summaryrefslogtreecommitdiff
path: root/src/regexp.c
blob: 04e068d3d581be515464783f63def84652f23f70 (plain)
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
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
5910
5911
5912
5913
5914
5915
5916
5917
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
5939
5940
5941
5942
5943
5944
5945
5946
5947
5948
5949
5950
5951
5952
5953
5954
5955
5956
5957
5958
5959
5960
5961
5962
5963
5964
5965
5966
5967
5968
5969
5970
5971
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
6001
6002
6003
6004
6005
6006
6007
6008
6009
6010
6011
6012
6013
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
6037
6038
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
6063
6064
6065
6066
6067
6068
6069
6070
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
6086
6087
6088
6089
6090
6091
6092
6093
6094
6095
6096
6097
6098
6099
6100
6101
6102
6103
6104
6105
6106
6107
6108
6109
6110
6111
6112
6113
6114
6115
6116
6117
6118
6119
6120
6121
6122
6123
6124
6125
6126
6127
6128
6129
6130
6131
6132
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147
6148
6149
6150
6151
6152
6153
6154
6155
6156
6157
6158
6159
6160
6161
6162
6163
6164
6165
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
6181
6182
6183
6184
6185
6186
6187
6188
6189
6190
6191
6192
6193
6194
6195
6196
6197
6198
6199
6200
6201
6202
6203
6204
6205
6206
6207
6208
6209
6210
6211
6212
6213
6214
6215
6216
6217
6218
6219
6220
6221
6222
6223
6224
6225
6226
6227
6228
6229
6230
6231
6232
6233
6234
6235
6236
6237
6238
6239
6240
6241
6242
6243
6244
6245
6246
6247
6248
6249
6250
6251
6252
6253
6254
6255
6256
6257
6258
6259
6260
6261
6262
6263
6264
6265
6266
6267
6268
6269
6270
6271
6272
6273
6274
6275
6276
6277
6278
6279
6280
6281
6282
6283
6284
6285
6286
6287
6288
6289
6290
6291
6292
6293
6294
6295
6296
6297
6298
6299
6300
6301
6302
6303
6304
6305
6306
6307
6308
6309
6310
6311
6312
6313
6314
6315
6316
6317
6318
6319
6320
6321
6322
6323
6324
6325
6326
6327
6328
6329
6330
6331
6332
6333
6334
6335
6336
6337
6338
6339
6340
6341
6342
6343
6344
6345
6346
6347
6348
6349
6350
6351
6352
6353
6354
6355
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
6371
6372
6373
6374
6375
6376
6377
6378
6379
6380
6381
6382
6383
6384
6385
6386
6387
6388
6389
6390
6391
6392
6393
6394
6395
6396
6397
6398
6399
6400
6401
6402
6403
6404
6405
6406
6407
6408
6409
6410
6411
6412
6413
6414
6415
6416
6417
6418
6419
6420
6421
6422
6423
6424
6425
6426
6427
6428
6429
6430
6431
6432
6433
6434
6435
6436
6437
6438
6439
6440
6441
6442
6443
6444
6445
6446
6447
6448
6449
6450
6451
6452
6453
6454
6455
6456
6457
6458
6459
6460
6461
6462
6463
6464
6465
6466
6467
6468
6469
6470
6471
6472
6473
6474
6475
6476
6477
6478
6479
6480
6481
6482
6483
6484
6485
6486
6487
6488
6489
6490
6491
6492
6493
6494
6495
6496
6497
6498
6499
6500
6501
6502
6503
6504
6505
6506
6507
6508
6509
6510
6511
6512
6513
6514
6515
6516
6517
6518
6519
6520
6521
6522
6523
6524
6525
6526
6527
6528
6529
6530
6531
6532
6533
6534
6535
6536
6537
6538
6539
6540
6541
6542
6543
6544
6545
6546
6547
6548
6549
6550
6551
6552
6553
6554
6555
6556
6557
6558
6559
6560
6561
6562
6563
6564
6565
6566
6567
6568
6569
6570
6571
6572
6573
6574
6575
6576
6577
6578
6579
6580
6581
6582
6583
6584
6585
6586
6587
6588
6589
6590
6591
6592
6593
6594
6595
6596
6597
6598
6599
6600
6601
6602
6603
6604
6605
6606
6607
6608
6609
6610
6611
6612
6613
6614
6615
6616
6617
6618
6619
6620
6621
6622
6623
6624
6625
6626
6627
6628
6629
6630
6631
6632
6633
6634
6635
6636
6637
6638
6639
6640
6641
6642
6643
6644
6645
6646
6647
6648
6649
6650
6651
6652
6653
6654
6655
6656
6657
6658
6659
6660
6661
6662
6663
6664
6665
6666
6667
6668
6669
6670
6671
6672
6673
6674
6675
6676
6677
6678
6679
6680
6681
6682
6683
6684
6685
6686
6687
6688
6689
6690
6691
6692
6693
6694
6695
6696
6697
6698
6699
6700
6701
6702
6703
6704
6705
6706
6707
6708
6709
6710
6711
6712
6713
6714
6715
6716
6717
6718
6719
6720
6721
6722
6723
6724
6725
6726
6727
6728
6729
6730
6731
6732
6733
6734
6735
6736
6737
6738
6739
6740
6741
6742
6743
6744
6745
6746
6747
6748
6749
6750
6751
6752
6753
6754
6755
6756
6757
6758
6759
6760
6761
6762
6763
6764
6765
6766
6767
6768
6769
6770
6771
6772
6773
6774
6775
6776
6777
6778
6779
6780
6781
6782
6783
6784
6785
6786
6787
6788
6789
6790
6791
6792
6793
6794
6795
6796
6797
6798
6799
6800
6801
6802
6803
6804
6805
6806
6807
6808
6809
6810
6811
6812
6813
6814
6815
6816
6817
6818
6819
6820
6821
6822
6823
6824
6825
6826
6827
6828
6829
6830
6831
6832
6833
6834
6835
6836
6837
6838
6839
6840
6841
6842
6843
6844
6845
6846
6847
6848
6849
6850
6851
6852
6853
6854
6855
6856
6857
6858
6859
6860
6861
6862
6863
6864
6865
6866
6867
6868
6869
6870
6871
6872
6873
6874
6875
6876
6877
6878
6879
6880
6881
6882
6883
6884
6885
6886
6887
6888
6889
6890
6891
6892
6893
6894
6895
6896
6897
6898
6899
6900
6901
6902
6903
6904
6905
6906
6907
6908
6909
6910
6911
6912
6913
6914
6915
6916
6917
6918
6919
6920
6921
6922
6923
6924
6925
6926
6927
6928
6929
6930
6931
6932
6933
6934
6935
6936
6937
6938
6939
6940
6941
6942
6943
6944
6945
6946
6947
6948
6949
6950
6951
6952
6953
6954
6955
6956
6957
6958
6959
6960
6961
6962
6963
6964
6965
6966
6967
6968
6969
6970
6971
6972
6973
6974
6975
6976
6977
6978
6979
6980
6981
6982
6983
6984
6985
6986
6987
6988
6989
6990
6991
6992
6993
6994
6995
6996
6997
6998
6999
7000
7001
7002
7003
7004
7005
7006
7007
7008
7009
7010
7011
7012
7013
7014
7015
7016
7017
7018
7019
7020
7021
7022
7023
7024
7025
7026
7027
7028
7029
7030
7031
7032
7033
7034
7035
7036
7037
7038
7039
7040
7041
7042
7043
7044
7045
7046
7047
7048
7049
7050
7051
7052
7053
7054
7055
7056
7057
7058
7059
7060
7061
7062
7063
7064
7065
7066
7067
7068
7069
7070
7071
7072
7073
7074
7075
7076
7077
7078
7079
7080
7081
7082
7083
7084
7085
7086
7087
7088
7089
7090
7091
7092
7093
7094
7095
7096
7097
7098
7099
7100
7101
7102
7103
7104
7105
7106
7107
7108
7109
7110
7111
7112
7113
7114
7115
7116
7117
7118
7119
7120
7121
7122
7123
7124
7125
7126
7127
7128
7129
7130
7131
7132
7133
7134
7135
7136
7137
7138
7139
7140
7141
7142
7143
7144
7145
7146
7147
7148
7149
7150
7151
7152
7153
7154
7155
7156
7157
7158
7159
7160
7161
7162
7163
7164
7165
7166
7167
7168
7169
7170
7171
7172
7173
7174
7175
7176
7177
7178
7179
7180
7181
7182
7183
7184
7185
7186
7187
7188
7189
7190
7191
7192
7193
7194
7195
7196
7197
7198
7199
7200
7201
7202
7203
/* vi:set ts=8 sts=4 sw=4:
 *
 * Handling of regular expressions: vim_regcomp(), vim_regexec(), vim_regsub()
 *
 * NOTICE:
 *
 * This is NOT the original regular expression code as written by Henry
 * Spencer.  This code has been modified specifically for use with the VIM
 * editor, and should not be used separately from Vim.  If you want a good
 * regular expression library, get the original code.  The copyright notice
 * that follows is from the original.
 *
 * END NOTICE
 *
 *	Copyright (c) 1986 by University of Toronto.
 *	Written by Henry Spencer.  Not derived from licensed software.
 *
 *	Permission is granted to anyone to use this software for any
 *	purpose on any computer system, and to redistribute it freely,
 *	subject to the following restrictions:
 *
 *	1. The author is not responsible for the consequences of use of
 *		this software, no matter how awful, even if they arise
 *		from defects in it.
 *
 *	2. The origin of this software must not be misrepresented, either
 *		by explicit claim or by omission.
 *
 *	3. Altered versions must be plainly marked as such, and must not
 *		be misrepresented as being the original software.
 *
 * Beware that some of this code is subtly aware of the way operator
 * precedence is structured in regular expressions.  Serious changes in
 * regular-expression syntax might require a total rethink.
 *
 * Changes have been made by Tony Andrews, Olaf 'Rhialto' Seibert, Robert
 * Webb, Ciaran McCreesh and Bram Moolenaar.
 * Named character class support added by Walter Briscoe (1998 Jul 01)
 */

#include "vim.h"

#undef DEBUG

/*
 * The "internal use only" fields in regexp.h are present to pass info from
 * compile to execute that permits the execute phase to run lots faster on
 * simple cases.  They are:
 *
 * regstart	char that must begin a match; NUL if none obvious; Can be a
 *		multi-byte character.
 * reganch	is the match anchored (at beginning-of-line only)?
 * regmust	string (pointer into program) that match must include, or NULL
 * regmlen	length of regmust string
 * regflags	RF_ values or'ed together
 *
 * Regstart and reganch permit very fast decisions on suitable starting points
 * for a match, cutting down the work a lot.  Regmust permits fast rejection
 * of lines that cannot possibly match.  The regmust tests are costly enough
 * that vim_regcomp() supplies a regmust only if the r.e. contains something
 * potentially expensive (at present, the only such thing detected is * or +
 * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 * supplied because the test in vim_regexec() needs it and vim_regcomp() is
 * computing it anyway.
 */

/*
 * Structure for regexp "program".  This is essentially a linear encoding
 * of a nondeterministic finite-state machine (aka syntax charts or
 * "railroad normal form" in parsing technology).  Each node is an opcode
 * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 * all nodes except BRANCH and BRACES_COMPLEX implement concatenation; a "next"
 * pointer with a BRANCH on both ends of it is connecting two alternatives.
 * (Here we have one of the subtle syntax dependencies:	an individual BRANCH
 * (as opposed to a collection of them) is never concatenated with anything
 * because of operator precedence).  The "next" pointer of a BRACES_COMPLEX
 * node points to the node after the stuff to be repeated.
 * The operand of some types of node is a literal string; for others, it is a
 * node leading into a sub-FSM.  In particular, the operand of a BRANCH node
 * is the first node of the branch.
 * (NB this is *not* a tree structure: the tail of the branch connects to the
 * thing following the set of BRANCHes.)
 *
 * pattern	is coded like:
 *
 *			  +-----------------+
 *			  |		    V
 * <aa>\|<bb>	BRANCH <aa> BRANCH <bb> --> END
 *		     |	    ^	 |	    ^
 *		     +------+	 +----------+
 *
 *
 *		       +------------------+
 *		       V		  |
 * <aa>*	BRANCH BRANCH <aa> --> BACK BRANCH --> NOTHING --> END
 *		     |	    |		    ^			   ^
 *		     |	    +---------------+			   |
 *		     +---------------------------------------------+
 *
 *
 *		       +----------------------+
 *		       V		      |
 * <aa>\+	BRANCH <aa> --> BRANCH --> BACK  BRANCH --> NOTHING --> END
 *		     |		     |		 ^			^
 *		     |		     +-----------+			|
 *		     +--------------------------------------------------+
 *
 *
 *					+-------------------------+
 *					V			  |
 * <aa>\{}	BRANCH BRACE_LIMITS --> BRACE_COMPLEX <aa> --> BACK  END
 *		     |				    |		     ^
 *		     |				    +----------------+
 *		     +-----------------------------------------------+
 *
 *
 * <aa>\@!<bb>	BRANCH NOMATCH <aa> --> END  <bb> --> END
 *		     |	     |		      ^       ^
 *		     |	     +----------------+       |
 *		     +--------------------------------+
 *
 *						      +---------+
 *						      |		V
 * \z[abc]	BRANCH BRANCH  a  BRANCH  b  BRANCH  c	BRANCH	NOTHING --> END
 *		     |	    |	       |	  |	^		    ^
 *		     |	    |	       |	  +-----+		    |
 *		     |	    |	       +----------------+		    |
 *		     |	    +---------------------------+		    |
 *		     +------------------------------------------------------+
 *
 * They all start with a BRANCH for "\|" alternatives, even when there is only
 * one alternative.
 */

/*
 * The opcodes are:
 */

/* definition	number		   opnd?    meaning */
#define END		0	/*	End of program or NOMATCH operand. */
#define BOL		1	/*	Match "" at beginning of line. */
#define EOL		2	/*	Match "" at end of line. */
#define BRANCH		3	/* node Match this alternative, or the
				 *	next... */
#define BACK		4	/*	Match "", "next" ptr points backward. */
#define EXACTLY		5	/* str	Match this string. */
#define NOTHING		6	/*	Match empty string. */
#define STAR		7	/* node Match this (simple) thing 0 or more
				 *	times. */
#define PLUS		8	/* node Match this (simple) thing 1 or more
				 *	times. */
#define MATCH		9	/* node match the operand zero-width */
#define NOMATCH		10	/* node check for no match with operand */
#define BEHIND		11	/* node look behind for a match with operand */
#define NOBEHIND	12	/* node look behind for no match with operand */
#define SUBPAT		13	/* node match the operand here */
#define BRACE_SIMPLE	14	/* node Match this (simple) thing between m and
				 *	n times (\{m,n\}). */
#define BOW		15	/*	Match "" after [^a-zA-Z0-9_] */
#define EOW		16	/*	Match "" at    [^a-zA-Z0-9_] */
#define BRACE_LIMITS	17	/* nr nr  define the min & max for BRACE_SIMPLE
				 *	and BRACE_COMPLEX. */
#define NEWL		18	/*	Match line-break */
#define BHPOS		19	/*	End position for BEHIND or NOBEHIND */


/* character classes: 20-48 normal, 50-78 include a line-break */
#define ADD_NL		30
#define FIRST_NL	ANY + ADD_NL
#define ANY		20	/*	Match any one character. */
#define ANYOF		21	/* str	Match any character in this string. */
#define ANYBUT		22	/* str	Match any character not in this
				 *	string. */
#define IDENT		23	/*	Match identifier char */
#define SIDENT		24	/*	Match identifier char but no digit */
#define KWORD		25	/*	Match keyword char */
#define SKWORD		26	/*	Match word char but no digit */
#define FNAME		27	/*	Match file name char */
#define SFNAME		28	/*	Match file name char but no digit */
#define PRINT		29	/*	Match printable char */
#define SPRINT		30	/*	Match printable char but no digit */
#define WHITE		31	/*	Match whitespace char */
#define NWHITE		32	/*	Match non-whitespace char */
#define DIGIT		33	/*	Match digit char */
#define NDIGIT		34	/*	Match non-digit char */
#define HEX		35	/*	Match hex char */
#define NHEX		36	/*	Match non-hex char */
#define OCTAL		37	/*	Match octal char */
#define NOCTAL		38	/*	Match non-octal char */
#define WORD		39	/*	Match word char */
#define NWORD		40	/*	Match non-word char */
#define HEAD		41	/*	Match head char */
#define NHEAD		42	/*	Match non-head char */
#define ALPHA		43	/*	Match alpha char */
#define NALPHA		44	/*	Match non-alpha char */
#define LOWER		45	/*	Match lowercase char */
#define NLOWER		46	/*	Match non-lowercase char */
#define UPPER		47	/*	Match uppercase char */
#define NUPPER		48	/*	Match non-uppercase char */
#define LAST_NL		NUPPER + ADD_NL
#define WITH_NL(op)	((op) >= FIRST_NL && (op) <= LAST_NL)

#define MOPEN		80  /* -89	 Mark this point in input as start of
				 *	 \( subexpr.  MOPEN + 0 marks start of
				 *	 match. */
#define MCLOSE		90  /* -99	 Analogous to MOPEN.  MCLOSE + 0 marks
				 *	 end of match. */
#define BACKREF		100 /* -109 node Match same string again \1-\9 */

#ifdef FEAT_SYN_HL
# define ZOPEN		110 /* -119	 Mark this point in input as start of
				 *	 \z( subexpr. */
# define ZCLOSE		120 /* -129	 Analogous to ZOPEN. */
# define ZREF		130 /* -139 node Match external submatch \z1-\z9 */
#endif

#define BRACE_COMPLEX	140 /* -149 node Match nodes between m & n times */

#define NOPEN		150	/*	Mark this point in input as start of
					\%( subexpr. */
#define NCLOSE		151	/*	Analogous to NOPEN. */

#define MULTIBYTECODE	200	/* mbc	Match one multi-byte character */
#define RE_BOF		201	/*	Match "" at beginning of file. */
#define RE_EOF		202	/*	Match "" at end of file. */
#define CURSOR		203	/*	Match location of cursor. */

#define RE_LNUM		204	/* nr cmp  Match line number */
#define RE_COL		205	/* nr cmp  Match column number */
#define RE_VCOL		206	/* nr cmp  Match virtual column number */

#define RE_MARK		207	/* mark cmp  Match mark position */
#define RE_VISUAL	208	/*	Match Visual area */

/*
 * Magic characters have a special meaning, they don't match literally.
 * Magic characters are negative.  This separates them from literal characters
 * (possibly multi-byte).  Only ASCII characters can be Magic.
 */
#define Magic(x)	((int)(x) - 256)
#define un_Magic(x)	((x) + 256)
#define is_Magic(x)	((x) < 0)

static int no_Magic __ARGS((int x));
static int toggle_Magic __ARGS((int x));

    static int
no_Magic(x)
    int		x;
{
    if (is_Magic(x))
	return un_Magic(x);
    return x;
}

    static int
toggle_Magic(x)
    int		x;
{
    if (is_Magic(x))
	return un_Magic(x);
    return Magic(x);
}

/*
 * The first byte of the regexp internal "program" is actually this magic
 * number; the start node begins in the second byte.  It's used to catch the
 * most severe mutilation of the program by the caller.
 */

#define REGMAGIC	0234

/*
 * Opcode notes:
 *
 * BRANCH	The set of branches constituting a single choice are hooked
 *		together with their "next" pointers, since precedence prevents
 *		anything being concatenated to any individual branch.  The
 *		"next" pointer of the last BRANCH in a choice points to the
 *		thing following the whole choice.  This is also where the
 *		final "next" pointer of each individual branch points; each
 *		branch starts with the operand node of a BRANCH node.
 *
 * BACK		Normal "next" pointers all implicitly point forward; BACK
 *		exists to make loop structures possible.
 *
 * STAR,PLUS	'=', and complex '*' and '+', are implemented as circular
 *		BRANCH structures using BACK.  Simple cases (one character
 *		per match) are implemented with STAR and PLUS for speed
 *		and to minimize recursive plunges.
 *
 * BRACE_LIMITS	This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX
 *		node, and defines the min and max limits to be used for that
 *		node.
 *
 * MOPEN,MCLOSE	...are numbered at compile time.
 * ZOPEN,ZCLOSE	...ditto
 */

/*
 * A node is one char of opcode followed by two chars of "next" pointer.
 * "Next" pointers are stored as two 8-bit bytes, high order first.  The
 * value is a positive offset from the opcode of the node containing it.
 * An operand, if any, simply follows the node.  (Note that much of the
 * code generation knows about this implicit relationship.)
 *
 * Using two bytes for the "next" pointer is vast overkill for most things,
 * but allows patterns to get big without disasters.
 */
#define OP(p)		((int)*(p))
#define NEXT(p)		(((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377))
#define OPERAND(p)	((p) + 3)
/* Obtain an operand that was stored as four bytes, MSB first. */
#define OPERAND_MIN(p)	(((long)(p)[3] << 24) + ((long)(p)[4] << 16) \
			+ ((long)(p)[5] << 8) + (long)(p)[6])
/* Obtain a second operand stored as four bytes. */
#define OPERAND_MAX(p)	OPERAND_MIN((p) + 4)
/* Obtain a second single-byte operand stored after a four bytes operand. */
#define OPERAND_CMP(p)	(p)[7]

/*
 * Utility definitions.
 */
#define UCHARAT(p)	((int)*(char_u *)(p))

/* Used for an error (down from) vim_regcomp(): give the error message, set
 * rc_did_emsg and return NULL */
#define EMSG_RET_NULL(m) return (EMSG(m), rc_did_emsg = TRUE, (void *)NULL)
#define EMSG_M_RET_NULL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, (void *)NULL)
#define EMSG_RET_FAIL(m) return (EMSG(m), rc_did_emsg = TRUE, FAIL)
#define EMSG_ONE_RET_NULL EMSG_M_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL)

#define MAX_LIMIT	(32767L << 16L)

static int re_multi_type __ARGS((int));
static int cstrncmp __ARGS((char_u *s1, char_u *s2, int *n));
static char_u *cstrchr __ARGS((char_u *, int));

#ifdef DEBUG
static void	regdump __ARGS((char_u *, regprog_T *));
static char_u	*regprop __ARGS((char_u *));
#endif

#define NOT_MULTI	0
#define MULTI_ONE	1
#define MULTI_MULT	2
/*
 * Return NOT_MULTI if c is not a "multi" operator.
 * Return MULTI_ONE if c is a single "multi" operator.
 * Return MULTI_MULT if c is a multi "multi" operator.
 */
    static int
re_multi_type(c)
    int c;
{
    if (c == Magic('@') || c == Magic('=') || c == Magic('?'))
	return MULTI_ONE;
    if (c == Magic('*') || c == Magic('+') || c == Magic('{'))
	return MULTI_MULT;
    return NOT_MULTI;
}

/*
 * Flags to be passed up and down.
 */
#define HASWIDTH	0x1	/* Known never to match null string. */
#define SIMPLE		0x2	/* Simple enough to be STAR/PLUS operand. */
#define SPSTART		0x4	/* Starts with * or +. */
#define HASNL		0x8	/* Contains some \n. */
#define HASLOOKBH	0x10	/* Contains "\@<=" or "\@<!". */
#define WORST		0	/* Worst case. */

/*
 * When regcode is set to this value, code is not emitted and size is computed
 * instead.
 */
#define JUST_CALC_SIZE	((char_u *) -1)

static char_u		*reg_prev_sub = NULL;

/*
 * REGEXP_INRANGE contains all characters which are always special in a []
 * range after '\'.
 * REGEXP_ABBR contains all characters which act as abbreviations after '\'.
 * These are:
 *  \n	- New line (NL).
 *  \r	- Carriage Return (CR).
 *  \t	- Tab (TAB).
 *  \e	- Escape (ESC).
 *  \b	- Backspace (Ctrl_H).
 *  \d  - Character code in decimal, eg \d123
 *  \o	- Character code in octal, eg \o80
 *  \x	- Character code in hex, eg \x4a
 *  \u	- Multibyte character code, eg \u20ac
 *  \U	- Long multibyte character code, eg \U12345678
 */
static char_u REGEXP_INRANGE[] = "]^-n\\";
static char_u REGEXP_ABBR[] = "nrtebdoxuU";

static int	backslash_trans __ARGS((int c));
static int	get_char_class __ARGS((char_u **pp));
static int	get_equi_class __ARGS((char_u **pp));
static void	reg_equi_class __ARGS((int c));
static int	get_coll_element __ARGS((char_u **pp));
static char_u	*skip_anyof __ARGS((char_u *p));
static void	init_class_tab __ARGS((void));

/*
 * Translate '\x' to its control character, except "\n", which is Magic.
 */
    static int
backslash_trans(c)
    int		c;
{
    switch (c)
    {
	case 'r':   return CAR;
	case 't':   return TAB;
	case 'e':   return ESC;
	case 'b':   return BS;
    }
    return c;
}

/*
 * Check for a character class name "[:name:]".  "pp" points to the '['.
 * Returns one of the CLASS_ items. CLASS_NONE means that no item was
 * recognized.  Otherwise "pp" is advanced to after the item.
 */
    static int
get_char_class(pp)
    char_u	**pp;
{
    static const char *(class_names[]) =
    {
	"alnum:]",
#define CLASS_ALNUM 0
	"alpha:]",
#define CLASS_ALPHA 1
	"blank:]",
#define CLASS_BLANK 2
	"cntrl:]",
#define CLASS_CNTRL 3
	"digit:]",
#define CLASS_DIGIT 4
	"graph:]",
#define CLASS_GRAPH 5
	"lower:]",
#define CLASS_LOWER 6
	"print:]",
#define CLASS_PRINT 7
	"punct:]",
#define CLASS_PUNCT 8
	"space:]",
#define CLASS_SPACE 9
	"upper:]",
#define CLASS_UPPER 10
	"xdigit:]",
#define CLASS_XDIGIT 11
	"tab:]",
#define CLASS_TAB 12
	"return:]",
#define CLASS_RETURN 13
	"backspace:]",
#define CLASS_BACKSPACE 14
	"escape:]",
#define CLASS_ESCAPE 15
    };
#define CLASS_NONE 99
    int i;

    if ((*pp)[1] == ':')
    {
	for (i = 0; i < sizeof(class_names) / sizeof(*class_names); ++i)
	    if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0)
	    {
		*pp += STRLEN(class_names[i]) + 2;
		return i;
	    }
    }
    return CLASS_NONE;
}

/*
 * Specific version of character class functions.
 * Using a table to keep this fast.
 */
static short	class_tab[256];

#define	    RI_DIGIT	0x01
#define	    RI_HEX	0x02
#define	    RI_OCTAL	0x04
#define	    RI_WORD	0x08
#define	    RI_HEAD	0x10
#define	    RI_ALPHA	0x20
#define	    RI_LOWER	0x40
#define	    RI_UPPER	0x80
#define	    RI_WHITE	0x100

    static void
init_class_tab()
{
    int		i;
    static int	done = FALSE;

    if (done)
	return;

    for (i = 0; i < 256; ++i)
    {
	if (i >= '0' && i <= '7')
	    class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD;
	else if (i >= '8' && i <= '9')
	    class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD;
	else if (i >= 'a' && i <= 'f')
	    class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER;
#ifdef EBCDIC
	else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r')
						    || (i >= 's' && i <= 'z'))
#else
	else if (i >= 'g' && i <= 'z')
#endif
	    class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER;
	else if (i >= 'A' && i <= 'F')
	    class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER;
#ifdef EBCDIC
	else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R')
						    || (i >= 'S' && i <= 'Z'))
#else
	else if (i >= 'G' && i <= 'Z')
#endif
	    class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER;
	else if (i == '_')
	    class_tab[i] = RI_WORD + RI_HEAD;
	else
	    class_tab[i] = 0;
    }
    class_tab[' '] |= RI_WHITE;
    class_tab['\t'] |= RI_WHITE;
    done = TRUE;
}

#ifdef FEAT_MBYTE
# define ri_digit(c)	(c < 0x100 && (class_tab[c] & RI_DIGIT))
# define ri_hex(c)	(c < 0x100 && (class_tab[c] & RI_HEX))
# define ri_octal(c)	(c < 0x100 && (class_tab[c] & RI_OCTAL))
# define ri_word(c)	(c < 0x100 && (class_tab[c] & RI_WORD))
# define ri_head(c)	(c < 0x100 && (class_tab[c] & RI_HEAD))
# define ri_alpha(c)	(c < 0x100 && (class_tab[c] & RI_ALPHA))
# define ri_lower(c)	(c < 0x100 && (class_tab[c] & RI_LOWER))
# define ri_upper(c)	(c < 0x100 && (class_tab[c] & RI_UPPER))
# define ri_white(c)	(c < 0x100 && (class_tab[c] & RI_WHITE))
#else
# define ri_digit(c)	(class_tab[c] & RI_DIGIT)
# define ri_hex(c)	(class_tab[c] & RI_HEX)
# define ri_octal(c)	(class_tab[c] & RI_OCTAL)
# define ri_word(c)	(class_tab[c] & RI_WORD)
# define ri_head(c)	(class_tab[c] & RI_HEAD)
# define ri_alpha(c)	(class_tab[c] & RI_ALPHA)
# define ri_lower(c)	(class_tab[c] & RI_LOWER)
# define ri_upper(c)	(class_tab[c] & RI_UPPER)
# define ri_white(c)	(class_tab[c] & RI_WHITE)
#endif

/* flags for regflags */
#define RF_ICASE    1	/* ignore case */
#define RF_NOICASE  2	/* don't ignore case */
#define RF_HASNL    4	/* can match a NL */
#define RF_ICOMBINE 8	/* ignore combining characters */
#define RF_LOOKBH   16	/* uses "\@<=" or "\@<!" */

/*
 * Global work variables for vim_regcomp().
 */

static char_u	*regparse;	/* Input-scan pointer. */
static int	prevchr_len;	/* byte length of previous char */
static int	num_complex_braces; /* Complex \{...} count */
static int	regnpar;	/* () count. */
#ifdef FEAT_SYN_HL
static int	regnzpar;	/* \z() count. */
static int	re_has_z;	/* \z item detected */
#endif
static char_u	*regcode;	/* Code-emit pointer, or JUST_CALC_SIZE */
static long	regsize;	/* Code size. */
static char_u	had_endbrace[NSUBEXP];	/* flags, TRUE if end of () found */
static unsigned	regflags;	/* RF_ flags for prog */
static long	brace_min[10];	/* Minimums for complex brace repeats */
static long	brace_max[10];	/* Maximums for complex brace repeats */
static int	brace_count[10]; /* Current counts for complex brace repeats */
#if defined(FEAT_SYN_HL) || defined(PROTO)
static int	had_eol;	/* TRUE when EOL found by vim_regcomp() */
#endif
static int	one_exactly = FALSE;	/* only do one char for EXACTLY */

static int	reg_magic;	/* magicness of the pattern: */
#define MAGIC_NONE	1	/* "\V" very unmagic */
#define MAGIC_OFF	2	/* "\M" or 'magic' off */
#define MAGIC_ON	3	/* "\m" or 'magic' */
#define MAGIC_ALL	4	/* "\v" very magic */

static int	reg_string;	/* matching with a string instead of a buffer
				   line */
static int	reg_strict;	/* "[abc" is illegal */

/*
 * META contains all characters that may be magic, except '^' and '$'.
 */

#ifdef EBCDIC
static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~";
#else
/* META[] is used often enough to justify turning it into a table. */
static char_u META_flags[] = {
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
/*		   %  &     (  )  *  +	      .    */
    0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0,
/*     1  2  3	4  5  6  7  8  9	<  =  >  ? */
    0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1,
/*  @  A     C	D     F     H  I     K	L  M	 O */
    1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1,
/*  P	     S	   U  V  W  X	  Z  [		 _ */
    1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1,
/*     a     c	d     f     h  i     k	l  m  n  o */
    0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1,
/*  p	     s	   u  v  w  x	  z  {	|     ~    */
    1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1
};
#endif

static int	curchr;

/* arguments for reg() */
#define REG_NOPAREN	0	/* toplevel reg() */
#define REG_PAREN	1	/* \(\) */
#define REG_ZPAREN	2	/* \z(\) */
#define REG_NPAREN	3	/* \%(\) */

/*
 * Forward declarations for vim_regcomp()'s friends.
 */
static void	initchr __ARGS((char_u *));
static int	getchr __ARGS((void));
static void	skipchr_keepstart __ARGS((void));
static int	peekchr __ARGS((void));
static void	skipchr __ARGS((void));
static void	ungetchr __ARGS((void));
static int	gethexchrs __ARGS((int maxinputlen));
static int	getoctchrs __ARGS((void));
static int	getdecchrs __ARGS((void));
static int	coll_get_char __ARGS((void));
static void	regcomp_start __ARGS((char_u *expr, int flags));
static char_u	*reg __ARGS((int, int *));
static char_u	*regbranch __ARGS((int *flagp));
static char_u	*regconcat __ARGS((int *flagp));
static char_u	*regpiece __ARGS((int *));
static char_u	*regatom __ARGS((int *));
static char_u	*regnode __ARGS((int));
#ifdef FEAT_MBYTE
static int	use_multibytecode __ARGS((int c));
#endif
static int	prog_magic_wrong __ARGS((void));
static char_u	*regnext __ARGS((char_u *));
static void	regc __ARGS((int b));
#ifdef FEAT_MBYTE
static void	regmbc __ARGS((int c));
#else
# define regmbc(c) regc(c)
#endif
static void	reginsert __ARGS((int, char_u *));
static void	reginsert_limits __ARGS((int, long, long, char_u *));
static char_u	*re_put_long __ARGS((char_u *pr, long_u val));
static int	read_limits __ARGS((long *, long *));
static void	regtail __ARGS((char_u *, char_u *));
static void	regoptail __ARGS((char_u *, char_u *));

/*
 * Return TRUE if compiled regular expression "prog" can match a line break.
 */
    int
re_multiline(prog)
    regprog_T *prog;
{
    return (prog->regflags & RF_HASNL);
}

/*
 * Return TRUE if compiled regular expression "prog" looks before the start
 * position (pattern contains "\@<=" or "\@<!").
 */
    int
re_lookbehind(prog)
    regprog_T *prog;
{
    return (prog->regflags & RF_LOOKBH);
}

/*
 * Check for an equivalence class name "[=a=]".  "pp" points to the '['.
 * Returns a character representing the class. Zero means that no item was
 * recognized.  Otherwise "pp" is advanced to after the item.
 */
    static int
get_equi_class(pp)
    char_u	**pp;
{
    int		c;
    int		l = 1;
    char_u	*p = *pp;

    if (p[1] == '=')
    {
#ifdef FEAT_MBYTE
	if (has_mbyte)
	    l = (*mb_ptr2len)(p + 2);
#endif
	if (p[l + 2] == '=' && p[l + 3] == ']')
	{
#ifdef FEAT_MBYTE
	    if (has_mbyte)
		c = mb_ptr2char(p + 2);
	    else
#endif
		c = p[2];
	    *pp += l + 4;
	    return c;
	}
    }
    return 0;
}

/*
 * Produce the bytes for equivalence class "c".
 * Currently only handles latin1, latin9 and utf-8.
 */
    static void
reg_equi_class(c)
    int	    c;
{
#ifdef FEAT_MBYTE
    if (enc_utf8 || STRCMP(p_enc, "latin1") == 0
					 || STRCMP(p_enc, "iso-8859-15") == 0)
#endif
    {
	switch (c)
	{
	    case 'A': case '\300': case '\301': case '\302':
	    case '\303': case '\304': case '\305':
		      regmbc('A'); regmbc('\300'); regmbc('\301');
		      regmbc('\302'); regmbc('\303'); regmbc('\304');
		      regmbc('\305');
		      return;
	    case 'C': case '\307':
		      regmbc('C'); regmbc('\307');
		      return;
	    case 'E': case '\310': case '\311': case '\312': case '\313':
		      regmbc('E'); regmbc('\310'); regmbc('\311');
		      regmbc('\312'); regmbc('\313');
		      return;
	    case 'I': case '\314': case '\315': case '\316': case '\317':
		      regmbc('I'); regmbc('\314'); regmbc('\315');
		      regmbc('\316'); regmbc('\317');
		      return;
	    case 'N': case '\321':
		      regmbc('N'); regmbc('\321');
		      return;
	    case 'O': case '\322': case '\323': case '\324': case '\325':
	    case '\326':
		      regmbc('O'); regmbc('\322'); regmbc('\323');
		      regmbc('\324'); regmbc('\325'); regmbc('\326');
		      return;
	    case 'U': case '\331': case '\332': case '\333': case '\334':
		      regmbc('U'); regmbc('\331'); regmbc('\332');
		      regmbc('\333'); regmbc('\334');
		      return;
	    case 'Y': case '\335':
		      regmbc('Y'); regmbc('\335');
		      return;
	    case 'a': case '\340': case '\341': case '\342':
	    case '\343': case '\344': case '\345':
		      regmbc('a'); regmbc('\340'); regmbc('\341');
		      regmbc('\342'); regmbc('\343'); regmbc('\344');
		      regmbc('\345');
		      return;
	    case 'c': case '\347':
		      regmbc('c'); regmbc('\347');
		      return;
	    case 'e': case '\350': case '\351': case '\352': case '\353':
		      regmbc('e'); regmbc('\350'); regmbc('\351');
		      regmbc('\352'); regmbc('\353');
		      return;
	    case 'i': case '\354': case '\355': case '\356': case '\357':
		      regmbc('i'); regmbc('\354'); regmbc('\355');
		      regmbc('\356'); regmbc('\357');
		      return;
	    case 'n': case '\361':
		      regmbc('n'); regmbc('\361');
		      return;
	    case 'o': case '\362': case '\363': case '\364': case '\365':
	    case '\366':
		      regmbc('o'); regmbc('\362'); regmbc('\363');
		      regmbc('\364'); regmbc('\365'); regmbc('\366');
		      return;
	    case 'u': case '\371': case '\372': case '\373': case '\374':
		      regmbc('u'); regmbc('\371'); regmbc('\372');
		      regmbc('\373'); regmbc('\374');
		      return;
	    case 'y': case '\375': case '\377':
		      regmbc('y'); regmbc('\375'); regmbc('\377');
		      return;
	}
    }
    regmbc(c);
}

/*
 * Check for a collating element "[.a.]".  "pp" points to the '['.
 * Returns a character. Zero means that no item was recognized.  Otherwise
 * "pp" is advanced to after the item.
 * Currently only single characters are recognized!
 */
    static int
get_coll_element(pp)
    char_u	**pp;
{
    int		c;
    int		l = 1;
    char_u	*p = *pp;

    if (p[1] == '.')
    {
#ifdef FEAT_MBYTE
	if (has_mbyte)
	    l = (*mb_ptr2len)(p + 2);
#endif
	if (p[l + 2] == '.' && p[l + 3] == ']')
	{
#ifdef FEAT_MBYTE
	    if (has_mbyte)
		c = mb_ptr2char(p + 2);
	    else
#endif
		c = p[2];
	    *pp += l + 4;
	    return c;
	}
    }
    return 0;
}


/*
 * Skip over a "[]" range.
 * "p" must point to the character after the '['.
 * The returned pointer is on the matching ']', or the terminating NUL.
 */
    static char_u *
skip_anyof(p)
    char_u	*p;
{
    int		cpo_lit;	/* 'cpoptions' contains 'l' flag */
    int		cpo_bsl;	/* 'cpoptions' contains '\' flag */
#ifdef FEAT_MBYTE
    int		l;
#endif

    cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL;
    cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL;

    if (*p == '^')	/* Complement of range. */
	++p;
    if (*p == ']' || *p == '-')
	++p;
    while (*p != NUL && *p != ']')
    {
#ifdef FEAT_MBYTE
	if (has_mbyte && (l = (*mb_ptr2len)(p)) > 1)
	    p += l;
	else
#endif
	    if (*p == '-')
	    {
		++p;
		if (*p != ']' && *p != NUL)
		    mb_ptr_adv(p);
	    }
	else if (*p == '\\'
		&& !cpo_bsl
		&& (vim_strchr(REGEXP_INRANGE, p[1]) != NULL
		    || (!cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL)))
	    p += 2;
	else if (*p == '[')
	{
	    if (get_char_class(&p) == CLASS_NONE
		    && get_equi_class(&p) == 0
		    && get_coll_element(&p) == 0)
		++p; /* It was not a class name */
	}
	else
	    ++p;
    }

    return p;
}

/*
 * Skip past regular expression.
 * Stop at end of "startp" or where "dirc" is found ('/', '?', etc).
 * Take care of characters with a backslash in front of it.
 * Skip strings inside [ and ].
 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the
 * expression and change "\?" to "?".  If "*newp" is not NULL the expression
 * is changed in-place.
 */
    char_u *
skip_regexp(startp, dirc, magic, newp)
    char_u	*startp;
    int		dirc;
    int		magic;
    char_u	**newp;
{
    int		mymagic;
    char_u	*p = startp;

    if (magic)
	mymagic = MAGIC_ON;
    else
	mymagic = MAGIC_OFF;

    for (; p[0] != NUL; mb_ptr_adv(p))
    {
	if (p[0] == dirc)	/* found end of regexp */
	    break;
	if ((p[0] == '[' && mymagic >= MAGIC_ON)
		|| (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF))
	{
	    p = skip_anyof(p + 1);
	    if (p[0] == NUL)
		break;
	}
	else if (p[0] == '\\' && p[1] != NUL)
	{
	    if (dirc == '?' && newp != NULL && p[1] == '?')
	    {
		/* change "\?" to "?", make a copy first. */
		if (*newp == NULL)
		{
		    *newp = vim_strsave(startp);
		    if (*newp != NULL)
			p = *newp + (p - startp);
		}
		if (*newp != NULL)
		    mch_memmove(p, p + 1, STRLEN(p));
		else
		    ++p;
	    }
	    else
		++p;    /* skip next character */
	    if (*p == 'v')
		mymagic = MAGIC_ALL;
	    else if (*p == 'V')
		mymagic = MAGIC_NONE;
	}
    }
    return p;
}

/*
 * vim_regcomp() - compile a regular expression into internal code
 * Returns the program in allocated space.  Returns NULL for an error.
 *
 * We can't allocate space until we know how big the compiled form will be,
 * but we can't compile it (and thus know how big it is) until we've got a
 * place to put the code.  So we cheat:  we compile it twice, once with code
 * generation turned off and size counting turned on, and once "for real".
 * This also means that we don't allocate space until we are sure that the
 * thing really will compile successfully, and we never have to move the
 * code and thus invalidate pointers into it.  (Note that it has to be in
 * one piece because vim_free() must be able to free it all.)
 *
 * Whether upper/lower case is to be ignored is decided when executing the
 * program, it does not matter here.
 *
 * Beware that the optimization-preparation code in here knows about some
 * of the structure of the compiled regexp.
 * "re_flags": RE_MAGIC and/or RE_STRING.
 */
    regprog_T *
vim_regcomp(expr, re_flags)
    char_u	*expr;
    int		re_flags;
{
    regprog_T	*r;
    char_u	*scan;
    char_u	*longest;
    int		len;
    int		flags;

    if (expr == NULL)
	EMSG_RET_NULL(_(e_null));

    init_class_tab();

    /*
     * First pass: determine size, legality.
     */
    regcomp_start(expr, re_flags);
    regcode = JUST_CALC_SIZE;
    regc(REGMAGIC);
    if (reg(REG_NOPAREN, &flags) == NULL)
	return NULL;

    /* Small enough for pointer-storage convention? */
#ifdef SMALL_MALLOC		/* 16 bit storage allocation */
    if (regsize >= 65536L - 256L)
	EMSG_RET_NULL(_("E339: Pattern too long"));
#endif

    /* Allocate space. */
    r = (regprog_T *)lalloc(sizeof(regprog_T) + regsize, TRUE);
    if (r == NULL)
	return NULL;

    /*
     * Second pass: emit code.
     */
    regcomp_start(expr, re_flags);
    regcode = r->program;
    regc(REGMAGIC);
    if (reg(REG_NOPAREN, &flags) == NULL)
    {
	vim_free(r);
	return NULL;
    }

    /* Dig out information for optimizations. */
    r->regstart = NUL;		/* Worst-case defaults. */
    r->reganch = 0;
    r->regmust = NULL;
    r->regmlen = 0;
    r->regflags = regflags;
    if (flags & HASNL)
	r->regflags |= RF_HASNL;
    if (flags & HASLOOKBH)
	r->regflags |= RF_LOOKBH;
#ifdef FEAT_SYN_HL
    /* Remember whether this pattern has any \z specials in it. */
    r->reghasz = re_has_z;
#endif
    scan = r->program + 1;	/* First BRANCH. */
    if (OP(regnext(scan)) == END)   /* Only one top-level choice. */
    {
	scan = OPERAND(scan);

	/* Starting-point info. */
	if (OP(scan) == BOL || OP(scan) == RE_BOF)
	{
	    r->reganch++;
	    scan = regnext(scan);
	}

	if (OP(scan) == EXACTLY)
	{
#ifdef FEAT_MBYTE
	    if (has_mbyte)
		r->regstart = (*mb_ptr2char)(OPERAND(scan));
	    else
#endif
		r->regstart = *OPERAND(scan);
	}
	else if ((OP(scan) == BOW
		    || OP(scan) == EOW
		    || OP(scan) == NOTHING
		    || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN
		    || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE)
		 && OP(regnext(scan)) == EXACTLY)
	{
#ifdef FEAT_MBYTE
	    if (has_mbyte)
		r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan)));
	    else
#endif
		r->regstart = *OPERAND(regnext(scan));
	}

	/*
	 * If there's something expensive in the r.e., find the longest
	 * literal string that must appear and make it the regmust.  Resolve
	 * ties in favor of later strings, since the regstart check works
	 * with the beginning of the r.e. and avoiding duplication
	 * strengthens checking.  Not a strong reason, but sufficient in the
	 * absence of others.
	 */
	/*
	 * When the r.e. starts with BOW, it is faster to look for a regmust
	 * first. Used a lot for "#" and "*" commands. (Added by mool).
	 */
	if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW)
							  && !(flags & HASNL))
	{
	    longest = NULL;
	    len = 0;
	    for (; scan != NULL; scan = regnext(scan))
		if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len)
		{
		    longest = OPERAND(scan);
		    len = (int)STRLEN(OPERAND(scan));
		}
	    r->regmust = longest;
	    r->regmlen = len;
	}
    }
#ifdef DEBUG
    regdump(expr, r);
#endif
    return r;
}

/*
 * Setup to parse the regexp.  Used once to get the length and once to do it.
 */
    static void
regcomp_start(expr, re_flags)
    char_u	*expr;
    int		re_flags;	    /* see vim_regcomp() */
{
    initchr(expr);
    if (re_flags & RE_MAGIC)
	reg_magic = MAGIC_ON;
    else
	reg_magic = MAGIC_OFF;
    reg_string = (re_flags & RE_STRING);
    reg_strict = (re_flags & RE_STRICT);

    num_complex_braces = 0;
    regnpar = 1;
    vim_memset(had_endbrace, 0, sizeof(had_endbrace));
#ifdef FEAT_SYN_HL
    regnzpar = 1;
    re_has_z = 0;
#endif
    regsize = 0L;
    regflags = 0;
#if defined(FEAT_SYN_HL) || defined(PROTO)
    had_eol = FALSE;
#endif
}

#if defined(FEAT_SYN_HL) || defined(PROTO)
/*
 * Check if during the previous call to vim_regcomp the EOL item "$" has been
 * found.  This is messy, but it works fine.
 */
    int
vim_regcomp_had_eol()
{
    return had_eol;
}
#endif

/*
 * reg - regular expression, i.e. main body or parenthesized thing
 *
 * Caller must absorb opening parenthesis.
 *
 * Combining parenthesis handling with the base level of regular expression
 * is a trifle forced, but the need to tie the tails of the branches to what
 * follows makes it hard to avoid.
 */
    static char_u *
reg(paren, flagp)
    int		paren;	/* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */
    int		*flagp;
{
    char_u	*ret;
    char_u	*br;
    char_u	*ender;
    int		parno = 0;
    int		flags;

    *flagp = HASWIDTH;		/* Tentatively. */

#ifdef FEAT_SYN_HL
    if (paren == REG_ZPAREN)
    {
	/* Make a ZOPEN node. */
	if (regnzpar >= NSUBEXP)
	    EMSG_RET_NULL(_("E50: Too many \\z("));
	parno = regnzpar;
	regnzpar++;
	ret = regnode(ZOPEN + parno);
    }
    else
#endif
	if (paren == REG_PAREN)
    {
	/* Make a MOPEN node. */
	if (regnpar >= NSUBEXP)
	    EMSG_M_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL);
	parno = regnpar;
	++regnpar;
	ret = regnode(MOPEN + parno);
    }
    else if (paren == REG_NPAREN)
    {
	/* Make a NOPEN node. */
	ret = regnode(NOPEN);
    }
    else
	ret = NULL;

    /* Pick up the branches, linking them together. */
    br = regbranch(&flags);
    if (br == NULL)
	return NULL;
    if (ret != NULL)
	regtail(ret, br);	/* [MZ]OPEN -> first. */
    else
	ret = br;
    /* If one of the branches can be zero-width, the whole thing can.
     * If one of the branches has * at start or matches a line-break, the
     * whole thing can. */
    if (!(flags & HASWIDTH))
	*flagp &= ~HASWIDTH;
    *flagp |= flags & (SPSTART | HASNL | HASLOOKBH);
    while (peekchr() == Magic('|'))
    {
	skipchr();
	br = regbranch(&flags);
	if (br == NULL)
	    return NULL;
	regtail(ret, br);	/* BRANCH -> BRANCH. */
	if (!(flags & HASWIDTH))
	    *flagp &= ~HASWIDTH;
	*flagp |= flags & (SPSTART | HASNL | HASLOOKBH);
    }

    /* Make a closing node, and hook it on the end. */
    ender = regnode(
#ifdef FEAT_SYN_HL
	    paren == REG_ZPAREN ? ZCLOSE + parno :
#endif
	    paren == REG_PAREN ? MCLOSE + parno :
	    paren == REG_NPAREN ? NCLOSE : END);
    regtail(ret, ender);

    /* Hook the tails of the branches to the closing node. */
    for (br = ret; br != NULL; br = regnext(br))
	regoptail(br, ender);

    /* Check for proper termination. */
    if (paren != REG_NOPAREN && getchr() != Magic(')'))
    {
#ifdef FEAT_SYN_HL
	if (paren == REG_ZPAREN)
	    EMSG_RET_NULL(_("E52: Unmatched \\z("));
	else
#endif
	    if (paren == REG_NPAREN)
	    EMSG_M_RET_NULL(_("E53: Unmatched %s%%("), reg_magic == MAGIC_ALL);
	else
	    EMSG_M_RET_NULL(_("E54: Unmatched %s("), reg_magic == MAGIC_ALL);
    }
    else if (paren == REG_NOPAREN && peekchr() != NUL)
    {
	if (curchr == Magic(')'))
	    EMSG_M_RET_NULL(_("E55: Unmatched %s)"), reg_magic == MAGIC_ALL);
	else
	    EMSG_RET_NULL(_(e_trailing));	/* "Can't happen". */
	/* NOTREACHED */
    }
    /*
     * Here we set the flag allowing back references to this set of
     * parentheses.
     */
    if (paren == REG_PAREN)
	had_endbrace[parno] = TRUE;	/* have seen the close paren */
    return ret;
}

/*
 * Handle one alternative of an | operator.
 * Implements the & operator.
 */
    static char_u *
regbranch(flagp)
    int		*flagp;
{
    char_u	*ret;
    char_u	*chain = NULL;
    char_u	*latest;
    int		flags;

    *flagp = WORST | HASNL;		/* Tentatively. */

    ret = regnode(BRANCH);
    for (;;)
    {
	latest = regconcat(&flags);
	if (latest == NULL)
	    return NULL;
	/* If one of the branches has width, the whole thing has.  If one of
	 * the branches anchors at start-of-line, the whole thing does.
	 * If one of the branches uses look-behind, the whole thing does. */
	*flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH);
	/* If one of the branches doesn't match a line-break, the whole thing
	 * doesn't. */
	*flagp &= ~HASNL | (flags & HASNL);
	if (chain != NULL)
	    regtail(chain, latest);
	if (peekchr() != Magic('&'))
	    break;
	skipchr();
	regtail(latest, regnode(END)); /* operand ends */
	reginsert(MATCH, latest);
	chain = latest;
    }

    return ret;
}

/*
 * Handle one alternative of an | or & operator.
 * Implements the concatenation operator.
 */
    static char_u *
regconcat(flagp)
    int		*flagp;
{
    char_u	*first = NULL;
    char_u	*chain = NULL;
    char_u	*latest;
    int		flags;
    int		cont = TRUE;

    *flagp = WORST;		/* Tentatively. */

    while (cont)
    {
	switch (peekchr())
	{
	    case NUL:
	    case Magic('|'):
	    case Magic('&'):
	    case Magic(')'):
			    cont = FALSE;
			    break;
	    case Magic('Z'):
#ifdef FEAT_MBYTE
			    regflags |= RF_ICOMBINE;
#endif
			    skipchr_keepstart();
			    break;
	    case Magic('c'):
			    regflags |= RF_ICASE;
			    skipchr_keepstart();
			    break;
	    case Magic('C'):
			    regflags |= RF_NOICASE;
			    skipchr_keepstart();
			    break;
	    case Magic('v'):
			    reg_magic = MAGIC_ALL;
			    skipchr_keepstart();
			    curchr = -1;
			    break;
	    case Magic('m'):
			    reg_magic = MAGIC_ON;
			    skipchr_keepstart();
			    curchr = -1;
			    break;
	    case Magic('M'):
			    reg_magic = MAGIC_OFF;
			    skipchr_keepstart();
			    curchr = -1;
			    break;
	    case Magic('V'):
			    reg_magic = MAGIC_NONE;
			    skipchr_keepstart();
			    curchr = -1;
			    break;
	    default:
			    latest = regpiece(&flags);
			    if (latest == NULL)
				return NULL;
			    *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH);
			    if (chain == NULL)	/* First piece. */
				*flagp |= flags & SPSTART;
			    else
				regtail(chain, latest);
			    chain = latest;
			    if (first == NULL)
				first = latest;
			    break;
	}
    }
    if (first == NULL)		/* Loop ran zero times. */
	first = regnode(NOTHING);
    return first;
}

/*
 * regpiece - something followed by possible [*+=]
 *
 * Note that the branching code sequences used for = and the general cases
 * of * and + are somewhat optimized:  they use the same NOTHING node as
 * both the endmarker for their branch list and the body of the last branch.
 * It might seem that this node could be dispensed with entirely, but the
 * endmarker role is not redundant.
 */
    static char_u *
regpiece(flagp)
    int		    *flagp;
{
    char_u	    *ret;
    int		    op;
    char_u	    *next;
    int		    flags;
    long	    minval;
    long	    maxval;

    ret = regatom(&flags);
    if (ret == NULL)
	return NULL;

    op = peekchr();
    if (re_multi_type(op) == NOT_MULTI)
    {
	*flagp = flags;
	return ret;
    }
    /* default flags */
    *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH)));

    skipchr();
    switch (op)
    {
	case Magic('*'):
	    if (flags & SIMPLE)
		reginsert(STAR, ret);
	    else
	    {
		/* Emit x* as (x&|), where & means "self". */
		reginsert(BRANCH, ret); /* Either x */
		regoptail(ret, regnode(BACK));	/* and loop */
		regoptail(ret, ret);	/* back */
		regtail(ret, regnode(BRANCH));	/* or */
		regtail(ret, regnode(NOTHING)); /* null. */
	    }
	    break;

	case Magic('+'):
	    if (flags & SIMPLE)
		reginsert(PLUS, ret);
	    else
	    {
		/* Emit x+ as x(&|), where & means "self". */
		next = regnode(BRANCH); /* Either */
		regtail(ret, next);
		regtail(regnode(BACK), ret);	/* loop back */
		regtail(next, regnode(BRANCH)); /* or */
		regtail(ret, regnode(NOTHING)); /* null. */
	    }
	    *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH)));
	    break;

	case Magic('@'):
	    {
		int	lop = END;

		switch (no_Magic(getchr()))
		{
		    case '=': lop = MATCH; break;		  /* \@= */
		    case '!': lop = NOMATCH; break;		  /* \@! */
		    case '>': lop = SUBPAT; break;		  /* \@> */
		    case '<': switch (no_Magic(getchr()))
			      {
				  case '=': lop = BEHIND; break;   /* \@<= */
				  case '!': lop = NOBEHIND; break; /* \@<! */
			      }
		}
		if (lop == END)
		    EMSG_M_RET_NULL(_("E59: invalid character after %s@"),
						      reg_magic == MAGIC_ALL);
		/* Look behind must match with behind_pos. */
		if (lop == BEHIND || lop == NOBEHIND)
		{
		    regtail(ret, regnode(BHPOS));
		    *flagp |= HASLOOKBH;
		}
		regtail(ret, regnode(END)); /* operand ends */
		reginsert(lop, ret);
		break;
	    }

	case Magic('?'):
	case Magic('='):
	    /* Emit x= as (x|) */
	    reginsert(BRANCH, ret);		/* Either x */
	    regtail(ret, regnode(BRANCH));	/* or */
	    next = regnode(NOTHING);		/* null. */
	    regtail(ret, next);
	    regoptail(ret, next);
	    break;

	case Magic('{'):
	    if (!read_limits(&minval, &maxval))
		return NULL;
	    if (flags & SIMPLE)
	    {
		reginsert(BRACE_SIMPLE, ret);
		reginsert_limits(BRACE_LIMITS, minval, maxval, ret);
	    }
	    else
	    {
		if (num_complex_braces >= 10)
		    EMSG_M_RET_NULL(_("E60: Too many complex %s{...}s"),
						      reg_magic == MAGIC_ALL);
		reginsert(BRACE_COMPLEX + num_complex_braces, ret);
		regoptail(ret, regnode(BACK));
		regoptail(ret, ret);
		reginsert_limits(BRACE_LIMITS, minval, maxval, ret);
		++num_complex_braces;
	    }
	    if (minval > 0 && maxval > 0)
		*flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH)));
	    break;
    }
    if (re_multi_type(peekchr()) != NOT_MULTI)
    {
	/* Can't have a multi follow a multi. */
	if (peekchr() == Magic('*'))
	    sprintf((char *)IObuff, _("E61: Nested %s*"),
					    reg_magic >= MAGIC_ON ? "" : "\\");
	else
	    sprintf((char *)IObuff, _("E62: Nested %s%c"),
		reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr()));
	EMSG_RET_NULL(IObuff);
    }

    return ret;
}

/*
 * regatom - the lowest level
 *
 * Optimization:  gobbles an entire sequence of ordinary characters so that
 * it can turn them into a single node, which is smaller to store and
 * faster to run.  Don't do this when one_exactly is set.
 */
    static char_u *
regatom(flagp)
    int		   *flagp;
{
    char_u	    *ret;
    int		    flags;
    int		    cpo_lit;	    /* 'cpoptions' contains 'l' flag */
    int		    cpo_bsl;	    /* 'cpoptions' contains '\' flag */
    int		    c;
    static char_u   *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU";
    static int	    classcodes[] = {ANY, IDENT, SIDENT, KWORD, SKWORD,
				    FNAME, SFNAME, PRINT, SPRINT,
				    WHITE, NWHITE, DIGIT, NDIGIT,
				    HEX, NHEX, OCTAL, NOCTAL,
				    WORD, NWORD, HEAD, NHEAD,
				    ALPHA, NALPHA, LOWER, NLOWER,
				    UPPER, NUPPER
				    };
    char_u	    *p;
    int		    extra = 0;

    *flagp = WORST;		/* Tentatively. */
    cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL;
    cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL;

    c = getchr();
    switch (c)
    {
      case Magic('^'):
	ret = regnode(BOL);
	break;

      case Magic('$'):
	ret = regnode(EOL);
#if defined(FEAT_SYN_HL) || defined(PROTO)
	had_eol = TRUE;
#endif
	break;

      case Magic('<'):
	ret = regnode(BOW);
	break;

      case Magic('>'):
	ret = regnode(EOW);
	break;

      case Magic('_'):
	c = no_Magic(getchr());
	if (c == '^')		/* "\_^" is start-of-line */
	{
	    ret = regnode(BOL);
	    break;
	}
	if (c == '$')		/* "\_$" is end-of-line */
	{
	    ret = regnode(EOL);
#if defined(FEAT_SYN_HL) || defined(PROTO)
	    had_eol = TRUE;
#endif
	    break;
	}

	extra = ADD_NL;
	*flagp |= HASNL;

	/* "\_[" is character range plus newline */
	if (c == '[')
	    goto collection;

	/* "\_x" is character class plus newline */
	/*FALLTHROUGH*/

	/*
	 * Character classes.
	 */
      case Magic('.'):
      case Magic('i'):
      case Magic('I'):
      case Magic('k'):
      case Magic('K'):
      case Magic('f'):
      case Magic('F'):
      case Magic('p'):
      case Magic('P'):
      case Magic('s'):
      case Magic('S'):
      case Magic('d'):
      case Magic('D'):
      case Magic('x'):
      case Magic('X'):
      case Magic('o'):
      case Magic('O'):
      case Magic('w'):
      case Magic('W'):
      case Magic('h'):
      case Magic('H'):
      case Magic('a'):
      case Magic('A'):
      case Magic('l'):
      case Magic('L'):
      case Magic('u'):
      case Magic('U'):
	p = vim_strchr(classchars, no_Magic(c));
	if (p == NULL)
	    EMSG_RET_NULL(_("E63: invalid use of \\_"));
#ifdef FEAT_MBYTE
	/* When '.' is followed by a composing char ignore the dot, so that
	 * the composing char is matched here. */
	if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr()))
	{
	    c = getchr();
	    goto do_multibyte;
	}
#endif
	ret = regnode(classcodes[p - classchars] + extra);
	*flagp |= HASWIDTH | SIMPLE;
	break;

      case Magic('n'):
	if (reg_string)
	{
	    /* In a string "\n" matches a newline character. */
	    ret = regnode(EXACTLY);
	    regc(NL);
	    regc(NUL);
	    *flagp |= HASWIDTH | SIMPLE;
	}
	else
	{
	    /* In buffer text "\n" matches the end of a line. */
	    ret = regnode(NEWL);
	    *flagp |= HASWIDTH | HASNL;
	}
	break;

      case Magic('('):
	if (one_exactly)
	    EMSG_ONE_RET_NULL;
	ret = reg(REG_PAREN, &flags);
	if (ret == NULL)
	    return NULL;
	*flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH);
	break;

      case NUL:
      case Magic('|'):
      case Magic('&'):
      case Magic(')'):
	if (one_exactly)
	    EMSG_ONE_RET_NULL;
	EMSG_RET_NULL(_(e_internal));	/* Supposed to be caught earlier. */
	/* NOTREACHED */

      case Magic('='):
      case Magic('?'):
      case Magic('+'):
      case Magic('@'):
      case Magic('{'):
      case Magic('*'):
	c = no_Magic(c);
	sprintf((char *)IObuff, _("E64: %s%c follows nothing"),
		(c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL)
		? "" : "\\", c);
	EMSG_RET_NULL(IObuff);
	/* NOTREACHED */

      case Magic('~'):		/* previous substitute pattern */
	    if (reg_prev_sub != NULL)
	    {
		char_u	    *lp;

		ret = regnode(EXACTLY);
		lp = reg_prev_sub;
		while (*lp != NUL)
		    regc(*lp++);
		regc(NUL);
		if (*reg_prev_sub != NUL)
		{
		    *flagp |= HASWIDTH;
		    if ((lp - reg_prev_sub) == 1)
			*flagp |= SIMPLE;
		}
	    }
	    else
		EMSG_RET_NULL(_(e_nopresub));
	    break;

      case Magic('1'):
      case Magic('2'):
      case Magic('3'):
      case Magic('4'):
      case Magic('5'):
      case Magic('6'):
      case Magic('7'):
      case Magic('8'):
      case Magic('9'):
	    {
		int		    refnum;

		refnum = c - Magic('0');
		/*
		 * Check if the back reference is legal. We must have seen the
		 * close brace.
		 * TODO: Should also check that we don't refer to something
		 * that is repeated (+*=): what instance of the repetition
		 * should we match?
		 */
		if (!had_endbrace[refnum])
		{
		    /* Trick: check if "@<=" or "@<!" follows, in which case
		     * the \1 can appear before the referenced match. */
		    for (p = regparse; *p != NUL; ++p)
			if (p[0] == '@' && p[1] == '<'
					      && (p[2] == '!' || p[2] == '='))
			    break;
		    if (*p == NUL)
			EMSG_RET_NULL(_("E65: Illegal back reference"));
		}
		ret = regnode(BACKREF + refnum);
	    }
	    break;

      case Magic('z'):
	{
	    c = no_Magic(getchr());
	    switch (c)
	    {
#ifdef FEAT_SYN_HL
		case '(': if (reg_do_extmatch != REX_SET)
			      EMSG_RET_NULL(_("E66: \\z( not allowed here"));
			  if (one_exactly)
			      EMSG_ONE_RET_NULL;
			  ret = reg(REG_ZPAREN, &flags);
			  if (ret == NULL)
			      return NULL;
			  *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH);
			  re_has_z = REX_SET;
			  break;

		case '1':
		case '2':
		case '3':
		case '4':
		case '5':
		case '6':
		case '7':
		case '8':
		case '9': if (reg_do_extmatch != REX_USE)
			      EMSG_RET_NULL(_("E67: \\z1 et al. not allowed here"));
			  ret = regnode(ZREF + c - '0');
			  re_has_z = REX_USE;
			  break;
#endif

		case 's': ret = regnode(MOPEN + 0);
			  break;

		case 'e': ret = regnode(MCLOSE + 0);
			  break;

		default:  EMSG_RET_NULL(_("E68: Invalid character after \\z"));
	    }
	}
	break;

      case Magic('%'):
	{
	    c = no_Magic(getchr());
	    switch (c)
	    {
		/* () without a back reference */
		case '(':
		    if (one_exactly)
			EMSG_ONE_RET_NULL;
		    ret = reg(REG_NPAREN, &flags);
		    if (ret == NULL)
			return NULL;
		    *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH);
		    break;

		/* Catch \%^ and \%$ regardless of where they appear in the
		 * pattern -- regardless of whether or not it makes sense. */
		case '^':
		    ret = regnode(RE_BOF);
		    break;

		case '$':
		    ret = regnode(RE_EOF);
		    break;

		case '#':
		    ret = regnode(CURSOR);
		    break;

		case 'V':
		    ret = regnode(RE_VISUAL);
		    break;

		/* \%[abc]: Emit as a list of branches, all ending at the last
		 * branch which matches nothing. */
		case '[':
			  if (one_exactly)	/* doesn't nest */
			      EMSG_ONE_RET_NULL;
			  {
			      char_u	*lastbranch;
			      char_u	*lastnode = NULL;
			      char_u	*br;

			      ret = NULL;
			      while ((c = getchr()) != ']')
			      {
				  if (c == NUL)
				      EMSG_M_RET_NULL(_("E69: Missing ] after %s%%["),
						      reg_magic == MAGIC_ALL);
				  br = regnode(BRANCH);
				  if (ret == NULL)
				      ret = br;
				  else
				      regtail(lastnode, br);

				  ungetchr();
				  one_exactly = TRUE;
				  lastnode = regatom(flagp);
				  one_exactly = FALSE;
				  if (lastnode == NULL)
				      return NULL;
			      }
			      if (ret == NULL)
				  EMSG_M_RET_NULL(_("E70: Empty %s%%[]"),
						      reg_magic == MAGIC_ALL);
			      lastbranch = regnode(BRANCH);
			      br = regnode(NOTHING);
			      if (ret != JUST_CALC_SIZE)
			      {
				  regtail(lastnode, br);
				  regtail(lastbranch, br);
				  /* connect all branches to the NOTHING
				   * branch at the end */
				  for (br = ret; br != lastnode; )
				  {
				      if (OP(br) == BRANCH)
				      {
					  regtail(br, lastbranch);
					  br = OPERAND(br);
				      }
				      else
					  br = regnext(br);
				  }
			      }
			      *flagp &= ~HASWIDTH;
			      break;
			  }

		case 'd':   /* %d123 decimal */
		case 'o':   /* %o123 octal */
		case 'x':   /* %xab hex 2 */
		case 'u':   /* %uabcd hex 4 */
		case 'U':   /* %U1234abcd hex 8 */
			  {
			      int i;

			      switch (c)
			      {
				  case 'd': i = getdecchrs(); break;
				  case 'o': i = getoctchrs(); break;
				  case 'x': i = gethexchrs(2); break;
				  case 'u': i = gethexchrs(4); break;
				  case 'U': i = gethexchrs(8); break;
				  default:  i = -1; break;
			      }

			      if (i < 0)
				  EMSG_M_RET_NULL(
					_("E678: Invalid character after %s%%[dxouU]"),
					reg_magic == MAGIC_ALL);
#ifdef FEAT_MBYTE
			      if (use_multibytecode(i))
				  ret = regnode(MULTIBYTECODE);
			      else
#endif
				  ret = regnode(EXACTLY);
			      if (i == 0)
				  regc(0x0a);
			      else
#ifdef FEAT_MBYTE
				  regmbc(i);
#else
				  regc(i);
#endif
			      regc(NUL);
			      *flagp |= HASWIDTH;
			      break;
			  }

		default:
			  if (VIM_ISDIGIT(c) || c == '<' || c == '>'
								 || c == '\'')
			  {
			      long_u	n = 0;
			      int	cmp;

			      cmp = c;
			      if (cmp == '<' || cmp == '>')
				  c = getchr();
			      while (VIM_ISDIGIT(c))
			      {
				  n = n * 10 + (c - '0');
				  c = getchr();
			      }
			      if (c == '\'' && n == 0)
			      {
				  /* "\%'m", "\%<'m" and "\%>'m": Mark */
				  c = getchr();
				  ret = regnode(RE_MARK);
				  if (ret == JUST_CALC_SIZE)
				      regsize += 2;
				  else
				  {
				      *regcode++ = c;
				      *regcode++ = cmp;
				  }
				  break;
			      }
			      else if (c == 'l' || c == 'c' || c == 'v')
			      {
				  if (c == 'l')
				      ret = regnode(RE_LNUM);
				  else if (c == 'c')
				      ret = regnode(RE_COL);
				  else
				      ret = regnode(RE_VCOL);
				  if (ret == JUST_CALC_SIZE)
				      regsize += 5;
				  else
				  {
				      /* put the number and the optional
				       * comparator after the opcode */
				      regcode = re_put_long(regcode, n);
				      *regcode++ = cmp;
				  }
				  break;
			      }
			  }

			  EMSG_M_RET_NULL(_("E71: Invalid character after %s%%"),
						      reg_magic == MAGIC_ALL);
	    }
	}
	break;

      case Magic('['):
collection:
	{
	    char_u	*lp;

	    /*
	     * If there is no matching ']', we assume the '[' is a normal
	     * character.  This makes 'incsearch' and ":help [" work.
	     */
	    lp = skip_anyof(regparse);
	    if (*lp == ']')	/* there is a matching ']' */
	    {
		int	startc = -1;	/* > 0 when next '-' is a range */
		int	endc;

		/*
		 * In a character class, different parsing rules apply.
		 * Not even \ is special anymore, nothing is.
		 */
		if (*regparse == '^')	    /* Complement of range. */
		{
		    ret = regnode(ANYBUT + extra);
		    regparse++;
		}
		else
		    ret = regnode(ANYOF + extra);

		/* At the start ']' and '-' mean the literal character. */
		if (*regparse == ']' || *regparse == '-')
		{
		    startc = *regparse;
		    regc(*regparse++);
		}

		while (*regparse != NUL && *regparse != ']')
		{
		    if (*regparse == '-')
		    {
			++regparse;
			/* The '-' is not used for a range at the end and
			 * after or before a '\n'. */
			if (*regparse == ']' || *regparse == NUL
				|| startc == -1
				|| (regparse[0] == '\\' && regparse[1] == 'n'))
			{
			    regc('-');
			    startc = '-';	/* [--x] is a range */
			}
			else
			{
			    /* Also accept "a-[.z.]" */
			    endc = 0;
			    if (*regparse == '[')
				endc = get_coll_element(&regparse);
			    if (endc == 0)
			    {
#ifdef FEAT_MBYTE
				if (has_mbyte)
				    endc = mb_ptr2char_adv(&regparse);
				else
#endif
				    endc = *regparse++;
			    }

			    /* Handle \o40, \x20 and \u20AC style sequences */
			    if (endc == '\\' && !cpo_lit && !cpo_bsl)
				endc = coll_get_char();

			    if (startc > endc)
				EMSG_RET_NULL(_(e_invrange));
#ifdef FEAT_MBYTE
			    if (has_mbyte && ((*mb_char2len)(startc) > 1
						 || (*mb_char2len)(endc) > 1))
			    {
				/* Limit to a range of 256 chars */
				if (endc > startc + 256)
				    EMSG_RET_NULL(_(e_invrange));
				while (++startc <= endc)
				    regmbc(startc);
			    }
			    else
#endif
			    {
#ifdef EBCDIC
				int	alpha_only = FALSE;

				/* for alphabetical range skip the gaps
				 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'.  */
				if (isalpha(startc) && isalpha(endc))
				    alpha_only = TRUE;
#endif
				while (++startc <= endc)
#ifdef EBCDIC
				    if (!alpha_only || isalpha(startc))
#endif
					regc(startc);
			    }
			    startc = -1;
			}
		    }
		    /*
		     * Only "\]", "\^", "\]" and "\\" are special in Vi.  Vim
		     * accepts "\t", "\e", etc., but only when the 'l' flag in
		     * 'cpoptions' is not included.
		     * Posix doesn't recognize backslash at all.
		     */
		    else if (*regparse == '\\'
			    && !cpo_bsl
			    && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL
				|| (!cpo_lit
				    && vim_strchr(REGEXP_ABBR,
						       regparse[1]) != NULL)))
		    {
			regparse++;
			if (*regparse == 'n')
			{
			    /* '\n' in range: also match NL */
			    if (ret != JUST_CALC_SIZE)
			    {
				if (*ret == ANYBUT)
				    *ret = ANYBUT + ADD_NL;
				else if (*ret == ANYOF)
				    *ret = ANYOF + ADD_NL;
				/* else: must have had a \n already */
			    }
			    *flagp |= HASNL;
			    regparse++;
			    startc = -1;
			}
			else if (*regparse == 'd'
				|| *regparse == 'o'
				|| *regparse == 'x'
				|| *regparse == 'u'
				|| *regparse == 'U')
			{
			    startc = coll_get_char();
			    if (startc == 0)
				regc(0x0a);
			    else
#ifdef FEAT_MBYTE
				regmbc(startc);
#else
				regc(startc);
#endif
			}
			else
			{
			    startc = backslash_trans(*regparse++);
			    regc(startc);
			}
		    }
		    else if (*regparse == '[')
		    {
			int c_class;
			int cu;

			c_class = get_char_class(&regparse);
			startc = -1;
			/* Characters assumed to be 8 bits! */
			switch (c_class)
			{
			    case CLASS_NONE:
				c_class = get_equi_class(&regparse);
				if (c_class != 0)
				{
				    /* produce equivalence class */
				    reg_equi_class(c_class);
				}
				else if ((c_class =
					    get_coll_element(&regparse)) != 0)
				{
				    /* produce a collating element */
				    regmbc(c_class);
				}
				else
				{
				    /* literal '[', allow [[-x] as a range */
				    startc = *regparse++;
				    regc(startc);
				}
				break;
			    case CLASS_ALNUM:
				for (cu = 1; cu <= 255; cu++)
				    if (isalnum(cu))
					regc(cu);
				break;
			    case CLASS_ALPHA:
				for (cu = 1; cu <= 255; cu++)
				    if (isalpha(cu))
					regc(cu);
				break;
			    case CLASS_BLANK:
				regc(' ');
				regc('\t');
				break;
			    case CLASS_CNTRL:
				for (cu = 1; cu <= 255; cu++)
				    if (iscntrl(cu))
					regc(cu);
				break;
			    case CLASS_DIGIT:
				for (cu = 1; cu <= 255; cu++)
				    if (VIM_ISDIGIT(cu))
					regc(cu);
				break;
			    case CLASS_GRAPH:
				for (cu = 1; cu <= 255; cu++)
				    if (isgraph(cu))
					regc(cu);
				break;
			    case CLASS_LOWER:
				for (cu = 1; cu <= 255; cu++)
				    if (MB_ISLOWER(cu))
					regc(cu);
				break;
			    case CLASS_PRINT:
				for (cu = 1; cu <= 255; cu++)
				    if (vim_isprintc(cu))
					regc(cu);
				break;
			    case CLASS_PUNCT:
				for (cu = 1; cu <= 255; cu++)
				    if (ispunct(cu))
					regc(cu);
				break;
			    case CLASS_SPACE:
				for (cu = 9; cu <= 13; cu++)
				    regc(cu);
				regc(' ');
				break;
			    case CLASS_UPPER:
				for (cu = 1; cu <= 255; cu++)
				    if (MB_ISUPPER(cu))
					regc(cu);
				break;
			    case CLASS_XDIGIT:
				for (cu = 1; cu <= 255; cu++)
				    if (vim_isxdigit(cu))
					regc(cu);
				break;
			    case CLASS_TAB:
				regc('\t');
				break;
			    case CLASS_RETURN:
				regc('\r');
				break;
			    case CLASS_BACKSPACE:
				regc('\b');
				break;
			    case CLASS_ESCAPE:
				regc('\033');
				break;
			}
		    }
		    else
		    {
#ifdef FEAT_MBYTE
			if (has_mbyte)
			{
			    int	len;

			    /* produce a multibyte character, including any
			     * following composing characters */
			    startc = mb_ptr2char(regparse);
			    len = (*mb_ptr2len)(regparse);
			    if (enc_utf8 && utf_char2len(startc) != len)
				startc = -1;	/* composing chars */
			    while (--len >= 0)
				regc(*regparse++);
			}
			else
#endif
			{
			    startc = *regparse++;
			    regc(startc);
			}
		    }
		}
		regc(NUL);
		prevchr_len = 1;	/* last char was the ']' */
		if (*regparse != ']')
		    EMSG_RET_NULL(_(e_toomsbra));	/* Cannot happen? */
		skipchr();	    /* let's be friends with the lexer again */
		*flagp |= HASWIDTH | SIMPLE;
		break;
	    }
	    else if (reg_strict)
		EMSG_M_RET_NULL(_("E769: Missing ] after %s["),
						       reg_magic > MAGIC_OFF);
	}
	/* FALLTHROUGH */

      default:
	{
	    int		len;

#ifdef FEAT_MBYTE
	    /* A multi-byte character is handled as a separate atom if it's
	     * before a multi and when it's a composing char. */
	    if (use_multibytecode(c))
	    {
do_multibyte:
		ret = regnode(MULTIBYTECODE);
		regmbc(c);
		*flagp |= HASWIDTH | SIMPLE;
		break;
	    }
#endif

	    ret = regnode(EXACTLY);

	    /*
	     * Append characters as long as:
	     * - there is no following multi, we then need the character in
	     *   front of it as a single character operand
	     * - not running into a Magic character
	     * - "one_exactly" is not set
	     * But always emit at least one character.  Might be a Multi,
	     * e.g., a "[" without matching "]".
	     */
	    for (len = 0; c != NUL && (len == 0
			|| (re_multi_type(peekchr()) == NOT_MULTI
			    && !one_exactly
			    && !is_Magic(c))); ++len)
	    {
		c = no_Magic(c);
#ifdef FEAT_MBYTE
		if (has_mbyte)
		{
		    regmbc(c);
		    if (enc_utf8)
		    {
			int	l;

			/* Need to get composing character too. */
			for (;;)
			{
			    l = utf_ptr2len(regparse);
			    if (!UTF_COMPOSINGLIKE(regparse, regparse + l))
				break;
			    regmbc(utf_ptr2char(regparse));
			    skipchr();
			}
		    }
		}
		else
#endif
		    regc(c);
		c = getchr();
	    }
	    ungetchr();

	    regc(NUL);
	    *flagp |= HASWIDTH;
	    if (len == 1)
		*flagp |= SIMPLE;
	}
	break;
    }

    return ret;
}

#ifdef FEAT_MBYTE
/*
 * Return TRUE if MULTIBYTECODE should be used instead of EXACTLY for
 * character "c".
 */
    static int
use_multibytecode(c)
    int c;
{
    return has_mbyte && (*mb_char2len)(c) > 1
		     && (re_multi_type(peekchr()) != NOT_MULTI
			     || (enc_utf8 && utf_iscomposing(c)));
}
#endif

/*
 * emit a node
 * Return pointer to generated code.
 */
    static char_u *
regnode(op)
    int		op;
{
    char_u  *ret;

    ret = regcode;
    if (ret == JUST_CALC_SIZE)
	regsize += 3;
    else
    {
	*regcode++ = op;
	*regcode++ = NUL;		/* Null "next" pointer. */
	*regcode++ = NUL;
    }
    return ret;
}

/*
 * Emit (if appropriate) a byte of code
 */
    static void
regc(b)
    int		b;
{
    if (regcode == JUST_CALC_SIZE)
	regsize++;
    else
	*regcode++ = b;
}

#ifdef FEAT_MBYTE
/*
 * Emit (if appropriate) a multi-byte character of code
 */
    static void
regmbc(c)
    int		c;
{
    if (regcode == JUST_CALC_SIZE)
	regsize += (*mb_char2len)(c);
    else
	regcode += (*mb_char2bytes)(c, regcode);
}
#endif

/*
 * reginsert - insert an operator in front of already-emitted operand
 *
 * Means relocating the operand.
 */
    static void
reginsert(op, opnd)
    int		op;
    char_u     *opnd;
{
    char_u	*src;
    char_u	*dst;
    char_u	*place;

    if (regcode == JUST_CALC_SIZE)
    {
	regsize += 3;
	return;
    }
    src = regcode;
    regcode += 3;
    dst = regcode;
    while (src > opnd)
	*--dst = *--src;

    place = opnd;		/* Op node, where operand used to be. */
    *place++ = op;
    *place++ = NUL;
    *place = NUL;
}

/*
 * reginsert_limits - insert an operator in front of already-emitted operand.
 * The operator has the given limit values as operands.  Also set next pointer.
 *
 * Means relocating the operand.
 */
    static void
reginsert_limits(op, minval, maxval, opnd)
    int		op;
    long	minval;
    long	maxval;
    char_u	*opnd;
{
    char_u	*src;
    char_u	*dst;
    char_u	*place;

    if (regcode == JUST_CALC_SIZE)
    {
	regsize += 11;
	return;
    }
    src = regcode;
    regcode += 11;
    dst = regcode;
    while (src > opnd)
	*--dst = *--src;

    place = opnd;		/* Op node, where operand used to be. */
    *place++ = op;
    *place++ = NUL;
    *place++ = NUL;
    place = re_put_long(place, (long_u)minval);
    place = re_put_long(place, (long_u)maxval);
    regtail(opnd, place);
}

/*
 * Write a long as four bytes at "p" and return pointer to the next char.
 */
    static char_u *
re_put_long(p, val)
    char_u	*p;
    long_u	val;
{
    *p++ = (char_u) ((val >> 24) & 0377);
    *p++ = (char_u) ((val >> 16) & 0377);
    *p++ = (char_u) ((val >> 8) & 0377);
    *p++ = (char_u) (val & 0377);
    return p;
}

/*
 * regtail - set the next-pointer at the end of a node chain
 */
    static void
regtail(p, val)
    char_u	*p;
    char_u	*val;
{
    char_u	*scan;
    char_u	*temp;
    int		offset;

    if (p == JUST_CALC_SIZE)
	return;

    /* Find last node. */
    scan = p;
    for (;;)
    {
	temp = regnext(scan);
	if (temp == NULL)
	    break;
	scan = temp;
    }

    if (OP(scan) == BACK)
	offset = (int)(scan - val);
    else
	offset = (int)(val - scan);
    *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377);
    *(scan + 2) = (char_u) (offset & 0377);
}

/*
 * regoptail - regtail on item after a BRANCH; nop if none
 */
    static void
regoptail(p, val)
    char_u	*p;
    char_u	*val;
{
    /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */
    if (p == NULL || p == JUST_CALC_SIZE
	    || (OP(p) != BRANCH
		&& (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9)))
	return;
    regtail(OPERAND(p), val);
}

/*
 * getchr() - get the next character from the pattern. We know about
 * magic and such, so therefore we need a lexical analyzer.
 */

/* static int	    curchr; */
static int	prevprevchr;
static int	prevchr;
static int	nextchr;    /* used for ungetchr() */
/*
 * Note: prevchr is sometimes -1 when we are not at the start,
 * eg in /[ ^I]^ the pattern was never found even if it existed, because ^ was
 * taken to be magic -- webb
 */
static int	at_start;	/* True when on the first character */
static int	prev_at_start;  /* True when on the second character */

    static void
initchr(str)
    char_u *str;
{
    regparse = str;
    prevchr_len = 0;
    curchr = prevprevchr = prevchr = nextchr = -1;
    at_start = TRUE;
    prev_at_start = FALSE;
}

    static int
peekchr()
{
    static int	after_slash = FALSE;

    if (curchr == -1)
    {
	switch (curchr = regparse[0])
	{
	case '.':
	case '[':
	case '~':
	    /* magic when 'magic' is on */
	    if (reg_magic >= MAGIC_ON)
		curchr = Magic(curchr);
	    break;
	case '(':
	case ')':
	case '{':
	case '%':
	case '+':
	case '=':
	case '?':
	case '@':
	case '!':
	case '&':
	case '|':
	case '<':
	case '>':
	case '#':	/* future ext. */
	case '"':	/* future ext. */
	case '\'':	/* future ext. */
	case ',':	/* future ext. */
	case '-':	/* future ext. */
	case ':':	/* future ext. */
	case ';':	/* future ext. */
	case '`':	/* future ext. */
	case '/':	/* Can't be used in / command */
	    /* magic only after "\v" */
	    if (reg_magic == MAGIC_ALL)
		curchr = Magic(curchr);
	    break;
	case '*':
	    /* * is not magic as the very first character, eg "?*ptr", when
	     * after '^', eg "/^*ptr" and when after "\(", "\|", "\&".  But
	     * "\(\*" is not magic, thus must be magic if "after_slash" */
	    if (reg_magic >= MAGIC_ON
		    && !at_start
		    && !(prev_at_start && prevchr == Magic('^'))
		    && (after_slash
			|| (prevchr != Magic('(')
			    && prevchr != Magic('&')
			    && prevchr != Magic('|'))))
		curchr = Magic('*');
	    break;
	case '^':
	    /* '^' is only magic as the very first character and if it's after
	     * "\(", "\|", "\&' or "\n" */
	    if (reg_magic >= MAGIC_OFF
		    && (at_start
			|| reg_magic == MAGIC_ALL
			|| prevchr == Magic('(')
			|| prevchr == Magic('|')
			|| prevchr == Magic('&')
			|| prevchr == Magic('n')
			|| (no_Magic(prevchr) == '('
			    && prevprevchr == Magic('%'))))
	    {
		curchr = Magic('^');
		at_start = TRUE;
		prev_at_start = FALSE;
	    }
	    break;
	case '$':
	    /* '$' is only magic as the very last char and if it's in front of
	     * either "\|", "\)", "\&", or "\n" */
	    if (reg_magic >= MAGIC_OFF)
	    {
		char_u *p = regparse + 1;

		/* ignore \c \C \m and \M after '$' */
		while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C'
				|| p[1] == 'm' || p[1] == 'M' || p[1] == 'Z'))
		    p += 2;
		if (p[0] == NUL
			|| (p[0] == '\\'
			    && (p[1] == '|' || p[1] == '&' || p[1] == ')'
				|| p[1] == 'n'))
			|| reg_magic == MAGIC_ALL)
		    curchr = Magic('$');
	    }
	    break;
	case '\\':
	    {
		int c = regparse[1];

		if (c == NUL)
		    curchr = '\\';	/* trailing '\' */
		else if (
#ifdef EBCDIC
			vim_strchr(META, c)
#else
			c <= '~' && META_flags[c]
#endif
			)
		{
		    /*
		     * META contains everything that may be magic sometimes,
		     * except ^ and $ ("\^" and "\$" are only magic after
		     * "\v").  We now fetch the next character and toggle its
		     * magicness.  Therefore, \ is so meta-magic that it is
		     * not in META.
		     */
		    curchr = -1;
		    prev_at_start = at_start;
		    at_start = FALSE;	/* be able to say "/\*ptr" */
		    ++regparse;
		    ++after_slash;
		    peekchr();
		    --regparse;
		    --after_slash;
		    curchr = toggle_Magic(curchr);
		}
		else if (vim_strchr(REGEXP_ABBR, c))
		{
		    /*
		     * Handle abbreviations, like "\t" for TAB -- webb
		     */
		    curchr = backslash_trans(c);
		}
		else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^'))
		    curchr = toggle_Magic(c);
		else
		{
		    /*
		     * Next character can never be (made) magic?
		     * Then backslashing it won't do anything.
		     */
#ifdef FEAT_MBYTE
		    if (has_mbyte)
			curchr = (*mb_ptr2char)(regparse + 1);
		    else
#endif
			curchr = c;
		}
		break;
	    }

#ifdef FEAT_MBYTE
	default:
	    if (has_mbyte)
		curchr = (*mb_ptr2char)(regparse);
#endif
	}
    }

    return curchr;
}

/*
 * Eat one lexed character.  Do this in a way that we can undo it.
 */
    static void
skipchr()
{
    /* peekchr() eats a backslash, do the same here */
    if (*regparse == '\\')
	prevchr_len = 1;
    else
	prevchr_len = 0;
    if (regparse[prevchr_len] != NUL)
    {
#ifdef FEAT_MBYTE
	if (enc_utf8)
	    /* exclude composing chars that mb_ptr2len does include */
	    prevchr_len += utf_ptr2len(regparse + prevchr_len);
	else if (has_mbyte)
	    prevchr_len += (*mb_ptr2len)(regparse + prevchr_len);
	else
#endif
	    ++prevchr_len;
    }
    regparse += prevchr_len;
    prev_at_start = at_start;
    at_start = FALSE;
    prevprevchr = prevchr;
    prevchr = curchr;
    curchr = nextchr;	    /* use previously unget char, or -1 */
    nextchr = -1;
}

/*
 * Skip a character while keeping the value of prev_at_start for at_start.
 * prevchr and prevprevchr are also kept.
 */
    static void
skipchr_keepstart()
{
    int as = prev_at_start;
    int pr = prevchr;
    int prpr = prevprevchr;

    skipchr();
    at_start = as;
    prevchr = pr;
    prevprevchr = prpr;
}

    static int
getchr()
{
    int chr = peekchr();

    skipchr();
    return chr;
}

/*
 * put character back.  Works only once!
 */
    static void
ungetchr()
{
    nextchr = curchr;
    curchr = prevchr;
    prevchr = prevprevchr;
    at_start = prev_at_start;
    prev_at_start = FALSE;

    /* Backup regparse, so that it's at the same position as before the
     * getchr(). */
    regparse -= prevchr_len;
}

/*
 * Get and return the value of the hex string at the current position.
 * Return -1 if there is no valid hex number.
 * The position is updated:
 *     blahblah\%x20asdf
 *	   before-^ ^-after
 * The parameter controls the maximum number of input characters. This will be
 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence.
 */
    static int
gethexchrs(maxinputlen)
    int		maxinputlen;
{
    int		nr = 0;
    int		c;
    int		i;

    for (i = 0; i < maxinputlen; ++i)
    {
	c = regparse[0];
	if (!vim_isxdigit(c))
	    break;
	nr <<= 4;
	nr |= hex2nr(c);
	++regparse;
    }

    if (i == 0)
	return -1;
    return nr;
}

/*
 * get and return the value of the decimal string immediately after the
 * current position. Return -1 for invalid.  Consumes all digits.
 */
    static int
getdecchrs()
{
    int		nr = 0;
    int		c;
    int		i;

    for (i = 0; ; ++i)
    {
	c = regparse[0];
	if (c < '0' || c > '9')
	    break;
	nr *= 10;
	nr += c - '0';
	++regparse;
    }

    if (i == 0)
	return -1;
    return nr;
}

/*
 * get and return the value of the octal string immediately after the current
 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle
 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't
 * treat 8 or 9 as recognised characters. Position is updated:
 *     blahblah\%o210asdf
 *	   before-^  ^-after
 */
    static int
getoctchrs()
{
    int		nr = 0;
    int		c;
    int		i;

    for (i = 0; i < 3 && nr < 040; ++i)
    {
	c = regparse[0];
	if (c < '0' || c > '7')
	    break;
	nr <<= 3;
	nr |= hex2nr(c);
	++regparse;
    }

    if (i == 0)
	return -1;
    return nr;
}

/*
 * Get a number after a backslash that is inside [].
 * When nothing is recognized return a backslash.
 */
    static int
coll_get_char()
{
    int	    nr = -1;

    switch (*regparse++)
    {
	case 'd': nr = getdecchrs(); break;
	case 'o': nr = getoctchrs(); break;
	case 'x': nr = gethexchrs(2); break;
	case 'u': nr = gethexchrs(4); break;
	case 'U': nr = gethexchrs(8); break;
    }
    if (nr < 0)
    {
	/* If getting the number fails be backwards compatible: the character
	 * is a backslash. */
	--regparse;
	nr = '\\';
    }
    return nr;
}

/*
 * read_limits - Read two integers to be taken as a minimum and maximum.
 * If the first character is '-', then the range is reversed.
 * Should end with 'end'.  If minval is missing, zero is default, if maxval is
 * missing, a very big number is the default.
 */
    static int
read_limits(minval, maxval)
    long	*minval;
    long	*maxval;
{
    int		reverse = FALSE;
    char_u	*first_char;
    long	tmp;

    if (*regparse == '-')
    {
	/* Starts with '-', so reverse the range later */
	regparse++;
	reverse = TRUE;
    }
    first_char = regparse;
    *minval = getdigits(&regparse);
    if (*regparse == ',')	    /* There is a comma */
    {
	if (vim_isdigit(*++regparse))
	    *maxval = getdigits(&regparse);
	else
	    *maxval = MAX_LIMIT;
    }
    else if (VIM_ISDIGIT(*first_char))
	*maxval = *minval;	    /* It was \{n} or \{-n} */
    else
	*maxval = MAX_LIMIT;	    /* It was \{} or \{-} */
    if (*regparse == '\\')
	regparse++;	/* Allow either \{...} or \{...\} */
    if (*regparse != '}')
    {
	sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"),
					  reg_magic == MAGIC_ALL ? "" : "\\");
	EMSG_RET_FAIL(IObuff);
    }

    /*
     * Reverse the range if there was a '-', or make sure it is in the right
     * order otherwise.
     */
    if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval))
    {
	tmp = *minval;
	*minval = *maxval;
	*maxval = tmp;
    }
    skipchr();		/* let's be friends with the lexer again */
    return OK;
}

/*
 * vim_regexec and friends
 */

/*
 * Global work variables for vim_regexec().
 */

/* The current match-position is remembered with these variables: */
static linenr_T	reglnum;	/* line number, relative to first line */
static char_u	*regline;	/* start of current line */
static char_u	*reginput;	/* current input, points into "regline" */

static int	need_clear_subexpr;	/* subexpressions still need to be
					 * cleared */
#ifdef FEAT_SYN_HL
static int	need_clear_zsubexpr = FALSE;	/* extmatch subexpressions
						 * still need to be cleared */
#endif

/*
 * Structure used to save the current input state, when it needs to be
 * restored after trying a match.  Used by reg_save() and reg_restore().
 * Also stores the length of "backpos".
 */
typedef struct
{
    union
    {
	char_u	*ptr;	/* reginput pointer, for single-line regexp */
	lpos_T	pos;	/* reginput pos, for multi-line regexp */
    } rs_u;
    int		rs_len;
} regsave_T;

/* struct to save start/end pointer/position in for \(\) */
typedef struct
{
    union
    {
	char_u	*ptr;
	lpos_T	pos;
    } se_u;
} save_se_T;

static char_u	*reg_getline __ARGS((linenr_T lnum));
static long	vim_regexec_both __ARGS((char_u *line, colnr_T col, proftime_T *tm));
static long	regtry __ARGS((regprog_T *prog, colnr_T col));
static void	cleanup_subexpr __ARGS((void));
#ifdef FEAT_SYN_HL
static void	cleanup_zsubexpr __ARGS((void));
#endif
static void	reg_nextline __ARGS((void));
static void	reg_save __ARGS((regsave_T *save, garray_T *gap));
static void	reg_restore __ARGS((regsave_T *save, garray_T *gap));
static int	reg_save_equal __ARGS((regsave_T *save));
static void	save_se_multi __ARGS((save_se_T *savep, lpos_T *posp));
static void	save_se_one __ARGS((save_se_T *savep, char_u **pp));

/* Save the sub-expressions before attempting a match. */
#define save_se(savep, posp, pp) \
    REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp))

/* After a failed match restore the sub-expressions. */
#define restore_se(savep, posp, pp) { \
    if (REG_MULTI) \
	*(posp) = (savep)->se_u.pos; \
    else \
	*(pp) = (savep)->se_u.ptr; }

static int	re_num_cmp __ARGS((long_u val, char_u *scan));
static int	regmatch __ARGS((char_u *prog));
static int	regrepeat __ARGS((char_u *p, long maxcount));

#ifdef DEBUG
int		regnarrate = 0;
#endif

/*
 * Internal copy of 'ignorecase'.  It is set at each call to vim_regexec().
 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern
 * contains '\c' or '\C' the value is overruled.
 */
static int	ireg_ic;

#ifdef FEAT_MBYTE
/*
 * Similar to ireg_ic, but only for 'combining' characters.  Set with \Z flag
 * in the regexp.  Defaults to false, always.
 */
static int	ireg_icombine;
#endif

/*
 * Copy of "rmm_maxcol": maximum column to search for a match.  Zero when
 * there is no maximum.
 */
static colnr_T	ireg_maxcol;

/*
 * Sometimes need to save a copy of a line.  Since alloc()/free() is very
 * slow, we keep one allocated piece of memory and only re-allocate it when
 * it's too small.  It's freed in vim_regexec_both() when finished.
 */
static char_u	*reg_tofree = NULL;
static unsigned	reg_tofreelen;

/*
 * These variables are set when executing a regexp to speed up the execution.
 * Which ones are set depends on whether a single-line or multi-line match is
 * done:
 *			single-line		multi-line
 * reg_match		&regmatch_T		NULL
 * reg_mmatch		NULL			&regmmatch_T
 * reg_startp		reg_match->startp	<invalid>
 * reg_endp		reg_match->endp		<invalid>
 * reg_startpos		<invalid>		reg_mmatch->startpos
 * reg_endpos		<invalid>		reg_mmatch->endpos
 * reg_win		NULL			window in which to search
 * reg_buf		<invalid>		buffer in which to search
 * reg_firstlnum	<invalid>		first line in which to search
 * reg_maxline		0			last line nr
 * reg_line_lbr		FALSE or TRUE		FALSE
 */
static regmatch_T	*reg_match;
static regmmatch_T	*reg_mmatch;
static char_u		**reg_startp = NULL;
static char_u		**reg_endp = NULL;
static lpos_T		*reg_startpos = NULL;
static lpos_T		*reg_endpos = NULL;
static win_T		*reg_win;
static buf_T		*reg_buf;
static linenr_T		reg_firstlnum;
static linenr_T		reg_maxline;
static int		reg_line_lbr;	    /* "\n" in string is line break */

/* Values for rs_state in regitem_T. */
typedef enum regstate_E
{
    RS_NOPEN = 0	/* NOPEN and NCLOSE */
    , RS_MOPEN		/* MOPEN + [0-9] */
    , RS_MCLOSE		/* MCLOSE + [0-9] */
#ifdef FEAT_SYN_HL
    , RS_ZOPEN		/* ZOPEN + [0-9] */
    , RS_ZCLOSE		/* ZCLOSE + [0-9] */
#endif
    , RS_BRANCH		/* BRANCH */
    , RS_BRCPLX_MORE	/* BRACE_COMPLEX and trying one more match */
    , RS_BRCPLX_LONG	/* BRACE_COMPLEX and trying longest match */
    , RS_BRCPLX_SHORT	/* BRACE_COMPLEX and trying shortest match */
    , RS_NOMATCH	/* NOMATCH */
    , RS_BEHIND1	/* BEHIND / NOBEHIND matching rest */
    , RS_BEHIND2	/* BEHIND / NOBEHIND matching behind part */
    , RS_STAR_LONG	/* STAR/PLUS/BRACE_SIMPLE longest match */
    , RS_STAR_SHORT	/* STAR/PLUS/BRACE_SIMPLE shortest match */
} regstate_T;

/*
 * When there are alternatives a regstate_T is put on the regstack to remember
 * what we are doing.
 * Before it may be another type of item, depending on rs_state, to remember
 * more things.
 */
typedef struct regitem_S
{
    regstate_T	rs_state;	/* what we are doing, one of RS_ above */
    char_u	*rs_scan;	/* current node in program */
    union
    {
	save_se_T  sesave;
	regsave_T  regsave;
    } rs_un;			/* room for saving reginput */
    short	rs_no;		/* submatch nr */
} regitem_T;

static regitem_T *regstack_push __ARGS((regstate_T state, char_u *scan));
static void regstack_pop __ARGS((char_u **scan));

/* used for BEHIND and NOBEHIND matching */
typedef struct regbehind_S
{
    regsave_T	save_after;
    regsave_T	save_behind;
} regbehind_T;

/* used for STAR, PLUS and BRACE_SIMPLE matching */
typedef struct regstar_S
{
    int		nextb;		/* next byte */
    int		nextb_ic;	/* next byte reverse case */
    long	count;
    long	minval;
    long	maxval;
} regstar_T;

/* used to store input position when a BACK was encountered, so that we now if
 * we made any progress since the last time. */
typedef struct backpos_S
{
    char_u	*bp_scan;	/* "scan" where BACK was encountered */
    regsave_T	bp_pos;		/* last input position */
} backpos_T;

/*
 * "regstack" and "backpos" are used by regmatch().  They are kept over calls
 * to avoid invoking malloc() and free() often.
 * "regstack" is a stack with regitem_T items, sometimes preceded by regstar_T
 * or regbehind_T.
 * "backpos_T" is a table with backpos_T for BACK
 */
static garray_T	regstack = {0, 0, 0, 0, NULL};
static garray_T	backpos = {0, 0, 0, 0, NULL};

/*
 * Both for regstack and backpos tables we use the following strategy of
 * allocation (to reduce malloc/free calls):
 * - Initial size is fairly small.
 * - When needed, the tables are grown bigger (8 times at first, double after
 *   that).
 * - After executing the match we free the memory only if the array has grown.
 *   Thus the memory is kept allocated when it's at the initial size.
 * This makes it fast while not keeping a lot of memory allocated.
 * A three times speed increase was observed when using many simple patterns.
 */
#define REGSTACK_INITIAL	2048
#define BACKPOS_INITIAL		64

#if defined(EXITFREE) || defined(PROTO)
    void
free_regexp_stuff()
{
    ga_clear(&regstack);
    ga_clear(&backpos);
    vim_free(reg_tofree);
    vim_free(reg_prev_sub);
}
#endif

/*
 * Get pointer to the line "lnum", which is relative to "reg_firstlnum".
 */
    static char_u *
reg_getline(lnum)
    linenr_T	lnum;
{
    /* when looking behind for a match/no-match lnum is negative.  But we
     * can't go before line 1 */
    if (reg_firstlnum + lnum < 1)
	return NULL;
    if (lnum > reg_maxline)
	/* Must have matched the "\n" in the last line. */
	return (char_u *)"";
    return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE);
}

static regsave_T behind_pos;

#ifdef FEAT_SYN_HL
static char_u	*reg_startzp[NSUBEXP];	/* Workspace to mark beginning */
static char_u	*reg_endzp[NSUBEXP];	/*   and end of \z(...\) matches */
static lpos_T	reg_startzpos[NSUBEXP];	/* idem, beginning pos */
static lpos_T	reg_endzpos[NSUBEXP];	/* idem, end pos */
#endif

/* TRUE if using multi-line regexp. */
#define REG_MULTI	(reg_match == NULL)

/*
 * Match a regexp against a string.
 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp().
 * Uses curbuf for line count and 'iskeyword'.
 *
 * Return TRUE if there is a match, FALSE if not.
 */
    int
vim_regexec(rmp, line, col)
    regmatch_T	*rmp;
    char_u	*line;	/* string to match against */
    colnr_T	col;	/* column to start looking for match */
{
    reg_match = rmp;
    reg_mmatch = NULL;
    reg_maxline = 0;
    reg_line_lbr = FALSE;
    reg_win = NULL;
    ireg_ic = rmp->rm_ic;
#ifdef FEAT_MBYTE
    ireg_icombine = FALSE;
#endif
    ireg_maxcol = 0;
    return (vim_regexec_both(line, col, NULL) != 0);
}

#if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \
	|| defined(FIND_REPLACE_DIALOG) || defined(PROTO)
/*
 * Like vim_regexec(), but consider a "\n" in "line" to be a line break.
 */
    int
vim_regexec_nl(rmp, line, col)
    regmatch_T	*rmp;
    char_u	*line;	/* string to match against */
    colnr_T	col;	/* column to start looking for match */
{
    reg_match = rmp;
    reg_mmatch = NULL;
    reg_maxline = 0;
    reg_line_lbr = TRUE;
    reg_win = NULL;
    ireg_ic = rmp->rm_ic;
#ifdef FEAT_MBYTE
    ireg_icombine = FALSE;
#endif
    ireg_maxcol = 0;
    return (vim_regexec_both(line, col, NULL) != 0);
}
#endif

/*
 * Match a regexp against multiple lines.
 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp().
 * Uses curbuf for line count and 'iskeyword'.
 *
 * Return zero if there is no match.  Return number of lines contained in the
 * match otherwise.
 */
    long
vim_regexec_multi(rmp, win, buf, lnum, col, tm)
    regmmatch_T	*rmp;
    win_T	*win;		/* window in which to search or NULL */
    buf_T	*buf;		/* buffer in which to search */
    linenr_T	lnum;		/* nr of line to start looking for match */
    colnr_T	col;		/* column to start looking for match */
    proftime_T	*tm;		/* timeout limit or NULL */
{
    long	r;
    buf_T	*save_curbuf = curbuf;

    reg_match = NULL;
    reg_mmatch = rmp;
    reg_buf = buf;
    reg_win = win;
    reg_firstlnum = lnum;
    reg_maxline = reg_buf->b_ml.ml_line_count - lnum;
    reg_line_lbr = FALSE;
    ireg_ic = rmp->rmm_ic;
#ifdef FEAT_MBYTE
    ireg_icombine = FALSE;
#endif
    ireg_maxcol = rmp->rmm_maxcol;

    /* Need to switch to buffer "buf" to make vim_iswordc() work. */
    curbuf = buf;
    r = vim_regexec_both(NULL, col, tm);
    curbuf = save_curbuf;

    return r;
}

/*
 * Match a regexp against a string ("line" points to the string) or multiple
 * lines ("line" is NULL, use reg_getline()).
 */
/*ARGSUSED*/
    static long
vim_regexec_both(line, col, tm)
    char_u	*line;
    colnr_T	col;		/* column to start looking for match */
    proftime_T	*tm;		/* timeout limit or NULL */
{
    regprog_T	*prog;
    char_u	*s;
    long	retval = 0L;

    /* Create "regstack" and "backpos" if they are not allocated yet.
     * We allocate *_INITIAL amount of bytes first and then set the grow size
     * to much bigger value to avoid many malloc calls in case of deep regular
     * expressions.  */
    if (regstack.ga_data == NULL)
    {
	/* Use an item size of 1 byte, since we push different things
	 * onto the regstack. */
	ga_init2(&regstack, 1, REGSTACK_INITIAL);
	ga_grow(&regstack, REGSTACK_INITIAL);
	regstack.ga_growsize = REGSTACK_INITIAL * 8;
    }

    if (backpos.ga_data == NULL)
    {
	ga_init2(&backpos, sizeof(backpos_T), BACKPOS_INITIAL);
	ga_grow(&backpos, BACKPOS_INITIAL);
	backpos.ga_growsize = BACKPOS_INITIAL * 8;
    }

    if (REG_MULTI)
    {
	prog = reg_mmatch->regprog;
	line = reg_getline((linenr_T)0);
	reg_startpos = reg_mmatch->startpos;
	reg_endpos = reg_mmatch->endpos;
    }
    else
    {
	prog = reg_match->regprog;
	reg_startp = reg_match->startp;
	reg_endp = reg_match->endp;
    }

    /* Be paranoid... */
    if (prog == NULL || line == NULL)
    {
	EMSG(_(e_null));
	goto theend;
    }

    /* Check validity of program. */
    if (prog_magic_wrong())
	goto theend;

    /* If the start column is past the maximum column: no need to try. */
    if (ireg_maxcol > 0 && col >= ireg_maxcol)
	goto theend;

    /* If pattern contains "\c" or "\C": overrule value of ireg_ic */
    if (prog->regflags & RF_ICASE)
	ireg_ic = TRUE;
    else if (prog->regflags & RF_NOICASE)
	ireg_ic = FALSE;

#ifdef FEAT_MBYTE
    /* If pattern contains "\Z" overrule value of ireg_icombine */
    if (prog->regflags & RF_ICOMBINE)
	ireg_icombine = TRUE;
#endif

    /* If there is a "must appear" string, look for it. */
    if (prog->regmust != NULL)
    {
	int c;

#ifdef FEAT_MBYTE
	if (has_mbyte)
	    c = (*mb_ptr2char)(prog->regmust);
	else
#endif
	    c = *prog->regmust;
	s = line + col;

	/*
	 * This is used very often, esp. for ":global".  Use three versions of
	 * the loop to avoid overhead of conditions.
	 */
	if (!ireg_ic
#ifdef FEAT_MBYTE
		&& !has_mbyte
#endif
		)
	    while ((s = vim_strbyte(s, c)) != NULL)
	    {
		if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0)
		    break;		/* Found it. */
		++s;
	    }
#ifdef FEAT_MBYTE
	else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1))
	    while ((s = vim_strchr(s, c)) != NULL)
	    {
		if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0)
		    break;		/* Found it. */
		mb_ptr_adv(s);
	    }
#endif
	else
	    while ((s = cstrchr(s, c)) != NULL)
	    {
		if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0)
		    break;		/* Found it. */
		mb_ptr_adv(s);
	    }
	if (s == NULL)		/* Not present. */
	    goto theend;
    }

    regline = line;
    reglnum = 0;

    /* Simplest case: Anchored match need be tried only once. */
    if (prog->reganch)
    {
	int	c;

#ifdef FEAT_MBYTE
	if (has_mbyte)
	    c = (*mb_ptr2char)(regline + col);
	else
#endif
	    c = regline[col];
	if (prog->regstart == NUL
		|| prog->regstart == c
		|| (ireg_ic && ((
#ifdef FEAT_MBYTE
			(enc_utf8 && utf_fold(prog->regstart) == utf_fold(c)))
			|| (c < 255 && prog->regstart < 255 &&
#endif
			    MB_TOLOWER(prog->regstart) == MB_TOLOWER(c)))))
	    retval = regtry(prog, col);
	else
	    retval = 0;
    }
    else
    {
#ifdef FEAT_RELTIME
	int tm_count = 0;
#endif
	/* Messy cases:  unanchored match. */
	while (!got_int)
	{
	    if (prog->regstart != NUL)
	    {
		/* Skip until the char we know it must start with.
		 * Used often, do some work to avoid call overhead. */
		if (!ireg_ic
#ifdef FEAT_MBYTE
			    && !has_mbyte
#endif
			    )
		    s = vim_strbyte(regline + col, prog->regstart);
		else
		    s = cstrchr(regline + col, prog->regstart);
		if (s == NULL)
		{
		    retval = 0;
		    break;
		}
		col = (int)(s - regline);
	    }

	    /* Check for maximum column to try. */
	    if (ireg_maxcol > 0 && col >= ireg_maxcol)
	    {
		retval = 0;
		break;
	    }

	    retval = regtry(prog, col);
	    if (retval > 0)
		break;

	    /* if not currently on the first line, get it again */
	    if (reglnum != 0)
	    {
		reglnum = 0;
		regline = reg_getline((linenr_T)0);
	    }
	    if (regline[col] == NUL)
		break;
#ifdef FEAT_MBYTE
	    if (has_mbyte)
		col += (*mb_ptr2len)(regline + col);
	    else
#endif
		++col;
#ifdef FEAT_RELTIME
	    /* Check for timeout once in a twenty times to avoid overhead. */
	    if (tm != NULL && ++tm_count == 20)
	    {
		tm_count = 0;
		if (profile_passed_limit(tm))
		    break;
	    }
#endif
	}
    }

theend:
    /* Free "reg_tofree" when it's a bit big.
     * Free regstack and backpos if they are bigger than their initial size. */
    if (reg_tofreelen > 400)
    {
	vim_free(reg_tofree);
	reg_tofree = NULL;
    }
    if (regstack.ga_maxlen > REGSTACK_INITIAL)
	ga_clear(&regstack);
    if (backpos.ga_maxlen > BACKPOS_INITIAL)
	ga_clear(&backpos);

    return retval;
}

#ifdef FEAT_SYN_HL
static reg_extmatch_T *make_extmatch __ARGS((void));

/*
 * Create a new extmatch and mark it as referenced once.
 */
    static reg_extmatch_T *
make_extmatch()
{
    reg_extmatch_T	*em;

    em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T));
    if (em != NULL)
	em->refcnt = 1;
    return em;
}

/*
 * Add a reference to an extmatch.
 */
    reg_extmatch_T *
ref_extmatch(em)
    reg_extmatch_T	*em;
{
    if (em != NULL)
	em->refcnt++;
    return em;
}

/*
 * Remove a reference to an extmatch.  If there are no references left, free
 * the info.
 */
    void
unref_extmatch(em)
    reg_extmatch_T	*em;
{
    int i;

    if (em != NULL && --em->refcnt <= 0)
    {
	for (i = 0; i < NSUBEXP; ++i)
	    vim_free(em->matches[i]);
	vim_free(em);
    }
}
#endif

/*
 * regtry - try match of "prog" with at regline["col"].
 * Returns 0 for failure, number of lines contained in the match otherwise.
 */
    static long
regtry(prog, col)
    regprog_T	*prog;
    colnr_T	col;
{
    reginput = regline + col;
    need_clear_subexpr = TRUE;
#ifdef FEAT_SYN_HL
    /* Clear the external match subpointers if necessary. */
    if (prog->reghasz == REX_SET)
	need_clear_zsubexpr = TRUE;
#endif

    if (regmatch(prog->program + 1) == 0)
	return 0;

    cleanup_subexpr();
    if (REG_MULTI)
    {
	if (reg_startpos[0].lnum < 0)
	{
	    reg_startpos[0].lnum = 0;
	    reg_startpos[0].col = col;
	}
	if (reg_endpos[0].lnum < 0)
	{
	    reg_endpos[0].lnum = reglnum;
	    reg_endpos[0].col = (int)(reginput - regline);
	}
	else
	    /* Use line number of "\ze". */
	    reglnum = reg_endpos[0].lnum;
    }
    else
    {
	if (reg_startp[0] == NULL)
	    reg_startp[0] = regline + col;
	if (reg_endp[0] == NULL)
	    reg_endp[0] = reginput;
    }
#ifdef FEAT_SYN_HL
    /* Package any found \z(...\) matches for export. Default is none. */
    unref_extmatch(re_extmatch_out);
    re_extmatch_out = NULL;

    if (prog->reghasz == REX_SET)
    {
	int		i;

	cleanup_zsubexpr();
	re_extmatch_out = make_extmatch();
	for (i = 0; i < NSUBEXP; i++)
	{
	    if (REG_MULTI)
	    {
		/* Only accept single line matches. */
		if (reg_startzpos[i].lnum >= 0
			&& reg_endzpos[i].lnum == reg_startzpos[i].lnum)
		    re_extmatch_out->matches[i] =
			vim_strnsave(reg_getline(reg_startzpos[i].lnum)
						       + reg_startzpos[i].col,
				   reg_endzpos[i].col - reg_startzpos[i].col);
	    }
	    else
	    {
		if (reg_startzp[i] != NULL && reg_endzp[i] != NULL)
		    re_extmatch_out->matches[i] =
			    vim_strnsave(reg_startzp[i],
					(int)(reg_endzp[i] - reg_startzp[i]));
	    }
	}
    }
#endif
    return 1 + reglnum;
}

#ifdef FEAT_MBYTE
static int reg_prev_class __ARGS((void));

/*
 * Get class of previous character.
 */
    static int
reg_prev_class()
{
    if (reginput > regline)
	return mb_get_class(reginput - 1
				     - (*mb_head_off)(regline, reginput - 1));
    return -1;
}

#endif
#define ADVANCE_REGINPUT() mb_ptr_adv(reginput)

/*
 * The arguments from BRACE_LIMITS are stored here.  They are actually local
 * to regmatch(), but they are here to reduce the amount of stack space used
 * (it can be called recursively many times).
 */
static long	bl_minval;
static long	bl_maxval;

/*
 * regmatch - main matching routine
 *
 * Conceptually the strategy is simple: Check to see whether the current node
 * matches, push an item onto the regstack and loop to see whether the rest
 * matches, and then act accordingly.  In practice we make some effort to
 * avoid using the regstack, in particular by going through "ordinary" nodes
 * (that don't need to know whether the rest of the match failed) by a nested
 * loop.
 *
 * Returns TRUE when there is a match.  Leaves reginput and reglnum just after
 * the last matched character.
 * Returns FALSE when there is no match.  Leaves reginput and reglnum in an
 * undefined state!
 */
    static int
regmatch(scan)
    char_u	*scan;		/* Current node. */
{
  char_u	*next;		/* Next node. */
  int		op;
  int		c;
  regitem_T	*rp;
  int		no;
  int		status;		/* one of the RA_ values: */
#define RA_FAIL		1	/* something failed, abort */
#define RA_CONT		2	/* continue in inner loop */
#define RA_BREAK	3	/* break inner loop */
#define RA_MATCH	4	/* successful match */
#define RA_NOMATCH	5	/* didn't match */

  /* Make "regstack" and "backpos" empty.  They are allocated and freed in
   * vim_regexec_both() to reduce malloc()/free() calls. */
  regstack.ga_len = 0;
  backpos.ga_len = 0;

  /*
   * Repeat until "regstack" is empty.
   */
  for (;;)
  {
    /* Some patterns my cause a long time to match, even though they are not
     * illegal.  E.g., "\([a-z]\+\)\+Q".  Allow breaking them with CTRL-C. */
    fast_breakcheck();

#ifdef DEBUG
    if (scan != NULL && regnarrate)
    {
	mch_errmsg(regprop(scan));
	mch_errmsg("(\n");
    }
#endif

    /*
     * Repeat for items that can be matched sequentially, without using the
     * regstack.
     */
    for (;;)
    {
	if (got_int || scan == NULL)
	{
	    status = RA_FAIL;
	    break;
	}
	status = RA_CONT;

#ifdef DEBUG
	if (regnarrate)
	{
	    mch_errmsg(regprop(scan));
	    mch_errmsg("...\n");
# ifdef FEAT_SYN_HL
	    if (re_extmatch_in != NULL)
	    {
		int i;

		mch_errmsg(_("External submatches:\n"));
		for (i = 0; i < NSUBEXP; i++)
		{
		    mch_errmsg("    \"");
		    if (re_extmatch_in->matches[i] != NULL)
			mch_errmsg(re_extmatch_in->matches[i]);
		    mch_errmsg("\"\n");
		}
	    }
# endif
	}
#endif
	next = regnext(scan);

	op = OP(scan);
	/* Check for character class with NL added. */
	if (!reg_line_lbr && WITH_NL(op) && REG_MULTI
				&& *reginput == NUL && reglnum <= reg_maxline)
	{
	    reg_nextline();
	}
	else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n')
	{
	    ADVANCE_REGINPUT();
	}
	else
	{
	  if (WITH_NL(op))
	      op -= ADD_NL;
#ifdef FEAT_MBYTE
	  if (has_mbyte)
	      c = (*mb_ptr2char)(reginput);
	  else
#endif
	      c = *reginput;
	  switch (op)
	  {
	  case BOL:
	    if (reginput != regline)
		status = RA_NOMATCH;
	    break;

	  case EOL:
	    if (c != NUL)
		status = RA_NOMATCH;
	    break;

	  case RE_BOF:
	    /* We're not at the beginning of the file when below the first
	     * line where we started, not at the start of the line or we
	     * didn't start at the first line of the buffer. */
	    if (reglnum != 0 || reginput != regline
					  || (REG_MULTI && reg_firstlnum > 1))
		status = RA_NOMATCH;
	    break;

	  case RE_EOF:
	    if (reglnum != reg_maxline || c != NUL)
		status = RA_NOMATCH;
	    break;

	  case CURSOR:
	    /* Check if the buffer is in a window and compare the
	     * reg_win->w_cursor position to the match position. */
	    if (reg_win == NULL
		    || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum)
		    || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col))
		status = RA_NOMATCH;
	    break;

	  case RE_MARK:
	    /* Compare the mark position to the match position.  NOTE: Always
	     * uses the current buffer. */
	    {
		int	mark = OPERAND(scan)[0];
		int	cmp = OPERAND(scan)[1];
		pos_T	*pos;

		pos = getmark(mark, FALSE);
		if (pos == NULL		     /* mark doesn't exist */
			|| pos->lnum <= 0    /* mark isn't set (in curbuf) */
			|| (pos->lnum == reglnum + reg_firstlnum
				? (pos->col == (colnr_T)(reginput - regline)
				    ? (cmp == '<' || cmp == '>')
				    : (pos->col < (colnr_T)(reginput - regline)
					? cmp != '>'
					: cmp != '<'))
				: (pos->lnum < reglnum + reg_firstlnum
				    ? cmp != '>'
				    : cmp != '<')))
		    status = RA_NOMATCH;
	    }
	    break;

	  case RE_VISUAL:
#ifdef FEAT_VISUAL
	    /* Check if the buffer is the current buffer. and whether the
	     * position is inside the Visual area. */
	    if (reg_buf != curbuf || VIsual.lnum == 0)
		status = RA_NOMATCH;
	    else
	    {
		pos_T	    top, bot;
		linenr_T    lnum;
		colnr_T	    col;
		win_T	    *wp = reg_win == NULL ? curwin : reg_win;
		int	    mode;

		if (VIsual_active)
		{
		    if (lt(VIsual, wp->w_cursor))
		    {
			top = VIsual;
			bot = wp->w_cursor;
		    }
		    else
		    {
			top = wp->w_cursor;
			bot = VIsual;
		    }
		    mode = VIsual_mode;
		}
		else
		{
		    if (lt(curbuf->b_visual.vi_start, curbuf->b_visual.vi_end))
		    {
			top = curbuf->b_visual.vi_start;
			bot = curbuf->b_visual.vi_end;
		    }
		    else
		    {
			top = curbuf->b_visual.vi_end;
			bot = curbuf->b_visual.vi_start;
		    }
		    mode = curbuf->b_visual.vi_mode;
		}
		lnum = reglnum + reg_firstlnum;
		col = (colnr_T)(reginput - regline);
		if (lnum < top.lnum || lnum > bot.lnum)
		    status = RA_NOMATCH;
		else if (mode == 'v')
		{
		    if ((lnum == top.lnum && col < top.col)
			    || (lnum == bot.lnum
					 && col >= bot.col + (*p_sel != 'e')))
			status = RA_NOMATCH;
		}
		else if (mode == Ctrl_V)
		{
		    colnr_T	    start, end;
		    colnr_T	    start2, end2;
		    colnr_T	    cols;

		    getvvcol(wp, &top, &start, NULL, &end);
		    getvvcol(wp, &bot, &start2, NULL, &end2);
		    if (start2 < start)
			start = start2;
		    if (end2 > end)
			end = end2;
		    if (top.col == MAXCOL || bot.col == MAXCOL)
			end = MAXCOL;
		    cols = win_linetabsize(wp,
				      regline, (colnr_T)(reginput - regline));
		    if (cols < start || cols > end - (*p_sel == 'e'))
			status = RA_NOMATCH;
		}
	    }
#else
	    status = RA_NOMATCH;
#endif
	    break;

	  case RE_LNUM:
	    if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum),
									scan))
		status = RA_NOMATCH;
	    break;

	  case RE_COL:
	    if (!re_num_cmp((long_u)(reginput - regline) + 1, scan))
		status = RA_NOMATCH;
	    break;

	  case RE_VCOL:
	    if (!re_num_cmp((long_u)win_linetabsize(
			    reg_win == NULL ? curwin : reg_win,
			    regline, (colnr_T)(reginput - regline)) + 1, scan))
		status = RA_NOMATCH;
	    break;

	  case BOW:	/* \<word; reginput points to w */
	    if (c == NUL)	/* Can't match at end of line */
		status = RA_NOMATCH;
#ifdef FEAT_MBYTE
	    else if (has_mbyte)
	    {
		int this_class;

		/* Get class of current and previous char (if it exists). */
		this_class = mb_get_class(reginput);
		if (this_class <= 1)
		    status = RA_NOMATCH;  /* not on a word at all */
		else if (reg_prev_class() == this_class)
		    status = RA_NOMATCH;  /* previous char is in same word */
	    }
#endif
	    else
	    {
		if (!vim_iswordc(c)
			|| (reginput > regline && vim_iswordc(reginput[-1])))
		    status = RA_NOMATCH;
	    }
	    break;

	  case EOW:	/* word\>; reginput points after d */
	    if (reginput == regline)    /* Can't match at start of line */
		status = RA_NOMATCH;
#ifdef FEAT_MBYTE
	    else if (has_mbyte)
	    {
		int this_class, prev_class;

		/* Get class of current and previous char (if it exists). */
		this_class = mb_get_class(reginput);
		prev_class = reg_prev_class();
		if (this_class == prev_class
			|| prev_class == 0 || prev_class == 1)
		    status = RA_NOMATCH;
	    }
#endif
	    else
	    {
		if (!vim_iswordc(reginput[-1])
			|| (reginput[0] != NUL && vim_iswordc(c)))
		    status = RA_NOMATCH;
	    }
	    break; /* Matched with EOW */

	  case ANY:
	    if (c == NUL)
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case IDENT:
	    if (!vim_isIDc(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case SIDENT:
	    if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case KWORD:
	    if (!vim_iswordp(reginput))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case SKWORD:
	    if (VIM_ISDIGIT(*reginput) || !vim_iswordp(reginput))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case FNAME:
	    if (!vim_isfilec(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case SFNAME:
	    if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case PRINT:
	    if (ptr2cells(reginput) != 1)
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case SPRINT:
	    if (VIM_ISDIGIT(*reginput) || ptr2cells(reginput) != 1)
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case WHITE:
	    if (!vim_iswhite(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case NWHITE:
	    if (c == NUL || vim_iswhite(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case DIGIT:
	    if (!ri_digit(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case NDIGIT:
	    if (c == NUL || ri_digit(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case HEX:
	    if (!ri_hex(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case NHEX:
	    if (c == NUL || ri_hex(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case OCTAL:
	    if (!ri_octal(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case NOCTAL:
	    if (c == NUL || ri_octal(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case WORD:
	    if (!ri_word(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case NWORD:
	    if (c == NUL || ri_word(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case HEAD:
	    if (!ri_head(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case NHEAD:
	    if (c == NUL || ri_head(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case ALPHA:
	    if (!ri_alpha(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case NALPHA:
	    if (c == NUL || ri_alpha(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case LOWER:
	    if (!ri_lower(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case NLOWER:
	    if (c == NUL || ri_lower(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case UPPER:
	    if (!ri_upper(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case NUPPER:
	    if (c == NUL || ri_upper(c))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

	  case EXACTLY:
	    {
		int	len;
		char_u	*opnd;

		opnd = OPERAND(scan);
		/* Inline the first byte, for speed. */
		if (*opnd != *reginput
			&& (!ireg_ic || (
#ifdef FEAT_MBYTE
			    !enc_utf8 &&
#endif
			    MB_TOLOWER(*opnd) != MB_TOLOWER(*reginput))))
		    status = RA_NOMATCH;
		else if (*opnd == NUL)
		{
		    /* match empty string always works; happens when "~" is
		     * empty. */
		}
		else if (opnd[1] == NUL
#ifdef FEAT_MBYTE
			    && !(enc_utf8 && ireg_ic)
#endif
			)
		    ++reginput;		/* matched a single char */
		else
		{
		    len = (int)STRLEN(opnd);
		    /* Need to match first byte again for multi-byte. */
		    if (cstrncmp(opnd, reginput, &len) != 0)
			status = RA_NOMATCH;
#ifdef FEAT_MBYTE
		    /* Check for following composing character. */
		    else if (enc_utf8
			       && UTF_COMPOSINGLIKE(reginput, reginput + len))
		    {
			/* raaron: This code makes a composing character get
			 * ignored, which is the correct behavior (sometimes)
			 * for voweled Hebrew texts. */
			if (!ireg_icombine)
			    status = RA_NOMATCH;
		    }
#endif
		    else
			reginput += len;
		}
	    }
	    break;

	  case ANYOF:
	  case ANYBUT:
	    if (c == NUL)
		status = RA_NOMATCH;
	    else if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF))
		status = RA_NOMATCH;
	    else
		ADVANCE_REGINPUT();
	    break;

#ifdef FEAT_MBYTE
	  case MULTIBYTECODE:
	    if (has_mbyte)
	    {
		int	i, len;
		char_u	*opnd;
		int	opndc = 0, inpc;

		opnd = OPERAND(scan);
		/* Safety check (just in case 'encoding' was changed since
		 * compiling the program). */
		if ((len = (*mb_ptr2len)(opnd)) < 2)
		{
		    status = RA_NOMATCH;
		    break;
		}
		if (enc_utf8)
		    opndc = mb_ptr2char(opnd);
		if (enc_utf8 && utf_iscomposing(opndc))
		{
		    /* When only a composing char is given match at any
		     * position where that composing char appears. */
		    status = RA_NOMATCH;
		    for (i = 0; reginput[i] != NUL; i += utf_char2len(inpc))
		    {
			inpc = mb_ptr2char(reginput + i);
			if (!utf_iscomposing(inpc))
			{
			    if (i > 0)
				break;
			}
			else if (opndc == inpc)
			{
			    /* Include all following composing chars. */
			    len = i + mb_ptr2len(reginput + i);
			    status = RA_MATCH;
			    break;
			}
		    }
		}
		else
		    for (i = 0; i < len; ++i)
			if (opnd[i] != reginput[i])
			{
			    status = RA_NOMATCH;
			    break;
			}
		reginput += len;
	    }
	    else
		status = RA_NOMATCH;
	    break;
#endif

	  case NOTHING:
	    break;

	  case BACK:
	    {
		int		i;
		backpos_T	*bp;

		/*
		 * When we run into BACK we need to check if we don't keep
		 * looping without matching any input.  The second and later
		 * times a BACK is encountered it fails if the input is still
		 * at the same position as the previous time.
		 * The positions are stored in "backpos" and found by the
		 * current value of "scan", the position in the RE program.
		 */
		bp = (backpos_T *)backpos.ga_data;
		for (i = 0; i < backpos.ga_len; ++i)
		    if (bp[i].bp_scan == scan)
			break;
		if (i == backpos.ga_len)
		{
		    /* First time at this BACK, make room to store the pos. */
		    if (ga_grow(&backpos, 1) == FAIL)
			status = RA_FAIL;
		    else
		    {
			/* get "ga_data" again, it may have changed */
			bp = (backpos_T *)backpos.ga_data;
			bp[i].bp_scan = scan;
			++backpos.ga_len;
		    }
		}
		else if (reg_save_equal(&bp[i].bp_pos))
		    /* Still at same position as last time, fail. */
		    status = RA_NOMATCH;

		if (status != RA_FAIL && status != RA_NOMATCH)
		    reg_save(&bp[i].bp_pos, &backpos);
	    }
	    break;

	  case MOPEN + 0:   /* Match start: \zs */
	  case MOPEN + 1:   /* \( */
	  case MOPEN + 2:
	  case MOPEN + 3:
	  case MOPEN + 4:
	  case MOPEN + 5:
	  case MOPEN + 6:
	  case MOPEN + 7:
	  case MOPEN + 8:
	  case MOPEN + 9:
	    {
		no = op - MOPEN;
		cleanup_subexpr();
		rp = regstack_push(RS_MOPEN, scan);
		if (rp == NULL)
		    status = RA_FAIL;
		else
		{
		    rp->rs_no = no;
		    save_se(&rp->rs_un.sesave, &reg_startpos[no],
							     &reg_startp[no]);
		    /* We simply continue and handle the result when done. */
		}
	    }
	    break;

	  case NOPEN:	    /* \%( */
	  case NCLOSE:	    /* \) after \%( */
		if (regstack_push(RS_NOPEN, scan) == NULL)
		    status = RA_FAIL;
		/* We simply continue and handle the result when done. */
		break;

#ifdef FEAT_SYN_HL
	  case ZOPEN + 1:
	  case ZOPEN + 2:
	  case ZOPEN + 3:
	  case ZOPEN + 4:
	  case ZOPEN + 5:
	  case ZOPEN + 6:
	  case ZOPEN + 7:
	  case ZOPEN + 8:
	  case ZOPEN + 9:
	    {
		no = op - ZOPEN;
		cleanup_zsubexpr();
		rp = regstack_push(RS_ZOPEN, scan);
		if (rp == NULL)
		    status = RA_FAIL;
		else
		{
		    rp->rs_no = no;
		    save_se(&rp->rs_un.sesave, &reg_startzpos[no],
							     &reg_startzp[no]);
		    /* We simply continue and handle the result when done. */
		}
	    }
	    break;
#endif

	  case MCLOSE + 0:  /* Match end: \ze */
	  case MCLOSE + 1:  /* \) */
	  case MCLOSE + 2:
	  case MCLOSE + 3:
	  case MCLOSE + 4:
	  case MCLOSE + 5:
	  case MCLOSE + 6:
	  case MCLOSE + 7:
	  case MCLOSE + 8:
	  case MCLOSE + 9:
	    {
		no = op - MCLOSE;
		cleanup_subexpr();
		rp = regstack_push(RS_MCLOSE, scan);
		if (rp == NULL)
		    status = RA_FAIL;
		else
		{
		    rp->rs_no = no;
		    save_se(&rp->rs_un.sesave, &reg_endpos[no], &reg_endp[no]);
		    /* We simply continue and handle the result when done. */
		}
	    }
	    break;

#ifdef FEAT_SYN_HL
	  case ZCLOSE + 1:  /* \) after \z( */
	  case ZCLOSE + 2:
	  case ZCLOSE + 3:
	  case ZCLOSE + 4:
	  case ZCLOSE + 5:
	  case ZCLOSE + 6:
	  case ZCLOSE + 7:
	  case ZCLOSE + 8:
	  case ZCLOSE + 9:
	    {
		no = op - ZCLOSE;
		cleanup_zsubexpr();
		rp = regstack_push(RS_ZCLOSE, scan);
		if (rp == NULL)
		    status = RA_FAIL;
		else
		{
		    rp->rs_no = no;
		    save_se(&rp->rs_un.sesave, &reg_endzpos[no],
							      &reg_endzp[no]);
		    /* We simply continue and handle the result when done. */
		}
	    }
	    break;
#endif

	  case BACKREF + 1:
	  case BACKREF + 2:
	  case BACKREF + 3:
	  case BACKREF + 4:
	  case BACKREF + 5:
	  case BACKREF + 6:
	  case BACKREF + 7:
	  case BACKREF + 8:
	  case BACKREF + 9:
	    {
		int		len;
		linenr_T	clnum;
		colnr_T		ccol;
		char_u		*p;

		no = op - BACKREF;
		cleanup_subexpr();
		if (!REG_MULTI)		/* Single-line regexp */
		{
		    if (reg_endp[no] == NULL)
		    {
			/* Backref was not set: Match an empty string. */
			len = 0;
		    }
		    else
		    {
			/* Compare current input with back-ref in the same
			 * line. */
			len = (int)(reg_endp[no] - reg_startp[no]);
			if (cstrncmp(reg_startp[no], reginput, &len) != 0)
			    status = RA_NOMATCH;
		    }
		}
		else				/* Multi-line regexp */
		{
		    if (reg_endpos[no].lnum < 0)
		    {
			/* Backref was not set: Match an empty string. */
			len = 0;
		    }
		    else
		    {
			if (reg_startpos[no].lnum == reglnum
				&& reg_endpos[no].lnum == reglnum)
			{
			    /* Compare back-ref within the current line. */
			    len = reg_endpos[no].col - reg_startpos[no].col;
			    if (cstrncmp(regline + reg_startpos[no].col,
							  reginput, &len) != 0)
				status = RA_NOMATCH;
			}
			else
			{
			    /* Messy situation: Need to compare between two
			     * lines. */
			    ccol = reg_startpos[no].col;
			    clnum = reg_startpos[no].lnum;
			    for (;;)
			    {
				/* Since getting one line may invalidate
				 * the other, need to make copy.  Slow! */
				if (regline != reg_tofree)
				{
				    len = (int)STRLEN(regline);
				    if (reg_tofree == NULL
						 || len >= (int)reg_tofreelen)
				    {
					len += 50;	/* get some extra */
					vim_free(reg_tofree);
					reg_tofree = alloc(len);
					if (reg_tofree == NULL)
					{
					    status = RA_FAIL; /* outof memory!*/
					    break;
					}
					reg_tofreelen = len;
				    }
				    STRCPY(reg_tofree, regline);
				    reginput = reg_tofree
						       + (reginput - regline);
				    regline = reg_tofree;
				}

				/* Get the line to compare with. */
				p = reg_getline(clnum);
				if (clnum == reg_endpos[no].lnum)
				    len = reg_endpos[no].col - ccol;
				else
				    len = (int)STRLEN(p + ccol);

				if (cstrncmp(p + ccol, reginput, &len) != 0)
				{
				    status = RA_NOMATCH;  /* doesn't match */
				    break;
				}
				if (clnum == reg_endpos[no].lnum)
				    break;		/* match and at end! */
				if (reglnum >= reg_maxline)
				{
				    status = RA_NOMATCH;  /* text too short */
				    break;
				}

				/* Advance to next line. */
				reg_nextline();
				++clnum;
				ccol = 0;
				if (got_int)
				{
				    status = RA_FAIL;
				    break;
				}
			    }

			    /* found a match!  Note that regline may now point
			     * to a copy of the line, that should not matter. */
			}
		    }
		}

		/* Matched the backref, skip over it. */
		reginput += len;
	    }
	    break;

#ifdef FEAT_SYN_HL
	  case ZREF + 1:
	  case ZREF + 2:
	  case ZREF + 3:
	  case ZREF + 4:
	  case ZREF + 5:
	  case ZREF + 6:
	  case ZREF + 7:
	  case ZREF + 8:
	  case ZREF + 9:
	    {
		int	len;

		cleanup_zsubexpr();
		no = op - ZREF;
		if (re_extmatch_in != NULL
			&& re_extmatch_in->matches[no] != NULL)
		{
		    len = (int)STRLEN(re_extmatch_in->matches[no]);
		    if (cstrncmp(re_extmatch_in->matches[no],
							  reginput, &len) != 0)
			status = RA_NOMATCH;
		    else
			reginput += len;
		}
		else
		{
		    /* Backref was not set: Match an empty string. */
		}
	    }
	    break;
#endif

	  case BRANCH:
	    {
		if (OP(next) != BRANCH) /* No choice. */
		    next = OPERAND(scan);	/* Avoid recursion. */
		else
		{
		    rp = regstack_push(RS_BRANCH, scan);
		    if (rp == NULL)
			status = RA_FAIL;
		    else
			status = RA_BREAK;	/* rest is below */
		}
	    }
	    break;

	  case BRACE_LIMITS:
	    {
		if (OP(next) == BRACE_SIMPLE)
		{
		    bl_minval = OPERAND_MIN(scan);
		    bl_maxval = OPERAND_MAX(scan);
		}
		else if (OP(next) >= BRACE_COMPLEX
			&& OP(next) < BRACE_COMPLEX + 10)
		{
		    no = OP(next) - BRACE_COMPLEX;
		    brace_min[no] = OPERAND_MIN(scan);
		    brace_max[no] = OPERAND_MAX(scan);
		    brace_count[no] = 0;
		}
		else
		{
		    EMSG(_(e_internal));	    /* Shouldn't happen */
		    status = RA_FAIL;
		}
	    }
	    break;

	  case BRACE_COMPLEX + 0:
	  case BRACE_COMPLEX + 1:
	  case BRACE_COMPLEX + 2:
	  case BRACE_COMPLEX + 3:
	  case BRACE_COMPLEX + 4:
	  case BRACE_COMPLEX + 5:
	  case BRACE_COMPLEX + 6:
	  case BRACE_COMPLEX + 7:
	  case BRACE_COMPLEX + 8:
	  case BRACE_COMPLEX + 9:
	    {
		no = op - BRACE_COMPLEX;
		++brace_count[no];

		/* If not matched enough times yet, try one more */
		if (brace_count[no] <= (brace_min[no] <= brace_max[no]
					     ? brace_min[no] : brace_max[no]))
		{
		    rp = regstack_push(RS_BRCPLX_MORE, scan);
		    if (rp == NULL)
			status = RA_FAIL;
		    else
		    {
			rp->rs_no = no;
			reg_save(&rp->rs_un.regsave, &backpos);
			next = OPERAND(scan);
			/* We continue and handle the result when done. */
		    }
		    break;
		}

		/* If matched enough times, may try matching some more */
		if (brace_min[no] <= brace_max[no])
		{
		    /* Range is the normal way around, use longest match */
		    if (brace_count[no] <= brace_max[no])
		    {
			rp = regstack_push(RS_BRCPLX_LONG, scan);
			if (rp == NULL)
			    status = RA_FAIL;
			else
			{
			    rp->rs_no = no;
			    reg_save(&rp->rs_un.regsave, &backpos);
			    next = OPERAND(scan);
			    /* We continue and handle the result when done. */
			}
		    }
		}
		else
		{
		    /* Range is backwards, use shortest match first */
		    if (brace_count[no] <= brace_min[no])
		    {
			rp = regstack_push(RS_BRCPLX_SHORT, scan);
			if (rp == NULL)
			    status = RA_FAIL;
			else
			{
			    reg_save(&rp->rs_un.regsave, &backpos);
			    /* We continue and handle the result when done. */
			}
		    }
		}
	    }
	    break;

	  case BRACE_SIMPLE:
	  case STAR:
	  case PLUS:
	    {
		regstar_T	rst;

		/*
		 * Lookahead to avoid useless match attempts when we know
		 * what character comes next.
		 */
		if (OP(next) == EXACTLY)
		{
		    rst.nextb = *OPERAND(next);
		    if (ireg_ic)
		    {
			if (MB_ISUPPER(rst.nextb))
			    rst.nextb_ic = MB_TOLOWER(rst.nextb);
			else
			    rst.nextb_ic = MB_TOUPPER(rst.nextb);
		    }
		    else
			rst.nextb_ic = rst.nextb;
		}
		else
		{
		    rst.nextb = NUL;
		    rst.nextb_ic = NUL;
		}
		if (op != BRACE_SIMPLE)
		{
		    rst.minval = (op == STAR) ? 0 : 1;
		    rst.maxval = MAX_LIMIT;
		}
		else
		{
		    rst.minval = bl_minval;
		    rst.maxval = bl_maxval;
		}

		/*
		 * When maxval > minval, try matching as much as possible, up
		 * to maxval.  When maxval < minval, try matching at least the
		 * minimal number (since the range is backwards, that's also
		 * maxval!).
		 */
		rst.count = regrepeat(OPERAND(scan), rst.maxval);
		if (got_int)
		{
		    status = RA_FAIL;
		    break;
		}
		if (rst.minval <= rst.maxval
			  ? rst.count >= rst.minval : rst.count >= rst.maxval)
		{
		    /* It could match.  Prepare for trying to match what
		     * follows.  The code is below.  Parameters are stored in
		     * a regstar_T on the regstack. */
		    if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp)
		    {
			EMSG(_(e_maxmempat));
			status = RA_FAIL;
		    }
		    else if (ga_grow(&regstack, sizeof(regstar_T)) == FAIL)
			status = RA_FAIL;
		    else
		    {
			regstack.ga_len += sizeof(regstar_T);
			rp = regstack_push(rst.minval <= rst.maxval
					? RS_STAR_LONG : RS_STAR_SHORT, scan);
			if (rp == NULL)
			    status = RA_FAIL;
			else
			{
			    *(((regstar_T *)rp) - 1) = rst;
			    status = RA_BREAK;	    /* skip the restore bits */
			}
		    }
		}
		else
		    status = RA_NOMATCH;

	    }
	    break;

	  case NOMATCH:
	  case MATCH:
	  case SUBPAT:
	    rp = regstack_push(RS_NOMATCH, scan);
	    if (rp == NULL)
		status = RA_FAIL;
	    else
	    {
		rp->rs_no = op;
		reg_save(&rp->rs_un.regsave, &backpos);
		next = OPERAND(scan);
		/* We continue and handle the result when done. */
	    }
	    break;

	  case BEHIND:
	  case NOBEHIND:
	    /* Need a bit of room to store extra positions. */
	    if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp)
	    {
		EMSG(_(e_maxmempat));
		status = RA_FAIL;
	    }
	    else if (ga_grow(&regstack, sizeof(regbehind_T)) == FAIL)
		status = RA_FAIL;
	    else
	    {
		regstack.ga_len += sizeof(regbehind_T);
		rp = regstack_push(RS_BEHIND1, scan);
		if (rp == NULL)
		    status = RA_FAIL;
		else
		{
		    rp->rs_no = op;
		    reg_save(&rp->rs_un.regsave, &backpos);
		    /* First try if what follows matches.  If it does then we
		     * check the behind match by looping. */
		}
	    }
	    break;

	  case BHPOS:
	    if (REG_MULTI)
	    {
		if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline)
			|| behind_pos.rs_u.pos.lnum != reglnum)
		    status = RA_NOMATCH;
	    }
	    else if (behind_pos.rs_u.ptr != reginput)
		status = RA_NOMATCH;
	    break;

	  case NEWL:
	    if ((c != NUL || !REG_MULTI || reglnum > reg_maxline
			     || reg_line_lbr) && (c != '\n' || !reg_line_lbr))
		status = RA_NOMATCH;
	    else if (reg_line_lbr)
		ADVANCE_REGINPUT();
	    else
		reg_nextline();
	    break;

	  case END:
	    status = RA_MATCH;	/* Success! */
	    break;

	  default:
	    EMSG(_(e_re_corr));
#ifdef DEBUG
	    printf("Illegal op code %d\n", op);
#endif
	    status = RA_FAIL;
	    break;
	  }
	}

	/* If we can't continue sequentially, break the inner loop. */
	if (status != RA_CONT)
	    break;

	/* Continue in inner loop, advance to next item. */
	scan = next;

    } /* end of inner loop */

    /*
     * If there is something on the regstack execute the code for the state.
     * If the state is popped then loop and use the older state.
     */
    while (regstack.ga_len > 0 && status != RA_FAIL)
    {
	rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1;
	switch (rp->rs_state)
	{
	  case RS_NOPEN:
	    /* Result is passed on as-is, simply pop the state. */
	    regstack_pop(&scan);
	    break;

	  case RS_MOPEN:
	    /* Pop the state.  Restore pointers when there is no match. */
	    if (status == RA_NOMATCH)
		restore_se(&rp->rs_un.sesave, &reg_startpos[rp->rs_no],
						  &reg_startp[rp->rs_no]);
	    regstack_pop(&scan);
	    break;

#ifdef FEAT_SYN_HL
	  case RS_ZOPEN:
	    /* Pop the state.  Restore pointers when there is no match. */
	    if (status == RA_NOMATCH)
		restore_se(&rp->rs_un.sesave, &reg_startzpos[rp->rs_no],
						 &reg_startzp[rp->rs_no]);
	    regstack_pop(&scan);
	    break;
#endif

	  case RS_MCLOSE:
	    /* Pop the state.  Restore pointers when there is no match. */
	    if (status == RA_NOMATCH)
		restore_se(&rp->rs_un.sesave, &reg_endpos[rp->rs_no],
						    &reg_endp[rp->rs_no]);
	    regstack_pop(&scan);
	    break;

#ifdef FEAT_SYN_HL
	  case RS_ZCLOSE:
	    /* Pop the state.  Restore pointers when there is no match. */
	    if (status == RA_NOMATCH)
		restore_se(&rp->rs_un.sesave, &reg_endzpos[rp->rs_no],
						   &reg_endzp[rp->rs_no]);
	    regstack_pop(&scan);
	    break;
#endif

	  case RS_BRANCH:
	    if (status == RA_MATCH)
		/* this branch matched, use it */
		regstack_pop(&scan);
	    else
	    {
		if (status != RA_BREAK)
		{
		    /* After a non-matching branch: try next one. */
		    reg_restore(&rp->rs_un.regsave, &backpos);
		    scan = rp->rs_scan;
		}
		if (scan == NULL || OP(scan) != BRANCH)
		{
		    /* no more branches, didn't find a match */
		    status = RA_NOMATCH;
		    regstack_pop(&scan);
		}
		else
		{
		    /* Prepare to try a branch. */
		    rp->rs_scan = regnext(scan);
		    reg_save(&rp->rs_un.regsave, &backpos);
		    scan = OPERAND(scan);
		}
	    }
	    break;

	  case RS_BRCPLX_MORE:
	    /* Pop the state.  Restore pointers when there is no match. */
	    if (status == RA_NOMATCH)
	    {
		reg_restore(&rp->rs_un.regsave, &backpos);
		--brace_count[rp->rs_no];	/* decrement match count */
	    }
	    regstack_pop(&scan);
	    break;

	  case RS_BRCPLX_LONG:
	    /* Pop the state.  Restore pointers when there is no match. */
	    if (status == RA_NOMATCH)
	    {
		/* There was no match, but we did find enough matches. */
		reg_restore(&rp->rs_un.regsave, &backpos);
		--brace_count[rp->rs_no];
		/* continue with the items after "\{}" */
		status = RA_CONT;
	    }
	    regstack_pop(&scan);
	    if (status == RA_CONT)
		scan = regnext(scan);
	    break;

	  case RS_BRCPLX_SHORT:
	    /* Pop the state.  Restore pointers when there is no match. */
	    if (status == RA_NOMATCH)
		/* There was no match, try to match one more item. */
		reg_restore(&rp->rs_un.regsave, &backpos);
	    regstack_pop(&scan);
	    if (status == RA_NOMATCH)
	    {
		scan = OPERAND(scan);
		status = RA_CONT;
	    }
	    break;

	  case RS_NOMATCH:
	    /* Pop the state.  If the operand matches for NOMATCH or
	     * doesn't match for MATCH/SUBPAT, we fail.  Otherwise backup,
	     * except for SUBPAT, and continue with the next item. */
	    if (status == (rp->rs_no == NOMATCH ? RA_MATCH : RA_NOMATCH))
		status = RA_NOMATCH;
	    else
	    {
		status = RA_CONT;
		if (rp->rs_no != SUBPAT)	/* zero-width */
		    reg_restore(&rp->rs_un.regsave, &backpos);
	    }
	    regstack_pop(&scan);
	    if (status == RA_CONT)
		scan = regnext(scan);
	    break;

	  case RS_BEHIND1:
	    if (status == RA_NOMATCH)
	    {
		regstack_pop(&scan);
		regstack.ga_len -= sizeof(regbehind_T);
	    }
	    else
	    {
		/* The stuff after BEHIND/NOBEHIND matches.  Now try if
		 * the behind part does (not) match before the current
		 * position in the input.  This must be done at every
		 * position in the input and checking if the match ends at
		 * the current position. */

		/* save the position after the found match for next */
		reg_save(&(((regbehind_T *)rp) - 1)->save_after, &backpos);

		/* start looking for a match with operand at the current
		 * position.  Go back one character until we find the
		 * result, hitting the start of the line or the previous
		 * line (for multi-line matching).
		 * Set behind_pos to where the match should end, BHPOS
		 * will match it.  Save the current value. */
		(((regbehind_T *)rp) - 1)->save_behind = behind_pos;
		behind_pos = rp->rs_un.regsave;

		rp->rs_state = RS_BEHIND2;

		reg_restore(&rp->rs_un.regsave, &backpos);
		scan = OPERAND(rp->rs_scan);
	    }
	    break;

	  case RS_BEHIND2:
	    /*
	     * Looping for BEHIND / NOBEHIND match.
	     */
	    if (status == RA_MATCH && reg_save_equal(&behind_pos))
	    {
		/* found a match that ends where "next" started */
		behind_pos = (((regbehind_T *)rp) - 1)->save_behind;
		if (rp->rs_no == BEHIND)
		    reg_restore(&(((regbehind_T *)rp) - 1)->save_after,
								    &backpos);
		else
		    /* But we didn't want a match. */
		    status = RA_NOMATCH;
		regstack_pop(&scan);
		regstack.ga_len -= sizeof(regbehind_T);
	    }
	    else
	    {
		/* No match: Go back one character.  May go to previous
		 * line once. */
		no = OK;
		if (REG_MULTI)
		{
		    if (rp->rs_un.regsave.rs_u.pos.col == 0)
		    {
			if (rp->rs_un.regsave.rs_u.pos.lnum
					< behind_pos.rs_u.pos.lnum
				|| reg_getline(
					--rp->rs_un.regsave.rs_u.pos.lnum)
								  == NULL)
			    no = FAIL;
			else
			{
			    reg_restore(&rp->rs_un.regsave, &backpos);
			    rp->rs_un.regsave.rs_u.pos.col =
						 (colnr_T)STRLEN(regline);
			}
		    }
		    else
			--rp->rs_un.regsave.rs_u.pos.col;
		}
		else
		{
		    if (rp->rs_un.regsave.rs_u.ptr == regline)
			no = FAIL;
		    else
			--rp->rs_un.regsave.rs_u.ptr;
		}
		if (no == OK)
		{
		    /* Advanced, prepare for finding match again. */
		    reg_restore(&rp->rs_un.regsave, &backpos);
		    scan = OPERAND(rp->rs_scan);
		}
		else
		{
		    /* Can't advance.  For NOBEHIND that's a match. */
		    behind_pos = (((regbehind_T *)rp) - 1)->save_behind;
		    if (rp->rs_no == NOBEHIND)
		    {
			reg_restore(&(((regbehind_T *)rp) - 1)->save_after,
								    &backpos);
			status = RA_MATCH;
		    }
		    else
			status = RA_NOMATCH;
		    regstack_pop(&scan);
		    regstack.ga_len -= sizeof(regbehind_T);
		}
	    }
	    break;

	  case RS_STAR_LONG:
	  case RS_STAR_SHORT:
	    {
		regstar_T	    *rst = ((regstar_T *)rp) - 1;

		if (status == RA_MATCH)
		{
		    regstack_pop(&scan);
		    regstack.ga_len -= sizeof(regstar_T);
		    break;
		}

		/* Tried once already, restore input pointers. */
		if (status != RA_BREAK)
		    reg_restore(&rp->rs_un.regsave, &backpos);

		/* Repeat until we found a position where it could match. */
		for (;;)
		{
		    if (status != RA_BREAK)
		    {
			/* Tried first position already, advance. */
			if (rp->rs_state == RS_STAR_LONG)
			{
			    /* Trying for longest match, but couldn't or
			     * didn't match -- back up one char. */
			    if (--rst->count < rst->minval)
				break;
			    if (reginput == regline)
			    {
				/* backup to last char of previous line */
				--reglnum;
				regline = reg_getline(reglnum);
				/* Just in case regrepeat() didn't count
				 * right. */
				if (regline == NULL)
				    break;
				reginput = regline + STRLEN(regline);
				fast_breakcheck();
			    }
			    else
				mb_ptr_back(regline, reginput);
			}
			else
			{
			    /* Range is backwards, use shortest match first.
			     * Careful: maxval and minval are exchanged!
			     * Couldn't or didn't match: try advancing one
			     * char. */
			    if (rst->count == rst->minval
				  || regrepeat(OPERAND(rp->rs_scan), 1L) == 0)
				break;
			    ++rst->count;
			}
			if (got_int)
			    break;
		    }
		    else
			status = RA_NOMATCH;

		    /* If it could match, try it. */
		    if (rst->nextb == NUL || *reginput == rst->nextb
					     || *reginput == rst->nextb_ic)
		    {
			reg_save(&rp->rs_un.regsave, &backpos);
			scan = regnext(rp->rs_scan);
			status = RA_CONT;
			break;
		    }
		}
		if (status != RA_CONT)
		{
		    /* Failed. */
		    regstack_pop(&scan);
		    regstack.ga_len -= sizeof(regstar_T);
		    status = RA_NOMATCH;
		}
	    }
	    break;
	}

	/* If we want to continue the inner loop or didn't pop a state
	 * continue matching loop */
	if (status == RA_CONT || rp == (regitem_T *)
			     ((char *)regstack.ga_data + regstack.ga_len) - 1)
	    break;
    }

    /* May need to continue with the inner loop, starting at "scan". */
    if (status == RA_CONT)
	continue;

    /*
     * If the regstack is empty or something failed we are done.
     */
    if (regstack.ga_len == 0 || status == RA_FAIL)
    {
	if (scan == NULL)
	{
	    /*
	     * We get here only if there's trouble -- normally "case END" is
	     * the terminating point.
	     */
	    EMSG(_(e_re_corr));
#ifdef DEBUG
	    printf("Premature EOL\n");
#endif
	}
	if (status == RA_FAIL)
	    got_int = TRUE;
	return (status == RA_MATCH);
    }

  } /* End of loop until the regstack is empty. */

  /* NOTREACHED */
}

/*
 * Push an item onto the regstack.
 * Returns pointer to new item.  Returns NULL when out of memory.
 */
    static regitem_T *
regstack_push(state, scan)
    regstate_T	state;
    char_u	*scan;
{
    regitem_T	*rp;

    if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp)
    {
	EMSG(_(e_maxmempat));
	return NULL;
    }
    if (ga_grow(&regstack, sizeof(regitem_T)) == FAIL)
	return NULL;

    rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len);
    rp->rs_state = state;
    rp->rs_scan = scan;

    regstack.ga_len += sizeof(regitem_T);
    return rp;
}

/*
 * Pop an item from the regstack.
 */
    static void
regstack_pop(scan)
    char_u	**scan;
{
    regitem_T	*rp;

    rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1;
    *scan = rp->rs_scan;

    regstack.ga_len -= sizeof(regitem_T);
}

/*
 * regrepeat - repeatedly match something simple, return how many.
 * Advances reginput (and reglnum) to just after the matched chars.
 */
    static int
regrepeat(p, maxcount)
    char_u	*p;
    long	maxcount;   /* maximum number of matches allowed */
{
    long	count = 0;
    char_u	*scan;
    char_u	*opnd;
    int		mask;
    int		testval = 0;

    scan = reginput;	    /* Make local copy of reginput for speed. */
    opnd = OPERAND(p);
    switch (OP(p))
    {
      case ANY:
      case ANY + ADD_NL:
	while (count < maxcount)
	{
	    /* Matching anything means we continue until end-of-line (or
	     * end-of-file for ANY + ADD_NL), only limited by maxcount. */
	    while (*scan != NUL && count < maxcount)
	    {
		++count;
		mb_ptr_adv(scan);
	    }
	    if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline
					 || reg_line_lbr || count == maxcount)
		break;
	    ++count;		/* count the line-break */
	    reg_nextline();
	    scan = reginput;
	    if (got_int)
		break;
	}
	break;

      case IDENT:
      case IDENT + ADD_NL:
	testval = TRUE;
	/*FALLTHROUGH*/
      case SIDENT:
      case SIDENT + ADD_NL:
	while (count < maxcount)
	{
	    if (vim_isIDc(*scan) && (testval || !VIM_ISDIGIT(*scan)))
	    {
		mb_ptr_adv(scan);
	    }
	    else if (*scan == NUL)
	    {
		if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline
							      || reg_line_lbr)
		    break;
		reg_nextline();
		scan = reginput;
		if (got_int)
		    break;
	    }
	    else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
		++scan;
	    else
		break;
	    ++count;
	}
	break;

      case KWORD:
      case KWORD + ADD_NL:
	testval = TRUE;
	/*FALLTHROUGH*/
      case SKWORD:
      case SKWORD + ADD_NL:
	while (count < maxcount)
	{
	    if (vim_iswordp(scan) && (testval || !VIM_ISDIGIT(*scan)))
	    {
		mb_ptr_adv(scan);
	    }
	    else if (*scan == NUL)
	    {
		if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline
							      || reg_line_lbr)
		    break;
		reg_nextline();
		scan = reginput;
		if (got_int)
		    break;
	    }
	    else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
		++scan;
	    else
		break;
	    ++count;
	}
	break;

      case FNAME:
      case FNAME + ADD_NL:
	testval = TRUE;
	/*FALLTHROUGH*/
      case SFNAME:
      case SFNAME + ADD_NL:
	while (count < maxcount)
	{
	    if (vim_isfilec(*scan) && (testval || !VIM_ISDIGIT(*scan)))
	    {
		mb_ptr_adv(scan);
	    }
	    else if (*scan == NUL)
	    {
		if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline
							      || reg_line_lbr)
		    break;
		reg_nextline();
		scan = reginput;
		if (got_int)
		    break;
	    }
	    else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
		++scan;
	    else
		break;
	    ++count;
	}
	break;

      case PRINT:
      case PRINT + ADD_NL:
	testval = TRUE;
	/*FALLTHROUGH*/
      case SPRINT:
      case SPRINT + ADD_NL:
	while (count < maxcount)
	{
	    if (*scan == NUL)
	    {
		if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline
							      || reg_line_lbr)
		    break;
		reg_nextline();
		scan = reginput;
		if (got_int)
		    break;
	    }
	    else if (ptr2cells(scan) == 1 && (testval || !VIM_ISDIGIT(*scan)))
	    {
		mb_ptr_adv(scan);
	    }
	    else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
		++scan;
	    else
		break;
	    ++count;
	}
	break;

      case WHITE:
      case WHITE + ADD_NL:
	testval = mask = RI_WHITE;
do_class:
	while (count < maxcount)
	{
#ifdef FEAT_MBYTE
	    int		l;
#endif
	    if (*scan == NUL)
	    {
		if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline
							      || reg_line_lbr)
		    break;
		reg_nextline();
		scan = reginput;
		if (got_int)
		    break;
	    }
#ifdef FEAT_MBYTE
	    else if (has_mbyte && (l = (*mb_ptr2len)(scan)) > 1)
	    {
		if (testval != 0)
		    break;
		scan += l;
	    }
#endif
	    else if ((class_tab[*scan] & mask) == testval)
		++scan;
	    else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
		++scan;
	    else
		break;
	    ++count;
	}
	break;

      case NWHITE:
      case NWHITE + ADD_NL:
	mask = RI_WHITE;
	goto do_class;
      case DIGIT:
      case DIGIT + ADD_NL:
	testval = mask = RI_DIGIT;
	goto do_class;
      case NDIGIT:
      case NDIGIT + ADD_NL:
	mask = RI_DIGIT;
	goto do_class;
      case HEX:
      case HEX + ADD_NL:
	testval = mask = RI_HEX;
	goto do_class;
      case NHEX:
      case NHEX + ADD_NL:
	mask = RI_HEX;
	goto do_class;
      case OCTAL:
      case OCTAL + ADD_NL:
	testval = mask = RI_OCTAL;
	goto do_class;
      case NOCTAL:
      case NOCTAL + ADD_NL:
	mask = RI_OCTAL;
	goto do_class;
      case WORD:
      case WORD + ADD_NL:
	testval = mask = RI_WORD;
	goto do_class;
      case NWORD:
      case NWORD + ADD_NL:
	mask = RI_WORD;
	goto do_class;
      case HEAD:
      case HEAD + ADD_NL:
	testval = mask = RI_HEAD;
	goto do_class;
      case NHEAD:
      case NHEAD + ADD_NL:
	mask = RI_HEAD;
	goto do_class;
      case ALPHA:
      case ALPHA + ADD_NL:
	testval = mask = RI_ALPHA;
	goto do_class;
      case NALPHA:
      case NALPHA + ADD_NL:
	mask = RI_ALPHA;
	goto do_class;
      case LOWER:
      case LOWER + ADD_NL:
	testval = mask = RI_LOWER;
	goto do_class;
      case NLOWER:
      case NLOWER + ADD_NL:
	mask = RI_LOWER;
	goto do_class;
      case UPPER:
      case UPPER + ADD_NL:
	testval = mask = RI_UPPER;
	goto do_class;
      case NUPPER:
      case NUPPER + ADD_NL:
	mask = RI_UPPER;
	goto do_class;

      case EXACTLY:
	{
	    int	    cu, cl;

	    /* This doesn't do a multi-byte character, because a MULTIBYTECODE
	     * would have been used for it.  It does handle single-byte
	     * characters, such as latin1. */
	    if (ireg_ic)
	    {
		cu = MB_TOUPPER(*opnd);
		cl = MB_TOLOWER(*opnd);
		while (count < maxcount && (*scan == cu || *scan == cl))
		{
		    count++;
		    scan++;
		}
	    }
	    else
	    {
		cu = *opnd;
		while (count < maxcount && *scan == cu)
		{
		    count++;
		    scan++;
		}
	    }
	    break;
	}

#ifdef FEAT_MBYTE
      case MULTIBYTECODE:
	{
	    int		i, len, cf = 0;

	    /* Safety check (just in case 'encoding' was changed since
	     * compiling the program). */
	    if ((len = (*mb_ptr2len)(opnd)) > 1)
	    {
		if (ireg_ic && enc_utf8)
		    cf = utf_fold(utf_ptr2char(opnd));
		while (count < maxcount)
		{
		    for (i = 0; i < len; ++i)
			if (opnd[i] != scan[i])
			    break;
		    if (i < len && (!ireg_ic || !enc_utf8
					|| utf_fold(utf_ptr2char(scan)) != cf))
			break;
		    scan += len;
		    ++count;
		}
	    }
	}
	break;
#endif

      case ANYOF:
      case ANYOF + ADD_NL:
	testval = TRUE;
	/*FALLTHROUGH*/

      case ANYBUT:
      case ANYBUT + ADD_NL:
	while (count < maxcount)
	{
#ifdef FEAT_MBYTE
	    int len;
#endif
	    if (*scan == NUL)
	    {
		if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline
							      || reg_line_lbr)
		    break;
		reg_nextline();
		scan = reginput;
		if (got_int)
		    break;
	    }
	    else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
		++scan;
#ifdef FEAT_MBYTE
	    else if (has_mbyte && (len = (*mb_ptr2len)(scan)) > 1)
	    {
		if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval)
		    break;
		scan += len;
	    }
#endif
	    else
	    {
		if ((cstrchr(opnd, *scan) == NULL) == testval)
		    break;
		++scan;
	    }
	    ++count;
	}
	break;

      case NEWL:
	while (count < maxcount
		&& ((*scan == NUL && reglnum <= reg_maxline && !reg_line_lbr
			    && REG_MULTI) || (*scan == '\n' && reg_line_lbr)))
	{
	    count++;
	    if (reg_line_lbr)
		ADVANCE_REGINPUT();
	    else
		reg_nextline();
	    scan = reginput;
	    if (got_int)
		break;
	}
	break;

      default:			/* Oh dear.  Called inappropriately. */
	EMSG(_(e_re_corr));
#ifdef DEBUG
	printf("Called regrepeat with op code %d\n", OP(p));
#endif
	break;
    }

    reginput = scan;

    return (int)count;
}

/*
 * regnext - dig the "next" pointer out of a node
 */
    static char_u *
regnext(p)
    char_u  *p;
{
    int	    offset;

    if (p == JUST_CALC_SIZE)
	return NULL;

    offset = NEXT(p);
    if (offset == 0)
	return NULL;

    if (OP(p) == BACK)
	return p - offset;
    else
	return p + offset;
}

/*
 * Check the regexp program for its magic number.
 * Return TRUE if it's wrong.
 */
    static int
prog_magic_wrong()
{
    if (UCHARAT(REG_MULTI
		? reg_mmatch->regprog->program
		: reg_match->regprog->program) != REGMAGIC)
    {
	EMSG(_(e_re_corr));
	return TRUE;
    }
    return FALSE;
}

/*
 * Cleanup the subexpressions, if this wasn't done yet.
 * This construction is used to clear the subexpressions only when they are
 * used (to increase speed).
 */
    static void
cleanup_subexpr()
{
    if (need_clear_subexpr)
    {
	if (REG_MULTI)
	{
	    /* Use 0xff to set lnum to -1 */
	    vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP);
	    vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP);
	}
	else
	{
	    vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP);
	    vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP);
	}
	need_clear_subexpr = FALSE;
    }
}

#ifdef FEAT_SYN_HL
    static void
cleanup_zsubexpr()
{
    if (need_clear_zsubexpr)
    {
	if (REG_MULTI)
	{
	    /* Use 0xff to set lnum to -1 */
	    vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP);
	    vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP);
	}
	else
	{
	    vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP);
	    vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP);
	}
	need_clear_zsubexpr = FALSE;
    }
}
#endif

/*
 * Advance reglnum, regline and reginput to the next line.
 */
    static void
reg_nextline()
{
    regline = reg_getline(++reglnum);
    reginput = regline;
    fast_breakcheck();
}

/*
 * Save the input line and position in a regsave_T.
 */
    static void
reg_save(save, gap)
    regsave_T	*save;
    garray_T	*gap;
{
    if (REG_MULTI)
    {
	save->rs_u.pos.col = (colnr_T)(reginput - regline);
	save->rs_u.pos.lnum = reglnum;
    }
    else
	save->rs_u.ptr = reginput;
    save->rs_len = gap->ga_len;
}

/*
 * Restore the input line and position from a regsave_T.
 */
    static void
reg_restore(save, gap)
    regsave_T	*save;
    garray_T	*gap;
{
    if (REG_MULTI)
    {
	if (reglnum != save->rs_u.pos.lnum)
	{
	    /* only call reg_getline() when the line number changed to save
	     * a bit of time */
	    reglnum = save->rs_u.pos.lnum;
	    regline = reg_getline(reglnum);
	}
	reginput = regline + save->rs_u.pos.col;
    }
    else
	reginput = save->rs_u.ptr;
    gap->ga_len = save->rs_len;
}

/*
 * Return TRUE if current position is equal to saved position.
 */
    static int
reg_save_equal(save)
    regsave_T	*save;
{
    if (REG_MULTI)
	return reglnum == save->rs_u.pos.lnum
				  && reginput == regline + save->rs_u.pos.col;
    return reginput == save->rs_u.ptr;
}

/*
 * Tentatively set the sub-expression start to the current position (after
 * calling regmatch() they will have changed).  Need to save the existing
 * values for when there is no match.
 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()),
 * depending on REG_MULTI.
 */
    static void
save_se_multi(savep, posp)
    save_se_T	*savep;
    lpos_T	*posp;
{
    savep->se_u.pos = *posp;
    posp->lnum = reglnum;
    posp->col = (colnr_T)(reginput - regline);
}

    static void
save_se_one(savep, pp)
    save_se_T	*savep;
    char_u	**pp;
{
    savep->se_u.ptr = *pp;
    *pp = reginput;
}

/*
 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL.
 */
    static int
re_num_cmp(val, scan)
    long_u	val;
    char_u	*scan;
{
    long_u  n = OPERAND_MIN(scan);

    if (OPERAND_CMP(scan) == '>')
	return val > n;
    if (OPERAND_CMP(scan) == '<')
	return val < n;
    return val == n;
}


#ifdef DEBUG

/*
 * regdump - dump a regexp onto stdout in vaguely comprehensible form
 */
    static void
regdump(pattern, r)
    char_u	*pattern;
    regprog_T	*r;
{
    char_u  *s;
    int	    op = EXACTLY;	/* Arbitrary non-END op. */
    char_u  *next;
    char_u  *end = NULL;

    printf("\r\nregcomp(%s):\r\n", pattern);

    s = r->program + 1;
    /*
     * Loop until we find the END that isn't before a referred next (an END
     * can also appear in a NOMATCH operand).
     */
    while (op != END || s <= end)
    {
	op = OP(s);
	printf("%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */
	next = regnext(s);
	if (next == NULL)	/* Next ptr. */
	    printf("(0)");
	else
	    printf("(%d)", (int)((s - r->program) + (next - s)));
	if (end < next)
	    end = next;
	if (op == BRACE_LIMITS)
	{
	    /* Two short ints */
	    printf(" minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s));
	    s += 8;
	}
	s += 3;
	if (op == ANYOF || op == ANYOF + ADD_NL
		|| op == ANYBUT || op == ANYBUT + ADD_NL
		|| op == EXACTLY)
	{
	    /* Literal string, where present. */
	    while (*s != NUL)
		printf("%c", *s++);
	    s++;
	}
	printf("\r\n");
    }

    /* Header fields of interest. */
    if (r->regstart != NUL)
	printf("start `%s' 0x%x; ", r->regstart < 256
		? (char *)transchar(r->regstart)
		: "multibyte", r->regstart);
    if (r->reganch)
	printf("anchored; ");
    if (r->regmust != NULL)
	printf("must have \"%s\"", r->regmust);
    printf("\r\n");
}

/*
 * regprop - printable representation of opcode
 */
    static char_u *
regprop(op)
    char_u	   *op;
{
    char_u	    *p;
    static char_u   buf[50];

    (void) strcpy(buf, ":");

    switch (OP(op))
    {
      case BOL:
	p = "BOL";
	break;
      case EOL:
	p = "EOL";
	break;
      case RE_BOF:
	p = "BOF";
	break;
      case RE_EOF:
	p = "EOF";
	break;
      case CURSOR:
	p = "CURSOR";
	break;
      case RE_VISUAL:
	p = "RE_VISUAL";
	break;
      case RE_LNUM:
	p = "RE_LNUM";
	break;
      case RE_MARK:
	p = "RE_MARK";
	break;
      case RE_COL:
	p = "RE_COL";
	break;
      case RE_VCOL:
	p = "RE_VCOL";
	break;
      case BOW:
	p = "BOW";
	break;
      case EOW:
	p = "EOW";
	break;
      case ANY:
	p = "ANY";
	break;
      case ANY + ADD_NL:
	p = "ANY+NL";
	break;
      case ANYOF:
	p = "ANYOF";
	break;
      case ANYOF + ADD_NL:
	p = "ANYOF+NL";
	break;
      case ANYBUT:
	p = "ANYBUT";
	break;
      case ANYBUT + ADD_NL:
	p = "ANYBUT+NL";
	break;
      case IDENT:
	p = "IDENT";
	break;
      case IDENT + ADD_NL:
	p = "IDENT+NL";
	break;
      case SIDENT:
	p = "SIDENT";
	break;
      case SIDENT + ADD_NL:
	p = "SIDENT+NL";
	break;
      case KWORD:
	p = "KWORD";
	break;
      case KWORD + ADD_NL:
	p = "KWORD+NL";
	break;
      case SKWORD:
	p = "SKWORD";
	break;
      case SKWORD + ADD_NL:
	p = "SKWORD+NL";
	break;
      case FNAME:
	p = "FNAME";
	break;
      case FNAME + ADD_NL:
	p = "FNAME+NL";
	break;
      case SFNAME:
	p = "SFNAME";
	break;
      case SFNAME + ADD_NL:
	p = "SFNAME+NL";
	break;
      case PRINT:
	p = "PRINT";
	break;
      case PRINT + ADD_NL:
	p = "PRINT+NL";
	break;
      case SPRINT:
	p = "SPRINT";
	break;
      case SPRINT + ADD_NL:
	p = "SPRINT+NL";
	break;
      case WHITE:
	p = "WHITE";
	break;
      case WHITE + ADD_NL:
	p = "WHITE+NL";
	break;
      case NWHITE:
	p = "NWHITE";
	break;
      case NWHITE + ADD_NL:
	p = "NWHITE+NL";
	break;
      case DIGIT:
	p = "DIGIT";
	break;
      case DIGIT + ADD_NL:
	p = "DIGIT+NL";
	break;
      case NDIGIT:
	p = "NDIGIT";
	break;
      case NDIGIT + ADD_NL:
	p = "NDIGIT+NL";
	break;
      case HEX:
	p = "HEX";
	break;
      case HEX + ADD_NL:
	p = "HEX+NL";
	break;
      case NHEX:
	p = "NHEX";
	break;
      case NHEX + ADD_NL:
	p = "NHEX+NL";
	break;
      case OCTAL:
	p = "OCTAL";
	break;
      case OCTAL + ADD_NL:
	p = "OCTAL+NL";
	break;
      case NOCTAL:
	p = "NOCTAL";
	break;
      case NOCTAL + ADD_NL:
	p = "NOCTAL+NL";
	break;
      case WORD:
	p = "WORD";
	break;
      case WORD + ADD_NL:
	p = "WORD+NL";
	break;
      case NWORD:
	p = "NWORD";
	break;
      case NWORD + ADD_NL:
	p = "NWORD+NL";
	break;
      case HEAD:
	p = "HEAD";
	break;
      case HEAD + ADD_NL:
	p = "HEAD+NL";
	break;
      case NHEAD:
	p = "NHEAD";
	break;
      case NHEAD + ADD_NL:
	p = "NHEAD+NL";
	break;
      case ALPHA:
	p = "ALPHA";
	break;
      case ALPHA + ADD_NL:
	p = "ALPHA+NL";
	break;
      case NALPHA:
	p = "NALPHA";
	break;
      case NALPHA + ADD_NL:
	p = "NALPHA+NL";
	break;
      case LOWER:
	p = "LOWER";
	break;
      case LOWER + ADD_NL:
	p = "LOWER+NL";
	break;
      case NLOWER:
	p = "NLOWER";
	break;
      case NLOWER + ADD_NL:
	p = "NLOWER+NL";
	break;
      case UPPER:
	p = "UPPER";
	break;
      case UPPER + ADD_NL:
	p = "UPPER+NL";
	break;
      case NUPPER:
	p = "NUPPER";
	break;
      case NUPPER + ADD_NL:
	p = "NUPPER+NL";
	break;
      case BRANCH:
	p = "BRANCH";
	break;
      case EXACTLY:
	p = "EXACTLY";
	break;
      case NOTHING:
	p = "NOTHING";
	break;
      case BACK:
	p = "BACK";
	break;
      case END:
	p = "END";
	break;
      case MOPEN + 0:
	p = "MATCH START";
	break;
      case MOPEN + 1:
      case MOPEN + 2:
      case MOPEN + 3:
      case MOPEN + 4:
      case MOPEN + 5:
      case MOPEN + 6:
      case MOPEN + 7:
      case MOPEN + 8:
      case MOPEN + 9:
	sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN);
	p = NULL;
	break;
      case MCLOSE + 0:
	p = "MATCH END";
	break;
      case MCLOSE + 1:
      case MCLOSE + 2:
      case MCLOSE + 3:
      case MCLOSE + 4:
      case MCLOSE + 5:
      case MCLOSE + 6:
      case MCLOSE + 7:
      case MCLOSE + 8:
      case MCLOSE + 9:
	sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE);
	p = NULL;
	break;
      case BACKREF + 1:
      case BACKREF + 2:
      case BACKREF + 3:
      case BACKREF + 4:
      case BACKREF + 5:
      case BACKREF + 6:
      case BACKREF + 7:
      case BACKREF + 8:
      case BACKREF + 9:
	sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF);
	p = NULL;
	break;
      case NOPEN:
	p = "NOPEN";
	break;
      case NCLOSE:
	p = "NCLOSE";
	break;
#ifdef FEAT_SYN_HL
      case ZOPEN + 1:
      case ZOPEN + 2:
      case ZOPEN + 3:
      case ZOPEN + 4:
      case ZOPEN + 5:
      case ZOPEN + 6:
      case ZOPEN + 7:
      case ZOPEN + 8:
      case ZOPEN + 9:
	sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN);
	p = NULL;
	break;
      case ZCLOSE + 1:
      case ZCLOSE + 2:
      case ZCLOSE + 3:
      case ZCLOSE + 4:
      case ZCLOSE + 5:
      case ZCLOSE + 6:
      case ZCLOSE + 7:
      case ZCLOSE + 8:
      case ZCLOSE + 9:
	sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE);
	p = NULL;
	break;
      case ZREF + 1:
      case ZREF + 2:
      case ZREF + 3:
      case ZREF + 4:
      case ZREF + 5:
      case ZREF + 6:
      case ZREF + 7:
      case ZREF + 8:
      case ZREF + 9:
	sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF);
	p = NULL;
	break;
#endif
      case STAR:
	p = "STAR";
	break;
      case PLUS:
	p = "PLUS";
	break;
      case NOMATCH:
	p = "NOMATCH";
	break;
      case MATCH:
	p = "MATCH";
	break;
      case BEHIND:
	p = "BEHIND";
	break;
      case NOBEHIND:
	p = "NOBEHIND";
	break;
      case SUBPAT:
	p = "SUBPAT";
	break;
      case BRACE_LIMITS:
	p = "BRACE_LIMITS";
	break;
      case BRACE_SIMPLE:
	p = "BRACE_SIMPLE";
	break;
      case BRACE_COMPLEX + 0:
      case BRACE_COMPLEX + 1:
      case BRACE_COMPLEX + 2:
      case BRACE_COMPLEX + 3:
      case BRACE_COMPLEX + 4:
      case BRACE_COMPLEX + 5:
      case BRACE_COMPLEX + 6:
      case BRACE_COMPLEX + 7:
      case BRACE_COMPLEX + 8:
      case BRACE_COMPLEX + 9:
	sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX);
	p = NULL;
	break;
#ifdef FEAT_MBYTE
      case MULTIBYTECODE:
	p = "MULTIBYTECODE";
	break;
#endif
      case NEWL:
	p = "NEWL";
	break;
      default:
	sprintf(buf + STRLEN(buf), "corrupt %d", OP(op));
	p = NULL;
	break;
    }
    if (p != NULL)
	(void) strcat(buf, p);
    return buf;
}
#endif

#ifdef FEAT_MBYTE
static void mb_decompose __ARGS((int c, int *c1, int *c2, int *c3));

typedef struct
{
    int a, b, c;
} decomp_T;


/* 0xfb20 - 0xfb4f */
static decomp_T decomp_table[0xfb4f-0xfb20+1] =
{
    {0x5e2,0,0},		/* 0xfb20	alt ayin */
    {0x5d0,0,0},		/* 0xfb21	alt alef */
    {0x5d3,0,0},		/* 0xfb22	alt dalet */
    {0x5d4,0,0},		/* 0xfb23	alt he */
    {0x5db,0,0},		/* 0xfb24	alt kaf */
    {0x5dc,0,0},		/* 0xfb25	alt lamed */
    {0x5dd,0,0},		/* 0xfb26	alt mem-sofit */
    {0x5e8,0,0},		/* 0xfb27	alt resh */
    {0x5ea,0,0},		/* 0xfb28	alt tav */
    {'+', 0, 0},		/* 0xfb29	alt plus */
    {0x5e9, 0x5c1, 0},		/* 0xfb2a	shin+shin-dot */
    {0x5e9, 0x5c2, 0},		/* 0xfb2b	shin+sin-dot */
    {0x5e9, 0x5c1, 0x5bc},	/* 0xfb2c	shin+shin-dot+dagesh */
    {0x5e9, 0x5c2, 0x5bc},	/* 0xfb2d	shin+sin-dot+dagesh */
    {0x5d0, 0x5b7, 0},		/* 0xfb2e	alef+patah */
    {0x5d0, 0x5b8, 0},		/* 0xfb2f	alef+qamats */
    {0x5d0, 0x5b4, 0},		/* 0xfb30	alef+hiriq */
    {0x5d1, 0x5bc, 0},		/* 0xfb31	bet+dagesh */
    {0x5d2, 0x5bc, 0},		/* 0xfb32	gimel+dagesh */
    {0x5d3, 0x5bc, 0},		/* 0xfb33	dalet+dagesh */
    {0x5d4, 0x5bc, 0},		/* 0xfb34	he+dagesh */
    {0x5d5, 0x5bc, 0},		/* 0xfb35	vav+dagesh */
    {0x5d6, 0x5bc, 0},		/* 0xfb36	zayin+dagesh */
    {0xfb37, 0, 0},		/* 0xfb37 -- UNUSED */
    {0x5d8, 0x5bc, 0},		/* 0xfb38	tet+dagesh */
    {0x5d9, 0x5bc, 0},		/* 0xfb39	yud+dagesh */
    {0x5da, 0x5bc, 0},		/* 0xfb3a	kaf sofit+dagesh */
    {0x5db, 0x5bc, 0},		/* 0xfb3b	kaf+dagesh */
    {0x5dc, 0x5bc, 0},		/* 0xfb3c	lamed+dagesh */
    {0xfb3d, 0, 0},		/* 0xfb3d -- UNUSED */
    {0x5de, 0x5bc, 0},		/* 0xfb3e	mem+dagesh */
    {0xfb3f, 0, 0},		/* 0xfb3f -- UNUSED */
    {0x5e0, 0x5bc, 0},		/* 0xfb40	nun+dagesh */
    {0x5e1, 0x5bc, 0},		/* 0xfb41	samech+dagesh */
    {0xfb42, 0, 0},		/* 0xfb42 -- UNUSED */
    {0x5e3, 0x5bc, 0},		/* 0xfb43	pe sofit+dagesh */
    {0x5e4, 0x5bc,0},		/* 0xfb44	pe+dagesh */
    {0xfb45, 0, 0},		/* 0xfb45 -- UNUSED */
    {0x5e6, 0x5bc, 0},		/* 0xfb46	tsadi+dagesh */
    {0x5e7, 0x5bc, 0},		/* 0xfb47	qof+dagesh */
    {0x5e8, 0x5bc, 0},		/* 0xfb48	resh+dagesh */
    {0x5e9, 0x5bc, 0},		/* 0xfb49	shin+dagesh */
    {0x5ea, 0x5bc, 0},		/* 0xfb4a	tav+dagesh */
    {0x5d5, 0x5b9, 0},		/* 0xfb4b	vav+holam */
    {0x5d1, 0x5bf, 0},		/* 0xfb4c	bet+rafe */
    {0x5db, 0x5bf, 0},		/* 0xfb4d	kaf+rafe */
    {0x5e4, 0x5bf, 0},		/* 0xfb4e	pe+rafe */
    {0x5d0, 0x5dc, 0}		/* 0xfb4f	alef-lamed */
};

    static void
mb_decompose(c, c1, c2, c3)
    int c, *c1, *c2, *c3;
{
    decomp_T d;

    if (c >= 0x4b20 && c <= 0xfb4f)
    {
	d = decomp_table[c - 0xfb20];
	*c1 = d.a;
	*c2 = d.b;
	*c3 = d.c;
    }
    else
    {
	*c1 = c;
	*c2 = *c3 = 0;
    }
}
#endif

/*
 * Compare two strings, ignore case if ireg_ic set.
 * Return 0 if strings match, non-zero otherwise.
 * Correct the length "*n" when composing characters are ignored.
 */
    static int
cstrncmp(s1, s2, n)
    char_u	*s1, *s2;
    int		*n;
{
    int		result;

    if (!ireg_ic)
	result = STRNCMP(s1, s2, *n);
    else
	result = MB_STRNICMP(s1, s2, *n);

#ifdef FEAT_MBYTE
    /* if it failed and it's utf8 and we want to combineignore: */
    if (result != 0 && enc_utf8 && ireg_icombine)
    {
	char_u	*str1, *str2;
	int	c1, c2, c11, c12;
	int	junk;

	/* we have to handle the strcmp ourselves, since it is necessary to
	 * deal with the composing characters by ignoring them: */
	str1 = s1;
	str2 = s2;
	c1 = c2 = 0;
	while ((int)(str1 - s1) < *n)
	{
	    c1 = mb_ptr2char_adv(&str1);
	    c2 = mb_ptr2char_adv(&str2);

	    /* decompose the character if necessary, into 'base' characters
	     * because I don't care about Arabic, I will hard-code the Hebrew
	     * which I *do* care about!  So sue me... */
	    if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2)))
	    {
		/* decomposition necessary? */
		mb_decompose(c1, &c11, &junk, &junk);
		mb_decompose(c2, &c12, &junk, &junk);
		c1 = c11;
		c2 = c12;
		if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12)))
		    break;
	    }
	}
	result = c2 - c1;
	if (result == 0)
	    *n = (int)(str2 - s2);
    }
#endif

    return result;
}

/*
 * cstrchr: This function is used a lot for simple searches, keep it fast!
 */
    static char_u *
cstrchr(s, c)
    char_u	*s;
    int		c;
{
    char_u	*p;
    int		cc;

    if (!ireg_ic
#ifdef FEAT_MBYTE
	    || (!enc_utf8 && mb_char2len(c) > 1)
#endif
	    )
	return vim_strchr(s, c);

    /* tolower() and toupper() can be slow, comparing twice should be a lot
     * faster (esp. when using MS Visual C++!).
     * For UTF-8 need to use folded case. */
#ifdef FEAT_MBYTE
    if (enc_utf8 && c > 0x80)
	cc = utf_fold(c);
    else
#endif
	 if (MB_ISUPPER(c))
	cc = MB_TOLOWER(c);
    else if (MB_ISLOWER(c))
	cc = MB_TOUPPER(c);
    else
	return vim_strchr(s, c);

#ifdef FEAT_MBYTE
    if (has_mbyte)
    {
	for (p = s; *p != NUL; p += (*mb_ptr2len)(p))
	{
	    if (enc_utf8 && c > 0x80)
	    {
		if (utf_fold(utf_ptr2char(p)) == cc)
		    return p;
	    }
	    else if (*p == c || *p == cc)
		return p;
	}
    }
    else
#endif
	/* Faster version for when there are no multi-byte characters. */
	for (p = s; *p != NUL; ++p)
	    if (*p == c || *p == cc)
		return p;

    return NULL;
}

/***************************************************************
 *		      regsub stuff			       *
 ***************************************************************/

/* This stuff below really confuses cc on an SGI -- webb */
#ifdef __sgi
# undef __ARGS
# define __ARGS(x)  ()
#endif

/*
 * We should define ftpr as a pointer to a function returning a pointer to
 * a function returning a pointer to a function ...
 * This is impossible, so we declare a pointer to a function returning a
 * pointer to a function returning void. This should work for all compilers.
 */
typedef void (*(*fptr_T) __ARGS((int *, int)))();

static fptr_T do_upper __ARGS((int *, int));
static fptr_T do_Upper __ARGS((int *, int));
static fptr_T do_lower __ARGS((int *, int));
static fptr_T do_Lower __ARGS((int *, int));

static int vim_regsub_both __ARGS((char_u *source, char_u *dest, int copy, int magic, int backslash));

    static fptr_T
do_upper(d, c)
    int		*d;
    int		c;
{
    *d = MB_TOUPPER(c);

    return (fptr_T)NULL;
}

    static fptr_T
do_Upper(d, c)
    int		*d;
    int		c;
{
    *d = MB_TOUPPER(c);

    return (fptr_T)do_Upper;
}

    static fptr_T
do_lower(d, c)
    int		*d;
    int		c;
{
    *d = MB_TOLOWER(c);

    return (fptr_T)NULL;
}

    static fptr_T
do_Lower(d, c)
    int		*d;
    int		c;
{
    *d = MB_TOLOWER(c);

    return (fptr_T)do_Lower;
}

/*
 * regtilde(): Replace tildes in the pattern by the old pattern.
 *
 * Short explanation of the tilde: It stands for the previous replacement
 * pattern.  If that previous pattern also contains a ~ we should go back a
 * step further...  But we insert the previous pattern into the current one
 * and remember that.
 * This still does not handle the case where "magic" changes.  So require the
 * user to keep his hands off of "magic".
 *
 * The tildes are parsed once before the first call to vim_regsub().
 */
    char_u *
regtilde(source, magic)
    char_u	*source;
    int		magic;
{
    char_u	*newsub = source;
    char_u	*tmpsub;
    char_u	*p;
    int		len;
    int		prevlen;

    for (p = newsub; *p; ++p)
    {
	if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic))
	{
	    if (reg_prev_sub != NULL)
	    {
		/* length = len(newsub) - 1 + len(prev_sub) + 1 */
		prevlen = (int)STRLEN(reg_prev_sub);
		tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen));
		if (tmpsub != NULL)
		{
		    /* copy prefix */
		    len = (int)(p - newsub);	/* not including ~ */
		    mch_memmove(tmpsub, newsub, (size_t)len);
		    /* interpret tilde */
		    mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen);
		    /* copy postfix */
		    if (!magic)
			++p;			/* back off \ */
		    STRCPY(tmpsub + len + prevlen, p + 1);

		    if (newsub != source)	/* already allocated newsub */
			vim_free(newsub);
		    newsub = tmpsub;
		    p = newsub + len + prevlen;
		}
	    }
	    else if (magic)
		mch_memmove(p, p + 1, STRLEN(p));	/* remove '~' */
	    else
		mch_memmove(p, p + 2, STRLEN(p) - 1);	/* remove '\~' */
	    --p;
	}
	else
	{
	    if (*p == '\\' && p[1])		/* skip escaped characters */
		++p;
#ifdef FEAT_MBYTE
	    if (has_mbyte)
		p += (*mb_ptr2len)(p) - 1;
#endif
	}
    }

    vim_free(reg_prev_sub);
    if (newsub != source)	/* newsub was allocated, just keep it */
	reg_prev_sub = newsub;
    else			/* no ~ found, need to save newsub  */
	reg_prev_sub = vim_strsave(newsub);
    return newsub;
}

#ifdef FEAT_EVAL
static int can_f_submatch = FALSE;	/* TRUE when submatch() can be used */

/* These pointers are used instead of reg_match and reg_mmatch for
 * reg_submatch().  Needed for when the substitution string is an expression
 * that contains a call to substitute() and submatch(). */
static regmatch_T	*submatch_match;
static regmmatch_T	*submatch_mmatch;
#endif

#if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO)
/*
 * vim_regsub() - perform substitutions after a vim_regexec() or
 * vim_regexec_multi() match.
 *
 * If "copy" is TRUE really copy into "dest".
 * If "copy" is FALSE nothing is copied, this is just to find out the length
 * of the result.
 *
 * If "backslash" is TRUE, a backslash will be removed later, need to double
 * them to keep them, and insert a backslash before a CR to avoid it being
 * replaced with a line break later.
 *
 * Note: The matched text must not change between the call of
 * vim_regexec()/vim_regexec_multi() and vim_regsub()!  It would make the back
 * references invalid!
 *
 * Returns the size of the replacement, including terminating NUL.
 */
    int
vim_regsub(rmp, source, dest, copy, magic, backslash)
    regmatch_T	*rmp;
    char_u	*source;
    char_u	*dest;
    int		copy;
    int		magic;
    int		backslash;
{
    reg_match = rmp;
    reg_mmatch = NULL;
    reg_maxline = 0;
    return vim_regsub_both(source, dest, copy, magic, backslash);
}
#endif

    int
vim_regsub_multi(rmp, lnum, source, dest, copy, magic, backslash)
    regmmatch_T	*rmp;
    linenr_T	lnum;
    char_u	*source;
    char_u	*dest;
    int		copy;
    int		magic;
    int		backslash;
{
    reg_match = NULL;
    reg_mmatch = rmp;
    reg_buf = curbuf;		/* always works on the current buffer! */
    reg_firstlnum = lnum;
    reg_maxline = curbuf->b_ml.ml_line_count - lnum;
    return vim_regsub_both(source, dest, copy, magic, backslash);
}

    static int
vim_regsub_both(source, dest, copy, magic, backslash)
    char_u	*source;
    char_u	*dest;
    int		copy;
    int		magic;
    int		backslash;
{
    char_u	*src;
    char_u	*dst;
    char_u	*s;
    int		c;
    int		cc;
    int		no = -1;
    fptr_T	func = (fptr_T)NULL;
    linenr_T	clnum = 0;	/* init for GCC */
    int		len = 0;	/* init for GCC */
#ifdef FEAT_EVAL
    static char_u *eval_result = NULL;
#endif

    /* Be paranoid... */
    if (source == NULL || dest == NULL)
    {
	EMSG(_(e_null));
	return 0;
    }
    if (prog_magic_wrong())
	return 0;
    src = source;
    dst = dest;

    /*
     * When the substitute part starts with "\=" evaluate it as an expression.
     */
    if (source[0] == '\\' && source[1] == '='
#ifdef FEAT_EVAL
	    && !can_f_submatch	    /* can't do this recursively */
#endif
	    )
    {
#ifdef FEAT_EVAL
	/* To make sure that the length doesn't change between checking the
	 * length and copying the string, and to speed up things, the
	 * resulting string is saved from the call with "copy" == FALSE to the
	 * call with "copy" == TRUE. */
	if (copy)
	{
	    if (eval_result != NULL)
	    {
		STRCPY(dest, eval_result);
		dst += STRLEN(eval_result);
		vim_free(eval_result);
		eval_result = NULL;
	    }
	}
	else
	{
	    linenr_T	save_reg_maxline;
	    win_T	*save_reg_win;
	    int		save_ireg_ic;

	    vim_free(eval_result);

	    /* The expression may contain substitute(), which calls us
	     * recursively.  Make sure submatch() gets the text from the first
	     * level.  Don't need to save "reg_buf", because
	     * vim_regexec_multi() can't be called recursively. */
	    submatch_match = reg_match;
	    submatch_mmatch = reg_mmatch;
	    save_reg_maxline = reg_maxline;
	    save_reg_win = reg_win;
	    save_ireg_ic = ireg_ic;
	    can_f_submatch = TRUE;

	    eval_result = eval_to_string(source + 2, NULL, TRUE);
	    if (eval_result != NULL)
	    {
		for (s = eval_result; *s != NUL; mb_ptr_adv(s))
		{
		    /* Change NL to CR, so that it becomes a line break.
		     * Skip over a backslashed character. */
		    if (*s == NL)
			*s = CAR;
		    else if (*s == '\\' && s[1] != NUL)
			++s;
		}

		dst += STRLEN(eval_result);
	    }

	    reg_match = submatch_match;
	    reg_mmatch = submatch_mmatch;
	    reg_maxline = save_reg_maxline;
	    reg_win = save_reg_win;
	    ireg_ic = save_ireg_ic;
	    can_f_submatch = FALSE;
	}
#endif
    }
    else
      while ((c = *src++) != NUL)
      {
	if (c == '&' && magic)
	    no = 0;
	else if (c == '\\' && *src != NUL)
	{
	    if (*src == '&' && !magic)
	    {
		++src;
		no = 0;
	    }
	    else if ('0' <= *src && *src <= '9')
	    {
		no = *src++ - '0';
	    }
	    else if (vim_strchr((char_u *)"uUlLeE", *src))
	    {
		switch (*src++)
		{
		case 'u':   func = (fptr_T)do_upper;
			    continue;
		case 'U':   func = (fptr_T)do_Upper;
			    continue;
		case 'l':   func = (fptr_T)do_lower;
			    continue;
		case 'L':   func = (fptr_T)do_Lower;
			    continue;
		case 'e':
		case 'E':   func = (fptr_T)NULL;
			    continue;
		}
	    }
	}
	if (no < 0)	      /* Ordinary character. */
	{
	    if (c == K_SPECIAL && src[0] != NUL && src[1] != NUL)
	    {
		/* Copy a special key as-is. */
		if (copy)
		{
		    *dst++ = c;
		    *dst++ = *src++;
		    *dst++ = *src++;
		}
		else
		{
		    dst += 3;
		    src += 2;
		}
		continue;
	    }

	    if (c == '\\' && *src != NUL)
	    {
		/* Check for abbreviations -- webb */
		switch (*src)
		{
		    case 'r':	c = CAR;	++src;	break;
		    case 'n':	c = NL;		++src;	break;
		    case 't':	c = TAB;	++src;	break;
		 /* Oh no!  \e already has meaning in subst pat :-( */
		 /* case 'e':   c = ESC;	++src;	break; */
		    case 'b':	c = Ctrl_H;	++src;	break;

		    /* If "backslash" is TRUE the backslash will be removed
		     * later.  Used to insert a literal CR. */
		    default:	if (backslash)
				{
				    if (copy)
					*dst = '\\';
				    ++dst;
				}
				c = *src++;
		}
	    }
#ifdef FEAT_MBYTE
	    else if (has_mbyte)
		c = mb_ptr2char(src - 1);
#endif

	    /* Write to buffer, if copy is set. */
	    if (func == (fptr_T)NULL)	/* just copy */
		cc = c;
	    else
		/* Turbo C complains without the typecast */
		func = (fptr_T)(func(&cc, c));

#ifdef FEAT_MBYTE
	    if (has_mbyte)
	    {
		src += mb_ptr2len(src - 1) - 1;
		if (copy)
		    mb_char2bytes(cc, dst);
		dst += mb_char2len(cc) - 1;
	    }
	    else
#endif
		if (copy)
		    *dst = cc;
	    dst++;
	}
	else
	{
	    if (REG_MULTI)
	    {
		clnum = reg_mmatch->startpos[no].lnum;
		if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0)
		    s = NULL;
		else
		{
		    s = reg_getline(clnum) + reg_mmatch->startpos[no].col;
		    if (reg_mmatch->endpos[no].lnum == clnum)
			len = reg_mmatch->endpos[no].col
					       - reg_mmatch->startpos[no].col;
		    else
			len = (int)STRLEN(s);
		}
	    }
	    else
	    {
		s = reg_match->startp[no];
		if (reg_match->endp[no] == NULL)
		    s = NULL;
		else
		    len = (int)(reg_match->endp[no] - s);
	    }
	    if (s != NULL)
	    {
		for (;;)
		{
		    if (len == 0)
		    {
			if (REG_MULTI)
			{
			    if (reg_mmatch->endpos[no].lnum == clnum)
				break;
			    if (copy)
				*dst = CAR;
			    ++dst;
			    s = reg_getline(++clnum);
			    if (reg_mmatch->endpos[no].lnum == clnum)
				len = reg_mmatch->endpos[no].col;
			    else
				len = (int)STRLEN(s);
			}
			else
			    break;
		    }
		    else if (*s == NUL) /* we hit NUL. */
		    {
			if (copy)
			    EMSG(_(e_re_damg));
			goto exit;
		    }
		    else
		    {
			if (backslash && (*s == CAR || *s == '\\'))
			{
			    /*
			     * Insert a backslash in front of a CR, otherwise
			     * it will be replaced by a line break.
			     * Number of backslashes will be halved later,
			     * double them here.
			     */
			    if (copy)
			    {
				dst[0] = '\\';
				dst[1] = *s;
			    }
			    dst += 2;
			}
			else
			{
#ifdef FEAT_MBYTE
			    if (has_mbyte)
				c = mb_ptr2char(s);
			    else
#endif
				c = *s;

			    if (func == (fptr_T)NULL)	/* just copy */
				cc = c;
			    else
				/* Turbo C complains without the typecast */
				func = (fptr_T)(func(&cc, c));

#ifdef FEAT_MBYTE
			    if (has_mbyte)
			    {
				int l;

				/* Copy composing characters separately, one
				 * at a time. */
				if (enc_utf8)
				    l = utf_ptr2len(s) - 1;
				else
				    l = mb_ptr2len(s) - 1;

				s += l;
				len -= l;
				if (copy)
				    mb_char2bytes(cc, dst);
				dst += mb_char2len(cc) - 1;
			    }
			    else
#endif
				if (copy)
				    *dst = cc;
			    dst++;
			}

			++s;
			--len;
		    }
		}
	    }
	    no = -1;
	}
      }
    if (copy)
	*dst = NUL;

exit:
    return (int)((dst - dest) + 1);
}

#ifdef FEAT_EVAL
/*
 * Used for the submatch() function: get the string from the n'th submatch in
 * allocated memory.
 * Returns NULL when not in a ":s" command and for a non-existing submatch.
 */
    char_u *
reg_submatch(no)
    int		no;
{
    char_u	*retval = NULL;
    char_u	*s;
    int		len;
    int		round;
    linenr_T	lnum;

    if (!can_f_submatch || no < 0)
	return NULL;

    if (submatch_match == NULL)
    {
	/*
	 * First round: compute the length and allocate memory.
	 * Second round: copy the text.
	 */
	for (round = 1; round <= 2; ++round)
	{
	    lnum = submatch_mmatch->startpos[no].lnum;
	    if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0)
		return NULL;

	    s = reg_getline(lnum) + submatch_mmatch->startpos[no].col;
	    if (s == NULL)  /* anti-crash check, cannot happen? */
		break;
	    if (submatch_mmatch->endpos[no].lnum == lnum)
	    {
		/* Within one line: take form start to end col. */
		len = submatch_mmatch->endpos[no].col
					  - submatch_mmatch->startpos[no].col;
		if (round == 2)
		    vim_strncpy(retval, s, len);
		++len;
	    }
	    else
	    {
		/* Multiple lines: take start line from start col, middle
		 * lines completely and end line up to end col. */
		len = (int)STRLEN(s);
		if (round == 2)
		{
		    STRCPY(retval, s);
		    retval[len] = '\n';
		}
		++len;
		++lnum;
		while (lnum < submatch_mmatch->endpos[no].lnum)
		{
		    s = reg_getline(lnum++);
		    if (round == 2)
			STRCPY(retval + len, s);
		    len += (int)STRLEN(s);
		    if (round == 2)
			retval[len] = '\n';
		    ++len;
		}
		if (round == 2)
		    STRNCPY(retval + len, reg_getline(lnum),
					     submatch_mmatch->endpos[no].col);
		len += submatch_mmatch->endpos[no].col;
		if (round == 2)
		    retval[len] = NUL;
		++len;
	    }

	    if (retval == NULL)
	    {
		retval = lalloc((long_u)len, TRUE);
		if (retval == NULL)
		    return NULL;
	    }
	}
    }
    else
    {
	if (submatch_match->endp[no] == NULL)
	    retval = NULL;
	else
	{
	    s = submatch_match->startp[no];
	    retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s));
	}
    }

    return retval;
}
#endif