/[pcre]/code/tags/pcre-7.8/pcre_compile.c
ViewVC logotype

Diff of /code/tags/pcre-7.8/pcre_compile.c

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

revision 96 by nigel, Fri Mar 2 13:10:43 2007 UTC revision 240 by ph10, Tue Sep 11 15:47:20 2007 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) 1997-2006 University of Cambridge             Copyright (c) 1997-2007 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    
# Line 58  used by pcretest. DEBUG is not defined w Line 61  used by pcretest. DEBUG is not defined w
61  #endif  #endif
62    
63    
64    /* Macro for setting individual bits in class bitmaps. */
65    
66    #define SETBIT(a,b) a[b/8] |= (1 << (b%8))
67    
68    /* Maximum length value to check against when making sure that the integer that
69    holds the compiled pattern length does not overflow. We make it a bit less than
70    INT_MAX to allow for adding in group terminating bytes, so that we don't have
71    to check them every time. */
72    
73    #define OFLOW_MAX (INT_MAX - 20)
74    
75    
76  /*************************************************  /*************************************************
77  *      Code parameters and static tables         *  *      Code parameters and static tables         *
78  *************************************************/  *************************************************/
# Line 82  are simple data values; negative values Line 97  are simple data values; negative values
97  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
98  is invalid. */  is invalid. */
99    
100  #if !EBCDIC   /* This is the "normal" table for ASCII systems */  #ifndef EBCDIC  /* This is the "normal" table for ASCII systems */
101  static const short int escapes[] = {  static const short int escapes[] = {
102       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */
103       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */
104     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */
105       0,      0,      0,      0,      0,      0,      0,      0,   /* H - O */  -ESC_H,      0,      0, -ESC_K,      0,      0,      0,      0,   /* H - O */
106  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0,      0, -ESC_W,   /* P - W */  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0, -ESC_V, -ESC_W,   /* P - W */
107  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */
108     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */
109       0,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */  -ESC_h,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */
110  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0,      0, -ESC_w,   /* p - w */  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0, -ESC_v, -ESC_w,   /* p - w */
111       0,      0, -ESC_z                                            /* x - z */       0,      0, -ESC_z                                            /* x - z */
112  };  };
113    
114  #else         /* This is the "abnormal" table for EBCDIC systems */  #else           /* This is the "abnormal" table for EBCDIC systems */
115  static const short int escapes[] = {  static const short int escapes[] = {
116  /*  48 */     0,     0,      0,     '.',    '<',   '(',    '+',    '|',  /*  48 */     0,     0,      0,     '.',    '<',   '(',    '+',    '|',
117  /*  50 */   '&',     0,      0,       0,      0,     0,      0,      0,  /*  50 */   '&',     0,      0,       0,      0,     0,      0,      0,
# Line 106  static const short int escapes[] = { Line 121  static const short int escapes[] = {
121  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,
122  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',
123  /*  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,
124  /*  88 */     0,     0,      0,     '{',      0,     0,      0,      0,  /*  88 */-ESC_h,     0,      0,     '{',      0,     0,      0,      0,
125  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,
126  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,
127  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,     0, -ESC_w,      0,  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,-ESC_v, -ESC_w,      0,
128  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,
129  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,
130  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',
131  /*  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,
132  /*  C8 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,
133  /*  D0 */   '}',     0,      0,       0,      0,     0,      0, -ESC_P,  /*  D0 */   '}',     0, -ESC_K,       0,      0,     0,      0, -ESC_P,
134  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,
135  /*  E0 */  '\\',     0, -ESC_S,       0,      0,     0, -ESC_W, -ESC_X,  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,
136  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,
137  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,
138  /*  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 140  static const short int escapes[] = {
140  #endif  #endif
141    
142    
143  /* 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
144  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
145  as this is assumed for handling case independence. */  the number of relocations when a shared library is dynamically linked. */
146    
147  static const char *const posix_names[] = {  typedef struct verbitem {
148    "alpha", "lower", "upper",    int   len;
149    "alnum", "ascii", "blank", "cntrl", "digit", "graph",    int   op;
150    "print", "punct", "space", "word",  "xdigit" };  } verbitem;
151    
152    static const char verbnames[] =
153      "ACCEPT\0"
154      "COMMIT\0"
155      "F\0"
156      "FAIL\0"
157      "PRUNE\0"
158      "SKIP\0"
159      "THEN";
160    
161    static verbitem verbs[] = {
162      { 6, OP_ACCEPT },
163      { 6, OP_COMMIT },
164      { 1, OP_FAIL },
165      { 4, OP_FAIL },
166      { 5, OP_PRUNE },
167      { 4, OP_SKIP  },
168      { 4, OP_THEN  }
169    };
170    
171    static int verbcount = sizeof(verbs)/sizeof(verbitem);
172    
173    
174    /* Tables of names of POSIX character classes and their lengths. The names are
175    now all in a single string, to reduce the number of relocations when a shared
176    library is dynamically loaded. The list of lengths is terminated by a zero
177    length entry. The first three must be alpha, lower, upper, as this is assumed
178    for handling case independence. */
179    
180    static const char posix_names[] =
181      "alpha\0"  "lower\0"  "upper\0"  "alnum\0"  "ascii\0"  "blank\0"
182      "cntrl\0"  "digit\0"  "graph\0"  "print\0"  "punct\0"  "space\0"
183      "word\0"   "xdigit";
184    
185  static const uschar posix_name_lengths[] = {  static const uschar posix_name_lengths[] = {
186    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 219  static const int posix_class_maps[] = {
219  /* The texts of compile-time error messages. These are "char *" because they  /* The texts of compile-time error messages. These are "char *" because they
220  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
221  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
222  are no longer used. */  are no longer used. This used to be a table of strings, but in order to reduce
223    the number of relocations needed when a shared library is loaded dynamically,
224  static const char *error_texts[] = {  it is now one long string. We cannot use a table of offsets, because the
225    "no error",  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we
226    "\\ at end of pattern",  simply count through to the one we want - this isn't a performance issue
227    "\\c at end of pattern",  because these strings are used only when there is a compilation error. */
228    "unrecognized character follows \\",  
229    "numbers out of order in {} quantifier",  static const char error_texts[] =
230      "no error\0"
231      "\\ at end of pattern\0"
232      "\\c at end of pattern\0"
233      "unrecognized character follows \\\0"
234      "numbers out of order in {} quantifier\0"
235    /* 5 */    /* 5 */
236    "number too big in {} quantifier",    "number too big in {} quantifier\0"
237    "missing terminating ] for character class",    "missing terminating ] for character class\0"
238    "invalid escape sequence in character class",    "invalid escape sequence in character class\0"
239    "range out of order in character class",    "range out of order in character class\0"
240    "nothing to repeat",    "nothing to repeat\0"
241    /* 10 */    /* 10 */
242    "operand of unlimited repeat could match the empty string",  /** DEAD **/    "operand of unlimited repeat could match the empty string\0"  /** DEAD **/
243    "internal error: unexpected repeat",    "internal error: unexpected repeat\0"
244    "unrecognized character after (?",    "unrecognized character after (?\0"
245    "POSIX named classes are supported only within a class",    "POSIX named classes are supported only within a class\0"
246    "missing )",    "missing )\0"
247    /* 15 */    /* 15 */
248    "reference to non-existent subpattern",    "reference to non-existent subpattern\0"
249    "erroffset passed as NULL",    "erroffset passed as NULL\0"
250    "unknown option bit(s) set",    "unknown option bit(s) set\0"
251    "missing ) after comment",    "missing ) after comment\0"
252    "parentheses nested too deeply",  /** DEAD **/    "parentheses nested too deeply\0"  /** DEAD **/
253    /* 20 */    /* 20 */
254    "regular expression too large",    "regular expression is too large\0"
255    "failed to get memory",    "failed to get memory\0"
256    "unmatched parentheses",    "unmatched parentheses\0"
257    "internal error: code overflow",    "internal error: code overflow\0"
258    "unrecognized character after (?<",    "unrecognized character after (?<\0"
259    /* 25 */    /* 25 */
260    "lookbehind assertion is not fixed length",    "lookbehind assertion is not fixed length\0"
261    "malformed number or name after (?(",    "malformed number or name after (?(\0"
262    "conditional group contains more than two branches",    "conditional group contains more than two branches\0"
263    "assertion expected after (?(",    "assertion expected after (?(\0"
264    "(?R or (?digits must be followed by )",    "(?R or (?[+-]digits must be followed by )\0"
265    /* 30 */    /* 30 */
266    "unknown POSIX class name",    "unknown POSIX class name\0"
267    "POSIX collating elements are not supported",    "POSIX collating elements are not supported\0"
268    "this version of PCRE is not compiled with PCRE_UTF8 support",    "this version of PCRE is not compiled with PCRE_UTF8 support\0"
269    "spare error",  /** DEAD **/    "spare error\0"  /** DEAD **/
270    "character value in \\x{...} sequence is too large",    "character value in \\x{...} sequence is too large\0"
271    /* 35 */    /* 35 */
272    "invalid condition (?(0)",    "invalid condition (?(0)\0"
273    "\\C not allowed in lookbehind assertion",    "\\C not allowed in lookbehind assertion\0"
274    "PCRE does not support \\L, \\l, \\N, \\U, or \\u",    "PCRE does not support \\L, \\l, \\N, \\U, or \\u\0"
275    "number after (?C is > 255",    "number after (?C is > 255\0"
276    "closing ) for (?C expected",    "closing ) for (?C expected\0"
277    /* 40 */    /* 40 */
278    "recursive call could loop indefinitely",    "recursive call could loop indefinitely\0"
279    "unrecognized character after (?P",    "unrecognized character after (?P\0"
280    "syntax error in subpattern name (missing terminator)",    "syntax error in subpattern name (missing terminator)\0"
281    "two named subpatterns have the same name",    "two named subpatterns have the same name\0"
282    "invalid UTF-8 string",    "invalid UTF-8 string\0"
283    /* 45 */    /* 45 */
284    "support for \\P, \\p, and \\X has not been compiled",    "support for \\P, \\p, and \\X has not been compiled\0"
285    "malformed \\P or \\p sequence",    "malformed \\P or \\p sequence\0"
286    "unknown property name after \\P or \\p",    "unknown property name after \\P or \\p\0"
287    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)\0"
288    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
289    /* 50 */    /* 50 */
290    "repeated subpattern is too long",    "repeated subpattern is too long\0"    /** DEAD **/
291    "octal value is greater than \\377 (not in UTF-8 mode)",    "octal value is greater than \\377 (not in UTF-8 mode)\0"
292    "internal error: overran compiling workspace",    "internal error: overran compiling workspace\0"
293    "internal error: previously-checked referenced subpattern not found",    "internal error: previously-checked referenced subpattern not found\0"
294    "DEFINE group contains more than one branch",    "DEFINE group contains more than one branch\0"
295    /* 55 */    /* 55 */
296    "repeating a DEFINE group is not allowed",    "repeating a DEFINE group is not allowed\0"
297    "inconsistent NEWLINE options",    "inconsistent NEWLINE options\0"
298    "\\g is not followed by an (optionally braced) non-zero number"    "\\g is not followed by a braced name or an optionally braced non-zero number\0"
299  };    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number\0"
300      "(*VERB) with an argument is not supported\0"
301      /* 60 */
302      "(*VERB) not recognized\0"
303      "number is too big";
304    
305    
306  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 262  For convenience, we use the same bit def Line 319  For convenience, we use the same bit def
319    
320  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
321    
322  #if !EBCDIC    /* This is the "normal" case, for ASCII systems */  #ifndef EBCDIC  /* This is the "normal" case, for ASCII systems */
323  static const unsigned char digitab[] =  static const unsigned char digitab[] =
324    {    {
325    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
# Line 298  static const unsigned char digitab[] = Line 355  static const unsigned char digitab[] =
355    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */
356    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */
357    
358  #else          /* This is the "abnormal" case, for EBCDIC systems */  #else           /* This is the "abnormal" case, for EBCDIC systems */
359  static const unsigned char digitab[] =  static const unsigned char digitab[] =
360    {    {
361    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
# Line 312  static const unsigned char digitab[] = Line 369  static const unsigned char digitab[] =
369    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*    - 71 40 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*    - 71 40 */
370    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  72- |     */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  72- |     */
371    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  & - 87 50 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  & - 87 50 */
372    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  88-     */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  88- 95    */
373    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  - -103 60 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  - -103 60 */
374    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 104- ?     */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 104- ?     */
375    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 70 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 70 */
# Line 346  static const unsigned char ebcdic_charta Line 403  static const unsigned char ebcdic_charta
403    0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*    - 71 */    0x01,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*    - 71 */
404    0x00,0x00,0x00,0x80,0x00,0x80,0x80,0x80, /*  72- |  */    0x00,0x00,0x00,0x80,0x00,0x80,0x80,0x80, /*  72- |  */
405    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  & - 87 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  & - 87 */
406    0x00,0x00,0x00,0x80,0x80,0x80,0x00,0x00, /*  88-  */    0x00,0x00,0x00,0x80,0x80,0x80,0x00,0x00, /*  88- 95 */
407    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  - -103 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*  - -103 */
408    0x00,0x00,0x00,0x00,0x00,0x10,0x00,0x80, /* 104- ?  */    0x00,0x00,0x00,0x00,0x00,0x10,0x00,0x80, /* 104- ?  */
409    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 112-119 */
# Line 373  static const unsigned char ebcdic_charta Line 430  static const unsigned char ebcdic_charta
430  /* Definition to allow mutual recursion */  /* Definition to allow mutual recursion */
431    
432  static BOOL  static BOOL
433    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, int, int *,    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,
434      int *, branch_chain *, compile_data *, int *);      int *, int *, branch_chain *, compile_data *, int *);
435    
436    
437    
438    /*************************************************
439    *            Find an error text                  *
440    *************************************************/
441    
442    /* The error texts are now all in one long string, to save on relocations. As
443    some of the text is of unknown length, we can't use a table of offsets.
444    Instead, just count through the strings. This is not a performance issue
445    because it happens only when there has been a compilation error.
446    
447    Argument:   the error number
448    Returns:    pointer to the error string
449    */
450    
451    static const char *
452    find_error_text(int n)
453    {
454    const char *s = error_texts;
455    for (; n > 0; n--) while (*s++ != 0);
456    return s;
457    }
458    
459    
460  /*************************************************  /*************************************************
# Line 399  Arguments: Line 478  Arguments:
478    
479  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
480                   negative => a special escape sequence                   negative => a special escape sequence
481                   on error, errorptr is set                   on error, errorcodeptr is set
482  */  */
483    
484  static int  static int
# Line 421  if (c == 0) *errorcodeptr = ERR1; Line 500  if (c == 0) *errorcodeptr = ERR1;
500  a table. A non-zero result is something that can be returned immediately.  a table. A non-zero result is something that can be returned immediately.
501  Otherwise further processing may be required. */  Otherwise further processing may be required. */
502    
503  #if !EBCDIC    /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
504  else if (c < '0' || c > 'z') {}                           /* Not alphameric */  else if (c < '0' || c > 'z') {}                           /* Not alphameric */
505  else if ((i = escapes[c - '0']) != 0) c = i;  else if ((i = escapes[c - '0']) != 0) c = i;
506    
507  #else          /* EBCDIC coding */  #else           /* EBCDIC coding */
508  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphameric */  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphameric */
509  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
510  #endif  #endif
# Line 452  else Line 531  else
531    
532      /* \g must be followed by a number, either plain or braced. If positive, it      /* \g must be followed by a number, either plain or braced. If positive, it
533      is an absolute backreference. If negative, it is a relative backreference.      is an absolute backreference. If negative, it is a relative backreference.
534      This is a Perl 5.10 feature. */      This is a Perl 5.10 feature. Perl 5.10 also supports \g{name} as a
535        reference to a named group. This is part of Perl's movement towards a
536        unified syntax for back references. As this is synonymous with \k{name}, we
537        fudge it up by pretending it really was \k. */
538    
539      case 'g':      case 'g':
540      if (ptr[1] == '{')      if (ptr[1] == '{')
541        {        {
542          const uschar *p;
543          for (p = ptr+2; *p != 0 && *p != '}'; p++)
544            if (*p != '-' && (digitab[*p] & ctype_digit) == 0) break;
545          if (*p != 0 && *p != '}')
546            {
547            c = -ESC_k;
548            break;
549            }
550        braced = TRUE;        braced = TRUE;
551        ptr++;        ptr++;
552        }        }
# Line 473  else Line 563  else
563      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
564        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
565    
566        if (c < 0)
567          {
568          *errorcodeptr = ERR61;
569          break;
570          }
571    
572      if (c == 0 || (braced && *(++ptr) != '}'))      if (c == 0 || (braced && *(++ptr) != '}'))
573        {        {
574        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
575        return 0;        break;
576        }        }
577    
578      if (negated)      if (negated)
# Line 484  else Line 580  else
580        if (c > bracount)        if (c > bracount)
581          {          {
582          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
583          return 0;          break;
584          }          }
585        c = bracount - (c - 1);        c = bracount - (c - 1);
586        }        }
# Line 513  else Line 609  else
609        c -= '0';        c -= '0';
610        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
611          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - '0';
612          if (c < 0)
613            {
614            *errorcodeptr = ERR61;
615            break;
616            }
617        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
618          {          {
619          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 562  else Line 663  else
663          if (c == 0 && cc == '0') continue;     /* Leading zeroes */          if (c == 0 && cc == '0') continue;     /* Leading zeroes */
664          count++;          count++;
665    
666  #if !EBCDIC    /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
667          if (cc >= 'a') cc -= 32;               /* Convert to upper case */          if (cc >= 'a') cc -= 32;               /* Convert to upper case */
668          c = (c << 4) + cc - ((cc < 'A')? '0' : ('A' - 10));          c = (c << 4) + cc - ((cc < 'A')? '0' : ('A' - 10));
669  #else          /* EBCDIC coding */  #else           /* EBCDIC coding */
670          if (cc >= 'a' && cc <= 'z') cc += 64;  /* Convert to upper case */          if (cc >= 'a' && cc <= 'z') cc += 64;  /* Convert to upper case */
671          c = (c << 4) + cc - ((cc >= '0')? '0' : ('A' - 10));          c = (c << 4) + cc - ((cc >= '0')? '0' : ('A' - 10));
672  #endif  #endif
# Line 589  else Line 690  else
690        {        {
691        int cc;                               /* Some compilers don't like ++ */        int cc;                               /* Some compilers don't like ++ */
692        cc = *(++ptr);                        /* in initializers */        cc = *(++ptr);                        /* in initializers */
693  #if !EBCDIC    /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
694        if (cc >= 'a') cc -= 32;              /* Convert to upper case */        if (cc >= 'a') cc -= 32;              /* Convert to upper case */
695        c = c * 16 + cc - ((cc < 'A')? '0' : ('A' - 10));        c = c * 16 + cc - ((cc < 'A')? '0' : ('A' - 10));
696  #else          /* EBCDIC coding */  #else           /* EBCDIC coding */
697        if (cc <= 'z') cc += 64;              /* Convert to upper case */        if (cc <= 'z') cc += 64;              /* Convert to upper case */
698        c = c * 16 + cc - ((cc >= '0')? '0' : ('A' - 10));        c = c * 16 + cc - ((cc >= '0')? '0' : ('A' - 10));
699  #endif  #endif
# Line 608  else Line 709  else
709      if (c == 0)      if (c == 0)
710        {        {
711        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
712        return 0;        break;
713        }        }
714    
715  #if !EBCDIC    /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
716      if (c >= 'a' && c <= 'z') c -= 32;      if (c >= 'a' && c <= 'z') c -= 32;
717      c ^= 0x40;      c ^= 0x40;
718  #else          /* EBCDIC coding */  #else           /* EBCDIC coding */
719      if (c >= 'a' && c <= 'z') c += 64;      if (c >= 'a' && c <= 'z') c += 64;
720      c ^= 0xC0;      c ^= 0xC0;
721  #endif  #endif
# Line 684  if (c == '{') Line 785  if (c == '{')
785      *negptr = TRUE;      *negptr = TRUE;
786      ptr++;      ptr++;
787      }      }
788    for (i = 0; i < sizeof(name) - 1; i++)    for (i = 0; i < (int)sizeof(name) - 1; i++)
789      {      {
790      c = *(++ptr);      c = *(++ptr);
791      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 713  top = _pcre_utt_size; Line 814  top = _pcre_utt_size;
814  while (bot < top)  while (bot < top)
815    {    {
816    i = (bot + top) >> 1;    i = (bot + top) >> 1;
817    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
818    if (c == 0)    if (c == 0)
819      {      {
820      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 887  for (; *ptr != 0; ptr++) Line 988  for (; *ptr != 0; ptr++)
988      {      {
989      while (*(++ptr) != ']')      while (*(++ptr) != ']')
990        {        {
991          if (*ptr == 0) return -1;
992        if (*ptr == '\\')        if (*ptr == '\\')
993          {          {
994          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 914  for (; *ptr != 0; ptr++) Line 1016  for (; *ptr != 0; ptr++)
1016    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
1017    
1018    if (*ptr != '(') continue;    if (*ptr != '(') continue;
1019    if (ptr[1] != '?')    if (ptr[1] != '?' && ptr[1] != '*')
1020      {      {
1021      count++;      count++;
1022      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1042  for (;;) Line 1144  for (;;)
1144    {    {
1145    int d;    int d;
1146    register int op = *cc;    register int op = *cc;
   
1147    switch (op)    switch (op)
1148      {      {
1149      case OP_CBRA:      case OP_CBRA:
# Line 1131  for (;;) Line 1232  for (;;)
1232    
1233      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1234      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1235        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1236      cc += 4;      cc += 4;
1237      break;      break;
1238    
# Line 1239  for (;;) Line 1341  for (;;)
1341      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1342      }      }
1343    
1344    /* 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
1345    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
1346    arrange to skip the extra bytes. */    two bytes of parameters. */
1347    
1348    else    else
1349      {      {
1350        switch(c)
1351          {
1352          case OP_TYPESTAR:
1353          case OP_TYPEMINSTAR:
1354          case OP_TYPEPLUS:
1355          case OP_TYPEMINPLUS:
1356          case OP_TYPEQUERY:
1357          case OP_TYPEMINQUERY:
1358          case OP_TYPEPOSSTAR:
1359          case OP_TYPEPOSPLUS:
1360          case OP_TYPEPOSQUERY:
1361          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1362          break;
1363    
1364          case OP_TYPEUPTO:
1365          case OP_TYPEMINUPTO:
1366          case OP_TYPEEXACT:
1367          case OP_TYPEPOSUPTO:
1368          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1369          break;
1370          }
1371    
1372        /* Add in the fixed length from the table */
1373    
1374      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1375    
1376      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1377      a multi-byte character. The length in the table is a minimum, so we have to
1378      arrange to skip the extra bytes. */
1379    
1380    #ifdef SUPPORT_UTF8
1381      if (utf8) switch(c)      if (utf8) switch(c)
1382        {        {
1383        case OP_CHAR:        case OP_CHAR:
# Line 1266  for (;;) Line 1398  for (;;)
1398        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1399        break;        break;
1400        }        }
1401    #endif
1402      }      }
1403    }    }
1404  }  }
# Line 1301  for (;;) Line 1434  for (;;)
1434    
1435    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1436    
1437    /* 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
1438    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
1439    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1440    
1441    else    else
1442      {      {
1443        switch(c)
1444          {
1445          case OP_TYPESTAR:
1446          case OP_TYPEMINSTAR:
1447          case OP_TYPEPLUS:
1448          case OP_TYPEMINPLUS:
1449          case OP_TYPEQUERY:
1450          case OP_TYPEMINQUERY:
1451          case OP_TYPEPOSSTAR:
1452          case OP_TYPEPOSPLUS:
1453          case OP_TYPEPOSQUERY:
1454          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1455          break;
1456    
1457          case OP_TYPEPOSUPTO:
1458          case OP_TYPEUPTO:
1459          case OP_TYPEMINUPTO:
1460          case OP_TYPEEXACT:
1461          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1462          break;
1463          }
1464    
1465        /* Add in the fixed length from the table */
1466    
1467      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1468    
1469        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1470        by a multi-byte character. The length in the table is a minimum, so we have
1471        to arrange to skip the extra bytes. */
1472    
1473    #ifdef SUPPORT_UTF8
1474      if (utf8) switch(c)      if (utf8) switch(c)
1475        {        {
1476        case OP_CHAR:        case OP_CHAR:
# Line 1329  for (;;) Line 1491  for (;;)
1491        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1492        break;        break;
1493        }        }
1494    #endif
1495      }      }
1496    }    }
1497  }  }
# Line 1366  for (code = first_significant_code(code Line 1529  for (code = first_significant_code(code
1529    
1530    c = *code;    c = *code;
1531    
1532    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    /* Groups with zero repeats can of course be empty; skip them. */
1533    
1534      if (c == OP_BRAZERO || c == OP_BRAMINZERO)
1535        {
1536        code += _pcre_OP_lengths[c];
1537        do code += GET(code, 1); while (*code == OP_ALT);
1538        c = *code;
1539        continue;
1540        }
1541    
1542      /* For other groups, scan the branches. */
1543    
1544      if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1545      {      {
1546      BOOL empty_branch;      BOOL empty_branch;
1547      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1382  for (code = first_significant_code(code Line 1557  for (code = first_significant_code(code
1557        }        }
1558      while (*code == OP_ALT);      while (*code == OP_ALT);
1559      if (!empty_branch) return FALSE;   /* All branches are non-empty */      if (!empty_branch) return FALSE;   /* All branches are non-empty */
1560        c = *code;
     /* Move past the KET and fudge things so that the increment in the "for"  
     above has no effect. */  
   
     c = OP_END;  
     code += 1 + LINK_SIZE - _pcre_OP_lengths[c];  
1561      continue;      continue;
1562      }      }
1563    
# Line 1395  for (code = first_significant_code(code Line 1565  for (code = first_significant_code(code
1565    
1566    switch (c)    switch (c)
1567      {      {
1568      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1569        cannot be represented just by a bit map. This includes negated single
1570        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1571        actual length is stored in the compiled code, so we must update "code"
1572        here. */
1573    
1574  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1575      case OP_XCLASS:      case OP_XCLASS:
1576      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1577      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1578  #endif  #endif
1579    
# Line 1461  for (code = first_significant_code(code Line 1635  for (code = first_significant_code(code
1635      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1636      return FALSE;      return FALSE;
1637    
1638        /* These are going to continue, as they may be empty, but we have to
1639        fudge the length for the \p and \P cases. */
1640    
1641        case OP_TYPESTAR:
1642        case OP_TYPEMINSTAR:
1643        case OP_TYPEPOSSTAR:
1644        case OP_TYPEQUERY:
1645        case OP_TYPEMINQUERY:
1646        case OP_TYPEPOSQUERY:
1647        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1648        break;
1649    
1650        /* Same for these */
1651    
1652        case OP_TYPEUPTO:
1653        case OP_TYPEMINUPTO:
1654        case OP_TYPEPOSUPTO:
1655        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1656        break;
1657    
1658      /* End of branch */      /* End of branch */
1659    
1660      case OP_KET:      case OP_KET:
# Line 1577  Returns: a value representing the na Line 1771  Returns: a value representing the na
1771  static int  static int
1772  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1773  {  {
1774    const char *pn = posix_names;
1775  register int yield = 0;  register int yield = 0;
1776  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1777    {    {
1778    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1779      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1780      pn += posix_name_lengths[yield] + 1;
1781    yield++;    yield++;
1782    }    }
1783  return -1;  return -1;
# Line 1623  adjust_recurse(uschar *group, int adjust Line 1819  adjust_recurse(uschar *group, int adjust
1819    uschar *save_hwm)    uschar *save_hwm)
1820  {  {
1821  uschar *ptr = group;  uschar *ptr = group;
1822    
1823  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1824    {    {
1825    int offset;    int offset;
# Line 1921  if (next >= 0) switch(op_code) Line 2118  if (next >= 0) switch(op_code)
2118    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2119    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2120    
2121      case OP_HSPACE:
2122      case OP_NOT_HSPACE:
2123      switch(next)
2124        {
2125        case 0x09:
2126        case 0x20:
2127        case 0xa0:
2128        case 0x1680:
2129        case 0x180e:
2130        case 0x2000:
2131        case 0x2001:
2132        case 0x2002:
2133        case 0x2003:
2134        case 0x2004:
2135        case 0x2005:
2136        case 0x2006:
2137        case 0x2007:
2138        case 0x2008:
2139        case 0x2009:
2140        case 0x200A:
2141        case 0x202f:
2142        case 0x205f:
2143        case 0x3000:
2144        return op_code != OP_HSPACE;
2145        default:
2146        return op_code == OP_HSPACE;
2147        }
2148    
2149      case OP_VSPACE:
2150      case OP_NOT_VSPACE:
2151      switch(next)
2152        {
2153        case 0x0a:
2154        case 0x0b:
2155        case 0x0c:
2156        case 0x0d:
2157        case 0x85:
2158        case 0x2028:
2159        case 0x2029:
2160        return op_code != OP_VSPACE;
2161        default:
2162        return op_code == OP_VSPACE;
2163        }
2164    
2165    default:    default:
2166    return FALSE;    return FALSE;
2167    }    }
# Line 1955  switch(op_code) Line 2196  switch(op_code)
2196      case ESC_W:      case ESC_W:
2197      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2198    
2199        case ESC_h:
2200        case ESC_H:
2201        switch(item)
2202          {
2203          case 0x09:
2204          case 0x20:
2205          case 0xa0:
2206          case 0x1680:
2207          case 0x180e:
2208          case 0x2000:
2209          case 0x2001:
2210          case 0x2002:
2211          case 0x2003:
2212          case 0x2004:
2213          case 0x2005:
2214          case 0x2006:
2215          case 0x2007:
2216          case 0x2008:
2217          case 0x2009:
2218          case 0x200A:
2219          case 0x202f:
2220          case 0x205f:
2221          case 0x3000:
2222          return -next != ESC_h;
2223          default:
2224          return -next == ESC_h;
2225          }
2226    
2227        case ESC_v:
2228        case ESC_V:
2229        switch(item)
2230          {
2231          case 0x0a:
2232          case 0x0b:
2233          case 0x0c:
2234          case 0x0d:
2235          case 0x85:
2236          case 0x2028:
2237          case 0x2029:
2238          return -next != ESC_v;
2239          default:
2240          return -next == ESC_v;
2241          }
2242    
2243      default:      default:
2244      return FALSE;      return FALSE;
2245      }      }
2246    
2247    case OP_DIGIT:    case OP_DIGIT:
2248    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2249             next == -ESC_h || next == -ESC_v;
2250    
2251    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2252    return next == -ESC_d;    return next == -ESC_d;
# Line 1969  switch(op_code) Line 2255  switch(op_code)
2255    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2256    
2257    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2258    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2259    
2260      case OP_HSPACE:
2261      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2262    
2263      case OP_NOT_HSPACE:
2264      return next == -ESC_h;
2265    
2266      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2267      case OP_VSPACE:
2268      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2269    
2270      case OP_NOT_VSPACE:
2271      return next == -ESC_v;
2272    
2273    case OP_WORDCHAR:    case OP_WORDCHAR:
2274    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2275    
2276    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2277    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2087  for (;; ptr++) Line 2386  for (;; ptr++)
2386    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2387    BOOL is_quantifier;    BOOL is_quantifier;
2388    BOOL is_recurse;    BOOL is_recurse;
2389      BOOL reset_bracount;
2390    int class_charcount;    int class_charcount;
2391    int class_lastchar;    int class_lastchar;
2392    int newoptions;    int newoptions;
2393    int recno;    int recno;
2394      int refsign;
2395    int skipbytes;    int skipbytes;
2396    int subreqbyte;    int subreqbyte;
2397    int subfirstbyte;    int subfirstbyte;
# Line 2123  for (;; ptr++) Line 2424  for (;; ptr++)
2424      */      */
2425    
2426      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2427    
2428        /* Paranoid check for integer overflow */
2429    
2430        if (OFLOW_MAX - *lengthptr < code - last_code)
2431          {
2432          *errorcodeptr = ERR20;
2433          goto FAILED;
2434          }
2435    
2436      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2437      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));
2438    
# Line 2235  for (;; ptr++) Line 2545  for (;; ptr++)
2545      *ptrptr = ptr;      *ptrptr = ptr;
2546      if (lengthptr != NULL)      if (lengthptr != NULL)
2547        {        {
2548          if (OFLOW_MAX - *lengthptr < code - last_code)
2549            {
2550            *errorcodeptr = ERR20;
2551            goto FAILED;
2552            }
2553        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2554        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2555        }        }
# Line 2297  for (;; ptr++) Line 2612  for (;; ptr++)
2612        goto FAILED;        goto FAILED;
2613        }        }
2614    
2615      /* 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,
2616        if the first few characters (either before or after ^) are \Q\E or \E we
2617        skip them too. This makes for compatibility with Perl. */
2618    
2619      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2620        for (;;)
2621        {        {
       negate_class = TRUE;  
2622        c = *(++ptr);        c = *(++ptr);
2623        }        if (c == '\\')
2624      else          {
2625        {          if (ptr[1] == 'E') ptr++;
2626        negate_class = FALSE;            else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2627                else break;
2628            }
2629          else if (!negate_class && c == '^')
2630            negate_class = TRUE;
2631          else break;
2632        }        }
2633    
2634      /* 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
# Line 2447  for (;; ptr++) Line 2769  for (;; ptr++)
2769        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
2770        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2771        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2772        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
2773        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2774    
2775        if (c == '\\')        if (c == '\\')
# Line 2467  for (;; ptr++) Line 2789  for (;; ptr++)
2789            else inescq = TRUE;            else inescq = TRUE;
2790            continue;            continue;
2791            }            }
2792            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2793    
2794          if (c < 0)          if (c < 0)
2795            {            {
# Line 2515  for (;; ptr++) Line 2838  for (;; ptr++)
2838            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2839                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2840    
2841              /* We need to deal with \H, \h, \V, and \v in both phases because
2842              they use extra memory. */
2843    
2844              if (-c == ESC_h)
2845                {
2846                SETBIT(classbits, 0x09); /* VT */
2847                SETBIT(classbits, 0x20); /* SPACE */
2848                SETBIT(classbits, 0xa0); /* NSBP */
2849    #ifdef SUPPORT_UTF8
2850                if (utf8)
2851                  {
2852                  class_utf8 = TRUE;
2853                  *class_utf8data++ = XCL_SINGLE;
2854                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2855                  *class_utf8data++ = XCL_SINGLE;
2856                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2857                  *class_utf8data++ = XCL_RANGE;
2858                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2859                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2860                  *class_utf8data++ = XCL_SINGLE;
2861                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2862                  *class_utf8data++ = XCL_SINGLE;
2863                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2864                  *class_utf8data++ = XCL_SINGLE;
2865                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2866                  }
2867    #endif
2868                continue;
2869                }
2870    
2871              if (-c == ESC_H)
2872                {
2873                for (c = 0; c < 32; c++)
2874                  {
2875                  int x = 0xff;
2876                  switch (c)
2877                    {
2878                    case 0x09/8: x ^= 1 << (0x09%8); break;
2879                    case 0x20/8: x ^= 1 << (0x20%8); break;
2880                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
2881                    default: break;
2882                    }
2883                  classbits[c] |= x;
2884                  }
2885    
2886    #ifdef SUPPORT_UTF8
2887                if (utf8)
2888                  {
2889                  class_utf8 = TRUE;
2890                  *class_utf8data++ = XCL_RANGE;
2891                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2892                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
2893                  *class_utf8data++ = XCL_RANGE;
2894                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
2895                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
2896                  *class_utf8data++ = XCL_RANGE;
2897                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
2898                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
2899                  *class_utf8data++ = XCL_RANGE;
2900                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
2901                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
2902                  *class_utf8data++ = XCL_RANGE;
2903                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
2904                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
2905                  *class_utf8data++ = XCL_RANGE;
2906                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
2907                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
2908                  *class_utf8data++ = XCL_RANGE;
2909                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
2910                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2911                  }
2912    #endif
2913                continue;
2914                }
2915    
2916              if (-c == ESC_v)
2917                {
2918                SETBIT(classbits, 0x0a); /* LF */
2919                SETBIT(classbits, 0x0b); /* VT */
2920                SETBIT(classbits, 0x0c); /* FF */
2921                SETBIT(classbits, 0x0d); /* CR */
2922                SETBIT(classbits, 0x85); /* NEL */
2923    #ifdef SUPPORT_UTF8
2924                if (utf8)
2925                  {
2926                  class_utf8 = TRUE;
2927                  *class_utf8data++ = XCL_RANGE;
2928                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
2929                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2930                  }
2931    #endif
2932                continue;
2933                }
2934    
2935              if (-c == ESC_V)
2936                {
2937                for (c = 0; c < 32; c++)
2938                  {
2939                  int x = 0xff;
2940                  switch (c)
2941                    {
2942                    case 0x0a/8: x ^= 1 << (0x0a%8);
2943                                 x ^= 1 << (0x0b%8);
2944                                 x ^= 1 << (0x0c%8);
2945                                 x ^= 1 << (0x0d%8);
2946                                 break;
2947                    case 0x85/8: x ^= 1 << (0x85%8); break;
2948                    default: break;
2949                    }
2950                  classbits[c] |= x;
2951                  }
2952    
2953    #ifdef SUPPORT_UTF8
2954                if (utf8)
2955                  {
2956                  class_utf8 = TRUE;
2957                  *class_utf8data++ = XCL_RANGE;
2958                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2959                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
2960                  *class_utf8data++ = XCL_RANGE;
2961                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2962                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2963                  }
2964    #endif
2965                continue;
2966                }
2967    
2968            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
2969    
2970  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2566  for (;; ptr++) Line 3016  for (;; ptr++)
3016    
3017        oldptr = ptr;        oldptr = ptr;
3018    
3019          /* Remember \r or \n */
3020    
3021          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3022    
3023          /* Check for range */
3024    
3025        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3026          {          {
3027          int d;          int d;
# Line 2633  for (;; ptr++) Line 3089  for (;; ptr++)
3089    
3090          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3091    
3092            /* Remember \r or \n */
3093    
3094            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3095    
3096          /* 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
3097          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3098          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2655  for (;; ptr++) Line 3115  for (;; ptr++)
3115              unsigned int origd = d;              unsigned int origd = d;
3116              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3117                {                {
3118                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3119                      ocd <= (unsigned int)d)
3120                    continue;                          /* Skip embedded ranges */
3121    
3122                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3123                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3124                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3125                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3126                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3127                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3128                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3129                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3130                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3131                  d = ocd;                  d = ocd;
3132                  continue;                  continue;
# Line 2782  for (;; ptr++) Line 3246  for (;; ptr++)
3246        goto FAILED;        goto FAILED;
3247        }        }
3248    
3249    
3250    /* This code has been disabled because it would mean that \s counts as
3251    an explicit \r or \n reference, and that's not really what is wanted. Now
3252    we set the flag only if there is a literal "\r" or "\n" in the class. */
3253    
3254    #if 0
3255        /* Remember whether \r or \n are in this class */
3256    
3257        if (negate_class)
3258          {
3259          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3260          }
3261        else
3262          {
3263          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3264          }
3265    #endif
3266    
3267    
3268      /* 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
3269      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
3270      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
3271      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3272      single-bytes only. This is an historical hangover. Maybe one day we can  
3273      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3274        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3275        operate on single-bytes only. This is an historical hangover. Maybe one day
3276        we can tidy these opcodes to handle multi-byte characters.
3277    
3278      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
3279      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 2797  for (;; ptr++) Line 3283  for (;; ptr++)
3283      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3284    
3285  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3286      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3287            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3288  #else  #else
3289      if (class_charcount == 1)      if (class_charcount == 1)
3290  #endif  #endif
# Line 3072  for (;; ptr++) Line 3556  for (;; ptr++)
3556        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3557        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3558    
3559        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3560    
3561        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3562    
# Line 3222  for (;; ptr++) Line 3706  for (;; ptr++)
3706        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3707        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3708    
3709        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3710    
3711        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3712          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3258  for (;; ptr++) Line 3742  for (;; ptr++)
3742          goto FAILED;          goto FAILED;
3743          }          }
3744    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3745        /* 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
3746        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
3747        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 3354  for (;; ptr++) Line 3830  for (;; ptr++)
3830          if (repeat_min > 1)          if (repeat_min > 1)
3831            {            {
3832            /* 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
3833            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3834              potential integer overflow. */
3835    
3836            if (lengthptr != NULL)            if (lengthptr != NULL)
3837              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3838                int delta = (repeat_min - 1)*length_prevgroup;
3839                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3840                                                                (double)INT_MAX ||
3841                    OFLOW_MAX - *lengthptr < delta)
3842                  {
3843                  *errorcodeptr = ERR20;
3844                  goto FAILED;
3845                  }
3846                *lengthptr += delta;
3847                }
3848    
3849            /* This is compiling for real */            /* This is compiling for real */
3850    
# Line 3395  for (;; ptr++) Line 3882  for (;; ptr++)
3882          /* 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
3883          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
3884          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
3885          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3886            paranoid checks to avoid integer overflow. */
3887    
3888          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3889            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3890              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3891                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3892              if ((double)repeat_max *
3893                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3894                      > (double)INT_MAX ||
3895                  OFLOW_MAX - *lengthptr < delta)
3896                {
3897                *errorcodeptr = ERR20;
3898                goto FAILED;
3899                }
3900              *lengthptr += delta;
3901              }
3902    
3903          /* This is compiling for real */          /* This is compiling for real */
3904    
# Line 3551  for (;; ptr++) Line 4050  for (;; ptr++)
4050      /* ===================================================================*/      /* ===================================================================*/
4051      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4052      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4053      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. */  
4054    
4055      case '(':      case '(':
4056      newoptions = options;      newoptions = options;
4057      skipbytes = 0;      skipbytes = 0;
4058      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4059      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4060        reset_bracount = FALSE;
4061    
4062        /* First deal with various "verbs" that can be introduced by '*'. */
4063    
4064        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4065          {
4066          int i, namelen;
4067          const char *vn = verbnames;
4068          const uschar *name = ++ptr;
4069          previous = NULL;
4070          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4071          if (*ptr == ':')
4072            {
4073            *errorcodeptr = ERR59;   /* Not supported */
4074            goto FAILED;
4075            }
4076          if (*ptr != ')')
4077            {
4078            *errorcodeptr = ERR60;
4079            goto FAILED;
4080            }
4081          namelen = ptr - name;
4082          for (i = 0; i < verbcount; i++)
4083            {
4084            if (namelen == verbs[i].len &&
4085                strncmp((char *)name, vn, namelen) == 0)
4086              {
4087              *code = verbs[i].op;
4088              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4089              break;
4090              }
4091            vn += verbs[i].len + 1;
4092            }
4093          if (i < verbcount) continue;
4094          *errorcodeptr = ERR60;
4095          goto FAILED;
4096          }
4097    
4098        /* Deal with the extended parentheses; all are introduced by '?', and the
4099        appearance of any of them means that this is not a capturing group. */
4100    
4101      if (*(++ptr) == '?')      else if (*ptr == '?')
4102        {        {
4103        int i, set, unset, namelen;        int i, set, unset, namelen;
4104        int *optset;        int *optset;
# Line 3582  for (;; ptr++) Line 4119  for (;; ptr++)
4119    
4120    
4121          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4122            case '|':                 /* Reset capture count for each branch */
4123            reset_bracount = TRUE;
4124            /* Fall through */
4125    
4126            /* ------------------------------------------------------------ */
4127          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
4128          bravalue = OP_BRA;          bravalue = OP_BRA;
4129          ptr++;          ptr++;
# Line 3617  for (;; ptr++) Line 4159  for (;; ptr++)
4159    
4160          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
4161          skipbytes = 3;          skipbytes = 3;
4162            refsign = -1;
4163    
4164          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4165    
# Line 3640  for (;; ptr++) Line 4183  for (;; ptr++)
4183            terminator = '\'';            terminator = '\'';
4184            ptr++;            ptr++;
4185            }            }
4186          else terminator = 0;          else
4187              {
4188              terminator = 0;
4189              if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);
4190              }
4191    
4192          /* 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 */
4193    
# Line 3676  for (;; ptr++) Line 4223  for (;; ptr++)
4223          if (lengthptr != NULL) break;          if (lengthptr != NULL) break;
4224    
4225          /* 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
4226          reference. */          reference. If the string started with "+" or "-" we require the rest to
4227            be digits, in which case recno will be set. */
4228    
4229            if (refsign > 0)
4230              {
4231              if (recno <= 0)
4232                {
4233                *errorcodeptr = ERR58;
4234                goto FAILED;
4235                }
4236              if (refsign == '-')
4237                {
4238                recno = cd->bracount - recno + 1;
4239                if (recno <= 0)
4240                  {
4241                  *errorcodeptr = ERR15;
4242                  goto FAILED;
4243                  }
4244                }
4245              else recno += cd->bracount;
4246              PUT2(code, 2+LINK_SIZE, recno);
4247              break;
4248              }
4249    
4250            /* Otherwise (did not start with "+" or "-"), start by looking for the
4251            name. */
4252    
4253          slot = cd->name_table;          slot = cd->name_table;
4254          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 3768  for (;; ptr++) Line 4340  for (;; ptr++)
4340    
4341          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4342          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4343          ptr++;          ptr++;
4344            if (*ptr == ')')          /* Optimize (?!) */
4345              {
4346              *code++ = OP_FAIL;
4347              previous = NULL;
4348              continue;
4349              }
4350            bravalue = OP_ASSERT_NOT;
4351          break;          break;
4352    
4353    
# Line 3995  for (;; ptr++) Line 4573  for (;; ptr++)
4573    
4574    
4575          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4576            case '-': case '+':
4577          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */
4578          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case '5': case '6': case '7': case '8': case '9':   /* subroutine */
4579            {            {
4580            const uschar *called;            const uschar *called;
4581    
4582              if ((refsign = *ptr) == '+') ptr++;
4583              else if (refsign == '-')
4584                {
4585                if ((digitab[ptr[1]] & ctype_digit) == 0)
4586                  goto OTHER_CHAR_AFTER_QUERY;
4587                ptr++;
4588                }
4589    
4590            recno = 0;            recno = 0;
4591            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4592              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - '0';
4593    
4594            if (*ptr != ')')            if (*ptr != ')')
4595              {              {
4596              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4597              goto FAILED;              goto FAILED;
4598              }              }
4599    
4600              if (refsign == '-')
4601                {
4602                if (recno == 0)
4603                  {
4604                  *errorcodeptr = ERR58;
4605                  goto FAILED;
4606                  }
4607                recno = cd->bracount - recno + 1;
4608                if (recno <= 0)
4609                  {
4610                  *errorcodeptr = ERR15;
4611                  goto FAILED;
4612                  }
4613                }
4614              else if (refsign == '+')
4615                {
4616                if (recno == 0)
4617                  {
4618                  *errorcodeptr = ERR58;
4619                  goto FAILED;
4620                  }
4621                recno += cd->bracount;
4622                }
4623    
4624            /* Come here from code above that handles a named recursion */            /* Come here from code above that handles a named recursion */
4625    
4626            HANDLE_RECURSION:            HANDLE_RECURSION:
# Line 4080  for (;; ptr++) Line 4693  for (;; ptr++)
4693    
4694          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4695          default:              /* Other characters: check option setting */          default:              /* Other characters: check option setting */
4696            OTHER_CHAR_AFTER_QUERY:
4697          set = unset = 0;          set = unset = 0;
4698          optset = &set;          optset = &set;
4699    
# Line 4091  for (;; ptr++) Line 4705  for (;; ptr++)
4705    
4706              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4707              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4708              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4709              break;              break;
4710    
4711              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4214  for (;; ptr++) Line 4828  for (;; ptr++)
4828           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4829           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4830            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4831             reset_bracount,               /* True if (?| group */
4832           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4833           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4834           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4230  for (;; ptr++) Line 4845  for (;; ptr++)
4845      is on the bracket. */      is on the bracket. */
4846    
4847      /* If this is a conditional bracket, check that there are no more than      /* If this is a conditional bracket, check that there are no more than
4848      two branches in the group, or just one if it's a DEFINE group. */      two branches in the group, or just one if it's a DEFINE group. We do this
4849        in the real compile phase, not in the pre-pass, where the whole group may
4850        not be available. */
4851    
4852      if (bravalue == OP_COND)      if (bravalue == OP_COND && lengthptr == NULL)
4853        {        {
4854        uschar *tc = code;        uschar *tc = code;
4855        int condcount = 0;        int condcount = 0;
# Line 4279  for (;; ptr++) Line 4896  for (;; ptr++)
4896        goto FAILED;        goto FAILED;
4897        }        }
4898    
4899      /* In the pre-compile phase, update the length by the length of the nested      /* In the pre-compile phase, update the length by the length of the group,
4900      group, less the brackets at either end. Then reduce the compiled code to      less the brackets at either end. Then reduce the compiled code to just a
4901      just the brackets so that it doesn't use much memory if it is duplicated by      set of non-capturing brackets so that it doesn't use much memory if it is
4902      a quantifier. */      duplicated by a quantifier.*/
4903    
4904      if (lengthptr != NULL)      if (lengthptr != NULL)
4905        {        {
4906          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4907            {
4908            *errorcodeptr = ERR20;
4909            goto FAILED;
4910            }
4911        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4912        code++;        *code++ = OP_BRA;
4913        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4914        *code++ = OP_KET;        *code++ = OP_KET;
4915        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4916          break;    /* No need to waste time with special character handling */
4917        }        }
4918    
4919      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4920    
4921      else code = tempcode;      code = tempcode;
4922    
4923      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4924      relevant. */      relevant. */
# Line 4392  for (;; ptr++) Line 5015  for (;; ptr++)
5015        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
5016        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
5017    
5018        /* \k<name> or \k'name' is a back reference by name (Perl syntax) */        /* \k<name> or \k'name' is a back reference by name (Perl syntax).
5019          We also support \k{name} (.NET syntax) */
5020    
5021        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\''))        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\'' || ptr[1] == '{'))
5022          {          {
5023          is_recurse = FALSE;          is_recurse = FALSE;
5024          terminator = (*(++ptr) == '<')? '>' : '\'';          terminator = (*(++ptr) == '<')? '>' : (*ptr == '\'')? '\'' : '}';
5025          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
5026          }          }
5027    
# Line 4498  for (;; ptr++) Line 5122  for (;; ptr++)
5122      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5123      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5124    
5125        /* Remember if \r or \n were seen */
5126    
5127        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5128          cd->external_flags |= PCRE_HASCRORLF;
5129    
5130      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5131      byte, set it from this character, but revert to none on a zero repeat.      byte, set it from this character, but revert to none on a zero repeat.
5132      Otherwise, leave the firstbyte value alone, and don't change it on a zero      Otherwise, leave the firstbyte value alone, and don't change it on a zero
# Line 4563  This function is used during the pre-com Line 5192  This function is used during the pre-com
5192  out the amount of memory needed, as well as during the real compile phase. The  out the amount of memory needed, as well as during the real compile phase. The
5193  value of lengthptr distinguishes the two phases.  value of lengthptr distinguishes the two phases.
5194    
5195  Argument:  Arguments:
5196    options        option bits, including any changes for this subpattern    options        option bits, including any changes for this subpattern
5197    oldims         previous settings of ims option bits    oldims         previous settings of ims option bits
5198    codeptr        -> the address of the current code pointer    codeptr        -> the address of the current code pointer
5199    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5200    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5201    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5202      reset_bracount TRUE to reset the count for each branch
5203    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5204    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5205    reqbyteptr     place to put the last required character, or a negative number    reqbyteptr     place to put the last required character, or a negative number
# Line 4583  Returns: TRUE on success Line 5213  Returns: TRUE on success
5213    
5214  static BOOL  static BOOL
5215  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5216    int *errorcodeptr, BOOL lookbehind, int skipbytes, int *firstbyteptr,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5217    int *reqbyteptr, branch_chain *bcptr, compile_data *cd, int *lengthptr)    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5218      int *lengthptr)
5219  {  {
5220  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
5221  uschar *code = *codeptr;  uschar *code = *codeptr;
# Line 4594  uschar *reverse_count = NULL; Line 5225  uschar *reverse_count = NULL;
5225  int firstbyte, reqbyte;  int firstbyte, reqbyte;
5226  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
5227  int length;  int length;
5228    int orig_bracount;
5229    int max_bracount;
5230  branch_chain bc;  branch_chain bc;
5231    
5232  bc.outer = bcptr;  bc.outer = bcptr;
# Line 4622  code += 1 + LINK_SIZE + skipbytes; Line 5255  code += 1 + LINK_SIZE + skipbytes;
5255    
5256  /* Loop for each alternative branch */  /* Loop for each alternative branch */
5257    
5258    orig_bracount = max_bracount = cd->bracount;
5259  for (;;)  for (;;)
5260    {    {
5261      /* For a (?| group, reset the capturing bracket count so that each branch
5262      uses the same numbers. */
5263    
5264      if (reset_bracount) cd->bracount = orig_bracount;
5265    
5266    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5267    
5268    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4653  for (;;) Line 5292  for (;;)
5292      return FALSE;      return FALSE;
5293      }      }
5294    
5295      /* Keep the highest bracket count in case (?| was used and some branch
5296      has fewer than the rest. */
5297    
5298      if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5299    
5300    /* In the real compile phase, there is some post-processing to be done. */    /* In the real compile phase, there is some post-processing to be done. */
5301    
5302    if (lengthptr == NULL)    if (lengthptr == NULL)
# Line 4716  for (;;) Line 5360  for (;;)
5360        }        }
5361      }      }
5362    
5363    /* Reached end of expression, either ')' or end of pattern. Go back through    /* Reached end of expression, either ')' or end of pattern. In the real
5364    the alternative branches and reverse the chain of offsets, with the field in    compile phase, go back through the alternative branches and reverse the chain
5365    the BRA item now becoming an offset to the first alternative. If there are    of offsets, with the field in the BRA item now becoming an offset to the
5366    no alternatives, it points to the end of the group. The length in the    first alternative. If there are no alternatives, it points to the end of the
5367    terminating ket is always the length of the whole bracketed item. If any of    group. The length in the terminating ket is always the length of the whole
5368    the ims options were changed inside the group, compile a resetting op-code    bracketed item. If any of the ims options were changed inside the group,
5369    following, except at the very end of the pattern. Return leaving the pointer    compile a resetting op-code following, except at the very end of the pattern.
5370    at the terminating char. */    Return leaving the pointer at the terminating char. */
5371    
5372    if (*ptr != '|')    if (*ptr != '|')
5373      {      {
5374      int branch_length = code - last_branch;      if (lengthptr == NULL)
     do  
5375        {        {
5376        int prev_length = GET(last_branch, 1);        int branch_length = code - last_branch;
5377        PUT(last_branch, 1, branch_length);        do
5378        branch_length = prev_length;          {
5379        last_branch -= branch_length;          int prev_length = GET(last_branch, 1);
5380            PUT(last_branch, 1, branch_length);
5381            branch_length = prev_length;
5382            last_branch -= branch_length;
5383            }
5384          while (branch_length > 0);
5385        }        }
     while (branch_length > 0);  
5386    
5387      /* Fill in the ket */      /* Fill in the ket */
5388    
# Line 4752  for (;;) Line 5399  for (;;)
5399        length += 2;        length += 2;
5400        }        }
5401    
5402        /* Retain the highest bracket number, in case resetting was used. */
5403    
5404        cd->bracount = max_bracount;
5405    
5406      /* Set values to pass back */      /* Set values to pass back */
5407    
5408      *codeptr = code;      *codeptr = code;
5409      *ptrptr = ptr;      *ptrptr = ptr;
5410      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5411      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5412      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5413          {
5414          if (OFLOW_MAX - *lengthptr < length)
5415            {
5416            *errorcodeptr = ERR20;
5417            return FALSE;
5418            }
5419          *lengthptr += length;
5420          }
5421      return TRUE;      return TRUE;
5422      }      }
5423    
5424    /* Another branch follows; insert an "or" node. Its length field points back    /* Another branch follows. In the pre-compile phase, we can move the code
5425      pointer back to where it was for the start of the first branch. (That is,
5426      pretend that each branch is the only one.)
5427    
5428      In the real compile phase, insert an ALT node. Its length field points back
5429    to the previous branch while the bracket remains open. At the end the chain    to the previous branch while the bracket remains open. At the end the chain
5430    is reversed. It's done like this so that the start of the bracket has a    is reversed. It's done like this so that the start of the bracket has a
5431    zero offset until it is closed, making it possible to detect recursion. */    zero offset until it is closed, making it possible to detect recursion. */
5432    
5433    *code = OP_ALT;    if (lengthptr != NULL)
5434    PUT(code, 1, code - last_branch);      {
5435    bc.current = last_branch = code;      code = *codeptr + 1 + LINK_SIZE + skipbytes;
5436    code += 1 + LINK_SIZE;      length += 1 + LINK_SIZE;
5437        }
5438      else
5439        {
5440        *code = OP_ALT;
5441        PUT(code, 1, code - last_branch);
5442        bc.current = last_branch = code;
5443        code += 1 + LINK_SIZE;
5444        }
5445    
5446    ptr++;    ptr++;
   length += 1 + LINK_SIZE;  
5447    }    }
5448  /* Control never reaches here */  /* Control never reaches here */
5449  }  }
# Line 5039  Returns: pointer to compiled data Line 5710  Returns: pointer to compiled data
5710                  with errorptr and erroroffset set                  with errorptr and erroroffset set
5711  */  */
5712    
5713  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5714  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
5715    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
5716  {  {
# Line 5047  return pcre_compile2(pattern, options, N Line 5718  return pcre_compile2(pattern, options, N
5718  }  }
5719    
5720    
5721  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5722  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
5723    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
5724  {  {
# Line 5055  real_pcre *re; Line 5726  real_pcre *re;
5726  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5727  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5728  int errorcode = 0;  int errorcode = 0;
5729    int skipatstart = 0;
5730  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5731  BOOL utf8;  BOOL utf8;
5732  #endif  #endif
# Line 5096  if (errorcodeptr != NULL) *errorcodeptr Line 5768  if (errorcodeptr != NULL) *errorcodeptr
5768  if (erroroffset == NULL)  if (erroroffset == NULL)
5769    {    {
5770    errorcode = ERR16;    errorcode = ERR16;
5771    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5772    }    }
5773    
5774  *erroroffset = 0;  *erroroffset = 0;
# Line 5109  if (utf8 && (options & PCRE_NO_UTF8_CHEC Line 5781  if (utf8 && (options & PCRE_NO_UTF8_CHEC
5781       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)
5782    {    {
5783    errorcode = ERR44;    errorcode = ERR44;
5784    goto PCRE_UTF8_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5785    }    }
5786  #else  #else
5787  if ((options & PCRE_UTF8) != 0)  if ((options & PCRE_UTF8) != 0)
# Line 5133  cd->fcc = tables + fcc_offset; Line 5805  cd->fcc = tables + fcc_offset;
5805  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5806  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5807    
5808    /* Check for global one-time settings at the start of the pattern, and remember
5809    the offset for later. */
5810    
5811    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5812      {
5813      int newnl = 0;
5814      int newbsr = 0;
5815    
5816      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5817        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5818      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5819        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5820      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5821        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5822      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5823        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5824      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5825        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5826    
5827      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5828        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5829      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5830        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5831    
5832      if (newnl != 0)
5833        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5834      else if (newbsr != 0)
5835        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5836      else break;
5837      }
5838    
5839    /* Check validity of \R options. */
5840    
5841    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5842      {
5843      case 0:
5844      case PCRE_BSR_ANYCRLF:
5845      case PCRE_BSR_UNICODE:
5846      break;
5847      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5848      }
5849    
5850  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5851  current code allows for fixed one- or two-byte sequences, plus "any". */  current code allows for fixed one- or two-byte sequences, plus "any" and
5852    "anycrlf". */
5853    
5854  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5855    {    {
5856    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5857    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5858    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5859    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
5860         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;
5861    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
5862      case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
5863    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5864    }    }
5865    
5866  if (newline < 0)  if (newline == -2)
5867      {
5868      cd->nltype = NLTYPE_ANYCRLF;
5869      }
5870    else if (newline < 0)
5871    {    {
5872    cd->nltype = NLTYPE_ANY;    cd->nltype = NLTYPE_ANY;
5873    }    }
# Line 5196  cd->hwm = cworkspace; Line 5916  cd->hwm = cworkspace;
5916  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5917  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5918  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5919  cd->external_options = options;  cd->external_options = options;
5920    cd->external_flags = 0;
5921    
5922  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we
5923  don't need to look at the result of the function here. The initial options have  don't need to look at the result of the function here. The initial options have
# Line 5205  been put into the cd block so that they Line 5925  been put into the cd block so that they
5925  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5926  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5927    
5928    ptr += skipatstart;
5929  code = cworkspace;  code = cworkspace;
5930  *code = OP_BRA;  *code = OP_BRA;
5931  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
5932    &code, &ptr, &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length);    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
5933      &length);
5934  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
5935    
5936  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 5234  if (re == NULL) Line 5956  if (re == NULL)
5956    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
5957    }    }
5958    
5959  /* Put in the magic number, and save the sizes, initial options, and character  /* Put in the magic number, and save the sizes, initial options, internal
5960  table pointer. NULL is used for the default character tables. The nullpad field  flags, and character table pointer. NULL is used for the default character
5961  is at the end; it's there to help in the case when a regex compiled on a system  tables. The nullpad field is at the end; it's there to help in the case when a
5962  with 4-byte pointers is run on another with 8-byte pointers. */  regex compiled on a system with 4-byte pointers is run on another with 8-byte
5963    pointers. */
5964    
5965  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
5966  re->size = size;  re->size = size;
5967  re->options = cd->external_options;  re->options = cd->external_options;
5968    re->flags = cd->external_flags;
5969  re->dummy1 = 0;  re->dummy1 = 0;
5970  re->first_byte = 0;  re->first_byte = 0;
5971  re->req_byte = 0;  re->req_byte = 0;
# Line 5266  codestart = cd->name_table + re->name_en Line 5990  codestart = cd->name_table + re->name_en
5990  cd->start_code = codestart;  cd->start_code = codestart;
5991  cd->hwm = cworkspace;  cd->hwm = cworkspace;
5992  cd->req_varyopt = 0;  cd->req_varyopt = 0;
5993  cd->nopartial = FALSE;  cd->had_accept = FALSE;
5994    
5995  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
5996  error, errorcode will be set non-zero, so we don't need to look at the result  error, errorcode will be set non-zero, so we don't need to look at the result
5997  of the function here. */  of the function here. */
5998    
5999  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6000  code = (uschar *)codestart;  code = (uschar *)codestart;
6001  *code = OP_BRA;  *code = OP_BRA;
6002  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6003    &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6004  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6005  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6006    re->flags = cd->external_flags;
6007    
6008  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6009    
6010  /* If not reached end of pattern on success, there's an excess bracket. */  /* If not reached end of pattern on success, there's an excess bracket. */
6011    
# Line 5321  if (errorcode != 0) Line 6046  if (errorcode != 0)
6046    (pcre_free)(re);    (pcre_free)(re);
6047    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6048    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6049  #ifdef SUPPORT_UTF8    PCRE_EARLY_ERROR_RETURN2:
6050    PCRE_UTF8_ERROR_RETURN:    *errorptr = find_error_text(errorcode);
 #endif  
   *errorptr = error_texts[errorcode];  
6051    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6052    return NULL;    return NULL;
6053    }    }
# Line 5353  if ((re->options & PCRE_ANCHORED) == 0) Line 6076  if ((re->options & PCRE_ANCHORED) == 0)
6076        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6077        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6078           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6079        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6080        }        }
6081      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6082        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6083      }      }
6084    }    }
6085    
# Line 5370  if (reqbyte >= 0 && Line 6093  if (reqbyte >= 0 &&
6093    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6094    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6095      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6096    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6097    }    }
6098    
6099  /* Print out the compiled data if debugging is enabled. This is never the  /* Print out the compiled data if debugging is enabled. This is never the
# Line 5381  case when building a production library. Line 6104  case when building a production library.
6104  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6105    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6106    
6107  if (re->options != 0)  printf("Options=%08x\n", re->options);
   {  
   printf("%s%s%s%s%s%s%s%s%s\n",  
     ((re->options & PCRE_NOPARTIAL) != 0)? "nopartial " : "",  
     ((re->options & PCRE_ANCHORED) != 0)? "anchored " : "",  
     ((re->options & PCRE_CASELESS) != 0)? "caseless " : "",  
     ((re->options & PCRE_EXTENDED) != 0)? "extended " : "",  
     ((re->options & PCRE_MULTILINE) != 0)? "multiline " : "",  
     ((re->options & PCRE_DOTALL) != 0)? "dotall " : "",  
     ((re->options & PCRE_DOLLAR_ENDONLY) != 0)? "endonly " : "",  
     ((re->options & PCRE_EXTRA) != 0)? "extra " : "",  
     ((re->options & PCRE_UNGREEDY) != 0)? "ungreedy " : "");  
   }  
6108    
6109  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6110    {    {
6111    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6112    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5404  if ((re->options & PCRE_FIRSTSET) != 0) Line 6115  if ((re->options & PCRE_FIRSTSET) != 0)
6115      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6116    }    }
6117    
6118  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6119    {    {
6120    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6121    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5413  if ((re->options & PCRE_REQCHSET) != 0) Line 6124  if ((re->options & PCRE_REQCHSET) != 0)
6124      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
6125    }    }
6126    
6127  pcre_printint(re, stdout);  pcre_printint(re, stdout, TRUE);
6128    
6129  /* This check is done here in the debugging case so that the code that  /* This check is done here in the debugging case so that the code that
6130  was compiled can be seen. */  was compiled can be seen. */
# Line 5421  was compiled can be seen. */ Line 6132  was compiled can be seen. */
6132  if (code - codestart > length)  if (code - codestart > length)
6133    {    {
6134    (pcre_free)(re);    (pcre_free)(re);
6135    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6136    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6137    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6138    return NULL;    return NULL;

Legend:
Removed from v.96  
changed lines
  Added in v.240

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12