Source

htsql-charset-option / src / htsql / ctl / regress.py

Full commit
   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
#
# Copyright (c) 2006-2008, Prometheus Research, LLC
# Authors: Kirill Simonov <xi@gamma.dn.ua>,
#          Clark C. Evans <cce@clarkevans.com>
#


"""
:mod:`htsql.ctl.regress`
========================

This module implements the `regress` routine.
"""


from __future__ import with_statement
from .error import ScriptError
from .routine import Argument, Routine
from .option import (InputOption, TrainOption, PurgeOption,
                     ForceOption, QuietOption)
from .request import Request
from ..validator import (Val, AnyVal, BoolVal, StrVal, WordVal, ChoiceVal,
                         IntVal, UFloatVal, DBVal, SeqVal, MapVal, ClassVal)
from ..util import maybe, trim_doc, DB
import traceback
import StringIO
import sys
import os, os.path
import shutil
import re
import difflib
import yaml, yaml.constructor


# Indicates that a field has no default value and therefore cannot be omitted.
MANDATORY_FIELD = object()

# Return values for `ask_*` methods indicating the user-chosen action.
DO_CONTINUE = object()
DO_DISCARD = object()
DO_HALT = object()
DO_RECORD = object()
DO_SAVE = object()
DO_SKIP = object()


class TermStringIO(StringIO.StringIO):
    """
    A readable file-like object with an "echo".  Whenever some content is read
    from it, the same content is echoed to the specified `output` stream.

    Use :class:`TermStringIO` to preserve the content of interactive
    sessions with pre-recorded input.  Assign::

        sys.stdout = StringIO.StringIO()
        sys.stdin = TermStringIO(input, sys.stdout)

    where `input` contains the pre-recorded input data.  After the
    session is done, the content of `sys.stdout` will be the same as
    if the session was performed on a real terminal with echo enabled.

    `buf` (a string)
        The content of the stream.

    `output` (a writable file-like object)
        A stream that records data being read.
    """

    def __init__(self, buf, output):
        StringIO.StringIO.__init__(self, buf)
        self.output = output

    def read(self, n=-1):
        data = StringIO.StringIO.read(self, n)
        self.output.write(data)
        return data

    def readline(self, length=None):
        data = StringIO.StringIO.readline(self, length)
        self.output.write(data)
        return data



class Field(object):
    """
    Describes a parameter of test data.

    `attribute` (a string)
        The name of the attribute that contains the field value.

    `validator` (:class:`htsql.validator.Val`)
        The validator for the field values.

    `default`
        The default value of the field.  If not provided, the field
        cannot be omitted.  The `is_mandatory` attribute indicates if
        the `default` value is provided.

    `hint` (a string or ``None``)
        A short one-line description of the field.
    """

    # Use it to filter out `AnyField` instances.
    is_any = False

    def __init__(self, attribute, validator,
                 default=MANDATORY_FIELD, hint=None):
        # Sanity check on the arguments.
        assert isinstance(attribute, str)
        assert re.match(r'^[a-zA-Z_][0-9a-zA-Z_]*$', attribute)
        assert isinstance(validator, Val)
        assert isinstance(hint, maybe(str))

        self.attribute = attribute
        self.validator = validator
        self.default = default
        self.is_mandatory = (default is MANDATORY_FIELD)
        self.hint = hint

    def get_hint(self):
        """
        Returns short one-line description of the field.
        """
        return self.hint

    def get_signature(self):
        """
        Returns the field name.
        """
        signature = self.attribute.replace('_', '-')
        if self.is_mandatory:
            signature += '*'
        return signature


class AnyField(object):
    """
    Indicates that test data may contain extra fields.

    Add ``AnyField()`` to the `fields` list to indicate that YAML
    representation of test data may contain some attributes not
    described by other fields.  These extra attributes will be
    silently ignored.
    """

    # Use it to filter out `AnyField` instances.
    is_any = True


class TestData(object):
    """
    Represents input or output data of a test case.

    This is an abstract class.  Create a subclass of :class:`TestData`
    to describe input or output data for a specific test kind.  You need
    to specify the format of test data using the `fields` class attribute.

    The `fields` attribute is a list of :class:`Field` instances.  Each
    field describes an attribute of test data.

    Instances if :class:`TestData` are YAML-serializable.  A instance of
    a :class:`TestData` subclass is represented as a mapping YAML node.
    The sets of keys and the format of the values come from the `fields`
    list.  Add an :class:`AnyField` instance to `fields` to indicate
    that the mapping node may contain some extra fields (which are to
    be ignored).

    The constructor of :class:`TestData` accepts the following arguments:

    `routine` (:class:`RegressRoutine`)
        The routine that started the testing.

    `case_class` (a subclass of :class:`TestCase`)
        A test type.  The object being constructed is an instance
        of either `case_class.Input` or `case_class.Output`.

    `attributes` (a dictionary)
        A dictionary of attributes and their values.  The set of
        attributes is declared using the `fields` class variable.

    `location` (a string or ``None``)
        When the test data is loaded from a YAML file, `location`
        indicates the location of the corresponding YAML node.
    """

    fields = []

    def __init__(self, routine, case_class, attributes, location=None):
        # Sanity check on the arguments.
        assert isinstance(routine, RegressRoutine)
        assert issubclass(case_class, TestCase)
        assert self.__class__ in [case_class.Input, case_class.Output]
        assert isinstance(attributes, dict)
        assert isinstance(location, maybe(str))

        self.routine = routine
        self.case_class = case_class
        for name in attributes:
            setattr(self, name, attributes[name])
        self.location = location
        self.init_attributes()

    def init_attributes(self):
        """
        Normalize field values.
        """
        # Override in a subclass if you need to massage some field values.

    def __str__(self):
        # Produces the value of the first mandatory field.
        title_attribute = None
        for field in self.fields:
            if field.is_any:
                continue
            if field.is_mandatory:
                title_attribute = field.attribute
        if title_attribute is None:
            return ''
        return repr(getattr(self, title_attribute))

    def __repr__(self):
        return "<%s.%s %s>" % (self.case_class.__name__,
                               self.__class__.__name__, self)


class TestCase(object):
    """
    Describes a test type.

    This an abstract class.  Create a subclass of :class:`TestCase`
    to describe a new type of test case.  When subclassing, define
    the following class attributes:

    `name` (a string)
        The name of the test.

    `hint` (a string)
        Short one-line description of the test.

    `help` (a string)
        Long description of the test.

    `Input` (a subclass of :class:`TestData`)
        The format of the test input.

    `Output` (a subclass of :class:`TestData` or ``None``)
        The format of the test output.

    You also need to override methods :meth:`verify` and :meth:`train`
    to specify how to execute the test case in a normal and in a train mode.

    The constructor of :class:`TestCase` takes the following arguments:

    `routine` (:class:`RegressRoutine`)
        The routine that started the testing.

    `state`
        An object keeping the mutable testing state.

    `input` (an instance of `Input`)
        Input test data.

    `output` (an instance of `Output` or ``None``)
        Expected output test data.
    """

    name = None
    hint = None
    help = None

    # Override to declare the format of input and output test data.
    Input = None
    Output = None

    @classmethod
    def get_hint(cls):
        """
        Returns short one-line description of the test case.
        """
        return cls.hint

    @classmethod
    def get_help(cls):
        """
        Returns long description of the test case.
        """
        # Produce:
        # {help}
        # 
        # Input data:
        #   {field.signature} - {field.hint}
        #   ...
        #
        # Output data:
        #   {field.signature} - {field.hint}
        #   ...
        lines = []
        help = trim_doc(cls.help)
        if help is not None:
            lines.append(help)
        for data_class in [cls.Input, cls.Output]:
            if data_class is None:
                continue
            if lines:
                lines.append("")
            lines.append("%s data:" % data_class.__name__)
            for field in data_class.fields:
                if field.is_any:
                    continue
                signature = field.get_signature()
                hint = field.get_hint()
                if hint is not None:
                    lines.append("  %-24s : %s" % (signature, hint))
                else:
                    lines.append("  %s" % signature)
        return "\n".join(lines)

    def __init__(self, routine, state, input, output):
        # Sanity check on the arguments.
        assert isinstance(routine, RegressRoutine)
        assert isinstance(state, routine.state_class)
        if self.Input is None:
            assert input is None
        else:
            assert isinstance(input, self.Input)
        if self.Output is None:
            assert output is None
        else:
            assert isinstance(output, maybe(self.Output))

        self.routine = routine
        self.state = state
        self.input = input
        self.output = output

        # When the test case is in the quiet mode (indicated by `is_quiet`),
        # all output is redirected to `quiet_buffer`.  If for some reason
        # the test case leaves the quiet mode, all the accumulated data
        # is dumped to the standard output stream.
        self.is_quiet = routine.quiet
        self.quiet_buffer = StringIO.StringIO()

    def make_output(self, **attributes):
        # Generate a new test output record with the given attributes.
        return self.Output(self.routine, self.__class__, attributes)

    @classmethod
    def matches(cls, input, output):
        """
        Checks if the given input and output records belong to the same
        test case.

        Note that we assume that both test input and test output have
        a field with the same attribute name.  This attribute is called
        the key attribute.  Input data matches output data when the
        values of their key attribute are equal.
        """
        # Sanity check on the arguments.
        assert isinstance(input, maybe(TestData))
        assert isinstance(output, maybe(TestData))

        # `input` and `output` must be instances of `Input` and `Output`
        # classes of the test case.
        if cls.Input is None or cls.Output is None:
            return False
        if not isinstance(input, cls.Input):
            return False
        if not isinstance(output, cls.Output):
            return False

        # Find the key attribute: one that is declared both as an input field
        # and as an output field.
        key_attribute = None
        input_attributes = [field.attribute for field in cls.Input.fields
                                            if not field.is_any]
        output_attributes = [field.attribute for field in cls.Output.fields
                                             if not field.is_any]
        for attribute in input_attributes:
            if attribute in output_attributes:
                key_attribute = attribute
                break
        if key_attribute is None:
            return False

        # `input` and `output` are matched when the values of their key
        # attributes are equal.
        if getattr(input, key_attribute) != getattr(output, key_attribute):
            return False
        return True

    def get_suites(self):
        """
        For container test cases, returns a set of test suites that belong
        to the test case; otherwise returns an empty set.
        """
        return set()

    def out(self, *values, **options):
        """
        Print values to the standard output stream.

        :meth:`out` supports the same options as
        :meth:`htsql.ctl.script.Script.out` and an extra option:

        `indent`
            A number of spaces to print before the first value,
            default is ``0``.
        """
        indent = options.pop('indent', 0)
        if indent:
            values = (' '*(indent-1),) + values
        # If the test case is in the quiet mode, redirect the output
        # to `quiet_buffer`.
        if self.is_quiet and 'file' not in options:
            options['file'] = self.quiet_buffer
        self.routine.ctl.out(*values, **options)

    def ask(self, message, choices):
        """
        Asks the user a question; returns the reply.

        `message` (a string)
            The question.

        `choices` (a list of strings)
            The list of valid replies.

        Typically the question has the form::
        
            Press ENTER to perform <the default action>,
                  'x'+ENTER to perform <another action>,
                  'y'+ENTER to perform <another action>,
                  'z'+ENTER to perform <another action>.

        In this case, `choices` should be equal to::

            ['', 'x', 'y', 'z']

        The reply is stripped of leading and trailing whitespaces
        and translated to the lower case.
        """
        # Leave the quiet mode and print the question.
        self.force_out()
        self.out()
        self.out(">>>", message)
        line = None

        # Repeat till we get a valid answer.
        while line not in choices:
            self.out("> ", end='')
            line = self.routine.ctl.stdin.readline().strip().lower()

        return line

    def ask_halt(self):
        """
        Ask if the user wants to halt the tests.

        Returns `DO_HALT` or `DO_CONTINUE`.
        """
        line = self.ask("Press ENTER to halt,"
                        " 'c'+ENTER to continue", ['', 'c'])
        if line == '':
            return DO_HALT
        if line == 'c':
            return DO_CONTINUE

    def ask_record(self):
        """
        Ask if the user wants to remember the new output of a test case.

        Returns `DO_RECORD`, `DO_SKIP`, or `DO_HALT`.
        """
        line = self.ask("Press ENTER to record,"
                        " 's'+ENTER to skip,"
                        " 'h'+ENTER to halt", ['', 's', 'h'])
        if line == '':
            return DO_RECORD
        if line == 's':
            return DO_SKIP
        if line == 'h':
            return DO_HALT

    def ask_save(self):
        """
        Ask if the user wants to save the updated output data.

        Returns `DO_SAVE` or `DO_DISCARD`.
        """
        line = self.ask("Press ENTER to save changes,"
                        " 'd'+ENTER to discard changes", ['', 'd'])
        if line == '':
            return DO_SAVE
        if line == 'd':
            return DO_DISCARD

    def out_exception(self, exc_info):
        """
        Prints an exception traceback.
        """
        # Obey the quiet mode: redirect to `quiet_buffer` if necessary.
        if self.is_quiet:
            file = self.quiet_buffer
        else:
            file = self.routine.ctl.stdout

        exc_type, exc_value, exc_traceback = exc_info
        traceback.print_exception(exc_type, exc_value, exc_traceback,
                                  file=file)

    def out_sep(self, sep="-", length=72):
        """
        Prints a separator: a long line of dashes.
        """
        self.out(sep*length)

    def out_header(self):
        """
        Prints a nice header describing the test case.
        """
        # Print:
        # ---------------- ... -
        #   {NAME} {value}
        #   ({input.location})
        # where {value} is the value of the first field of the input data.
        self.out_sep()
        if not self.input.fields or self.input.fields[0].is_any:
            return
        attribute = self.input.fields[0].attribute
        value = getattr(self.input, attribute)
        if value is not None:
            self.out("%s %s" % (self.name.upper(), value), indent=2)
        if self.input.location is not None:
            self.out("(%s)" % self.input.location, indent=2)

    def halted(self, message=None):
        """
        Indicate that the test case failed and stop the tests.
        """
        self.force_out()
        if message is not None:
            self.out(message)
        self.state.failed += 1
        self.state.is_exiting = True

    def failed(self, message=None):
        """
        Indicate that the test case failed; stop the tests unless
        ``--force`` or ``--train`` flags are set.
        """
        self.force_out()
        if message is not None:
            self.out(message)
        self.state.failed += 1
        if not (self.routine.force or self.routine.train):
            self.state.is_exiting = True

    def updated(self, message=None):
        """
        Indicate that the output of the test case has been updated.
        """
        self.force_out()
        if message is not None:
            self.out(message)
        self.state.updated += 1

    def passed(self, message=None):
        """
        Indicate that the test case passed.
        """
        if message is not None:
            self.out(message)
        self.state.passed += 1

    def force_out(self):
        # Leave the quiet mode; flush the content of `quiet_buffer`
        # to the standard output stream.
        if not self.is_quiet:
            return
        self.is_quiet = False
        buffer = self.quiet_buffer.getvalue()
        self.routine.ctl.stdout.write(buffer)
        self.routine.ctl.stdout.flush()

    def verify(self):
        """
        Executes the test case.

        This method runs the test case with the given input data.
        If the test completed without errors, compare the produced
        output with the given expected output.

        The test case fails if

        - the test failed to complete without errors;
        - or the expected test output is not provided;
        - or the expected test output is not equal to the actual test output.

        Some test cases may not generate output; in this case the test
        passes if it is completed without errors.
        """
        # Override when subclassing.
        raise ScriptError("test %r is not implemented" % self.name)

    def train(self):
        """
        Executes the test case in the training mode; returns the output data.

        In the train mode, when the expected test output is not equal to the
        actual test output, the user is given a choice to update the expected
        test output.

        Note that when the output has not been changed or the user refused
        to update it, the method must return the original output data,
        ``self.output``.
        """
        # Override when subclassing if the test case requires test output data.
        # Otherwise, just run the test case in the normal mode.
        self.verify()
        return None


class RunAndCompareTestCase(TestCase):
    """
    Implements common methods for a broad category of test cases.

    This class implements common scenario: run the test, get the output
    and compare it with the expected output.

    This is an abstract class; create a subclass to implement a concrete
    test case.  The following methods has to be overriden: :meth:`execute`,
    :meth:`render` and :meth:`differs`.
    """

    def out_lines(self, lines, indent=0):
        """
        Prints the lines with the specified identation.
        """
        for line in lines:
            # If `line` is UTF-8 encoded, print it literally;
            # otherwise, replace special and non-ASCII characters
            # with dots.
            try:
                line.decode('utf-8')
            except UnicodeDecodeError:
                line = re.sub(r'[\x00-\x1F\x7E-\xFF]', '.', line)
            self.out(line.rstrip(), indent=indent)

    def out_diff(self, old_output, new_output):
        """
        Prints the delta between two test outputs.
        """
        # Sanity check on the arguments.
        assert isinstance(old_output, maybe(self.Output))
        assert isinstance(new_output, self.Output)

        # Render the outputs to the lists of lines.
        old_lines = self.render(old_output)
        new_lines = self.render(new_output)

        # This function is supposed to be called in two cases:
        # when there is no expected output, but only the actual output,
        # and when the expected output differs from the actual output.
        # However it may also happen that the function is called with
        # two identical outputs, or that the `render` method hides
        # the difference.
        if old_lines is None:
            self.out("=== the test output is new")
        elif old_lines != new_lines:
            self.out("=== the test output is changed")
        else:
            self.out("=== the test output is not changed")
        self.out()

        # Display the actual output if there is no expected output;
        # otherwise display the delta between the expected and the actual
        # output in the unified diff format.
        if old_lines is None or old_lines == new_lines:
            lines = new_lines
        else:
            diff = difflib.unified_diff(old_lines, new_lines,
                                        n=2, lineterm='')
            # Strip the leading `---` and `+++` lines of the unified diff.
            lines = list(diff)[2:]
        self.out_lines(lines, indent=2)

    def render(self, output):
        """
        Converts the output data to a list of lines.
        """
        # Override when subclassing.
        raise NotImplementedError()

    def execute(self):
        """
        Runs the test case; returns the produced output.

        Returns ``None`` if an error occured when running the test case.
        """
        # Override when subclassing.
        raise NotImplementedError()

    def differs(self, old_output, new_output):
        """
        Checks if the actual test output differs from the expected test output.
        """
        # Override when subclassing.
        raise NotImplementedError()

    def verify(self):
        # Display the header.
        self.out_header()

        # When no expected test output, fail without executing the test.
        if self.output is None:
            return self.failed("*** no output data found")

        # Execute the test; get the actual test output.
        new_output = self.execute()

        # `None` indicates that an error occurred; `execute()` is responsible
        # for displaying an error message, so we just update the status and
        # exit.
        if new_output is None:
            return self.failed()

        # Compare the expected and the actual outputs, fail if they are
        # different.
        if self.differs(self.output, new_output):
            self.out_diff(self.output, new_output)
            return self.failed("*** unexpected test output")

        # The actual output coincides with the expected output; we are good.
        return self.passed()

    def train(self):
        # Display the header.
        self.out_header()

        # Execute the test; get the actual test output.
        new_output = self.execute()

        # We need to handle three possible outcomes: an error occurred
        # when running the test, the expected output differs from the
        # actual output and the expected output coincides with the actual
        # output.

        # An error occurred while running the test.
        if new_output is None:
            # Ask the user if they want to stop the testing; the expected
            # output is not updated.
            reply = self.ask_halt()
            if reply is DO_HALT:
                self.halted("*** halting")
            else:
                self.failed()
            return self.output

        # The actual output differs from the expected output.
        if self.differs(self.output, new_output):
            # Display the difference.
            self.out_diff(self.output, new_output)
            # Ask the user if they want to record the new output,
            # keep the old output, or halt the testing.
            reply = self.ask_record()
            if reply is DO_HALT:
                self.halted("*** halting")
                return self.output
            if reply is DO_RECORD:
                if self.output is None:
                    self.updated("*** recording new test output")
                else:
                    self.updated("*** recording updated test output")
                return new_output
            self.failed()
            return self.output

        # The actual output coincides with the expected output; note that
        # the caller checks if ``case.train() is case.output`` to learn
        # if the output is updated.
        self.passed()
        return self.output


class SkipTestCase(TestCase):
    """
    Disables an existing test case.
    """

    name = "skip"
    hint = """disable a test case"""
    help = """
    Add a `skip` field to any input test record to disable the test.
    """

    class Input(TestData):
        fields = [
                Field('skip', AnyVal(),
                      hint="""indicates that the test is disabled"""),
                AnyField(),
        ]

    def verify(self):
        return 


class AppTestCase(TestCase):
    """
    Configures the HTSQL application.
    """

    name = "app"
    hint = """configure the HTSQL application"""
    help = """
    To run HTSQL requests, the testing engine needs to create an HTSQL
    application.  This test case allows you to configure the application
    parameters.
    """

    class Input(TestData):
        fields = [
                Field('db', DBVal(),
                      hint="""the connection URI"""),
        ]

    def out_header(self):
        # Overriden to avoid printing the password to the database.

        # Clone `input.db`, but omit the password.
        db = self.input.db
        sanitized_db = DB(engine=db.engine,
                          username=db.username,
                          password=None,
                          host=db.host,
                          port=db.port,
                          database=db.database,
                          options=db.options)

        # Print:
        # ---------------- ... -
        #   APP {sanitized_db}
        #   ({input.location})
        self.out_sep()
        self.out("%s %s" % (self.name.upper(), sanitized_db), indent=2)
        if self.input.location is not None:
            self.out("(%s)" % self.input.location, indent=2)

    def verify(self):
        # Display the header.
        self.out_header()

        # Create an application and update the testing state.  The created
        # application will be in effect for the subsequent tests in the
        # current suite and all the nested suites unless overriden.
        from htsql.application import Application
        self.state.app = None
        try:
            self.state.app = Application(self.input.db)
        except Exception, exc:
            self.out_exception(sys.exc_info())
            return self.failed("*** an exception occured while"
                               " initializing an HTSQL application")

        return self.passed()


class IncludeTestCase(TestCase):
    """
    Loads input test data from a file.
    """

    name = "include"
    hint = """load input data from a file"""
    help = """
    This test case allows you to execute a test case or a test suite defined
    in a separate file.
    """

    class Input(TestData):
        fields = [
                Field('include', StrVal(),
                      hint="""file containing input test data"""),
        ]

    class Output(TestData):
        fields = [
                Field('include', StrVal(),
                      hint="""file containing input test data"""),
                Field('output', SeqVal(ClassVal(TestData)),
                      hint="""the corresponding output test data"""),
        ]

    def __init__(self, routine, state, input, output):
        super(IncludeTestCase, self).__init__(routine, state, input, output)

        # Load the input data and create the corresponding test case.
        self.included_input = routine.load_input(self.input.include)
        case_class = self.included_input.case_class
        self.included_output = None
        if self.output is not None:
            if case_class.matches(self.included_input, self.output.output):
                self.included_output = self.output.output
        self.case = case_class(routine, state,
                               self.included_input,
                               self.included_output)

    def get_suites(self):
        # Get the set of nested suites.
        return self.case.get_suites()

    def verify(self):
        # Run the included test.
        self.case.verify()

    def train(self):
        # Run the included test; get the output.
        new_output = self.case.train()

        # Three outcomes are possible: the test generated no output, in this
        # case we don't need to create an output record either; the test
        # generated new or updated output, we have to update our output as
        # well; and finally, the test output didn't change, we could keep
        # ours too.
        if new_output is None:
            output = None
        elif new_output is not self.included_output:
            output = self.make_output(include=self.input.include,
                                      output=new_output)
        else:
            output = self.output
        return output


class SuiteTestCase(TestCase):
    """
    Implements a container of test cases.
    """

    name = "suite"
    hint = """contains other test cases"""
    help = """
    A test suite is a container of test cases.  Typically, it is the
    top-level test case in a test file.
    
    The testing engine allows you to specify what suites to run by their
    ids.
    """

    class Input(TestData):
        fields = [
                Field('title', StrVal(),
                      hint="""the description of the suite"""),
                Field('id', StrVal(), None,
                      hint="""the code name of the suite"""),
                Field('output', StrVal(), None,
                      hint="""file to save the output of the tests"""),
                Field('tests', SeqVal(ClassVal(TestData)),
                      hint="""a list of test inputs"""),
        ]

        def init_attributes(self):
            # When `id` is not specified, generate it from the title.
            if self.id is None:
                self.id = self.title.lower().replace(' ', '-')

    class Output(TestData):
        fields = [
                Field('id', StrVal(),
                      hint="""the code name of the suite"""),
                Field('tests', SeqVal(ClassVal(TestData)),
                      hint="""a list of test outputs"""),
        ]

    def __init__(self, routine, state, input, output):
        super(SuiteTestCase, self).__init__(routine, state, input, output)

        # A test suite has an ability to save its test output to a separate
        # file.  In this case, `self.ext_output` contains the test data
        # loaded from the file.
        self.ext_output = None
        if input.output is not None and os.path.exists(input.output):
            ext_output = routine.load_output(input.output)
            if self.matches(input, ext_output):
                self.ext_output = ext_output

        # Generate a list of test cases.
        self.cases = []
        self.cases_state = TestState()
        self.init_cases()

    def init_cases(self):
        # Generate a list of test cases.  We have two independent lists:
        # one containing input test records and the other containing
        # output test records.  Our goal is to find matching pairs and
        # generate the corresponding test cases.

        # The matching pairs of input and output data.
        pairs = []

        # List of available output records.  We need to copy it since
        # it is going to be modified.
        available_outputs = []
        if self.ext_output is not None:
            available_outputs = self.ext_output.tests[:]
        elif self.output is not None:
            available_outputs = self.output.tests[:]

        # For each input record, find the matching output record.
        for input in self.input.tests:
            case_class = input.case_class
            for idx, output in enumerate(available_outputs):
                if case_class.matches(input, output):
                    pairs.append((input, output))
                    del available_outputs[idx]
                    break
            else:
                pairs.append((input, None))

        # Initialize the test cases.
        for input, output in pairs:
            case_class = input.case_class
            case = case_class(self.routine, self.cases_state, input, output)
            self.cases.append(case)

    def get_suites(self):
        # Get a set of (this and) the nested suites.
        suites = set([self.input.id])
        for case in self.cases:
            suites |= case.get_suites()
        return suites

    def out_header(self):
        # Print the header:
        # ================ ... =
        #   {input.title}
        #   ({input.location})
        self.out_sep("=")
        self.out(self.input.title, indent=2)
        if self.input.location is not None:
            self.out("(%s)" % self.input.location, indent=2)

    def skipped(self):
        # Check if the suite should not be executed.

        # The suite is skipped when:
        # - the user specified an explicit list of the suites to run;
        # - and the suite is not one of them;
        # - and the suite does not contain any selected nested suite;
        # - and the suite is not nested in some selected suite.
        if not self.routine.suites:
            return False
        if self.state.with_all_suites:
            return False
        if self.input.id in self.routine.suites:
            self.cases_state.with_all_suites = True
            return False
        if self.get_suites() & set(self.routine.suites):
            return False
        return True

    def verify(self):
        # Run the suite.

        # Push the current state to the cases state.
        self.state.push(self.cases_state)
        # Check if the user specified the suites to run and
        # this one is not among them.
        if self.skipped():
            return
        # Display the headers.
        self.out_header()
        # Run the nested test cases.
        for case in self.cases:
            case.verify()
            # Check if the user asked to halt the testing.
            if self.cases_state.is_exiting:
                break
        # Pull the statistical information from the cases state.
        self.state.pull(self.cases_state)

    def train(self):
        # Run the suite; update the test output if necessary.

        # Push the current state to the cases state.
        self.state.push(self.cases_state)
        # Check if the user specified the suites to run and
        # this one is not among them.
        if self.skipped():
            return self.output
        # A dictionary containing the output (or `None`) generated by test
        # cases when it differs from the existing test output.
        new_output_by_case = {}
        # Display the header.
        self.out_header()
        # Run the nested tests.
        for case in self.cases:
            new_output = case.train()
            # Record modified output data.
            if new_output is not case.output:
                new_output_by_case[case] = new_output
            # Check if the user asked to halt the testing.
            if self.cases_state.is_exiting:
                break
        # Pull the statistical information from the cases state.
        self.state.pull(self.cases_state)
        # Generate a new output record.
        output = self.make_output(new_output_by_case)
        # The output is kept in a separate file.
        if self.input.output is not None:
            # If the output has been updated, ask the user if they want
            # to save it.
            if output is not self.ext_output:
                self.out_sep()
                reply = self.ask_save()
                if reply is DO_DISCARD:
                    # `self.output` may still be not ``None`` if the `output`
                    # field was recently added.  In that case, we don't want
                    # to delete the regular output data until it is saved
                    # to a separate file.
                    return self.output
                self.out("*** saving test output data to %r"
                         % self.input.output)
                self.routine.save_output(self.input.output, output)
            # Returning `None` since the output is saved to a separate file.
            return None
        return output

    def make_output(self, new_output_by_case):
        # Generate the output test data.

        # Here we update the list of output test records.  Note that the list
        # may contain some inactive output records.  These output records
        # do not correspond to any input records and thus have no respective
        # test case.  It may happen if the user removed or modified the input
        # data.  Since a test case may be only temporarily disabled, we never
        # remove inactive output records unless the `--purge` option is enabled.
        
        # The list of the output records.
        tests = []

        # Start with the original list of output records.
        if self.output is not None:
            tests = self.output.tests[:]
        if self.ext_output is not None:
            tests = self.ext_output.tests[:]

        # `--purge` is enabled, we don't have to keep inactive records,
        # so simply generate the list from scratch.
        if self.routine.purge and not self.state.is_exiting:
            tests = []
            for case in self.cases:
                output = case.output
                if case in new_output_by_case:
                    output = new_output_by_case[case]
                if output is not None:
                    tests.append(output)

        # Some test cases generated new output, so we need to update the list.
        elif new_output_by_case:

            # Here we take the original list of records and replace those
            # that have been updated.  We may also encounter a new output
            # record, which has no corresponding old record in the list.
            # For that new record, we need to find a position in the list.
            # We want the order of the output records to match the order
            # of their respective input records, so to ensure this, we
            # put any new record immediately after all other records processed
            # so far.

            # Position to put new records.
            next_idx = 0
            for case in self.cases:
                # The record has been added, removed or updated.
                if case in new_output_by_case:
                    new_output = new_output_by_case[case]
                    # The record is rarely entirely removed so we should almost
                    # never get ``None`` here.  If we do, do nothing.
                    if new_output is not None:
                        # This is an updated record: replace the old record
                        # and update the position for the following new
                        # records.
                        if case.output in tests:
                            idx = tests.index(case.output)
                            tests[idx] = new_output
                            if idx >= next_idx:
                                next_idx = idx+1
                        # This is a new record: place it to the designated
                        # position.
                        else:
                            tests.insert(next_idx, new_output)
                            next_idx += 1

                # The record has not been changed.
                else:
                    # Make sure any new record will go after this one.
                    if case.output in tests:
                        idx = tests.index(case.output)
                        if idx >= next_idx:
                            next_idx = idx+1

        # When there are no test output data, skip creating the output record.
        if not tests:
            return None

        # Now we need to check if the new output list coincides with the old
        # one, in which case we don't want to create a new output record.
        if self.input.output is not None:
            if self.ext_output is not None and self.ext_output.tests == tests:
                return self.ext_output
        else:
            if self.output is not None and self.output.tests == tests:
                return self.output

        # Generate and return new output data.
        output = super(SuiteTestCase, self).make_output(id=self.input.id,
                                                        tests=tests)
        return output


class QueryTestCase(RunAndCompareTestCase):
    """
    Performs an HTSQL query.
    """

    name = "query"
    hint = """execute an HTSQL query"""
    help = """
    This test case executes an HTSQL query.
    """

    class Input(TestData):
        fields = [
                Field('uri', StrVal(),
                      hint="""the HTSQL query"""),
                Field('method', ChoiceVal(['GET', 'POST']), 'GET',
                      hint="""the HTTP method (GET or POST)"""),
                Field('remote_user', StrVal(), None,
                      hint="""the HTTP remote user"""),
                Field('headers', MapVal(StrVal(), StrVal()), None,
                      hint="""the HTTP headers"""),
                Field('content_type', StrVal(), None,
                      hint="""the content type of HTTP POST data"""),
                Field('content_body', StrVal(), None,
                      hint="""the HTTP POST data"""),
                Field('expect', IntVal(), 200,
                      hint="""the HTTP status code to expect"""),
                Field('ignore', BoolVal(), False,
                      hint="""ignore the response body"""),
        ]

        def init_attributes(self):
            # Check that `content-type` and `content-body` are set only if
            # the HTTP method is `POST`.
            if self.method == 'GET':
                if self.content_type is not None:
                    raise ValueError("unexpected content-type parameter"
                                     " for a GET request")
                if self.content_body is not None:
                    raise ValueError("unexpected content-body parameter"
                                     " for a GET request")
            if self.method == 'POST':
                if self.content_body is None:
                    raise ValueError("no expected content-body parameter"
                                     " for a POST request")


    class Output(TestData):
        fields = [
                Field('uri', StrVal(),
                      hint="""the HTSQL query"""),
                Field('status', StrVal(),
                      hint="""the response status line"""),
                Field('headers', SeqVal(SeqVal(StrVal(), length=2)),
                      hint="""the response headers"""),
                Field('body', StrVal(),
                      hint="""the response body"""),
        ]

        def init_attributes(self):
            # Convert the list of two-element lists to a list of pairs.
            self.headers = [(key, value) for key, value in self.headers]

    def out_header(self):
        # Display the header:
        # ---------------- ... -
        #   {method} {uri}
        #   ({input.location})
        #   Remote-User: {remote_user}
        #   {header}: value
        #   ...
        #   Content-Type: {content_type}
        #
        #   {content_body}
        self.out_sep()
        self.out("%s %s" % (self.input.method, self.input.uri), indent=2)
        self.out("(%s)" % self.input.location, indent=2)
        if self.input.remote_user is not None:
            self.out("Remote-User: %s" % self.input.remote_user, indent=2)
        if self.input.headers:
            for key in sorted(self.input.headers):
                value = self.input.headers[key]
                self.out("%s: %s" % (key, value), indent=2)
        if self.input.content_type is not None:
            self.out("Content-Type: %s" % self.input.content_type, indent=2)
        self.out()
        if self.input.content_body:
            self.out_lines(self.input.content_body.splitlines(), indent=2)

    def differs(self, old_output, new_output):
        # Check if the actual output differs from the expected output.
        if old_output is None or new_output is None:
            return True
        if old_output.status != new_output.status:
            return True
        if old_output.headers != new_output.headers:
            return True
        if not self.input.ignore:
            if old_output.body != new_output.body:
                return True
        return False

    def render(self, output):
        # Convert the output record to a list of lines.
        if output is None:
            return None
        lines = []
        lines.append(output.status)
        for header, value in output.headers:
            lines.append("%s: %s" % (header, value))
        lines.append("")
        lines.extend(output.body.splitlines())
        return lines

    def execute(self):
        # Execute the query; return the output.

        # Prepare the HTSQL application.
        app = self.state.app
        if app is None:
            return self.failed("*** no HTSQL application is defined")

        # Prepare and execute the query.
        request = Request.prepare(method=self.input.method,
                                  query=self.input.uri,
                                  remote_user=self.input.remote_user,
                                  content_type=self.input.content_type,
                                  content_body=self.input.content_body,
                                  extra_headers=self.input.headers)
        response = request.execute(app)

        # Check if the response is valid.
        if response.exc_info is not None:
            self.out_exception(response.exc_info)
            return self.out("*** an exception occured"
                            " while executing the query")
        if not response.complete():
            return self.out("*** the response is not complete")

        # Generate the output record.
        new_output = self.make_output(uri=self.input.uri,
                                      status=response.status,
                                      headers=response.headers,
                                      body=response.body)

        # Check if we get the expected status code (200, by default).
        # If not, display the response and discard the output.
        if not response.status.startswith(str(self.input.expect)):
            self.out_diff(self.output, new_output)
            return self.out("*** unexpected status code: %s"
                            % response.status)

        return new_output


class CtlTestCase(RunAndCompareTestCase):
    """
    Executes a script routine.
    """

    name = "ctl"
    hint = """execute a routine"""
    help = """
    This test case simulates a run of the HTSQL command-line application.
    """

    class Input(TestData):
        fields = [
                Field('ctl', SeqVal(StrVal()),
                      hint="""a list of command-line parameters"""),
                Field('stdin', StrVal(), '',
                      hint="""the content of the standard input"""),
                Field('expect', IntVal(), 0,
                      hint="""the exit code to expect"""),
                Field('ignore', BoolVal(), False,
                      hint="""ignore the exit code and the standard output"""),
        ]

    class Output(TestData):
        fields = [
                Field('ctl', SeqVal(StrVal()),
                      hint="""a list of command-line parameters"""),
                Field('stdout', StrVal(),
                      hint="""the content of the standard output"""),
                Field('exit', IntVal(),
                      hint="""the exit code"""),
        ]

    def out_header(self):
        # Display the header:
        # ---------------- ... -
        #   {EXECUTABLE} {ctl}
        #   ({input.location})
        self.out_sep()
        executable = os.path.basename(self.routine.executable)
        command_line = " ".join([executable.upper()]+self.input.ctl)
        self.out(command_line, indent=2)
        self.out("(%s)" % self.input.location, indent=2)

    def differs(self, old_output, new_output):
        # Check if the actual output differs from the expected output.
        if old_output is None or new_output is None:
            return True
        if not self.input.ignore:
            if old_output.exit != new_output.exit:
                return True
            if old_output.stdout != new_output.stdout:
                return True
        return False

    def render(self, output):
        # Convert the output to a list of lines.
        if output is None:
            return None
        return output.stdout.splitlines()

    def execute(self):
        # Run the routine; return the output

        # Prepare the standard streams and the script instance.
        stdout = StringIO.StringIO()
        stderr = stdout
        stdin = TermStringIO(stdout, stderr)
        command_line = [self.routine.executable]+self.input.ctl

        # The script class.
        ctl_class = self.routine.ctl.__class__

        # Initialize and execute the script; check for exceptions.
        try:
            ctl = ctl_class(stdin, stdout, stderr)
            exit = ctl.main(command_line)
        except:
            self.out_exception(sys.exc_info())
            return self.out("*** an exception occured"
                            " while running the application")

        # Normalize the exit code.
        if exit is None:
            exit = 0
        elif not isinstance(exit, int):
            stderr.write(str(exit))
            exit = 1

        # Generate a new output record.
        new_output = self.make_output(ctl=self.input.ctl,
                                      stdout=stdout.getvalue(),
                                      exit=exit)

        # Check if we get the expected exit code; if not, display
        # the content of stdout and discard the output record.
        if not self.input.ignore:
            if new_output.exit != self.input.expect:
                self.out_diff(self.output, new_output)
                return self.out("*** unexpected exit code: %s" % exit)

        return new_output


class Fork(object):
    """
    Keeps information on the started processes.

    Class attributes:

    `active_forks`
        The global list of active processes.

    `is_atexit_registered`
        Indicates whether an :func:`atexit.atexit` callable was registered.
        The callable is called when the script is about to finish and kills
        any remaining active processes.

    Attributes:

    `process` (an instance of :class:`subprocess.Popen`)
        The wrapped process.

    `temp_path` (a string)
        A directory containing two files: `input` and `output`, which
        keeps the content of the standard input and the standard output
        respectively.
    """

    active_forks = []
    is_atexit_registered = False

    @classmethod
    def start(cls, executable, arguments, input):
        """
        Starts a new process.

        `executable`
            The path to the executable.

        `arguments`
            The list of arguments (not including the executable).

        `input`
            The content of the standard input.

        Returns a new :class:`Fork` instance.
        """
        # Create a temporary directory with the files 'input' and 'output'.
        temp_path = tempfile.mkdtemp()
        stream = open("%s/input" % temp_path, 'wb')
        stream.write(input)
        stream.close()
        # Prepare the standard input and the standard output streams.
        stdin = open("%s/input" % temp_path, 'rb')
        stdout = open("%s/output" % temp_path, 'wb')
        # Start the process.
        try:
            process = subprocess.Popen([executable]+arguments,
                                       stdin=stdin,
                                       stdout=stdout,
                                       stderr=subprocess.STDOUT)
        except:
            shutil.rmtree(temp_path)
            raise
        # Return a new `Fork` instance.
        return cls(process, temp_path)

    @classmethod
    def atexit(cls):
        # Finalize any remaining active processes.
        for fork in cls.active_forks:
            fork.end()

    @classmethod
    def atexit_register(cls):
        # Register the `atexit` callable if not done already.
        if not cls.is_atexit_registered:
            atexit.register(cls.atexit)
            cls.is_atexit_registered = True

    def __init__(self, process, temp_path):
        # Sanity check on the arguments.
        assert isinstance(process, subprocess.Popen)
        assert isinstance(temp_path, str) and os.path.isdir(temp_path)

        self.process = process
        self.temp_path = temp_path

        # Save themselves in the global list of active processes.
        self.active_forks.append(self)
        # Register the `atexit` callback.
        self.atexit_register()

    def end(self):
        """
        Ends the process.

        Returns the content of the standard output.
        """
        # Terminate the process if it is still alive.
        if self.process.poll() is None:
            # FIXME: Python 2.5 compatibility!
            self.process.terminate()
            time.sleep(1.0)
        # Read the standard output.
        stream = open("%s/output" % self.temp_path, 'rb')
        output = stream.read()
        stream.close()
        # Remove the temporary directory.
        shutil.rmtree(self.temp_path)
        # Remove it from the list of active processes.
        self.active_forks.remove(self)
        return output


class StartCtlTestCase(TestCase):
    """
    Starts a long-running routine.
    """

    name = "start-ctl"
    hint = """execute a long-running routine"""
    help = """
    This test case starts a long-running the HTSQL command-line
    application.  Use the `end-ctl` test case to finalize the application
    and check the output.
    """

    class Input(TestData):
        fields = [
                Field('start_ctl', SeqVal(StrVal()),
                      hint="""a list of command-line parameters"""),
                Field('stdin', StrVal(), '',
                      hint="""the content of the standard output"""),
                Field('sleep', UFloatVal(), 0,
                      hint="""sleep for the specified number of seconds"""),
        ]

    def verify(self):
        # Execute the test.

        # Check if an application with the same command-line parameters
        # has already been started.
        key = tuple(self.input.start_ctl)
        if key in self.state.forks:
            return self.fork("*** the application is already started")

        # Start and save the process.
        fork = Fork.start(self.routine.executable,
                          self.input.start_ctl,
                          self.input.stdin)
        self.state.forks[key] = fork


class EndCtlTestCase(RunAndCompareTestCase):
    """
    Terminates a long-running routine.
    """

    name = "end-ctl"
    hint = """terminate a long-running routine"""
    help = """
    This test case allows you to terminate a long-running routine started
    with `start-ctl`.
    """

    class Input(TestData):
        fields = [
                Field('end_ctl', SeqVal(StrVal()),
                      hint="""a list of command-line parameters"""),
                Field('ignore', BoolVal(), False,
                      hint="""ignore the exit code and the standard output"""),
        ]

    class Output(TestData):
        fields = [
                Field('end_ctl', SeqVal(StrVal()),
                      hint="""a list of command-line parameters"""),
                Field('stdout', StrVal(),
                      hint="""ignore the standard output"""),
        ]

    def differs(self, old_output, new_output):
        # Check if the actual output differs from the expected output.
        if old_output is None or new_output is None:
            return True
        if not self.input.ignore:
            if old_output.stdout != new_output.stdout:
                return True
        return False

    def render(self, output):
        # Convert the output record to a list of lines.
        if output is None:
            return None
        return output.stdout.splitlines()

    def execute(self):
        # Execute the test case.

        # Find the active process with the same command-line artguments.
        key = tuple(self.input.end_ctl)
        if key not in self.state.forks:
            return self.out("*** the application has not been started")
        fork = self.state.forks.pop(key)

        # Terminate the process; get the standard output.
        stdout = fork.end()

        # Create and return the output record.
        new_output = self.make_output(end_ctl=self.input.end_ctl,
                                      stdout=stdout)
        return new_output


class PythonCodeTestCase(RunAndCompareTestCase):
    """
    Executes arbitrary Python code.
    """

    name = "python"
    hint = """execute Python code"""
    help = """
    This test case allows you to execute arbitrary Python code.
    """

    class Input(TestData):
        fields = [
                Field('py', WordVal(),
                      hint="""the code name"""),
                Field('code', StrVal(),
                      hint="""Python code"""),
                Field('stdin', StrVal(), '',
                      hint="""the content of the standard input"""),
                Field('expect', StrVal(), None,
                      hint="""the name of an exception to expect"""),
                Field('ignore', BoolVal(), False,
                      hint="""ignore the standard output"""),
        ]

    class Output(TestData):
        fields = [
                Field('py', WordVal(),
                      hint="""the code name"""),
                Field('stdout', StrVal(),
                      hint="""the content of the standard output"""),
        ]

    def differs(self, old_output, new_output):
        # Check if the actual output differs from the expected output.
        if old_output is None or new_output is None:
            return True
        if not self.input.ignore:
            if old_output.stdout != new_output.stdout:
                return True
        return False

    def render(self, output):
        # Convert the output record to a list of lines.
        if output is None:
            return None
        return output.stdout.splitlines()

    def execute(self):
        # Execute the test case.

        # Prepare new standard streams.
        old_stdin = sys.stdin
        old_stdout = sys.stdout
        old_stderr = sys.stderr
        sys.stdin = StringIO.StringIO(self.input.stdin)
        sys.stdout = StringIO.StringIO()
        sys.stderr = sys.stdout
        # Execute the code.
        exc_info = None
        try:
            exec self.input.code in {}
        except:
            exc_info = sys.exc_info()
        # Make new output record.
        new_output = self.make_output(py=self.input.py,
                                      stdout=sys.stdout.getvalue())
        # Restore old standard streams.
        sys.stdin = old_stdin
        sys.stdout = old_stdout
        sys.stderr = old_stderr
        # An exception occured while running the code.
        if exc_info is not None:
            # Display the output and the exception
            self.out_diff(self.output, new_output)
            self.out_exception(exc_info)
            exc_name = exc_info[0].__name__
            # The exception was unexpected: discard the output.
            if self.input.expect is None or self.input.expect != exc_name:
                return self.out("*** an unexpected exception occured")
        else:
            # We didn't get the expected exception: discard the output.
            if self.input.expect is not None:
                return self.out("*** an expected exception did not occur")
        return new_output


class SQLTestCase(TestCase):
    """
    Executes a SQL query.
    """
    # TODO: Can't implement until the SQL splitter is done.

    name = "sql"
    hint = """execute a SQL statement"""
    help = """
    This test case executes one or multiple SQL statements.
    """

    class Input(TestData):
        fields = [
                Field('connect', DBVal(),
                      hint="""the connection URI"""),
                Field('sql', StrVal(),
                      hint="""the statements to execute"""),
                Field('autocommit', BoolVal(), False,
                      hint="""use the auto-commit mode"""),
                Field('ignore', BoolVal(), False,
                      hint="""ignore any errors"""),
        ]

    def out_header(self):
        # Print:
        # ---------------- ... -
        #   {first line of input.sql}
        #   ({input.location})
        self.out_sep()
        first_line = self.input.sql.split('\n', 1)[0]
        self.out(first_line, indent=2)
        if self.input.location is not None:
            self.out("(%s)" % self.input.location, indent=2)

    def verify(self):
        # Display the header.
        self.out_header()

        # Load the SQL input data.
        sql = self.load()

        # Generate an HTSQL application.  We need an application instance
        # to split the SQL data and to connect to the database, but we
        # never use it for executing HTSQL queries.
        from htsql.application import Application
        from htsql.connect import Connect, DBError
        from htsql.split_sql import SplitSQL
        try:
            app = Application(self.input.connect)
        except Exception, exc:
            self.out_exception(sys.exc_info())
            return self.failed("*** an exception occured while"
                               " initializing an HTSQL application")

        # Activate the application so that we could use the splitter
        # and the connection adapters.
        with app:
            # Realize a splitter and split the input data to individual
            # SQL statements.
            split_sql = SplitSQL()
            try:
                statements = list(split_sql(sql))
            except ValueError, exc:
                return self.failed("*** invalid SQL: %s" % exc)

            # Realize the connector and connect to the database.
            connect = Connect()
            try:
                connection = connect(with_autocommit=self.input.autocommit)
                cursor = connection.cursor()
            except DBError, exc:
                return self.failed("*** failed to connect to the database:"
                                   " %s" % exc)

            # Execute the given SQL statements.
            for statement in statements:
                try:
                    # Execute the statement in the current connection.
                    cursor.execute(statement)
                except DBError, exc:
                    # Display the statement that caused a problem.
                    for line in statement.splitlines():
                        self.out(line, indent=4)
                    # Normally, we end the test case when an error occurs,
                    # but if `ignore` is set, we just break the loop.
                    if not self.input.ignore:
                        return self.failed("*** failed to execute SQL:"
                                           " %s" % exc)
                    break

            # No error occurred while executing the SQL statements.
            else:
                # Commit the transaction unless `autocommit` mode is set.
                # Again, respect the `ignore` flag.
                if not self.input.autocommit:
                    try:
                        connection.commit()
                    except DBError, exc:
                        if not self.input.ignore:
                            return self.failed("*** failed to commit"
                                               " a transaction: %s" % exc)

            # Close the connection.  Note that we insist that connection
            # is opened and closed successfully regardless of the value
            # of the `ignore` flag.
            try:
                connection.close()
            except DBError, exc:
                return self.failed("*** failed to close the connection:"
                                   " %s" % exc)

        # If we reached that far, we passed the test.
        return self.passed()

    def load(self):
        """
        Returns the SQL data to execute.
        """
        # Override when subclassing.
        return self.input.sql


class SQLIncludeTestCase(SQLTestCase):
    """
    Loads SQL queries from a file and executes them.
    """

    name = "sql-include"
    hint = """load and execute SQL statements"""
    help = """
    This test case loads SQL statements from a file and execute them.
    """

    class Input(TestData):
        fields = [
                Field('connect', DBVal(),
                      hint="""the connection URI"""),
                Field('sql_include', StrVal(),
                      hint="""the file containing SQL statements"""),
                Field('autocommit', BoolVal(), False,
                      hint="""use the auto-commit mode"""),
                Field('ignore', BoolVal(), False,
                      hint="""ignore any errors"""),
        ]

    def out_header(self):
        # Print:
        # ---------------- ... -
        #   SQL-INCLUDE {input.sql_include}
        #   ({input.location})
        self.out_sep()
        self.out("%s %s" % (self.name.upper(), self.input.sql_include),
                 indent=2)
        if self.input.location is not None:
            self.out("(%s)" % self.input.location, indent=2)

    def load(self):
        # Load SQL from the given file.
        stream = open(self.input.sql_include, 'rb')
        sql = stream.read()
        stream.close()
        return sql


class WriteToFileTestCase(TestCase):
    """
    Writes some data to a file.
    """

    name = "write-to-file"
    hint = """write some data to a file"""
    help = None

    class Input(TestData):
        fields = [
                Field('write', StrVal(),
                      hint="""the file name"""),
                Field('data', StrVal(),
                      hint="""the data to write"""),
        ]

    def verify(self):
        # Display the header.
        self.out_header()
        # Write the data to the file.
        stream = open(input.write, 'wb')
        stream.write(input.data)
        stream.close()


class ReadFromFileTestCase(RunAndCompareTestCase):
    """
    Reads the file content.
    """

    name = "read-from-file"
    hint = """read the content of a file"""
    help = None

    class Input(TestData):
        fields = [
                Field('read', StrVal(),
                      hint="""the file name"""),
        ]

    class Output(TestData):
        fields = [
                Field('read', StrVal(),
                      hint="""the file name"""),
                Field('data', StrVal(),
                      hint="""the content of the file"""),
        ]

    def differs(self, old_output, new_output):
        # Check if the actual output differs from the expected output.
        if old_output is None or new_output is None:
            return True
        return (old_output.data != new_output.data)

    def render(self, output):
        # Convert the output record to a list of lines.
        if output is None:
            return None
        return output.data.splitlines()

    def execute(self):
        # Execute the test.

        # Check if the file exists.
        if not os.path.exists(self.input.read):
            return self.out("*** file %r does not exist" % self.input.read)

        # Read the data and create the output record.
        stream = open(self.input.read, 'rb')
        data = stream.read()
        stream.close()
        new_output = self.make_output(read=self.input.read, data=data)
        return new_output


class RemoveFilesTestCase(TestCase):
    """
    Removes the specified files.
    """

    name = "remove-files"
    hint = """remove the specified files"""
    help = """
    Remove a list of files.  It is not an error if some of the files do not
    exist.
    """

    class Input(TestData):
        fields = [
                Field('remove', SeqVal(StrVal()),
                      hint="""a list of files to remove"""),
        ]

    def verify(self):
        # Display the header.
        self.out_header()
        # Remove the given files.
        for path in self.input.remove:
            if os.path.exists(path):
                os.unlink(path)


class MakeDirTestCase(TestCase):
    """
    Creates a directory.
    """

    name = "make-dir"
    hint = """create a directory"""
    help = """
    Create a directory.  If necessary, all intermediate directories are also
    created.
    """

    class Input(TestData):
        fields = [
                Field('mkdir', StrVal(),
                      hint="""the directory name"""),
        ]

    def verify(self):
        # Display the header.
        self.out_header()
        # Create the directory if it does not already exist.
        if not os.path.isdir(self.input.mkdir):
            os.makedirs(self.input.mkdir)


class RemoveDirTestCase(TestCase):
    """
    Removes a directory.
    """

    name = "remove-dir"
    hint = """remove a directory"""
    help = """
    Removes a directory with all its content.  It is not an error if the
    directory does not exist.
    """

    class Input(TestData):
        fields = [
                Field('rmdir', StrVal(),
                      hint="""the directory name"""),
        ]

    def verify(self):
        # Display the header.
        self.out_header()
        # Remove the directory with all its content (DANGEROUS!).
        if os.path.exists(self.input.rmdir):
            shutil.rmtree(self.input.rmdir)


class TestState(object):
    """
    Keeps the mutable state of the testing process.

    `app`
        The current HTSQL application.

    `forks`
        A mapping from command-line parameters to :class:`Fork`
        instances; contains long-running applications.

    `with_all_suites`
        Indicates that the current suite or one of its ancestors
        was explicitly selected by the user.

    `passed`
        The current number of passed tests.

    `failed`
        The current number of failed tests.

    `updated`
        The current number of updated tests.

    `is_exiting`
        Indicates whether the user asked to halt the testing.
    """

    def __init__(self, app=None, forks=None, with_all_suites=False,
                 passed=0, failed=0, updated=0, is_exiting=False):
        self.app = app
        self.forks = forks or {}
        self.with_all_suites = with_all_suites
        self.passed = passed
        self.failed = failed
        self.updated = updated
        self.is_exiting = is_exiting

    def push(self, other):
        """
        Push the state data to a derived state.

        `other` (:class:`TestState`)
            A derived state, the state created by a suite for
            the suite test cases.
        """
        other.app = self.app
        other.forks = self.forks
        other.with_all_suites = self.with_all_suites
        other.passed = self.passed
        other.failed = self.failed
        other.updated = self.updated
        other.is_exiting = self.is_exiting

    def pull(self, other):
        """
        Pull the state from a derived state.

        Note that only statistical information is pulled from
        the derived state.

        `other` (:class:`TestState`)
            A derived state, the state created by a suite for
            the suite test cases.
        """
        self.passed = other.passed
        self.failed = other.failed
        self.updated = other.updated
        self.is_exiting = other.is_exiting


# The base classes for the YAML loaders and dumpers.  When available,
# use the fast, LibYAML-based variants, if not, use the slow pure-Python
# versions.
BaseYAMLLoader = yaml.SafeLoader
if hasattr(yaml, 'CSafeLoader'):
    BaseYAMLLoader = yaml.CSafeLoader
BaseYAMLDumper = yaml.SafeDumper
if hasattr(yaml, 'CSafeDumper'):
    BaseYAMLDumper = yaml.CSafeDumper


class RegressYAMLLoader(BaseYAMLLoader):
    """
    Loads test data from a YAML file.

    `routine` (:class:`RegressRoutine`)
        The testing engine.

    `with_input` (Boolean)
        Indicates that the YAML file contains input records.

    `with_output` (Boolean)
        Indicates that the YAML file contains output records.

    `stream` (a file or a file-like object)
        The YAML stream.
    """

    # A pattern to match `!environ` nodes.
    environ_pattern = r"""^ \$ \{ [a-zA-Z_][0-9a-zA-Z_.-]* \} $"""
    environ_regexp = re.compile(environ_pattern, re.X)

    def __init__(self, routine, with_input, with_output, stream):
        super(RegressYAMLLoader, self).__init__(stream)
        self.routine = routine
        # The list of permitted record classes.
        self.records = []
        # A mapping of record_class -> case_class.
        self.case_by_record = {}
        # A mapping of record_class -> the set of all attributes.
        self.all_keys_by_record = {}
        # A mapping of record_class -> the set of mandatory attributes.
        self.mandatory_keys_by_record = {}
        # Generate a list of permitted record classes.
        self.init_records(with_input, with_output)

    def init_records(self, with_input, with_output):
        # Gather the record classes from the available test cases.
        for case_class in self.routine.cases:
            if with_input and case_class.Input is not None:
                self.records.append(case_class.Input)
                self.case_by_record[case_class.Input] = case_class
            if with_output and case_class.Output is not None:
                self.records.append(case_class.Output)
                self.case_by_record[case_class.Output] = case_class

        # For each record class, prepare the set of all attributes and
        # the set of mandatory attributes.
        for record_class in self.records:
            all_keys = set()
            for field in record_class.fields:
                if field.is_any:
                    all_keys = None
                    break
                all_keys.add(field.attribute.replace('_', '-'))
            self.all_keys_by_record[record_class] = all_keys
            mandatory_keys = set()
            for field in record_class.fields:
                if field.is_any or not field.is_mandatory:
                    continue
                mandatory_keys.add(field.attribute.replace('_', '-'))
            if not mandatory_keys:
                mandatory_keys = None
            self.mandatory_keys_by_record[record_class] = mandatory_keys

    def load(self):
        """
        Loads test data from the YAML stream.
        """
        # That ensures the stream contains one document, parses it and
        # returns the corresponding object.
        return self.get_single_data()

    def construct_document(self, node):
        # We override this to ensure that any produced document is
        # a test record of expected type.
        data = super(RegressYAMLLoader, self).construct_document(node)
        if type(data) not in self.records:
            raise yaml.constructor.ConstructorError(None, None,
                    "unexpected document type",
                    node.start_mark)
        return data

    def construct_yaml_str(self, node):
        # Always convert a `!!str` scalar node to a byte string.
        # By default, PyYAML converts an `!!str`` node containing non-ASCII
        # characters to a Unicode string.
        value = self.construct_scalar(node)
        value = value.encode('utf-8')
        return value

    def construct_yaml_map(self, node):
        # Detect if a node represent test data and convert it to a test record.

        # We assume that the node represents a test record if it contains
        # all mandatory keys of the record class.  Otherwise, we assume it
        # is a regular dictionary.
        #
        # It would be much better to perform this detection on the tag
        # resolution phase.  However this phase does not give us access
        # to the mapping keys, so we have no choice but do it during the
        # construction phase.

        # Check if we got a mapping node.
        if not isinstance(node, yaml.MappingNode):
            raise yaml.constructor.ConstructorError(None, None,
                    "expected a mapping node, but found %s" % node.id,
                    node.start_mark)

        # Objects corresponding to the key nodes.
        keys = []
        # Objects corresponding to the value nodes.
        values = []
        # The mapping of key object -> value object.
        value_by_key = {}
        # The mapping of key object -> the mark of the key node.
        key_mark_by_key = {}
        # The mapping of key object -> the mark of the value node.
        value_mark_by_key = {}

        # Convert the key and the value nodes.
        for key_node, value_node in node.value:
            key = self.construct_object(key_node, deep=True)
            try:
                hash(key)
            except TypeError, exc:
                raise yaml.constructor.ConstructorError(
                        "while constructing a mapping",
                        node.start_mark,
                        "found unacceptable key (%s)" % exc,
                        key_node.start_mark)
            keys.append(key)
            value = self.construct_object(value_node, deep=True)
            values.append(value)
            value_by_key[key] = value
            key_mark_by_key[key] = key_node.start_mark
            value_mark_by_key[key] = value_node.start_mark

        # Find a record class such that the node contains all
        # the mandatory record fields.
        detected_record_class = None
        key_set = set(keys)
        for record_class in self.records:
            mandatory_keys = self.mandatory_keys_by_record[record_class]
            if mandatory_keys is None:
                continue
            if key_set.issuperset(mandatory_keys):
                detected_record_class = record_class
                break

        # If we can't find a suitable record class, it must be a regular
        # dictionary.
        if detected_record_class is None:
            return dict(zip(keys, values))

        # Check that the node does not contain any keys other than
        # the record fields.
        all_keys = self.all_keys_by_record[detected_record_class]
        if all_keys is not None:
            for key in keys:
                if key not in all_keys:
                    raise yaml.constructor.ConstructorError(None, None,
                            "unexpected key %r; expected one of %s"
                            % (key, ", ".join(sorted(all_keys))),
                            key_mark_by_key[key])

        # Generate the record attributes: validate and normalize
        # the field values.
        attributes = {}
        for field in detected_record_class.fields:
            if field.is_any:
                continue
            key = field.attribute.replace('_', '-')
            if key in value_by_key:
                value = value_by_key[key]
                try:
                    value = field.validator(value)
                except ValueError, exc:
                    raise yaml.constructor.ConstructorError(None, None,
                            "invalid field %r (%s)" % (key, exc),
                            value_mark_by_key[key])
            else:
                value = field.default
            attributes[field.attribute] = value

        # Record where the node was found.
        location = "\"%s\", line %s" \
                   % (node.start_mark.name, node.start_mark.line+1)

        # Instantiate and return the test record.
        case_class = self.case_by_record[detected_record_class]
        try:
            record = detected_record_class(self.routine, case_class,
                                           attributes, location)
        except ValueError, exc:
            raise yaml.constructor.ConstructorError(None, None,
                    "invalid test data (%s)" % exc,
                    node.start_mark)
        return record

    def construct_environ(self, node):
        # Replace an `!environ` scalar with the value of the corresponding
        # environment variable.

        # Get the scalar value and check that it has the form ${...}.
        value = self.construct_scalar(node)
        value = value.encode('utf-8')
        if not (value.startswith('${') and value.endswith('}')):
            raise yaml.constructor.ConstructorError(None, None,
                    "invalid environment variable", node.start_mark)

        # The name of the environment variable.
        name = value[2:-1]

        # Return the value of the environment variables.  Blank values
        # are returned as `None`.
        value = os.environ.get(name, '')
        if not value:
            return None
        return value


# Register custom constructors for `!!str``, `!!map`` and ``!environ``.
RegressYAMLLoader.add_constructor(
        u'tag:yaml.org,2002:str',
        RegressYAMLLoader.construct_yaml_str)
RegressYAMLLoader.add_constructor(
        u'tag:yaml.org,2002:map',
        RegressYAMLLoader.construct_yaml_map)
RegressYAMLLoader.add_constructor(
        u'!environ',
        RegressYAMLLoader.construct_environ)


# Register a resolver for ``!environ``.
RegressYAMLLoader.add_implicit_resolver(
        u'!environ', RegressYAMLLoader.environ_regexp, [u'$'])


class RegressYAMLDumper(BaseYAMLDumper):
    """
    Dumps test data to a YAML file.

    `routine` (:class:`RegressRoutine`)
        The testing engine.

    `with_input` (Boolean)
        Indicates that the YAML file will contain input records.

    `with_output` (Boolean)
        Indicates that the YAML file will contain output records.

    `stream` (a file or a file-like object)
        The stream where the YAML document is written.
    """

    def __init__(self, routine, with_input, with_output, stream, **keywords):
        # FIXME: we don't really need extra `with_*` parameters, this
        # constructor is always called with with_input=False, with_output=True.
        super(RegressYAMLDumper, self).__init__(stream, **keywords)
        self.routine = routine
        # The set of permitted record classes.
        self.records = set()
        # Gather the permitted record classes.
        self.init_records(with_input, with_output)
        # Check if the PyYAML version is suitable for dumping.
        self.check_version()

    def init_records(self, with_input, with_output):
        # Gather permitted record classes.
        for case_class in self.routine.cases:
            if with_input and case_class.Input is not None:
                self.records.add(case_class.Input)
            if with_output and case_class.Output is not None:
                self.records.add(case_class.Output)

    def check_version(self):
        # We require PyYAML >= 3.07 built with LibYAML >= 0.1.2 to dump
        # YAML data.  Other versions may produce slightly different output.
        # Since the YAML files may be kept in a VCS repository, we don't
        # want minor formatting changes generate unnecessarily large diffs.
        try:
            pyyaml_version = yaml.__version__
        except AttributeError:
            pyyaml_version = '3.05'
        try:
            import _yaml
            libyaml_version = _yaml.get_version_string()
        except ImportError:
            libyaml_version = None
        if pyyaml_version < '3.07':
            raise ScriptError("PyYAML >= 3.07 is required"
                              " to dump test output")
        if libyaml_version is None:
            raise ScriptError("PyYAML built with LibYAML bindings"
                              " is required to dump test output")
        if libyaml_version < '0.1.2':
            raise ScriptError("LibYAML >= 0.1.2 is required"
                              " to dump test output")

    def dump(self, data):
        """
        Dumps the data to the YAML stream.
        """
        self.open()
        self.represent(data)
        self.close()

    def represent_str(self, data):
        # Serialize a string.  We override the default string serializer
        # to use the literal block style for multi-line strings.
        tag = None
        style = None
        if data.endswith('\n'):
            style = '|'
        try:
            data = data.decode('utf-8')
            tag = u'tag:yaml.org,2002:str'
        except UnicodeDecodeError:
            data = data.encode('base64')
            tag = u'tag:yaml.org,2002:binary'
            style = '|'
        return self.represent_scalar(tag, data, style=style)

    def represent_record(self, data):
        # Complain when given a record of unexpected type.
        if type(data) not in self.records:
            return super(RegressYAMLDumper, self).represent_undefined(data)
        # Extract the fields skipping those with the default value.
        mapping = []
        for field in data.fields:
            if field.is_any:
                continue
            name = field.attribute.replace('_', '-')
            value = getattr(data, field.attribute)
            if value == field.default:
                continue
            mapping.append((name, value))
        # Generate a mapping node.
        return self.represent_mapping(u'tag:yaml.org,2002:map', mapping,
                                      flow_style=False)


# Register custom representers for `str` and `TestData`.
RegressYAMLDumper.add_representer(
        str, RegressYAMLDumper.represent_str)
RegressYAMLDumper.add_multi_representer(
        TestData, RegressYAMLDumper.represent_record)


class RegressRoutine(Routine):
    """
    Implements the `regress` routine.
    """

    name = 'regress'
    aliases = ['test']
    arguments = [
            Argument('suites', SeqVal(WordVal()), None, is_list=True),
    ]
    options = [
            InputOption,
            TrainOption,
            PurgeOption,
            ForceOption,
            QuietOption,
    ]
    hint = """run regression tests"""
    help = """
    This routine runs a series of test cases.
    
    A test case takes input data and produces output data.  The test
    succeeds if it runs without errors and its output data coincides with
    the expected output.

    Input and output test data are stored in the YAML format.  Run
    '%(executable)s help regress <case>' to get the description of the
    format for a specific test type.

    Test cases are organized into suites.  A test suite is a special type of
    a test case that contains other test cases.

    By default, the routine executes all tests in the given YAML file.  To
    run only specific test suites, list their identifiers in the command
    line.

    Unless option `--force` is used, the testing process will halt on the
    first test failure.

    The routine reads the input data from the standard input stream.  Use
    option `--input FILE` to read the input data from a file instead.

    The routine supports training mode, in which it allows you to add
    expected output for new tests and updated expected output for existing
    tests.  Use option `--train` to run the routine in the training mode.

    When a test case is removed, the routine does not remove obsolete
    expected output records automatically.  Use option `--purge` to remove
    stale output records.

    By default, the routine prints the header of every executed tests.  Use
    option `--quiet` to print only errors and final statistics.
    """

    # This text is written to YAML files generated by the routine.
    output_help = """
    #
    # This file contains expected test output data for regression tests.
    # It was generated automatically by the `regress` routine.
    #
    """

    # List of supported types of test cases.
    cases = [
            SkipTestCase,
            AppTestCase,
            IncludeTestCase,
            SuiteTestCase,
            QueryTestCase,
            CtlTestCase,
            StartCtlTestCase,
            EndCtlTestCase,
            PythonCodeTestCase,
            SQLTestCase,
            SQLIncludeTestCase,
            WriteToFileTestCase,
            ReadFromFileTestCase,
            RemoveFilesTestCase,
            MakeDirTestCase,
            RemoveDirTestCase,
    ]

    # Represents the mutable state of the testing process.
    state_class = TestState

    @classmethod
    def get_help(cls, **substitutes):
        """
        Returns a long description of the routine.
        """
        # Produce routine description of the form:
        # {help}
        # 
        # Test cases: (run ... for more help)
        #   {case.name} : {case.hint}
        #   ...
        lines = []
        help = super(RegressRoutine, cls).get_help(**substitutes)
        if help is not None:
            lines.append(help)
        if cls.cases:
            if lines:
                lines.append("")
            lines.append("Test cases:"
                         " (run '%(executable)s help regress <case>'"
                         " for more help)" % substitutes)
            for case_class in cls.cases:
                case_name = case_class.name
                case_hint = case_class.get_hint()
                if case_hint is not None:
                    lines.append("  %-24s : %s" % (case_name, case_hint))
                else:
                    lines.append("  %s" % case_name)
        return "\n".join(lines)

    @classmethod
    def get_feature(cls, name):
        """
        Finds the test case by name.
        """
        for case_class in cls.cases:
            if case_class.name == name:
                return case_class
        raise ScriptError("unknown test case %r" % name)

    def run(self):
        # Get the test input data.
        input = self.load_input(self.input)
        # Initialize the testing state.
        state = self.state_class()
        # Create a test case.
        case = input.case_class(self, state, input, None)

        # Check if all test suites specified by the user exist.
        if self.suites:
            available_suites = case.get_suites()
            for suite in self.suites:
                if suite not in available_suites:
                    raise ScriptError("unknown suite %r" % suite)

        # Start the testing in the selected mode.
        if self.train:
            case.train()
        else:
            case.verify()

        # Display the statistics.
        self.ctl.out("="*72)
        if state.passed:
            self.ctl.out("TESTS PASSED: %s" % state.passed)
        if state.failed:
            self.ctl.out("TESTS FAILED: %s" % state.failed)
        if state.updated:
            self.ctl.out("TESTS UPDATED: %s" % state.updated)
        self.ctl.out()

        # Produce a fatal error if at least one test failed.
        if state.failed:
            if state.failed == 1:
                message = "a test failed"
            else:
                message = "%s tests failed" % state.failed
            raise ScriptError(message)

    def load_input(self, path):
        # Load test input data from a file.  If `path` is `None`,
        # load from the standard input.
        assert isinstance(path, maybe(str))
        if path is not None:
            stream = open(path, 'rb')
        else:
            stream = self.ctl.stdin
        loader = RegressYAMLLoader(self, True, False, stream)
        try:
            input = loader.load()
        except yaml.YAMLError, exc:
            raise ScriptError("failed to load test input data: %s" % exc)
        return input

    def load_output(self, path):
        # Load test output data from a file.
        assert isinstance(path, str)
        stream = open(path, 'rb')
        loader = RegressYAMLLoader(self, False, True, stream)
        try:
            input = loader.load()
        except yaml.YAMLError, exc:
            raise ScriptError("failed to load test output data: %s" % exc)
        return input

    def save_output(self, path, output):
        # Serialize and write test output data to a file.
        assert isinstance(path, str)
        assert isinstance(output, TestData)
        stream = open(path, 'wb')
        if self.output_help is not None:
            self.ctl.out(trim_doc(self.output_help), file=stream)
            self.ctl.out(file=stream)
        dumper = RegressYAMLDumper(self, False, True, stream)
        try:
            dumper.dump(output)
        except yaml.YAMLError, exc:
            raise ScriptError("failed to write test output data: %s" % exc)