/[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 149 by ph10, Mon Apr 16 15:28:08 2007 UTC revision 504 by ph10, Mon Mar 8 08:57:04 2010 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) 1997-2007 University of Cambridge             Copyright (c) 1997-2010 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 42  POSSIBILITY OF SUCH DAMAGE. Line 42  POSSIBILITY OF SUCH DAMAGE.
42  supporting internal functions that are not used by other modules. */  supporting internal functions that are not used by other modules. */
43    
44    
45    #ifdef HAVE_CONFIG_H
46    #include "config.h"
47    #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
50  #define PSSTART start_pattern  /* Field containing processed string start */  #define PSSTART start_pattern  /* Field containing processed string start */
51  #define PSEND   end_pattern    /* Field containing processed string end */  #define PSEND   end_pattern    /* Field containing processed string end */
52    
   
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When DEBUG is defined, we need the pcre_printint() function, which is also  /* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is
57  used by pcretest. DEBUG is not defined when building a production library. */  also used by pcretest. PCRE_DEBUG is not defined when building a production
58    library. */
59    
60  #ifdef DEBUG  #ifdef PCRE_DEBUG
61  #include "pcre_printint.src"  #include "pcre_printint.src"
62  #endif  #endif
63    
64    
65    /* Macro for setting individual bits in class bitmaps. */
66    
67    #define SETBIT(a,b) a[b/8] |= (1 << (b%8))
68    
69    /* Maximum length value to check against when making sure that the integer that
70    holds the compiled pattern length does not overflow. We make it a bit less than
71    INT_MAX to allow for adding in group terminating bytes, so that we don't have
72    to check them every time. */
73    
74    #define OFLOW_MAX (INT_MAX - 20)
75    
76    
77  /*************************************************  /*************************************************
78  *      Code parameters and static tables         *  *      Code parameters and static tables         *
79  *************************************************/  *************************************************/
# Line 82  are simple data values; negative values Line 98  are simple data values; negative values
98  on. Zero means further processing is needed (for things like \x), or the escape  on. Zero means further processing is needed (for things like \x), or the escape
99  is invalid. */  is invalid. */
100    
101  #ifndef EBCDIC  /* This is the "normal" table for ASCII systems */  #ifndef EBCDIC
102    
103    /* This is the "normal" table for ASCII systems or for EBCDIC systems running
104    in UTF-8 mode. */
105    
106  static const short int escapes[] = {  static const short int escapes[] = {
107       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */       0,                       0,
108       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */       0,                       0,
109     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */       0,                       0,
110       0,      0,      0,      0,      0,      0,      0,      0,   /* H - O */       0,                       0,
111  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0,      0, -ESC_W,   /* P - W */       0,                       0,
112  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */       CHAR_COLON,              CHAR_SEMICOLON,
113     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */       CHAR_LESS_THAN_SIGN,     CHAR_EQUALS_SIGN,
114       0,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */       CHAR_GREATER_THAN_SIGN,  CHAR_QUESTION_MARK,
115  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0,      0, -ESC_w,   /* p - w */       CHAR_COMMERCIAL_AT,      -ESC_A,
116       0,      0, -ESC_z                                            /* x - z */       -ESC_B,                  -ESC_C,
117         -ESC_D,                  -ESC_E,
118         0,                       -ESC_G,
119         -ESC_H,                  0,
120         0,                       -ESC_K,
121         0,                       0,
122         0,                       0,
123         -ESC_P,                  -ESC_Q,
124         -ESC_R,                  -ESC_S,
125         0,                       0,
126         -ESC_V,                  -ESC_W,
127         -ESC_X,                  0,
128         -ESC_Z,                  CHAR_LEFT_SQUARE_BRACKET,
129         CHAR_BACKSLASH,          CHAR_RIGHT_SQUARE_BRACKET,
130         CHAR_CIRCUMFLEX_ACCENT,  CHAR_UNDERSCORE,
131         CHAR_GRAVE_ACCENT,       7,
132         -ESC_b,                  0,
133         -ESC_d,                  ESC_e,
134         ESC_f,                   0,
135         -ESC_h,                  0,
136         0,                       -ESC_k,
137         0,                       0,
138         ESC_n,                   0,
139         -ESC_p,                  0,
140         ESC_r,                   -ESC_s,
141         ESC_tee,                 0,
142         -ESC_v,                  -ESC_w,
143         0,                       0,
144         -ESC_z
145  };  };
146    
147  #else           /* This is the "abnormal" table for EBCDIC systems */  #else
148    
149    /* This is the "abnormal" table for EBCDIC systems without UTF-8 support. */
150    
151  static const short int escapes[] = {  static const short int escapes[] = {
152  /*  48 */     0,     0,      0,     '.',    '<',   '(',    '+',    '|',  /*  48 */     0,     0,      0,     '.',    '<',   '(',    '+',    '|',
153  /*  50 */   '&',     0,      0,       0,      0,     0,      0,      0,  /*  50 */   '&',     0,      0,       0,      0,     0,      0,      0,
# Line 106  static const short int escapes[] = { Line 157  static const short int escapes[] = {
157  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,
158  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',
159  /*  80 */     0,     7, -ESC_b,       0, -ESC_d, ESC_e,  ESC_f,      0,  /*  80 */     0,     7, -ESC_b,       0, -ESC_d, ESC_e,  ESC_f,      0,
160  /*  88 */     0,     0,      0,     '{',      0,     0,      0,      0,  /*  88 */-ESC_h,     0,      0,     '{',      0,     0,      0,      0,
161  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,
162  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,
163  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,     0, -ESC_w,      0,  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,-ESC_v, -ESC_w,      0,
164  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,
165  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,
166  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',
167  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,
168  /*  C8 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,
169  /*  D0 */   '}',     0,      0,       0,      0,     0,      0, -ESC_P,  /*  D0 */   '}',     0, -ESC_K,       0,      0,     0,      0, -ESC_P,
170  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,
171  /*  E0 */  '\\',     0, -ESC_S,       0,      0,     0, -ESC_W, -ESC_X,  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,
172  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,
173  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,
174  /*  F8 */     0,     0,      0,       0,      0,     0,      0,      0  /*  F8 */     0,     0,      0,       0,      0,     0,      0,      0
# Line 125  static const short int escapes[] = { Line 176  static const short int escapes[] = {
176  #endif  #endif
177    
178    
179  /* Tables of names of POSIX character classes and their lengths. The list is  /* Table of special "verbs" like (*PRUNE). This is a short table, so it is
180  terminated by a zero length entry. The first three must be alpha, lower, upper,  searched linearly. Put all the names into a single string, in order to reduce
181  as this is assumed for handling case independence. */  the number of relocations when a shared library is dynamically linked. The
182    string is built from string macros so that it works in UTF-8 mode on EBCDIC
183  static const char *const posix_names[] = {  platforms. */
184    "alpha", "lower", "upper",  
185    "alnum", "ascii", "blank", "cntrl", "digit", "graph",  typedef struct verbitem {
186    "print", "punct", "space", "word",  "xdigit" };    int   len;
187      int   op;
188    } verbitem;
189    
190    static const char verbnames[] =
191      STRING_ACCEPT0
192      STRING_COMMIT0
193      STRING_F0
194      STRING_FAIL0
195      STRING_PRUNE0
196      STRING_SKIP0
197      STRING_THEN;
198    
199    static const verbitem verbs[] = {
200      { 6, OP_ACCEPT },
201      { 6, OP_COMMIT },
202      { 1, OP_FAIL },
203      { 4, OP_FAIL },
204      { 5, OP_PRUNE },
205      { 4, OP_SKIP  },
206      { 4, OP_THEN  }
207    };
208    
209    static const int verbcount = sizeof(verbs)/sizeof(verbitem);
210    
211    
212    /* Tables of names of POSIX character classes and their lengths. The names are
213    now all in a single string, to reduce the number of relocations when a shared
214    library is dynamically loaded. The list of lengths is terminated by a zero
215    length entry. The first three must be alpha, lower, upper, as this is assumed
216    for handling case independence. */
217    
218    static const char posix_names[] =
219      STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0
220      STRING_ascii0 STRING_blank0 STRING_cntrl0 STRING_digit0
221      STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
222      STRING_word0  STRING_xdigit;
223    
224  static const uschar posix_name_lengths[] = {  static const uschar posix_name_lengths[] = {
225    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 };
# Line 171  static const int posix_class_maps[] = { Line 258  static const int posix_class_maps[] = {
258  /* The texts of compile-time error messages. These are "char *" because they  /* The texts of compile-time error messages. These are "char *" because they
259  are passed to the outside world. Do not ever re-use any error number, because  are passed to the outside world. Do not ever re-use any error number, because
260  they are documented. Always add a new error instead. Messages marked DEAD below  they are documented. Always add a new error instead. Messages marked DEAD below
261  are no longer used. */  are no longer used. This used to be a table of strings, but in order to reduce
262    the number of relocations needed when a shared library is loaded dynamically,
263  static const char *error_texts[] = {  it is now one long string. We cannot use a table of offsets, because the
264    "no error",  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we
265    "\\ at end of pattern",  simply count through to the one we want - this isn't a performance issue
266    "\\c at end of pattern",  because these strings are used only when there is a compilation error.
267    "unrecognized character follows \\",  
268    "numbers out of order in {} quantifier",  Each substring ends with \0 to insert a null character. This includes the final
269    substring, so that the whole string ends with \0\0, which can be detected when
270    counting through. */
271    
272    static const char error_texts[] =
273      "no error\0"
274      "\\ at end of pattern\0"
275      "\\c at end of pattern\0"
276      "unrecognized character follows \\\0"
277      "numbers out of order in {} quantifier\0"
278    /* 5 */    /* 5 */
279    "number too big in {} quantifier",    "number too big in {} quantifier\0"
280    "missing terminating ] for character class",    "missing terminating ] for character class\0"
281    "invalid escape sequence in character class",    "invalid escape sequence in character class\0"
282    "range out of order in character class",    "range out of order in character class\0"
283    "nothing to repeat",    "nothing to repeat\0"
284    /* 10 */    /* 10 */
285    "operand of unlimited repeat could match the empty string",  /** DEAD **/    "operand of unlimited repeat could match the empty string\0"  /** DEAD **/
286    "internal error: unexpected repeat",    "internal error: unexpected repeat\0"
287    "unrecognized character after (?",    "unrecognized character after (? or (?-\0"
288    "POSIX named classes are supported only within a class",    "POSIX named classes are supported only within a class\0"
289    "missing )",    "missing )\0"
290    /* 15 */    /* 15 */
291    "reference to non-existent subpattern",    "reference to non-existent subpattern\0"
292    "erroffset passed as NULL",    "erroffset passed as NULL\0"
293    "unknown option bit(s) set",    "unknown option bit(s) set\0"
294    "missing ) after comment",    "missing ) after comment\0"
295    "parentheses nested too deeply",  /** DEAD **/    "parentheses nested too deeply\0"  /** DEAD **/
296    /* 20 */    /* 20 */
297    "regular expression too large",    "regular expression is too large\0"
298    "failed to get memory",    "failed to get memory\0"
299    "unmatched parentheses",    "unmatched parentheses\0"
300    "internal error: code overflow",    "internal error: code overflow\0"
301    "unrecognized character after (?<",    "unrecognized character after (?<\0"
302    /* 25 */    /* 25 */
303    "lookbehind assertion is not fixed length",    "lookbehind assertion is not fixed length\0"
304    "malformed number or name after (?(",    "malformed number or name after (?(\0"
305    "conditional group contains more than two branches",    "conditional group contains more than two branches\0"
306    "assertion expected after (?(",    "assertion expected after (?(\0"
307    "(?R or (?digits must be followed by )",    "(?R or (?[+-]digits must be followed by )\0"
308    /* 30 */    /* 30 */
309    "unknown POSIX class name",    "unknown POSIX class name\0"
310    "POSIX collating elements are not supported",    "POSIX collating elements are not supported\0"
311    "this version of PCRE is not compiled with PCRE_UTF8 support",    "this version of PCRE is not compiled with PCRE_UTF8 support\0"
312    "spare error",  /** DEAD **/    "spare error\0"  /** DEAD **/
313    "character value in \\x{...} sequence is too large",    "character value in \\x{...} sequence is too large\0"
314    /* 35 */    /* 35 */
315    "invalid condition (?(0)",    "invalid condition (?(0)\0"
316    "\\C not allowed in lookbehind assertion",    "\\C not allowed in lookbehind assertion\0"
317    "PCRE does not support \\L, \\l, \\N, \\U, or \\u",    "PCRE does not support \\L, \\l, \\N, \\U, or \\u\0"
318    "number after (?C is > 255",    "number after (?C is > 255\0"
319    "closing ) for (?C expected",    "closing ) for (?C expected\0"
320    /* 40 */    /* 40 */
321    "recursive call could loop indefinitely",    "recursive call could loop indefinitely\0"
322    "unrecognized character after (?P",    "unrecognized character after (?P\0"
323    "syntax error in subpattern name (missing terminator)",    "syntax error in subpattern name (missing terminator)\0"
324    "two named subpatterns have the same name",    "two named subpatterns have the same name\0"
325    "invalid UTF-8 string",    "invalid UTF-8 string\0"
326    /* 45 */    /* 45 */
327    "support for \\P, \\p, and \\X has not been compiled",    "support for \\P, \\p, and \\X has not been compiled\0"
328    "malformed \\P or \\p sequence",    "malformed \\P or \\p sequence\0"
329    "unknown property name after \\P or \\p",    "unknown property name after \\P or \\p\0"
330    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)\0"
331    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
332    /* 50 */    /* 50 */
333    "repeated subpattern is too long",    "repeated subpattern is too long\0"    /** DEAD **/
334    "octal value is greater than \\377 (not in UTF-8 mode)",    "octal value is greater than \\377 (not in UTF-8 mode)\0"
335    "internal error: overran compiling workspace",    "internal error: overran compiling workspace\0"
336    "internal error: previously-checked referenced subpattern not found",    "internal error: previously-checked referenced subpattern not found\0"
337    "DEFINE group contains more than one branch",    "DEFINE group contains more than one branch\0"
338    /* 55 */    /* 55 */
339    "repeating a DEFINE group is not allowed",    "repeating a DEFINE group is not allowed\0"
340    "inconsistent NEWLINE options",    "inconsistent NEWLINE options\0"
341    "\\g is not followed by an (optionally braced) non-zero number"    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"
342  };    "a numbered reference must not be zero\0"
343      "(*VERB) with an argument is not supported\0"
344      /* 60 */
345      "(*VERB) not recognized\0"
346      "number is too big\0"
347      "subpattern name expected\0"
348      "digit expected after (?+\0"
349      "] is an invalid data character in JavaScript compatibility mode\0"
350      /* 65 */
351      "different names for subpatterns of the same number are not allowed\0";
352    
353  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
354  patterns. Note that the tables in chartables are dependent on the locale, and  patterns. Note that the tables in chartables are dependent on the locale, and
# Line 262  For convenience, we use the same bit def Line 366  For convenience, we use the same bit def
366    
367  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
368    
369  #ifndef EBCDIC  /* This is the "normal" case, for ASCII systems */  #ifndef EBCDIC
370    
371    /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in
372    UTF-8 mode. */
373    
374  static const unsigned char digitab[] =  static const unsigned char digitab[] =
375    {    {
376    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
# Line 298  static const unsigned char digitab[] = Line 406  static const unsigned char digitab[] =
406    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */
407    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */
408    
409  #else           /* This is the "abnormal" case, for EBCDIC systems */  #else
410    
411    /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */
412    
413  static const unsigned char digitab[] =  static const unsigned char digitab[] =
414    {    {
415    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
# Line 373  static const unsigned char ebcdic_charta Line 484  static const unsigned char ebcdic_charta
484  /* Definition to allow mutual recursion */  /* Definition to allow mutual recursion */
485    
486  static BOOL  static BOOL
487    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, int, int *,    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,
488      int *, branch_chain *, compile_data *, int *);      int *, int *, branch_chain *, compile_data *, int *);
489    
490    
491    
492  /*************************************************  /*************************************************
493    *            Find an error text                  *
494    *************************************************/
495    
496    /* The error texts are now all in one long string, to save on relocations. As
497    some of the text is of unknown length, we can't use a table of offsets.
498    Instead, just count through the strings. This is not a performance issue
499    because it happens only when there has been a compilation error.
500    
501    Argument:   the error number
502    Returns:    pointer to the error string
503    */
504    
505    static const char *
506    find_error_text(int n)
507    {
508    const char *s = error_texts;
509    for (; n > 0; n--)
510      {
511      while (*s++ != 0) {};
512      if (*s == 0) return "Error text not found (please report)";
513      }
514    return s;
515    }
516    
517    
518    /*************************************************
519  *            Handle escapes                      *  *            Handle escapes                      *
520  *************************************************/  *************************************************/
521    
# Line 399  Arguments: Line 536  Arguments:
536    
537  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
538                   negative => a special escape sequence                   negative => a special escape sequence
539                   on error, errorptr is set                   on error, errorcodeptr is set
540  */  */
541    
542  static int  static int
# Line 417  ptr--;                            /* Set Line 554  ptr--;                            /* Set
554    
555  if (c == 0) *errorcodeptr = ERR1;  if (c == 0) *errorcodeptr = ERR1;
556    
557  /* Non-alphamerics are literals. For digits or letters, do an initial lookup in  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
558  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.
559  Otherwise further processing may be required. */  Otherwise further processing may be required. */
560    
561  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
562  else if (c < '0' || c > 'z') {}                           /* Not alphameric */  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */
563  else if ((i = escapes[c - '0']) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) c = i;
564    
565  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
566  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphameric */  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */
567  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
568  #endif  #endif
569    
# Line 442  else Line 579  else
579      /* A number of Perl escapes are not handled by PCRE. We give an explicit      /* A number of Perl escapes are not handled by PCRE. We give an explicit
580      error. */      error. */
581    
582      case 'l':      case CHAR_l:
583      case 'L':      case CHAR_L:
584      case 'N':      case CHAR_N:
585      case 'u':      case CHAR_u:
586      case 'U':      case CHAR_U:
587      *errorcodeptr = ERR37;      *errorcodeptr = ERR37;
588      break;      break;
589    
590      /* \g must be followed by a number, either plain or braced. If positive, it      /* \g must be followed by one of a number of specific things:
591      is an absolute backreference. If negative, it is a relative backreference.  
592      This is a Perl 5.10 feature. */      (1) A number, either plain or braced. If positive, it is an absolute
593        backreference. If negative, it is a relative backreference. This is a Perl
594        5.10 feature.
595    
596        (2) Perl 5.10 also supports \g{name} as a reference to a named group. This
597        is part of Perl's movement towards a unified syntax for back references. As
598        this is synonymous with \k{name}, we fudge it up by pretending it really
599        was \k.
600    
601        (3) For Oniguruma compatibility we also support \g followed by a name or a
602        number either in angle brackets or in single quotes. However, these are
603        (possibly recursive) subroutine calls, _not_ backreferences. Just return
604        the -ESC_g code (cf \k). */
605    
606      case 'g':      case CHAR_g:
607      if (ptr[1] == '{')      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
608        {        {
609          c = -ESC_g;
610          break;
611          }
612    
613        /* Handle the Perl-compatible cases */
614    
615        if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
616          {
617          const uschar *p;
618          for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
619            if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;
620          if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
621            {
622            c = -ESC_k;
623            break;
624            }
625        braced = TRUE;        braced = TRUE;
626        ptr++;        ptr++;
627        }        }
628      else braced = FALSE;      else braced = FALSE;
629    
630      if (ptr[1] == '-')      if (ptr[1] == CHAR_MINUS)
631        {        {
632        negated = TRUE;        negated = TRUE;
633        ptr++;        ptr++;
# Line 471  else Line 636  else
636    
637      c = 0;      c = 0;
638      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
639        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - CHAR_0;
640    
641      if (c == 0 || (braced && *(++ptr) != '}'))      if (c < 0)   /* Integer overflow */
642          {
643          *errorcodeptr = ERR61;
644          break;
645          }
646    
647        if (braced && *(++ptr) != CHAR_RIGHT_CURLY_BRACKET)
648        {        {
649        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
650        return 0;        break;
651          }
652    
653        if (c == 0)
654          {
655          *errorcodeptr = ERR58;
656          break;
657        }        }
658    
659      if (negated)      if (negated)
# Line 484  else Line 661  else
661        if (c > bracount)        if (c > bracount)
662          {          {
663          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
664          return 0;          break;
665          }          }
666        c = bracount - (c - 1);        c = bracount - (c - 1);
667        }        }
# Line 504  else Line 681  else
681      value is greater than 377, the least significant 8 bits are taken. Inside a      value is greater than 377, the least significant 8 bits are taken. Inside a
682      character class, \ followed by a digit is always an octal number. */      character class, \ followed by a digit is always an octal number. */
683    
684      case '1': case '2': case '3': case '4': case '5':      case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5:
685      case '6': case '7': case '8': case '9':      case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
686    
687      if (!isclass)      if (!isclass)
688        {        {
689        oldptr = ptr;        oldptr = ptr;
690        c -= '0';        c -= CHAR_0;
691        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
692          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - CHAR_0;
693          if (c < 0)    /* Integer overflow */
694            {
695            *errorcodeptr = ERR61;
696            break;
697            }
698        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
699          {          {
700          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 525  else Line 707  else
707      generates a binary zero byte and treats the digit as a following literal.      generates a binary zero byte and treats the digit as a following literal.
708      Thus we have to pull back the pointer by one. */      Thus we have to pull back the pointer by one. */
709    
710      if ((c = *ptr) >= '8')      if ((c = *ptr) >= CHAR_8)
711        {        {
712        ptr--;        ptr--;
713        c = 0;        c = 0;
# Line 538  else Line 720  else
720      to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more      to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more
721      than 3 octal digits. */      than 3 octal digits. */
722    
723      case '0':      case CHAR_0:
724      c -= '0';      c -= CHAR_0;
725      while(i++ < 2 && ptr[1] >= '0' && ptr[1] <= '7')      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
726          c = c * 8 + *(++ptr) - '0';          c = c * 8 + *(++ptr) - CHAR_0;
727      if (!utf8 && c > 255) *errorcodeptr = ERR51;      if (!utf8 && c > 255) *errorcodeptr = ERR51;
728      break;      break;
729    
# Line 549  else Line 731  else
731      than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is      than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is
732      treated as a data character. */      treated as a data character. */
733    
734      case 'x':      case CHAR_x:
735      if (ptr[1] == '{')      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
736        {        {
737        const uschar *pt = ptr + 2;        const uschar *pt = ptr + 2;
738        int count = 0;        int count = 0;
# Line 559  else Line 741  else
741        while ((digitab[*pt] & ctype_xdigit) != 0)        while ((digitab[*pt] & ctype_xdigit) != 0)
742          {          {
743          register int cc = *pt++;          register int cc = *pt++;
744          if (c == 0 && cc == '0') continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
745          count++;          count++;
746    
747  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
748          if (cc >= 'a') cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
749          c = (c << 4) + cc - ((cc < 'A')? '0' : ('A' - 10));          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
750  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
751          if (cc >= 'a' && cc <= 'z') cc += 64;  /* Convert to upper case */          if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
752          c = (c << 4) + cc - ((cc >= '0')? '0' : ('A' - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
753  #endif  #endif
754          }          }
755    
756        if (*pt == '}')        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
757          {          {
758          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;
759          ptr = pt;          ptr = pt;
# Line 587  else Line 769  else
769      c = 0;      c = 0;
770      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)
771        {        {
772        int cc;                               /* Some compilers don't like ++ */        int cc;                                  /* Some compilers don't like */
773        cc = *(++ptr);                        /* in initializers */        cc = *(++ptr);                           /* ++ in initializers */
774  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
775        if (cc >= 'a') cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
776        c = c * 16 + cc - ((cc < 'A')? '0' : ('A' - 10));        c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
777  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
778        if (cc <= 'z') cc += 64;              /* Convert to upper case */        if (cc <= CHAR_z) cc += 64;              /* Convert to upper case */
779        c = c * 16 + cc - ((cc >= '0')? '0' : ('A' - 10));        c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
780  #endif  #endif
781        }        }
782      break;      break;
# Line 603  else Line 785  else
785      This coding is ASCII-specific, but then the whole concept of \cx is      This coding is ASCII-specific, but then the whole concept of \cx is
786      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */
787    
788      case 'c':      case CHAR_c:
789      c = *(++ptr);      c = *(++ptr);
790      if (c == 0)      if (c == 0)
791        {        {
792        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
793        return 0;        break;
794        }        }
795    
796  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
797      if (c >= 'a' && c <= 'z') c -= 32;      if (c >= CHAR_a && c <= CHAR_z) c -= 32;
798      c ^= 0x40;      c ^= 0x40;
799  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
800      if (c >= 'a' && c <= 'z') c += 64;      if (c >= CHAR_a && c <= CHAR_z) c += 64;
801      c ^= 0xC0;      c ^= 0xC0;
802  #endif  #endif
803      break;      break;
804    
805      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any
806      other alphameric following \ is an error if PCRE_EXTRA was set; otherwise,      other alphanumeric following \ is an error if PCRE_EXTRA was set;
807      for Perl compatibility, it is a literal. This code looks a bit odd, but      otherwise, for Perl compatibility, it is a literal. This code looks a bit
808      there used to be some cases other than the default, and there may be again      odd, but there used to be some cases other than the default, and there may
809      in future, so I haven't "optimized" it. */      be again in future, so I haven't "optimized" it. */
810    
811      default:      default:
812      if ((options & PCRE_EXTRA) != 0) switch(c)      if ((options & PCRE_EXTRA) != 0) switch(c)
# Line 677  if (c == 0) goto ERROR_RETURN; Line 859  if (c == 0) goto ERROR_RETURN;
859  /* \P or \p can be followed by a name in {}, optionally preceded by ^ for  /* \P or \p can be followed by a name in {}, optionally preceded by ^ for
860  negation. */  negation. */
861    
862  if (c == '{')  if (c == CHAR_LEFT_CURLY_BRACKET)
863    {    {
864    if (ptr[1] == '^')    if (ptr[1] == CHAR_CIRCUMFLEX_ACCENT)
865      {      {
866      *negptr = TRUE;      *negptr = TRUE;
867      ptr++;      ptr++;
868      }      }
869    for (i = 0; i < sizeof(name) - 1; i++)    for (i = 0; i < (int)sizeof(name) - 1; i++)
870      {      {
871      c = *(++ptr);      c = *(++ptr);
872      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
873      if (c == '}') break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
874      name[i] = c;      name[i] = c;
875      }      }
876    if (c !='}') goto ERROR_RETURN;    if (c != CHAR_RIGHT_CURLY_BRACKET) goto ERROR_RETURN;
877    name[i] = 0;    name[i] = 0;
878    }    }
879    
# Line 713  top = _pcre_utt_size; Line 895  top = _pcre_utt_size;
895  while (bot < top)  while (bot < top)
896    {    {
897    i = (bot + top) >> 1;    i = (bot + top) >> 1;
898    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
899    if (c == 0)    if (c == 0)
900      {      {
901      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 756  is_counted_repeat(const uschar *p) Line 938  is_counted_repeat(const uschar *p)
938  {  {
939  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;
940  while ((digitab[*p] & ctype_digit) != 0) p++;  while ((digitab[*p] & ctype_digit) != 0) p++;
941  if (*p == '}') return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
942    
943  if (*p++ != ',') return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
944  if (*p == '}') return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
945    
946  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;
947  while ((digitab[*p] & ctype_digit) != 0) p++;  while ((digitab[*p] & ctype_digit) != 0) p++;
948    
949  return (*p == '}');  return (*p == CHAR_RIGHT_CURLY_BRACKET);
950  }  }
951    
952    
# Line 797  int max = -1; Line 979  int max = -1;
979  /* Read the minimum value and do a paranoid check: a negative value indicates  /* Read the minimum value and do a paranoid check: a negative value indicates
980  an integer overflow. */  an integer overflow. */
981    
982  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - '0';  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;
983  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
984    {    {
985    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 807  if (min < 0 || min > 65535) Line 989  if (min < 0 || min > 65535)
989  /* Read the maximum value if there is one, and again do a paranoid on its size.  /* Read the maximum value if there is one, and again do a paranoid on its size.
990  Also, max must not be less than min. */  Also, max must not be less than min. */
991    
992  if (*p == '}') max = min; else  if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else
993    {    {
994    if (*(++p) != '}')    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
995      {      {
996      max = 0;      max = 0;
997      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - '0';      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;
998      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
999        {        {
1000        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 837  return p; Line 1019  return p;
1019    
1020    
1021  /*************************************************  /*************************************************
1022  *       Find forward referenced subpattern       *  *  Subroutine for finding forward reference      *
1023  *************************************************/  *************************************************/
1024    
1025  /* This function scans along a pattern's text looking for capturing  /* This recursive function is called only from find_parens() below. The
1026    top-level call starts at the beginning of the pattern. All other calls must
1027    start at a parenthesis. It scans along a pattern's text looking for capturing
1028  subpatterns, and counting them. If it finds a named pattern that matches the  subpatterns, and counting them. If it finds a named pattern that matches the
1029  name it is given, it returns its number. Alternatively, if the name is NULL, it  name it is given, it returns its number. Alternatively, if the name is NULL, it
1030  returns when it reaches a given numbered subpattern. This is used for forward  returns when it reaches a given numbered subpattern. We know that if (?P< is
1031  references to subpatterns. We know that if (?P< is encountered, the name will  encountered, the name will be terminated by '>' because that is checked in the
1032  be terminated by '>' because that is checked in the first pass.  first pass. Recursion is used to keep track of subpatterns that reset the
1033    capturing group numbers - the (?| feature.
1034    
1035  Arguments:  Arguments:
1036    ptr          current position in the pattern    ptrptr       address of the current character pointer (updated)
1037    count        current count of capturing parens so far encountered    cd           compile background data
1038    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1039    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1040    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1041      count        pointer to the current capturing subpattern number (updated)
1042    
1043  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1044  */  */
1045    
1046  static int  static int
1047  find_parens(const uschar *ptr, int count, const uschar *name, int lorn,  find_parens_sub(uschar **ptrptr, compile_data *cd, const uschar *name, int lorn,
1048    BOOL xmode)    BOOL xmode, int *count)
1049  {  {
1050  const uschar *thisname;  uschar *ptr = *ptrptr;
1051    int start_count = *count;
1052    int hwm_count = start_count;
1053    BOOL dup_parens = FALSE;
1054    
1055  for (; *ptr != 0; ptr++)  /* If the first character is a parenthesis, check on the type of group we are
1056    dealing with. The very first call may not start with a parenthesis. */
1057    
1058    if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1059    {    {
1060    int term;    if (ptr[1] == CHAR_QUESTION_MARK &&
1061          ptr[2] == CHAR_VERTICAL_LINE)
1062        {
1063        ptr += 3;
1064        dup_parens = TRUE;
1065        }
1066    
1067      /* Handle a normal, unnamed capturing parenthesis */
1068    
1069      else if (ptr[1] != CHAR_QUESTION_MARK && ptr[1] != CHAR_ASTERISK)
1070        {
1071        *count += 1;
1072        if (name == NULL && *count == lorn) return *count;
1073        ptr++;
1074        }
1075    
1076      /* Handle a condition. If it is an assertion, just carry on so that it
1077      is processed as normal. If not, skip to the closing parenthesis of the
1078      condition (there can't be any nested parens. */
1079    
1080      else if (ptr[2] == CHAR_LEFT_PARENTHESIS)
1081        {
1082        ptr += 2;
1083        if (ptr[1] != CHAR_QUESTION_MARK)
1084          {
1085          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1086          if (*ptr != 0) ptr++;
1087          }
1088        }
1089    
1090      /* We have either (? or (* and not a condition */
1091    
1092      else
1093        {
1094        ptr += 2;
1095        if (*ptr == CHAR_P) ptr++;                      /* Allow optional P */
1096    
1097        /* We have to disambiguate (?<! and (?<= from (?<name> for named groups */
1098    
1099        if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1100            ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1101          {
1102          int term;
1103          const uschar *thisname;
1104          *count += 1;
1105          if (name == NULL && *count == lorn) return *count;
1106          term = *ptr++;
1107          if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1108          thisname = ptr;
1109          while (*ptr != term) ptr++;
1110          if (name != NULL && lorn == ptr - thisname &&
1111              strncmp((const char *)name, (const char *)thisname, lorn) == 0)
1112            return *count;
1113          term++;
1114          }
1115        }
1116      }
1117    
1118    /* Past any initial parenthesis handling, scan for parentheses or vertical
1119    bars. */
1120    
1121    for (; *ptr != 0; ptr++)
1122      {
1123    /* Skip over backslashed characters and also entire \Q...\E */    /* Skip over backslashed characters and also entire \Q...\E */
1124    
1125    if (*ptr == '\\')    if (*ptr == CHAR_BACKSLASH)
1126      {      {
1127      if (*(++ptr) == 0) return -1;      if (*(++ptr) == 0) goto FAIL_EXIT;
1128      if (*ptr == 'Q') for (;;)      if (*ptr == CHAR_Q) for (;;)
1129        {        {
1130        while (*(++ptr) != 0 && *ptr != '\\');        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};
1131        if (*ptr == 0) return -1;        if (*ptr == 0) goto FAIL_EXIT;
1132        if (*(++ptr) == 'E') break;        if (*(++ptr) == CHAR_E) break;
1133        }        }
1134      continue;      continue;
1135      }      }
1136    
1137    /* Skip over character classes */    /* Skip over character classes; this logic must be similar to the way they
1138      are handled for real. If the first character is '^', skip it. Also, if the
1139      first few characters (either before or after ^) are \Q\E or \E we skip them
1140      too. This makes for compatibility with Perl. Note the use of STR macros to
1141      encode "Q\\E" so that it works in UTF-8 on EBCDIC platforms. */
1142    
1143    if (*ptr == '[')    if (*ptr == CHAR_LEFT_SQUARE_BRACKET)
1144      {      {
1145      while (*(++ptr) != ']')      BOOL negate_class = FALSE;
1146        for (;;)
1147        {        {
1148        if (*ptr == '\\')        if (ptr[1] == CHAR_BACKSLASH)
1149            {
1150            if (ptr[2] == CHAR_E)
1151              ptr+= 2;
1152            else if (strncmp((const char *)ptr+2,
1153                     STR_Q STR_BACKSLASH STR_E, 3) == 0)
1154              ptr += 4;
1155            else
1156              break;
1157            }
1158          else if (!negate_class && ptr[1] == CHAR_CIRCUMFLEX_ACCENT)
1159          {          {
1160          if (*(++ptr) == 0) return -1;          negate_class = TRUE;
1161          if (*ptr == 'Q') for (;;)          ptr++;
1162            }
1163          else break;
1164          }
1165    
1166        /* If the next character is ']', it is a data character that must be
1167        skipped, except in JavaScript compatibility mode. */
1168    
1169        if (ptr[1] == CHAR_RIGHT_SQUARE_BRACKET &&
1170            (cd->external_options & PCRE_JAVASCRIPT_COMPAT) == 0)
1171          ptr++;
1172    
1173        while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1174          {
1175          if (*ptr == 0) return -1;
1176          if (*ptr == CHAR_BACKSLASH)
1177            {
1178            if (*(++ptr) == 0) goto FAIL_EXIT;
1179            if (*ptr == CHAR_Q) for (;;)
1180            {            {
1181            while (*(++ptr) != 0 && *ptr != '\\');            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};
1182            if (*ptr == 0) return -1;            if (*ptr == 0) goto FAIL_EXIT;
1183            if (*(++ptr) == 'E') break;            if (*(++ptr) == CHAR_E) break;
1184            }            }
1185          continue;          continue;
1186          }          }
# Line 904  for (; *ptr != 0; ptr++) Line 1190  for (; *ptr != 0; ptr++)
1190    
1191    /* Skip comments in /x mode */    /* Skip comments in /x mode */
1192    
1193    if (xmode && *ptr == '#')    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1194      {      {
1195      while (*(++ptr) != 0 && *ptr != '\n');      while (*(++ptr) != 0 && *ptr != CHAR_NL) {};
1196      if (*ptr == 0) return -1;      if (*ptr == 0) goto FAIL_EXIT;
1197      continue;      continue;
1198      }      }
1199    
1200    /* An opening parens must now be a real metacharacter */    /* Check for the special metacharacters */
1201    
1202    if (*ptr != '(') continue;    if (*ptr == CHAR_LEFT_PARENTHESIS)
   if (ptr[1] != '?')  
1203      {      {
1204      count++;      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, count);
1205      if (name == NULL && count == lorn) return count;      if (rc > 0) return rc;
1206      continue;      if (*ptr == 0) goto FAIL_EXIT;
1207      }      }
1208    
1209    ptr += 2;    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
1210    if (*ptr == 'P') ptr++;                      /* Allow optional P */      {
1211        if (dup_parens && *count < hwm_count) *count = hwm_count;
1212        *ptrptr = ptr;
1213        return -1;
1214        }
1215    
1216    /* We have to disambiguate (?<! and (?<= from (?<name> */    else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)
1217        {
1218        if (*count > hwm_count) hwm_count = *count;
1219        *count = start_count;
1220        }
1221      }
1222    
1223    if ((*ptr != '<' || ptr[1] == '!' || ptr[1] == '=') &&  FAIL_EXIT:
1224         *ptr != '\'')  *ptrptr = ptr;
1225      continue;  return -1;
1226    }
1227    
1228    
1229    
1230    
1231    /*************************************************
1232    *       Find forward referenced subpattern       *
1233    *************************************************/
1234    
1235    /* This function scans along a pattern's text looking for capturing
1236    subpatterns, and counting them. If it finds a named pattern that matches the
1237    name it is given, it returns its number. Alternatively, if the name is NULL, it
1238    returns when it reaches a given numbered subpattern. This is used for forward
1239    references to subpatterns. We used to be able to start this scan from the
1240    current compiling point, using the current count value from cd->bracount, and
1241    do it all in a single loop, but the addition of the possibility of duplicate
1242    subpattern numbers means that we have to scan from the very start, in order to
1243    take account of such duplicates, and to use a recursive function to keep track
1244    of the different types of group.
1245    
1246    Arguments:
1247      cd           compile background data
1248      name         name to seek, or NULL if seeking a numbered subpattern
1249      lorn         name length, or subpattern number if name is NULL
1250      xmode        TRUE if we are in /x mode
1251    
1252    Returns:       the number of the found subpattern, or -1 if not found
1253    */
1254    
1255    count++;  static int
1256    find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode)
1257    {
1258    uschar *ptr = (uschar *)cd->start_pattern;
1259    int count = 0;
1260    int rc;
1261    
1262    /* If the pattern does not start with an opening parenthesis, the first call
1263    to find_parens_sub() will scan right to the end (if necessary). However, if it
1264    does start with a parenthesis, find_parens_sub() will return when it hits the
1265    matching closing parens. That is why we have to have a loop. */
1266    
1267    if (name == NULL && count == lorn) return count;  for (;;)
1268    term = *ptr++;    {
1269    if (term == '<') term = '>';    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, &count);
1270    thisname = ptr;    if (rc > 0 || *ptr++ == 0) break;
   while (*ptr != term) ptr++;  
   if (name != NULL && lorn == ptr - thisname &&  
       strncmp((const char *)name, (const char *)thisname, lorn) == 0)  
     return count;  
1271    }    }
1272    
1273  return -1;  return rc;
1274  }  }
1275    
1276    
1277    
1278    
1279  /*************************************************  /*************************************************
1280  *      Find first significant op code            *  *      Find first significant op code            *
1281  *************************************************/  *************************************************/
# Line 996  for (;;) Line 1325  for (;;)
1325    
1326      case OP_CALLOUT:      case OP_CALLOUT:
1327      case OP_CREF:      case OP_CREF:
1328        case OP_NCREF:
1329      case OP_RREF:      case OP_RREF:
1330        case OP_NRREF:
1331      case OP_DEF:      case OP_DEF:
1332      code += _pcre_OP_lengths[*code];      code += _pcre_OP_lengths[*code];
1333      break;      break;
# Line 1012  for (;;) Line 1343  for (;;)
1343    
1344    
1345  /*************************************************  /*************************************************
1346  *        Find the fixed length of a pattern      *  *        Find the fixed length of a branch       *
1347  *************************************************/  *************************************************/
1348    
1349  /* Scan a pattern and compute the fixed length of subject that will match it,  /* Scan a branch and compute the fixed length of subject that will match it,
1350  if the length is fixed. This is needed for dealing with backward assertions.  if the length is fixed. This is needed for dealing with backward assertions.
1351  In UTF8 mode, the result is in characters rather than bytes.  In UTF8 mode, the result is in characters rather than bytes. The branch is
1352    temporarily terminated with OP_END when this function is called.
1353    
1354    This function is called when a backward assertion is encountered, so that if it
1355    fails, the error message can point to the correct place in the pattern.
1356    However, we cannot do this when the assertion contains subroutine calls,
1357    because they can be forward references. We solve this by remembering this case
1358    and doing the check at the end; a flag specifies which mode we are running in.
1359    
1360  Arguments:  Arguments:
1361    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1362    options  the compiling options    options  the compiling options
1363      atend    TRUE if called when the pattern is complete
1364      cd       the "compile data" structure
1365    
1366  Returns:   the fixed length, or -1 if there is no fixed length,  Returns:   the fixed length,
1367                 or -1 if there is no fixed length,
1368               or -2 if \C was encountered               or -2 if \C was encountered
1369                 or -3 if an OP_RECURSE item was encountered and atend is FALSE
1370  */  */
1371    
1372  static int  static int
1373  find_fixedlength(uschar *code, int options)  find_fixedlength(uschar *code, int options, BOOL atend, compile_data *cd)
1374  {  {
1375  int length = -1;  int length = -1;
1376    
# Line 1041  branch, check the length against that of Line 1383  branch, check the length against that of
1383  for (;;)  for (;;)
1384    {    {
1385    int d;    int d;
1386      uschar *ce, *cs;
1387    register int op = *cc;    register int op = *cc;
   
1388    switch (op)    switch (op)
1389      {      {
1390      case OP_CBRA:      case OP_CBRA:
1391      case OP_BRA:      case OP_BRA:
1392      case OP_ONCE:      case OP_ONCE:
1393      case OP_COND:      case OP_COND:
1394      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), options);      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), options, atend, cd);
1395      if (d < 0) return d;      if (d < 0) return d;
1396      branchlength += d;      branchlength += d;
1397      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
# Line 1072  for (;;) Line 1414  for (;;)
1414      branchlength = 0;      branchlength = 0;
1415      break;      break;
1416    
1417        /* A true recursion implies not fixed length, but a subroutine call may
1418        be OK. If the subroutine is a forward reference, we can't deal with
1419        it until the end of the pattern, so return -3. */
1420    
1421        case OP_RECURSE:
1422        if (!atend) return -3;
1423        cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1424        do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */
1425        if (cc > cs && cc < ce) return -1;                /* Recursion */
1426        d = find_fixedlength(cs + 2, options, atend, cd);
1427        if (d < 0) return d;
1428        branchlength += d;
1429        cc += 1 + LINK_SIZE;
1430        break;
1431    
1432      /* Skip over assertive subpatterns */      /* Skip over assertive subpatterns */
1433    
1434      case OP_ASSERT:      case OP_ASSERT:
# Line 1085  for (;;) Line 1442  for (;;)
1442    
1443      case OP_REVERSE:      case OP_REVERSE:
1444      case OP_CREF:      case OP_CREF:
1445        case OP_NCREF:
1446      case OP_RREF:      case OP_RREF:
1447        case OP_NRREF:
1448      case OP_DEF:      case OP_DEF:
1449      case OP_OPT:      case OP_OPT:
1450      case OP_CALLOUT:      case OP_CALLOUT:
1451      case OP_SOD:      case OP_SOD:
1452      case OP_SOM:      case OP_SOM:
1453        case OP_SET_SOM:
1454      case OP_EOD:      case OP_EOD:
1455      case OP_EODN:      case OP_EODN:
1456      case OP_CIRC:      case OP_CIRC:
# Line 1108  for (;;) Line 1468  for (;;)
1468      branchlength++;      branchlength++;
1469      cc += 2;      cc += 2;
1470  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1471      if ((options & PCRE_UTF8) != 0)      if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0)
1472        {        cc += _pcre_utf8_table4[cc[-1] & 0x3f];
       while ((*cc & 0xc0) == 0x80) cc++;  
       }  
1473  #endif  #endif
1474      break;      break;
1475    
# Line 1122  for (;;) Line 1480  for (;;)
1480      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1481      cc += 4;      cc += 4;
1482  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1483      if ((options & PCRE_UTF8) != 0)      if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0)
1484        {        cc += _pcre_utf8_table4[cc[-1] & 0x3f];
       while((*cc & 0x80) == 0x80) cc++;  
       }  
1485  #endif  #endif
1486      break;      break;
1487    
1488      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1489      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1490        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1491      cc += 4;      cc += 4;
1492      break;      break;
1493    
# Line 1148  for (;;) Line 1505  for (;;)
1505      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
1506      case OP_WORDCHAR:      case OP_WORDCHAR:
1507      case OP_ANY:      case OP_ANY:
1508        case OP_ALLANY:
1509      branchlength++;      branchlength++;
1510      cc++;      cc++;
1511      break;      break;
# Line 1202  for (;;) Line 1560  for (;;)
1560    
1561    
1562  /*************************************************  /*************************************************
1563  *    Scan compiled regex for numbered bracket    *  *    Scan compiled regex for specific bracket    *
1564  *************************************************/  *************************************************/
1565    
1566  /* This little function scans through a compiled pattern until it finds a  /* This little function scans through a compiled pattern until it finds a
1567  capturing bracket with the given number.  capturing bracket with the given number, or, if the number is negative, an
1568    instance of OP_REVERSE for a lookbehind. The function is global in the C sense
1569    so that it can be called from pcre_study() when finding the minimum matching
1570    length.
1571    
1572  Arguments:  Arguments:
1573    code        points to start of expression    code        points to start of expression
1574    utf8        TRUE in UTF-8 mode    utf8        TRUE in UTF-8 mode
1575    number      the required bracket number    number      the required bracket number or negative to find a lookbehind
1576    
1577  Returns:      pointer to the opcode for the bracket, or NULL if not found  Returns:      pointer to the opcode for the bracket, or NULL if not found
1578  */  */
1579    
1580  static const uschar *  const uschar *
1581  find_bracket(const uschar *code, BOOL utf8, int number)  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)
1582  {  {
1583  for (;;)  for (;;)
1584    {    {
# Line 1230  for (;;) Line 1591  for (;;)
1591    
1592    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1593    
1594      /* Handle recursion */
1595    
1596      else if (c == OP_REVERSE)
1597        {
1598        if (number < 0) return (uschar *)code;
1599        code += _pcre_OP_lengths[c];
1600        }
1601    
1602    /* Handle capturing bracket */    /* Handle capturing bracket */
1603    
1604    else if (c == OP_CBRA)    else if (c == OP_CBRA)
# Line 1239  for (;;) Line 1608  for (;;)
1608      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1609      }      }
1610    
1611    /* In UTF-8 mode, opcodes that are followed by a character may be followed by    /* Otherwise, we can get the item's length from the table, except that for
1612    a multi-byte character. The length in the table is a minimum, so we have to    repeated character types, we have to test for \p and \P, which have an extra
1613    arrange to skip the extra bytes. */    two bytes of parameters. */
1614    
1615    else    else
1616      {      {
1617        switch(c)
1618          {
1619          case OP_TYPESTAR:
1620          case OP_TYPEMINSTAR:
1621          case OP_TYPEPLUS:
1622          case OP_TYPEMINPLUS:
1623          case OP_TYPEQUERY:
1624          case OP_TYPEMINQUERY:
1625          case OP_TYPEPOSSTAR:
1626          case OP_TYPEPOSPLUS:
1627          case OP_TYPEPOSQUERY:
1628          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1629          break;
1630    
1631          case OP_TYPEUPTO:
1632          case OP_TYPEMINUPTO:
1633          case OP_TYPEEXACT:
1634          case OP_TYPEPOSUPTO:
1635          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1636          break;
1637          }
1638    
1639        /* Add in the fixed length from the table */
1640    
1641      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1642    
1643      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1644      a multi-byte character. The length in the table is a minimum, so we have to
1645      arrange to skip the extra bytes. */
1646    
1647  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1648      if (utf8) switch(c)      if (utf8) switch(c)
1649        {        {
# Line 1267  for (;;) Line 1665  for (;;)
1665        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1666        break;        break;
1667        }        }
1668    #else
1669        (void)(utf8);  /* Keep compiler happy by referencing function argument */
1670  #endif  #endif
1671      }      }
1672    }    }
# Line 1303  for (;;) Line 1703  for (;;)
1703    
1704    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1705    
1706    /* Otherwise, we get the item's length from the table. In UTF-8 mode, opcodes    /* Otherwise, we can get the item's length from the table, except that for
1707    that are followed by a character may be followed by a multi-byte character.    repeated character types, we have to test for \p and \P, which have an extra
1708    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1709    
1710    else    else
1711      {      {
1712        switch(c)
1713          {
1714          case OP_TYPESTAR:
1715          case OP_TYPEMINSTAR:
1716          case OP_TYPEPLUS:
1717          case OP_TYPEMINPLUS:
1718          case OP_TYPEQUERY:
1719          case OP_TYPEMINQUERY:
1720          case OP_TYPEPOSSTAR:
1721          case OP_TYPEPOSPLUS:
1722          case OP_TYPEPOSQUERY:
1723          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1724          break;
1725    
1726          case OP_TYPEPOSUPTO:
1727          case OP_TYPEUPTO:
1728          case OP_TYPEMINUPTO:
1729          case OP_TYPEEXACT:
1730          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1731          break;
1732          }
1733    
1734        /* Add in the fixed length from the table */
1735    
1736      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1737    
1738        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1739        by a multi-byte character. The length in the table is a minimum, so we have
1740        to arrange to skip the extra bytes. */
1741    
1742  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1743      if (utf8) switch(c)      if (utf8) switch(c)
1744        {        {
# Line 1332  for (;;) Line 1760  for (;;)
1760        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1761        break;        break;
1762        }        }
1763    #else
1764        (void)(utf8);  /* Keep compiler happy by referencing function argument */
1765  #endif  #endif
1766      }      }
1767    }    }
# Line 1347  for (;;) Line 1777  for (;;)
1777  can match the empty string or not. It is called from could_be_empty()  can match the empty string or not. It is called from could_be_empty()
1778  below and from compile_branch() when checking for an unlimited repeat of a  below and from compile_branch() when checking for an unlimited repeat of a
1779  group that can match nothing. Note that first_significant_code() skips over  group that can match nothing. Note that first_significant_code() skips over
1780  assertions. If we hit an unclosed bracket, we return "empty" - this means we've  backward and negative forward assertions when its final argument is TRUE. If we
1781  struck an inner bracket whose current branch will already have been scanned.  hit an unclosed bracket, we return "empty" - this means we've struck an inner
1782    bracket whose current branch will already have been scanned.
1783    
1784  Arguments:  Arguments:
1785    code        points to start of search    code        points to start of search
1786    endcode     points to where to stop    endcode     points to where to stop
1787    utf8        TRUE if in UTF8 mode    utf8        TRUE if in UTF8 mode
1788      cd          contains pointers to tables etc.
1789    
1790  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
1791  */  */
1792    
1793  static BOOL  static BOOL
1794  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8)  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,
1795      compile_data *cd)
1796  {  {
1797  register int c;  register int c;
1798  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);
# Line 1369  for (code = first_significant_code(code Line 1802  for (code = first_significant_code(code
1802    const uschar *ccode;    const uschar *ccode;
1803    
1804    c = *code;    c = *code;
1805    
1806      /* Skip over forward assertions; the other assertions are skipped by
1807      first_significant_code() with a TRUE final argument. */
1808    
1809    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    if (c == OP_ASSERT)
1810      {      {
1811      BOOL empty_branch;      do code += GET(code, 1); while (*code == OP_ALT);
1812      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      c = *code;
1813        continue;
1814        }
1815    
1816      /* Scan a closed bracket */    /* Groups with zero repeats can of course be empty; skip them. */
1817    
1818      empty_branch = FALSE;    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO)
1819        {
1820        code += _pcre_OP_lengths[c];
1821        do code += GET(code, 1); while (*code == OP_ALT);
1822        c = *code;
1823        continue;
1824        }
1825    
1826      /* For a recursion/subroutine call, if its end has been reached, which
1827      implies a subroutine call, we can scan it. */
1828    
1829      if (c == OP_RECURSE)
1830        {
1831        BOOL empty_branch = FALSE;
1832        const uschar *scode = cd->start_code + GET(code, 1);
1833        if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
1834      do      do
1835        {        {
1836        if (!empty_branch && could_be_empty_branch(code, endcode, utf8))        if (could_be_empty_branch(scode, endcode, utf8, cd))
1837            {
1838          empty_branch = TRUE;          empty_branch = TRUE;
1839        code += GET(code, 1);          break;
1840            }
1841          scode += GET(scode, 1);
1842        }        }
1843      while (*code == OP_ALT);      while (*scode == OP_ALT);
1844      if (!empty_branch) return FALSE;   /* All branches are non-empty */      if (!empty_branch) return FALSE;  /* All branches are non-empty */
1845        continue;
1846        }
1847    
1848      /* For other groups, scan the branches. */
1849    
1850      /* Move past the KET and fudge things so that the increment in the "for"    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1851      above has no effect. */      {
1852        BOOL empty_branch;
1853        if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
1854    
1855      c = OP_END;      /* If a conditional group has only one branch, there is a second, implied,
1856      code += 1 + LINK_SIZE - _pcre_OP_lengths[c];      empty branch, so just skip over the conditional, because it could be empty.
1857        Otherwise, scan the individual branches of the group. */
1858    
1859        if (c == OP_COND && code[GET(code, 1)] != OP_ALT)
1860          code += GET(code, 1);
1861        else
1862          {
1863          empty_branch = FALSE;
1864          do
1865            {
1866            if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))
1867              empty_branch = TRUE;
1868            code += GET(code, 1);
1869            }
1870          while (*code == OP_ALT);
1871          if (!empty_branch) return FALSE;   /* All branches are non-empty */
1872          }
1873    
1874        c = *code;
1875      continue;      continue;
1876      }      }
1877    
# Line 1399  for (code = first_significant_code(code Line 1879  for (code = first_significant_code(code
1879    
1880    switch (c)    switch (c)
1881      {      {
1882      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1883        cannot be represented just by a bit map. This includes negated single
1884        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1885        actual length is stored in the compiled code, so we must update "code"
1886        here. */
1887    
1888  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1889      case OP_XCLASS:      case OP_XCLASS:
1890      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1891      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1892  #endif  #endif
1893    
# Line 1447  for (code = first_significant_code(code Line 1931  for (code = first_significant_code(code
1931      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
1932      case OP_WORDCHAR:      case OP_WORDCHAR:
1933      case OP_ANY:      case OP_ANY:
1934        case OP_ALLANY:
1935      case OP_ANYBYTE:      case OP_ANYBYTE:
1936      case OP_CHAR:      case OP_CHAR:
1937      case OP_CHARNC:      case OP_CHARNC:
# Line 1465  for (code = first_significant_code(code Line 1950  for (code = first_significant_code(code
1950      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1951      return FALSE;      return FALSE;
1952    
1953        /* These are going to continue, as they may be empty, but we have to
1954        fudge the length for the \p and \P cases. */
1955    
1956        case OP_TYPESTAR:
1957        case OP_TYPEMINSTAR:
1958        case OP_TYPEPOSSTAR:
1959        case OP_TYPEQUERY:
1960        case OP_TYPEMINQUERY:
1961        case OP_TYPEPOSQUERY:
1962        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1963        break;
1964    
1965        /* Same for these */
1966    
1967        case OP_TYPEUPTO:
1968        case OP_TYPEMINUPTO:
1969        case OP_TYPEPOSUPTO:
1970        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1971        break;
1972    
1973      /* End of branch */      /* End of branch */
1974    
1975      case OP_KET:      case OP_KET:
# Line 1483  for (code = first_significant_code(code Line 1988  for (code = first_significant_code(code
1988      case OP_QUERY:      case OP_QUERY:
1989      case OP_MINQUERY:      case OP_MINQUERY:
1990      case OP_POSQUERY:      case OP_POSQUERY:
1991        if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];
1992        break;
1993    
1994      case OP_UPTO:      case OP_UPTO:
1995      case OP_MINUPTO:      case OP_MINUPTO:
1996      case OP_POSUPTO:      case OP_POSUPTO:
1997      if (utf8) while ((code[2] & 0xc0) == 0x80) code++;      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];
1998      break;      break;
1999  #endif  #endif
2000    
2001        /* None of the remaining opcodes are required to match a character. */
2002    
2003        default:
2004        break;
2005      }      }
2006    }    }
2007    
# Line 1511  Arguments: Line 2024  Arguments:
2024    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2025    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2026    utf8        TRUE if in UTF-8 mode    utf8        TRUE if in UTF-8 mode
2027      cd          pointers to tables etc
2028    
2029  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2030  */  */
2031    
2032  static BOOL  static BOOL
2033  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,
2034    BOOL utf8)    BOOL utf8, compile_data *cd)
2035  {  {
2036  while (bcptr != NULL && bcptr->current >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2037    {    {
2038    if (!could_be_empty_branch(bcptr->current, endcode, utf8)) return FALSE;    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))
2039        return FALSE;
2040    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2041    }    }
2042  return TRUE;  return TRUE;
# Line 1534  return TRUE; Line 2049  return TRUE;
2049  *************************************************/  *************************************************/
2050    
2051  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
2052  encountered in a character class. It checks whether this is followed by an  encountered in a character class. It checks whether this is followed by a
2053  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
2054  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
2055    
2056    Originally, this function only recognized a sequence of letters between the
2057    terminators, but it seems that Perl recognizes any sequence of characters,
2058    though of course unknown POSIX names are subsequently rejected. Perl gives an
2059    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
2060    didn't consider this to be a POSIX class. Likewise for [:1234:].
2061    
2062    The problem in trying to be exactly like Perl is in the handling of escapes. We
2063    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
2064    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
2065    below handles the special case of \], but does not try to do any other escape
2066    processing. This makes it different from Perl for cases such as [:l\ower:]
2067    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
2068    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
2069    I think.
2070    
2071  Argument:  Arguments:
2072    ptr      pointer to the initial [    ptr      pointer to the initial [
2073    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
2074    
2075  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
2076  */  */
2077    
2078  static BOOL  static BOOL
2079  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
2080  {  {
2081  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
2082  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2083  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
2084    {    {
2085    *endptr = ptr;    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else
2086    return TRUE;      {
2087        if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2088        if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2089          {
2090          *endptr = ptr;
2091          return TRUE;
2092          }
2093        }
2094    }    }
2095  return FALSE;  return FALSE;
2096  }  }
# Line 1581  Returns:     a value representing the na Line 2115  Returns:     a value representing the na
2115  static int  static int
2116  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
2117  {  {
2118    const char *pn = posix_names;
2119  register int yield = 0;  register int yield = 0;
2120  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2121    {    {
2122    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2123      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
2124      pn += posix_name_lengths[yield] + 1;
2125    yield++;    yield++;
2126    }    }
2127  return -1;  return -1;
# Line 1600  return -1; Line 2136  return -1;
2136  that is referenced. This means that groups can be replicated for fixed  that is referenced. This means that groups can be replicated for fixed
2137  repetition simply by copying (because the recursion is allowed to refer to  repetition simply by copying (because the recursion is allowed to refer to
2138  earlier groups that are outside the current group). However, when a group is  earlier groups that are outside the current group). However, when a group is
2139  optional (i.e. the minimum quantifier is zero), OP_BRAZERO is inserted before  optional (i.e. the minimum quantifier is zero), OP_BRAZERO or OP_SKIPZERO is
2140  it, after it has been compiled. This means that any OP_RECURSE items within it  inserted before it, after it has been compiled. This means that any OP_RECURSE
2141  that refer to the group itself or any contained groups have to have their  items within it that refer to the group itself or any contained groups have to
2142  offsets adjusted. That one of the jobs of this function. Before it is called,  have their offsets adjusted. That one of the jobs of this function. Before it
2143  the partially compiled regex must be temporarily terminated with OP_END.  is called, the partially compiled regex must be temporarily terminated with
2144    OP_END.
2145    
2146  This function has been extended with the possibility of forward references for  This function has been extended with the possibility of forward references for
2147  recursions and subroutine calls. It must also check the list of such references  recursions and subroutine calls. It must also check the list of such references
# Line 1627  adjust_recurse(uschar *group, int adjust Line 2164  adjust_recurse(uschar *group, int adjust
2164    uschar *save_hwm)    uschar *save_hwm)
2165  {  {
2166  uschar *ptr = group;  uschar *ptr = group;
2167    
2168  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
2169    {    {
2170    int offset;    int offset;
# Line 1738  get_othercase_range(unsigned int *cptr, Line 2276  get_othercase_range(unsigned int *cptr,
2276  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2277    
2278  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2279    { if ((othercase = _pcre_ucp_othercase(c)) != NOTACHAR) break; }    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }
2280    
2281  if (c > d) return FALSE;  if (c > d) return FALSE;
2282    
# Line 1747  next = othercase + 1; Line 2285  next = othercase + 1;
2285    
2286  for (++c; c <= d; c++)  for (++c; c <= d; c++)
2287    {    {
2288    if (_pcre_ucp_othercase(c) != next) break;    if (UCD_OTHERCASE(c) != next) break;
2289    next++;    next++;
2290    }    }
2291    
# Line 1793  if ((options & PCRE_EXTENDED) != 0) Line 2331  if ((options & PCRE_EXTENDED) != 0)
2331    for (;;)    for (;;)
2332      {      {
2333      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2334      if (*ptr == '#')      if (*ptr == CHAR_NUMBER_SIGN)
2335        {        {
2336        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2337          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
# Line 1805  if ((options & PCRE_EXTENDED) != 0) Line 2343  if ((options & PCRE_EXTENDED) != 0)
2343  /* If the next item is one that we can handle, get its value. A non-negative  /* If the next item is one that we can handle, get its value. A non-negative
2344  value is a character, a negative value is an escape value. */  value is a character, a negative value is an escape value. */
2345    
2346  if (*ptr == '\\')  if (*ptr == CHAR_BACKSLASH)
2347    {    {
2348    int temperrorcode = 0;    int temperrorcode = 0;
2349    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);
# Line 1830  if ((options & PCRE_EXTENDED) != 0) Line 2368  if ((options & PCRE_EXTENDED) != 0)
2368    for (;;)    for (;;)
2369      {      {
2370      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2371      if (*ptr == '#')      if (*ptr == CHAR_NUMBER_SIGN)
2372        {        {
2373        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2374          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
# Line 1841  if ((options & PCRE_EXTENDED) != 0) Line 2379  if ((options & PCRE_EXTENDED) != 0)
2379    
2380  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
2381    
2382  if (*ptr == '*' || *ptr == '?' || strncmp((char *)ptr, "{0,", 3) == 0)  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
2383    return FALSE;    strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
2384        return FALSE;
2385    
2386  /* Now compare the next item with the previous opcode. If the previous is a  /* Now compare the next item with the previous opcode. If the previous is a
2387  positive single character match, "item" either contains the character or, if  positive single character match, "item" either contains the character or, if
# Line 1857  if (next >= 0) switch(op_code) Line 2396  if (next >= 0) switch(op_code)
2396    case OP_CHAR:    case OP_CHAR:
2397  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2398    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }
2399    #else
2400      (void)(utf8_char);  /* Keep compiler happy by referencing function argument */
2401  #endif  #endif
2402    return item != next;    return item != next;
2403    
# Line 1875  if (next >= 0) switch(op_code) Line 2416  if (next >= 0) switch(op_code)
2416      unsigned int othercase;      unsigned int othercase;
2417      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
2418  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2419      othercase = _pcre_ucp_othercase((unsigned int)next);      othercase = UCD_OTHERCASE((unsigned int)next);
2420  #else  #else
2421      othercase = NOTACHAR;      othercase = NOTACHAR;
2422  #endif  #endif
# Line 1888  if (next >= 0) switch(op_code) Line 2429  if (next >= 0) switch(op_code)
2429    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, "item" must be a single-byte character. */
2430    
2431    case OP_NOT:    case OP_NOT:
   if (next < 0) return FALSE;  /* Not a character */  
2432    if (item == next) return TRUE;    if (item == next) return TRUE;
2433    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2434  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1897  if (next >= 0) switch(op_code) Line 2437  if (next >= 0) switch(op_code)
2437      unsigned int othercase;      unsigned int othercase;
2438      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
2439  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2440      othercase = _pcre_ucp_othercase(next);      othercase = UCD_OTHERCASE(next);
2441  #else  #else
2442      othercase = NOTACHAR;      othercase = NOTACHAR;
2443  #endif  #endif
# Line 1925  if (next >= 0) switch(op_code) Line 2465  if (next >= 0) switch(op_code)
2465    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2466    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2467    
2468      case OP_HSPACE:
2469      case OP_NOT_HSPACE:
2470      switch(next)
2471        {
2472        case 0x09:
2473        case 0x20:
2474        case 0xa0:
2475        case 0x1680:
2476        case 0x180e:
2477        case 0x2000:
2478        case 0x2001:
2479        case 0x2002:
2480        case 0x2003:
2481        case 0x2004:
2482        case 0x2005:
2483        case 0x2006:
2484        case 0x2007:
2485        case 0x2008:
2486        case 0x2009:
2487        case 0x200A:
2488        case 0x202f:
2489        case 0x205f:
2490        case 0x3000:
2491        return op_code != OP_HSPACE;
2492        default:
2493        return op_code == OP_HSPACE;
2494        }
2495    
2496      case OP_VSPACE:
2497      case OP_NOT_VSPACE:
2498      switch(next)
2499        {
2500        case 0x0a:
2501        case 0x0b:
2502        case 0x0c:
2503        case 0x0d:
2504        case 0x85:
2505        case 0x2028:
2506        case 0x2029:
2507        return op_code != OP_VSPACE;
2508        default:
2509        return op_code == OP_VSPACE;
2510        }
2511    
2512    default:    default:
2513    return FALSE;    return FALSE;
2514    }    }
# Line 1959  switch(op_code) Line 2543  switch(op_code)
2543      case ESC_W:      case ESC_W:
2544      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2545    
2546        case ESC_h:
2547        case ESC_H:
2548        switch(item)
2549          {
2550          case 0x09:
2551          case 0x20:
2552          case 0xa0:
2553          case 0x1680:
2554          case 0x180e:
2555          case 0x2000:
2556          case 0x2001:
2557          case 0x2002:
2558          case 0x2003:
2559          case 0x2004:
2560          case 0x2005:
2561          case 0x2006:
2562          case 0x2007:
2563          case 0x2008:
2564          case 0x2009:
2565          case 0x200A:
2566          case 0x202f:
2567          case 0x205f:
2568          case 0x3000:
2569          return -next != ESC_h;
2570          default:
2571          return -next == ESC_h;
2572          }
2573    
2574        case ESC_v:
2575        case ESC_V:
2576        switch(item)
2577          {
2578          case 0x0a:
2579          case 0x0b:
2580          case 0x0c:
2581          case 0x0d:
2582          case 0x85:
2583          case 0x2028:
2584          case 0x2029:
2585          return -next != ESC_v;
2586          default:
2587          return -next == ESC_v;
2588          }
2589    
2590      default:      default:
2591      return FALSE;      return FALSE;
2592      }      }
2593    
2594    case OP_DIGIT:    case OP_DIGIT:
2595    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2596             next == -ESC_h || next == -ESC_v;
2597    
2598    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2599    return next == -ESC_d;    return next == -ESC_d;
# Line 1973  switch(op_code) Line 2602  switch(op_code)
2602    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2603    
2604    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2605    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2606    
2607      case OP_HSPACE:
2608      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2609    
2610      case OP_NOT_HSPACE:
2611      return next == -ESC_h;
2612    
2613      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2614      case OP_VSPACE:
2615      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2616    
2617      case OP_NOT_VSPACE:
2618      return next == -ESC_v;
2619    
2620    case OP_WORDCHAR:    case OP_WORDCHAR:
2621    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2622    
2623    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2624    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2049  uschar classbits[32]; Line 2691  uschar classbits[32];
2691  BOOL class_utf8;  BOOL class_utf8;
2692  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
2693  uschar *class_utf8data;  uschar *class_utf8data;
2694    uschar *class_utf8data_base;
2695  uschar utf8_char[6];  uschar utf8_char[6];
2696  #else  #else
2697  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
2698  uschar *utf8_char = NULL;  uschar *utf8_char = NULL;
2699  #endif  #endif
2700    
2701  #ifdef DEBUG  #ifdef PCRE_DEBUG
2702  if (lengthptr != NULL) DPRINTF((">> start branch\n"));  if (lengthptr != NULL) DPRINTF((">> start branch\n"));
2703  #endif  #endif
2704    
# Line 2088  req_caseopt = ((options & PCRE_CASELESS) Line 2731  req_caseopt = ((options & PCRE_CASELESS)
2731  for (;; ptr++)  for (;; ptr++)
2732    {    {
2733    BOOL negate_class;    BOOL negate_class;
2734      BOOL should_flip_negation;
2735    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2736    BOOL is_quantifier;    BOOL is_quantifier;
2737    BOOL is_recurse;    BOOL is_recurse;
2738      BOOL reset_bracount;
2739    int class_charcount;    int class_charcount;
2740    int class_lastchar;    int class_lastchar;
2741    int newoptions;    int newoptions;
2742    int recno;    int recno;
2743      int refsign;
2744    int skipbytes;    int skipbytes;
2745    int subreqbyte;    int subreqbyte;
2746    int subfirstbyte;    int subfirstbyte;
# Line 2111  for (;; ptr++) Line 2757  for (;; ptr++)
2757    
2758    if (lengthptr != NULL)    if (lengthptr != NULL)
2759      {      {
2760  #ifdef DEBUG  #ifdef PCRE_DEBUG
2761      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
2762  #endif  #endif
2763      if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */      if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */
# Line 2127  for (;; ptr++) Line 2773  for (;; ptr++)
2773      */      */
2774    
2775      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2776    
2777        /* Paranoid check for integer overflow */
2778    
2779        if (OFLOW_MAX - *lengthptr < code - last_code)
2780          {
2781          *errorcodeptr = ERR20;
2782          goto FAILED;
2783          }
2784    
2785      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2786      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));
2787    
# Line 2164  for (;; ptr++) Line 2819  for (;; ptr++)
2819    
2820    if (inescq && c != 0)    if (inescq && c != 0)
2821      {      {
2822      if (c == '\\' && ptr[1] == 'E')      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
2823        {        {
2824        inescq = FALSE;        inescq = FALSE;
2825        ptr++;        ptr++;
# Line 2190  for (;; ptr++) Line 2845  for (;; ptr++)
2845    /* Fill in length of a previous callout, except when the next thing is    /* Fill in length of a previous callout, except when the next thing is
2846    a quantifier. */    a quantifier. */
2847    
2848    is_quantifier = c == '*' || c == '+' || c == '?' ||    is_quantifier =
2849      (c == '{' && is_counted_repeat(ptr+1));      c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK ||
2850        (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1));
2851    
2852    if (!is_quantifier && previous_callout != NULL &&    if (!is_quantifier && previous_callout != NULL &&
2853         after_manual_callout-- <= 0)         after_manual_callout-- <= 0)
# Line 2206  for (;; ptr++) Line 2862  for (;; ptr++)
2862    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
2863      {      {
2864      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if ((cd->ctypes[c] & ctype_space) != 0) continue;
2865      if (c == '#')      if (c == CHAR_NUMBER_SIGN)
2866        {        {
2867        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2868          {          {
# Line 2231  for (;; ptr++) Line 2887  for (;; ptr++)
2887      {      {
2888      /* ===================================================================*/      /* ===================================================================*/
2889      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
2890      case '|':                      /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
2891      case ')':      case CHAR_RIGHT_PARENTHESIS:
2892      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
2893      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
2894      *codeptr = code;      *codeptr = code;
2895      *ptrptr = ptr;      *ptrptr = ptr;
2896      if (lengthptr != NULL)      if (lengthptr != NULL)
2897        {        {
2898          if (OFLOW_MAX - *lengthptr < code - last_code)
2899            {
2900            *errorcodeptr = ERR20;
2901            goto FAILED;
2902            }
2903        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2904        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2905        }        }
# Line 2249  for (;; ptr++) Line 2910  for (;; ptr++)
2910      /* Handle single-character metacharacters. In multiline mode, ^ disables      /* Handle single-character metacharacters. In multiline mode, ^ disables
2911      the setting of any following char as a first character. */      the setting of any following char as a first character. */
2912    
2913      case '^':      case CHAR_CIRCUMFLEX_ACCENT:
2914      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
2915        {        {
2916        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
# Line 2258  for (;; ptr++) Line 2919  for (;; ptr++)
2919      *code++ = OP_CIRC;      *code++ = OP_CIRC;
2920      break;      break;
2921    
2922      case '$':      case CHAR_DOLLAR_SIGN:
2923      previous = NULL;      previous = NULL;
2924      *code++ = OP_DOLL;      *code++ = OP_DOLL;
2925      break;      break;
# Line 2266  for (;; ptr++) Line 2927  for (;; ptr++)
2927      /* There can never be a first char if '.' is first, whatever happens about      /* There can never be a first char if '.' is first, whatever happens about
2928      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqbyte doesn't change either. */
2929    
2930      case '.':      case CHAR_DOT:
2931      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
2932      zerofirstbyte = firstbyte;      zerofirstbyte = firstbyte;
2933      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
2934      previous = code;      previous = code;
2935      *code++ = OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
2936      break;      break;
2937    
2938    
# Line 2286  for (;; ptr++) Line 2947  for (;; ptr++)
2947      opcode is compiled. It may optionally have a bit map for characters < 256,      opcode is compiled. It may optionally have a bit map for characters < 256,
2948      but those above are are explicitly listed afterwards. A flag byte tells      but those above are are explicitly listed afterwards. A flag byte tells
2949      whether the bitmap is present, and whether this is a negated class or not.      whether the bitmap is present, and whether this is a negated class or not.
     */  
2950    
2951      case '[':      In JavaScript compatibility mode, an isolated ']' causes an error. In
2952        default (Perl) mode, it is treated as a data character. */
2953    
2954        case CHAR_RIGHT_SQUARE_BRACKET:
2955        if ((cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
2956          {
2957          *errorcodeptr = ERR64;
2958          goto FAILED;
2959          }
2960        goto NORMAL_CHAR;
2961    
2962        case CHAR_LEFT_SQUARE_BRACKET:
2963      previous = code;      previous = code;
2964    
2965      /* PCRE supports POSIX class stuff inside a class. Perl gives an error if      /* PCRE supports POSIX class stuff inside a class. Perl gives an error if
2966      they are encountered at the top level, so we'll do that too. */      they are encountered at the top level, so we'll do that too. */
2967    
2968      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2969          check_posix_syntax(ptr, &tempptr, cd))           ptr[1] == CHAR_EQUALS_SIGN) &&
2970            check_posix_syntax(ptr, &tempptr))
2971        {        {
2972        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == CHAR_COLON)? ERR13 : ERR31;
2973        goto FAILED;        goto FAILED;
2974        }        }
2975    
2976      /* If the first character is '^', set the negation flag and skip it. */      /* If the first character is '^', set the negation flag and skip it. Also,
2977        if the first few characters (either before or after ^) are \Q\E or \E we
2978        skip them too. This makes for compatibility with Perl. */
2979    
2980      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2981        for (;;)
2982        {        {
       negate_class = TRUE;  
2983        c = *(++ptr);        c = *(++ptr);
2984          if (c == CHAR_BACKSLASH)
2985            {
2986            if (ptr[1] == CHAR_E)
2987              ptr++;
2988            else if (strncmp((const char *)ptr+1,
2989                              STR_Q STR_BACKSLASH STR_E, 3) == 0)
2990              ptr += 3;
2991            else
2992              break;
2993            }
2994          else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT)
2995            negate_class = TRUE;
2996          else break;
2997        }        }
2998      else  
2999        /* Empty classes are allowed in JavaScript compatibility mode. Otherwise,
3000        an initial ']' is taken as a data character -- the code below handles
3001        that. In JS mode, [] must always fail, so generate OP_FAIL, whereas
3002        [^] must match any character, so generate OP_ALLANY. */
3003    
3004        if (c == CHAR_RIGHT_SQUARE_BRACKET &&
3005            (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
3006        {        {
3007        negate_class = FALSE;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
3008          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
3009          zerofirstbyte = firstbyte;
3010          break;
3011        }        }
3012    
3013        /* If a class contains a negative special such as \S, we need to flip the
3014        negation flag at the end, so that support for characters > 255 works
3015        correctly (they are all included in the class). */
3016    
3017        should_flip_negation = FALSE;
3018    
3019      /* Keep a count of chars with values < 256 so that we can optimize the case      /* Keep a count of chars with values < 256 so that we can optimize the case
3020      of just a single character (as long as it's < 256). However, For higher      of just a single character (as long as it's < 256). However, For higher
3021      valued UTF-8 characters, we don't yet do any optimization. */      valued UTF-8 characters, we don't yet do any optimization. */
# Line 2330  for (;; ptr++) Line 3033  for (;; ptr++)
3033  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3034      class_utf8 = FALSE;                       /* No chars >= 256 */      class_utf8 = FALSE;                       /* No chars >= 256 */
3035      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */
3036        class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */
3037  #endif  #endif
3038    
3039      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
# Line 2345  for (;; ptr++) Line 3049  for (;; ptr++)
3049          {                           /* Braces are required because the */          {                           /* Braces are required because the */
3050          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
3051          }          }
3052    
3053          /* In the pre-compile phase, accumulate the length of any UTF-8 extra
3054          data and reset the pointer. This is so that very large classes that
3055          contain a zillion UTF-8 characters no longer overwrite the work space
3056          (which is on the stack). */
3057    
3058          if (lengthptr != NULL)
3059            {
3060            *lengthptr += class_utf8data - class_utf8data_base;
3061            class_utf8data = class_utf8data_base;
3062            }
3063    
3064  #endif  #endif
3065    
3066        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
3067    
3068        if (inescq)        if (inescq)
3069          {          {
3070          if (c == '\\' && ptr[1] == 'E')     /* If we are at \E */          if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)  /* If we are at \E */
3071            {            {
3072            inescq = FALSE;                   /* Reset literal state */            inescq = FALSE;                   /* Reset literal state */
3073            ptr++;                            /* Skip the 'E' */            ptr++;                            /* Skip the 'E' */
# Line 2366  for (;; ptr++) Line 3082  for (;; ptr++)
3082        [.ch.] and [=ch=] ("collating elements") and fault them, as Perl        [.ch.] and [=ch=] ("collating elements") and fault them, as Perl
3083        5.6 and 5.8 do. */        5.6 and 5.8 do. */
3084    
3085        if (c == '[' &&        if (c == CHAR_LEFT_SQUARE_BRACKET &&
3086            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
3087            check_posix_syntax(ptr, &tempptr, cd))             ptr[1] == CHAR_EQUALS_SIGN) && check_posix_syntax(ptr, &tempptr))
3088          {          {
3089          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
3090          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
3091          register const uschar *cbits = cd->cbits;          register const uschar *cbits = cd->cbits;
3092          uschar pbits[32];          uschar pbits[32];
3093    
3094          if (ptr[1] != ':')          if (ptr[1] != CHAR_COLON)
3095            {            {
3096            *errorcodeptr = ERR31;            *errorcodeptr = ERR31;
3097            goto FAILED;            goto FAILED;
3098            }            }
3099    
3100          ptr += 2;          ptr += 2;
3101          if (*ptr == '^')          if (*ptr == CHAR_CIRCUMFLEX_ACCENT)
3102            {            {
3103            local_negate = TRUE;            local_negate = TRUE;
3104              should_flip_negation = TRUE;  /* Note negative special */
3105            ptr++;            ptr++;
3106            }            }
3107    
# Line 2451  for (;; ptr++) Line 3168  for (;; ptr++)
3168        of the specials, which just set a flag. The sequence \b is a special        of the specials, which just set a flag. The sequence \b is a special
3169        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
3170        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
3171        to or into the one we are building. We assume they have more than one        to 'or' into the one we are building. We assume they have more than one
3172        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
3173    
3174        if (c == '\\')        if (c == CHAR_BACKSLASH)
3175          {          {
3176          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3177          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
3178    
3179          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = CHAR_BS;       /* \b is backspace in a class */
3180          else if (-c == ESC_X) c = 'X';   /* \X is literal X in a class */          else if (-c == ESC_X) c = CHAR_X;   /* \X is literal X in a class */
3181          else if (-c == ESC_R) c = 'R';   /* \R is literal R in a class */          else if (-c == ESC_R) c = CHAR_R;   /* \R is literal R in a class */
3182          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
3183            {            {
3184            if (ptr[1] == '\\' && ptr[2] == 'E')            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
3185              {              {
3186              ptr += 2; /* avoid empty string */              ptr += 2; /* avoid empty string */
3187              }              }
3188            else inescq = TRUE;            else inescq = TRUE;
3189            continue;            continue;
3190            }            }
3191            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
3192    
3193          if (c < 0)          if (c < 0)
3194            {            {
# Line 2486  for (;; ptr++) Line 3204  for (;; ptr++)
3204              continue;              continue;
3205    
3206              case ESC_D:              case ESC_D:
3207                should_flip_negation = TRUE;
3208              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
3209              continue;              continue;
3210    
# Line 2494  for (;; ptr++) Line 3213  for (;; ptr++)
3213              continue;              continue;
3214    
3215              case ESC_W:              case ESC_W:
3216                should_flip_negation = TRUE;
3217              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
3218              continue;              continue;
3219    
# Line 2503  for (;; ptr++) Line 3223  for (;; ptr++)
3223              continue;              continue;
3224    
3225              case ESC_S:              case ESC_S:
3226                should_flip_negation = TRUE;
3227              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
3228              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
3229              continue;              continue;
3230    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
3231              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
3232              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
3233              }              }
# Line 2519  for (;; ptr++) Line 3237  for (;; ptr++)
3237            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
3238                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
3239    
3240              /* We need to deal with \H, \h, \V, and \v in both phases because
3241              they use extra memory. */
3242    
3243              if (-c == ESC_h)
3244                {
3245                SETBIT(classbits, 0x09); /* VT */
3246                SETBIT(classbits, 0x20); /* SPACE */
3247                SETBIT(classbits, 0xa0); /* NSBP */
3248    #ifdef SUPPORT_UTF8
3249                if (utf8)
3250                  {
3251                  class_utf8 = TRUE;
3252                  *class_utf8data++ = XCL_SINGLE;
3253                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
3254                  *class_utf8data++ = XCL_SINGLE;
3255                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
3256                  *class_utf8data++ = XCL_RANGE;
3257                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
3258                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
3259                  *class_utf8data++ = XCL_SINGLE;
3260                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
3261                  *class_utf8data++ = XCL_SINGLE;
3262                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
3263                  *class_utf8data++ = XCL_SINGLE;
3264                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
3265                  }
3266    #endif
3267                continue;
3268                }
3269    
3270              if (-c == ESC_H)
3271                {
3272                for (c = 0; c < 32; c++)
3273                  {
3274                  int x = 0xff;
3275                  switch (c)
3276                    {
3277                    case 0x09/8: x ^= 1 << (0x09%8); break;
3278                    case 0x20/8: x ^= 1 << (0x20%8); break;
3279                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
3280                    default: break;
3281                    }
3282                  classbits[c] |= x;
3283                  }
3284    
3285    #ifdef SUPPORT_UTF8
3286                if (utf8)
3287                  {
3288                  class_utf8 = TRUE;
3289                  *class_utf8data++ = XCL_RANGE;
3290                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
3291                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
3292                  *class_utf8data++ = XCL_RANGE;
3293                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
3294                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
3295                  *class_utf8data++ = XCL_RANGE;
3296                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
3297                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
3298                  *class_utf8data++ = XCL_RANGE;
3299                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
3300                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
3301                  *class_utf8data++ = XCL_RANGE;
3302                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
3303                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
3304                  *class_utf8data++ = XCL_RANGE;
3305                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
3306                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
3307                  *class_utf8data++ = XCL_RANGE;
3308                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
3309                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
3310                  }
3311    #endif
3312                continue;
3313                }
3314    
3315              if (-c == ESC_v)
3316                {
3317                SETBIT(classbits, 0x0a); /* LF */
3318                SETBIT(classbits, 0x0b); /* VT */
3319                SETBIT(classbits, 0x0c); /* FF */
3320                SETBIT(classbits, 0x0d); /* CR */
3321                SETBIT(classbits, 0x85); /* NEL */
3322    #ifdef SUPPORT_UTF8
3323                if (utf8)
3324                  {
3325                  class_utf8 = TRUE;
3326                  *class_utf8data++ = XCL_RANGE;
3327                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
3328                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3329                  }
3330    #endif
3331                continue;
3332                }
3333    
3334              if (-c == ESC_V)
3335                {
3336                for (c = 0; c < 32; c++)
3337                  {
3338                  int x = 0xff;
3339                  switch (c)
3340                    {
3341                    case 0x0a/8: x ^= 1 << (0x0a%8);
3342                                 x ^= 1 << (0x0b%8);
3343                                 x ^= 1 << (0x0c%8);
3344                                 x ^= 1 << (0x0d%8);
3345                                 break;
3346                    case 0x85/8: x ^= 1 << (0x85%8); break;
3347                    default: break;
3348                    }
3349                  classbits[c] |= x;
3350                  }
3351    
3352    #ifdef SUPPORT_UTF8
3353                if (utf8)
3354                  {
3355                  class_utf8 = TRUE;
3356                  *class_utf8data++ = XCL_RANGE;
3357                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
3358                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
3359                  *class_utf8data++ = XCL_RANGE;
3360                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3361                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
3362                  }
3363    #endif
3364                continue;
3365                }
3366    
3367            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
3368    
3369  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2562  for (;; ptr++) Line 3407  for (;; ptr++)
3407        entirely. The code for handling \Q and \E is messy. */        entirely. The code for handling \Q and \E is messy. */
3408    
3409        CHECK_RANGE:        CHECK_RANGE:
3410        while (ptr[1] == '\\' && ptr[2] == 'E')        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
3411          {          {
3412          inescq = FALSE;          inescq = FALSE;
3413          ptr += 2;          ptr += 2;
# Line 2570  for (;; ptr++) Line 3415  for (;; ptr++)
3415    
3416        oldptr = ptr;        oldptr = ptr;
3417    
3418        if (!inescq && ptr[1] == '-')        /* Remember \r or \n */
3419    
3420          if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
3421    
3422          /* Check for range */
3423    
3424          if (!inescq && ptr[1] == CHAR_MINUS)
3425          {          {
3426          int d;          int d;
3427          ptr += 2;          ptr += 2;
3428          while (*ptr == '\\' && ptr[1] == 'E') ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
3429    
3430          /* If we hit \Q (not followed by \E) at this point, go into escaped          /* If we hit \Q (not followed by \E) at this point, go into escaped
3431          mode. */          mode. */
3432    
3433          while (*ptr == '\\' && ptr[1] == 'Q')          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_Q)
3434            {            {
3435            ptr += 2;            ptr += 2;
3436            if (*ptr == '\\' && ptr[1] == 'E') { ptr += 2; continue; }            if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3437                { ptr += 2; continue; }
3438            inescq = TRUE;            inescq = TRUE;
3439            break;            break;
3440            }            }
3441    
3442          if (*ptr == 0 || (!inescq && *ptr == ']'))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
3443            {            {
3444            ptr = oldptr;            ptr = oldptr;
3445            goto LONE_SINGLE_CHARACTER;            goto LONE_SINGLE_CHARACTER;
# Line 2606  for (;; ptr++) Line 3458  for (;; ptr++)
3458          not any of the other escapes. Perl 5.6 treats a hyphen as a literal          not any of the other escapes. Perl 5.6 treats a hyphen as a literal
3459          in such circumstances. */          in such circumstances. */
3460    
3461          if (!inescq && d == '\\')          if (!inescq && d == CHAR_BACKSLASH)
3462            {            {
3463            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3464            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3465    
3466            /* \b is backslash; \X is literal X; \R is literal R; any other            /* \b is backspace; \X is literal X; \R is literal R; any other
3467            special means the '-' was literal */            special means the '-' was literal */
3468    
3469            if (d < 0)            if (d < 0)
3470              {              {
3471              if (d == -ESC_b) d = '\b';              if (d == -ESC_b) d = CHAR_BS;
3472              else if (d == -ESC_X) d = 'X';              else if (d == -ESC_X) d = CHAR_X;
3473              else if (d == -ESC_R) d = 'R'; else              else if (d == -ESC_R) d = CHAR_R; else
3474                {                {
3475                ptr = oldptr;                ptr = oldptr;
3476                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
# Line 2637  for (;; ptr++) Line 3489  for (;; ptr++)
3489    
3490          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3491    
3492            /* Remember \r or \n */
3493    
3494            if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
3495    
3496          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless
3497          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3498          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2659  for (;; ptr++) Line 3515  for (;; ptr++)
3515              unsigned int origd = d;              unsigned int origd = d;
3516              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3517                {                {
3518                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3519                      ocd <= (unsigned int)d)
3520                    continue;                          /* Skip embedded ranges */
3521    
3522                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3523                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3524                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3525                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3526                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3527                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3528                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3529                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3530                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3531                  d = ocd;                  d = ocd;
3532                  continue;                  continue;
# Line 2751  for (;; ptr++) Line 3611  for (;; ptr++)
3611          if ((options & PCRE_CASELESS) != 0)          if ((options & PCRE_CASELESS) != 0)
3612            {            {
3613            unsigned int othercase;            unsigned int othercase;
3614            if ((othercase = _pcre_ucp_othercase(c)) != NOTACHAR)            if ((othercase = UCD_OTHERCASE(c)) != c)
3615              {              {
3616              *class_utf8data++ = XCL_SINGLE;              *class_utf8data++ = XCL_SINGLE;
3617              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);
# Line 2778  for (;; ptr++) Line 3638  for (;; ptr++)
3638    
3639      /* Loop until ']' reached. This "while" is the end of the "do" above. */      /* Loop until ']' reached. This "while" is the end of the "do" above. */
3640    
3641      while ((c = *(++ptr)) != 0 && (c != ']' || inescq));      while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
3642    
3643      if (c == 0)                          /* Missing terminating ']' */      if (c == 0)                          /* Missing terminating ']' */
3644        {        {
# Line 2786  for (;; ptr++) Line 3646  for (;; ptr++)
3646        goto FAILED;        goto FAILED;
3647        }        }
3648    
3649    
3650    /* This code has been disabled because it would mean that \s counts as
3651    an explicit \r or \n reference, and that's not really what is wanted. Now
3652    we set the flag only if there is a literal "\r" or "\n" in the class. */
3653    
3654    #if 0
3655        /* Remember whether \r or \n are in this class */
3656    
3657        if (negate_class)
3658          {
3659          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3660          }
3661        else
3662          {
3663          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3664          }
3665    #endif
3666    
3667    
3668      /* If class_charcount is 1, we saw precisely one character whose value is      /* If class_charcount is 1, we saw precisely one character whose value is
3669      less than 256. In non-UTF-8 mode we can always optimize. In UTF-8 mode, we      less than 256. As long as there were no characters >= 128 and there was no
3670      can optimize the negative case only if there were no characters >= 128      use of \p or \P, in other words, no use of any XCLASS features, we can
3671      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3672      single-bytes only. This is an historical hangover. Maybe one day we can  
3673      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3674        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3675        operate on single-bytes only. This is an historical hangover. Maybe one day
3676        we can tidy these opcodes to handle multi-byte characters.
3677    
3678      The optimization throws away the bit map. We turn the item into a      The optimization throws away the bit map. We turn the item into a
3679      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note
# Line 2801  for (;; ptr++) Line 3683  for (;; ptr++)
3683      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3684    
3685  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3686      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3687            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3688  #else  #else
3689      if (class_charcount == 1)      if (class_charcount == 1)
3690  #endif  #endif
# Line 2847  for (;; ptr++) Line 3727  for (;; ptr++)
3727      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3728    
3729      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3730      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3731      we can omit the bitmap in the actual compiled code. */      such as \S in the class, because in that case all characters > 255 are in
3732        the class, so any that were explicitly given as well can be ignored. If
3733        (when there are explicit characters > 255 that must be listed) there are no
3734        characters < 256, we can omit the bitmap in the actual compiled code. */
3735    
3736  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3737      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3738        {        {
3739        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3740        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 2877  for (;; ptr++) Line 3760  for (;; ptr++)
3760        }        }
3761  #endif  #endif
3762    
3763      /* If there are no characters > 255, negate the 32-byte map if necessary,      /* If there are no characters > 255, set the opcode to OP_CLASS or
3764      and copy it into the code vector. If this is the first thing in the branch,      OP_NCLASS, depending on whether the whole class was negated and whether
3765      there can be no first char setting, whatever the repeat count. Any reqbyte      there were negative specials such as \S in the class. Then copy the 32-byte
3766      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3767    
3768        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3769      if (negate_class)      if (negate_class)
3770        {        {
       *code++ = OP_NCLASS;  
3771        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3772          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3773        }        }
3774      else      else
3775        {        {
       *code++ = OP_CLASS;  
3776        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3777        }        }
3778      code += 32;      code += 32;
# Line 2901  for (;; ptr++) Line 3783  for (;; ptr++)
3783      /* Various kinds of repeat; '{' is not necessarily a quantifier, but this      /* Various kinds of repeat; '{' is not necessarily a quantifier, but this
3784      has been tested above. */      has been tested above. */
3785    
3786      case '{':      case CHAR_LEFT_CURLY_BRACKET:
3787      if (!is_quantifier) goto NORMAL_CHAR;      if (!is_quantifier) goto NORMAL_CHAR;
3788      ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr);      ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr);
3789      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
3790      goto REPEAT;      goto REPEAT;
3791    
3792      case '*':      case CHAR_ASTERISK:
3793      repeat_min = 0;      repeat_min = 0;
3794      repeat_max = -1;      repeat_max = -1;
3795      goto REPEAT;      goto REPEAT;
3796    
3797      case '+':      case CHAR_PLUS:
3798      repeat_min = 1;      repeat_min = 1;
3799      repeat_max = -1;      repeat_max = -1;
3800      goto REPEAT;      goto REPEAT;
3801    
3802      case '?':      case CHAR_QUESTION_MARK:
3803      repeat_min = 0;      repeat_min = 0;
3804      repeat_max = 1;      repeat_max = 1;
3805    
# Line 2952  for (;; ptr++) Line 3834  for (;; ptr++)
3834      but if PCRE_UNGREEDY is set, it works the other way round. We change the      but if PCRE_UNGREEDY is set, it works the other way round. We change the
3835      repeat type to the non-default. */      repeat type to the non-default. */
3836    
3837      if (ptr[1] == '+')      if (ptr[1] == CHAR_PLUS)
3838        {        {
3839        repeat_type = 0;                  /* Force greedy */        repeat_type = 0;                  /* Force greedy */
3840        possessive_quantifier = TRUE;        possessive_quantifier = TRUE;
3841        ptr++;        ptr++;
3842        }        }
3843      else if (ptr[1] == '?')      else if (ptr[1] == CHAR_QUESTION_MARK)
3844        {        {
3845        repeat_type = greedy_non_default;        repeat_type = greedy_non_default;
3846        ptr++;        ptr++;
# Line 3073  for (;; ptr++) Line 3955  for (;; ptr++)
3955    
3956        if (repeat_max == 0) goto END_REPEAT;        if (repeat_max == 0) goto END_REPEAT;
3957    
3958          /*--------------------------------------------------------------------*/
3959          /* This code is obsolete from release 8.00; the restriction was finally
3960          removed: */
3961    
3962        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3963        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3964    
3965        if (repeat_max != 1) cd->nopartial = TRUE;        /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */
3966          /*--------------------------------------------------------------------*/
3967    
3968        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3969    
# Line 3223  for (;; ptr++) Line 4110  for (;; ptr++)
4110          goto END_REPEAT;          goto END_REPEAT;
4111          }          }
4112    
4113          /*--------------------------------------------------------------------*/
4114          /* This code is obsolete from release 8.00; the restriction was finally
4115          removed: */
4116    
4117        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
4118        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
4119    
4120        if (repeat_max != 1) cd->nopartial = TRUE;        /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */
4121          /*--------------------------------------------------------------------*/
4122    
4123        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
4124          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3262  for (;; ptr++) Line 4154  for (;; ptr++)
4154          goto FAILED;          goto FAILED;
4155          }          }
4156    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
4157        /* If the maximum repeat count is unlimited, find the end of the bracket        /* If the maximum repeat count is unlimited, find the end of the bracket
4158        by scanning through from the start, and compute the offset back to it        by scanning through from the start, and compute the offset back to it
4159        from the current code pointer. There may be an OP_OPT setting following        from the current code pointer. There may be an OP_OPT setting following
# Line 3292  for (;; ptr++) Line 4176  for (;; ptr++)
4176    
4177        if (repeat_min == 0)        if (repeat_min == 0)
4178          {          {
4179          /* If the maximum is also zero, we just omit the group from the output          /* If the maximum is also zero, we used to just omit the group from the
4180          altogether. */          output altogether, like this:
4181    
4182          if (repeat_max == 0)          ** if (repeat_max == 0)
4183            {          **   {
4184            code = previous;          **   code = previous;
4185            goto END_REPEAT;          **   goto END_REPEAT;
4186            }          **   }
4187    
4188            However, that fails when a group is referenced as a subroutine from
4189            elsewhere in the pattern, so now we stick in OP_SKIPZERO in front of it
4190            so that it is skipped on execution. As we don't have a list of which
4191            groups are referenced, we cannot do this selectively.
4192    
4193            If the maximum is 1 or unlimited, we just have to stick in the BRAZERO
4194            and do no more at this point. However, we do need to adjust any
4195            OP_RECURSE calls inside the group that refer to the group itself or any
4196            internal or forward referenced group, because the offset is from the
4197            start of the whole regex. Temporarily terminate the pattern while doing
4198            this. */
4199    
4200          /* If the maximum is 1 or unlimited, we just have to stick in the          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */
         BRAZERO and do no more at this point. However, we do need to adjust  
         any OP_RECURSE calls inside the group that refer to the group itself or  
         any internal or forward referenced group, because the offset is from  
         the start of the whole regex. Temporarily terminate the pattern while  
         doing this. */  
   
         if (repeat_max <= 1)  
4201            {            {
4202            *code = OP_END;            *code = OP_END;
4203            adjust_recurse(previous, 1, utf8, cd, save_hwm);            adjust_recurse(previous, 1, utf8, cd, save_hwm);
4204            memmove(previous+1, previous, len);            memmove(previous+1, previous, len);
4205            code++;            code++;
4206              if (repeat_max == 0)
4207                {
4208                *previous++ = OP_SKIPZERO;
4209                goto END_REPEAT;
4210                }
4211            *previous++ = OP_BRAZERO + repeat_type;            *previous++ = OP_BRAZERO + repeat_type;
4212            }            }
4213    
# Line 3358  for (;; ptr++) Line 4252  for (;; ptr++)
4252          if (repeat_min > 1)          if (repeat_min > 1)
4253            {            {
4254            /* In the pre-compile phase, we don't actually do the replication. We            /* In the pre-compile phase, we don't actually do the replication. We
4255            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
4256              potential integer overflow. The INT64_OR_DOUBLE type is a 64-bit
4257              integer type when available, otherwise double. */
4258    
4259            if (lengthptr != NULL)            if (lengthptr != NULL)
4260              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
4261                int delta = (repeat_min - 1)*length_prevgroup;
4262                if ((INT64_OR_DOUBLE)(repeat_min - 1)*
4263                      (INT64_OR_DOUBLE)length_prevgroup >
4264                        (INT64_OR_DOUBLE)INT_MAX ||
4265                    OFLOW_MAX - *lengthptr < delta)
4266                  {
4267                  *errorcodeptr = ERR20;
4268                  goto FAILED;
4269                  }
4270                *lengthptr += delta;
4271                }
4272    
4273            /* This is compiling for real */            /* This is compiling for real */
4274    
# Line 3399  for (;; ptr++) Line 4306  for (;; ptr++)
4306          /* In the pre-compile phase, we don't actually do the replication. We          /* In the pre-compile phase, we don't actually do the replication. We
4307          just adjust the length as if we had. For each repetition we must add 1          just adjust the length as if we had. For each repetition we must add 1
4308          to the length for BRAZERO and for all but the last repetition we must          to the length for BRAZERO and for all but the last repetition we must
4309          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
4310            paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is
4311            a 64-bit integer type when available, otherwise double. */
4312    
4313          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
4314            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
4315              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
4316                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
4317              if ((INT64_OR_DOUBLE)repeat_max *
4318                    (INT64_OR_DOUBLE)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
4319                      > (INT64_OR_DOUBLE)INT_MAX ||
4320                  OFLOW_MAX - *lengthptr < delta)
4321                {
4322                *errorcodeptr = ERR20;
4323                goto FAILED;
4324                }
4325              *lengthptr += delta;
4326              }
4327    
4328          /* This is compiling for real */          /* This is compiling for real */
4329    
# Line 3473  for (;; ptr++) Line 4393  for (;; ptr++)
4393            uschar *scode = bracode;            uschar *scode = bracode;
4394            do            do
4395              {              {
4396              if (could_be_empty_branch(scode, ketcode, utf8))              if (could_be_empty_branch(scode, ketcode, utf8, cd))
4397                {                {
4398                *bracode += OP_SBRA - OP_BRA;                *bracode += OP_SBRA - OP_BRA;
4399                break;                break;
# Line 3485  for (;; ptr++) Line 4405  for (;; ptr++)
4405          }          }
4406        }        }
4407    
4408        /* If previous is OP_FAIL, it was generated by an empty class [] in
4409        JavaScript mode. The other ways in which OP_FAIL can be generated, that is
4410        by (*FAIL) or (?!) set previous to NULL, which gives a "nothing to repeat"
4411        error above. We can just ignore the repeat in JS case. */
4412    
4413        else if (*previous == OP_FAIL) goto END_REPEAT;
4414    
4415      /* Else there's some kind of shambles */      /* Else there's some kind of shambles */
4416    
4417      else      else
# Line 3509  for (;; ptr++) Line 4436  for (;; ptr++)
4436      if (possessive_quantifier)      if (possessive_quantifier)
4437        {        {
4438        int len;        int len;
4439        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||  
4440            *tempcode == OP_NOTEXACT)        if (*tempcode == OP_TYPEEXACT)
4441            tempcode += _pcre_OP_lengths[*tempcode] +
4442              ((tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP)? 2 : 0);
4443    
4444          else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)
4445            {
4446          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode];
4447    #ifdef SUPPORT_UTF8
4448            if (utf8 && tempcode[-1] >= 0xc0)
4449              tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f];
4450    #endif
4451            }
4452    
4453        len = code - tempcode;        len = code - tempcode;
4454        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4455          {          {
# Line 3530  for (;; ptr++) Line 4468  for (;; ptr++)
4468          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;
4469          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;
4470    
4471            /* Because we are moving code along, we must ensure that any
4472            pending recursive references are updated. */
4473    
4474          default:          default:
4475            *code = OP_END;
4476            adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);
4477          memmove(tempcode + 1+LINK_SIZE, tempcode, len);          memmove(tempcode + 1+LINK_SIZE, tempcode, len);
4478          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
4479          len += 1 + LINK_SIZE;          len += 1 + LINK_SIZE;
# Line 3555  for (;; ptr++) Line 4498  for (;; ptr++)
4498      /* ===================================================================*/      /* ===================================================================*/
4499      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4500      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4501      parenthesis forms. First deal with the specials; all are introduced by ?,      parenthesis forms.  */
     and the appearance of any of them means that this is not a capturing  
     group. */  
4502    
4503      case '(':      case CHAR_LEFT_PARENTHESIS:
4504      newoptions = options;      newoptions = options;
4505      skipbytes = 0;      skipbytes = 0;
4506      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4507      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4508        reset_bracount = FALSE;
4509    
4510        /* First deal with various "verbs" that can be introduced by '*'. */
4511    
4512        if (*(++ptr) == CHAR_ASTERISK && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4513          {
4514          int i, namelen;
4515          const char *vn = verbnames;
4516          const uschar *name = ++ptr;
4517          previous = NULL;
4518          while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
4519          if (*ptr == CHAR_COLON)
4520            {
4521            *errorcodeptr = ERR59;   /* Not supported */
4522            goto FAILED;
4523            }
4524          if (*ptr != CHAR_RIGHT_PARENTHESIS)
4525            {
4526            *errorcodeptr = ERR60;
4527            goto FAILED;
4528            }
4529          namelen = ptr - name;
4530          for (i = 0; i < verbcount; i++)
4531            {
4532            if (namelen == verbs[i].len &&
4533                strncmp((char *)name, vn, namelen) == 0)
4534              {
4535              /* Check for open captures before ACCEPT */
4536    
4537              if (verbs[i].op == OP_ACCEPT)
4538                {
4539                open_capitem *oc;
4540                cd->had_accept = TRUE;
4541                for (oc = cd->open_caps; oc != NULL; oc = oc->next)
4542                  {
4543                  *code++ = OP_CLOSE;
4544                  PUT2INC(code, 0, oc->number);
4545                  }
4546                }
4547              *code++ = verbs[i].op;
4548              break;
4549              }
4550            vn += verbs[i].len + 1;
4551            }
4552          if (i < verbcount) continue;
4553          *errorcodeptr = ERR60;
4554          goto FAILED;
4555          }
4556    
4557      if (*(++ptr) == '?')      /* Deal with the extended parentheses; all are introduced by '?', and the
4558        appearance of any of them means that this is not a capturing group. */
4559    
4560        else if (*ptr == CHAR_QUESTION_MARK)
4561        {        {
4562        int i, set, unset, namelen;        int i, set, unset, namelen;
4563        int *optset;        int *optset;
# Line 3574  for (;; ptr++) Line 4566  for (;; ptr++)
4566    
4567        switch (*(++ptr))        switch (*(++ptr))
4568          {          {
4569          case '#':                 /* Comment; skip to ket */          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */
4570          ptr++;          ptr++;
4571          while (*ptr != 0 && *ptr != ')') ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
4572          if (*ptr == 0)          if (*ptr == 0)
4573            {            {
4574            *errorcodeptr = ERR18;            *errorcodeptr = ERR18;
# Line 3586  for (;; ptr++) Line 4578  for (;; ptr++)
4578    
4579    
4580          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4581          case ':':                 /* Non-capturing bracket */          case CHAR_VERTICAL_LINE:  /* Reset capture count for each branch */
4582            reset_bracount = TRUE;
4583            /* Fall through */
4584    
4585            /* ------------------------------------------------------------ */
4586            case CHAR_COLON:          /* Non-capturing bracket */
4587          bravalue = OP_BRA;          bravalue = OP_BRA;
4588          ptr++;          ptr++;
4589          break;          break;
4590    
4591    
4592          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4593          case '(':          case CHAR_LEFT_PARENTHESIS:
4594          bravalue = OP_COND;       /* Conditional group */          bravalue = OP_COND;       /* Conditional group */
4595    
4596          /* A condition can be an assertion, a number (referring to a numbered          /* A condition can be an assertion, a number (referring to a numbered
# Line 3613  for (;; ptr++) Line 4610  for (;; ptr++)
4610          the switch. This will take control down to where bracketed groups,          the switch. This will take control down to where bracketed groups,
4611          including assertions, are processed. */          including assertions, are processed. */
4612    
4613          if (ptr[1] == '?' && (ptr[2] == '=' || ptr[2] == '!' || ptr[2] == '<'))          if (ptr[1] == CHAR_QUESTION_MARK && (ptr[2] == CHAR_EQUALS_SIGN ||
4614                ptr[2] == CHAR_EXCLAMATION_MARK || ptr[2] == CHAR_LESS_THAN_SIGN))
4615            break;            break;
4616    
4617          /* Most other conditions use OP_CREF (a couple change to OP_RREF          /* Most other conditions use OP_CREF (a couple change to OP_RREF
# Line 3621  for (;; ptr++) Line 4619  for (;; ptr++)
4619    
4620          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
4621          skipbytes = 3;          skipbytes = 3;
4622            refsign = -1;
4623    
4624          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4625    
4626          if (ptr[1] == 'R' && ptr[2] == '&')          if (ptr[1] == CHAR_R && ptr[2] == CHAR_AMPERSAND)
4627            {            {
4628            terminator = -1;            terminator = -1;
4629            ptr += 2;            ptr += 2;
# Line 3634  for (;; ptr++) Line 4633  for (;; ptr++)
4633          /* Check for a test for a named group's having been set, using the Perl          /* Check for a test for a named group's having been set, using the Perl
4634          syntax (?(<name>) or (?('name') */          syntax (?(<name>) or (?('name') */
4635    
4636          else if (ptr[1] == '<')          else if (ptr[1] == CHAR_LESS_THAN_SIGN)
4637            {            {
4638            terminator = '>';            terminator = CHAR_GREATER_THAN_SIGN;
4639            ptr++;            ptr++;
4640            }            }
4641          else if (ptr[1] == '\'')          else if (ptr[1] == CHAR_APOSTROPHE)
4642            {            {
4643            terminator = '\'';            terminator = CHAR_APOSTROPHE;
4644            ptr++;            ptr++;
4645            }            }
4646          else terminator = 0;          else
4647              {
4648              terminator = 0;
4649              if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);
4650              }
4651    
4652          /* We now expect to read a name; any thing else is an error */          /* We now expect to read a name; any thing else is an error */
4653    
# Line 3663  for (;; ptr++) Line 4666  for (;; ptr++)
4666            {            {
4667            if (recno >= 0)            if (recno >= 0)
4668              recno = ((digitab[*ptr] & ctype_digit) != 0)?              recno = ((digitab[*ptr] & ctype_digit) != 0)?
4669                recno * 10 + *ptr - '0' : -1;                recno * 10 + *ptr - CHAR_0 : -1;
4670            ptr++;            ptr++;
4671            }            }
4672          namelen = ptr - name;          namelen = ptr - name;
4673    
4674          if ((terminator > 0 && *ptr++ != terminator) || *ptr++ != ')')          if ((terminator > 0 && *ptr++ != terminator) ||
4675                *ptr++ != CHAR_RIGHT_PARENTHESIS)
4676            {            {
4677            ptr--;      /* Error offset */            ptr--;      /* Error offset */
4678            *errorcodeptr = ERR26;            *errorcodeptr = ERR26;
# Line 3680  for (;; ptr++) Line 4684  for (;; ptr++)
4684          if (lengthptr != NULL) break;          if (lengthptr != NULL) break;
4685    
4686          /* In the real compile we do the work of looking for the actual          /* In the real compile we do the work of looking for the actual
4687          reference. */          reference. If the string started with "+" or "-" we require the rest to
4688            be digits, in which case recno will be set. */
4689    
4690            if (refsign > 0)
4691              {
4692              if (recno <= 0)
4693                {
4694                *errorcodeptr = ERR58;
4695                goto FAILED;
4696                }
4697              recno = (refsign == CHAR_MINUS)?
4698                cd->bracount - recno + 1 : recno +cd->bracount;
4699              if (recno <= 0 || recno > cd->final_bracount)
4700                {
4701                *errorcodeptr = ERR15;
4702                goto FAILED;
4703                }
4704              PUT2(code, 2+LINK_SIZE, recno);
4705              break;
4706              }
4707    
4708            /* Otherwise (did not start with "+" or "-"), start by looking for the
4709            name. If we find a name, add one to the opcode to change OP_CREF or
4710            OP_RREF into OP_NCREF or OP_NRREF. These behave exactly the same,
4711            except they record that the reference was originally to a name. The
4712            information is used to check duplicate names. */
4713    
4714          slot = cd->name_table;          slot = cd->name_table;
4715          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 3695  for (;; ptr++) Line 4724  for (;; ptr++)
4724            {            {
4725            recno = GET2(slot, 0);            recno = GET2(slot, 0);
4726            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4727              code[1+LINK_SIZE]++;
4728            }            }
4729    
4730          /* Search the pattern for a forward reference */          /* Search the pattern for a forward reference */
4731    
4732          else if ((i = find_parens(ptr, cd->bracount, name, namelen,          else if ((i = find_parens(cd, name, namelen,
4733                          (options & PCRE_EXTENDED) != 0)) > 0)                          (options & PCRE_EXTENDED) != 0)) > 0)
4734            {            {
4735            PUT2(code, 2+LINK_SIZE, i);            PUT2(code, 2+LINK_SIZE, i);
4736              code[1+LINK_SIZE]++;
4737            }            }
4738    
4739          /* If terminator == 0 it means that the name followed directly after          /* If terminator == 0 it means that the name followed directly after
# Line 3720  for (;; ptr++) Line 4751  for (;; ptr++)
4751          /* Check for (?(R) for recursion. Allow digits after R to specify a          /* Check for (?(R) for recursion. Allow digits after R to specify a
4752          specific group number. */          specific group number. */
4753    
4754          else if (*name == 'R')          else if (*name == CHAR_R)
4755            {            {
4756            recno = 0;            recno = 0;
4757            for (i = 1; i < namelen; i++)            for (i = 1; i < namelen; i++)
# Line 3730  for (;; ptr++) Line 4761  for (;; ptr++)
4761                *errorcodeptr = ERR15;                *errorcodeptr = ERR15;
4762                goto FAILED;                goto FAILED;
4763                }                }
4764              recno = recno * 10 + name[i] - '0';              recno = recno * 10 + name[i] - CHAR_0;
4765              }              }
4766            if (recno == 0) recno = RREF_ANY;            if (recno == 0) recno = RREF_ANY;
4767            code[1+LINK_SIZE] = OP_RREF;      /* Change test type */            code[1+LINK_SIZE] = OP_RREF;      /* Change test type */
# Line 3740  for (;; ptr++) Line 4771  for (;; ptr++)
4771          /* Similarly, check for the (?(DEFINE) "condition", which is always          /* Similarly, check for the (?(DEFINE) "condition", which is always
4772          false. */          false. */
4773    
4774          else if (namelen == 6 && strncmp((char *)name, "DEFINE", 6) == 0)          else if (namelen == 6 && strncmp((char *)name, STRING_DEFINE, 6) == 0)
4775            {            {
4776            code[1+LINK_SIZE] = OP_DEF;            code[1+LINK_SIZE] = OP_DEF;
4777            skipbytes = 1;            skipbytes = 1;
4778            }            }
4779    
4780          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4781            in the second pass here, so final_bracount is set. */
4782    
4783          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4784            {            {
4785            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4786            }            }
# Line 3764  for (;; ptr++) Line 4796  for (;; ptr++)
4796    
4797    
4798          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4799          case '=':                 /* Positive lookahead */          case CHAR_EQUALS_SIGN:                 /* Positive lookahead */
4800          bravalue = OP_ASSERT;          bravalue = OP_ASSERT;
4801          ptr++;          ptr++;
4802          break;          break;
4803    
4804    
4805          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4806          case '!':                 /* Negative lookahead */          case CHAR_EXCLAMATION_MARK:            /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4807          ptr++;          ptr++;
4808            if (*ptr == CHAR_RIGHT_PARENTHESIS)    /* Optimize (?!) */
4809              {
4810              *code++ = OP_FAIL;
4811              previous = NULL;
4812              continue;
4813              }
4814            bravalue = OP_ASSERT_NOT;
4815          break;          break;
4816    
4817    
4818          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4819          case '<':                 /* Lookbehind or named define */          case CHAR_LESS_THAN_SIGN:              /* Lookbehind or named define */
4820          switch (ptr[1])          switch (ptr[1])
4821            {            {
4822            case '=':               /* Positive lookbehind */            case CHAR_EQUALS_SIGN:               /* Positive lookbehind */
4823            bravalue = OP_ASSERTBACK;            bravalue = OP_ASSERTBACK;
4824            ptr += 2;            ptr += 2;
4825            break;            break;
4826    
4827            case '!':               /* Negative lookbehind */            case CHAR_EXCLAMATION_MARK:          /* Negative lookbehind */
4828            bravalue = OP_ASSERTBACK_NOT;            bravalue = OP_ASSERTBACK_NOT;
4829            ptr += 2;            ptr += 2;
4830            break;            break;
# Line 3801  for (;; ptr++) Line 4839  for (;; ptr++)
4839    
4840    
4841          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4842          case '>':                 /* One-time brackets */          case CHAR_GREATER_THAN_SIGN:           /* One-time brackets */
4843          bravalue = OP_ONCE;          bravalue = OP_ONCE;
4844          ptr++;          ptr++;
4845          break;          break;
4846    
4847    
4848          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4849          case 'C':                 /* Callout - may be followed by digits; */          case CHAR_C:                 /* Callout - may be followed by digits; */
4850          previous_callout = code;  /* Save for later completion */          previous_callout = code;  /* Save for later completion */
4851          after_manual_callout = 1; /* Skip one item before completing */          after_manual_callout = 1; /* Skip one item before completing */
4852          *code++ = OP_CALLOUT;          *code++ = OP_CALLOUT;
4853            {            {
4854            int n = 0;            int n = 0;
4855            while ((digitab[*(++ptr)] & ctype_digit) != 0)            while ((digitab[*(++ptr)] & ctype_digit) != 0)
4856              n = n * 10 + *ptr - '0';              n = n * 10 + *ptr - CHAR_0;
4857            if (*ptr != ')')            if (*ptr != CHAR_RIGHT_PARENTHESIS)
4858              {              {
4859              *errorcodeptr = ERR39;              *errorcodeptr = ERR39;
4860              goto FAILED;              goto FAILED;
# Line 3836  for (;; ptr++) Line 4874  for (;; ptr++)
4874    
4875    
4876          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4877          case 'P':                 /* Python-style named subpattern handling */          case CHAR_P:              /* Python-style named subpattern handling */
4878          if (*(++ptr) == '=' || *ptr == '>')  /* Reference or recursion */          if (*(++ptr) == CHAR_EQUALS_SIGN ||
4879                *ptr == CHAR_GREATER_THAN_SIGN)  /* Reference or recursion */
4880            {            {
4881            is_recurse = *ptr == '>';            is_recurse = *ptr == CHAR_GREATER_THAN_SIGN;
4882            terminator = ')';            terminator = CHAR_RIGHT_PARENTHESIS;
4883            goto NAMED_REF_OR_RECURSE;            goto NAMED_REF_OR_RECURSE;
4884            }            }
4885          else if (*ptr != '<')    /* Test for Python-style definition */          else if (*ptr != CHAR_LESS_THAN_SIGN)  /* Test for Python-style defn */
4886            {            {
4887            *errorcodeptr = ERR41;            *errorcodeptr = ERR41;
4888            goto FAILED;            goto FAILED;
# Line 3853  for (;; ptr++) Line 4892  for (;; ptr++)
4892    
4893          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4894          DEFINE_NAME:    /* Come here from (?< handling */          DEFINE_NAME:    /* Come here from (?< handling */
4895          case '\'':          case CHAR_APOSTROPHE:
4896            {            {
4897            terminator = (*ptr == '<')? '>' : '\'';            terminator = (*ptr == CHAR_LESS_THAN_SIGN)?
4898                CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
4899            name = ++ptr;            name = ++ptr;
4900    
4901            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
# Line 3886  for (;; ptr++) Line 4926  for (;; ptr++)
4926                }                }
4927              }              }
4928    
4929            /* In the real compile, create the entry in the table */            /* In the real compile, create the entry in the table, maintaining
4930              alphabetical order. Duplicate names for different numbers are
4931              permitted only if PCRE_DUPNAMES is set. Duplicate names for the same
4932              number are always OK. (An existing number can be re-used if (?|
4933              appears in the pattern.) In either event, a duplicate name results in
4934              a duplicate entry in the table, even if the number is the same. This
4935              is because the number of names, and hence the table size, is computed
4936              in the pre-compile, and it affects various numbers and pointers which
4937              would all have to be modified, and the compiled code moved down, if
4938              duplicates with the same number were omitted from the table. This
4939              doesn't seem worth the hassle. However, *different* names for the
4940              same number are not permitted. */
4941    
4942            else            else
4943              {              {
4944                BOOL dupname = FALSE;
4945              slot = cd->name_table;              slot = cd->name_table;
4946    
4947              for (i = 0; i < cd->names_found; i++)              for (i = 0; i < cd->names_found; i++)
4948                {                {
4949                int crc = memcmp(name, slot+2, namelen);                int crc = memcmp(name, slot+2, namelen);
# Line 3898  for (;; ptr++) Line 4951  for (;; ptr++)
4951                  {                  {
4952                  if (slot[2+namelen] == 0)                  if (slot[2+namelen] == 0)
4953                    {                    {
4954                    if ((options & PCRE_DUPNAMES) == 0)                    if (GET2(slot, 0) != cd->bracount + 1 &&
4955                          (options & PCRE_DUPNAMES) == 0)
4956                      {                      {
4957                      *errorcodeptr = ERR43;                      *errorcodeptr = ERR43;
4958                      goto FAILED;                      goto FAILED;
4959                      }                      }
4960                      else dupname = TRUE;
4961                    }                    }
4962                  else crc = -1;      /* Current name is substring */                  else crc = -1;      /* Current name is a substring */
4963                  }                  }
4964    
4965                  /* Make space in the table and break the loop for an earlier
4966                  name. For a duplicate or later name, carry on. We do this for
4967                  duplicates so that in the simple case (when ?(| is not used) they
4968                  are in order of their numbers. */
4969    
4970                if (crc < 0)                if (crc < 0)
4971                  {                  {
4972                  memmove(slot + cd->name_entry_size, slot,                  memmove(slot + cd->name_entry_size, slot,
4973                    (cd->names_found - i) * cd->name_entry_size);                    (cd->names_found - i) * cd->name_entry_size);
4974                  break;                  break;
4975                  }                  }
4976    
4977                  /* Continue the loop for a later or duplicate name */
4978    
4979                slot += cd->name_entry_size;                slot += cd->name_entry_size;
4980                }                }
4981    
4982                /* For non-duplicate names, check for a duplicate number before
4983                adding the new name. */
4984    
4985                if (!dupname)
4986                  {
4987                  uschar *cslot = cd->name_table;
4988                  for (i = 0; i < cd->names_found; i++)
4989                    {
4990                    if (cslot != slot)
4991                      {
4992                      if (GET2(cslot, 0) == cd->bracount + 1)
4993                        {
4994                        *errorcodeptr = ERR65;
4995                        goto FAILED;
4996                        }
4997                      }
4998                    else i--;
4999                    cslot += cd->name_entry_size;
5000                    }
5001                  }
5002    
5003              PUT2(slot, 0, cd->bracount + 1);              PUT2(slot, 0, cd->bracount + 1);
5004              memcpy(slot + 2, name, namelen);              memcpy(slot + 2, name, namelen);
5005              slot[2+namelen] = 0;              slot[2+namelen] = 0;
5006              }              }
5007            }            }
5008    
5009          /* In both cases, count the number of names we've encountered. */          /* In both pre-compile and compile, count the number of names we've
5010            encountered. */
5011    
         ptr++;                    /* Move past > or ' */  
5012          cd->names_found++;          cd->names_found++;
5013            ptr++;                    /* Move past > or ' */
5014          goto NUMBERED_GROUP;          goto NUMBERED_GROUP;
5015    
5016    
5017          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
5018          case '&':                 /* Perl recursion/subroutine syntax */          case CHAR_AMPERSAND:            /* Perl recursion/subroutine syntax */
5019          terminator = ')';          terminator = CHAR_RIGHT_PARENTHESIS;
5020          is_recurse = TRUE;          is_recurse = TRUE;
5021          /* Fall through */          /* Fall through */
5022    
5023          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
5024          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
5025          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
5026            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
5027            .NET syntax, and the Oniguruma \g<...> and \g'...' subroutine syntax. */
5028    
5029          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
5030          name = ++ptr;          name = ++ptr;
# Line 3948  for (;; ptr++) Line 5036  for (;; ptr++)
5036    
5037          if (lengthptr != NULL)          if (lengthptr != NULL)
5038            {            {
5039              if (namelen == 0)
5040                {
5041                *errorcodeptr = ERR62;
5042                goto FAILED;
5043                }
5044            if (*ptr != terminator)            if (*ptr != terminator)
5045              {              {
5046              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 3961  for (;; ptr++) Line 5054  for (;; ptr++)
5054            recno = 0;            recno = 0;
5055            }            }
5056    
5057          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
5058            first, and then check that we have reached the end of the name in the
5059            table. That way, if the name that is longer than any in the table,
5060            the comparison will fail without reading beyond the table entry. */
5061    
5062          else          else
5063            {            {
5064            slot = cd->name_table;            slot = cd->name_table;
5065            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
5066              {              {
5067              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
5068                    slot[2+namelen] == 0)
5069                  break;
5070              slot += cd->name_entry_size;              slot += cd->name_entry_size;
5071              }              }
5072    
# Line 3977  for (;; ptr++) Line 5075  for (;; ptr++)
5075              recno = GET2(slot, 0);              recno = GET2(slot, 0);
5076              }              }
5077            else if ((recno =                /* Forward back reference */            else if ((recno =                /* Forward back reference */
5078                      find_parens(ptr, cd->bracount, name, namelen,                      find_parens(cd, name, namelen,
5079                        (options & PCRE_EXTENDED) != 0)) <= 0)                        (options & PCRE_EXTENDED) != 0)) <= 0)
5080              {              {
5081              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
# Line 3993  for (;; ptr++) Line 5091  for (;; ptr++)
5091    
5092    
5093          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
5094          case 'R':                 /* Recursion */          case CHAR_R:              /* Recursion */
5095          ptr++;                    /* Same as (?0)      */          ptr++;                    /* Same as (?0)      */
5096          /* Fall through */          /* Fall through */
5097    
5098    
5099          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
5100          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */          case CHAR_MINUS: case CHAR_PLUS:  /* Recursion or subroutine */
5101          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4:
5102            case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
5103            {            {
5104            const uschar *called;            const uschar *called;
5105              terminator = CHAR_RIGHT_PARENTHESIS;
5106    
5107              /* Come here from the \g<...> and \g'...' code (Oniguruma
5108              compatibility). However, the syntax has been checked to ensure that
5109              the ... are a (signed) number, so that neither ERR63 nor ERR29 will
5110              be called on this path, nor with the jump to OTHER_CHAR_AFTER_QUERY
5111              ever be taken. */
5112    
5113              HANDLE_NUMERICAL_RECURSION:
5114    
5115              if ((refsign = *ptr) == CHAR_PLUS)
5116                {
5117                ptr++;
5118                if ((digitab[*ptr] & ctype_digit) == 0)
5119                  {
5120                  *errorcodeptr = ERR63;
5121                  goto FAILED;
5122                  }
5123                }
5124              else if (refsign == CHAR_MINUS)
5125                {
5126                if ((digitab[ptr[1]] & ctype_digit) == 0)
5127                  goto OTHER_CHAR_AFTER_QUERY;
5128                ptr++;
5129                }
5130    
5131            recno = 0;            recno = 0;
5132            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
5133              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - CHAR_0;
5134            if (*ptr != ')')  
5135              if (*ptr != terminator)
5136              {              {
5137              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
5138              goto FAILED;              goto FAILED;
5139              }              }
5140    
5141              if (refsign == CHAR_MINUS)
5142                {
5143                if (recno == 0)
5144                  {
5145                  *errorcodeptr = ERR58;
5146                  goto FAILED;
5147                  }
5148                recno = cd->bracount - recno + 1;
5149                if (recno <= 0)
5150                  {
5151                  *errorcodeptr = ERR15;
5152                  goto FAILED;
5153                  }
5154                }
5155              else if (refsign == CHAR_PLUS)
5156                {
5157                if (recno == 0)
5158                  {
5159                  *errorcodeptr = ERR58;
5160                  goto FAILED;
5161                  }
5162                recno += cd->bracount;
5163                }
5164    
5165            /* Come here from code above that handles a named recursion */            /* Come here from code above that handles a named recursion */
5166    
5167            HANDLE_RECURSION:            HANDLE_RECURSION:
# Line 4029  for (;; ptr++) Line 5179  for (;; ptr++)
5179            if (lengthptr == NULL)            if (lengthptr == NULL)
5180              {              {
5181              *code = OP_END;              *code = OP_END;
5182              if (recno != 0) called = find_bracket(cd->start_code, utf8, recno);              if (recno != 0)
5183                  called = _pcre_find_bracket(cd->start_code, utf8, recno);
5184    
5185              /* Forward reference */              /* Forward reference */
5186    
5187              if (called == NULL)              if (called == NULL)
5188                {                {
5189                if (find_parens(ptr, cd->bracount, NULL, recno,                if (find_parens(cd, NULL, recno,
5190                     (options & PCRE_EXTENDED) != 0) < 0)                      (options & PCRE_EXTENDED) != 0) < 0)
5191                  {                  {
5192                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
5193                  goto FAILED;                  goto FAILED;
5194                  }                  }
5195    
5196                  /* Fudge the value of "called" so that when it is inserted as an
5197                  offset below, what it actually inserted is the reference number
5198                  of the group. */
5199    
5200                called = cd->start_code + recno;                called = cd->start_code + recno;
5201                PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code);                PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code);
5202                }                }
# Line 4050  for (;; ptr++) Line 5206  for (;; ptr++)
5206              recursion that could loop for ever, and diagnose that case. */              recursion that could loop for ever, and diagnose that case. */
5207    
5208              else if (GET(called, 1) == 0 &&              else if (GET(called, 1) == 0 &&
5209                       could_be_empty(called, code, bcptr, utf8))                       could_be_empty(called, code, bcptr, utf8, cd))
5210                {                {
5211                *errorcodeptr = ERR40;                *errorcodeptr = ERR40;
5212                goto FAILED;                goto FAILED;
# Line 4084  for (;; ptr++) Line 5240  for (;; ptr++)
5240    
5241          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
5242          default:              /* Other characters: check option setting */          default:              /* Other characters: check option setting */
5243            OTHER_CHAR_AFTER_QUERY:
5244          set = unset = 0;          set = unset = 0;
5245          optset = &set;          optset = &set;
5246    
5247          while (*ptr != ')' && *ptr != ':')          while (*ptr != CHAR_RIGHT_PARENTHESIS && *ptr != CHAR_COLON)
5248            {            {
5249            switch (*ptr++)            switch (*ptr++)
5250              {              {
5251              case '-': optset = &unset; break;              case CHAR_MINUS: optset = &unset; break;
5252    
5253              case 'J':    /* Record that it changed in the external options */              case CHAR_J:    /* Record that it changed in the external options */
5254              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
5255              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
5256              break;              break;
5257    
5258              case 'i': *optset |= PCRE_CASELESS; break;              case CHAR_i: *optset |= PCRE_CASELESS; break;
5259              case 'm': *optset |= PCRE_MULTILINE; break;              case CHAR_m: *optset |= PCRE_MULTILINE; break;
5260              case 's': *optset |= PCRE_DOTALL; break;              case CHAR_s: *optset |= PCRE_DOTALL; break;
5261              case 'x': *optset |= PCRE_EXTENDED; break;              case CHAR_x: *optset |= PCRE_EXTENDED; break;
5262              case 'U': *optset |= PCRE_UNGREEDY; break;              case CHAR_U: *optset |= PCRE_UNGREEDY; break;
5263              case 'X': *optset |= PCRE_EXTRA; break;              case CHAR_X: *optset |= PCRE_EXTRA; break;
5264    
5265              default:  *errorcodeptr = ERR12;              default:  *errorcodeptr = ERR12;
5266                        ptr--;    /* Correct the offset */                        ptr--;    /* Correct the offset */
# Line 4134  for (;; ptr++) Line 5291  for (;; ptr++)
5291          both phases.          both phases.
5292    
5293          If we are not at the pattern start, compile code to change the ims          If we are not at the pattern start, compile code to change the ims
5294          options if this setting actually changes any of them. We also pass the          options if this setting actually changes any of them, and reset the
5295          new setting back so that it can be put at the start of any following          greedy defaults and the case value for firstbyte and reqbyte. */
         branches, and when this group ends (if we are in a group), a resetting  
         item can be compiled. */  
5296    
5297          if (*ptr == ')')          if (*ptr == CHAR_RIGHT_PARENTHESIS)
5298            {            {
5299            if (code == cd->start_code + 1 + LINK_SIZE &&            if (code == cd->start_code + 1 + LINK_SIZE &&
5300                 (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE))                 (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE))
5301              {              {
5302              cd->external_options = newoptions;              cd->external_options = newoptions;
             options = newoptions;  
5303              }              }
5304           else            else
5305              {              {
5306