/[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 166 by ph10, Wed May 9 14:48:28 2007 UTC revision 223 by ph10, Mon Aug 20 11:07:53 2007 UTC
# 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 87  static const short int escapes[] = { Line 102  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    
# 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    /* Table of special "verbs" like (*PRUNE) */
144    
145    typedef struct verbitem {
146      const char *name;
147      int   len;
148      int   op;
149    } verbitem;
150    
151    static verbitem verbs[] = {
152      { "ACCEPT", 6, OP_ACCEPT },
153      { "COMMIT", 6, OP_COMMIT },
154      { "F",      1, OP_FAIL },
155      { "FAIL",   4, OP_FAIL },
156      { "PRUNE",  5, OP_PRUNE },
157      { "SKIP",   4, OP_SKIP  },
158      { "THEN",   4, OP_THEN  }
159    };
160    
161    static int verbcount = sizeof(verbs)/sizeof(verbitem);
162    
163    
164  /* Tables of names of POSIX character classes and their lengths. The list is  /* Tables of names of POSIX character classes and their lengths. The list is
165  terminated by a zero length entry. The first three must be alpha, lower, upper,  terminated by a zero length entry. The first three must be alpha, lower, upper,
166  as this is assumed for handling case independence. */  as this is assumed for handling case independence. */
# Line 198  static const char *error_texts[] = { Line 234  static const char *error_texts[] = {
234    "missing ) after comment",    "missing ) after comment",
235    "parentheses nested too deeply",  /** DEAD **/    "parentheses nested too deeply",  /** DEAD **/
236    /* 20 */    /* 20 */
237    "regular expression too large",    "regular expression is too large",
238    "failed to get memory",    "failed to get memory",
239    "unmatched parentheses",    "unmatched parentheses",
240    "internal error: code overflow",    "internal error: code overflow",
# Line 234  static const char *error_texts[] = { Line 270  static const char *error_texts[] = {
270    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",
271    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",
272    /* 50 */    /* 50 */
273    "repeated subpattern is too long",    "repeated subpattern is too long",    /** DEAD **/
274    "octal value is greater than \\377 (not in UTF-8 mode)",    "octal value is greater than \\377 (not in UTF-8 mode)",
275    "internal error: overran compiling workspace",    "internal error: overran compiling workspace",
276    "internal error: previously-checked referenced subpattern not found",    "internal error: previously-checked referenced subpattern not found",
# Line 242  static const char *error_texts[] = { Line 278  static const char *error_texts[] = {
278    /* 55 */    /* 55 */
279    "repeating a DEFINE group is not allowed",    "repeating a DEFINE group is not allowed",
280    "inconsistent NEWLINE options",    "inconsistent NEWLINE options",
281    "\\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",
282    "(?+ or (?- must be followed by a non-zero number"    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number",
283      "(*VERB) with an argument is not supported",
284      /* 60 */
285      "(*VERB) not recognized",
286      "number is too big"
287  };  };
288    
289    
# Line 374  static const unsigned char ebcdic_charta Line 414  static const unsigned char ebcdic_charta
414  /* Definition to allow mutual recursion */  /* Definition to allow mutual recursion */
415    
416  static BOOL  static BOOL
417    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, int, int *,    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,
418      int *, branch_chain *, compile_data *, int *);      int *, int *, branch_chain *, compile_data *, int *);
419    
420    
421    
# Line 400  Arguments: Line 440  Arguments:
440    
441  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
442                   negative => a special escape sequence                   negative => a special escape sequence
443                   on error, errorptr is set                   on error, errorcodeptr is set
444  */  */
445    
446  static int  static int
# Line 453  else Line 493  else
493    
494      /* \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
495      is an absolute backreference. If negative, it is a relative backreference.      is an absolute backreference. If negative, it is a relative backreference.
496      This is a Perl 5.10 feature. */      This is a Perl 5.10 feature. Perl 5.10 also supports \g{name} as a
497        reference to a named group. This is part of Perl's movement towards a
498        unified syntax for back references. As this is synonymous with \k{name}, we
499        fudge it up by pretending it really was \k. */
500    
501      case 'g':      case 'g':
502      if (ptr[1] == '{')      if (ptr[1] == '{')
503        {        {
504          const uschar *p;
505          for (p = ptr+2; *p != 0 && *p != '}'; p++)
506            if (*p != '-' && (digitab[*p] & ctype_digit) == 0) break;
507          if (*p != 0 && *p != '}')
508            {
509            c = -ESC_k;
510            break;
511            }
512        braced = TRUE;        braced = TRUE;
513        ptr++;        ptr++;
514        }        }
# Line 474  else Line 525  else
525      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
526        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
527    
528        if (c < 0)
529          {
530          *errorcodeptr = ERR61;
531          break;
532          }
533    
534      if (c == 0 || (braced && *(++ptr) != '}'))      if (c == 0 || (braced && *(++ptr) != '}'))
535        {        {
536        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
537        return 0;        break;
538        }        }
539    
540      if (negated)      if (negated)
# Line 485  else Line 542  else
542        if (c > bracount)        if (c > bracount)
543          {          {
544          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
545          return 0;          break;
546          }          }
547        c = bracount - (c - 1);        c = bracount - (c - 1);
548        }        }
# Line 514  else Line 571  else
571        c -= '0';        c -= '0';
572        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
573          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - '0';
574          if (c < 0)
575            {
576            *errorcodeptr = ERR61;
577            break;
578            }
579        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
580          {          {
581          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 609  else Line 671  else
671      if (c == 0)      if (c == 0)
672        {        {
673        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
674        return 0;        break;
675        }        }
676    
677  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
# Line 685  if (c == '{') Line 747  if (c == '{')
747      *negptr = TRUE;      *negptr = TRUE;
748      ptr++;      ptr++;
749      }      }
750    for (i = 0; i < sizeof(name) - 1; i++)    for (i = 0; i < (int)sizeof(name) - 1; i++)
751      {      {
752      c = *(++ptr);      c = *(++ptr);
753      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 888  for (; *ptr != 0; ptr++) Line 950  for (; *ptr != 0; ptr++)
950      {      {
951      while (*(++ptr) != ']')      while (*(++ptr) != ']')
952        {        {
953          if (*ptr == 0) return -1;
954        if (*ptr == '\\')        if (*ptr == '\\')
955          {          {
956          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 915  for (; *ptr != 0; ptr++) Line 978  for (; *ptr != 0; ptr++)
978    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
979    
980    if (*ptr != '(') continue;    if (*ptr != '(') continue;
981    if (ptr[1] != '?')    if (ptr[1] != '?' && ptr[1] != '*')
982      {      {
983      count++;      count++;
984      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1043  for (;;) Line 1106  for (;;)
1106    {    {
1107    int d;    int d;
1108    register int op = *cc;    register int op = *cc;
   
1109    switch (op)    switch (op)
1110      {      {
1111      case OP_CBRA:      case OP_CBRA:
# Line 1132  for (;;) Line 1194  for (;;)
1194    
1195      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1196      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1197        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1198      cc += 4;      cc += 4;
1199      break;      break;
1200    
# Line 1240  for (;;) Line 1303  for (;;)
1303      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1304      }      }
1305    
1306    /* 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
1307    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
1308    arrange to skip the extra bytes. */    two bytes of parameters. */
1309    
1310    else    else
1311      {      {
1312        switch(c)
1313          {
1314          case OP_TYPESTAR:
1315          case OP_TYPEMINSTAR:
1316          case OP_TYPEPLUS:
1317          case OP_TYPEMINPLUS:
1318          case OP_TYPEQUERY:
1319          case OP_TYPEMINQUERY:
1320          case OP_TYPEPOSSTAR:
1321          case OP_TYPEPOSPLUS:
1322          case OP_TYPEPOSQUERY:
1323          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1324          break;
1325    
1326          case OP_TYPEUPTO:
1327          case OP_TYPEMINUPTO:
1328          case OP_TYPEEXACT:
1329          case OP_TYPEPOSUPTO:
1330          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1331          break;
1332          }
1333    
1334        /* Add in the fixed length from the table */
1335    
1336      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1337    
1338      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1339      a multi-byte character. The length in the table is a minimum, so we have to
1340      arrange to skip the extra bytes. */
1341    
1342  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1343      if (utf8) switch(c)      if (utf8) switch(c)
1344        {        {
# Line 1304  for (;;) Line 1396  for (;;)
1396    
1397    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1398    
1399    /* 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
1400    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
1401    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1402    
1403    else    else
1404      {      {
1405        switch(c)
1406          {
1407          case OP_TYPESTAR:
1408          case OP_TYPEMINSTAR:
1409          case OP_TYPEPLUS:
1410          case OP_TYPEMINPLUS:
1411          case OP_TYPEQUERY:
1412          case OP_TYPEMINQUERY:
1413          case OP_TYPEPOSSTAR:
1414          case OP_TYPEPOSPLUS:
1415          case OP_TYPEPOSQUERY:
1416          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1417          break;
1418    
1419          case OP_TYPEPOSUPTO:
1420          case OP_TYPEUPTO:
1421          case OP_TYPEMINUPTO:
1422          case OP_TYPEEXACT:
1423          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1424          break;
1425          }
1426    
1427        /* Add in the fixed length from the table */
1428    
1429      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1430    
1431        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1432        by a multi-byte character. The length in the table is a minimum, so we have
1433        to arrange to skip the extra bytes. */
1434    
1435  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1436      if (utf8) switch(c)      if (utf8) switch(c)
1437        {        {
# Line 1371  for (code = first_significant_code(code Line 1491  for (code = first_significant_code(code
1491    
1492    c = *code;    c = *code;
1493    
1494    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    /* Groups with zero repeats can of course be empty; skip them. */
1495    
1496      if (c == OP_BRAZERO || c == OP_BRAMINZERO)
1497        {
1498        code += _pcre_OP_lengths[c];
1499        do code += GET(code, 1); while (*code == OP_ALT);
1500        c = *code;
1501        continue;
1502        }
1503    
1504      /* For other groups, scan the branches. */
1505    
1506      if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1507      {      {
1508      BOOL empty_branch;      BOOL empty_branch;
1509      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1387  for (code = first_significant_code(code Line 1519  for (code = first_significant_code(code
1519        }        }
1520      while (*code == OP_ALT);      while (*code == OP_ALT);
1521      if (!empty_branch) return FALSE;   /* All branches are non-empty */      if (!empty_branch) return FALSE;   /* All branches are non-empty */
1522        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];  
1523      continue;      continue;
1524      }      }
1525    
# Line 1400  for (code = first_significant_code(code Line 1527  for (code = first_significant_code(code
1527    
1528    switch (c)    switch (c)
1529      {      {
1530      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1531        cannot be represented just by a bit map. This includes negated single
1532        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1533        actual length is stored in the compiled code, so we must update "code"
1534        here. */
1535    
1536  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1537      case OP_XCLASS:      case OP_XCLASS:
1538      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1539      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1540  #endif  #endif
1541    
# Line 1926  if (next >= 0) switch(op_code) Line 2057  if (next >= 0) switch(op_code)
2057    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2058    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2059    
2060      case OP_HSPACE:
2061      case OP_NOT_HSPACE:
2062      switch(next)
2063        {
2064        case 0x09:
2065        case 0x20:
2066        case 0xa0:
2067        case 0x1680:
2068        case 0x180e:
2069        case 0x2000:
2070        case 0x2001:
2071        case 0x2002:
2072        case 0x2003:
2073        case 0x2004:
2074        case 0x2005:
2075        case 0x2006:
2076        case 0x2007:
2077        case 0x2008:
2078        case 0x2009:
2079        case 0x200A:
2080        case 0x202f:
2081        case 0x205f:
2082        case 0x3000:
2083        return op_code != OP_HSPACE;
2084        default:
2085        return op_code == OP_HSPACE;
2086        }
2087    
2088      case OP_VSPACE:
2089      case OP_NOT_VSPACE:
2090      switch(next)
2091        {
2092        case 0x0a:
2093        case 0x0b:
2094        case 0x0c:
2095        case 0x0d:
2096        case 0x85:
2097        case 0x2028:
2098        case 0x2029:
2099        return op_code != OP_VSPACE;
2100        default:
2101        return op_code == OP_VSPACE;
2102        }
2103    
2104    default:    default:
2105    return FALSE;    return FALSE;
2106    }    }
# Line 1960  switch(op_code) Line 2135  switch(op_code)
2135      case ESC_W:      case ESC_W:
2136      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2137    
2138        case ESC_h:
2139        case ESC_H:
2140        switch(item)
2141          {
2142          case 0x09:
2143          case 0x20:
2144          case 0xa0:
2145          case 0x1680:
2146          case 0x180e:
2147          case 0x2000:
2148          case 0x2001:
2149          case 0x2002:
2150          case 0x2003:
2151          case 0x2004:
2152          case 0x2005:
2153          case 0x2006:
2154          case 0x2007:
2155          case 0x2008:
2156          case 0x2009:
2157          case 0x200A:
2158          case 0x202f:
2159          case 0x205f:
2160          case 0x3000:
2161          return -next != ESC_h;
2162          default:
2163          return -next == ESC_h;
2164          }
2165    
2166        case ESC_v:
2167        case ESC_V:
2168        switch(item)
2169          {
2170          case 0x0a:
2171          case 0x0b:
2172          case 0x0c:
2173          case 0x0d:
2174          case 0x85:
2175          case 0x2028:
2176          case 0x2029:
2177          return -next != ESC_v;
2178          default:
2179          return -next == ESC_v;
2180          }
2181    
2182      default:      default:
2183      return FALSE;      return FALSE;
2184      }      }
2185    
2186    case OP_DIGIT:    case OP_DIGIT:
2187    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2188             next == -ESC_h || next == -ESC_v;
2189    
2190    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2191    return next == -ESC_d;    return next == -ESC_d;
# Line 1974  switch(op_code) Line 2194  switch(op_code)
2194    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2195    
2196    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2197    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2198    
2199      case OP_HSPACE:
2200      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2201    
2202      case OP_NOT_HSPACE:
2203      return next == -ESC_h;
2204    
2205      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2206      case OP_VSPACE:
2207      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2208    
2209      case OP_NOT_VSPACE:
2210      return next == -ESC_v;
2211    
2212    case OP_WORDCHAR:    case OP_WORDCHAR:
2213    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2214    
2215    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2216    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2092  for (;; ptr++) Line 2325  for (;; ptr++)
2325    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2326    BOOL is_quantifier;    BOOL is_quantifier;
2327    BOOL is_recurse;    BOOL is_recurse;
2328      BOOL reset_bracount;
2329    int class_charcount;    int class_charcount;
2330    int class_lastchar;    int class_lastchar;
2331    int newoptions;    int newoptions;
2332    int recno;    int recno;
2333      int refsign;
2334    int skipbytes;    int skipbytes;
2335    int subreqbyte;    int subreqbyte;
2336    int subfirstbyte;    int subfirstbyte;
# Line 2128  for (;; ptr++) Line 2363  for (;; ptr++)
2363      */      */
2364    
2365      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2366    
2367        /* Paranoid check for integer overflow */
2368    
2369        if (OFLOW_MAX - *lengthptr < code - last_code)
2370          {
2371          *errorcodeptr = ERR20;
2372          goto FAILED;
2373          }
2374    
2375      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2376      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));
2377    
# Line 2240  for (;; ptr++) Line 2484  for (;; ptr++)
2484      *ptrptr = ptr;      *ptrptr = ptr;
2485      if (lengthptr != NULL)      if (lengthptr != NULL)
2486        {        {
2487          if (OFLOW_MAX - *lengthptr < code - last_code)
2488            {
2489            *errorcodeptr = ERR20;
2490            goto FAILED;
2491            }
2492        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2493        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2494        }        }
# Line 2302  for (;; ptr++) Line 2551  for (;; ptr++)
2551        goto FAILED;        goto FAILED;
2552        }        }
2553    
2554      /* 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,
2555        if the first few characters (either before or after ^) are \Q\E or \E we
2556        skip them too. This makes for compatibility with Perl. */
2557    
2558      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2559        for (;;)
2560        {        {
       negate_class = TRUE;  
2561        c = *(++ptr);        c = *(++ptr);
2562        }        if (c == '\\')
2563      else          {
2564        {          if (ptr[1] == 'E') ptr++;
2565        negate_class = FALSE;            else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2566                else break;
2567            }
2568          else if (!negate_class && c == '^')
2569            negate_class = TRUE;
2570          else break;
2571        }        }
2572    
2573      /* 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 2452  for (;; ptr++) Line 2708  for (;; ptr++)
2708        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
2709        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2710        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2711        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
2712        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2713    
2714        if (c == '\\')        if (c == '\\')
# Line 2472  for (;; ptr++) Line 2728  for (;; ptr++)
2728            else inescq = TRUE;            else inescq = TRUE;
2729            continue;            continue;
2730            }            }
2731            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2732    
2733          if (c < 0)          if (c < 0)
2734            {            {
# Line 2520  for (;; ptr++) Line 2777  for (;; ptr++)
2777            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2778                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2779    
2780              /* We need to deal with \H, \h, \V, and \v in both phases because
2781              they use extra memory. */
2782    
2783              if (-c == ESC_h)
2784                {
2785                SETBIT(classbits, 0x09); /* VT */
2786                SETBIT(classbits, 0x20); /* SPACE */
2787                SETBIT(classbits, 0xa0); /* NSBP */
2788    #ifdef SUPPORT_UTF8
2789                if (utf8)
2790                  {
2791                  class_utf8 = TRUE;
2792                  *class_utf8data++ = XCL_SINGLE;
2793                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2794                  *class_utf8data++ = XCL_SINGLE;
2795                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2796                  *class_utf8data++ = XCL_RANGE;
2797                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2798                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2799                  *class_utf8data++ = XCL_SINGLE;
2800                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2801                  *class_utf8data++ = XCL_SINGLE;
2802                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2803                  *class_utf8data++ = XCL_SINGLE;
2804                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2805                  }
2806    #endif
2807                continue;
2808                }
2809    
2810              if (-c == ESC_H)
2811                {
2812                for (c = 0; c < 32; c++)
2813                  {
2814                  int x = 0xff;
2815                  switch (c)
2816                    {
2817                    case 0x09/8: x ^= 1 << (0x09%8); break;
2818                    case 0x20/8: x ^= 1 << (0x20%8); break;
2819                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
2820                    default: break;
2821                    }
2822                  classbits[c] |= x;
2823                  }
2824    
2825    #ifdef SUPPORT_UTF8
2826                if (utf8)
2827                  {
2828                  class_utf8 = TRUE;
2829                  *class_utf8data++ = XCL_RANGE;
2830                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2831                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
2832                  *class_utf8data++ = XCL_RANGE;
2833                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
2834                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
2835                  *class_utf8data++ = XCL_RANGE;
2836                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
2837                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
2838                  *class_utf8data++ = XCL_RANGE;
2839                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
2840                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
2841                  *class_utf8data++ = XCL_RANGE;
2842                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
2843                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
2844                  *class_utf8data++ = XCL_RANGE;
2845                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
2846                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
2847                  *class_utf8data++ = XCL_RANGE;
2848                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
2849                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2850                  }
2851    #endif
2852                continue;
2853                }
2854    
2855              if (-c == ESC_v)
2856                {
2857                SETBIT(classbits, 0x0a); /* LF */
2858                SETBIT(classbits, 0x0b); /* VT */
2859                SETBIT(classbits, 0x0c); /* FF */
2860                SETBIT(classbits, 0x0d); /* CR */
2861                SETBIT(classbits, 0x85); /* NEL */
2862    #ifdef SUPPORT_UTF8
2863                if (utf8)
2864                  {
2865                  class_utf8 = TRUE;
2866                  *class_utf8data++ = XCL_RANGE;
2867                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
2868                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2869                  }
2870    #endif
2871                continue;
2872                }
2873    
2874              if (-c == ESC_V)
2875                {
2876                for (c = 0; c < 32; c++)
2877                  {
2878                  int x = 0xff;
2879                  switch (c)
2880                    {
2881                    case 0x0a/8: x ^= 1 << (0x0a%8);
2882                                 x ^= 1 << (0x0b%8);
2883                                 x ^= 1 << (0x0c%8);
2884                                 x ^= 1 << (0x0d%8);
2885                                 break;
2886                    case 0x85/8: x ^= 1 << (0x85%8); break;
2887                    default: break;
2888                    }
2889                  classbits[c] |= x;
2890                  }
2891    
2892    #ifdef SUPPORT_UTF8
2893                if (utf8)
2894                  {
2895                  class_utf8 = TRUE;
2896                  *class_utf8data++ = XCL_RANGE;
2897                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2898                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
2899                  *class_utf8data++ = XCL_RANGE;
2900                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2901                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2902                  }
2903    #endif
2904                continue;
2905                }
2906    
2907            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
2908    
2909  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2660  for (;; ptr++) Line 3044  for (;; ptr++)
3044              unsigned int origd = d;              unsigned int origd = d;
3045              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3046                {                {
3047                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3048                      ocd <= (unsigned int)d)
3049                    continue;                          /* Skip embedded ranges */
3050    
3051                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3052                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3053                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3054                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3055                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3056                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3057                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3058                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3059                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3060                  d = ocd;                  d = ocd;
3061                  continue;                  continue;
# Line 2788  for (;; ptr++) Line 3176  for (;; ptr++)
3176        }        }
3177    
3178      /* 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
3179      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
3180      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
3181      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3182      single-bytes only. This is an historical hangover. Maybe one day we can  
3183      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3184        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3185        operate on single-bytes only. This is an historical hangover. Maybe one day
3186        we can tidy these opcodes to handle multi-byte characters.
3187    
3188      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
3189      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 2802  for (;; ptr++) Line 3193  for (;; ptr++)
3193      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3194    
3195  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3196      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3197            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3198  #else  #else
3199      if (class_charcount == 1)      if (class_charcount == 1)
3200  #endif  #endif
# Line 3263  for (;; ptr++) Line 3652  for (;; ptr++)
3652          goto FAILED;          goto FAILED;
3653          }          }
3654    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3655        /* 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
3656        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
3657        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 3359  for (;; ptr++) Line 3740  for (;; ptr++)
3740          if (repeat_min > 1)          if (repeat_min > 1)
3741            {            {
3742            /* 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
3743            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3744              potential integer overflow. */
3745    
3746            if (lengthptr != NULL)            if (lengthptr != NULL)
3747              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3748                int delta = (repeat_min - 1)*length_prevgroup;
3749                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3750                                                                (double)INT_MAX ||
3751                    OFLOW_MAX - *lengthptr < delta)
3752                  {
3753                  *errorcodeptr = ERR20;
3754                  goto FAILED;
3755                  }
3756                *lengthptr += delta;
3757                }
3758    
3759            /* This is compiling for real */            /* This is compiling for real */
3760    
# Line 3400  for (;; ptr++) Line 3792  for (;; ptr++)
3792          /* 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
3793          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
3794          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
3795          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3796            paranoid checks to avoid integer overflow. */
3797    
3798          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3799            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3800              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3801                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3802              if ((double)repeat_max *
3803                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3804                      > (double)INT_MAX ||
3805                  OFLOW_MAX - *lengthptr < delta)
3806                {
3807                *errorcodeptr = ERR20;
3808                goto FAILED;
3809                }
3810              *lengthptr += delta;
3811              }
3812    
3813          /* This is compiling for real */          /* This is compiling for real */
3814    
# Line 3556  for (;; ptr++) Line 3960  for (;; ptr++)
3960      /* ===================================================================*/      /* ===================================================================*/
3961      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
3962      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
3963      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. */  
3964    
3965      case '(':      case '(':
3966      newoptions = options;      newoptions = options;
3967      skipbytes = 0;      skipbytes = 0;
3968      bravalue = OP_CBRA;      bravalue = OP_CBRA;
3969      save_hwm = cd->hwm;      save_hwm = cd->hwm;
3970        reset_bracount = FALSE;
3971    
3972      if (*(++ptr) == '?')      /* First deal with various "verbs" that can be introduced by '*'. */
3973    
3974        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
3975          {
3976          int i, namelen;
3977          const uschar *name = ++ptr;
3978          previous = NULL;
3979          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
3980          if (*ptr == ':')
3981            {
3982            *errorcodeptr = ERR59;   /* Not supported */
3983            goto FAILED;
3984            }
3985          if (*ptr != ')')
3986            {
3987            *errorcodeptr = ERR60;
3988            goto FAILED;
3989            }
3990          namelen = ptr - name;
3991          for (i = 0; i < verbcount; i++)
3992            {
3993            if (namelen == verbs[i].len &&
3994                strncmp((char *)name, verbs[i].name, namelen) == 0)
3995              {
3996              *code = verbs[i].op;
3997              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
3998              break;
3999              }
4000            }
4001          if (i < verbcount) continue;
4002          *errorcodeptr = ERR60;
4003          goto FAILED;
4004          }
4005    
4006        /* Deal with the extended parentheses; all are introduced by '?', and the
4007        appearance of any of them means that this is not a capturing group. */
4008    
4009        else if (*ptr == '?')
4010        {        {
4011        int i, set, unset, namelen;        int i, set, unset, namelen;
4012        int *optset;        int *optset;
# Line 3587  for (;; ptr++) Line 4027  for (;; ptr++)
4027    
4028    
4029          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4030            case '|':                 /* Reset capture count for each branch */
4031            reset_bracount = TRUE;
4032            /* Fall through */
4033    
4034            /* ------------------------------------------------------------ */
4035          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
4036          bravalue = OP_BRA;          bravalue = OP_BRA;
4037          ptr++;          ptr++;
# Line 3622  for (;; ptr++) Line 4067  for (;; ptr++)
4067    
4068          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
4069          skipbytes = 3;          skipbytes = 3;
4070            refsign = -1;
4071    
4072          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4073    
# Line 3645  for (;; ptr++) Line 4091  for (;; ptr++)
4091            terminator = '\'';            terminator = '\'';
4092            ptr++;            ptr++;
4093            }            }
4094          else terminator = 0;          else
4095              {
4096              terminator = 0;
4097              if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);
4098              }
4099    
4100          /* 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 */
4101    
# Line 3681  for (;; ptr++) Line 4131  for (;; ptr++)
4131          if (lengthptr != NULL) break;          if (lengthptr != NULL) break;
4132    
4133          /* 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
4134          reference. */          reference. If the string started with "+" or "-" we require the rest to
4135            be digits, in which case recno will be set. */
4136    
4137            if (refsign > 0)
4138              {
4139              if (recno <= 0)
4140                {
4141                *errorcodeptr = ERR58;
4142                goto FAILED;
4143                }
4144              if (refsign == '-')
4145                {
4146                recno = cd->bracount - recno + 1;
4147                if (recno <= 0)
4148                  {
4149                  *errorcodeptr = ERR15;
4150                  goto FAILED;
4151                  }
4152                }
4153              else recno += cd->bracount;
4154              PUT2(code, 2+LINK_SIZE, recno);
4155              break;
4156              }
4157    
4158            /* Otherwise (did not start with "+" or "-"), start by looking for the
4159            name. */
4160    
4161          slot = cd->name_table;          slot = cd->name_table;
4162          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 3773  for (;; ptr++) Line 4248  for (;; ptr++)
4248    
4249          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4250          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4251          ptr++;          ptr++;
4252            if (*ptr == ')')          /* Optimize (?!) */
4253              {
4254              *code++ = OP_FAIL;
4255              previous = NULL;
4256              continue;
4257              }
4258            bravalue = OP_ASSERT_NOT;
4259          break;          break;
4260    
4261    
# Line 4005  for (;; ptr++) Line 4486  for (;; ptr++)
4486          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case '5': case '6': case '7': case '8': case '9':   /* subroutine */
4487            {            {
4488            const uschar *called;            const uschar *called;
           int sign = *ptr;  
4489    
4490            if (sign == '+') ptr++;            if ((refsign = *ptr) == '+') ptr++;
4491            else if (sign == '-')            else if (refsign == '-')
4492              {              {
4493              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
4494                goto OTHER_CHAR_AFTER_QUERY;                goto OTHER_CHAR_AFTER_QUERY;
4495              ptr++;              ptr++;
4496              }              }
4497    
4498            recno = 0;            recno = 0;
4499            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4500              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - '0';
# Line 4024  for (;; ptr++) Line 4504  for (;; ptr++)
4504              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4505              goto FAILED;              goto FAILED;
4506              }              }
4507    
4508            if (sign == '-')            if (refsign == '-')
4509              {              {
4510              if (recno == 0)              if (recno == 0)
4511                {                {
4512                *errorcodeptr = ERR58;                *errorcodeptr = ERR58;
4513                goto FAILED;                goto FAILED;
4514                }                }
4515              recno = cd->bracount - recno + 1;              recno = cd->bracount - recno + 1;
4516              if (recno <= 0)              if (recno <= 0)
4517                {                {
4518                *errorcodeptr = ERR15;                *errorcodeptr = ERR15;
4519                goto FAILED;                goto FAILED;
4520                }                }
4521              }              }
4522            else if (sign == '+')            else if (refsign == '+')
4523              {              {
4524              if (recno == 0)              if (recno == 0)
4525                {                {
4526                *errorcodeptr = ERR58;                *errorcodeptr = ERR58;
4527                goto FAILED;                goto FAILED;
4528                }                }
4529              recno += cd->bracount;              recno += cd->bracount;
4530              }              }
4531    
4532            /* Come here from code above that handles a named recursion */            /* Come here from code above that handles a named recursion */
4533    
# Line 4121  for (;; ptr++) Line 4601  for (;; ptr++)
4601    
4602          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4603          default:              /* Other characters: check option setting */          default:              /* Other characters: check option setting */
4604          OTHER_CHAR_AFTER_QUERY:          OTHER_CHAR_AFTER_QUERY:
4605          set = unset = 0;          set = unset = 0;
4606          optset = &set;          optset = &set;
4607    
# Line 4256  for (;; ptr++) Line 4736  for (;; ptr++)
4736           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4737           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4738            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4739             reset_bracount,               /* True if (?| group */
4740           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4741           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4742           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4272  for (;; ptr++) Line 4753  for (;; ptr++)
4753      is on the bracket. */      is on the bracket. */
4754    
4755      /* 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
4756      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
4757        in the real compile phase, not in the pre-pass, where the whole group may
4758        not be available. */
4759    
4760      if (bravalue == OP_COND)      if (bravalue == OP_COND && lengthptr == NULL)
4761        {        {
4762        uschar *tc = code;        uschar *tc = code;
4763        int condcount = 0;        int condcount = 0;
# Line 4321  for (;; ptr++) Line 4804  for (;; ptr++)
4804        goto FAILED;        goto FAILED;
4805        }        }
4806    
4807      /* 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,
4808      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
4809      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
4810      a quantifier. */      duplicated by a quantifier.*/
4811    
4812      if (lengthptr != NULL)      if (lengthptr != NULL)
4813        {        {
4814          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4815            {
4816            *errorcodeptr = ERR20;
4817            goto FAILED;
4818            }
4819        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4820        code++;        *code++ = OP_BRA;
4821        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4822        *code++ = OP_KET;        *code++ = OP_KET;
4823        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4824          break;    /* No need to waste time with special character handling */
4825        }        }
4826    
4827      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4828    
4829      else code = tempcode;      code = tempcode;
4830    
4831      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4832      relevant. */      relevant. */
# Line 4434  for (;; ptr++) Line 4923  for (;; ptr++)
4923        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
4924        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
4925    
4926        /* \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).
4927          We also support \k{name} (.NET syntax) */
4928    
4929        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\''))        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\'' || ptr[1] == '{'))
4930          {          {
4931          is_recurse = FALSE;          is_recurse = FALSE;
4932          terminator = (*(++ptr) == '<')? '>' : '\'';          terminator = (*(++ptr) == '<')? '>' : (*ptr == '\'')? '\'' : '}';
4933          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
4934          }          }
4935    
# Line 4605  This function is used during the pre-com Line 5095  This function is used during the pre-com
5095  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
5096  value of lengthptr distinguishes the two phases.  value of lengthptr distinguishes the two phases.
5097    
5098  Argument:  Arguments:
5099    options        option bits, including any changes for this subpattern    options        option bits, including any changes for this subpattern
5100    oldims         previous settings of ims option bits    oldims         previous settings of ims option bits
5101    codeptr        -> the address of the current code pointer    codeptr        -> the address of the current code pointer
5102    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5103    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5104    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5105      reset_bracount TRUE to reset the count for each branch
5106    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5107    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5108    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 4625  Returns:         TRUE on success Line 5116  Returns:         TRUE on success
5116    
5117  static BOOL  static BOOL
5118  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5119    int *errorcodeptr, BOOL lookbehind, int skipbytes, int *firstbyteptr,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5120    int *reqbyteptr, branch_chain *bcptr, compile_data *cd, int *lengthptr)    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5121      int *lengthptr)
5122  {  {
5123  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
5124  uschar *code = *codeptr;  uschar *code = *codeptr;
# Line 4636  uschar *reverse_count = NULL; Line 5128  uschar *reverse_count = NULL;
5128  int firstbyte, reqbyte;  int firstbyte, reqbyte;
5129  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
5130  int length;  int length;
5131    int orig_bracount;
5132    int max_bracount;
5133  branch_chain bc;  branch_chain bc;
5134    
5135  bc.outer = bcptr;  bc.outer = bcptr;
# Line 4664  code += 1 + LINK_SIZE + skipbytes; Line 5158  code += 1 + LINK_SIZE + skipbytes;
5158    
5159  /* Loop for each alternative branch */  /* Loop for each alternative branch */
5160    
5161    orig_bracount = max_bracount = cd->bracount;
5162  for (;;)  for (;;)
5163    {    {
5164      /* For a (?| group, reset the capturing bracket count so that each branch
5165      uses the same numbers. */
5166    
5167      if (reset_bracount) cd->bracount = orig_bracount;
5168    
5169    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5170    
5171    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4695  for (;;) Line 5195  for (;;)
5195      return FALSE;      return FALSE;
5196      }      }
5197    
5198      /* Keep the highest bracket count in case (?| was used and some branch
5199      has fewer than the rest. */
5200    
5201      if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5202    
5203    /* 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. */
5204    
5205    if (lengthptr == NULL)    if (lengthptr == NULL)
# Line 4758  for (;;) Line 5263  for (;;)
5263        }        }
5264      }      }
5265    
5266    /* Reached end of expression, either ')' or end of pattern. Go back through    /* Reached end of expression, either ')' or end of pattern. In the real
5267    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
5268    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
5269    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
5270    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
5271    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,
5272    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.
5273    at the terminating char. */    Return leaving the pointer at the terminating char. */
5274    
5275    if (*ptr != '|')    if (*ptr != '|')
5276      {      {
5277      int branch_length = code - last_branch;      if (lengthptr == NULL)
     do  
5278        {        {
5279        int prev_length = GET(last_branch, 1);        int branch_length = code - last_branch;
5280        PUT(last_branch, 1, branch_length);        do
5281        branch_length = prev_length;          {
5282        last_branch -= branch_length;          int prev_length = GET(last_branch, 1);
5283            PUT(last_branch, 1, branch_length);
5284            branch_length = prev_length;
5285            last_branch -= branch_length;
5286            }
5287          while (branch_length > 0);
5288        }        }
     while (branch_length > 0);  
5289    
5290      /* Fill in the ket */      /* Fill in the ket */
5291    
# Line 4794  for (;;) Line 5302  for (;;)
5302        length += 2;        length += 2;
5303        }        }
5304    
5305        /* Retain the highest bracket number, in case resetting was used. */
5306    
5307        cd->bracount = max_bracount;
5308    
5309      /* Set values to pass back */      /* Set values to pass back */
5310    
5311      *codeptr = code;      *codeptr = code;
5312      *ptrptr = ptr;      *ptrptr = ptr;
5313      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5314      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5315      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5316          {
5317          if (OFLOW_MAX - *lengthptr < length)
5318            {
5319            *errorcodeptr = ERR20;
5320            return FALSE;
5321            }
5322          *lengthptr += length;
5323          }
5324      return TRUE;      return TRUE;
5325      }      }
5326    
5327    /* 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
5328      pointer back to where it was for the start of the first branch. (That is,
5329      pretend that each branch is the only one.)
5330    
5331      In the real compile phase, insert an ALT node. Its length field points back
5332    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
5333    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
5334    zero offset until it is closed, making it possible to detect recursion. */    zero offset until it is closed, making it possible to detect recursion. */
5335    
5336    *code = OP_ALT;    if (lengthptr != NULL)
5337    PUT(code, 1, code - last_branch);      {
5338    bc.current = last_branch = code;      code = *codeptr + 1 + LINK_SIZE + skipbytes;
5339    code += 1 + LINK_SIZE;      length += 1 + LINK_SIZE;
5340        }
5341      else
5342        {
5343        *code = OP_ALT;
5344        PUT(code, 1, code - last_branch);
5345        bc.current = last_branch = code;
5346        code += 1 + LINK_SIZE;
5347        }
5348    
5349    ptr++;    ptr++;
   length += 1 + LINK_SIZE;  
5350    }    }
5351  /* Control never reaches here */  /* Control never reaches here */
5352  }  }
# Line 5256  outside can help speed up starting point Line 5788  outside can help speed up starting point
5788  code = cworkspace;  code = cworkspace;
5789  *code = OP_BRA;  *code = OP_BRA;
5790  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
5791    &code, &ptr, &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length);    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
5792      &length);
5793  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
5794    
5795  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 5315  cd->start_code = codestart; Line 5848  cd->start_code = codestart;
5848  cd->hwm = cworkspace;  cd->hwm = cworkspace;
5849  cd->req_varyopt = 0;  cd->req_varyopt = 0;
5850  cd->nopartial = FALSE;  cd->nopartial = FALSE;
5851    cd->had_accept = FALSE;
5852    
5853  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
5854  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
# Line 5324  ptr = (const uschar *)pattern; Line 5858  ptr = (const uschar *)pattern;
5858  code = (uschar *)codestart;  code = (uschar *)codestart;
5859  *code = OP_BRA;  *code = OP_BRA;
5860  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
5861    &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
5862  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
5863  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
5864    
5865  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;
5866    if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
5867    
5868  /* 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. */
5869    

Legend:
Removed from v.166  
changed lines
  Added in v.223

  ViewVC Help
Powered by ViewVC 1.1.5