harbour-project /tags/harbour-2.0.0/external/pcre/pcreexec.c

Language C Lines 5803
MD5 Hash d6aa033f5405db69dcebe3bf35ce5e24 Estimated Cost $125,482 (why?)
Repository https://harbour-project.svn.sourceforge.net/svnroot/harbour-project View Raw File
   1
   2
   3
   4
   5
   6
   7
   8
   9
  10
  11
  12
  13
  14
  15
  16
  17
  18
  19
  20
  21
  22
  23
  24
  25
  26
  27
  28
  29
  30
  31
  32
  33
  34
  35
  36
  37
  38
  39
  40
  41
  42
  43
  44
  45
  46
  47
  48
  49
  50
  51
  52
  53
  54
  55
  56
  57
  58
  59
  60
  61
  62
  63
  64
  65
  66
  67
  68
  69
  70
  71
  72
  73
  74
  75
  76
  77
  78
  79
  80
  81
  82
  83
  84
  85
  86
  87
  88
  89
  90
  91
  92
  93
  94
  95
  96
  97
  98
  99
 100
 101
 102
 103
 104
 105
 106
 107
 108
 109
 110
 111
 112
 113
 114
 115
 116
 117
 118
 119
 120
 121
 122
 123
 124
 125
 126
 127
 128
 129
 130
 131
 132
 133
 134
 135
 136
 137
 138
 139
 140
 141
 142
 143
 144
 145
 146
 147
 148
 149
 150
 151
 152
 153
 154
 155
 156
 157
 158
 159
 160
 161
 162
 163
 164
 165
 166
 167
 168
 169
 170
 171
 172
 173
 174
 175
 176
 177
 178
 179
 180
 181
 182
 183
 184
 185
 186
 187
 188
 189
 190
 191
 192
 193
 194
 195
 196
 197
 198
 199
 200
 201
 202
 203
 204
 205
 206
 207
 208
 209
 210
 211
 212
 213
 214
 215
 216
 217
 218
 219
 220
 221
 222
 223
 224
 225
 226
 227
 228
 229
 230
 231
 232
 233
 234
 235
 236
 237
 238
 239
 240
 241
 242
 243
 244
 245
 246
 247
 248
 249
 250
 251
 252
 253
 254
 255
 256
 257
 258
 259
 260
 261
 262
 263
 264
 265
 266
 267
 268
 269
 270
 271
 272
 273
 274
 275
 276
 277
 278
 279
 280
 281
 282
 283
 284
 285
 286
 287
 288
 289
 290
 291
 292
 293
 294
 295
 296
 297
 298
 299
 300
 301
 302
 303
 304
 305
 306
 307
 308
 309
 310
 311
 312
 313
 314
 315
 316
 317
 318
 319
 320
 321
 322
 323
 324
 325
 326
 327
 328
 329
 330
 331
 332
 333
 334
 335
 336
 337
 338
 339
 340
 341
 342
 343
 344
 345
 346
 347
 348
 349
 350
 351
 352
 353
 354
 355
 356
 357
 358
 359
 360
 361
 362
 363
 364
 365
 366
 367
 368
 369
 370
 371
 372
 373
 374
 375
 376
 377
 378
 379
 380
 381
 382
 383
 384
 385
 386
 387
 388
 389
 390
 391
 392
 393
 394
 395
 396
 397
 398
 399
 400
 401
 402
 403
 404
 405
 406
 407
 408
 409
 410
 411
 412
 413
 414
 415
 416
 417
 418
 419
 420
 421
 422
 423
 424
 425
 426
 427
 428
 429
 430
 431
 432
 433
 434
 435
 436
 437
 438
 439
 440
 441
 442
 443
 444
 445
 446
 447
 448
 449
 450
 451
 452
 453
 454
 455
 456
 457
 458
 459
 460
 461
 462
 463
 464
 465
 466
 467
 468
 469
 470
 471
 472
 473
 474
 475
 476
 477
 478
 479
 480
 481
 482
 483
 484
 485
 486
 487
 488
 489
 490
 491
 492
 493
 494
 495
 496
 497
 498
 499
 500
 501
 502
 503
 504
 505
 506
 507
 508
 509
 510
 511
 512
 513
 514
 515
 516
 517
 518
 519
 520
 521
 522
 523
 524
 525
 526
 527
 528
 529
 530
 531
 532
 533
 534
 535
 536
 537
 538
 539
 540
 541
 542
 543
 544
 545
 546
 547
 548
 549
 550
 551
 552
 553
 554
 555
 556
 557
 558
 559
 560
 561
 562
 563
 564
 565
 566
 567
 568
 569
 570
 571
 572
 573
 574
 575
 576
 577
 578
 579
 580
 581
 582
 583
 584
 585
 586
 587
 588
 589
 590
 591
 592
 593
 594
 595
 596
 597
 598
 599
 600
 601
 602
 603
 604
 605
 606
 607
 608
 609
 610
 611
 612
 613
 614
 615
 616
 617
 618
 619
 620
 621
 622
 623
 624
 625
 626
 627
 628
 629
 630
 631
 632
 633
 634
 635
 636
 637
 638
 639
 640
 641
 642
 643
 644
 645
 646
 647
 648
 649
 650
 651
 652
 653
 654
 655
 656
 657
 658
 659
 660
 661
 662
 663
 664
 665
 666
 667
 668
 669
 670
 671
 672
 673
 674
 675
 676
 677
 678
 679
 680
 681
 682
 683
 684
 685
 686
 687
 688
 689
 690
 691
 692
 693
 694
 695
 696
 697
 698
 699
 700
 701
 702
 703
 704
 705
 706
 707
 708
 709
 710
 711
 712
 713
 714
 715
 716
 717
 718
 719
 720
 721
 722
 723
 724
 725
 726
 727
 728
 729
 730
 731
 732
 733
 734
 735
 736
 737
 738
 739
 740
 741
 742
 743
 744
 745
 746
 747
 748
 749
 750
 751
 752
 753
 754
 755
 756
 757
 758
 759
 760
 761
 762
 763
 764
 765
 766
 767
 768
 769
 770
 771
 772
 773
 774
 775
 776
 777
 778
 779
 780
 781
 782
 783
 784
 785
 786
 787
 788
 789
 790
 791
 792
 793
 794
 795
 796
 797
 798
 799
 800
 801
 802
 803
 804
 805
 806
 807
 808
 809
 810
 811
 812
 813
 814
 815
 816
 817
 818
 819
 820
 821
 822
 823
 824
 825
 826
 827
 828
 829
 830
 831
 832
 833
 834
 835
 836
 837
 838
 839
 840
 841
 842
 843
 844
 845
 846
 847
 848
 849
 850
 851
 852
 853
 854
 855
 856
 857
 858
 859
 860
 861
 862
 863
 864
 865
 866
 867
 868
 869
 870
 871
 872
 873
 874
 875
 876
 877
 878
 879
 880
 881
 882
 883
 884
 885
 886
 887
 888
 889
 890
 891
 892
 893
 894
 895
 896
 897
 898
 899
 900
 901
 902
 903
 904
 905
 906
 907
 908
 909
 910
 911
 912
 913
 914
 915
 916
 917
 918
 919
 920
 921
 922
 923
 924
 925
 926
 927
 928
 929
 930
 931
 932
 933
 934
 935
 936
 937
 938
 939
 940
 941
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
/*************************************************
*      Perl-Compatible Regular Expressions       *
*************************************************/

/* PCRE is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language.

                       Written by Philip Hazel
           Copyright (c) 1997-2009 University of Cambridge

-----------------------------------------------------------------------------
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:

    * Redistributions of source code must retain the above copyright notice,
      this list of conditions and the following disclaimer.

    * Redistributions in binary form must reproduce the above copyright
      notice, this list of conditions and the following disclaimer in the
      documentation and/or other materials provided with the distribution.

    * Neither the name of the University of Cambridge nor the names of its
      contributors may be used to endorse or promote products derived from
      this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
-----------------------------------------------------------------------------
*/


/* This module contains pcre_exec(), the externally visible function that does
pattern matching using an NFA algorithm, trying to mimic Perl as closely as
possible. There are also some static supporting functions. */

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#define NLBLOCK md             /* Block containing newline information */
#define PSSTART start_subject  /* Field containing processed string start */
#define PSEND   end_subject    /* Field containing processed string end */

#include "pcreinal.h"

/* Undefine some potentially clashing cpp symbols */

#undef min
#undef max

/* Flag bits for the match() function */

#define match_condassert     0x01  /* Called to check a condition assertion */
#define match_cbegroup       0x02  /* Could-be-empty unlimited repeat group */

/* Non-error returns from the match() function. Error returns are externally
defined PCRE_ERROR_xxx codes, which are all negative. */

#define MATCH_MATCH        1
#define MATCH_NOMATCH      0

/* Special internal returns from the match() function. Make them sufficiently
negative to avoid the external error codes. */

#define MATCH_COMMIT       (-999)
#define MATCH_PRUNE        (-998)
#define MATCH_SKIP         (-997)
#define MATCH_THEN         (-996)

/* Maximum number of ints of offset to save on the stack for recursive calls.
If the offset vector is bigger, malloc is used. This should be a multiple of 3,
because the offset vector is always a multiple of 3 long. */

#define REC_STACK_SAVE_MAX 30

/* Min and max values for the common repeats; for the maxima, 0 => infinity */

static const char rep_min[] = { 0, 0, 1, 1, 0, 0 };
static const char rep_max[] = { 0, 0, 0, 0, 1, 1 };



#ifdef DEBUG
/*************************************************
*        Debugging function to print chars       *
*************************************************/

/* Print a sequence of chars in printable format, stopping at the end of the
subject if the requested.

Arguments:
  p           points to characters
  length      number to print
  is_subject  TRUE if printing from within md->start_subject
  md          pointer to matching data block, if is_subject is TRUE

Returns:     nothing
*/

static void
pchars(const uschar *p, int length, BOOL is_subject, match_data *md)
{
unsigned int c;
if (is_subject && length > md->end_subject - p) length = md->end_subject - p;
while (length-- > 0)
  if (isprint(c = *(p++))) printf("%c", c); else printf("\\x%02x", c);
}
#endif



/*************************************************
*          Match a back-reference                *
*************************************************/

/* If a back reference hasn't been set, the length that is passed is greater
than the number of characters left in the string, so the match fails.

Arguments:
  offset      index into the offset vector
  eptr        points into the subject
  length      length to be matched
  md          points to match data block
  ims         the ims flags

Returns:      TRUE if matched
*/

static BOOL
match_ref(int offset, register USPTR eptr, int length, match_data *md,
  unsigned long int ims)
{
USPTR p = md->start_subject + md->offset_vector[offset];

#ifdef DEBUG
if (eptr >= md->end_subject)
  printf("matching subject <null>");
else
  {
  printf("matching subject ");
  pchars(eptr, length, TRUE, md);
  }
printf(" against backref ");
pchars(p, length, FALSE, md);
printf("\n");
#endif

/* Always fail if not enough characters left */

if (length > md->end_subject - eptr) return FALSE;

/* Separate the caseless case for speed. In UTF-8 mode we can only do this
properly if Unicode properties are supported. Otherwise, we can check only
ASCII characters. */

if ((ims & PCRE_CASELESS) != 0)
  {
#ifdef SUPPORT_UTF8
#ifdef SUPPORT_UCP
  if (md->utf8)
    {
    USPTR endptr = eptr + length;
    while (eptr < endptr)
      {
      int c, d;
      GETCHARINC(c, eptr);
      GETCHARINC(d, p);
      if (c != d && c != UCD_OTHERCASE(d)) return FALSE;
      }
    }
  else
#endif
#endif

  /* The same code works when not in UTF-8 mode and in UTF-8 mode when there
  is no UCP support. */

  while (length-- > 0)
    { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; }
  }

/* In the caseful case, we can just compare the bytes, whether or not we
are in UTF-8 mode. */

else
  { while (length-- > 0) if (*p++ != *eptr++) return FALSE; }

return TRUE;
}



/***************************************************************************
****************************************************************************
                   RECURSION IN THE match() FUNCTION

The match() function is highly recursive, though not every recursive call
increases the recursive depth. Nevertheless, some regular expressions can cause
it to recurse to a great depth. I was writing for Unix, so I just let it call
itself recursively. This uses the stack for saving everything that has to be
saved for a recursive call. On Unix, the stack can be large, and this works
fine.

It turns out that on some non-Unix-like systems there are problems with
programs that use a lot of stack. (This despite the fact that every last chip
has oodles of memory these days, and techniques for extending the stack have
been known for decades.) So....

There is a fudge, triggered by defining NO_RECURSE, which avoids recursive
calls by keeping local variables that need to be preserved in blocks of memory
obtained from malloc() instead instead of on the stack. Macros are used to
achieve this so that the actual code doesn't look very different to what it
always used to.

The original heap-recursive code used longjmp(). However, it seems that this
can be very slow on some operating systems. Following a suggestion from Stan
Switzer, the use of longjmp() has been abolished, at the cost of having to
provide a unique number for each call to RMATCH. There is no way of generating
a sequence of numbers at compile time in C. I have given them names, to make
them stand out more clearly.

Crude tests on x86 Linux show a small speedup of around 5-8%. However, on
FreeBSD, avoiding longjmp() more than halves the time taken to run the standard
tests. Furthermore, not using longjmp() means that local dynamic variables
don't have indeterminate values; this has meant that the frame size can be
reduced because the result can be "passed back" by straight setting of the
variable instead of being passed in the frame.
****************************************************************************
***************************************************************************/

/* Numbers for RMATCH calls. When this list is changed, the code at HEAP_RETURN
below must be updated in sync.  */

enum { RM1=1, RM2,  RM3,  RM4,  RM5,  RM6,  RM7,  RM8,  RM9,  RM10,
       RM11,  RM12, RM13, RM14, RM15, RM16, RM17, RM18, RM19, RM20,
       RM21,  RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30,
       RM31,  RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40,
       RM41,  RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50,
       RM51,  RM52, RM53, RM54 };

/* These versions of the macros use the stack, as normal. There are debugging
versions and production versions. Note that the "rw" argument of RMATCH isn't
actuall used in this definition. */

#ifndef NO_RECURSE
#define REGISTER register

#ifdef DEBUG
#define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \
  { \
  printf("match() called in line %d\n", __LINE__); \
  rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \
  printf("to line %d\n", __LINE__); \
  }
#define RRETURN(ra) \
  { \
  printf("match() returned %d from line %d ", ra, __LINE__); \
  return ra; \
  }
#else
#define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \
  rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1)
#define RRETURN(ra) return ra
#endif

#else


/* These versions of the macros manage a private stack on the heap. Note that
the "rd" argument of RMATCH isn't actually used in this definition. It's the md
argument of match(), which never changes. */

#define REGISTER

#define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\
  {\
  heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\
  frame->Xwhere = rw; \
  newframe->Xeptr = ra;\
  newframe->Xecode = rb;\
  newframe->Xmstart = mstart;\
  newframe->Xoffset_top = rc;\
  newframe->Xims = re;\
  newframe->Xeptrb = rf;\
  newframe->Xflags = rg;\
  newframe->Xrdepth = frame->Xrdepth + 1;\
  newframe->Xprevframe = frame;\
  frame = newframe;\
  DPRINTF(("restarting from line %d\n", __LINE__));\
  goto HEAP_RECURSE;\
  L_##rw:\
  DPRINTF(("jumped back to line %d\n", __LINE__));\
  }

#define RRETURN(ra)\
  {\
  heapframe *newframe = frame;\
  frame = newframe->Xprevframe;\
  (pcre_stack_free)(newframe);\
  if (frame != NULL)\
    {\
    rrc = ra;\
    goto HEAP_RETURN;\
    }\
  return ra;\
  }


/* Structure for remembering the local variables in a private frame */

typedef struct heapframe {
  struct heapframe *Xprevframe;

  /* Function arguments that may change */

  USPTR Xeptr;
  const uschar *Xecode;
  USPTR Xmstart;
  int Xoffset_top;
  long int Xims;
  eptrblock *Xeptrb;
  int Xflags;
  unsigned int Xrdepth;

  /* Function local variables */

  USPTR Xcallpat;
#ifdef SUPPORT_UTF8
  USPTR Xcharptr;
#endif
  USPTR Xdata;
  USPTR Xnext;
  USPTR Xpp;
  USPTR Xprev;
  USPTR Xsaved_eptr;

  recursion_info Xnew_recursive;

  BOOL Xcur_is_word;
  BOOL Xcondition;
  BOOL Xprev_is_word;

  unsigned long int Xoriginal_ims;

#ifdef SUPPORT_UCP
  int Xprop_type;
  int Xprop_value;
  int Xprop_fail_result;
  int Xprop_category;
  int Xprop_chartype;
  int Xprop_script;
  int Xoclength;
  uschar Xocchars[8];
#endif

  int Xcodelink;
  int Xctype;
  unsigned int Xfc;
  int Xfi;
  int Xlength;
  int Xmax;
  int Xmin;
  int Xnumber;
  int Xoffset;
  int Xop;
  int Xsave_capture_last;
  int Xsave_offset1, Xsave_offset2, Xsave_offset3;
  int Xstacksave[REC_STACK_SAVE_MAX];

  eptrblock Xnewptrb;

  /* Where to jump back to */

  int Xwhere;

} heapframe;

#endif


/***************************************************************************
***************************************************************************/



/*************************************************
*         Match from current position            *
*************************************************/

/* This function is called recursively in many circumstances. Whenever it
returns a negative (error) response, the outer incarnation must also return the
same response. */

/* These macros pack up tests that are used for partial matching, and which
appears several times in the code. We set the "hit end" flag if the pointer is
at the end of the subject and also past the start of the subject (i.e.
something has been matched). For hard partial matching, we then return
immediately. The second one is used when we already know we are past the end of
the subject. */

#define CHECK_PARTIAL()\
  if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\
    {\
    md->hitend = TRUE;\
    if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\
    }

#define SCHECK_PARTIAL()\
  if (md->partial != 0 && eptr > mstart)\
    {\
    md->hitend = TRUE;\
    if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL);\
    }


/* Performance note: It might be tempting to extract commonly used fields from
the md structure (e.g. utf8, end_subject) into individual variables to improve
performance. Tests using gcc on a SPARC disproved this; in the first case, it
made performance worse.

Arguments:
   eptr        pointer to current character in subject
   ecode       pointer to current position in compiled code
   mstart      pointer to the current match start position (can be modified
                 by encountering \K)
   offset_top  current top pointer
   md          pointer to "static" info for the match
   ims         current /i, /m, and /s options
   eptrb       pointer to chain of blocks containing eptr at start of
                 brackets - for testing for empty matches
   flags       can contain
                 match_condassert - this is an assertion condition
                 match_cbegroup - this is the start of an unlimited repeat
                   group that can match an empty string
   rdepth      the recursion depth

Returns:       MATCH_MATCH if matched            )  these values are >= 0
               MATCH_NOMATCH if failed to match  )
               a negative PCRE_ERROR_xxx value if aborted by an error condition
                 (e.g. stopped by repeated call or recursion limit)
*/

static int
match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart,
  int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb,
  int flags, unsigned int rdepth)
{
/* These variables do not need to be preserved over recursion in this function,
so they can be ordinary variables in all cases. Mark some of them with
"register" because they are used a lot in loops. */

register int  rrc;         /* Returns from recursive calls */
register int  i;           /* Used for loops not involving calls to RMATCH() */
register unsigned int c;   /* Character values not kept over RMATCH() calls */
register BOOL utf8;        /* Local copy of UTF-8 flag for speed */

BOOL minimize, possessive; /* Quantifier options */
int condcode;

/* When recursion is not being used, all "local" variables that have to be
preserved over calls to RMATCH() are part of a "frame" which is obtained from
heap storage. Set up the top-level frame here; others are obtained from the
heap whenever RMATCH() does a "recursion". See the macro definitions above. */

#ifdef NO_RECURSE
heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe));
frame->Xprevframe = NULL;            /* Marks the top level */

/* Copy in the original argument variables */

frame->Xeptr = eptr;
frame->Xecode = ecode;
frame->Xmstart = mstart;
frame->Xoffset_top = offset_top;
frame->Xims = ims;
frame->Xeptrb = eptrb;
frame->Xflags = flags;
frame->Xrdepth = rdepth;

/* This is where control jumps back to to effect "recursion" */

HEAP_RECURSE:

/* Macros make the argument variables come from the current frame */

#define eptr               frame->Xeptr
#define ecode              frame->Xecode
#define mstart             frame->Xmstart
#define offset_top         frame->Xoffset_top
#define ims                frame->Xims
#define eptrb              frame->Xeptrb
#define flags              frame->Xflags
#define rdepth             frame->Xrdepth

/* Ditto for the local variables */

#ifdef SUPPORT_UTF8
#define charptr            frame->Xcharptr
#endif
#define callpat            frame->Xcallpat
#define codelink           frame->Xcodelink
#define data               frame->Xdata
#define next               frame->Xnext
#define pp                 frame->Xpp
#define prev               frame->Xprev
#define saved_eptr         frame->Xsaved_eptr

#define new_recursive      frame->Xnew_recursive

#define cur_is_word        frame->Xcur_is_word
#define condition          frame->Xcondition
#define prev_is_word       frame->Xprev_is_word

#define original_ims       frame->Xoriginal_ims

#ifdef SUPPORT_UCP
#define prop_type          frame->Xprop_type
#define prop_value         frame->Xprop_value
#define prop_fail_result   frame->Xprop_fail_result
#define prop_category      frame->Xprop_category
#define prop_chartype      frame->Xprop_chartype
#define prop_script        frame->Xprop_script
#define oclength           frame->Xoclength
#define occhars            frame->Xocchars
#endif

#define ctype              frame->Xctype
#define fc                 frame->Xfc
#define fi                 frame->Xfi
#define length             frame->Xlength
#define max                frame->Xmax
#define min                frame->Xmin
#define number             frame->Xnumber
#define offset             frame->Xoffset
#define op                 frame->Xop
#define save_capture_last  frame->Xsave_capture_last
#define save_offset1       frame->Xsave_offset1
#define save_offset2       frame->Xsave_offset2
#define save_offset3       frame->Xsave_offset3
#define stacksave          frame->Xstacksave

#define newptrb            frame->Xnewptrb

/* When recursion is being used, local variables are allocated on the stack and
get preserved during recursion in the normal way. In this environment, fi and
i, and fc and c, can be the same variables. */

#else         /* NO_RECURSE not defined */
#define fi i
#define fc c


#ifdef SUPPORT_UTF8                /* Many of these variables are used only  */
const uschar *charptr;             /* in small blocks of the code. My normal */
#endif                             /* style of coding would have declared    */
const uschar *callpat;             /* them within each of those blocks.      */
const uschar *data;                /* However, in order to accommodate the   */
const uschar *next;                /* version of this code that uses an      */
USPTR         pp;                  /* external "stack" implemented on the    */
const uschar *prev;                /* heap, it is easier to declare them all */
USPTR         saved_eptr;          /* here, so the declarations can be cut   */
                                   /* out in a block. The only declarations  */
recursion_info new_recursive;      /* within blocks below are for variables  */
                                   /* that do not have to be preserved over  */
BOOL cur_is_word;                  /* a recursive call to RMATCH().          */
BOOL condition;
BOOL prev_is_word;

unsigned long int original_ims;

#ifdef SUPPORT_UCP
int prop_type;
int prop_value;
int prop_fail_result;
int prop_category;
int prop_chartype;
int prop_script;
int oclength;
uschar occhars[8];
#endif

int codelink;
int ctype;
int length;
int max;
int min;
int number;
int offset;
int op;
int save_capture_last;
int save_offset1, save_offset2, save_offset3;
int stacksave[REC_STACK_SAVE_MAX];

eptrblock newptrb;
#endif     /* NO_RECURSE */

/* These statements are here to stop the compiler complaining about unitialized
variables. */

#ifdef SUPPORT_UCP
prop_value = 0;
prop_fail_result = 0;
#endif


/* This label is used for tail recursion, which is used in a few cases even
when NO_RECURSE is not defined, in order to reduce the amount of stack that is
used. Thanks to Ian Taylor for noticing this possibility and sending the
original patch. */

TAIL_RECURSE:

/* OK, now we can get on with the real code of the function. Recursive calls
are specified by the macro RMATCH and RRETURN is used to return. When
NO_RECURSE is *not* defined, these just turn into a recursive call to match()
and a "return", respectively (possibly with some debugging if DEBUG is
defined). However, RMATCH isn't like a function call because it's quite a
complicated macro. It has to be used in one particular way. This shouldn't,
however, impact performance when true recursion is being used. */

#ifdef SUPPORT_UTF8
utf8 = md->utf8;       /* Local copy of the flag */
#else
utf8 = FALSE;
#endif

/* First check that we haven't called match() too many times, or that we
haven't exceeded the recursive call limit. */

if (md->match_call_count++ >= md->match_limit) RRETURN(PCRE_ERROR_MATCHLIMIT);
if (rdepth >= md->match_limit_recursion) RRETURN(PCRE_ERROR_RECURSIONLIMIT);

original_ims = ims;    /* Save for resetting on ')' */

/* At the start of a group with an unlimited repeat that may match an empty
string, the match_cbegroup flag is set. When this is the case, add the current
subject pointer to the chain of such remembered pointers, to be checked when we
hit the closing ket, in order to break infinite loops that match no characters.
When match() is called in other circumstances, don't add to the chain. The
match_cbegroup flag must NOT be used with tail recursion, because the memory
block that is used is on the stack, so a new one may be required for each
match(). */

if ((flags & match_cbegroup) != 0)
  {
  newptrb.epb_saved_eptr = eptr;
  newptrb.epb_prev = eptrb;
  eptrb = &newptrb;
  }

/* Now start processing the opcodes. */

for (;;)
  {
  minimize = possessive = FALSE;
  op = *ecode;

  switch(op)
    {
    case OP_FAIL:
    RRETURN(MATCH_NOMATCH);

    case OP_PRUNE:
    RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
      ims, eptrb, flags, RM51);
    if (rrc != MATCH_NOMATCH) RRETURN(rrc);
    RRETURN(MATCH_PRUNE);

    case OP_COMMIT:
    RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
      ims, eptrb, flags, RM52);
    if (rrc != MATCH_NOMATCH) RRETURN(rrc);
    RRETURN(MATCH_COMMIT);

    case OP_SKIP:
    RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
      ims, eptrb, flags, RM53);
    if (rrc != MATCH_NOMATCH) RRETURN(rrc);
    md->start_match_ptr = eptr;   /* Pass back current position */
    RRETURN(MATCH_SKIP);

    case OP_THEN:
    RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
      ims, eptrb, flags, RM54);
    if (rrc != MATCH_NOMATCH) RRETURN(rrc);
    RRETURN(MATCH_THEN);

    /* Handle a capturing bracket. If there is space in the offset vector, save
    the current subject position in the working slot at the top of the vector.
    We mustn't change the current values of the data slot, because they may be
    set from a previous iteration of this group, and be referred to by a
    reference inside the group.

    If the bracket fails to match, we need to restore this value and also the
    values of the final offsets, in case they were set by a previous iteration
    of the same bracket.

    If there isn't enough space in the offset vector, treat this as if it were
    a non-capturing bracket. Don't worry about setting the flag for the error
    case here; that is handled in the code for KET. */

    case OP_CBRA:
    case OP_SCBRA:
    number = GET2(ecode, 1+LINK_SIZE);
    offset = number << 1;

#ifdef DEBUG
    printf("start bracket %d\n", number);
    printf("subject=");
    pchars(eptr, 16, TRUE, md);
    printf("\n");
#endif

    if (offset < md->offset_max)
      {
      save_offset1 = md->offset_vector[offset];
      save_offset2 = md->offset_vector[offset+1];
      save_offset3 = md->offset_vector[md->offset_end - number];
      save_capture_last = md->capture_last;

      DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3));
      md->offset_vector[md->offset_end - number] = eptr - md->start_subject;

      flags = (op == OP_SCBRA)? match_cbegroup : 0;
      do
        {
        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md,
          ims, eptrb, flags, RM1);
        if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
        md->capture_last = save_capture_last;
        ecode += GET(ecode, 1);
        }
      while (*ecode == OP_ALT);

      DPRINTF(("bracket %d failed\n", number));

      md->offset_vector[offset] = save_offset1;
      md->offset_vector[offset+1] = save_offset2;
      md->offset_vector[md->offset_end - number] = save_offset3;

      RRETURN(MATCH_NOMATCH);
      }

    /* FALL THROUGH ... Insufficient room for saving captured contents. Treat
    as a non-capturing bracket. */

    /* VVVVVVVVVVVVVVVVVVVVVVVVV */
    /* VVVVVVVVVVVVVVVVVVVVVVVVV */

    DPRINTF(("insufficient capture room: treat as non-capturing\n"));

    /* VVVVVVVVVVVVVVVVVVVVVVVVV */
    /* VVVVVVVVVVVVVVVVVVVVVVVVV */

    /* Non-capturing bracket. Loop for all the alternatives. When we get to the
    final alternative within the brackets, we would return the result of a
    recursive call to match() whatever happened. We can reduce stack usage by
    turning this into a tail recursion, except in the case when match_cbegroup
    is set.*/

    case OP_BRA:
    case OP_SBRA:
    DPRINTF(("start non-capturing bracket\n"));
    flags = (op >= OP_SBRA)? match_cbegroup : 0;
    for (;;)
      {
      if (ecode[GET(ecode, 1)] != OP_ALT)   /* Final alternative */
        {
        if (flags == 0)    /* Not a possibly empty group */
          {
          ecode += _pcre_OP_lengths[*ecode];
          DPRINTF(("bracket 0 tail recursion\n"));
          goto TAIL_RECURSE;
          }

        /* Possibly empty group; can't use tail recursion. */

        RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,
          eptrb, flags, RM48);
        RRETURN(rrc);
        }

      /* For non-final alternatives, continue the loop for a NOMATCH result;
      otherwise return. */

      RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims,
        eptrb, flags, RM2);
      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
      ecode += GET(ecode, 1);
      }
    /* Control never reaches here. */

    /* Conditional group: compilation checked that there are no more than
    two branches. If the condition is false, skipping the first branch takes us
    past the end if there is only one branch, but that's OK because that is
    exactly what going to the ket would do. As there is only one branch to be
    obeyed, we can use tail recursion to avoid using another stack frame. */

    case OP_COND:
    case OP_SCOND:
    codelink= GET(ecode, 1);

    /* Because of the way auto-callout works during compile, a callout item is
    inserted between OP_COND and an assertion condition. */

    if (ecode[LINK_SIZE+1] == OP_CALLOUT)
      {
      if (pcre_callout != NULL)
        {
        pcre_callout_block cb;
        cb.version          = 1;   /* Version 1 of the callout block */
        cb.callout_number   = ecode[LINK_SIZE+2];
        cb.offset_vector    = md->offset_vector;
        cb.subject          = (PCRE_SPTR)md->start_subject;
        cb.subject_length   = md->end_subject - md->start_subject;
        cb.start_match      = mstart - md->start_subject;
        cb.current_position = eptr - md->start_subject;
        cb.pattern_position = GET(ecode, LINK_SIZE + 3);
        cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE);
        cb.capture_top      = offset_top/2;
        cb.capture_last     = md->capture_last;
        cb.callout_data     = md->callout_data;
        if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);
        if (rrc < 0) RRETURN(rrc);
        }
      ecode += _pcre_OP_lengths[OP_CALLOUT];
      }

    condcode = ecode[LINK_SIZE+1];

    /* Now see what the actual condition is */

    if (condcode == OP_RREF || condcode == OP_NRREF)    /* Recursion test */
      {
      if (md->recursive == NULL)                /* Not recursing => FALSE */
        {
        condition = FALSE;
        ecode += GET(ecode, 1);
        }
      else
        {
        int recno = GET2(ecode, LINK_SIZE + 2);   /* Recursion group number*/
        condition =  (recno == RREF_ANY || recno == md->recursive->group_num);

        /* If the test is for recursion into a specific subpattern, and it is
        false, but the test was set up by name, scan the table to see if the
        name refers to any other numbers, and test them. The condition is true
        if any one is set. */

        if (!condition && condcode == OP_NRREF && recno != RREF_ANY)
          {
          uschar *slotA = md->name_table;
          for (i = 0; i < md->name_count; i++)
            {
            if (GET2(slotA, 0) == recno) break;
            slotA += md->name_entry_size;
            }

          /* Found a name for the number - there can be only one; duplicate
          names for different numbers are allowed, but not vice versa. First
          scan down for duplicates. */

          if (i < md->name_count)
            {
            uschar *slotB = slotA;
            while (slotB > md->name_table)
              {
              slotB -= md->name_entry_size;
              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
                {
                condition = GET2(slotB, 0) == md->recursive->group_num;
                if (condition) break;
                }
              else break;
              }

            /* Scan up for duplicates */

            if (!condition)
              {
              slotB = slotA;
              for (i++; i < md->name_count; i++)
                {
                slotB += md->name_entry_size;
                if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
                  {
                  condition = GET2(slotB, 0) == md->recursive->group_num;
                  if (condition) break;
                  }
                else break;
                }
              }
            }
          }

        /* Chose branch according to the condition */

        ecode += condition? 3 : GET(ecode, 1);
        }
      }

    else if (condcode == OP_CREF || condcode == OP_NCREF)  /* Group used test */
      {
      offset = GET2(ecode, LINK_SIZE+2) << 1;  /* Doubled ref number */
      condition = offset < offset_top && md->offset_vector[offset] >= 0;

      /* If the numbered capture is unset, but the reference was by name,
      scan the table to see if the name refers to any other numbers, and test
      them. The condition is true if any one is set. This is tediously similar
      to the code above, but not close enough to try to amalgamate. */

      if (!condition && condcode == OP_NCREF)
        {
        int refno = offset >> 1;
        uschar *slotA = md->name_table;

        for (i = 0; i < md->name_count; i++)
          {
          if (GET2(slotA, 0) == refno) break;
          slotA += md->name_entry_size;
          }

        /* Found a name for the number - there can be only one; duplicate names
        for different numbers are allowed, but not vice versa. First scan down
        for duplicates. */

        if (i < md->name_count)
          {
          uschar *slotB = slotA;
          while (slotB > md->name_table)
            {
            slotB -= md->name_entry_size;
            if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
              {
              offset = GET2(slotB, 0) << 1;
              condition = offset < offset_top &&
                md->offset_vector[offset] >= 0;
              if (condition) break;
              }
            else break;
            }

          /* Scan up for duplicates */

          if (!condition)
            {
            slotB = slotA;
            for (i++; i < md->name_count; i++)
              {
              slotB += md->name_entry_size;
              if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0)
                {
                offset = GET2(slotB, 0) << 1;
                condition = offset < offset_top &&
                  md->offset_vector[offset] >= 0;
                if (condition) break;
                }
              else break;
              }
            }
          }
        }

      /* Chose branch according to the condition */

      ecode += condition? 3 : GET(ecode, 1);
      }

    else if (condcode == OP_DEF)     /* DEFINE - always false */
      {
      condition = FALSE;
      ecode += GET(ecode, 1);
      }

    /* The condition is an assertion. Call match() to evaluate it - setting
    the final argument match_condassert causes it to stop at the end of an
    assertion. */

    else
      {
      RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL,
          match_condassert, RM3);
      if (rrc == MATCH_MATCH)
        {
        condition = TRUE;
        ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2);
        while (*ecode == OP_ALT) ecode += GET(ecode, 1);
        }
      else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
        {
        RRETURN(rrc);         /* Need braces because of following else */
        }
      else
        {
        condition = FALSE;
        ecode += codelink;
        }
      }

    /* We are now at the branch that is to be obeyed. As there is only one,
    we can use tail recursion to avoid using another stack frame, except when
    match_cbegroup is required for an unlimited repeat of a possibly empty
    group. If the second alternative doesn't exist, we can just plough on. */

    if (condition || *ecode == OP_ALT)
      {
      ecode += 1 + LINK_SIZE;
      if (op == OP_SCOND)        /* Possibly empty group */
        {
        RMATCH(eptr, ecode, offset_top, md, ims, eptrb, match_cbegroup, RM49);
        RRETURN(rrc);
        }
      else                       /* Group must match something */
        {
        flags = 0;
        goto TAIL_RECURSE;
        }
      }
    else                         /* Condition false & no alternative */
      {
      ecode += 1 + LINK_SIZE;
      }
    break;


    /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes,
    to close any currently open capturing brackets. */

    case OP_CLOSE:
    number = GET2(ecode, 1);
    offset = number << 1;

#ifdef DEBUG
      printf("end bracket %d at *ACCEPT", number);
      printf("\n");
#endif

    md->capture_last = number;
    if (offset >= md->offset_max) md->offset_overflow = TRUE; else
      {
      md->offset_vector[offset] =
        md->offset_vector[md->offset_end - number];
      md->offset_vector[offset+1] = eptr - md->start_subject;
      if (offset_top <= offset) offset_top = offset + 2;
      }
    ecode += 3;
    break;


    /* End of the pattern, either real or forced. If we are in a top-level
    recursion, we should restore the offsets appropriately and continue from
    after the call. */

    case OP_ACCEPT:
    case OP_END:
    if (md->recursive != NULL && md->recursive->group_num == 0)
      {
      recursion_info *rec = md->recursive;
      DPRINTF(("End of pattern in a (?0) recursion\n"));
      md->recursive = rec->prevrec;
      memmove(md->offset_vector, rec->offset_save,
        rec->saved_max * sizeof(int));
      offset_top = rec->save_offset_top;
      mstart = rec->save_start;
      ims = original_ims;
      ecode = rec->after_call;
      break;
      }

    /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is
    set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of
    the subject. In both cases, backtracking will then try other alternatives,
    if any. */

    if (eptr == mstart &&
        (md->notempty ||
          (md->notempty_atstart &&
            mstart == md->start_subject + md->start_offset)))
      RRETURN(MATCH_NOMATCH);

    /* Otherwise, we have a match. */

    md->end_match_ptr = eptr;           /* Record where we ended */
    md->end_offset_top = offset_top;    /* and how many extracts were taken */
    md->start_match_ptr = mstart;       /* and the start (\K can modify) */
    RRETURN(MATCH_MATCH);

    /* Change option settings */

    case OP_OPT:
    ims = ecode[1];
    ecode += 2;
    DPRINTF(("ims set to %02lx\n", ims));
    break;

    /* Assertion brackets. Check the alternative branches in turn - the
    matching won't pass the KET for an assertion. If any one branch matches,
    the assertion is true. Lookbehind assertions have an OP_REVERSE item at the
    start of each branch to move the current point backwards, so the code at
    this level is identical to the lookahead case. */

    case OP_ASSERT:
    case OP_ASSERTBACK:
    do
      {
      RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
        RM4);
      if (rrc == MATCH_MATCH) break;
      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
      ecode += GET(ecode, 1);
      }
    while (*ecode == OP_ALT);
    if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH);

    /* If checking an assertion for a condition, return MATCH_MATCH. */

    if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);

    /* Continue from after the assertion, updating the offsets high water
    mark, since extracts may have been taken during the assertion. */

    do ecode += GET(ecode,1); while (*ecode == OP_ALT);
    ecode += 1 + LINK_SIZE;
    offset_top = md->end_offset_top;
    continue;

    /* Negative assertion: all branches must fail to match */

    case OP_ASSERT_NOT:
    case OP_ASSERTBACK_NOT:
    do
      {
      RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0,
        RM5);
      if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH);
      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
      ecode += GET(ecode,1);
      }
    while (*ecode == OP_ALT);

    if ((flags & match_condassert) != 0) RRETURN(MATCH_MATCH);

    ecode += 1 + LINK_SIZE;
    continue;

    /* Move the subject pointer back. This occurs only at the start of
    each branch of a lookbehind assertion. If we are too close to the start to
    move back, this match function fails. When working with UTF-8 we move
    back a number of characters, not bytes. */

    case OP_REVERSE:
#ifdef SUPPORT_UTF8
    if (utf8)
      {
      i = GET(ecode, 1);
      while (i-- > 0)
        {
        eptr--;
        if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
        BACKCHAR(eptr);
        }
      }
    else
#endif

    /* No UTF-8 support, or not in UTF-8 mode: count is byte count */

      {
      eptr -= GET(ecode, 1);
      if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH);
      }

    /* Save the earliest consulted character, then skip to next op code */

    if (eptr < md->start_used_ptr) md->start_used_ptr = eptr;
    ecode += 1 + LINK_SIZE;
    break;

    /* The callout item calls an external function, if one is provided, passing
    details of the match so far. This is mainly for debugging, though the
    function is able to force a failure. */

    case OP_CALLOUT:
    if (pcre_callout != NULL)
      {
      pcre_callout_block cb;
      cb.version          = 1;   /* Version 1 of the callout block */
      cb.callout_number   = ecode[1];
      cb.offset_vector    = md->offset_vector;
      cb.subject          = (PCRE_SPTR)md->start_subject;
      cb.subject_length   = md->end_subject - md->start_subject;
      cb.start_match      = mstart - md->start_subject;
      cb.current_position = eptr - md->start_subject;
      cb.pattern_position = GET(ecode, 2);
      cb.next_item_length = GET(ecode, 2 + LINK_SIZE);
      cb.capture_top      = offset_top/2;
      cb.capture_last     = md->capture_last;
      cb.callout_data     = md->callout_data;
      if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH);
      if (rrc < 0) RRETURN(rrc);
      }
    ecode += 2 + 2*LINK_SIZE;
    break;

    /* Recursion either matches the current regex, or some subexpression. The
    offset data is the offset to the starting bracket from the start of the
    whole pattern. (This is so that it works from duplicated subpatterns.)

    If there are any capturing brackets started but not finished, we have to
    save their starting points and reinstate them after the recursion. However,
    we don't know how many such there are (offset_top records the completed
    total) so we just have to save all the potential data. There may be up to
    65535 such values, which is too large to put on the stack, but using malloc
    for small numbers seems expensive. As a compromise, the stack is used when
    there are no more than REC_STACK_SAVE_MAX values to store; otherwise malloc
    is used. A problem is what to do if the malloc fails ... there is no way of
    returning to the top level with an error. Save the top REC_STACK_SAVE_MAX
    values on the stack, and accept that the rest may be wrong.

    There are also other values that have to be saved. We use a chained
    sequence of blocks that actually live on the stack. Thanks to Robin Houston
    for the original version of this logic. */

    case OP_RECURSE:
      {
      callpat = md->start_code + GET(ecode, 1);
      new_recursive.group_num = (callpat == md->start_code)? 0 :
        GET2(callpat, 1 + LINK_SIZE);

      /* Add to "recursing stack" */

      new_recursive.prevrec = md->recursive;
      md->recursive = &new_recursive;

      /* Find where to continue from afterwards */

      ecode += 1 + LINK_SIZE;
      new_recursive.after_call = ecode;

      /* Now save the offset data. */

      new_recursive.saved_max = md->offset_end;
      if (new_recursive.saved_max <= REC_STACK_SAVE_MAX)
        new_recursive.offset_save = stacksave;
      else
        {
        new_recursive.offset_save =
          (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int));
        if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY);
        }

      memcpy(new_recursive.offset_save, md->offset_vector,
            new_recursive.saved_max * sizeof(int));
      new_recursive.save_start = mstart;
      new_recursive.save_offset_top = offset_top;
      mstart = eptr;

      /* OK, now we can do the recursion. For each top-level alternative we
      restore the offset and recursion data. */

      DPRINTF(("Recursing into group %d\n", new_recursive.group_num));
      flags = (*callpat >= OP_SBRA)? match_cbegroup : 0;
      do
        {
        RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top,
          md, ims, eptrb, flags, RM6);
        if (rrc == MATCH_MATCH)
          {
          DPRINTF(("Recursion matched\n"));
          md->recursive = new_recursive.prevrec;
          if (new_recursive.offset_save != stacksave)
            (pcre_free)(new_recursive.offset_save);
          RRETURN(MATCH_MATCH);
          }
        else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN)
          {
          DPRINTF(("Recursion gave error %d\n", rrc));
          if (new_recursive.offset_save != stacksave)
            (pcre_free)(new_recursive.offset_save);
          RRETURN(rrc);
          }

        md->recursive = &new_recursive;
        memcpy(md->offset_vector, new_recursive.offset_save,
            new_recursive.saved_max * sizeof(int));
        callpat += GET(callpat, 1);
        }
      while (*callpat == OP_ALT);

      DPRINTF(("Recursion didn't match\n"));
      md->recursive = new_recursive.prevrec;
      if (new_recursive.offset_save != stacksave)
        (pcre_free)(new_recursive.offset_save);
      RRETURN(MATCH_NOMATCH);
      }
    /* Control never reaches here */

    /* "Once" brackets are like assertion brackets except that after a match,
    the point in the subject string is not moved back. Thus there can never be
    a move back into the brackets. Friedl calls these "atomic" subpatterns.
    Check the alternative branches in turn - the matching won't pass the KET
    for this kind of subpattern. If any one branch matches, we carry on as at
    the end of a normal bracket, leaving the subject pointer. */

    case OP_ONCE:
    prev = ecode;
    saved_eptr = eptr;

    do
      {
      RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7);
      if (rrc == MATCH_MATCH) break;
      if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc);
      ecode += GET(ecode,1);
      }
    while (*ecode == OP_ALT);

    /* If hit the end of the group (which could be repeated), fail */

    if (*ecode != OP_ONCE && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH);

    /* Continue as from after the assertion, updating the offsets high water
    mark, since extracts may have been taken. */

    do ecode += GET(ecode, 1); while (*ecode == OP_ALT);

    offset_top = md->end_offset_top;
    eptr = md->end_match_ptr;

    /* For a non-repeating ket, just continue at this level. This also
    happens for a repeating ket if no characters were matched in the group.
    This is the forcible breaking of infinite loops as implemented in Perl
    5.005. If there is an options reset, it will get obeyed in the normal
    course of events. */

    if (*ecode == OP_KET || eptr == saved_eptr)
      {
      ecode += 1+LINK_SIZE;
      break;
      }

    /* The repeating kets try the rest of the pattern or restart from the
    preceding bracket, in the appropriate order. The second "call" of match()
    uses tail recursion, to avoid using another stack frame. We need to reset
    any options that changed within the bracket before re-running it, so
    check the next opcode. */

    if (ecode[1+LINK_SIZE] == OP_OPT)
      {
      ims = (ims & ~PCRE_IMS) | ecode[4];
      DPRINTF(("ims set to %02lx at group repeat\n", ims));
      }

    if (*ecode == OP_KETRMIN)
      {
      RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM8);
      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
      ecode = prev;
      flags = 0;
      goto TAIL_RECURSE;
      }
    else  /* OP_KETRMAX */
      {
      RMATCH(eptr, prev, offset_top, md, ims, eptrb, match_cbegroup, RM9);
      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
      ecode += 1 + LINK_SIZE;
      flags = 0;
      goto TAIL_RECURSE;
      }
    /* Control never gets here */

    /* An alternation is the end of a branch; scan along to find the end of the
    bracketed group and go to there. */

    case OP_ALT:
    do ecode += GET(ecode,1); while (*ecode == OP_ALT);
    break;

    /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group,
    indicating that it may occur zero times. It may repeat infinitely, or not
    at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets
    with fixed upper repeat limits are compiled as a number of copies, with the
    optional ones preceded by BRAZERO or BRAMINZERO. */

    case OP_BRAZERO:
      {
      next = ecode+1;
      RMATCH(eptr, next, offset_top, md, ims, eptrb, 0, RM10);
      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
      do next += GET(next,1); while (*next == OP_ALT);
      ecode = next + 1 + LINK_SIZE;
      }
    break;

    case OP_BRAMINZERO:
      {
      next = ecode+1;
      do next += GET(next, 1); while (*next == OP_ALT);
      RMATCH(eptr, next + 1+LINK_SIZE, offset_top, md, ims, eptrb, 0, RM11);
      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
      ecode++;
      }
    break;

    case OP_SKIPZERO:
      {
      next = ecode+1;
      do next += GET(next,1); while (*next == OP_ALT);
      ecode = next + 1 + LINK_SIZE;
      }
    break;

    /* End of a group, repeated or non-repeating. */

    case OP_KET:
    case OP_KETRMIN:
    case OP_KETRMAX:
    prev = ecode - GET(ecode, 1);

    /* If this was a group that remembered the subject start, in order to break
    infinite repeats of empty string matches, retrieve the subject start from
    the chain. Otherwise, set it NULL. */

    if (*prev >= OP_SBRA)
      {
      saved_eptr = eptrb->epb_saved_eptr;   /* Value at start of group */
      eptrb = eptrb->epb_prev;              /* Backup to previous group */
      }
    else saved_eptr = NULL;

    /* If we are at the end of an assertion group, stop matching and return
    MATCH_MATCH, but record the current high water mark for use by positive
    assertions. Do this also for the "once" (atomic) groups. */

    if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT ||
        *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT ||
        *prev == OP_ONCE)
      {
      md->end_match_ptr = eptr;      /* For ONCE */
      md->end_offset_top = offset_top;
      RRETURN(MATCH_MATCH);
      }

    /* For capturing groups we have to check the group number back at the start
    and if necessary complete handling an extraction by setting the offsets and
    bumping the high water mark. Note that whole-pattern recursion is coded as
    a recurse into group 0, so it won't be picked up here. Instead, we catch it
    when the OP_END is reached. Other recursion is handled here. */

    if (*prev == OP_CBRA || *prev == OP_SCBRA)
      {
      number = GET2(prev, 1+LINK_SIZE);
      offset = number << 1;

#ifdef DEBUG
      printf("end bracket %d", number);
      printf("\n");
#endif

      md->capture_last = number;
      if (offset >= md->offset_max) md->offset_overflow = TRUE; else
        {
        md->offset_vector[offset] =
          md->offset_vector[md->offset_end - number];
        md->offset_vector[offset+1] = eptr - md->start_subject;
        if (offset_top <= offset) offset_top = offset + 2;
        }

      /* Handle a recursively called group. Restore the offsets
      appropriately and continue from after the call. */

      if (md->recursive != NULL && md->recursive->group_num == number)
        {
        recursion_info *rec = md->recursive;
        DPRINTF(("Recursion (%d) succeeded - continuing\n", number));
        md->recursive = rec->prevrec;
        mstart = rec->save_start;
        memcpy(md->offset_vector, rec->offset_save,
          rec->saved_max * sizeof(int));
        offset_top = rec->save_offset_top;
        ecode = rec->after_call;
        ims = original_ims;
        break;
        }
      }

    /* For both capturing and non-capturing groups, reset the value of the ims
    flags, in case they got changed during the group. */

    ims = original_ims;
    DPRINTF(("ims reset to %02lx\n", ims));

    /* For a non-repeating ket, just continue at this level. This also
    happens for a repeating ket if no characters were matched in the group.
    This is the forcible breaking of infinite loops as implemented in Perl
    5.005. If there is an options reset, it will get obeyed in the normal
    course of events. */

    if (*ecode == OP_KET || eptr == saved_eptr)
      {
      ecode += 1 + LINK_SIZE;
      break;
      }

    /* The repeating kets try the rest of the pattern or restart from the
    preceding bracket, in the appropriate order. In the second case, we can use
    tail recursion to avoid using another stack frame, unless we have an
    unlimited repeat of a group that can match an empty string. */

    flags = (*prev >= OP_SBRA)? match_cbegroup : 0;

    if (*ecode == OP_KETRMIN)
      {
      RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM12);
      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
      if (flags != 0)    /* Could match an empty string */
        {
        RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM50);
        RRETURN(rrc);
        }
      ecode = prev;
      goto TAIL_RECURSE;
      }
    else  /* OP_KETRMAX */
      {
      RMATCH(eptr, prev, offset_top, md, ims, eptrb, flags, RM13);
      if (rrc != MATCH_NOMATCH) RRETURN(rrc);
      ecode += 1 + LINK_SIZE;
      flags = 0;
      goto TAIL_RECURSE;
      }
    /* Control never gets here */

    /* Start of subject unless notbol, or after internal newline if multiline */

    case OP_CIRC:
    if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH);
    if ((ims & PCRE_MULTILINE) != 0)
      {
      if (eptr != md->start_subject &&
          (eptr == md->end_subject || !WAS_NEWLINE(eptr)))
        RRETURN(MATCH_NOMATCH);
      ecode++;
      break;
      }
    /* ... else fall through */

    /* Start of subject assertion */

    case OP_SOD:
    if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    /* Start of match assertion */

    case OP_SOM:
    if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    /* Reset the start of match point */

    case OP_SET_SOM:
    mstart = eptr;
    ecode++;
    break;

    /* Assert before internal newline if multiline, or before a terminating
    newline unless endonly is set, else end of subject unless noteol is set. */

    case OP_DOLL:
    if ((ims & PCRE_MULTILINE) != 0)
      {
      if (eptr < md->end_subject)
        { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); }
      else
        { if (md->noteol) RRETURN(MATCH_NOMATCH); }
      ecode++;
      break;
      }
    else
      {
      if (md->noteol) RRETURN(MATCH_NOMATCH);
      if (!md->endonly)
        {
        if (eptr != md->end_subject &&
            (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
          RRETURN(MATCH_NOMATCH);
        ecode++;
        break;
        }
      }
    /* ... else fall through for endonly */

    /* End of subject assertion (\z) */

    case OP_EOD:
    if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    /* End of subject or ending \n assertion (\Z) */

    case OP_EODN:
    if (eptr != md->end_subject &&
        (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen))
      RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    /* Word boundary assertions */

    case OP_NOT_WORD_BOUNDARY:
    case OP_WORD_BOUNDARY:
      {

      /* Find out if the previous and current characters are "word" characters.
      It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to
      be "non-word" characters. Remember the earliest consulted character for
      partial matching. */

#ifdef SUPPORT_UTF8
      if (utf8)
        {
        if (eptr == md->start_subject) prev_is_word = FALSE; else
          {
          USPTR lastptr = eptr - 1;
          while((*lastptr & 0xc0) == 0x80) lastptr--;
          if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr;
          GETCHAR(c, lastptr);
          prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
          }
        if (eptr >= md->end_subject)
          {
          SCHECK_PARTIAL();
          cur_is_word = FALSE;
          }
        else
          {
          GETCHAR(c, eptr);
          cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0;
          }
        }
      else
#endif

      /* Not in UTF-8 mode */

        {
        if (eptr == md->start_subject) prev_is_word = FALSE; else
          {
          if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1;
          prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0);
          }
        if (eptr >= md->end_subject)
          {
          SCHECK_PARTIAL();
          cur_is_word = FALSE;
          }
        else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0);
        }

      /* Now see if the situation is what we want */

      if ((*ecode++ == OP_WORD_BOUNDARY)?
           cur_is_word == prev_is_word : cur_is_word != prev_is_word)
        RRETURN(MATCH_NOMATCH);
      }
    break;

    /* Match a single character type; inline for speed */

    case OP_ANY:
    if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH);
    /* Fall through */

    case OP_ALLANY:
    if (eptr++ >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++;
    ecode++;
    break;

    /* Match a single byte, even in UTF-8 mode. This opcode really does match
    any byte, even newline, independent of the setting of PCRE_DOTALL. */

    case OP_ANYBYTE:
    if (eptr++ >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    ecode++;
    break;

    case OP_NOT_DIGIT:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    if (
#ifdef SUPPORT_UTF8
       c < 256 &&
#endif
       (md->ctypes[c] & ctype_digit) != 0
       )
      RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    case OP_DIGIT:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    if (
#ifdef SUPPORT_UTF8
       c >= 256 ||
#endif
       (md->ctypes[c] & ctype_digit) == 0
       )
      RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    case OP_NOT_WHITESPACE:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    if (
#ifdef SUPPORT_UTF8
       c < 256 &&
#endif
       (md->ctypes[c] & ctype_space) != 0
       )
      RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    case OP_WHITESPACE:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    if (
#ifdef SUPPORT_UTF8
       c >= 256 ||
#endif
       (md->ctypes[c] & ctype_space) == 0
       )
      RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    case OP_NOT_WORDCHAR:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    if (
#ifdef SUPPORT_UTF8
       c < 256 &&
#endif
       (md->ctypes[c] & ctype_word) != 0
       )
      RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    case OP_WORDCHAR:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    if (
#ifdef SUPPORT_UTF8
       c >= 256 ||
#endif
       (md->ctypes[c] & ctype_word) == 0
       )
      RRETURN(MATCH_NOMATCH);
    ecode++;
    break;

    case OP_ANYNL:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    switch(c)
      {
      default: RRETURN(MATCH_NOMATCH);
      case 0x000d:
      if (eptr < md->end_subject && *eptr == 0x0a) eptr++;
      break;

      case 0x000a:
      break;

      case 0x000b:
      case 0x000c:
      case 0x0085:
      case 0x2028:
      case 0x2029:
      if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH);
      break;
      }
    ecode++;
    break;

    case OP_NOT_HSPACE:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    switch(c)
      {
      default: break;
      case 0x09:      /* HT */
      case 0x20:      /* SPACE */
      case 0xa0:      /* NBSP */
      case 0x1680:    /* OGHAM SPACE MARK */
      case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
      case 0x2000:    /* EN QUAD */
      case 0x2001:    /* EM QUAD */
      case 0x2002:    /* EN SPACE */
      case 0x2003:    /* EM SPACE */
      case 0x2004:    /* THREE-PER-EM SPACE */
      case 0x2005:    /* FOUR-PER-EM SPACE */
      case 0x2006:    /* SIX-PER-EM SPACE */
      case 0x2007:    /* FIGURE SPACE */
      case 0x2008:    /* PUNCTUATION SPACE */
      case 0x2009:    /* THIN SPACE */
      case 0x200A:    /* HAIR SPACE */
      case 0x202f:    /* NARROW NO-BREAK SPACE */
      case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
      case 0x3000:    /* IDEOGRAPHIC SPACE */
      RRETURN(MATCH_NOMATCH);
      }
    ecode++;
    break;

    case OP_HSPACE:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    switch(c)
      {
      default: RRETURN(MATCH_NOMATCH);
      case 0x09:      /* HT */
      case 0x20:      /* SPACE */
      case 0xa0:      /* NBSP */
      case 0x1680:    /* OGHAM SPACE MARK */
      case 0x180e:    /* MONGOLIAN VOWEL SEPARATOR */
      case 0x2000:    /* EN QUAD */
      case 0x2001:    /* EM QUAD */
      case 0x2002:    /* EN SPACE */
      case 0x2003:    /* EM SPACE */
      case 0x2004:    /* THREE-PER-EM SPACE */
      case 0x2005:    /* FOUR-PER-EM SPACE */
      case 0x2006:    /* SIX-PER-EM SPACE */
      case 0x2007:    /* FIGURE SPACE */
      case 0x2008:    /* PUNCTUATION SPACE */
      case 0x2009:    /* THIN SPACE */
      case 0x200A:    /* HAIR SPACE */
      case 0x202f:    /* NARROW NO-BREAK SPACE */
      case 0x205f:    /* MEDIUM MATHEMATICAL SPACE */
      case 0x3000:    /* IDEOGRAPHIC SPACE */
      break;
      }
    ecode++;
    break;

    case OP_NOT_VSPACE:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    switch(c)
      {
      default: break;
      case 0x0a:      /* LF */
      case 0x0b:      /* VT */
      case 0x0c:      /* FF */
      case 0x0d:      /* CR */
      case 0x85:      /* NEL */
      case 0x2028:    /* LINE SEPARATOR */
      case 0x2029:    /* PARAGRAPH SEPARATOR */
      RRETURN(MATCH_NOMATCH);
      }
    ecode++;
    break;

    case OP_VSPACE:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
    switch(c)
      {
      default: RRETURN(MATCH_NOMATCH);
      case 0x0a:      /* LF */
      case 0x0b:      /* VT */
      case 0x0c:      /* FF */
      case 0x0d:      /* CR */
      case 0x85:      /* NEL */
      case 0x2028:    /* LINE SEPARATOR */
      case 0x2029:    /* PARAGRAPH SEPARATOR */
      break;
      }
    ecode++;
    break;

#ifdef SUPPORT_UCP
    /* Check the next character by Unicode property. We will get here only
    if the support is in the binary; otherwise a compile-time error occurs. */

    case OP_PROP:
    case OP_NOTPROP:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
      {
      const ucd_record *prop = GET_UCD(c);

      switch(ecode[1])
        {
        case PT_ANY:
        if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH);
        break;

        case PT_LAMP:
        if ((prop->chartype == ucp_Lu ||
             prop->chartype == ucp_Ll ||
             prop->chartype == ucp_Lt) == (op == OP_NOTPROP))
          RRETURN(MATCH_NOMATCH);
         break;

        case PT_GC:
        if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP))
          RRETURN(MATCH_NOMATCH);
        break;

        case PT_PC:
        if ((ecode[2] != prop->chartype) == (op == OP_PROP))
          RRETURN(MATCH_NOMATCH);
        break;

        case PT_SC:
        if ((ecode[2] != prop->script) == (op == OP_PROP))
          RRETURN(MATCH_NOMATCH);
        break;

        default:
        RRETURN(PCRE_ERROR_INTERNAL);
        }

      ecode += 3;
      }
    break;

    /* Match an extended Unicode sequence. We will get here only if the support
    is in the binary; otherwise a compile-time error occurs. */

    case OP_EXTUNI:
    if (eptr >= md->end_subject)
      {
      SCHECK_PARTIAL();
      RRETURN(MATCH_NOMATCH);
      }
    GETCHARINCTEST(c, eptr);
      {
      int category = UCD_CATEGORY(c);
      if (category == ucp_M) RRETURN(MATCH_NOMATCH);
      while (eptr < md->end_subject)
        {
        int len = 1;
        if (!utf8) c = *eptr; else
          {
          GETCHARLEN(c, eptr, len);
          }
        category = UCD_CATEGORY(c);
        if (category != ucp_M) break;
        eptr += len;
        }
      }
    ecode++;
    break;
#endif


    /* Match a back reference, possibly repeatedly. Look past the end of the
    item to see if there is repeat information following. The code is similar
    to that for character classes, but repeated for efficiency. Then obey
    similar code to character type repeats - written out again for speed.
    However, if the referenced string is the empty string, always treat
    it as matched, any number of times (otherwise there could be infinite
    loops). */

    case OP_REF:
      {
      offset = GET2(ecode, 1) << 1;               /* Doubled ref number */
      ecode += 3;

      /* If the reference is unset, there are two possibilities:

      (a) In the default, Perl-compatible state, set the length to be longer
      than the amount of subject left; this ensures that every attempt at a
      match fails. We can't just fail here, because of the possibility of
      quantifiers with zero minima.

      (b) If the JavaScript compatibility flag is set, set the length to zero
      so that the back reference matches an empty string.

      Otherwise, set the length to the length of what was matched by the
      referenced subpattern. */

      if (offset >= offset_top || md->offset_vector[offset] < 0)
        length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1;
      else
        length = md->offset_vector[offset+1] - md->offset_vector[offset];

      /* Set up for repetition, or handle the non-repeated case */

      switch (*ecode)
        {
        case OP_CRSTAR:
        case OP_CRMINSTAR:
        case OP_CRPLUS:
        case OP_CRMINPLUS:
        case OP_CRQUERY:
        case OP_CRMINQUERY:
        c
Back to Top