_NOTPROP))
3499            return FALSE;
3500          break;
3501    #endif
3502    
3503          case OP_NCLASS:
3504          if (chr > 255) return FALSE;
3505          /* Fall through */
3506    
3507          case OP_CLASS:
3508          if (chr > 255) break;
3509          class_bitset = (pcre_uint8 *)
3510            ((list_ptr == list ? code : base_end) - list_ptr[2]);
3511          if ((class_bitset[chr >> 3] & (1 << (chr & 7))) != 0) return FALSE;
3512          break;
3513    
3514    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3515          case OP_XCLASS:
3516          if (PRIV(xclass)(chr, (list_ptr == list ? code : base_end) -
3517              list_ptr[2] + LINK_SIZE, utf)) return FALSE;
3518          break;
3519  #endif  #endif
3520        return c != othercase;  
3521          default:
3522          return FALSE;
3523        }        }
3524    
3525        chr_ptr++;
3526        }
3527      while(*chr_ptr != NOTACHAR);
3528    
3529      /* At least one character must be matched from this opcode. */
3530    
3531      if (list[1] == 0) return TRUE;
3532      }
3533    
3534    return FALSE;
3535    }
3536    
3537    
3538    
3539    /*************************************************
3540    *    Scan compiled regex for auto-possession     *
3541    *************************************************/
3542    
3543    /* Replaces single character iterations with their possessive alternatives
3544    if appropriate. This function modifies the compiled opcode!
3545    
3546    Arguments:
3547      code        points to start of the byte code
3548      utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
3549      cd          static compile data
3550    
3551    Returns:      nothing
3552    */
3553    
3554    static void
3555    auto_possessify(pcre_uchar *code, BOOL utf, const compile_data *cd)
3556    {
3557    register pcre_uchar c;
3558    const pcre_uchar *end;
3559    pcre_uchar *repeat_opcode;
3560    pcre_uint32 list[8];
3561    
3562    for (;;)
3563      {
3564      c = *code;
3565    
3566      if (c >= OP_STAR && c <= OP_TYPEPOSUPTO)
3567        {
3568        c -= get_repeat_base(c) - OP_STAR;
3569        end = (c <= OP_MINUPTO) ?
3570          get_chr_property_list(code, utf, cd->fcc, list) : NULL;
3571        list[1] = c == OP_STAR || c == OP_PLUS || c == OP_QUERY || c == OP_UPTO;
3572    
3573        if (end != NULL && compare_opcodes(end, utf, cd, list, end))
3574          {
3575          switch(c)
3576            {
3577            case OP_STAR:
3578            *code += OP_POSSTAR - OP_STAR;
3579            break;
3580    
3581            case OP_MINSTAR:
3582            *code += OP_POSSTAR - OP_MINSTAR;
3583            break;
3584    
3585            case OP_PLUS:
3586            *code += OP_POSPLUS - OP_PLUS;
3587            break;
3588    
3589            case OP_MINPLUS:
3590            *code += OP_POSPLUS - OP_MINPLUS;
3591            break;
3592    
3593            case OP_QUERY:
3594            *code += OP_POSQUERY - OP_QUERY;
3595            break;
3596    
3597            case OP_MINQUERY:
3598            *code += OP_POSQUERY - OP_MINQUERY;
3599            break;
3600    
3601            case OP_UPTO:
3602            *code += OP_POSUPTO - OP_UPTO;
3603            break;
3604    
3605            case OP_MINUPTO:
3606            *code += OP_MINUPTO - OP_UPTO;
3607            break;
3608            }
3609          }
3610        c = *code;
3611        }
3612      else if (c == OP_CLASS || c == OP_NCLASS || c == OP_XCLASS)
3613        {
3614    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3615        if (c == OP_XCLASS)
3616          repeat_opcode = code + GET(code, 1);
3617      else      else
3618  #endif  /* SUPPORT_UTF */  #endif
3619      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */        repeat_opcode = code + 1 + (32 / sizeof(pcre_uchar));
3620    
3621        c = *repeat_opcode;
3622        if (c >= OP_CRSTAR && c <= OP_CRMINRANGE)
3623          {
3624          /* end must not be NULL. */
3625          end = get_chr_property_list(code, utf, cd->fcc, list);
3626    
3627          list[1] = (c & 1) == 0;
3628    
3629          if (compare_opcodes(end, utf, cd, list, end))
3630            {
3631            switch (c)
3632              {
3633              case OP_CRSTAR:
3634              case OP_CRMINSTAR:
3635              *repeat_opcode = OP_CRPOSSTAR;
3636              break;
3637    
3638              case OP_CRPLUS:
3639              case OP_CRMINPLUS:
3640              *repeat_opcode = OP_CRPOSPLUS;
3641              break;
3642    
3643              case OP_CRQUERY:
3644              case OP_CRMINQUERY:
3645              *repeat_opcode = OP_CRPOSQUERY;
3646              break;
3647    
3648              case OP_CRRANGE:
3649              case OP_CRMINRANGE:
3650              *repeat_opcode = OP_CRPOSRANGE;
3651              break;
3652              }
3653            }
3654          }
3655        c = *code;
3656        }
3657    
3658      switch(c)
3659        {
3660        case OP_END:
3661        return;
3662    
3663        case OP_TYPESTAR:
3664        case OP_TYPEMINSTAR:
3665        case OP_TYPEPLUS:
3666        case OP_TYPEMINPLUS:
3667        case OP_TYPEQUERY:
3668        case OP_TYPEMINQUERY:
3669        case OP_TYPEPOSSTAR:
3670        case OP_TYPEPOSPLUS:
3671        case OP_TYPEPOSQUERY:
3672        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
3673        break;
3674    
3675        case OP_TYPEUPTO:
3676        case OP_TYPEMINUPTO:
3677        case OP_TYPEEXACT:
3678        case OP_TYPEPOSUPTO:
3679        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
3680          code += 2;
3681        break;
3682    
3683    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3684        case OP_XCLASS:
3685        code += GET(code, 1);
3686        break;
3687    #endif
3688    
3689        case OP_MARK:
3690        case OP_PRUNE_ARG:
3691        case OP_SKIP_ARG:
3692        case OP_THEN_ARG:
3693        code += code[1];
3694        break;
3695        }
3696    
3697      /* Add in the fixed length from the table */
3698    
3699      code += PRIV(OP_lengths)[c];
3700    
3701      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
3702      a multi-byte character. The length in the table is a minimum, so we have to
3703      arrange to skip the extra bytes. */
3704    
3705    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3706      if (utf) switch(c)
3707        {
3708        case OP_CHAR:
3709        case OP_CHARI:
3710      case OP_NOT:      case OP_NOT:
     return c == next;  
   
3711      case OP_NOTI:      case OP_NOTI:
3712      if (c == next) return TRUE;      case OP_STAR:
3713  #ifdef SUPPORT_UTF      case OP_MINSTAR:
3714      if (utf)      case OP_PLUS:
3715        {      case OP_MINPLUS:
3716        pcre_uint32 othercase;      case OP_QUERY:
3717        if (next < 128) othercase = cd->fcc[next]; else      case OP_MINQUERY:
3718  #ifdef SUPPORT_UCP      case OP_UPTO:
3719        othercase = UCD_OTHERCASE(next);      case OP_MINUPTO:
3720        case OP_EXACT:
3721        case OP_POSSTAR:
3722        case OP_POSPLUS:
3723        case OP_POSQUERY:
3724        case OP_POSUPTO:
3725        case OP_STARI:
3726        case OP_MINSTARI:
3727        case OP_PLUSI:
3728        case OP_MINPLUSI:
3729        case OP_QUERYI:
3730        case OP_MINQUERYI:
3731        case OP_UPTOI:
3732        case OP_MINUPTOI:
3733        case OP_EXACTI:
3734        case OP_POSSTARI:
3735        case OP_POSPLUSI:
3736        case OP_POSQUERYI:
3737        case OP_POSUPTOI:
3738        case OP_NOTSTAR:
3739        case OP_NOTMINSTAR:
3740        case OP_NOTPLUS:
3741        case OP_NOTMINPLUS:
3742        case OP_NOTQUERY:
3743        case OP_NOTMINQUERY:
3744        case OP_NOTUPTO:
3745        case OP_NOTMINUPTO:
3746        case OP_NOTEXACT:
3747        case OP_NOTPOSSTAR:
3748        case OP_NOTPOSPLUS:
3749        case OP_NOTPOSQUERY:
3750        case OP_NOTPOSUPTO:
3751        case OP_NOTSTARI:
3752        case OP_NOTMINSTARI:
3753        case OP_NOTPLUSI:
3754        case OP_NOTMINPLUSI:
3755        case OP_NOTQUERYI:
3756        case OP_NOTMINQUERYI:
3757        case OP_NOTUPTOI:
3758        case OP_NOTMINUPTOI:
3759        case OP_NOTEXACTI:
3760        case OP_NOTPOSSTARI:
3761        case OP_NOTPOSPLUSI:
3762        case OP_NOTPOSQUERYI:
3763        case OP_NOTPOSUPTOI:
3764        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
3765        break;
3766        }
3767  #else  #else
3768        othercase = NOTACHAR;    (void)(utf);  /* Keep compiler happy by referencing function argument */
3769  #endif  #endif
3770        return c == othercase;    }
3771    }
3772    
3773    
3774    
3775    /*************************************************
3776    *           Check for POSIX class syntax         *
3777    *************************************************/
3778    
3779    /* This function is called when the sequence "[:" or "[." or "[=" is
3780    encountered in a character class. It checks whether this is followed by a
3781    sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
3782    reach an unescaped ']' without the special preceding character, return FALSE.
3783    
3784    Originally, this function only recognized a sequence of letters between the
3785    terminators, but it seems that Perl recognizes any sequence of characters,
3786    though of course unknown POSIX names are subsequently rejected. Perl gives an
3787    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
3788    didn't consider this to be a POSIX class. Likewise for [:1234:].
3789    
3790    The problem in trying to be exactly like Perl is in the handling of escapes. We
3791    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
3792    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
3793    below handles the special case of \], but does not try to do any other escape
3794    processing. This makes it different from Perl for cases such as [:l\ower:]
3795    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
3796    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
3797    I think.
3798    
3799    A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.
3800    It seems that the appearance of a nested POSIX class supersedes an apparent
3801    external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or
3802    a digit.
3803    
3804    In Perl, unescaped square brackets may also appear as part of class names. For
3805    example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for
3806    [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not
3807    seem right at all. PCRE does not allow closing square brackets in POSIX class
3808    names.
3809    
3810    Arguments:
3811      ptr      pointer to the initial [
3812      endptr   where to return the end pointer
3813    
3814    Returns:   TRUE or FALSE
3815    */
3816    
3817    static BOOL
3818    check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
3819    {
3820    pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
3821    terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
3822    for (++ptr; *ptr != CHAR_NULL; ptr++)
3823      {
3824      if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
3825        ptr++;
3826      else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
3827      else
3828        {
3829        if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
3830          {
3831          *endptr = ptr;
3832          return TRUE;
3833        }        }
3834      else      if (*ptr == CHAR_LEFT_SQUARE_BRACKET &&
3835  #endif  /* SUPPORT_UTF */           (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
3836      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */            ptr[1] == CHAR_EQUALS_SIGN) &&
3837            check_posix_syntax(ptr, endptr))
3838          return FALSE;
3839        }
3840      }
3841    return FALSE;
3842    }
3843    
     /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.  
     When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */  
3844    
     case OP_DIGIT:  
     return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;  
3845    
     case OP_NOT_DIGIT:  
     return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;  
3846    
3847      case OP_WHITESPACE:  /*************************************************
3848      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;  *          Check POSIX class name                *
3849    *************************************************/
3850    
3851      case OP_NOT_WHITESPACE:  /* This function is called to check the name given in a POSIX-style class entry
3852      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;  such as [:alnum:].
3853    
3854    Arguments:
3855      ptr        points to the first letter
3856      len        the length of the name
3857    
3858    Returns:     a value representing the name, or -1 if unknown
3859    */
3860    
3861    static int
3862    check_posix_name(const pcre_uchar *ptr, int len)
3863    {
3864    const char *pn = posix_names;
3865    register int yield = 0;
3866    while (posix_name_lengths[yield] != 0)
3867      {
3868      if (len == posix_name_lengths[yield] &&
3869        STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
3870      pn += posix_name_lengths[yield] + 1;
3871      yield++;
3872      }
3873    return -1;
3874    }
3875    
3876    
3877    /*************************************************
3878    *    Adjust OP_RECURSE items in repeated group   *
3879    *************************************************/
3880    
3881    /* OP_RECURSE items contain an offset from the start of the regex to the group
3882    that is referenced. This means that groups can be replicated for fixed
3883    repetition simply by copying (because the recursion is allowed to refer to
3884    earlier groups that are outside the current group). However, when a group is
3885    optional (i.e. the minimum quantifier is zero), OP_BRAZERO or OP_SKIPZERO is
3886    inserted before it, after it has been compiled. This means that any OP_RECURSE
3887    items within it that refer to the group itself or any contained groups have to
3888    have their offsets adjusted. That one of the jobs of this function. Before it
3889    is called, the partially compiled regex must be temporarily terminated with
3890    OP_END.
3891    
3892    This function has been extended with the possibility of forward references for
3893    recursions and subroutine calls. It must also check the list of such references
3894    for the group we are dealing with. If it finds that one of the recursions in
3895    the current group is on this list, it adjusts the offset in the list, not the
3896    value in the reference (which is a group number).
3897    
3898    Arguments:
3899      group      points to the start of the group
3900      adjust     the amount by which the group is to be moved
3901      utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
3902      cd         contains pointers to tables etc.
3903      save_hwm   the hwm forward reference pointer at the start of the group
3904    
3905      case OP_WORDCHAR:  Returns:     nothing
3906      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;  */
3907    
3908      case OP_NOT_WORDCHAR:  static void
3909      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
3910      pcre_uchar *save_hwm)
3911    {
3912    pcre_uchar *ptr = group;
3913    
3914      case OP_HSPACE:  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
3915      case OP_NOT_HSPACE:    {
3916      switch(next)    int offset;
3917        {    pcre_uchar *hc;
       HSPACE_CASES:  
       return op_code == OP_NOT_HSPACE;  
3918    
3919        default:    /* See if this recursion is on the forward reference list. If so, adjust the
3920        return op_code != OP_NOT_HSPACE;    reference. */
       }  
3921    
3922      case OP_ANYNL:    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
3923      case OP_VSPACE:      {
3924      case OP_NOT_VSPACE:      offset = (int)GET(hc, 0);
3925      switch(next)      if (cd->start_code + offset == ptr + 1)
3926        {        {
3927        VSPACE_CASES:        PUT(hc, 0, offset + adjust);
3928        return op_code == OP_NOT_VSPACE;        break;
   
       default:  
       return op_code != OP_NOT_VSPACE;  
3929        }        }
   
 #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;  
3930      }      }
   }  
3931    
3932  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP    /* Otherwise, adjust the recursion offset if it's after the start of this
3933  is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are    group. */
 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. */  
3934    
3935  switch(op_code)    if (hc >= cd->hwm)
   {  
   case OP_CHAR:  
   case OP_CHARI:  
   switch(escape)  
3936      {      {
3937      case ESC_d:      offset = (int)GET(ptr, 1);
3938      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
3939        }
     case ESC_D:  
     return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;  
   
     case ESC_s:  
     return c > 255 || (cd->ctypes[c] & ctype_space) == 0;  
3940    
3941      case ESC_S:    ptr += 1 + LINK_SIZE;
3942      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;    }
3943    }
3944    
     case ESC_w:  
     return c > 255 || (cd->ctypes[c] & ctype_word) == 0;  
3945    
     case ESC_W:  
     return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;  
3946    
3947      case ESC_h:  /*************************************************
3948      case ESC_H:  *        Insert an automatic callout point       *
3949      switch(c)  *************************************************/
       {  
       HSPACE_CASES:  
       return escape != ESC_h;  
   
       default:  
       return escape == ESC_h;  
       }  
3950    
3951      case ESC_v:  /* This function is called when the PCRE_AUTO_CALLOUT option is set, to insert
3952      case ESC_V:  callout points before each pattern item.
     switch(c)  
       {  
       VSPACE_CASES:  
       return escape != ESC_v;  
3953    
3954        default:  Arguments:
3955        return escape == ESC_v;    code           current code pointer
3956        }    ptr            current pattern pointer
3957      cd             pointers to tables etc
3958    
3959      /* When PCRE_UCP is set, these values get generated for \d etc. Find  Returns:         new code pointer
3960      their substitutions and process them. The result will always be either  */
     ESC_p or ESC_P. Then fall through to process those values. */  
3961    
3962  #ifdef SUPPORT_UCP  static pcre_uchar *
3963      case ESC_du:  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
3964      case ESC_DU:  {
3965      case ESC_wu:  *code++ = OP_CALLOUT;
3966      case ESC_WU:  *code++ = 255;
3967      case ESC_su:  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
3968      case ESC_SU:  PUT(code, LINK_SIZE, 0);                       /* Default length */
3969        {  return code + 2 * LINK_SIZE;
3970        int temperrorcode = 0;  }
       ptr = substitutes[escape - ESC_DU];  
       escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);  
       if (temperrorcode != 0) return FALSE;  
       ptr++;    /* For compatibility */  
       }  
     /* Fall through */  
3971    
     case ESC_p:  
     case ESC_P:  
       {  
       int ptype, pdata, errorcodeptr;  
       BOOL negated;  
3972    
       ptr--;      /* Make ptr point at the p or P */  
       ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);  
       if (ptype < 0) return FALSE;  
       ptr++;      /* Point past the final curly ket */  
3973    
3974        /* If the property item is optional, we have to give up. (When generated  /*************************************************
3975        from \d etc by PCRE_UCP, this test will have been applied much earlier,  *         Complete a callout item                *
3976        to the original \d etc. At this point, ptr will point to a zero byte. */  *************************************************/
3977    
3978        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  /* A callout item contains the length of the next item in the pattern, which
3979          STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)  we can't fill in till after we have reached the relevant point. This is used
3980            return FALSE;  for both automatic and manual callouts.
3981    
3982        /* Do the property check. */  Arguments:
3983      previous_callout   points to previous callout item
3984      ptr                current pattern pointer
3985      cd                 pointers to tables etc
3986    
3987        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);  Returns:             nothing
3988        }  */
 #endif  
3989    
3990      default:  static void
3991      return FALSE;  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
3992      }  {
3993    int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
3994    PUT(previous_callout, 2 + LINK_SIZE, length);
3995    }
3996    
   /* In principle, support for Unicode properties should be integrated here as  
   well. It means re-organizing the above code so as to get hold of the property  
   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.) */  
3997    
   case OP_DIGIT:  
   return escape == ESC_D || escape == ESC_s || escape == ESC_W ||  
          escape == ESC_h || escape == ESC_v || escape == ESC_R;  
3998    
3999    case OP_NOT_DIGIT:  #ifdef SUPPORT_UCP
4000    return escape == ESC_d;  /*************************************************
4001    *           Get othercase range                  *
4002    *************************************************/
4003    
4004    case OP_WHITESPACE:  /* This function is passed the start and end of a class range, in UTF-8 mode
4005    return escape == ESC_S || escape == ESC_d || escape == ESC_w;  with UCP support. It searches up the characters, looking for ranges of
4006    characters in the "other" case. Each call returns the next one, updating the
4007    start address. A character with multiple other cases is returned on its own
4008    with a special return value.
4009    
4010    case OP_NOT_WHITESPACE:  Arguments:
4011    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;    cptr        points to starting character value; updated
4012      d           end value
4013      ocptr       where to put start of othercase range
4014      odptr       where to put end of othercase range
4015    
4016    case OP_HSPACE:  Yield:        -1 when no more
4017    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||                 0 when a range is returned
4018           escape == ESC_w || escape == ESC_v || escape == ESC_R;                >0 the CASESET offset for char with multiple other cases
4019                    in this case, ocptr contains the original
4020    */
4021    
4022    case OP_NOT_HSPACE:  static int
4023    return escape == ESC_h;  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
4024      pcre_uint32 *odptr)
4025    {
4026    pcre_uint32 c, othercase, next;
4027    unsigned int co;
4028    
4029    /* Can't have \S in here because VT matches \S (Perl anomaly) */  /* Find the first character that has an other case. If it has multiple other
4030    case OP_ANYNL:  cases, return its case offset value. */
   case OP_VSPACE:  
   return escape == ESC_V || escape == ESC_d || escape == ESC_w;  
4031    
4032    case OP_NOT_VSPACE:  for (c = *cptr; c <= d; c++)
4033    return escape == ESC_v || escape == ESC_R;    {
4034      if ((co = UCD_CASESET(c)) != 0)
4035        {
4036        *ocptr = c++;   /* Character that has the set */
4037        *cptr = c;      /* Rest of input range */
4038        return (int)co;
4039        }
4040      if ((othercase = UCD_OTHERCASE(c)) != c) break;
4041      }
4042    
4043    case OP_WORDCHAR:  if (c > d) return -1;  /* Reached end of range */
   return escape == ESC_W || escape == ESC_s || escape == ESC_h ||  
          escape == ESC_v || escape == ESC_R;  
4044    
4045    case OP_NOT_WORDCHAR:  *ocptr = othercase;
4046    return escape == ESC_w || escape == ESC_d;  next = othercase + 1;
4047    
4048    default:  for (++c; c <= d; c++)
4049    return FALSE;    {
4050      if (UCD_OTHERCASE(c) != next) break;
4051      next++;
4052    }    }
4053    
4054  /* Control does not reach here */  *odptr = next - 1;     /* End of othercase range */
4055    *cptr = c;             /* Rest of input range */
4056    return 0;
4057  }  }
4058    #endif  /* SUPPORT_UCP */
4059    
4060    
4061    
# Line 3418  switch(op_code) Line 4064  switch(op_code)
4064  *************************************************/  *************************************************/
4065    
4066  /* This function packages up the logic of adding a character or range of  /* This function packages up the logic of adding a character or range of
4067  characters to a class. The character values in the arguments will be within the  characters to a class. The character values in the arguments will be within the
4068  valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is  valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
4069  mutually recursive with the function immediately below.  mutually recursive with the function immediately below.
4070    
4071  Arguments:  Arguments:
4072    classbits     the bit map for characters < 256    classbits     the bit map for characters < 256
4073    uchardptr     points to the pointer for extra data    uchardptr     points to the pointer for extra data
4074    options       the options word    options       the options word
4075    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
4076    start         start of range character    start         start of range character
4077    end           end of range character    end           end of range character
4078    
4079  Returns:        the number of < 256 characters added  Returns:        the number of < 256 characters added
4080                  the pointer to extra data is updated                  the pointer to extra data is updated
4081  */  */
# Line 3441  add_to_class(pcre_uint8 *classbits, pcre Line 4087  add_to_class(pcre_uint8 *classbits, pcre
4087  pcre_uint32 c;  pcre_uint32 c;
4088  int n8 = 0;  int n8 = 0;
4089    
4090  /* If caseless matching is required, scan the range and process alternate  /* If caseless matching is required, scan the range and process alternate
4091  cases. In Unicode, there are 8-bit characters that have alternate cases that  cases. In Unicode, there are 8-bit characters that have alternate cases that
4092  are greater than 255 and vice-versa. Sometimes we can just extend the original  are greater than 255 and vice-versa. Sometimes we can just extend the original
4093  range. */  range. */
4094    
4095  if ((options & PCRE_CASELESS) != 0)  if ((options & PCRE_CASELESS) != 0)
4096    {    {
4097  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4098    if ((options & PCRE_UTF8) != 0)    if ((options & PCRE_UTF8) != 0)
4099      {      {
4100      int rc;      int rc;
4101      pcre_uint32 oc, od;      pcre_uint32 oc, od;
4102    
4103      options &= ~PCRE_CASELESS;   /* Remove for recursive calls */      options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
4104      c = start;      c = start;
4105    
4106      while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)      while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
4107        {        {
4108        /* Handle a single character that has more than one other case. */        /* Handle a single character that has more than one other case. */
4109    
4110        if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,        if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
4111          PRIV(ucd_caseless_sets) + rc, oc);          PRIV(ucd_caseless_sets) + rc, oc);
4112    
4113        /* Do nothing if the other case range is within the original range. */        /* Do nothing if the other case range is within the original range. */
4114    
4115        else if (oc >= start && od <= end) continue;        else if (oc >= start && od <= end) continue;
4116    
4117        /* Extend the original range if there is overlap, noting that if oc < c, we        /* Extend the original range if there is overlap, noting that if oc < c, we
4118        can't have od > end because a subrange is always shorter than the basic        can't have od > end because a subrange is always shorter than the basic
4119        range. Otherwise, use a recursive call to add the additional range. */        range. Otherwise, use a recursive call to add the additional range. */
4120    
4121        else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */        else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
4122        else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */        else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
4123        else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);        else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
# Line 3481  if ((options & PCRE_CASELESS) != 0) Line 4127  if ((options & PCRE_CASELESS) != 0)
4127  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
4128    
4129    /* Not UTF-mode, or no UCP */    /* Not UTF-mode, or no UCP */
4130    
4131    for (c = start; c <= end && c < 256; c++)    for (c = start; c <= end && c < 256; c++)
4132      {      {
4133      SETBIT(classbits, cd->fcc[c]);      SETBIT(classbits, cd->fcc[c]);
4134      n8++;      n8++;
4135      }      }
4136    }    }
4137    
4138  /* Now handle the original range. Adjust the final value according to the bit  /* Now handle the original range. Adjust the final value according to the bit
4139  length - this means that the same lists of (e.g.) horizontal spaces can be used  length - this means that the same lists of (e.g.) horizontal spaces can be used
4140  in all cases. */  in all cases. */
# Line 3514  if (end < 0x100) Line 4160  if (end < 0x100)
4160    {    {
4161    for (c = start; c <= end; c++)    for (c = start; c <= end; c++)
4162      {      {
4163      n8++;      n8++;
4164      SETBIT(classbits, c);      SETBIT(classbits, c);
4165      }      }
4166    }    }
4167    
4168  else  else
4169    {    {
4170    pcre_uchar *uchardata = *uchardptr;    pcre_uchar *uchardata = *uchardptr;
4171    
4172  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4173    if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */    if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
4174      {      {
4175      if (start < end)      if (start < end)
4176        {        {
4177        *uchardata++ = XCL_RANGE;        *uchardata++ = XCL_RANGE;
4178        uchardata += PRIV(ord2utf)(start, uchardata);        uchardata += PRIV(ord2utf)(start, uchardata);
4179        uchardata += PRIV(ord2utf)(end, uchardata);        uchardata += PRIV(ord2utf)(end, uchardata);
4180        }        }
4181      else if (start == end)      else if (start == end)
4182        {        {
4183        *uchardata++ = XCL_SINGLE;        *uchardata++ = XCL_SINGLE;
4184        uchardata += PRIV(ord2utf)(start, uchardata);        uchardata += PRIV(ord2utf)(start, uchardata);
4185        }        }
4186      }      }
4187    else    else
4188  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
4189    
4190    /* Without UTF support, character values are constrained by the bit length,    /* Without UTF support, character values are constrained by the bit length,
4191    and can only be > 256 for 16-bit and 32-bit libraries. */    and can only be > 256 for 16-bit and 32-bit libraries. */
4192    
4193  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
4194      {}      {}
4195  #else  #else
4196    if (start < end)    if (start < end)
4197      {      {
4198      *uchardata++ = XCL_RANGE;      *uchardata++ = XCL_RANGE;
# Line 3557  else Line 4203  else
4203      {      {
4204      *uchardata++ = XCL_SINGLE;      *uchardata++ = XCL_SINGLE;
4205      *uchardata++ = start;      *uchardata++ = start;
4206      }      }
4207  #endif  #endif
4208    
4209    *uchardptr = uchardata;   /* Updata extra data pointer */    *uchardptr = uchardata;   /* Updata extra data pointer */
4210    }    }
4211    
4212  return n8;    /* Number of 8-bit characters */  return n8;    /* Number of 8-bit characters */
4213  }  }
4214    
4215    
4216    
4217    
4218  /*************************************************  /*************************************************
4219  *        Add a list of characters to a class     *  *        Add a list of characters to a class     *
4220  *************************************************/  *************************************************/
4221    
4222  /* This function is used for adding a list of case-equivalent characters to a  /* This function is used for adding a list of case-equivalent characters to a
4223  class, and also for adding a list of horizontal or vertical whitespace. If the  class, and also for adding a list of horizontal or vertical whitespace. If the
4224  list is in order (which it should be), ranges of characters are detected and  list is in order (which it should be), ranges of characters are detected and
4225  handled appropriately. This function is mutually recursive with the function  handled appropriately. This function is mutually recursive with the function
# Line 3583  Arguments: Line 4229  Arguments:
4229    classbits     the bit map for characters < 256    classbits     the bit map for characters < 256
4230    uchardptr     points to the pointer for extra data    uchardptr     points to the pointer for extra data
4231    options       the options word    options       the options word
4232    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
4233    p             points to row of 32-bit values, terminated by NOTACHAR    p             points to row of 32-bit values, terminated by NOTACHAR
4234    except        character to omit; this is used when adding lists of    except        character to omit; this is used when adding lists of
4235                    case-equivalent characters to avoid including the one we                    case-equivalent characters to avoid including the one we
4236                    already know about                    already know about
4237    
4238  Returns:        the number of < 256 characters added  Returns:        the number of < 256 characters added
4239                  the pointer to extra data is updated                  the pointer to extra data is updated
4240  */  */
# Line 3602  while (p[0] < NOTACHAR) Line 4248  while (p[0] < NOTACHAR)
4248    {    {
4249    int n = 0;    int n = 0;
4250    if (p[0] != except)    if (p[0] != except)
4251      {      {
4252      while(p[n+1] == p[0] + n + 1) n++;      while(p[n+1] == p[0] + n + 1) n++;
4253      n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);      n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
4254      }      }
4255    p += n + 1;    p += n + 1;
4256    }    }
4257  return n8;  return n8;
4258  }  }
4259    
4260    
4261    
# Line 3624  Arguments: Line 4270  Arguments:
4270    classbits     the bit map for characters < 256    classbits     the bit map for characters < 256
4271    uchardptr     points to the pointer for extra data    uchardptr     points to the pointer for extra data
4272    options       the options word    options       the options word
4273    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
4274    p             points to row of 32-bit values, terminated by NOTACHAR    p             points to row of 32-bit values, terminated by NOTACHAR
4275    
4276  Returns:        the number of < 256 characters added  Returns:        the number of < 256 characters added
4277                  the pointer to extra data is updated                  the pointer to extra data is updated
4278  */  */
4279    
4280  static int  static int
4281  add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,  add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
4282    int options, compile_data *cd, const pcre_uint32 *p)    int options, compile_data *cd, const pcre_uint32 *p)
4283  {  {
4284  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
# Line 3644  while (p[0] < NOTACHAR) Line 4290  while (p[0] < NOTACHAR)
4290    while (p[1] == p[0] + 1) p++;    while (p[1] == p[0] + 1) p++;
4291    n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,    n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
4292      (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);      (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
4293    p++;    p++;
4294    }    }
4295  return n8;  return n8;
4296  }  }
4297    
4298    
4299    
# Line 3662  to find out the amount of memory needed, Line 4308  to find out the amount of memory needed,
4308  phase. The value of lengthptr distinguishes the two phases.  phase. The value of lengthptr distinguishes the two phases.
4309    
4310  Arguments:  Arguments:
4311    optionsptr     pointer to the option bits    optionsptr        pointer to the option bits
4312    codeptr        points to the pointer to the current code point    codeptr           points to the pointer to the current code point
4313    ptrptr         points to the current pattern pointer    ptrptr            points to the current pattern pointer
4314    errorcodeptr   points to error code variable    errorcodeptr      points to error code variable
4315    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr      place to put the first required character
4316    reqcharptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
4317    bcptr          points to current branch chain    reqcharptr        place to put the last required character
4318    cond_depth     conditional nesting depth    reqcharflagsptr   place to put the last required character flags, or a negative number
4319    cd             contains pointers to tables etc.    bcptr             points to current branch chain
4320    lengthptr      NULL during the real compile phase    cond_depth        conditional nesting depth
4321                   points to length accumulator during pre-compile phase    cd                contains pointers to tables etc.
4322      lengthptr         NULL during the real compile phase
4323                        points to length accumulator during pre-compile phase
4324    
4325  Returns:         TRUE on success  Returns:            TRUE on success
4326                   FALSE, with *errorcodeptr set non-zero on error                      FALSE, with *errorcodeptr set non-zero on error
4327  */  */
4328    
4329  static BOOL  static BOOL
4330  compile_branch(int *optionsptr, pcre_uchar **codeptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
4331    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
4332    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
4333      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
4334      branch_chain *bcptr, int cond_depth,
4335    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
4336  {  {
4337  int repeat_type, op_type;  int repeat_type, op_type;
4338  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
4339  int bravalue = 0;  int bravalue = 0;
4340  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
4341  pcre_int32 firstchar, reqchar;  pcre_uint32 firstchar, reqchar;
4342  pcre_int32 zeroreqchar, zerofirstchar;  pcre_int32 firstcharflags, reqcharflags;
4343    pcre_uint32 zeroreqchar, zerofirstchar;
4344    pcre_int32 zeroreqcharflags, zerofirstcharflags;
4345  pcre_int32 req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
4346  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
4347  int after_manual_callout = 0;  int after_manual_callout = 0;
# Line 3717  dynamically as we process the pattern. * Line 4369  dynamically as we process the pattern. *
4369  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
4370  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
4371  BOOL utf = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
4372    #ifndef COMPILE_PCRE32
4373  pcre_uchar utf_chars[6];  pcre_uchar utf_chars[6];
4374    #endif
4375  #else  #else
4376  BOOL utf = FALSE;  BOOL utf = FALSE;
4377  #endif  #endif
4378    
4379  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define  /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
4380  class_uchardata always so that it can be passed to add_to_class() always,  class_uchardata always so that it can be passed to add_to_class() always,
4381  though it will not be used in non-UTF 8-bit cases. This avoids having to supply  though it will not be used in non-UTF 8-bit cases. This avoids having to supply
4382  alternative calls for the different cases. */  alternative calls for the different cases. */
4383    
4384  pcre_uchar *class_uchardata;  pcre_uchar *class_uchardata;
# Line 3752  to take the zero repeat into account. Th Line 4406  to take the zero repeat into account. Th
4406  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
4407  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
4408    
4409  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
4410    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
4411    
4412  /* The variable req_caseopt contains either the REQ_CASELESS value  /* The variable req_caseopt contains either the REQ_CASELESS value
4413  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 3778  for (;; ptr++) Line 4433  for (;; ptr++)
4433    int recno;    int recno;
4434    int refsign;    int refsign;
4435    int skipbytes;    int skipbytes;
4436    int subreqchar;    pcre_uint32 subreqchar, subfirstchar;
4437    int subfirstchar;    pcre_int32 subreqcharflags, subfirstcharflags;
4438    int terminator;    int terminator;
4439    int mclength;    unsigned int mclength;
4440    int tempbracount;    unsigned int tempbracount;
4441    int ec; // FIXMEchpe pcre_uint32    pcre_uint32 ec;
4442    pcre_uchar mcbuffer[8];    pcre_uchar mcbuffer[8];
4443    
4444    /* Get next character in the pattern */    /* Get next character in the pattern */
# Line 3793  for (;; ptr++) Line 4448  for (;; ptr++)
4448    /* 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
4449    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
4450    
4451    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
4452      {      {
4453      ptr = nestptr;      ptr = nestptr;
4454      nestptr = NULL;      nestptr = NULL;
# Line 3868  for (;; ptr++) Line 4523  for (;; ptr++)
4523    
4524    /* 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 */
4525    
4526    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
4527      {      {
4528      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
4529        {        {
# Line 3893  for (;; ptr++) Line 4548  for (;; ptr++)
4548        }        }
4549      }      }
4550    
   /* Fill in length of a previous callout, except when the next thing is  
   a quantifier. */  
   
4551    is_quantifier =    is_quantifier =
4552      c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK ||      c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK ||
4553      (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1));      (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1));
4554    
4555    if (!is_quantifier && previous_callout != NULL &&    /* Fill in length of a previous callout, except when the next thing is a
4556      quantifier or when processing a property substitution string in UCP mode. */
4557    
4558      if (!is_quantifier && previous_callout != NULL && nestptr == NULL &&
4559         after_manual_callout-- <= 0)         after_manual_callout-- <= 0)
4560      {      {
4561      if (lengthptr == NULL)      /* Don't attempt in pre-compile phase */      if (lengthptr == NULL)      /* Don't attempt in pre-compile phase */
# Line 3916  for (;; ptr++) Line 4571  for (;; ptr++)
4571      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
4572        {        {
4573        ptr++;        ptr++;
4574        while (*ptr != 0)        while (*ptr != CHAR_NULL)
4575          {          {
4576          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
4577          ptr++;          ptr++;
# Line 3924  for (;; ptr++) Line 4579  for (;; ptr++)
4579          if (utf) FORWARDCHAR(ptr);          if (utf) FORWARDCHAR(ptr);
4580  #endif  #endif
4581          }          }
4582        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
4583    
4584        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
4585        c = 0;        c = 0;
4586        }        }
4587      }      }
4588    
4589    /* No auto callout for quantifiers. */    /* No auto callout for quantifiers, or while processing property strings that
4590      are substituted for \w etc in UCP mode. */
4591    
4592    if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier)    if ((options & PCRE_AUTO_CALLOUT) != 0 && !is_quantifier && nestptr == NULL)
4593      {      {
4594      previous_callout = code;      previous_callout = code;
4595      code = auto_callout(code, ptr, cd);      code = auto_callout(code, ptr, cd);
# Line 3946  for (;; ptr++) Line 4602  for (;; ptr++)
4602      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
4603      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
4604      *firstcharptr = firstchar;      *firstcharptr = firstchar;
4605        *firstcharflagsptr = firstcharflags;
4606      *reqcharptr = reqchar;      *reqcharptr = reqchar;
4607        *reqcharflagsptr = reqcharflags;
4608      *codeptr = code;      *codeptr = code;
4609      *ptrptr = ptr;      *ptrptr = ptr;
4610      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3970  for (;; ptr++) Line 4628  for (;; ptr++)
4628      previous = NULL;      previous = NULL;
4629      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
4630        {        {
4631        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4632        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
4633        }        }
4634      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3985  for (;; ptr++) Line 4643  for (;; ptr++)
4643      repeats. The value of reqchar doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4644    
4645      case CHAR_DOT:      case CHAR_DOT:
4646      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4647      zerofirstchar = firstchar;      zerofirstchar = firstchar;
4648        zerofirstcharflags = firstcharflags;
4649      zeroreqchar = reqchar;      zeroreqchar = reqchar;
4650        zeroreqcharflags = reqcharflags;
4651      previous = code;      previous = code;
4652      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4653      break;      break;
# Line 4061  for (;; ptr++) Line 4721  for (;; ptr++)
4721          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4722        {        {
4723        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4724        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4725        zerofirstchar = firstchar;        zerofirstchar = firstchar;
4726          zerofirstcharflags = firstcharflags;
4727        break;        break;
4728        }        }
4729    
# Line 4097  for (;; ptr++) Line 4758  for (;; ptr++)
4758      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
4759      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4760    
4761      if (c != 0) do      if (c != CHAR_NULL) do
4762        {        {
4763        const pcre_uchar *oldptr;        const pcre_uchar *oldptr;
4764    
4765  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #ifdef SUPPORT_UTF
4766        if (utf && HAS_EXTRALEN(c))        if (utf && HAS_EXTRALEN(c))
4767          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4768          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
# Line 4112  for (;; ptr++) Line 4773  for (;; ptr++)
4773        /* In the pre-compile phase, accumulate the length of any extra        /* In the pre-compile phase, accumulate the length of any extra
4774        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
4775        contain a zillion > 255 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4776        (which is on the stack). We have to remember that there was XCLASS data,        (which is on the stack). We have to remember that there was XCLASS data,
4777        however. */        however. */
4778    
4779        if (lengthptr != NULL && class_uchardata > class_uchardata_base)        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
# Line 4176  for (;; ptr++) Line 4837  for (;; ptr++)
4837          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4838          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4839    
4840          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4841            posix_class = 0;            posix_class = 0;
4842    
4843          /* When PCRE_UCP is set, some of the POSIX classes are converted to          /* When PCRE_UCP is set, some of the POSIX classes are converted to
4844          different escape sequences that use Unicode properties. */          different escape sequences that use Unicode properties \p or \P. Others
4845            that are not available via \p or \P generate XCL_PROP/XCL_NOTPROP
4846            directly. */
4847    
4848  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4849          if ((options & PCRE_UCP) != 0)          if ((options & PCRE_UCP) != 0)
4850            {            {
4851              unsigned int ptype = 0;
4852            int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);            int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);
4853    
4854              /* The posix_substitutes table specifies which POSIX classes can be
4855              converted to \p or \P items. */
4856    
4857            if (posix_substitutes[pc] != NULL)            if (posix_substitutes[pc] != NULL)
4858              {              {
4859              nestptr = tempptr + 1;              nestptr = tempptr + 1;
4860              ptr = posix_substitutes[pc] - 1;              ptr = posix_substitutes[pc] - 1;
4861              continue;              continue;
4862              }              }
4863    
4864              /* There are three other classes that generate special property calls
4865              that are recognized only in an XCLASS. */
4866    
4867              else switch(posix_class)
4868                {
4869                case PC_GRAPH:
4870                ptype = PT_PXGRAPH;
4871                /* Fall through */
4872                case PC_PRINT:
4873                if (ptype == 0) ptype = PT_PXPRINT;
4874                /* Fall through */
4875                case PC_PUNCT:
4876                if (ptype == 0) ptype = PT_PXPUNCT;
4877                *class_uchardata++ = local_negate? XCL_NOTPROP : XCL_PROP;
4878                *class_uchardata++ = ptype;
4879                *class_uchardata++ = 0;
4880                ptr = tempptr + 1;
4881                continue;
4882    
4883                /* For all other POSIX classes, no special action is taken in UCP
4884                mode. Fall through to the non_UCP case. */
4885    
4886                default:
4887                break;
4888                }
4889            }            }
4890  #endif  #endif
4891          /* In the non-UCP case, we build the bit map for the POSIX class in a          /* In the non-UCP case, or when UCP makes no difference, we build the
4892          chunk of local store because we may be adding and subtracting from it,          bit map for the POSIX class in a chunk of local store because we may be
4893          and we don't want to subtract bits that may be in the main map already.          adding and subtracting from it, and we don't want to subtract bits that
4894          At the end we or the result into the bit map that is being built. */          may be in the main map already. At the end we or the result into the
4895            bit map that is being built. */
4896    
4897          posix_class *= 3;          posix_class *= 3;
4898    
# Line 4253  for (;; ptr++) Line 4948  for (;; ptr++)
4948