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

Diff of /code/trunk/pcre_compile.c

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

revision 226 by ph10, Tue Aug 21 11:46:08 2007 UTC revision 268 by ph10, Thu Nov 15 10:28:09 2007 UTC
# Line 43  supporting internal functions that are n Line 43  supporting internal functions that are n
43    
44    
45  #ifdef HAVE_CONFIG_H  #ifdef HAVE_CONFIG_H
46  #include <config.h>  #include "config.h"
47  #endif  #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
# Line 140  static const short int escapes[] = { Line 140  static const short int escapes[] = {
140  #endif  #endif
141    
142    
143  /* Table of special "verbs" like (*PRUNE) */  /* Table of special "verbs" like (*PRUNE). This is a short table, so it is
144    searched linearly. Put all the names into a single string, in order to reduce
145    the number of relocations when a shared library is dynamically linked. */
146    
147  typedef struct verbitem {  typedef struct verbitem {
   const char *name;  
148    int   len;    int   len;
149    int   op;    int   op;
150  } verbitem;  } 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[] = {  static verbitem verbs[] = {
162    { "ACCEPT", 6, OP_ACCEPT },    { 6, OP_ACCEPT },
163    { "COMMIT", 6, OP_COMMIT },    { 6, OP_COMMIT },
164    { "F",      1, OP_FAIL },    { 1, OP_FAIL },
165    { "FAIL",   4, OP_FAIL },    { 4, OP_FAIL },
166    { "PRUNE",  5, OP_PRUNE },    { 5, OP_PRUNE },
167    { "SKIP",   4, OP_SKIP  },    { 4, OP_SKIP  },
168    { "THEN",   4, OP_THEN  }    { 4, OP_THEN  }
169  };  };
170    
171  static int verbcount = sizeof(verbs)/sizeof(verbitem);  static int verbcount = sizeof(verbs)/sizeof(verbitem);
172    
173    
174  /* Tables of names of POSIX character classes and their lengths. The list is  /* Tables of names of POSIX character classes and their lengths. The names are
175  terminated by a zero length entry. The first three must be alpha, lower, upper,  now all in a single string, to reduce the number of relocations when a shared
176  as this is assumed for handling case independence. */  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  static const char *const posix_names[] = {  for handling case independence. */
179    "alpha", "lower", "upper",  
180    "alnum", "ascii", "blank", "cntrl", "digit", "graph",  static const char posix_names[] =
181    "print", "punct", "space", "word",  "xdigit" };    "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 207  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 is 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",    /** DEAD **/    "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 a braced name or 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",    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number\0"
300    "(*VERB) with an argument is not supported",    "(*VERB) with an argument is not supported\0"
301    /* 60 */    /* 60 */
302    "(*VERB) not recognized",    "(*VERB) not recognized\0"
303    "number is too big"    "number is too big\0"
304  };    "subpattern name expected after (?&";
305    
306    
307  /* 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 420  static BOOL Line 437  static BOOL
437    
438    
439  /*************************************************  /*************************************************
440    *            Find an error text                  *
441    *************************************************/
442    
443    /* The error texts are now all in one long string, to save on relocations. As
444    some of the text is of unknown length, we can't use a table of offsets.
445    Instead, just count through the strings. This is not a performance issue
446    because it happens only when there has been a compilation error.
447    
448    Argument:   the error number
449    Returns:    pointer to the error string
450    */
451    
452    static const char *
453    find_error_text(int n)
454    {
455    const char *s = error_texts;
456    for (; n > 0; n--) while (*s++ != 0);
457    return s;
458    }
459    
460    
461    /*************************************************
462  *            Handle escapes                      *  *            Handle escapes                      *
463  *************************************************/  *************************************************/
464    
# Line 776  top = _pcre_utt_size; Line 815  top = _pcre_utt_size;
815  while (bot < top)  while (bot < top)
816    {    {
817    i = (bot + top) >> 1;    i = (bot + top) >> 1;
818    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
819    if (c == 0)    if (c == 0)
820      {      {
821      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 1596  for (code = first_significant_code(code Line 1635  for (code = first_significant_code(code
1635      case OP_TYPEPOSPLUS:      case OP_TYPEPOSPLUS:
1636      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1637      return FALSE;      return FALSE;
1638    
1639      /* These are going to continue, as they may be empty, but we have to      /* These are going to continue, as they may be empty, but we have to
1640      fudge the length for the \p and \P cases. */      fudge the length for the \p and \P cases. */
1641    
1642      case OP_TYPESTAR:      case OP_TYPESTAR:
1643      case OP_TYPEMINSTAR:      case OP_TYPEMINSTAR:
1644      case OP_TYPEPOSSTAR:      case OP_TYPEPOSSTAR:
# Line 1607  for (code = first_significant_code(code Line 1646  for (code = first_significant_code(code
1646      case OP_TYPEMINQUERY:      case OP_TYPEMINQUERY:
1647      case OP_TYPEPOSQUERY:      case OP_TYPEPOSQUERY:
1648      if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;      if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1649      break;      break;
1650    
1651      /* Same for these */      /* Same for these */
1652    
1653      case OP_TYPEUPTO:      case OP_TYPEUPTO:
1654      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
1655      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
# Line 1733  Returns:     a value representing the na Line 1772  Returns:     a value representing the na
1772  static int  static int
1773  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1774  {  {
1775    const char *pn = posix_names;
1776  register int yield = 0;  register int yield = 0;
1777  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1778    {    {
1779    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1780      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1781      pn += posix_name_lengths[yield] + 1;
1782    yield++;    yield++;
1783    }    }
1784  return -1;  return -1;
# Line 2343  req_caseopt = ((options & PCRE_CASELESS) Line 2384  req_caseopt = ((options & PCRE_CASELESS)
2384  for (;; ptr++)  for (;; ptr++)
2385    {    {
2386    BOOL negate_class;    BOOL negate_class;
2387      BOOL should_flip_negation;
2388    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2389    BOOL is_quantifier;    BOOL is_quantifier;
2390    BOOL is_recurse;    BOOL is_recurse;
# Line 2591  for (;; ptr++) Line 2633  for (;; ptr++)
2633        else break;        else break;
2634        }        }
2635    
2636        /* If a class contains a negative special such as \S, we need to flip the
2637        negation flag at the end, so that support for characters > 255 works
2638        correctly (they are all included in the class). */
2639    
2640        should_flip_negation = FALSE;
2641    
2642      /* 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
2643      of just a single character (as long as it's < 256). However, For higher      of just a single character (as long as it's < 256). However, For higher
2644      valued UTF-8 characters, we don't yet do any optimization. */      valued UTF-8 characters, we don't yet do any optimization. */
# Line 2663  for (;; ptr++) Line 2711  for (;; ptr++)
2711          if (*ptr == '^')          if (*ptr == '^')
2712            {            {
2713            local_negate = TRUE;            local_negate = TRUE;
2714              should_flip_negation = TRUE;  /* Note negative special */
2715            ptr++;            ptr++;
2716            }            }
2717    
# Line 2765  for (;; ptr++) Line 2814  for (;; ptr++)
2814              continue;              continue;
2815    
2816              case ESC_D:              case ESC_D:
2817                should_flip_negation = TRUE;
2818              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2819              continue;              continue;
2820    
# Line 2773  for (;; ptr++) Line 2823  for (;; ptr++)
2823              continue;              continue;
2824    
2825              case ESC_W:              case ESC_W:
2826                should_flip_negation = TRUE;
2827              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2828              continue;              continue;
2829    
# Line 2782  for (;; ptr++) Line 2833  for (;; ptr++)
2833              continue;              continue;
2834    
2835              case ESC_S:              case ESC_S:
2836                should_flip_negation = TRUE;
2837              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2838              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2839              continue;              continue;
# Line 2976  for (;; ptr++) Line 3028  for (;; ptr++)
3028    
3029        oldptr = ptr;        oldptr = ptr;
3030    
3031          /* Remember \r or \n */
3032    
3033          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3034    
3035          /* Check for range */
3036    
3037        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3038          {          {
3039          int d;          int d;
# Line 3043  for (;; ptr++) Line 3101  for (;; ptr++)
3101    
3102          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3103    
3104            /* Remember \r or \n */
3105    
3106            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3107    
3108          /* 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
3109          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3110          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 3195  for (;; ptr++) Line 3257  for (;; ptr++)
3257        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
3258        goto FAILED;        goto FAILED;
3259        }        }
3260    
3261    
3262    /* This code has been disabled because it would mean that \s counts as
3263    an explicit \r or \n reference, and that's not really what is wanted. Now
3264    we set the flag only if there is a literal "\r" or "\n" in the class. */
3265    
3266    #if 0
3267      /* Remember whether \r or \n are in this class */      /* Remember whether \r or \n are in this class */
3268    
3269      if (negate_class)      if (negate_class)
3270        {        {
3271        if ((classbits[1] & 0x24) != 0x24) cd->external_options |= PCRE_HASCRORLF;        if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3272        }        }
3273      else      else
3274        {        {
3275        if ((classbits[1] & 0x24) != 0) cd->external_options |= PCRE_HASCRORLF;        if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3276        }        }
3277    #endif
3278    
3279    
3280      /* 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
3281      less than 256. As long as there were no characters >= 128 and there was no      less than 256. As long as there were no characters >= 128 and there was no
3282      use of \p or \P, in other words, no use of any XCLASS features, we can      use of \p or \P, in other words, no use of any XCLASS features, we can
3283      optimize.      optimize.
3284    
3285      In UTF-8 mode, we can optimize the negative case only if there were no      In UTF-8 mode, we can optimize the negative case only if there were no
3286      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3287      operate on single-bytes only. This is an historical hangover. Maybe one day      operate on single-bytes only. This is an historical hangover. Maybe one day
# Line 3225  for (;; ptr++) Line 3295  for (;; ptr++)
3295      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3296    
3297  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3298      if (class_charcount == 1 && !class_utf8 &&      if (class_charcount == 1 && !class_utf8 &&
3299        (!utf8 || !negate_class || class_lastchar < 128))        (!utf8 || !negate_class || class_lastchar < 128))
3300  #else  #else
3301      if (class_charcount == 1)      if (class_charcount == 1)
# Line 3269  for (;; ptr++) Line 3339  for (;; ptr++)
3339      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3340    
3341      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3342      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3343      we can omit the bitmap in the actual compiled code. */      such as \S in the class, because in that case all characters > 255 are in
3344        the class, so any that were explicitly given as well can be ignored. If
3345        (when there are explicit characters > 255 that must be listed) there are no
3346        characters < 256, we can omit the bitmap in the actual compiled code. */
3347    
3348  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3349      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3350        {        {
3351        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3352        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3299  for (;; ptr++) Line 3372  for (;; ptr++)
3372        }        }
3373  #endif  #endif
3374    
3375      /* If there are no characters > 255, negate the 32-byte map if necessary,      /* If there are no characters > 255, set the opcode to OP_CLASS or
3376      and copy it into the code vector. If this is the first thing in the branch,      OP_NCLASS, depending on whether the whole class was negated and whether
3377      there can be no first char setting, whatever the repeat count. Any reqbyte      there were negative specials such as \S in the class. Then copy the 32-byte
3378      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3379    
3380        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3381      if (negate_class)      if (negate_class)
3382        {        {
       *code++ = OP_NCLASS;  
3383        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3384          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3385        }        }
3386      else      else
3387        {        {
       *code++ = OP_CLASS;  
3388        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3389        }        }
3390      code += 32;      code += 32;
# Line 3498  for (;; ptr++) Line 3570  for (;; ptr++)
3570        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3571        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3572    
3573        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3574    
3575        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3576    
# Line 3648  for (;; ptr++) Line 3720  for (;; ptr++)
3720        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3721        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3722    
3723        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3724    
3725        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3726          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 4006  for (;; ptr++) Line 4078  for (;; ptr++)
4078      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4079        {        {
4080        int i, namelen;        int i, namelen;
4081          const char *vn = verbnames;
4082        const uschar *name = ++ptr;        const uschar *name = ++ptr;
4083        previous = NULL;        previous = NULL;
4084        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
# Line 4023  for (;; ptr++) Line 4096  for (;; ptr++)
4096        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4097          {          {
4098          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
4099              strncmp((char *)name, verbs[i].name, namelen) == 0)              strncmp((char *)name, vn, namelen) == 0)
4100            {            {
4101            *code = verbs[i].op;            *code = verbs[i].op;
4102            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4103            break;            break;
4104            }            }
4105            vn += verbs[i].len + 1;
4106          }          }
4107        if (i < verbcount) continue;        if (i < verbcount) continue;
4108        *errorcodeptr = ERR60;        *errorcodeptr = ERR60;
# Line 4462  for (;; ptr++) Line 4536  for (;; ptr++)
4536    
4537          if (lengthptr != NULL)          if (lengthptr != NULL)
4538            {            {
4539              if (namelen == 0)
4540                {
4541                *errorcodeptr = ERR62;
4542                goto FAILED;
4543                }
4544            if (*ptr != terminator)            if (*ptr != terminator)
4545              {              {
4546              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4475  for (;; ptr++) Line 4554  for (;; ptr++)
4554            recno = 0;            recno = 0;
4555            }            }
4556    
4557          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4558            first, and then check that we have reached the end of the name in the
4559            table. That way, if the name that is longer than any in the table,
4560            the comparison will fail without reading beyond the table entry. */
4561    
4562          else          else
4563            {            {
4564            slot = cd->name_table;            slot = cd->name_table;
4565            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4566              {              {
4567              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4568                    slot[2+namelen] == 0)
4569                  break;
4570              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4571              }              }
4572    
# Line 4645  for (;; ptr++) Line 4729  for (;; ptr++)
4729    
4730              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4731              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4732              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4733              break;              break;
4734    
4735              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 5061  for (;; ptr++) Line 5145  for (;; ptr++)
5145      previous = code;      previous = code;
5146      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5147      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5148    
5149      /* Remember if \r or \n were seen */      /* Remember if \r or \n were seen */
5150    
5151      if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')      if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5152        cd->external_options |= PCRE_HASCRORLF;        cd->external_flags |= PCRE_HASCRORLF;
5153    
5154      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5155      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.
# Line 5666  real_pcre *re; Line 5750  real_pcre *re;
5750  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5751  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5752  int errorcode = 0;  int errorcode = 0;
5753    int skipatstart = 0;
5754  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5755  BOOL utf8;  BOOL utf8;
5756  #endif  #endif
# Line 5744  cd->fcc = tables + fcc_offset; Line 5829  cd->fcc = tables + fcc_offset;
5829  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5830  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5831    
5832    /* Check for global one-time settings at the start of the pattern, and remember
5833    the offset for later. */
5834    
5835    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5836      {
5837      int newnl = 0;
5838      int newbsr = 0;
5839    
5840      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5841        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5842      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5843        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5844      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5845        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5846      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5847        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5848      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5849        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5850    
5851      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5852        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5853      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5854        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5855    
5856      if (newnl != 0)
5857        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5858      else if (newbsr != 0)
5859        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5860      else break;
5861      }
5862    
5863    /* Check validity of \R options. */
5864    
5865    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5866      {
5867      case 0:
5868      case PCRE_BSR_ANYCRLF:
5869      case PCRE_BSR_UNICODE:
5870      break;
5871      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5872      }
5873    
5874  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5875  current code allows for fixed one- or two-byte sequences, plus "any" and  current code allows for fixed one- or two-byte sequences, plus "any" and
5876  "anycrlf". */  "anycrlf". */
5877    
5878  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5879    {    {
5880    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5881    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5882    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5883    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5813  cd->hwm = cworkspace; Line 5940  cd->hwm = cworkspace;
5940  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5941  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5942  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5943  cd->external_options = options;  cd->external_options = options;
5944    cd->external_flags = 0;
5945    
5946  /* 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
5947  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 5822  been put into the cd block so that they Line 5949  been put into the cd block so that they
5949  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5950  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5951    
5952    ptr += skipatstart;
5953  code = cworkspace;  code = cworkspace;
5954  *code = OP_BRA;  *code = OP_BRA;
5955  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
# Line 5852  if (re == NULL) Line 5980  if (re == NULL)
5980    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
5981    }    }
5982    
5983  /* 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
5984  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
5985  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
5986  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
5987    pointers. */
5988    
5989  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
5990  re->size = size;  re->size = size;
5991  re->options = cd->external_options;  re->options = cd->external_options;
5992    re->flags = cd->external_flags;
5993  re->dummy1 = 0;  re->dummy1 = 0;
5994  re->first_byte = 0;  re->first_byte = 0;
5995  re->req_byte = 0;  re->req_byte = 0;
# Line 5884  codestart = cd->name_table + re->name_en Line 6014  codestart = cd->name_table + re->name_en
6014  cd->start_code = codestart;  cd->start_code = codestart;
6015  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6016  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
6017  cd->had_accept = FALSE;  cd->had_accept = FALSE;
6018    
6019  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6020  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
6021  of the function here. */  of the function here. */
6022    
6023  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6024  code = (uschar *)codestart;  code = (uschar *)codestart;
6025  *code = OP_BRA;  *code = OP_BRA;
6026  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6027    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6028  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6029  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6030    re->flags = cd->external_flags;
6031    
 if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  
6032  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6033    
6034  /* 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. */
# Line 5942  if (errorcode != 0) Line 6071  if (errorcode != 0)
6071    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6072    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6073    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6074    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6075    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6076    return NULL;    return NULL;
6077    }    }
# Line 5971  if ((re->options & PCRE_ANCHORED) == 0) Line 6100  if ((re->options & PCRE_ANCHORED) == 0)
6100        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6101        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6102           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6103        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6104        }        }
6105      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6106        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6107      }      }
6108    }    }
6109    
# Line 5988  if (reqbyte >= 0 && Line 6117  if (reqbyte >= 0 &&
6117    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6118    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6119      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6120    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6121    }    }
6122    
6123  /* 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 5998  case when building a production library. Line 6127  case when building a production library.
6127    
6128  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6129    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6130    
6131  printf("Options=%08x\n", re->options);  printf("Options=%08x\n", re->options);
6132    
6133  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6134    {    {
6135    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6136    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 6010  if ((re->options & PCRE_FIRSTSET) != 0) Line 6139  if ((re->options & PCRE_FIRSTSET) != 0)
6139      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6140    }    }
6141    
6142  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6143    {    {
6144    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6145    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 6027  was compiled can be seen. */ Line 6156  was compiled can be seen. */
6156  if (code - codestart > length)  if (code - codestart > length)
6157    {    {
6158    (pcre_free)(re);    (pcre_free)(re);
6159    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6160    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6161    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6162    return NULL;    return NULL;

Legend:
Removed from v.226  
changed lines
  Added in v.268

  ViewVC Help
Powered by ViewVC 1.1.5