/[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 170 by ph10, Mon Jun 4 11:21:13 2007 UTC revision 503 by ph10, Sun Mar 7 17:35:52 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, -ESC_K,      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    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number"    "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 263  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 299  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 374  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 400  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 418  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 443  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 CHAR_g:
607        if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
608          {
609          c = -ESC_g;
610          break;
611          }
612    
613      case 'g':      /* Handle the Perl-compatible cases */
614      if (ptr[1] == '{')  
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 472  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)   /* Integer overflow */
642          {
643          *errorcodeptr = ERR61;
644          break;
645          }
646    
647      if (c == 0 || (braced && *(++ptr) != '}'))      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 485  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 505  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 526  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 539  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 550  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 560  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 588  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 604  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 678  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 714  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 757  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 798  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 808  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 838  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[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            negate_class = TRUE;
1161            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 == '\\')        if (*ptr == 0) return -1;
1176          if (*ptr == CHAR_BACKSLASH)
1177          {          {
1178          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) goto FAIL_EXIT;
1179          if (*ptr == 'Q') for (;;)          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 905  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    FAIL_EXIT:
1224    *ptrptr = ptr;
1225    return -1;
1226    }
1227    
   if ((*ptr != '<' || ptr[1] == '!' || ptr[1] == '=') &&  
        *ptr != '\'')  
     continue;  
1228    
   count++;  
1229    
1230    if (name == NULL && count == lorn) return count;  
1231    term = *ptr++;  /*************************************************
1232    if (term == '<') term = '>';  *       Find forward referenced subpattern       *
1233    thisname = ptr;  *************************************************/
1234    while (*ptr != term) ptr++;  
1235    if (name != NULL && lorn == ptr - thisname &&  /* This function scans along a pattern's text looking for capturing
1236        strncmp((const char *)name, (const char *)thisname, lorn) == 0)  subpatterns, and counting them. If it finds a named pattern that matches the
1237      return count;  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    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    for (;;)
1268      {
1269      rc = find_parens_sub(&ptr, cd, name, lorn, xmode, &count);
1270      if (rc > 0 || *ptr++ == 0) break;
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 997  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 1013  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 1042  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 1073  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 1086  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 1109  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 1123  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 1149  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 1203  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 1231  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 1240  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 1268  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 1304  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 1333  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 1348  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 1371  for (code = first_significant_code(code Line 1803  for (code = first_significant_code(code
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_ASSERT)
1810        {
1811        do code += GET(code, 1); while (*code == OP_ALT);
1812        c = *code;
1813        continue;
1814        }
1815    
1816    /* Groups with zero repeats can of course be empty; skip them. */    /* Groups with zero repeats can of course be empty; skip them. */
1817    
1818    if (c == OP_BRAZERO || c == OP_BRAMINZERO)    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);      do code += GET(code, 1); while (*code == OP_ALT);
1822      c = *code;      c = *code;
1823      continue;      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        const uschar *scode = cd->start_code + GET(code, 1);
1832        if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
1833        do
1834          {
1835          if (!could_be_empty_branch(scode, endcode, utf8, cd)) return FALSE;
1836          scode += GET(scode, 1);
1837          }
1838        while (*scode == OP_ALT);
1839        continue;
1840        }
1841    
1842    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1843    
1844    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1845      {      {
1846      BOOL empty_branch;      BOOL empty_branch;
1847      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
1848    
1849      /* Scan a closed bracket */      /* If a conditional group has only one branch, there is a second, implied,
1850        empty branch, so just skip over the conditional, because it could be empty.
1851        Otherwise, scan the individual branches of the group. */
1852    
1853      empty_branch = FALSE;      if (c == OP_COND && code[GET(code, 1)] != OP_ALT)
     do  
       {  
       if (!empty_branch && could_be_empty_branch(code, endcode, utf8))  
         empty_branch = TRUE;  
1854        code += GET(code, 1);        code += GET(code, 1);
1855        else
1856          {
1857          empty_branch = FALSE;
1858          do
1859            {
1860            if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))
1861              empty_branch = TRUE;
1862            code += GET(code, 1);
1863            }
1864          while (*code == OP_ALT);
1865          if (!empty_branch) return FALSE;   /* All branches are non-empty */
1866        }        }
1867      while (*code == OP_ALT);  
1868      if (!empty_branch) return FALSE;   /* All branches are non-empty */      c = *code;
     c = *code;  
1869      continue;      continue;
1870      }      }
1871    
# Line 1406  for (code = first_significant_code(code Line 1873  for (code = first_significant_code(code
1873    
1874    switch (c)    switch (c)
1875      {      {
1876      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1877        cannot be represented just by a bit map. This includes negated single
1878        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1879        actual length is stored in the compiled code, so we must update "code"
1880        here. */
1881    
1882  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1883      case OP_XCLASS:      case OP_XCLASS:
1884      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1885      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1886  #endif  #endif
1887    
# Line 1454  for (code = first_significant_code(code Line 1925  for (code = first_significant_code(code
1925      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
1926      case OP_WORDCHAR:      case OP_WORDCHAR:
1927      case OP_ANY:      case OP_ANY:
1928        case OP_ALLANY:
1929      case OP_ANYBYTE:      case OP_ANYBYTE:
1930      case OP_CHAR:      case OP_CHAR:
1931      case OP_CHARNC:      case OP_CHARNC:
# Line 1472  for (code = first_significant_code(code Line 1944  for (code = first_significant_code(code
1944      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1945      return FALSE;      return FALSE;
1946    
1947        /* These are going to continue, as they may be empty, but we have to
1948        fudge the length for the \p and \P cases. */
1949    
1950        case OP_TYPESTAR:
1951        case OP_TYPEMINSTAR:
1952        case OP_TYPEPOSSTAR:
1953        case OP_TYPEQUERY:
1954        case OP_TYPEMINQUERY:
1955        case OP_TYPEPOSQUERY:
1956        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1957        break;
1958    
1959        /* Same for these */
1960    
1961        case OP_TYPEUPTO:
1962        case OP_TYPEMINUPTO:
1963        case OP_TYPEPOSUPTO:
1964        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1965        break;
1966    
1967      /* End of branch */      /* End of branch */
1968    
1969      case OP_KET:      case OP_KET:
# Line 1490  for (code = first_significant_code(code Line 1982  for (code = first_significant_code(code
1982      case OP_QUERY:      case OP_QUERY:
1983      case OP_MINQUERY:      case OP_MINQUERY:
1984      case OP_POSQUERY:      case OP_POSQUERY:
1985        if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];
1986        break;
1987    
1988      case OP_UPTO:      case OP_UPTO:
1989      case OP_MINUPTO:      case OP_MINUPTO:
1990      case OP_POSUPTO:      case OP_POSUPTO:
1991      if (utf8) while ((code[2] & 0xc0) == 0x80) code++;      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];
1992      break;      break;
1993  #endif  #endif
1994    
1995        /* None of the remaining opcodes are required to match a character. */
1996    
1997        default:
1998        break;
1999      }      }
2000    }    }
2001    
# Line 1518  Arguments: Line 2018  Arguments:
2018    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2019    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2020    utf8        TRUE if in UTF-8 mode    utf8        TRUE if in UTF-8 mode
2021      cd          pointers to tables etc
2022    
2023  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2024  */  */
2025    
2026  static BOOL  static BOOL
2027  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,
2028    BOOL utf8)    BOOL utf8, compile_data *cd)
2029  {  {
2030  while (bcptr != NULL && bcptr->current >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2031    {    {
2032    if (!could_be_empty_branch(bcptr->current, endcode, utf8)) return FALSE;    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))
2033        return FALSE;
2034    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2035    }    }
2036  return TRUE;  return TRUE;
# Line 1541  return TRUE; Line 2043  return TRUE;
2043  *************************************************/  *************************************************/
2044    
2045  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
2046  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
2047  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
2048  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
2049    
2050    Originally, this function only recognized a sequence of letters between the
2051    terminators, but it seems that Perl recognizes any sequence of characters,
2052    though of course unknown POSIX names are subsequently rejected. Perl gives an
2053    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
2054    didn't consider this to be a POSIX class. Likewise for [:1234:].
2055    
2056    The problem in trying to be exactly like Perl is in the handling of escapes. We
2057    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
2058    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
2059    below handles the special case of \], but does not try to do any other escape
2060    processing. This makes it different from Perl for cases such as [:l\ower:]
2061    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
2062    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
2063    I think.
2064    
2065  Argument:  Arguments:
2066    ptr      pointer to the initial [    ptr      pointer to the initial [
2067    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
2068    
2069  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
2070  */  */
2071    
2072  static BOOL  static BOOL
2073  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
2074  {  {
2075  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
2076  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2077  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
2078    {    {
2079    *endptr = ptr;    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else
2080    return TRUE;      {
2081        if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2082        if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2083          {
2084          *endptr = ptr;
2085          return TRUE;
2086          }
2087        }
2088    }    }
2089  return FALSE;  return FALSE;
2090  }  }
# Line 1588  Returns: a value representing the na Line 2109  Returns: a value representing the na
2109  static int  static int
2110  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
2111  {  {
2112    const char *pn = posix_names;
2113  register int yield = 0;  register int yield = 0;
2114  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2115    {    {
2116    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2117      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
2118      pn += posix_name_lengths[yield] + 1;
2119    yield++;    yield++;
2120    }    }
2121  return -1;  return -1;
# Line 1607  return -1; Line 2130  return -1;
2130  that is referenced. This means that groups can be replicated for fixed  that is referenced. This means that groups can be replicated for fixed
2131  repetition simply by copying (because the recursion is allowed to refer to  repetition simply by copying (because the recursion is allowed to refer to
2132  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
2133  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
2134  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
2135  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
2136  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
2137  the partially compiled regex must be temporarily terminated with OP_END.  is called, the partially compiled regex must be temporarily terminated with
2138    OP_END.
2139    
2140  This function has been extended with the possibility of forward references for  This function has been extended with the possibility of forward references for
2141  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 1634  adjust_recurse(uschar *group, int adjust Line 2158  adjust_recurse(uschar *group, int adjust
2158    uschar *save_hwm)    uschar *save_hwm)
2159  {  {
2160  uschar *ptr = group;  uschar *ptr = group;
2161    
2162  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
2163    {    {
2164    int offset;    int offset;
# Line 1745  get_othercase_range(unsigned int *cptr, Line 2270  get_othercase_range(unsigned int *cptr,
2270  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2271    
2272  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2273    { if ((othercase = _pcre_ucp_othercase(c)) != NOTACHAR) break; }    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }
2274    
2275  if (c > d) return FALSE;  if (c > d) return FALSE;
2276    
# Line 1754  next = othercase + 1; Line 2279  next = othercase + 1;
2279    
2280  for (++c; c <= d; c++)  for (++c; c <= d; c++)
2281    {    {
2282    if (_pcre_ucp_othercase(c) != next) break;    if (UCD_OTHERCASE(c) != next) break;
2283    next++;    next++;
2284    }    }
2285    
# Line 1800  if ((options & PCRE_EXTENDED) != 0) Line 2325  if ((options & PCRE_EXTENDED) != 0)
2325    for (;;)    for (;;)
2326      {      {
2327      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2328      if (*ptr == '#')      if (*ptr == CHAR_NUMBER_SIGN)
2329        {        {
2330        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2331          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
# Line 1812  if ((options & PCRE_EXTENDED) != 0) Line 2337  if ((options & PCRE_EXTENDED) != 0)
2337  /* 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
2338  value is a character, a negative value is an escape value. */  value is a character, a negative value is an escape value. */
2339    
2340  if (*ptr == '\\')  if (*ptr == CHAR_BACKSLASH)
2341    {    {
2342    int temperrorcode = 0;    int temperrorcode = 0;
2343    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);
# Line 1837  if ((options & PCRE_EXTENDED) != 0) Line 2362  if ((options & PCRE_EXTENDED) != 0)
2362    for (;;)    for (;;)
2363      {      {
2364      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2365      if (*ptr == '#')      if (*ptr == CHAR_NUMBER_SIGN)
2366        {        {
2367        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2368          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
# Line 1848  if ((options & PCRE_EXTENDED) != 0) Line 2373  if ((options & PCRE_EXTENDED) != 0)
2373    
2374  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
2375    
2376  if (*ptr == '*' || *ptr == '?' || strncmp((char *)ptr, "{0,", 3) == 0)  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
2377    return FALSE;    strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
2378        return FALSE;
2379    
2380  /* 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
2381  positive single character match, "item" either contains the character or, if  positive single character match, "item" either contains the character or, if
# Line 1864  if (next >= 0) switch(op_code) Line 2390  if (next >= 0) switch(op_code)
2390    case OP_CHAR:    case OP_CHAR:
2391  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2392    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }
2393    #else
2394      (void)(utf8_char);  /* Keep compiler happy by referencing function argument */
2395  #endif  #endif
2396    return item != next;    return item != next;
2397    
# Line 1882  if (next >= 0) switch(op_code) Line 2410  if (next >= 0) switch(op_code)
2410      unsigned int othercase;      unsigned int othercase;
2411      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
2412  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2413      othercase = _pcre_ucp_othercase((unsigned int)next);      othercase = UCD_OTHERCASE((unsigned int)next);
2414  #else  #else
2415      othercase = NOTACHAR;      othercase = NOTACHAR;
2416  #endif  #endif
# Line 1895  if (next >= 0) switch(op_code) Line 2423  if (next >= 0) switch(op_code)
2423    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, "item" must be a single-byte character. */
2424    
2425    case OP_NOT:    case OP_NOT:
   if (next < 0) return FALSE;  /* Not a character */  
2426    if (item == next) return TRUE;    if (item == next) return TRUE;
2427    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2428  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1904  if (next >= 0) switch(op_code) Line 2431  if (next >= 0) switch(op_code)
2431      unsigned int othercase;      unsigned int othercase;
2432      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
2433  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2434      othercase = _pcre_ucp_othercase(next);      othercase = UCD_OTHERCASE(next);
2435  #else  #else
2436      othercase = NOTACHAR;      othercase = NOTACHAR;
2437  #endif  #endif
# Line 1932  if (next >= 0) switch(op_code) Line 2459  if (next >= 0) switch(op_code)
2459    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2460    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2461    
2462      case OP_HSPACE:
2463      case OP_NOT_HSPACE:
2464      switch(next)
2465        {
2466        case 0x09:
2467        case 0x20:
2468        case 0xa0:
2469        case 0x1680:
2470        case 0x180e:
2471        case 0x2000:
2472        case 0x2001:
2473        case 0x2002:
2474        case 0x2003:
2475        case 0x2004:
2476        case 0x2005:
2477        case 0x2006:
2478        case 0x2007:
2479        case 0x2008:
2480        case 0x2009:
2481        case 0x200A:
2482        case 0x202f:
2483        case 0x205f:
2484        case 0x3000:
2485        return op_code != OP_HSPACE;
2486        default:
2487        return op_code == OP_HSPACE;
2488        }
2489    
2490      case OP_VSPACE:
2491      case OP_NOT_VSPACE:
2492      switch(next)
2493        {
2494        case 0x0a:
2495        case 0x0b:
2496        case 0x0c:
2497        case 0x0d:
2498        case 0x85:
2499        case 0x2028:
2500        case 0x2029:
2501        return op_code != OP_VSPACE;
2502        default:
2503        return op_code == OP_VSPACE;
2504        }
2505    
2506    default:    default:
2507    return FALSE;    return FALSE;
2508    }    }
# Line 1966  switch(op_code) Line 2537  switch(op_code)
2537      case ESC_W:      case ESC_W:
2538      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2539    
2540        case ESC_h:
2541        case ESC_H:
2542        switch(item)
2543          {
2544          case 0x09:
2545          case 0x20:
2546          case 0xa0:
2547          case 0x1680:
2548          case 0x180e:
2549          case 0x2000:
2550          case 0x2001:
2551          case 0x2002:
2552          case 0x2003:
2553          case 0x2004:
2554          case 0x2005:
2555          case 0x2006:
2556          case 0x2007:
2557          case 0x2008:
2558          case 0x2009:
2559          case 0x200A:
2560          case 0x202f:
2561          case 0x205f:
2562          case 0x3000:
2563          return -next != ESC_h;
2564          default:
2565          return -next == ESC_h;
2566          }
2567    
2568        case ESC_v:
2569        case ESC_V:
2570        switch(item)
2571          {
2572          case 0x0a:
2573          case 0x0b:
2574          case 0x0c:
2575          case 0x0d:
2576          case 0x85:
2577          case 0x2028:
2578          case 0x2029:
2579          return -next != ESC_v;
2580          default:
2581          return -next == ESC_v;
2582          }
2583    
2584      default:      default:
2585      return FALSE;      return FALSE;
2586      }      }
2587    
2588    case OP_DIGIT:    case OP_DIGIT:
2589    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2590             next == -ESC_h || next == -ESC_v;
2591    
2592    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2593    return next == -ESC_d;    return next == -ESC_d;
# Line 1980  switch(op_code) Line 2596  switch(op_code)
2596    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2597    
2598    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2599    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2600    
2601      case OP_HSPACE:
2602      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2603    
2604      case OP_NOT_HSPACE:
2605      return next == -ESC_h;
2606    
2607      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2608      case OP_VSPACE:
2609      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2610    
2611      case OP_NOT_VSPACE:
2612      return next == -ESC_v;
2613    
2614    case OP_WORDCHAR:    case OP_WORDCHAR:
2615    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2616    
2617    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2618    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2056  uschar classbits[32]; Line 2685  uschar classbits[32];
2685  BOOL class_utf8;  BOOL class_utf8;
2686  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
2687  uschar *class_utf8data;  uschar *class_utf8data;
2688    uschar *class_utf8data_base;
2689  uschar utf8_char[6];  uschar utf8_char[6];
2690  #else  #else
2691  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
2692  uschar *utf8_char = NULL;  uschar *utf8_char = NULL;
2693  #endif  #endif
2694    
2695  #ifdef DEBUG  #ifdef PCRE_DEBUG
2696  if (lengthptr != NULL) DPRINTF((">> start branch\n"));  if (lengthptr != NULL) DPRINTF((">> start branch\n"));
2697  #endif  #endif
2698    
# Line 2095  req_caseopt = ((options & PCRE_CASELESS) Line 2725  req_caseopt = ((options & PCRE_CASELESS)
2725  for (;; ptr++)  for (;; ptr++)
2726    {    {
2727    BOOL negate_class;    BOOL negate_class;
2728      BOOL should_flip_negation;
2729    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2730    BOOL is_quantifier;    BOOL is_quantifier;
2731    BOOL is_recurse;    BOOL is_recurse;
2732      BOOL reset_bracount;
2733    int class_charcount;    int class_charcount;
2734    int class_lastchar;    int class_lastchar;
2735    int newoptions;    int newoptions;
2736    int recno;    int recno;
2737    int refsign;    int refsign;
2738    int skipbytes;    int skipbytes;
2739    int subreqbyte;    int subreqbyte;
2740    int subfirstbyte;    int subfirstbyte;
# Line 2119  for (;; ptr++) Line 2751  for (;; ptr++)
2751    
2752    if (lengthptr != NULL)    if (lengthptr != NULL)
2753      {      {
2754  #ifdef DEBUG  #ifdef PCRE_DEBUG
2755      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
2756  #endif  #endif
2757      if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */      if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */
# Line 2135  for (;; ptr++) Line 2767  for (;; ptr++)
2767      */      */
2768    
2769      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2770    
2771        /* Paranoid check for integer overflow */
2772    
2773        if (OFLOW_MAX - *lengthptr < code - last_code)
2774          {
2775          *errorcodeptr = ERR20;
2776          goto FAILED;
2777          }
2778    
2779      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2780      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));
2781    
# Line 2172  for (;; ptr++) Line 2813  for (;; ptr++)
2813    
2814    if (inescq && c != 0)    if (inescq && c != 0)
2815      {      {
2816      if (c == '\\' && ptr[1] == 'E')      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
2817        {        {
2818        inescq = FALSE;        inescq = FALSE;
2819        ptr++;        ptr++;
# Line 2198  for (;; ptr++) Line 2839  for (;; ptr++)
2839    /* 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
2840    a quantifier. */    a quantifier. */
2841    
2842    is_quantifier = c == '*' || c == '+' || c == '?' ||    is_quantifier =
2843      (c == '{' && is_counted_repeat(ptr+1));      c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK ||
2844        (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1));
2845    
2846    if (!is_quantifier && previous_callout != NULL &&    if (!is_quantifier && previous_callout != NULL &&
2847         after_manual_callout-- <= 0)         after_manual_callout-- <= 0)
# Line 2214  for (;; ptr++) Line 2856  for (;; ptr++)
2856    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
2857      {      {
2858      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if ((cd->ctypes[c] & ctype_space) != 0) continue;
2859      if (c == '#')      if (c == CHAR_NUMBER_SIGN)
2860        {        {
2861        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2862          {          {
# Line 2239  for (;; ptr++) Line 2881  for (;; ptr++)
2881      {      {
2882      /* ===================================================================*/      /* ===================================================================*/
2883      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
2884      case '|':                      /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
2885      case ')':      case CHAR_RIGHT_PARENTHESIS:
2886      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
2887      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
2888      *codeptr = code;      *codeptr = code;
2889      *ptrptr = ptr;      *ptrptr = ptr;
2890      if (lengthptr != NULL)      if (lengthptr != NULL)
2891        {        {
2892          if (OFLOW_MAX - *lengthptr < code - last_code)
2893            {
2894            *errorcodeptr = ERR20;
2895            goto FAILED;
2896            }
2897        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2898        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2899        }        }
# Line 2257  for (;; ptr++) Line 2904  for (;; ptr++)
2904      /* Handle single-character metacharacters. In multiline mode, ^ disables      /* Handle single-character metacharacters. In multiline mode, ^ disables
2905      the setting of any following char as a first character. */      the setting of any following char as a first character. */
2906    
2907      case '^':      case CHAR_CIRCUMFLEX_ACCENT:
2908      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
2909        {        {
2910        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
# Line 2266  for (;; ptr++) Line 2913  for (;; ptr++)
2913      *code++ = OP_CIRC;      *code++ = OP_CIRC;
2914      break;      break;
2915    
2916      case '$':      case CHAR_DOLLAR_SIGN:
2917      previous = NULL;      previous = NULL;
2918      *code++ = OP_DOLL;      *code++ = OP_DOLL;
2919      break;      break;
# Line 2274  for (;; ptr++) Line 2921  for (;; ptr++)
2921      /* 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
2922      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqbyte doesn't change either. */
2923    
2924      case '.':      case CHAR_DOT:
2925      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
2926      zerofirstbyte = firstbyte;      zerofirstbyte = firstbyte;
2927      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
2928      previous = code;      previous = code;
2929      *code++ = OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
2930      break;      break;
2931    
2932    
# Line 2294  for (;; ptr++) Line 2941  for (;; ptr++)
2941      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,
2942      but those above are are explicitly listed afterwards. A flag byte tells      but those above are are explicitly listed afterwards. A flag byte tells
2943      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.
     */  
2944    
2945      case '[':      In JavaScript compatibility mode, an isolated ']' causes an error. In
2946        default (Perl) mode, it is treated as a data character. */
2947    
2948        case CHAR_RIGHT_SQUARE_BRACKET:
2949        if ((cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
2950          {
2951          *errorcodeptr = ERR64;
2952          goto FAILED;
2953          }
2954        goto NORMAL_CHAR;
2955    
2956        case CHAR_LEFT_SQUARE_BRACKET:
2957      previous = code;      previous = code;
2958    
2959      /* 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
2960      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. */
2961    
2962      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2963          check_posix_syntax(ptr, &tempptr, cd))           ptr[1] == CHAR_EQUALS_SIGN) &&
2964            check_posix_syntax(ptr, &tempptr))
2965        {        {
2966        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == CHAR_COLON)? ERR13 : ERR31;
2967        goto FAILED;        goto FAILED;
2968        }        }
2969    
2970      /* 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,
2971        if the first few characters (either before or after ^) are \Q\E or \E we
2972        skip them too. This makes for compatibility with Perl. */
2973    
2974      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2975        for (;;)
2976        {        {
       negate_class = TRUE;  
2977        c = *(++ptr);        c = *(++ptr);
2978          if (c == CHAR_BACKSLASH)
2979            {
2980            if (ptr[1] == CHAR_E)
2981              ptr++;
2982            else if (strncmp((const char *)ptr+1,
2983                              STR_Q STR_BACKSLASH STR_E, 3) == 0)
2984              ptr += 3;
2985            else
2986              break;
2987            }
2988          else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT)
2989            negate_class = TRUE;
2990          else break;
2991        }        }
2992      else  
2993        /* Empty classes are allowed in JavaScript compatibility mode. Otherwise,
2994        an initial ']' is taken as a data character -- the code below handles
2995        that. In JS mode, [] must always fail, so generate OP_FAIL, whereas
2996        [^] must match any character, so generate OP_ALLANY. */
2997    
2998        if (c == CHAR_RIGHT_SQUARE_BRACKET &&
2999            (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
3000        {        {
3001        negate_class = FALSE;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
3002          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
3003          zerofirstbyte = firstbyte;
3004          break;
3005        }        }
3006    
3007        /* If a class contains a negative special such as \S, we need to flip the
3008        negation flag at the end, so that support for characters > 255 works
3009        correctly (they are all included in the class). */
3010    
3011        should_flip_negation = FALSE;
3012    
3013      /* 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
3014      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
3015      valued UTF-8 characters, we don't yet do any optimization. */      valued UTF-8 characters, we don't yet do any optimization. */
# Line 2338  for (;; ptr++) Line 3027  for (;; ptr++)
3027  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3028      class_utf8 = FALSE;                       /* No chars >= 256 */      class_utf8 = FALSE;                       /* No chars >= 256 */
3029      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */
3030        class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */
3031  #endif  #endif
3032    
3033      /* 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 2353  for (;; ptr++) Line 3043  for (;; ptr++)
3043          {                           /* Braces are required because the */          {                           /* Braces are required because the */
3044          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
3045          }          }
3046    
3047          /* In the pre-compile phase, accumulate the length of any UTF-8 extra
3048          data and reset the pointer. This is so that very large classes that
3049          contain a zillion UTF-8 characters no longer overwrite the work space
3050          (which is on the stack). */
3051    
3052          if (lengthptr != NULL)
3053            {
3054            *lengthptr += class_utf8data - class_utf8data_base;
3055            class_utf8data = class_utf8data_base;
3056            }
3057    
3058  #endif  #endif
3059    
3060        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
3061    
3062        if (inescq)        if (inescq)
3063          {          {
3064          if (c == '\\' && ptr[1] == 'E')     /* If we are at \E */          if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)  /* If we are at \E */
3065            {            {
3066            inescq = FALSE;                   /* Reset literal state */            inescq = FALSE;                   /* Reset literal state */
3067            ptr++;                            /* Skip the 'E' */            ptr++;                            /* Skip the 'E' */
# Line 2374  for (;; ptr++) Line 3076  for (;; ptr++)
3076        [.ch.] and [=ch=] ("collating elements") and fault them, as Perl        [.ch.] and [=ch=] ("collating elements") and fault them, as Perl
3077        5.6 and 5.8 do. */        5.6 and 5.8 do. */
3078    
3079        if (c == '[' &&        if (c == CHAR_LEFT_SQUARE_BRACKET &&
3080            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
3081            check_posix_syntax(ptr, &tempptr, cd))             ptr[1] == CHAR_EQUALS_SIGN) && check_posix_syntax(ptr, &tempptr))
3082          {          {
3083          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
3084          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
3085          register const uschar *cbits = cd->cbits;          register const uschar *cbits = cd->cbits;
3086          uschar pbits[32];          uschar pbits[32];
3087    
3088          if (ptr[1] != ':')          if (ptr[1] != CHAR_COLON)
3089            {            {
3090            *errorcodeptr = ERR31;            *errorcodeptr = ERR31;
3091            goto FAILED;            goto FAILED;
3092            }            }
3093    
3094          ptr += 2;          ptr += 2;
3095          if (*ptr == '^')          if (*ptr == CHAR_CIRCUMFLEX_ACCENT)
3096            {            {
3097            local_negate = TRUE;            local_negate = TRUE;
3098              should_flip_negation = TRUE;  /* Note negative special */
3099            ptr++;            ptr++;
3100            }            }
3101    
# Line 2459  for (;; ptr++) Line 3162  for (;; ptr++)
3162        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
3163        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
3164        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
3165        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
3166        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
3167    
3168        if (c == '\\')        if (c == CHAR_BACKSLASH)
3169          {          {
3170          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3171          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
3172    
3173          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 */
3174          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 */
3175          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 */
3176          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
3177            {            {
3178            if (ptr[1] == '\\' && ptr[2] == 'E')            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
3179              {              {
3180              ptr += 2; /* avoid empty string */              ptr += 2; /* avoid empty string */
3181              }              }
3182            else inescq = TRUE;            else inescq = TRUE;
3183            continue;            continue;
3184            }            }
3185            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
3186    
3187          if (c < 0)          if (c < 0)
3188            {            {
# Line 2494  for (;; ptr++) Line 3198  for (;; ptr++)
3198              continue;              continue;
3199    
3200              case ESC_D:              case ESC_D:
3201                should_flip_negation = TRUE;
3202              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
3203              continue;              continue;
3204    
# Line 2502  for (;; ptr++) Line 3207  for (;; ptr++)
3207              continue;              continue;
3208    
3209              case ESC_W:              case ESC_W:
3210                should_flip_negation = TRUE;
3211              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
3212              continue;              continue;
3213    
# Line 2511  for (;; ptr++) Line 3217  for (;; ptr++)
3217              continue;              continue;
3218    
3219              case ESC_S:              case ESC_S:
3220                should_flip_negation = TRUE;
3221              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
3222              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
3223              continue;              continue;
3224    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
3225              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
3226              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
3227              }              }
# Line 2527  for (;; ptr++) Line 3231  for (;; ptr++)
3231            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
3232                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
3233    
3234              /* We need to deal with \H, \h, \V, and \v in both phases because
3235              they use extra memory. */
3236    
3237              if (-c == ESC_h)
3238                {
3239                SETBIT(classbits, 0x09); /* VT */
3240                SETBIT(classbits, 0x20); /* SPACE */
3241                SETBIT(classbits, 0xa0); /* NSBP */
3242    #ifdef SUPPORT_UTF8
3243                if (utf8)
3244                  {
3245                  class_utf8 = TRUE;
3246                  *class_utf8data++ = XCL_SINGLE;
3247                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
3248                  *class_utf8data++ = XCL_SINGLE;
3249                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
3250                  *class_utf8data++ = XCL_RANGE;
3251                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
3252                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
3253                  *class_utf8data++ = XCL_SINGLE;
3254                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
3255                  *class_utf8data++ = XCL_SINGLE;
3256                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
3257                  *class_utf8data++ = XCL_SINGLE;
3258                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
3259                  }
3260    #endif
3261                continue;
3262                }
3263    
3264              if (-c == ESC_H)
3265                {
3266                for (c = 0; c < 32; c++)
3267                  {
3268                  int x = 0xff;
3269                  switch (c)
3270                    {
3271                    case 0x09/8: x ^= 1 << (0x09%8); break;
3272                    case 0x20/8: x ^= 1 << (0x20%8); break;
3273                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
3274                    default: break;
3275                    }
3276                  classbits[c] |= x;
3277                  }
3278    
3279    #ifdef SUPPORT_UTF8
3280                if (utf8)
3281                  {
3282                  class_utf8 = TRUE;
3283                  *class_utf8data++ = XCL_RANGE;
3284                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
3285                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
3286                  *class_utf8data++ = XCL_RANGE;
3287                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
3288                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
3289                  *class_utf8data++ = XCL_RANGE;
3290                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
3291                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
3292                  *class_utf8data++ = XCL_RANGE;
3293                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
3294                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
3295                  *class_utf8data++ = XCL_RANGE;
3296                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
3297                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
3298                  *class_utf8data++ = XCL_RANGE;
3299                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
3300                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
3301                  *class_utf8data++ = XCL_RANGE;
3302                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
3303                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
3304                  }
3305    #endif
3306                continue;
3307                }
3308    
3309              if (-c == ESC_v)
3310                {
3311                SETBIT(classbits, 0x0a); /* LF */
3312                SETBIT(classbits, 0x0b); /* VT */
3313                SETBIT(classbits, 0x0c); /* FF */
3314                SETBIT(classbits, 0x0d); /* CR */
3315                SETBIT(classbits, 0x85); /* NEL */
3316    #ifdef SUPPORT_UTF8
3317                if (utf8)
3318                  {
3319                  class_utf8 = TRUE;
3320                  *class_utf8data++ = XCL_RANGE;
3321                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
3322                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3323                  }
3324    #endif
3325                continue;
3326                }
3327    
3328              if (-c == ESC_V)
3329                {
3330                for (c = 0; c < 32; c++)
3331                  {
3332                  int x = 0xff;
3333                  switch (c)
3334                    {
3335                    case 0x0a/8: x ^= 1 << (0x0a%8);
3336                                 x ^= 1 << (0x0b%8);
3337                                 x ^= 1 << (0x0c%8);
3338                                 x ^= 1 << (0x0d%8);
3339                                 break;
3340                    case 0x85/8: x ^= 1 << (0x85%8); break;
3341                    default: break;
3342                    }
3343                  classbits[c] |= x;
3344                  }
3345    
3346    #ifdef SUPPORT_UTF8
3347                if (utf8)
3348                  {
3349                  class_utf8 = TRUE;
3350                  *class_utf8data++ = XCL_RANGE;
3351                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
3352                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
3353                  *class_utf8data++ = XCL_RANGE;
3354                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3355                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
3356                  }
3357    #endif
3358                continue;
3359                }
3360    
3361            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
3362    
3363  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2570  for (;; ptr++) Line 3401  for (;; ptr++)
3401        entirely. The code for handling \Q and \E is messy. */        entirely. The code for handling \Q and \E is messy. */
3402    
3403        CHECK_RANGE:        CHECK_RANGE:
3404        while (ptr[1] == '\\' && ptr[2] == 'E')        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
3405          {          {
3406          inescq = FALSE;          inescq = FALSE;
3407          ptr += 2;          ptr += 2;
# Line 2578  for (;; ptr++) Line 3409  for (;; ptr++)
3409    
3410        oldptr = ptr;        oldptr = ptr;
3411    
3412        if (!inescq && ptr[1] == '-')        /* Remember \r or \n */
3413    
3414          if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
3415    
3416          /* Check for range */
3417    
3418          if (!inescq && ptr[1] == CHAR_MINUS)
3419          {          {
3420          int d;          int d;
3421          ptr += 2;          ptr += 2;
3422          while (*ptr == '\\' && ptr[1] == 'E') ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
3423    
3424          /* 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
3425          mode. */          mode. */
3426    
3427          while (*ptr == '\\' && ptr[1] == 'Q')          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_Q)
3428            {            {
3429            ptr += 2;            ptr += 2;
3430            if (*ptr == '\\' && ptr[1] == 'E') { ptr += 2; continue; }            if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3431                { ptr += 2; continue; }
3432            inescq = TRUE;            inescq = TRUE;
3433            break;            break;
3434            }            }
3435    
3436          if (*ptr == 0 || (!inescq && *ptr == ']'))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
3437            {            {
3438            ptr = oldptr;            ptr = oldptr;
3439            goto LONE_SINGLE_CHARACTER;            goto LONE_SINGLE_CHARACTER;
# Line 2614  for (;; ptr++) Line 3452  for (;; ptr++)
3452          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
3453          in such circumstances. */          in such circumstances. */
3454    
3455          if (!inescq && d == '\\')          if (!inescq && d == CHAR_BACKSLASH)
3456            {            {
3457            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3458            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3459    
3460            /* \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
3461            special means the '-' was literal */            special means the '-' was literal */
3462    
3463            if (d < 0)            if (d < 0)
3464              {              {
3465              if (d == -ESC_b) d = '\b';              if (d == -ESC_b) d = CHAR_BS;
3466              else if (d == -ESC_X) d = 'X';              else if (d == -ESC_X) d = CHAR_X;
3467              else if (d == -ESC_R) d = 'R'; else              else if (d == -ESC_R) d = CHAR_R; else
3468                {                {
3469                ptr = oldptr;                ptr = oldptr;
3470                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
# Line 2645  for (;; ptr++) Line 3483  for (;; ptr++)
3483    
3484          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3485    
3486            /* Remember \r or \n */
3487    
3488            if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
3489    
3490          /* 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
3491          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3492          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2667  for (;; ptr++) Line 3509  for (;; ptr++)
3509              unsigned int origd = d;              unsigned int origd = d;
3510              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3511                {                {
3512                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3513                      ocd <= (unsigned int)d)
3514                    continue;                          /* Skip embedded ranges */
3515    
3516                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3517                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3518                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3519                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3520                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3521                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3522                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3523                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3524                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3525                  d = ocd;                  d = ocd;
3526                  continue;                  continue;
# Line 2759  for (;; ptr++) Line 3605  for (;; ptr++)
3605          if ((options & PCRE_CASELESS) != 0)          if ((options & PCRE_CASELESS) != 0)
3606            {            {
3607            unsigned int othercase;            unsigned int othercase;
3608            if ((othercase = _pcre_ucp_othercase(c)) != NOTACHAR)            if ((othercase = UCD_OTHERCASE(c)) != c)
3609              {              {
3610              *class_utf8data++ = XCL_SINGLE;              *class_utf8data++ = XCL_SINGLE;
3611              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);
# Line 2786  for (;; ptr++) Line 3632  for (;; ptr++)
3632    
3633      /* Loop until ']' reached. This "while" is the end of the "do" above. */      /* Loop until ']' reached. This "while" is the end of the "do" above. */
3634    
3635      while ((c = *(++ptr)) != 0 && (c != ']' || inescq));      while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
3636    
3637      if (c == 0)                          /* Missing terminating ']' */      if (c == 0)                          /* Missing terminating ']' */
3638        {        {
# Line 2794  for (;; ptr++) Line 3640  for (;; ptr++)
3640        goto FAILED;        goto FAILED;
3641        }        }
3642    
3643    
3644    /* This code has been disabled because it would mean that \s counts as
3645    an explicit \r or \n reference, and that's not really what is wanted. Now
3646    we set the flag only if there is a literal "\r" or "\n" in the class. */
3647    
3648    #if 0
3649        /* Remember whether \r or \n are in this class */
3650    
3651        if (negate_class)
3652          {
3653          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3654          }
3655        else
3656          {
3657          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3658          }
3659    #endif
3660    
3661    
3662      /* 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
3663      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
3664      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
3665      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3666      single-bytes only. This is an historical hangover. Maybe one day we can  
3667      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3668        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3669        operate on single-bytes only. This is an historical hangover. Maybe one day
3670        we can tidy these opcodes to handle multi-byte characters.
3671    
3672      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
3673      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 2809  for (;; ptr++) Line 3677  for (;; ptr++)
3677      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3678    
3679  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3680      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3681            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3682  #else  #else
3683      if (class_charcount == 1)      if (class_charcount == 1)
3684  #endif  #endif
# Line 2855  for (;; ptr++) Line 3721  for (;; ptr++)
3721      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3722    
3723      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3724      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3725      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
3726        the class, so any that were explicitly given as well can be ignored. If
3727        (when there are explicit characters > 255 that must be listed) there are no
3728        characters < 256, we can omit the bitmap in the actual compiled code. */
3729    
3730  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3731      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3732        {        {
3733        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3734        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 2885  for (;; ptr++) Line 3754  for (;; ptr++)
3754        }        }
3755  #endif  #endif
3756    
3757      /* 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
3758      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
3759      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
3760      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3761    
3762        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3763      if (negate_class)      if (negate_class)
3764        {        {
       *code++ = OP_NCLASS;  
3765        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3766          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3767        }        }
3768      else      else
3769        {        {
       *code++ = OP_CLASS;  
3770        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3771        }        }
3772      code += 32;      code += 32;
# Line 2909  for (;; ptr++) Line 3777  for (;; ptr++)
3777      /* Various kinds of repeat; '{' is not necessarily a quantifier, but this      /* Various kinds of repeat; '{' is not necessarily a quantifier, but this
3778      has been tested above. */      has been tested above. */
3779    
3780      case '{':      case CHAR_LEFT_CURLY_BRACKET:
3781      if (!is_quantifier) goto NORMAL_CHAR;      if (!is_quantifier) goto NORMAL_CHAR;
3782      ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr);      ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr);
3783      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
3784      goto REPEAT;      goto REPEAT;
3785    
3786      case '*':      case CHAR_ASTERISK:
3787      repeat_min = 0;      repeat_min = 0;
3788      repeat_max = -1;      repeat_max = -1;
3789      goto REPEAT;      goto REPEAT;
3790    
3791      case '+':      case CHAR_PLUS:
3792      repeat_min = 1;      repeat_min = 1;
3793      repeat_max = -1;      repeat_max = -1;
3794      goto REPEAT;      goto REPEAT;
3795    
3796      case '?':      case CHAR_QUESTION_MARK:
3797      repeat_min = 0;      repeat_min = 0;
3798      repeat_max = 1;      repeat_max = 1;
3799    
# Line 2960  for (;; ptr++) Line 3828  for (;; ptr++)
3828      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
3829      repeat type to the non-default. */      repeat type to the non-default. */
3830    
3831      if (ptr[1] == '+')      if (ptr[1] == CHAR_PLUS)
3832        {        {
3833        repeat_type = 0;                  /* Force greedy */        repeat_type = 0;                  /* Force greedy */
3834        possessive_quantifier = TRUE;        possessive_quantifier = TRUE;
3835        ptr++;        ptr++;
3836        }        }
3837      else if (ptr[1] == '?')      else if (ptr[1] == CHAR_QUESTION_MARK)
3838        {        {
3839        repeat_type = greedy_non_default;        repeat_type = greedy_non_default;
3840        ptr++;        ptr++;
# Line 3081  for (;; ptr++) Line 3949  for (;; ptr++)
3949    
3950        if (repeat_max == 0) goto END_REPEAT;        if (repeat_max == 0) goto END_REPEAT;
3951    
3952          /*--------------------------------------------------------------------*/
3953          /* This code is obsolete from release 8.00; the restriction was finally
3954          removed: */
3955    
3956        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3957        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3958    
3959        if (repeat_max != 1) cd->nopartial = TRUE;        /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */
3960          /*--------------------------------------------------------------------*/
3961    
3962        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3963    
# Line 3231  for (;; ptr++) Line 4104  for (;; ptr++)
4104          goto END_REPEAT;          goto END_REPEAT;
4105          }          }
4106    
4107          /*--------------------------------------------------------------------*/
4108          /* This code is obsolete from release 8.00; the restriction was finally
4109          removed: */
4110    
4111        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
4112        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
4113    
4114        if (repeat_max != 1) cd->nopartial = TRUE;        /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */
4115          /*--------------------------------------------------------------------*/
4116    
4117        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
4118          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3270  for (;; ptr++) Line 4148  for (;; ptr++)
4148          goto FAILED;          goto FAILED;
4149          }          }
4150    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
4151        /* 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
4152        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
4153        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 3300  for (;; ptr++) Line 4170  for (;; ptr++)
4170    
4171        if (repeat_min == 0)        if (repeat_min == 0)
4172          {          {
4173          /* 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
4174          altogether. */          output altogether, like this:
   
         if (repeat_max == 0)  
           {  
           code = previous;  
           goto END_REPEAT;  
           }  
4175    
4176          /* If the maximum is 1 or unlimited, we just have to stick in the          ** if (repeat_max == 0)
4177          BRAZERO and do no more at this point. However, we do need to adjust          **   {
4178          any OP_RECURSE calls inside the group that refer to the group itself or          **   code = previous;
4179          any internal or forward referenced group, because the offset is from          **   goto END_REPEAT;
4180          the start of the whole regex. Temporarily terminate the pattern while          **   }
4181          doing this. */  
4182            However, that fails when a group is referenced as a subroutine from
4183            elsewhere in the pattern, so now we stick in OP_SKIPZERO in front of it
4184            so that it is skipped on execution. As we don't have a list of which
4185            groups are referenced, we cannot do this selectively.
4186    
4187            If the maximum is 1 or unlimited, we just have to stick in the BRAZERO
4188            and do no more at this point. However, we do need to adjust any
4189            OP_RECURSE calls inside the group that refer to the group itself or any
4190            internal or forward referenced group, because the offset is from the
4191            start of the whole regex. Temporarily terminate the pattern while doing
4192            this. */
4193    
4194          if (repeat_max <= 1)          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */
4195            {            {
4196            *code = OP_END;            *code = OP_END;
4197            adjust_recurse(previous, 1, utf8, cd, save_hwm);            adjust_recurse(previous, 1, utf8, cd, save_hwm);
4198            memmove(previous+1, previous, len);            memmove(previous+1, previous, len);
4199            code++;            code++;
4200              if (repeat_max == 0)
4201                {
4202                *previous++ = OP_SKIPZERO;
4203                goto END_REPEAT;
4204                }
4205            *previous++ = OP_BRAZERO + repeat_type;            *previous++ = OP_BRAZERO + repeat_type;
4206            }            }
4207    
# Line 3366  for (;; ptr++) Line 4246  for (;; ptr++)
4246          if (repeat_min > 1)          if (repeat_min > 1)
4247            {            {
4248            /* 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
4249            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
4250              potential integer overflow. The INT64_OR_DOUBLE type is a 64-bit
4251              integer type when available, otherwise double. */
4252    
4253            if (lengthptr != NULL)            if (lengthptr != NULL)
4254              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
4255                int delta = (repeat_min - 1)*length_prevgroup;
4256                if ((INT64_OR_DOUBLE)(repeat_min - 1)*
4257                      (INT64_OR_DOUBLE)length_prevgroup >
4258                        (INT64_OR_DOUBLE)INT_MAX ||
4259                    OFLOW_MAX - *lengthptr < delta)
4260                  {
4261                  *errorcodeptr = ERR20;
4262                  goto FAILED;
4263                  }
4264                *lengthptr += delta;
4265                }
4266    
4267            /* This is compiling for real */            /* This is compiling for real */
4268    
# Line 3407  for (;; ptr++) Line 4300  for (;; ptr++)
4300          /* 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
4301          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
4302          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
4303          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
4304            paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is
4305            a 64-bit integer type when available, otherwise double. */
4306    
4307          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
4308            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
4309              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
4310                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
4311              if ((INT64_OR_DOUBLE)repeat_max *
4312                    (INT64_OR_DOUBLE)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
4313                      > (INT64_OR_DOUBLE)INT_MAX ||
4314                  OFLOW_MAX - *lengthptr < delta)
4315                {
4316                *errorcodeptr = ERR20;
4317                goto FAILED;
4318                }
4319              *lengthptr += delta;
4320              }
4321    
4322          /* This is compiling for real */          /* This is compiling for real */
4323    
# Line 3481  for (;; ptr++) Line 4387  for (;; ptr++)
4387            uschar *scode = bracode;            uschar *scode = bracode;
4388            do            do
4389              {              {
4390              if (could_be_empty_branch(scode, ketcode, utf8))              if (could_be_empty_branch(scode, ketcode, utf8, cd))
4391                {                {
4392                *bracode += OP_SBRA - OP_BRA;                *bracode += OP_SBRA - OP_BRA;
4393                break;                break;
# Line 3493  for (;; ptr++) Line 4399  for (;; ptr++)
4399          }          }
4400        }        }
4401    
4402        /* If previous is OP_FAIL, it was generated by an empty class [] in
4403        JavaScript mode. The other ways in which OP_FAIL can be generated, that is
4404        by (*FAIL) or (?!) set previous to NULL, which gives a "nothing to repeat"
4405        error above. We can just ignore the repeat in JS case. */
4406    
4407        else if (*previous == OP_FAIL) goto END_REPEAT;
4408    
4409      /* Else there's some kind of shambles */      /* Else there's some kind of shambles */
4410    
4411      else      else
# Line 3517  for (;; ptr++) Line 4430  for (;; ptr++)
4430      if (possessive_quantifier)      if (possessive_quantifier)
4431        {        {
4432        int len;        int len;
4433        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||  
4434            *tempcode == OP_NOTEXACT)        if (*tempcode == OP_TYPEEXACT)
4435            tempcode += _pcre_OP_lengths[*tempcode] +
4436              ((tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP)? 2 : 0);
4437    
4438          else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)
4439            {
4440          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode];
4441    #ifdef SUPPORT_UTF8
4442            if (utf8 && tempcode[-1] >= 0xc0)
4443              tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f];
4444    #endif
4445            }
4446    
4447        len = code - tempcode;        len = code - tempcode;
4448        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4449          {          {
# Line 3538  for (;; ptr++) Line 4462  for (;; ptr++)
4462          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;
4463          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;
4464    
4465            /* Because we are moving code along, we must ensure that any
4466            pending recursive references are updated. */
4467    
4468          default:          default:
4469            *code = OP_END;
4470            adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);
4471          memmove(tempcode + 1+LINK_SIZE, tempcode, len);          memmove(tempcode + 1+LINK_SIZE, tempcode, len);
4472          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
4473          len += 1 + LINK_SIZE;          len += 1 + LINK_SIZE;
# Line 3563  for (;; ptr++) Line 4492  for (;; ptr++)
4492      /* ===================================================================*/      /* ===================================================================*/
4493      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4494      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4495      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. */  
4496    
4497      case '(':      case CHAR_LEFT_PARENTHESIS:
4498      newoptions = options;      newoptions = options;
4499      skipbytes = 0;      skipbytes = 0;
4500      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4501      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4502        reset_bracount = FALSE;
4503    
4504        /* First deal with various "verbs" that can be introduced by '*'. */
4505    
4506        if (*(++ptr) == CHAR_ASTERISK && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4507          {
4508          int i, namelen;
4509          const char *vn = verbnames;
4510          const uschar *name = ++ptr;
4511          previous = NULL;
4512          while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
4513          if (*ptr == CHAR_COLON)
4514            {
4515            *errorcodeptr = ERR59;   /* Not supported */
4516            goto FAILED;
4517            }
4518          if (*ptr != CHAR_RIGHT_PARENTHESIS)
4519            {
4520            *errorcodeptr = ERR60;
4521            goto FAILED;
4522            }
4523          namelen = ptr - name;
4524          for (i = 0; i < verbcount; i++)
4525            {
4526            if (namelen == verbs[i].len &&
4527                strncmp((char *)name, vn, namelen) == 0)
4528              {
4529              /* Check for open captures before ACCEPT */
4530    
4531              if (verbs[i].op == OP_ACCEPT)
4532                {
4533                open_capitem *oc;
4534                cd->had_accept = TRUE;
4535                for (oc = cd->open_caps; oc != NULL; oc = oc->next)
4536                  {
4537                  *code++ = OP_CLOSE;
4538                  PUT2INC(code, 0, oc->number);
4539                  }
4540                }
4541              *code++ = verbs[i].op;
4542              break;
4543              }
4544            vn += verbs[i].len + 1;
4545            }
4546          if (i < verbcount) continue;
4547          *errorcodeptr = ERR60;
4548          goto FAILED;
4549          }
4550    
4551        /* Deal with the extended parentheses; all are introduced by '?', and the
4552        appearance of any of them means that this is not a capturing group. */
4553    
4554      if (*(++ptr) == '?')      else if (*ptr == CHAR_QUESTION_MARK)
4555        {        {
4556        int i, set, unset, namelen;        int i, set, unset, namelen;
4557        int *optset;        int *optset;
# Line 3582  for (;; ptr++) Line 4560  for (;; ptr++)
4560    
4561        switch (*(++ptr))        switch (*(++ptr))
4562          {          {
4563          case '#':                 /* Comment; skip to ket */          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */
4564          ptr++;          ptr++;
4565          while (*ptr != 0 && *ptr != ')') ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
4566          if (*ptr == 0)          if (*ptr == 0)
4567            {            {
4568            *errorcodeptr = ERR18;            *errorcodeptr = ERR18;
# Line 3594  for (;; ptr++) Line 4572  for (;; ptr++)
4572    
4573    
4574          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4575          case ':':                 /* Non-capturing bracket */          case CHAR_VERTICAL_LINE:  /* Reset capture count for each branch */
4576            reset_bracount = TRUE;
4577            /* Fall through */
4578    
4579            /* ------------------------------------------------------------ */
4580            case CHAR_COLON:          /* Non-capturing bracket */
4581          bravalue = OP_BRA;          bravalue = OP_BRA;
4582          ptr++;          ptr++;
4583          break;          break;
4584    
4585    
4586          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4587          case '(':          case CHAR_LEFT_PARENTHESIS:
4588          bravalue = OP_COND;       /* Conditional group */          bravalue = OP_COND;       /* Conditional group */
4589    
4590          /* 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 3621  for (;; ptr++) Line 4604  for (;; ptr++)
4604          the switch. This will take control down to where bracketed groups,          the switch. This will take control down to where bracketed groups,
4605          including assertions, are processed. */          including assertions, are processed. */
4606    
4607          if (ptr[1] == '?' && (ptr[2] == '=' || ptr[2] == '!' || ptr[2] == '<'))          if (ptr[1] == CHAR_QUESTION_MARK && (ptr[2] == CHAR_EQUALS_SIGN ||
4608                ptr[2] == CHAR_EXCLAMATION_MARK || ptr[2] == CHAR_LESS_THAN_SIGN))
4609            break;            break;
4610    
4611          /* 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 3629  for (;; ptr++) Line 4613  for (;; ptr++)
4613    
4614          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
4615          skipbytes = 3;          skipbytes = 3;
4616          refsign = -1;          refsign = -1;
4617    
4618          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4619    
4620          if (ptr[1] == 'R' && ptr[2] == '&')          if (ptr[1] == CHAR_R && ptr[2] == CHAR_AMPERSAND)
4621            {            {
4622            terminator = -1;            terminator = -1;
4623            ptr += 2;            ptr += 2;
# Line 3643  for (;; ptr++) Line 4627  for (;; ptr++)
4627          /* 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
4628          syntax (?(<name>) or (?('name') */          syntax (?(<name>) or (?('name') */
4629    
4630          else if (ptr[1] == '<')          else if (ptr[1] == CHAR_LESS_THAN_SIGN)
4631            {            {
4632            terminator = '>';            terminator = CHAR_GREATER_THAN_SIGN;
4633            ptr++;            ptr++;
4634            }            }
4635          else if (ptr[1] == '\'')          else if (ptr[1] == CHAR_APOSTROPHE)
4636            {            {
4637            terminator = '\'';            terminator = CHAR_APOSTROPHE;
4638            ptr++;            ptr++;
4639            }            }
4640          else          else
4641            {            {
4642            terminator = 0;            terminator = 0;
4643            if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);
4644            }            }
4645    
4646          /* 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 */
4647    
# Line 3676  for (;; ptr++) Line 4660  for (;; ptr++)
4660            {            {
4661            if (recno >= 0)            if (recno >= 0)
4662              recno = ((digitab[*ptr] & ctype_digit) != 0)?              recno = ((digitab[*ptr] & ctype_digit) != 0)?
4663                recno * 10 + *ptr - '0' : -1;                recno * 10 + *ptr - CHAR_0 : -1;
4664            ptr++;            ptr++;
4665            }            }
4666          namelen = ptr - name;          namelen = ptr - name;
4667    
4668          if ((terminator > 0 && *ptr++ != terminator) || *ptr++ != ')')          if ((terminator > 0 && *ptr++ != terminator) ||
4669                *ptr++ != CHAR_RIGHT_PARENTHESIS)
4670            {            {
4671            ptr--;      /* Error offset */            ptr--;      /* Error offset */
4672            *errorcodeptr = ERR26;            *errorcodeptr = ERR26;
# Line 3695  for (;; ptr++) Line 4680  for (;; ptr++)
4680          /* 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
4681          reference. If the string started with "+" or "-" we require the rest to          reference. If the string started with "+" or "-" we require the rest to
4682          be digits, in which case recno will be set. */          be digits, in which case recno will be set. */
4683    
4684          if (refsign > 0)          if (refsign > 0)
4685            {            {
4686            if (recno <= 0)            if (recno <= 0)
4687              {              {
4688              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4689              goto FAILED;              goto FAILED;
4690              }              }
4691            if (refsign == '-')            recno = (refsign == CHAR_MINUS)?
4692                cd->bracount - recno + 1 : recno +cd->bracount;
4693              if (recno <= 0 || recno > cd->final_bracount)
4694              {              {
4695              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4696              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4697              }              }
           else recno += cd->bracount;  
4698            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4699            break;            break;
4700            }            }
4701    
4702          /* Otherwise (did not start with "+" or "-"), start by looking for the          /* Otherwise (did not start with "+" or "-"), start by looking for the
4703          name. */          name. If we find a name, add one to the opcode to change OP_CREF or
4704            OP_RREF into OP_NCREF or OP_NRREF. These behave exactly the same,
4705            except they record that the reference was originally to a name. The
4706            information is used to check duplicate names. */
4707    
4708          slot = cd->name_table;          slot = cd->name_table;
4709          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
4710            {            {
# Line 3733  for (;; ptr++) Line 4718  for (;; ptr++)
4718            {            {
4719            recno = GET2(slot, 0);            recno = GET2(slot, 0);
4720            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4721              code[1+LINK_SIZE]++;
4722            }            }
4723    
4724          /* Search the pattern for a forward reference */          /* Search the pattern for a forward reference */
4725    
4726          else if ((i = find_parens(ptr, cd->bracount, name, namelen,          else if ((i = find_parens(cd, name, namelen,
4727                          (options & PCRE_EXTENDED) != 0)) > 0)                          (options & PCRE_EXTENDED) != 0)) > 0)
4728            {            {
4729            PUT2(code, 2+LINK_SIZE, i);            PUT2(code, 2+LINK_SIZE, i);
4730              code[1+LINK_SIZE]++;
4731            }            }
4732    
4733          /* If terminator == 0 it means that the name followed directly after          /* If terminator == 0 it means that the name followed directly after
# Line 3758  for (;; ptr++) Line 4745  for (;; ptr++)
4745          /* Check for (?(R) for recursion. Allow digits after R to specify a          /* Check for (?(R) for recursion. Allow digits after R to specify a
4746          specific group number. */          specific group number. */
4747    
4748          else if (*name == 'R')          else if (*name == CHAR_R)
4749            {            {
4750            recno = 0;            recno = 0;
4751            for (i = 1; i < namelen; i++)            for (i = 1; i < namelen; i++)
# Line 3768  for (;; ptr++) Line 4755  for (;; ptr++)
4755                *errorcodeptr = ERR15;                *errorcodeptr = ERR15;
4756                goto FAILED;                goto FAILED;
4757                }                }
4758              recno = recno * 10 + name[i] - '0';              recno = recno * 10 + name[i] - CHAR_0;
4759              }              }
4760            if (recno == 0) recno = RREF_ANY;            if (recno == 0) recno = RREF_ANY;
4761            code[1+LINK_SIZE] = OP_RREF;      /* Change test type */            code[1+LINK_SIZE] = OP_RREF;      /* Change test type */
# Line 3778  for (;; ptr++) Line 4765  for (;; ptr++)
4765          /* Similarly, check for the (?(DEFINE) "condition", which is always          /* Similarly, check for the (?(DEFINE) "condition", which is always
4766          false. */          false. */
4767    
4768          else if (namelen == 6 && strncmp((char *)name, "DEFINE", 6) == 0)          else if (namelen == 6 && strncmp((char *)name, STRING_DEFINE, 6) == 0)
4769            {            {
4770            code[1+LINK_SIZE] = OP_DEF;            code[1+LINK_SIZE] = OP_DEF;
4771            skipbytes = 1;            skipbytes = 1;
4772            }            }
4773    
4774          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4775            in the second pass here, so final_bracount is set. */
4776    
4777          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4778            {            {
4779            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4780            }            }
# Line 3802  for (;; ptr++) Line 4790  for (;; ptr++)
4790    
4791    
4792          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4793          case '=':                 /* Positive lookahead */          case CHAR_EQUALS_SIGN:                 /* Positive lookahead */
4794          bravalue = OP_ASSERT;          bravalue = OP_ASSERT;
4795          ptr++;          ptr++;
4796          break;          break;
4797    
4798    
4799          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4800          case '!':                 /* Negative lookahead */          case CHAR_EXCLAMATION_MARK:            /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4801          ptr++;          ptr++;
4802            if (*ptr == CHAR_RIGHT_PARENTHESIS)    /* Optimize (?!) */
4803              {
4804              *code++ = OP_FAIL;
4805              previous = NULL;
4806              continue;
4807              }
4808            bravalue = OP_ASSERT_NOT;
4809          break;          break;
4810    
4811    
4812          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4813          case '<':                 /* Lookbehind or named define */          case CHAR_LESS_THAN_SIGN:              /* Lookbehind or named define */
4814          switch (ptr[1])          switch (ptr[1])
4815            {            {
4816            case '=':               /* Positive lookbehind */            case CHAR_EQUALS_SIGN:               /* Positive lookbehind */
4817            bravalue = OP_ASSERTBACK;            bravalue = OP_ASSERTBACK;
4818            ptr += 2;            ptr += 2;
4819            break;            break;
4820    
4821            case '!':               /* Negative lookbehind */            case CHAR_EXCLAMATION_MARK:          /* Negative lookbehind */
4822            bravalue = OP_ASSERTBACK_NOT;            bravalue = OP_ASSERTBACK_NOT;
4823            ptr += 2;            ptr += 2;
4824            break;            break;
# Line 3839  for (;; ptr++) Line 4833  for (;; ptr++)
4833    
4834    
4835          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4836          case '>':                 /* One-time brackets */          case CHAR_GREATER_THAN_SIGN:           /* One-time brackets */
4837          bravalue = OP_ONCE;          bravalue = OP_ONCE;
4838          ptr++;          ptr++;
4839          break;          break;
4840    
4841    
4842          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4843          case 'C':                 /* Callout - may be followed by digits; */          case CHAR_C:                 /* Callout - may be followed by digits; */
4844          previous_callout = code;  /* Save for later completion */          previous_callout = code;  /* Save for later completion */
4845          after_manual_callout = 1; /* Skip one item before completing */          after_manual_callout = 1; /* Skip one item before completing */
4846          *code++ = OP_CALLOUT;          *code++ = OP_CALLOUT;
4847            {            {
4848            int n = 0;            int n = 0;
4849            while ((digitab[*(++ptr)] & ctype_digit) != 0)            while ((digitab[*(++ptr)] & ctype_digit) != 0)
4850              n = n * 10 + *ptr - '0';              n = n * 10 + *ptr - CHAR_0;
4851            if (*ptr != ')')            if (*ptr != CHAR_RIGHT_PARENTHESIS)
4852              {              {
4853              *errorcodeptr = ERR39;              *errorcodeptr = ERR39;
4854              goto FAILED;              goto FAILED;
# Line 3874  for (;; ptr++) Line 4868  for (;; ptr++)
4868    
4869    
4870          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4871          case 'P':                 /* Python-style named subpattern handling */          case CHAR_P:              /* Python-style named subpattern handling */
4872          if (*(++ptr) == '=' || *ptr == '>')  /* Reference or recursion */          if (*(++ptr) == CHAR_EQUALS_SIGN ||
4873                *ptr == CHAR_GREATER_THAN_SIGN)  /* Reference or recursion */
4874            {            {
4875            is_recurse = *ptr == '>';            is_recurse = *ptr == CHAR_GREATER_THAN_SIGN;
4876            terminator = ')';            terminator = CHAR_RIGHT_PARENTHESIS;
4877            goto NAMED_REF_OR_RECURSE;            goto NAMED_REF_OR_RECURSE;
4878            }            }
4879          else if (*ptr != '<')    /* Test for Python-style definition */          else if (*ptr != CHAR_LESS_THAN_SIGN)  /* Test for Python-style defn */
4880            {            {
4881            *errorcodeptr = ERR41;            *errorcodeptr = ERR41;
4882            goto FAILED;            goto FAILED;
# Line 3891  for (;; ptr++) Line 4886  for (;; ptr++)
4886    
4887          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4888          DEFINE_NAME:    /* Come here from (?< handling */          DEFINE_NAME:    /* Come here from (?< handling */
4889          case '\'':          case CHAR_APOSTROPHE:
4890            {            {
4891            terminator = (*ptr == '<')? '>' : '\'';            terminator = (*ptr == CHAR_LESS_THAN_SIGN)?
4892                CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
4893            name = ++ptr;            name = ++ptr;
4894    
4895            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
# Line 3924  for (;; ptr++) Line 4920  for (;; ptr++)
4920                }                }
4921              }              }
4922    
4923            /* In the real compile, create the entry in the table */            /* In the real compile, create the entry in the table, maintaining
4924              alphabetical order. Duplicate names for different numbers are
4925              permitted only if PCRE_DUPNAMES is set. Duplicate names for the same
4926              number are always OK. (An existing number can be re-used if (?|
4927              appears in the pattern.) In either event, a duplicate name results in
4928              a duplicate entry in the table, even if the number is the same. This
4929              is because the number of names, and hence the table size, is computed
4930              in the pre-compile, and it affects various numbers and pointers which
4931              would all have to be modified, and the compiled code moved down, if
4932              duplicates with the same number were omitted from the table. This
4933              doesn't seem worth the hassle. However, *different* names for the
4934              same number are not permitted. */
4935    
4936            else            else
4937              {              {
4938                BOOL dupname = FALSE;
4939              slot = cd->name_table;              slot = cd->name_table;
4940    
4941              for (i = 0; i < cd->names_found; i++)              for (i = 0; i < cd->names_found; i++)
4942                {                {
4943                int crc = memcmp(name, slot+2, namelen);                int crc = memcmp(name, slot+2, namelen);
# Line 3936  for (;; ptr++) Line 4945  for (;; ptr++)
4945                  {                  {
4946                  if (slot[2+namelen] == 0)                  if (slot[2+namelen] == 0)
4947                    {                    {
4948                    if ((options & PCRE_DUPNAMES) == 0)                    if (GET2(slot, 0) != cd->bracount + 1 &&
4949                          (options & PCRE_DUPNAMES) == 0)
4950                      {                      {
4951                      *errorcodeptr = ERR43;                      *errorcodeptr = ERR43;
4952                      goto FAILED;                      goto FAILED;
4953                      }                      }
4954                      else dupname = TRUE;
4955                    }                    }
4956                  else crc = -1;      /* Current name is substring */                  else crc = -1;      /* Current name is a substring */
4957                  }                  }
4958    
4959                  /* Make space in the table and break the loop for an earlier
4960                  name. For a duplicate or later name, carry on. We do this for
4961                  duplicates so that in the simple case (when ?(| is not used) they
4962                  are in order of their numbers. */
4963    
4964                if (crc < 0)                if (crc < 0)
4965                  {                  {
4966                  memmove(slot + cd->name_entry_size, slot,                  memmove(slot + cd->name_entry_size, slot,
4967                    (cd->names_found - i) * cd->name_entry_size);                    (cd->names_found - i) * cd->name_entry_size);
4968                  break;                  break;
4969                  }                  }
4970    
4971                  /* Continue the loop for a later or duplicate name */
4972    
4973                slot += cd->name_entry_size;                slot += cd->name_entry_size;
4974                }                }
4975    
4976                /* For non-duplicate names, check for a duplicate number before
4977                adding the new name. */
4978    
4979                if (!dupname)
4980                  {
4981                  uschar *cslot = cd->name_table;
4982                  for (i = 0; i < cd->names_found; i++)
4983                    {
4984                    if (cslot != slot)
4985                      {
4986                      if (GET2(cslot, 0) == cd->bracount + 1)
4987                        {
4988                        *errorcodeptr = ERR65;
4989                        goto FAILED;
4990                        }
4991                      }
4992                    else i--;
4993                    cslot += cd->name_entry_size;
4994                    }
4995                  }
4996    
4997              PUT2(slot, 0, cd->bracount + 1);              PUT2(slot, 0, cd->bracount + 1);
4998              memcpy(slot + 2, name, namelen);              memcpy(slot + 2, name, namelen);
4999              slot[2+namelen] = 0;              slot[2+namelen] = 0;
5000              }              }
5001            }            }
5002    
5003          /* In both cases, count the number of names we've encountered. */          /* In both pre-compile and compile, count the number of names we've
5004            encountered. */
5005    
         ptr++;                    /* Move past > or ' */  
5006          cd->names_found++;          cd->names_found++;
5007            ptr++;                    /* Move past > or ' */
5008          goto NUMBERED_GROUP;          goto NUMBERED_GROUP;
5009    
5010    
5011          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
5012          case '&':                 /* Perl recursion/subroutine syntax */          case CHAR_AMPERSAND:            /* Perl recursion/subroutine syntax */
5013          terminator = ')';          terminator = CHAR_RIGHT_PARENTHESIS;
5014          is_recurse = TRUE;          is_recurse = TRUE;
5015          /* Fall through */          /* Fall through */
5016    
5017          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
5018          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
5019          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
5020            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
5021            .NET syntax, and the Oniguruma \g<...> and \g'...' subroutine syntax. */
5022    
5023          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
5024          name = ++ptr;          name = ++ptr;
# Line 3986  for (;; ptr++) Line 5030  for (;; ptr++)
5030    
5031          if (lengthptr != NULL)          if (lengthptr != NULL)
5032            {            {
5033              if (namelen == 0)
5034                {
5035                *errorcodeptr = ERR62;
5036                goto FAILED;
5037                }
5038            if (*ptr != terminator)            if (*ptr != terminator)
5039              {              {
5040              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 3999  for (;; ptr++) Line 5048  for (;; ptr++)
5048            recno = 0;            recno = 0;
5049            }            }
5050    
5051          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
5052            first, and then check that we have reached the end of the name in the
5053            table. That way, if the name that is longer than any in the table,
5054            the comparison will fail without reading beyond the table entry. */
5055    
5056          else          else
5057            {            {
5058            slot = cd->name_table;            slot = cd->name_table;
5059            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
5060              {              {
5061              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
5062                    slot[2+namelen] == 0)
5063                  break;
5064              slot += cd->name_entry_size;              slot += cd->name_entry_size;
5065              }              }
5066    
# Line 4015  for (;; ptr++) Line 5069  for (;; ptr++)
5069              recno = GET2(slot, 0);              recno = GET2(slot, 0);