/usr/share/perl5/Regexp/Debugger.pm is in libregexp-debugger-perl 0.001020-1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 | package Regexp::Debugger;
use warnings;
use strict;
eval "use feature 'evalbytes'"; # Experimental fix for Perl 5.16
our $VERSION = '0.001020';
# Handle Perl 5.18's new-found caution...
no if $] >= 5.018, warnings => "experimental::smartmatch";
# Give an accurate warning if used with an antique Perl...
BEGIN {
if ($] < 5.010001) {
die sprintf "Regexp::Debugger requires Perl v5.10.1 or later (at %s line %s)\n",
(caller 2)[1..2];
}
}
use 5.010001; # ...activate all the tasty 5.10 goodies
use List::Util qw< min max first sum >;
# Track configurable options lexically...
my @config;
# Track debugging history in various formats...
my %history_of;
# Persistent information within debugger...
my $prev_regex_pos; # ...track where we were previously in the regex
my $start_str_pos; # ...track where we started matching in the string
my $prev_str_pos; # ...track where we were previously in the string
my $prev_match_was_null; # ...under /g was previous match a null match?
my %capture; # ...track capture groups within regex
my $pre_is_pending; # ...did we try something last event?
my $interaction_quit; # ...did we get a quit request?
my $interaction_mode; # ...step-by-step, jump to match, or continue?
my $display_mode; # ...how is the match being visualized at present?
# Colours for heatmaps...
my @DEF_HEAT_COLOUR = (
'white on_black', # 0-20 percentile
'cyan on_blue', # 20-40 percentile
'blue on_cyan', # 40-60 percentile
'red on_yellow', # 60-80 percentile
'yellow on_red', # 80-100 percentile
);
# Colours for detailed regex descriptions...
my %DESCRIPTION_COLOUR = (
desc_sep_col => 'blue on_black underline',
desc_regex_col => 'white on_black',
desc_text_col => 'cyan on_black',
);
# Colour for error messages...
my $ERR_COL = 'red';
# Default config which any explicit config modifies...
my @SHOW_WS_OPTIONS = qw< compact visible original >;
my %DEFAULT_CONFIG = (
# How debugging info is displayed initially...
display_mode => 'visual',
# Colour scheme for debugging info...
info_col => ' white on_black',
try_col => 'bold magenta on_black',
match_col => ' bold cyan on_black',
fail_col => ' yellow on_red',
ws_col => ' bold blue underline',
# Colour scheme for regex descriptions...
%DESCRIPTION_COLOUR,
# Where debugging info is written to (undef --> STDOUT)...
save_to_fh => undef,
# How whitespace is managed...
show_ws => $SHOW_WS_OPTIONS[0],
);
# The current config...
my $lexical_config = \%DEFAULT_CONFIG;
# Simulate print() and say() on appropriate filehandle...
sub _print {
if (!$lexical_config->{save_to_fh}) {
no warnings 'utf8';
print @_;
}
}
sub _say {
if (!$lexical_config->{save_to_fh}) {
no warnings 'utf8';
say @_;
}
}
# How should matches be indicated???
my $MATCH_DRAG = ' ';
# Will heatmaps be visible???
my $heatmaps_invisible;
# Indent unit for hierarchical display...
my $INDENT = q{ };
# Simulate Term::ANSIColor badly (if necessary)...
CHECK {
my $can_color
= ( $^O ne 'MSWin32' or eval { require Win32::Console::ANSI } )
&& eval { require Term::ANSIColor };
if ( !$can_color ) {
*Term::ANSIColor::colored = sub { return shift };
$MATCH_DRAG = '_';
$heatmaps_invisible = 1;
}
}
# Load the module...
sub import {
use Carp;
# Don't need the module name...
shift;
# Export re 'eval' semantics...
$^H |= 0x00200000;
# Unpack the arguments...
if (@_ % 2) {
croak 'Odd number of configuration args after "use Regexp::Debugger"';
}
my %arg = @_;
# Creat a new lexically scoped config and remember its index...
push @config, { %DEFAULT_CONFIG };
$^H{'Regexp::Debugger::lexical_scope'} = $#config;
_load_config(\%arg);
# Signal lexical scoping (active, unless something was exported)...
$^H{'Regexp::Debugger::active'} = 1;
# Process any regexes in module's active lexical scope...
use overload;
overload::constant(
qr => sub {
my ($raw, $cooked, $type) = @_;
my $lexical_scope = (caller 1)[10]{'Regexp::Debugger::lexical_scope'};
# In active scope and really a regex and interactivity possible...
my $is_interactive = defined $arg{save_to} || -t *STDIN && -t *STDOUT;
if (_module_is_active() && $type =~ /qq?/ && $is_interactive) {
return bless {cooked=>$cooked, lexical_scope=>$lexical_scope}, 'Regexp::Debugger::Precursor';
}
# Ignore everything else...
else {
return $cooked;
}
}
);
}
# Deactivate module's regex effect when it is "anti-imported" with 'no'...
sub unimport {
# Signal lexical (non-)scoping...
$^H{'Regexp::Debugger::active'} = 0;
}
# Encapsulate the hoopy user-defined pragma interface...
sub _module_is_active {
return (caller 1)[10]->{'Regexp::Debugger::active'};
}
# Load ~/.rxrx config...
sub _load_config {
my $explicit_config_ref = shift();
my %config;
# Work out where to look...
my $home_dir = $ENV{HOME};
if (!$home_dir && eval { require File::HomeDir } ) {
$home_dir = File::HomeDir->my_home;
}
# Find config file...
CONFIG_FILE:
for my $config_file ( '.rxrx', ( $home_dir ? "$home_dir/.rxrx" : () ) ) {
# Is this a readable config file???
open my $fh, '<', $config_file
or next CONFIG_FILE;
# Read and parse config file...
CONFIG_LINE:
for my $config_line (<$fh>) {
if ($config_line =~ /^\s*(.*?)\s*[:=]\s*(.*?)\s*$/) {
$config{$1} = $2;
}
}
last CONFIG_FILE;
}
# Make any explicit args override .rxrxrc config...
%config = (display => 'visual', %config, %{$explicit_config_ref});
# Configure colour scheme for displays...
for my $colour (grep /_col$/, keys %DEFAULT_CONFIG) {
if (exists $config{$colour}) {
$config[-1]{$colour} = $config{$colour}
}
}
# Configure how whitespace is displayed...
my $show_ws = $config{show_ws};
if (defined $show_ws) {
if ($show_ws ~~ @SHOW_WS_OPTIONS) {
$config[-1]{show_ws} = $show_ws;
}
else {
croak "Unknown 'show_ws' option: '$show_ws'";
}
}
# Configure heatmap colour scheme...
my @heatmap_cols =
map { $config{$_} }
sort {
# Sort numerically (if feasible), else alphabetically...
my $a_key = $a =~ /(\d+)/ ? $1 : undef;
my $b_key = $b =~ /(\d+)/ ? $1 : undef;
defined $a_key && defined $b_key
? $a_key <=> $b_key
: $a cmp $b;
}
grep { /^heatmap/ }
keys %config;
if (!@heatmap_cols) {
@heatmap_cols = @DEF_HEAT_COLOUR;
}
$config[-1]{heatmap_col} = \@heatmap_cols;
# Configure initial display mode...
my $display = $config{display};
if (defined $display) {
$config[-1]{display_mode}
= $display =~ m{^events}i ? 'events'
: $display =~ m{^heatmap}i ? 'heatmap'
: $display =~ m{^visual}i ? 'visual'
: $display =~ m{^JSON}i ? 'JSON'
: croak "Unknown 'display' option: '$display'";
}
# Configure destination of debugging info...
my $save_to = $config{save_to};
if (defined $save_to) {
use Scalar::Util qw< openhandle >;
if (openhandle($save_to)) {
$config[-1]{save_to_fh} = $save_to;
}
elsif (!ref $save_to) {
my ($mode, $filename) = $save_to =~ m{ (>{0,2}) (.*) }x;
open my $fh, $mode||'>', $filename
or croak "Invalid 'save_to' option: '$save_to'\n($!)";
$config[-1]{save_to_fh} = $fh;
}
else {
croak "Invalid 'save_to' option: ", ref($save_to);
}
}
}
# General memory for each state of each regex...
# (structure is: $state{REGEX_NUM}{STATE_NUMBER}{ATTRIBUTE})
my %state;
my $next_regex_ID = 0;
#=====[ COMPILE-TIME INTERIM REPRESENTATION OF REGEXES ]===================
{
package Regexp::Debugger::Precursor;
# Only translate precursors once...
state %regex_cache;
use overload (
# Concatenation/interpolation just concatenates to the precursor...
q{.} => sub {
my ($x, $y, $reversed) = @_;
# Where are we from???
my $lexical_scope = $x->{lexical_scope} // 0;
# Reorder if necessary...
if ($reversed) { ($y,$x) = ($x,$y); }
# Unpack if objects...
if (ref $x) { $x = eval{ $x->{cooked} } // $x }
if (ref $y) { $y = eval{ $y->{cooked} } // $y }
# Undo overeager \Q if necessary...
if ($x =~ m{^\\\(\\\?\\\#R_d\\:(\d+)\\\)}) { $x = '\\Q' . $state{$1}{raw_regex} . '\\E' }
elsif ($x =~ m{^\(\?\#R_D:(\d+)\)}) { $x = '\\U' . uc($state{$1}{raw_regex}) . '\\E' }
elsif ($x =~ m{^\(\?\#r_d:(\d+)\)}) { $x = '\\L' . lc($state{$1}{raw_regex}) . '\\E' }
if ($y =~ m{^\\\(\\\?\\\#R_d\\:(\d+)\\\)}) { $y = '\\Q' . $state{$1}{raw_regex} . '\\E' }
elsif ($y =~ m{^\(\?\#R_D:(\d+)\)}) { $y = '\\U' . uc($state{$1}{raw_regex}) . '\\E' }
elsif ($y =~ m{^\(\?\#r_d:(\d+)\)}) { $y = '\\L' . lc($state{$1}{raw_regex}) . '\\E' }
# Do the concatenation...
$x .= $y//q{};
# Rebless as a precursor object...
return bless {cooked=>$x, lexical_scope=>$lexical_scope}, 'Regexp::Debugger::Precursor';
},
# Using as a string (i.e. matching) preprocesses the precursor...
q{""} => sub {
my ($obj) = @_;
use Scalar::Util qw< refaddr >;
return $regex_cache{ refaddr($obj) }
//= Regexp::Debugger::_build_debugging_regex( @{$obj}{'cooked', 'lexical_scope'} );
},
# Everything else, as usual...
fallback => 1,
);
}
#=====[ Augment a regex with debugging statements ]================
# Build code insertions for before and after elements in a regex...
# (the final $^R ensure these extra code blocks are "transparent")
sub _build_event {
my ($regex_ID, $event_ID, $event_desc_ref) = @_;
$event_desc_ref->{quantifier} //= q{};
$state{$regex_ID}{$event_ID} = $event_desc_ref;
return qq{(?{Regexp::Debugger::_report_event($regex_ID, $event_ID, pos()); \$^R })(?=)};
}
sub _build_whitespace_event {
my ($construct,$regex_ID, $event_ID, $event_desc_ref) = @_;
$event_desc_ref->{quantifier} //= q{};
my %event_desc_copy = %{$event_desc_ref};
$state{$regex_ID}{$event_ID} = { %event_desc_copy, event_type => 'pre' };
$state{$regex_ID}{$event_ID+1} = { %event_desc_copy, event_type => 'post', msg => 'Matched' };
return qq{(?>(?{local \$Regexp::Debugger::prevpos=pos})$construct(?{
if (defined \$Regexp::Debugger::prevpos && \$Regexp::Debugger::prevpos < pos){
Regexp::Debugger::_report_event($regex_ID, $event_ID, \$Regexp::Debugger::prevpos);
Regexp::Debugger::_report_event($regex_ID, $event_ID+1, pos());
}\$^R })|(?{
Regexp::Debugger::_report_event($regex_ID, $event_ID, pos());
\$^R
})(?!))};
}
# Translate lookaround markers...
my %LOOKTYPE = (
'(?=' => 'positive lookahead',
'(?!' => 'negative lookahead',
'(?<=' => 'positive lookbehind',
'(?<!' => 'negative lookbehind',
);
sub _build_debugging_regex {
my ( $raw_regex, $lexical_scope ) = @_;
$lexical_scope //= 0;
# How does this regexp show whitespace???
our $show_ws = $config[$lexical_scope]{show_ws};
# Build a clean, compacted version of the regex in this var...
my $clean_regex = q{};
# Track nested parentheticals so we can correctly mark each ')'...
my @paren_stack = ( {} );
# Give this regex a unique ID...
my $regex_ID = $next_regex_ID++;
# Remember raw data in case of over-eager quotemeta'ing...
$state{$regex_ID}{raw_regex} = $raw_regex;
# Remember location of regex...
my ($filename, $end_line) = (caller 1)[1,2];
my $regex_lines = $raw_regex =~ tr/\n//;
my $start_line = $end_line - $regex_lines;
$state{$regex_ID}{location}
= $start_line == $end_line ? qq{'$filename' line $start_line}
: qq{'$filename' lines $start_line-$end_line};
# Track each inserted debugging statement...
my $next_event_ID = 0;
# Track capture groups...
my $next_capture_group = 0;
my $max_capture_group = 0;
# Track named capture aliases...
my @capture_names_for;
# Track \Q...\E
my $in_quote = 0;
my $shared_quote_pos;
# Describe construct...
our $construct_desc;
our $quantifier_desc;
# Translate each component...
use re 'eval';
$raw_regex =~ s{
# Set-up...
(?{ $quantifier_desc = q{}; $construct_desc = q{}; })
# Match the next construct...
(?<construct>
(?<start> \A )
|
(?<end> \z )
|
(?<quote_start>
(??{!$Regexp::Debugger::in_quote ? q{} : q{(?!)} })
\\Q
(?{$Regexp::Debugger::in_quote = 1})
)
| (??{$Regexp::Debugger::in_quote ? q{} : q{(?!)} })
(
(?<quote_space> \s++ )
|
(?<quote_end> \\E )
(?{$Regexp::Debugger::in_quote = 0})
|
(?<quote_nonspace> \S )
)
|
(?<case_start>
\\U (?{$construct_desc = 'an auto-uppercased sequence'})
| \\L (?{$construct_desc = 'an auto-lowercased sequence'})
)
|
(?<case_end>
\\E
)
|
(?{$quantifier_desc = '';})
(?<closing_paren> [)] ) (?<quantifier> (?&QUANTIFIER) )?
|
(?<whitespace>
(?(?{ $show_ws eq 'compact' })
(?<whitespace_chars>
( (?: \s | (?&COMMENT) )+ )
(?! (?&UNSPACED_QUANTIFIER) ) (?{ $quantifier_desc = q{} })
(?{$construct_desc = $^N})
|
( \s )
(?{$construct_desc = $^N})
(?<quantifier> (?&UNSPACED_QUANTIFIER) )
)
|
(?!)
)
|
(?(?{ $show_ws eq 'visible' })
(?<whitespace_chars>
( [^\S\n\t]+ )
(?! (?&UNSPACED_QUANTIFIER) ) (?{ $quantifier_desc = q{} })
(?{$construct_desc = $^N})
|
( [^\S\n\t] )
(?{$construct_desc = $^N})
(?<quantifier> (?&UNSPACED_QUANTIFIER) )
)
|
(?!)
)
|
(?(?{ $show_ws eq 'original'})
(?<whitespace_chars>
( [^\S\n\t] )
(?{$construct_desc = $^N})
(?<quantifier> (?&UNSPACED_QUANTIFIER) )?
)
|
(?!)
)
|
(?<newline_char> \n )
(?<quantifier> (?&UNSPACED_QUANTIFIER) )?
(?{$construct_desc = 'a literal newline character'})
|
(?<tab_char> \t )
(?<quantifier> (?&UNSPACED_QUANTIFIER) )?
(?{$construct_desc = 'a literal tab character'})
)
|
(?<break_comment>
[(][?][#] \s* (?i: BREAK ) \s* [)]
)
|
(?<comment>
(?&COMMENT)
)
|
(?<modifier_set>
[(] [?] (?&MODIFIERS) [)]
)
|
(?<zero_width>
(?<_anchor>
\^
(?{$construct_desc = 'at start of string (or line)'})
|
\$
(?{$construct_desc = 'at end of string (or final newline)'})
|
\\ (?:
A (?{$construct_desc = 'at start of string'})
|
B (?{$construct_desc = 'not at an identifier boundary'})
|
b (?{$construct_desc = 'at an identifier boundary'})
|
G (?{$construct_desc = 'at previous match position'})
|
Z (?{$construct_desc = 'at end of string (or final newline)'})
|
z (?{$construct_desc = 'at end of string'})
)
)
)
|
(?<matchable_code_block>
[(] [?][?] (?&CODEBLOCK) [)]
)
|
(?<code_block>
[(] [?] (?&CODEBLOCK) [)]
)
|
# Control verbs like (*PRUNE) and (*MARK:name)...
(?<control>
\(\* [[:upper:]]*+ (?: : [^)]++ )? \)
)
|
(?<noncapturing_paren> [(] [?] (?&MODIFIERS)? : )
|
(?<lookaround_paren> [(] [?] [<]?[=!] )
|
(?<non_backtracking_paren> [(] [?] [>] )
(?{$construct_desc = 'a non-backtracking group'})
|
(?<branch_reset_paren> [(] [?] [|] )
|
(?<capturing_paren> [(] (?! [?]) )
|
(?<define_block> [(] [?] [(] DEFINE [)] )
|
(?<conditional_paren>
[(] [?] [(]
(?<condition>
\d+
| R \d*
| R& (?&IDENTIFIER)
| < (?&IDENTIFIER) >
| ' (?&IDENTIFIER) '
| [?] (?&CODEBLOCK)
)
[)]
)
|
(?<conditional_paren> (?<pending_condition>
[(] [?] (?= [(] [?] <? [=!] )
))
|
(?<named_capturing_paren>
[(] [?] P? < (?<capture_name> (?&IDENTIFIER) ) >
| [(] [?] ' (?<capture_name> (?&IDENTIFIER) ) '
)
|
(?<_alternation> [|] )
|
(?<keep_marker> \\K )
|
(?<atom>
(?<_self_matching> \w{2,} ) (?! (?&QUANTIFIER) )
(?{$quantifier_desc = ''; $construct_desc = qq{a literal sequence ("$+{_self_matching}")}})
|
(?{$quantifier_desc = '';})
(?<_self_matching> (?&NONMETA) ) (?<quantifier> (?&QUANTIFIER) )?
(?{$construct_desc = qq{a literal '$+{_self_matching}' character}})
|
(?{$quantifier_desc = '';})
(?<_metacharacter>
[.] (?{$construct_desc = 'any character (except newline)'})
|
\\
(?: (0[0-7]++) (?{$construct_desc = "a literal '".chr(oct($^N))."' character"})
| (\d++) (?{$construct_desc = "what was captured in \$$^N"})
| a (?{$construct_desc = 'an alarm/bell character'})
| c ([A-Z]) (?{$construct_desc = "a CTRL-$^N character"})
| C (?{$construct_desc = 'a C-language octet'})
| d (?{$construct_desc = 'a digit'})
| D (?{$construct_desc = 'a non-digit'})
| e (?{$construct_desc = 'an escape character'})
| f (?{$construct_desc = 'a form-feed character'})
| g (\d+) (?{$construct_desc = "what was captured in \$$^N"})
| g - (\d+) (?{$construct_desc = $^N == 1 ? "what was captured by the nearest preceding capture group"
: "what was captured $^N capture groups back" })
| g \{ (\d+) \} (?{$construct_desc = "what was captured in \$$^N"})
| g \{ - (\d+) \} (?{$construct_desc = $^N == 1 ? "what was captured by the nearest preceding capture group"
: "what was captured $^N capture groups back" })
| g \{ (\w++) \} (?{$construct_desc = "what the named capture <$^N> matched"})
| h (?{$construct_desc = 'a horizontal whitespace character'})
| H (?{$construct_desc = 'a non-horizontal-whitespace character'})
| k \< (\w++) \> (?{$construct_desc = "what the named capture <$^N> matched"})
| n (?{$construct_desc = 'a newline character'})
| N \{ ([^\}]++) \} (?{$construct_desc = "a single \L$^N\E character"})
| N (?{$construct_desc = 'a non-newline character'})
| p (\w++) (?{$construct_desc = "a character matching the Unicode property: $^N"})
| P (\w++) (?{$construct_desc = "a character not matching the Unicode property: $^N"})
| P \{ ([^\}]++) \} (?{$construct_desc = "a character not matching the Unicode property: $^N"})
| p \{ ([^\}]++) \} (?{$construct_desc = "a character matching the Unicode property: $^N"})
| r (?{$construct_desc = 'a return character'})
| R (?{$construct_desc = 'an end-of-line sequence'})
| S (?{$construct_desc = 'a non-whitespace character'})
| s (?{$construct_desc = 'a whitespace character'})
| t (?{$construct_desc = 'a tab character'})
| V (?{$construct_desc = 'a non-vertical-whitespace character'})
| v (?{$construct_desc = 'a vertical whitespace character'})
| w (?{$construct_desc = 'an identifier character'})
| W (?{$construct_desc = 'an non-identifier character'})
| x ([0-9A-Za-z]++) (?{$construct_desc = "a literal '".chr(oct('0x'.$^N))."' character"})
| x \{ ([0-9A-Za-z ]++) \} (?{$construct_desc = "a literal '".chr(oct('0x'.$^N))."' character"})
| X (?{$construct_desc = 'a Unicode grapheme cluster'})
| (.) (?{$construct_desc = "a literal '$^N' character"})
)
|
[(][?] P = (\w++) [)] # PCRE version of \k<NAME>
(?{$construct_desc = "what the named capture <$^N> matched"})
) (?<quantifier> (?&QUANTIFIER) )?
|
(?{$quantifier_desc = '';})
(?<_charset> (?&CHARSET) ) (?<quantifier> (?&QUANTIFIER) )?
(?{$construct_desc = substr($+{_charset},0,2) eq '[^'
? 'any character not listed'
: 'any of the listed characters'
})
|
(?{$quantifier_desc = '';})
(?<_named_subpattern_call>
[(][?]
(?:
[&] ((?&IDENTIFIER)) (?{$construct_desc = "a call to the subpattern named <$^N>"})
| P> ((?&IDENTIFIER)) (?{$construct_desc = "a call to the subpattern named <$^N>"})
| [+]? (\d++) (?{$construct_desc = 'a call to subpattern number $^N'})
| [-] (\d++) (?{$construct_desc = $^N == 1 ? "a call to the nearest preceding subpattern"
: "a call to the subpattern $^N back" })
| R (?{$construct_desc = 'a recursive call to the current regex'})
)
[)]
)
(?<quantifier> (?&QUANTIFIER) )?
)
|
(?<misc> . )
)
(?(DEFINE)
# Miscellaneous useful pattern fragments...
(?<COMMENT> [(][?][#] (?! \s* BREAK \s* ) .*? [)] | \# [^\n]* (?= \n | \z ) )
(?<CHARSET> \[ \^?+ \\?+ \]?+ (?: \[:\w+:\] | \\[\[\]\\] | [^]] )*+ \] )
(?<IDENTIFIER> [^\W\d]\w* )
(?<CODEBLOCK> \{ (?: (?&CODEBLOCK) | . )*? \} )
(?<MODIFIERS> [adlupimsx]+ (?: - [imsx]+ )?
| - [imsx]+
| \^ [alupimsx]
)
(?<QUANTIFIER> \s* (?&UNSPACED_QUANTIFIER) )
(?<UNSPACED_QUANTIFIER>
[*][+] (?{ $quantifier_desc = 'zero-or-more times (without backtracking)' })
| [*][?] (?{ $quantifier_desc = 'zero-or-more times (as few as possible)' })
| [*] (?{ $quantifier_desc = 'zero-or-more times (as many as possible)' })
| [+][+] (?{ $quantifier_desc = 'one-or-more times (without backtracking)' })
| [+][?] (?{ $quantifier_desc = 'one-or-more times (as few as possible)' })
| [+] (?{ $quantifier_desc = 'one-or-more times (as many as possible)' })
| [?][+] (?{ $quantifier_desc = 'one-or-zero times (without backtracking)' })
| [?][?] (?{ $quantifier_desc = 'zero-or-one times (as few as possible)' })
| [?] (?{ $quantifier_desc = 'one-or-zero times (as many as possible)' })
| {\d+,?\d*}[+] (?{ $quantifier_desc = 'the specified number of times (without backtracking)' })
| {\d+,?\d*}[?] (?{ $quantifier_desc = 'the specified number of times (as few as possible)' })
| {\d+,?\d*} (?{ $quantifier_desc = 'the specified number of times (as many as possible)' })
)
(?<NONMETA> [\w~`!%&=:;"'<>,/-] )
)
}{
# Which event is this???
my $event_ID = $next_event_ID++;
# What are we debugging???
my $construct = $+{construct};
# How deep in parens???
my $depth = scalar(@paren_stack);
my $indent = $INDENT x $depth;
# All events get this standard information...
my %std_info = (
construct_type => (first { /^_/ } keys %+),
construct => $construct,
regex_pos => length($clean_regex),
quantifier => $+{quantifier} // q{},
depth => $depth,
indent => $indent,
);
# use Data::Dumper 'Dumper';
# warn Dumper { std_info => \%std_info, '%+' => \%+ };
# Record the construct for display...
$clean_regex .=
exists $+{newline_char} ? ($std_info{construct} = q{\n} . $std_info{quantifier})
: exists $+{tab_char} ? ($std_info{construct} = q{\t} . $std_info{quantifier})
: exists $+{whitespace_chars} ? ($std_info{construct} = q{ } . $std_info{quantifier})
: $construct
;
# Determine and remember the necessary translation...
my $translation = do {
# Beginning and end of regex...
if (exists $+{start}) {
# Prime paren-tracking stack...
push @paren_stack, {};
# Insert an event to report (re-)starting...
_build_event($regex_ID, $event_ID => {
%std_info,
construct_type => '_START',
event_type => 'pre',
depth => 1,
lexical_scope => $lexical_scope,
})
. '(?:';
}
# At end of regex (if we get here, we matched)...
elsif (exists $+{end}) {
# Insert a final event to report successful match...
')'
. _build_event($regex_ID, $event_ID => {
%std_info,
construct_type => '_END',
event_type => 'post',
depth => 1,
msg => sub { my $steps = @{$history_of{visual}};
$steps .= ' step' . ($steps != 1 ? 's' : '');
# Was this a second null match???
my $match_was_null = (pos == $start_str_pos);
if ($match_was_null && $prev_match_was_null) {
return "Regex matched in $steps but failed to advance within string";
}
else {
$prev_match_was_null = $match_was_null;
return "Regex matched in $steps";
}
},
})
. '|'
. _build_event($regex_ID, $event_ID+1 => {
%std_info,
construct_type => '_END',
event_type => 'post',
regex_failed => 1,
depth => 1,
msg => sub { my $steps = @{$history_of{visual}//[]};
"Regex failed to match"
. ($steps ? " after $steps step" . ($steps != 1 ? 's' : '')
: ' (unable to advance within string)');
},
})
. '(?!)';
}
# Alternatives marked by a |...
elsif (exists $+{_alternation}) {
# Reset capture numbers if in reset group...
if (my $reset = $paren_stack[-1]{is_branch_reset}) {
$next_capture_group = $reset-1;
}
# We need two events, so add an extra one...
$event_ID = $next_event_ID++;
# Insert events to indicate which side of the | we're trying now...
_build_event($regex_ID, $event_ID-1 => {
%std_info,
event_type => 'end',
msg => 'End of successful alternative',
desc => 'Or...',
indent => $INDENT x ($depth-1),
})
. $construct
. _build_event($regex_ID, $event_ID => {
%std_info,
event_type => 'start',
msg => 'Trying next alternative',
})
}
# Whitespace has to be treated specially (because it may or may not be significant...
elsif (exists $+{whitespace}) {
# The two events communicate privately via this variable...
my $shared_str_pos;
# Two events required, so add an extra ID...
$next_event_ID++;
$construct_desc = join q{}, map { $_ eq "\n" ? '\n'
: $_ eq "\t" ? '\t'
: $_ eq " " ? '\N{SPACE}'
: $_
} split '', $construct_desc;
# Insert the appropriate events...
_build_whitespace_event($construct, $regex_ID, $event_ID => {
%std_info,
matchable => 1,
msg => "Trying literal whitespace ('$construct_desc') $quantifier_desc",
shared_str_pos => \$shared_str_pos,
})
}
# \L and \U start case-shifted sequences...
elsif (exists $+{case_start}) {
_build_event($regex_ID, $event_ID => {
%std_info,
event_type => 'pre',
msg => "Starting $construct_desc",
desc => 'The start of ' . $construct_desc,
})
}
elsif (exists $+{case_end}) {
_build_event($regex_ID, $event_ID => {
%std_info,
event_type => 'pre',
msg => 'End of autocasing',
desc => 'The end of autocasing',
})
}
# \Q starts a quoted sequence...
elsif (exists $+{quote_start}) {
# Set up communication channel between \Q and \E...
my $shared_pos;
$shared_quote_pos = \$shared_pos;
_build_event($regex_ID, $event_ID => {
%std_info,
event_type => 'pre',
msg => 'Starting quoted sequence',
desc => 'The start of a quoted sequence',
shared_str_pos => $shared_quote_pos,
})
}
# \E ends a quoted sequence...
elsif (exists $+{quote_end}) {
# Retrieve communication channel between \Q and \E...
my $shared_pos = $shared_quote_pos;
$shared_quote_pos = undef;
_build_event($regex_ID, $event_ID => {
%std_info,
event_type => 'post',
msg => 'End of quoted sequence',
desc => 'The end of a quoted sequence',
shared_str_pos => $shared_pos,
})
}
# Quoted subsequences...
elsif (exists $+{quote_space}) {
# The two events communicate privately via this variable...
my $shared_str_pos;
# Two events, so add an extra ID...
$event_ID = $next_event_ID++;
_build_event($regex_ID, $event_ID-1 => {
%std_info,
matchable => 1,
event_type => 'pre',
msg => 'Trying autoquoted literal whitespace',
shared_str_pos => \$shared_str_pos,
})
. quotemeta($construct)
. _build_event($regex_ID, $event_ID => {
%std_info,
matchable => 1,
event_type => 'post',
msg => 'Matched autoquoted literal whitespace',
shared_str_pos => \$shared_str_pos,
})
}
elsif (exists $+{quote_nonspace}) {
# The two events communicate privately via this variable...
my $shared_str_pos;
# Two events, so add an extra ID...
$event_ID = $next_event_ID++;
_build_event($regex_ID, $event_ID-1 => {
%std_info,
matchable => 1,
event_type => 'pre',
msg => 'Trying an autoquoted literal character',
desc => 'Match an autoquoted literal character',
shared_str_pos => \$shared_str_pos,
})
. quotemeta($construct)
. _build_event($regex_ID, $event_ID => {
%std_info,
matchable => 1,
event_type => 'post',
msg => 'Matched a literal character',
shared_str_pos => \$shared_str_pos,
})
}
# Atoms are any elements that match and emit debugging info before and after matching...
elsif (exists $+{atom}) {
# The two events communicate privately via this variable...
my $shared_str_pos;
# Track depth of subpattern calls...
my $is_subpattern_call = exists $+{_named_subpattern_call};
my $subpattern_call_prefix
= $is_subpattern_call
? q{(?{local $Regexp::Debugger::subpattern_depth = $Regexp::Debugger::subpattern_depth + 1})}
: q{};
my $subpattern_call_suffix
= $is_subpattern_call
? q{(?{local $Regexp::Debugger::subpattern_depth = $Regexp::Debugger::subpattern_depth - 1})}
: q{};
# Two events, so add an extra ID...
$event_ID = $next_event_ID++;
_build_event($regex_ID, $event_ID-1 => {
%std_info,
matchable => 1,
event_type => 'pre',
msg => "Trying $construct_desc" . (length($quantifier_desc) ? ", $quantifier_desc" : q{}),
desc => "Match $construct_desc" . (length($quantifier_desc) ? ", $quantifier_desc" : q{}),
shared_str_pos => \$shared_str_pos,
})
. $subpattern_call_prefix
. $construct
. $subpattern_call_suffix
. _build_event($regex_ID, $event_ID => {
%std_info,
matchable => 1,
event_type => 'post',
msg => 'Matched'
. ($is_subpattern_call ? " (discarding subpattern's captures)": q{}),
shared_str_pos => \$shared_str_pos,
})
}
# Code blocks (?{...})...
elsif (exists $+{code_block}) {
# Add an event beforehand to indicate execution of the block...
_build_event($regex_ID, $event_ID => {
%std_info,
matchable => 0,
event_type => 'action',
msg => 'Executing code block',
desc => 'Execute a block of code',
})
. $construct
}
# Code blocks that generate dynamic patterns (??{...})...
elsif (exists $+{matchable_code_block}) {
# These events communicate privately via this variable...
my $shared_str_pos;
# Modify construct to generate but not match...
substr($construct, 1, 1) = q{};
# Inserting three events, so add an extra two IDs...
$event_ID = ($next_event_ID+=3);
# First event pair reports executing the block...
_build_event($regex_ID, $event_ID-4 => {
%std_info,
matchable => 0,
event_type => 'action',
msg => 'Executing code block of postponed subpattern',
desc => "Execute a code block, then match the block's final value",
})
. $construct
. _build_event($regex_ID, $event_ID-3 => {
%std_info,
matchable => 0,
event_type => 'action',
msg => sub { "Code block returned: '$^R'" },
})
# Second event pair reports match of subpattern the block returned...
. _build_event($regex_ID, $event_ID-2 => {
%std_info,
matchable => 1,
event_type => 'pre',
msg => sub{ "Trying: qr{$^R}" },
shared_str_pos => \$shared_str_pos,
})
. '(??{ $^R })'
. _build_event($regex_ID, $event_ID-1 => {
%std_info,
matchable => 1,
event_type => 'post',
msg => 'Matched',
shared_str_pos => \$shared_str_pos,
})
}
# Keep marker...
elsif (exists $+{keep_marker}) {
# Insert events reporting testing the assertion, and if the test succeeds...
_build_event($regex_ID, $event_ID => {
%std_info,
matchable => 0,
event_type => 'action',
msg => "Forgetting everything matched to this point",
desc => 'Pretend the final match starts here',
})
. $construct
. '(?{ local $Regexp::Grammars::match_start_pos = pos() })'
}
# Zero-width assertions...
elsif (exists $+{zero_width}) {
# Two events, so add an extra ID...
$event_ID = $next_event_ID++;
# Insert events reporting testing the assertion, and if the test succeeds...
_build_event($regex_ID, $event_ID-1 => {
%std_info,
matchable => 1,
event_type => 'pre',
msg => "Testing if $construct_desc",
desc => "Match only if $construct_desc",
})
. $construct
. _build_event($regex_ID, $event_ID => {
%std_info,
matchable => 1,
event_type => 'post',
msg => 'Assertion satisfied',
})
}
# Control verbs: (*PRUNE) (*SKIP) (*FAIL) etc...
elsif (exists $+{control}) {
# Two events, so add an extra ID...
$event_ID = $next_event_ID++;
# Insert events to report both the attempt and its success...
_build_event($regex_ID, $event_ID-1 => {
%std_info,
matchable => 1,
event_type => 'pre',
msg => 'Executing a control',
desc => 'Execute a backtracking control',
})
. $construct
. _build_event($regex_ID, $event_ID => {
%std_info,
matchable => 1,
event_type => 'post',
msg => 'Control succeeded',
})
}
# Start of DEFINE block...
elsif (exists $+{define_block}) {
# It's an unbalanced opening paren, so remember it on the stack...
push @paren_stack, {
is_capture => 0,
construct_type => '_DEFINE_block',
is_definition => 1,
};
# Insert and event to report skipping the entire block...
_build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
matchable => 0,
event_type => 'pre',
msg => 'Skipping definitions',
desc => 'The start of a definition block (skipped during matching)',
})
. $construct . '(?:'
}
# Modifier set: (?is-mx) etc...
elsif (exists $+{modifier_set}) {
# Insert an event to report the change of active modifiers...
_build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
matchable => 0,
event_type => 'compile',
msg => 'Changing modifiers',
desc => 'Change current modifiers',
})
. $construct
}
# Conditional parens: (?(COND) X | Y )...
elsif (exists $+{conditional_paren}) {
# It's an unbalanced opening paren, so remember it on the stack...
push @paren_stack, {
is_capture => 0,
is_conditional => 1,
is_pending => exists $+{pending_condition}, # ...expecting a lookahead?
construct_type => '_conditional_group',
};
# Insert an event to report the test...
'(?:'
. _build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
event_type => 'pre',
msg => 'Testing condition',
desc => 'The start of a conditional block',
})
. $construct;
}
# Branch-reset parens...
elsif (exists $+{branch_reset_paren}) {
# It's an unbalanced opening paren, so remember it on the stack...
push @paren_stack, {
is_capture => 0,
is_branch_reset => $next_capture_group+1,
construct_type => '_branch_reset_group',
};
# Insert an event to report the start of branch-reseting...
'(?:'
. _build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
event_type => 'pre',
msg => 'Starting branch-resetting group',
desc => 'The start of a branch-resetting group',
})
. $construct;
}
# Non-capturing parens...
elsif (exists $+{noncapturing_paren}) {
# Do the non-capturing parens have embedded modifiers???
my $addendum = length($construct) > 3 ? ', changing modifiers' : q{};
# It's an unbalanced opening paren, so remember it on the stack...
push @paren_stack, {
is_capture => 0,
construct_type => '_noncapture_group',
};
# Insert an event to report the start of a non-capturing group...
'(?:'
. _build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
event_type => 'pre',
msg => 'Starting non-capturing group' . $addendum,
desc => 'The start of a non-capturing group',
})
. $construct;
}
# Non-backtracking parens...
elsif (exists $+{non_backtracking_paren}) {
# It's an unbalanced opening paren, so remember it on the stack...
push @paren_stack, {
is_capture => 0,
is_nonbacktrack => 1,
construct_type => '_nonbacktracking_group',
};
# Insert an event to report the start of a non-backtracking group...
'(?:'
. _build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
event_type => 'pre',
msg => 'Starting non-backtracking group',
desc => 'The start of a non-backtracking group',
})
. '(?>';
}
# Positive lookahead/lookbehind parens...
elsif (exists $+{lookaround_paren}) {
# It's an unbalanced opening paren, so remember it on the stack...
push @paren_stack, {
is_capture => 0,
is_lookaround => $LOOKTYPE{$construct},
construct_type => '_lookaround',
};
# Is this lookaround the test of a (?(COND) X | Y) conditional???
if ($paren_stack[-2]{is_conditional} && $paren_stack[-2]{is_pending}) {
# If so, the test is no longer pending...
delete $paren_stack[-2]{is_pending};
# Insert an event to report the test...
$construct
. '(?:'
. _build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
event_type => 'pre',
msg => 'Testing for ' . $LOOKTYPE{$construct},
desc => 'Match ' . lc $LOOKTYPE{$construct},
});
}
else {
# Otherwise, insert an event to report the start of the lookaround...
'(?:'
. _build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
event_type => 'pre',
msg => 'Starting ' . $LOOKTYPE{$construct},
desc => 'Match ' . $LOOKTYPE{$construct},
})
. $construct;
}
}
# Capturing parens...
elsif (exists $+{capturing_paren}) {
# The events communicate privately via this variable...
my $shared_str_pos;
# Get the corresponding capture group number...
$next_capture_group++;
# Track the maximum group number (for after branch resets)...
$max_capture_group = max($max_capture_group, $next_capture_group);
# It's an unbalanced opening paren, so remember it on the stack...
push @paren_stack, {
is_capture => 1,
construct_type => '_capture_group',
capture_name => '$'.$next_capture_group,
shared_str_pos => \$shared_str_pos,
};
# Insert an event to report the start of capturing...
'('
. _build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
event_type => 'pre',
msg => 'Capture to $'.$next_capture_group,
desc => "The start of a capturing block (\$$next_capture_group)",
})
. '(?:';
}
# Named capturing parens...
elsif (exists $+{named_capturing_paren}) {
# The events communicate privately via this variable...
my $shared_str_pos;
# Named capture groups are also numbered, so get the number...
$next_capture_group++;
# Track the maximum group number (for after branch resets)...
$max_capture_group = max($max_capture_group, $next_capture_group);
# If this creates a new numbered capture, remember the number...
if (!@{$capture_names_for[$next_capture_group]//[]}) {
push @{$capture_names_for[$next_capture_group]}, '$'.$next_capture_group;
}
# Add this name to the list of aliases for the same numbered capture...
# (Needed because named captures in two reset branches may alias
# to the same underlying numbered capture variable. See perlre)
push @{$capture_names_for[$next_capture_group]}, '$+{'.$+{capture_name}.'}';
# It's an unbalanced opening paren, so remember it on the stack...
push @paren_stack, {
is_capture => 1,
construct_type => '_capture_group',
capture_name => $capture_names_for[$next_capture_group],
shared_str_pos => \$shared_str_pos,
};
# Insert an event to report the start of the named capture...
$construct
. _build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
event_type => 'pre',
msg => $capture_names_for[$next_capture_group],
desc => "The start of a named capturing block (also \$$next_capture_group)",
})
. '(?:';
}
# Closing parens have to be deciphered...
elsif (exists $+{closing_paren}) {
# The top of the paren stack tells us what kind of group we're closing...
my $paren_data = pop(@paren_stack) // { type=>'unmatched closing )' };
# Update the next capture group number, if after a branch reset group...
if ($paren_data->{is_branch_reset}) {
$next_capture_group = $max_capture_group;
}
# Generate an appropriate message for the type of group being closed...
my $msg = $paren_data->{is_capture} && ref $paren_data->{capture_name}
? $paren_data->{capture_name}
: $paren_data->{is_capture} ? 'End of ' . $paren_data->{capture_name}
: $paren_data->{is_definition} ? 'End of definition block'
: $paren_data->{is_branch_reset} ? 'End of branch-resetting group'
: $paren_data->{is_lookaround} ? 'End of ' . $paren_data->{is_lookaround}
: $paren_data->{is_conditional} ? 'End of conditional group'
: $paren_data->{is_nonbacktrack} ? 'End of non-backtracking group'
: 'End of non-capturing group'
;
if (length($std_info{quantifier})) {
$msg .= " (matching $quantifier_desc)";
}
# Two events, so add an extra ID...
$event_ID = $next_event_ID++;
# Append an event reporting the completion of the group...
')'
. _build_event($regex_ID, $event_ID-1 => {
%std_info,
%{$paren_data},
event_type => 'post',
msg => $msg,
desc => ( ref $msg ? 'The end of the named capturing block' : 'The e' . substr($msg,1) ),
depth => $depth - 1,
indent => $INDENT x ($depth - 1),
})
. ($paren_data->{is_nonbacktrack}
? '|'
. _build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_data},
event_type => 'failed_nonbacktracking',
msg => 'non-backtracking group',
depth => $depth - 1,
indent => $INDENT x ($depth - 1),
})
. q{(?!)}
: q{}
)
. ')'
. $std_info{quantifier};
}
# Skip comments...
elsif (exists $+{break_comment}) {
# Insert an event reporting that the break comment is being skipped...
_build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
matchable => 0,
event_type => 'break',
msg => 'Breaking at (and skipping) comment',
desc => 'Ignore this comment (but Regexp::Debugger will break here)',
})
}
# Skip comments...
elsif (exists $+{comment}) {
# Insert an event reporting that the comment is being skipped...
_build_event($regex_ID, $event_ID => {
%std_info,
%{$paren_stack[-1]},
matchable => 0,
event_type => 'skip',
msg => 'Skipping comment',
desc => 'Ignore this comment',
})
}
# Ignore (but preserve) anything else...
else {
$construct;
}
};
}exmsg;
# Remember the regex...
$state{$regex_ID}{regex_src} = $clean_regex;
# Add a preface to reset state variables in the event handler...
$raw_regex = '(?>\A(?{Regexp::Debugger::_reset_debugger_state()})(?!)'
. '|\G(?{Regexp::Debugger::_reset_debugger_state_rematch()})(?!))'
. "|(?:$raw_regex)";
say "(?#R_d:$regex_ID)".$raw_regex;
return "(?#R_d:$regex_ID)".$raw_regex;
}
#====[ Dispatch in-regex events ]================================
# How big the display window is...
my $MAX_WIDTH = 80;
my $MAX_HEIGHT = 60;
# What to print so as to "clear" the screen...
my $CLEAR_SCREEN = "\n" x $MAX_HEIGHT;
# How wide is each column in event mode...
my $EVENT_COL_WIDTH = 15;
sub _show_if_active {
my ($data_mode, $display_mode, $event_desc) = @_;
# Accumulate history...
my $history_to_date
= @{$history_of{$data_mode}//[]} ? $history_of{$data_mode}[-1]{display} : q{};
# Remember, always....
push @{$history_of{$data_mode}}, {
display => $history_to_date . $event_desc . "\n"
};
# Show, if appropriate...
if ($display_mode eq $data_mode) {
if (!$lexical_config->{save_to_fh} || $data_mode ne 'JSON') {
_print $CLEAR_SCREEN;
_say $history_of{$data_mode}[-1]{display};
}
}
}
sub _show_JSON { _show_if_active('JSON', @_) }
sub _show_event { _show_if_active('events', @_) }
# Add a new animation "frame"...
sub _new_visualize {
my ($data_mode) = @_;
push @{$history_of{$data_mode}}, { display=>q{}, is_match => 0 };
}
# Output the args and also add them to the current animation "frame"
sub _visualize {
my ($data_mode, @output) = @_;
state $NO_MATCH = 0;
state $NO_FAIL = 0;
_visualize_matchfail($data_mode, $NO_MATCH, $NO_FAIL, @output);
}
sub _visualize_matchfail {
my ($data_mode, $is_match, $is_fail, @output) = @_;
my $output = join q{}, grep {defined} @output;
if ($display_mode eq $data_mode) {
_say $output;
}
$history_of{$data_mode}[-1]{is_fail} = 1 if $is_fail;
$history_of{$data_mode}[-1]{is_match} = 1 if $is_match;
$history_of{$data_mode}[-1]{display} .= $output . "\n";
}
# Show previous animation frames...
sub _revisualize {
my ($regex_ID, $step) = @_;
# Start at the previous step unless otherwise specified...
$step //= max(0, @{$history_of{$display_mode}}-2);
my $input = '-';
STEP:
while (1) {
# Did we fall out of available history???
last STEP if $step >= @{$history_of{$display_mode}};
# Clear display and show the requested step...
if ($input ne '?') {
_print $CLEAR_SCREEN;
_print $history_of{$display_mode}[$step]{display};
if ($display_mode eq 'events' || $display_mode eq 'JSON') {
if (!$lexical_config->{save_to_fh}) {
say _info_colourer(
qq{\n\n[\u$display_mode of regex at $state{$regex_ID}{location}]}
. qq{ [step: $step]}
);
}
}
}
# Next input (but use starting cmd if one given)...
$input = _interact();
# A <CTRL-C> terminates the process...
if ($input eq "\cC") {
kill 9, $$;
}
# An 'x' exits the process...
elsif ($input eq 'x') {
exit(0);
}
# A <CTRL-L> redraws the screen...
elsif ($input eq "\cL") {
next STEP;
}
# Step back (if possible)...
elsif ($input eq '-') {
$step = max(0, $step-1);
next STEP;
}
# Display explanation of regex...
elsif ($input eq 'd') {
_show_regex_description($regex_ID);
next STEP;
}
# Help!
elsif ($input eq '?') {
_show_help();
next STEP;
}
# Swap to requested mode...
elsif ($input eq 'v') {
$display_mode = 'visual';
next STEP;
}
elsif ($input eq 'h') {
# Can we use heatmap mode?
if ($heatmaps_invisible) {
say 'Cannot show heatmaps (Term::ANSIColor unavailable)';
say "Try 'H' instead";
$input = '?';
}
# If heatmaps available, check for misuse of 'h' instead of '?'...
else {
my $prompt_help = $display_mode eq 'heatmap';
$display_mode = 'heatmap';
if ($prompt_help) {
say "(Type '?' for help)";
}
}
next STEP;
}
elsif ($input eq 'e') {
$display_mode = 'events';
say _info_colourer(
qq{\n\n[Events of regex at $state{$regex_ID}{location}]}
. qq{ [step: $step]}
);
next STEP;
}
elsif ($input eq 'j') {
$display_mode = 'JSON';
next STEP;
say _info_colourer(
qq{\n\n[JSON data of regex at $state{$regex_ID}{location}]}
. qq{ [step: $step]}
);
}
# Continue with actual matching...
elsif ($input eq 'c') {
while ($step < @{$history_of{$display_mode}}) {
_print $CLEAR_SCREEN;
_print $history_of{$display_mode}[$step++]{display};
_pause(0.1);
}
last STEP;
}
# Quit entirely...
elsif ($input eq 'q' || $input eq "\cD") {
last STEP;
}
# Take a snapshot...
elsif ($input eq 'V') { _save_snapshot('full_visual', $step) ; next STEP; }
elsif ($input eq 'H') { _save_snapshot('full_heatmap', $step) ; next STEP; }
elsif ($input eq 'E') { _save_snapshot('events', $step) ; next STEP; }
elsif ($input eq 'J') { _save_snapshot('JSON', $step) ; next STEP; }
elsif ($input eq 'D') { _show_regex_description($regex_ID,'save') ; next STEP; }
# Step forward to match...
elsif ($input eq 'm') {
$step++;
while (!$history_of{$display_mode}[$step]{is_match}) {
last STEP if $step == @{$history_of{$display_mode}}-1;
$step++;
}
}
# Step forward to fail...
elsif ($input eq 'f') {
$step++;
while (!$history_of{$display_mode}[$step]{is_fail}) {
last STEP if $step == @{$history_of{$display_mode}}-1;
$step++;
}
}
# Otherwise just step forward...
else {
$step++;
}
}
# Return final command...
return ($input, $step);
}
sub _build_visualization {
# Unpack all the info needed...
my ($data_mode, $named_args_ref) = @_;
my ($regex_ID, $regex_src, $regex_pos, $construct_len,
$str_src, $str_pos,
$is_match, $is_fail, $is_trying, $is_capture,
$backtrack, $forward_step, $nested_because,
$msg, $colourer, $no_window, $step)
= @{$named_args_ref}{qw(
regex_ID regex_src regex_pos construct_len
str_src str_pos
is_match is_fail is_trying is_capture
backtrack forward_step nested_because
msg colourer no_window step
)};
# Clear screen...
_new_visualize($data_mode);
if (!$no_window) {
_visualize $data_mode, q{} for 1..$MAX_HEIGHT;
}
# Remember originals...
my $raw_str_src = $str_src;
my $raw_regex_src = $regex_src;
# Unwindowed displays show the title first...
if ($no_window) {
_visualize $data_mode,
_info_colourer(
qq{\n[\u$data_mode of regex at $state{$regex_ID}{location}]\n\n}
. qq{ [step: $step]}
);
}
# Visualize capture vars, if available...
my $max_name_width = 1 + max map {length} 0, keys %capture;
CAPVAR:
for my $name (do{ no warnings 'numeric'; sort { substr($a,1) <=> substr($b,1) } keys %capture}) {
# Remove any captures that are invalidated by backtracking...
if ($capture{$name}{start_pos} > $regex_pos) {
delete @{$capture{$name}}{'from','to'};
}
# Clean up and visualize each remaining variable...
my $start = $capture{$name}{from} // next CAPVAR;
my $end = $capture{$name}{to} // next CAPVAR;
my $cap_str = _quote_ws(substr($_,$start,$end-$start));
# Truncate captured value to maximum width by removing middle...
my $cap_len = length($cap_str);
if ($cap_len > $MAX_WIDTH) {
my $middle = $MAX_WIDTH/2 - 2;
substr($cap_str, $middle, -$middle, '....');
}
# Display capture var and value...
_visualize $data_mode,
_info_colourer(sprintf qq{%*s = '%s'}, $max_name_width, $name, $cap_str);
}
# Visualize special var, if used in regex...
_visualize $data_mode, q{};
if (index($raw_regex_src, '$^N') >= 0 && defined $^N) {
my $special_val = $^N;
# Truncate captured value to maximum width by removing middle...
my $cap_len = length($special_val);
if ($cap_len > $MAX_WIDTH) {
my $middle = $MAX_WIDTH/2 - 2;
substr($special_val, $middle, -$middle, '....');
}
# Display capture var and value...
_visualize $data_mode,
_info_colourer(sprintf qq{%*s = '%s'}, $max_name_width, '$^N', $special_val);
}
# Leave a gap...
_visualize $data_mode, q{} for 1..2;
# Show matching...
_visualize_matchfail $data_mode, $is_match, $is_fail;
# Reconfigure regex within visible window...
($regex_src, $regex_pos)
= _make_window(
text => $regex_src,
pos => $regex_pos,
heat => substr($data_mode, -7) eq 'heatmap' ? $history_of{match_heatmap} : [],
ws_colour => substr($data_mode, -7) eq 'heatmap',
no_window => $no_window,
);
# How wide is the display???
my $display_width
= $no_window ? $regex_pos
: min($regex_pos, $MAX_WIDTH - length($msg));
# Draw the regex with a message and a positional marker...
if ($data_mode ne 'full_heatmap') {
_visualize $data_mode, q{ }, q{ } x $display_width, $colourer->($msg);
_visualize $data_mode, q{ }, q{ } x $regex_pos , $colourer->('|');
_visualize $data_mode, q{ }, q{ } x $regex_pos , $colourer->('V') x ($construct_len || 1);
}
else {
_visualize $data_mode, q{ }, q{ } x $regex_pos , _info_colourer('|');
_visualize $data_mode, q{ }, q{ } x $regex_pos , _info_colourer('V' x ($construct_len || 1) );
}
# Draw regex itself...
_visualize $data_mode, q{/}, $regex_src, q{/};
# Leave a gap...
_visualize $data_mode, q{ } for 1..2;
# Create marker for any match or capture within string...
$forward_step = min($forward_step, $MAX_WIDTH);
my $last_match_marker
= (q{ } x ($str_pos - max(0,$forward_step)))
. ( $nested_because eq 'failed' ? q{}
: $is_capture && $forward_step == 1 ? 'V'
: $is_capture && $forward_step > 1 ? '\\' . ('_' x ($forward_step-2)) . '/'
: '^' x $forward_step
)
;
# Reconfigure string within visible window...
my $match_start;
($str_src, $str_pos, $match_start, $last_match_marker)
= _make_window(
text => $str_src,
pos => $str_pos,
start => $Regexp::Grammars::match_start_pos,
heat => substr($data_mode, -7) eq 'heatmap' ? $history_of{string_heatmap} : [],
ws_colour => substr($data_mode, -7) eq 'heatmap',
marker => $last_match_marker,
no_window => $no_window,
);
# Trim match start position...
if ($match_start > $str_pos) {
$match_start = $str_pos;
}
# Colour match marker...
$last_match_marker
= substr($last_match_marker,0,1) eq '^' ? _match_colourer($last_match_marker, 'reverse')
: _info_colourer($last_match_marker);
# Draw the string with a positional marker...
_visualize $data_mode,
q{ }, _info_colourer( substr(q{ } x $str_pos . '|' . $backtrack, 0, $MAX_WIDTH-2) );
_visualize $data_mode,
q{ }, q{ } x $match_start, _match_colourer($MATCH_DRAG x ($str_pos-$match_start)), _info_colourer('V');
$str_src = # Heatmap is already coloured...
substr($data_mode, -7) eq 'heatmap' ?
$str_src
# On failure, fail-colour to current position...
: $nested_because eq 'failed' ?
_fail_colourer( substr($str_src, 0, $str_pos), 'ws' )
. _ws_colourer( substr($str_src, $str_pos) )
# When trying, try-colour current position
: $is_trying ?
_fail_colourer( substr($str_src, 0, $match_start), 'ws' )
. _match_colourer( substr($str_src, $match_start, $str_pos-$match_start), 'underline', 'ws' )
. _try_colourer( substr($str_src, $str_pos, 1), 'underline bold', 'ws' )
. _ws_colourer( substr($str_src, min(length($str_src),$str_pos+1)) )
: # Otherwise, report pre-failure and current match...
_fail_colourer( substr($str_src, 0, $match_start), 'ws' )
. _match_colourer( substr($str_src, $match_start, $str_pos-$match_start), 'underline', 'ws' )
. _ws_colourer( substr($str_src, $str_pos) );
_visualize $data_mode, q{'}, $str_src, q{'}; # String itself
# Draw a marker for any match or capture within the string...
_visualize $data_mode, q{ }, $last_match_marker;
# Windowed displays show the title last...
if (!$no_window) {
_visualize $data_mode,
_info_colourer(
qq{\n[\u$data_mode of regex at $state{$regex_ID}{location}]}
. qq{ [step: $step]}
);
}
# Special case: full heatmaps are reported as a table too...
if ( $data_mode eq 'full_heatmap' ) {
# Tabulate regex...
_visualize $data_mode, _info_colourer("\n\nHeatmap for regex:\n");
_visualize $data_mode, _build_tabulated_heatmap($raw_regex_src, $history_of{match_heatmap});
# Tabulate string...
_visualize $data_mode, _info_colourer("\n\nHeatmap for string:\n");
_visualize $data_mode, _build_tabulated_heatmap($raw_str_src, $history_of{string_heatmap});
}
}
# Convert a heatmapped string to a table...
my $TABLE_STR_WIDTH = 15;
sub _build_tabulated_heatmap {
my ($str, $heatmap_ref) = @_;
# Normalized data...
my $max_heat = max(1, map { $_ // 0 } @{$heatmap_ref});
my @heat = map { ($_//0) / $max_heat } @{$heatmap_ref};
my $count_size = length($max_heat);
# Determine colours to be used...
my @HEAT_COLOUR = @{$lexical_config->{heatmap_col}};
# Accumulate graph
my @graph;
for my $index (0..length($str)-1) {
# Locate next char and its heat value...
my $char = substr($str, $index, 1);
my $abs_heat = $heatmap_ref->[$index] // 0;
my $display_char = $char eq "\n" ? '\n'
: $char eq "\t" ? '\t'
: $char;
# Graph it...
if (@graph && length($graph[-1]{text} . $display_char) < $TABLE_STR_WIDTH && $graph[-1]{heat} == $abs_heat) {
$graph[-1]{text} .= $display_char;
}
elsif ($char ne q{ } || $abs_heat != 0) {
my $rel_heat = $heat[$index] // 0;
push @graph, {
text => $display_char,
heat => $abs_heat,
rel_heat => $rel_heat,
bar => q{*} x (($MAX_WIDTH-$TABLE_STR_WIDTH) * $rel_heat),
};
}
}
# Draw table...
my $table;
for my $entry (@graph) {
my $colour_index = int( 0.5 + $#HEAT_COLOUR * $entry->{rel_heat} );
$table .=
q{ } .
Term::ANSIColor::colored(
substr($entry->{text} . q{ } x $TABLE_STR_WIDTH, 0, $TABLE_STR_WIDTH) .
sprintf("| %-*s |%s\n", $count_size, $entry->{heat} || q{ }, $entry->{bar}),
$HEAT_COLOUR[$colour_index]
);
}
return $table;
}
# These need to be localized within regexes, so have to be package vars...
our $subpattern_depth; # ...how many levels down in named subpatterns?
# Reset debugger variables at start of match...
sub _reset_debugger_state {
$prev_regex_pos = 0; # ...start of regex
$start_str_pos = 0; # ...starting point of match of string
$prev_str_pos = 0; # ...start of string
$prev_match_was_null = 0; # ...no previous match (to have been null)
$pre_is_pending = undef; # ...no try is pending
$interaction_mode = 's'; # ...always start in step-by-step mode
$interaction_quit = 0; # ...reset quit command for each regex
$subpattern_depth = 0; # ...start at top level of named subcalls
$Regexp::Grammars::match_start_pos = 0; # ...start matching at start of string
# Also leave a gap in the event history and JSON representations...
_show_event $lexical_config->{display_mode}, q{};
_show_JSON $lexical_config->{display_mode}, q{};
}
# Reset some debugger variables at restart of match...
sub _reset_debugger_state_rematch {
$prev_regex_pos = 0; # ...start of regex
$start_str_pos = pos; # ...starting point of match of string
$prev_str_pos = pos; # ...point of rematch
$pre_is_pending = undef; # ...no try is pending
$interaction_mode = 's'; # ...always start in step-by-step mode
$subpattern_depth = 0; # ...start at top level of named subcalls
$Regexp::Grammars::match_start_pos = pos; # ...start matching at rematch point
# Also leave a gap in the event history and JSON representations...
_show_event $lexical_config->{display_mode}, q{};
_show_JSON $lexical_config->{display_mode}, q{};
}
# Set up a JSON encoder...
my ($JSON_encoder, $JSON_decoder);
BEGIN {
($JSON_encoder, $JSON_decoder) =
eval{ require JSON::XS; } ? do {
my $json = JSON::XS->new->utf8(1)->pretty(1);
(
sub { return $json->encode(shift) },
sub { return $json->decode(shift) },
)
}
: eval{ require JSON; } ? do {
my $json = JSON->new->pretty(1);
(
sub { return $json->encode(shift) },
sub { return $json->decode(shift) },
)
}
: eval{ require 5.014;
require JSON::DWIW; } ? (
sub { JSON::DWIW->to_json(shift, {pretty=>1}) },
sub { JSON::DWIW->from_json(shift, {pretty=>1}) },
)
: eval{ require JSON::Syck; } ? (
\&JSON::Syck::Dump,
\&JSON::Syck::Load,
)
: (
sub { '{}' },
sub { {} },
);
}
# Report some activity within the regex match...
sub _report_event {
# Did the user quit the interactive debugger???
return if $interaction_quit;
# What are we matching....
my $str_src = $_;
# Which regex? Which event? Where in the string? Is this a recursive call?
my ($regex_ID, $event_ID, $str_pos, %opt) = @_;
my $nested_because = $opt{nested_because} // q{};
my $non_interactive = $opt{non_iteractive};
# Locate state info for this event...
my $state_ref = $state{$regex_ID};
my $event_ref = $state_ref->{$event_ID};
# Unpack the necessary info...
my ($matchable, $is_capture, $event_type, $construct)
= @{$event_ref}{qw< matchable is_capture event_type construct>};
my ($construct_type, $quantifier, $regex_pos, $capture_name, $msg)
= @{$event_ref}{qw< construct_type quantifier regex_pos capture_name msg>};
$construct_type //= q{};
# Reset display_mode, capture variables, and starting position on every restart...
if ($construct_type eq '_START') {
%capture = ();
$Regexp::Grammars::match_start_pos = pos();
$lexical_config = $config[$event_ref->{lexical_scope}];
# Reset display mode only on start (i.e. not on restart)...
if ($str_pos == 0) {
$display_mode = $lexical_config->{display_mode};
}
}
# Ignore final failure messages, except at the very end...
if ($event_ref->{regex_failed}) {
return if ($str_pos//0) < length($str_src);
}
# This variable allows us to query the start position of a submatch when at the end of the submatch...
my $shared_str_pos_ref = $event_ref->{shared_str_pos};
# Use the shared string pos on failure...
if ($nested_because eq 'failed') {
$str_pos = ${$shared_str_pos_ref // \$prev_str_pos} // $str_pos;
}
# Flatten aliased capture name(s)...
if (ref $capture_name) {
$capture_name = join ' and ', @{$capture_name}
}
# If we've matched, what did we match???
my $forward_step = 0; # ... will eventually contain how far forward we stepped
if (($matchable || $is_capture) && $event_type eq 'post' && $construct ne '|') {
$forward_step = $str_pos - ($shared_str_pos_ref ? ${$shared_str_pos_ref} : $str_pos);
}
my $backtrack = q{}; # ...will store the arrow demonstrating the backtracking
# Are we backtracking?
my $str_backtrack_len = min($EVENT_COL_WIDTH-1, $prev_str_pos-$str_pos);
my $regex_backtrack_len = min($EVENT_COL_WIDTH-1, $prev_regex_pos-$regex_pos);
my $event_str = '<' . '~' x $str_backtrack_len;
my $event_regex = '<' . '~' x $regex_backtrack_len;
if ($nested_because ne 'failed') {
# Generate backtracking arrow...
if ($str_pos < ($prev_str_pos//0)) {
$backtrack = '<' . '~' x ($prev_str_pos-$str_pos-1);
}
elsif ($regex_pos < ($prev_regex_pos//0)) {
$backtrack = ' ';
}
# Remember where we were...
$prev_str_pos = $str_pos;
$prev_regex_pos = $regex_pos;
}
# Was there a failed pending attempt???
my $pending_event_ID = $pre_is_pending;
$pre_is_pending = undef;
if (($event_type ne 'post' || $backtrack) && defined $pending_event_ID) {
_report_event($regex_ID, $pending_event_ID, undef, nested_because=>'failed');
}
# Get the source code of the regex...
my $regex_src = $state_ref->{regex_src};
# How long is this piece of the regex???
my $construct_len = length $construct;
# Build msg if it's dynamic...
if (ref($msg) eq 'CODE') {
$msg = $msg->();
}
# Construct status message (if necessary)...
$msg = $nested_because eq 'failed' ? q{Failed}
: $event_type eq 'pre' && ref $msg ? 'Capture to ' . join ' and ', @{$msg}
: $event_type eq 'post' && ref $msg ? 'End of ' . join ' and ', @{$msg}
: defined $msg ? $msg
: pos == $prev_str_pos && $construct_type eq '_START' ? q{Restarting regex match}
: $construct_type eq '_START' ? q{Starting regex match}
: q{}
;
# Report back-tracking occurred (but not when returning from named subpatterns)...
if ($backtrack && $event_type eq 'failed_nonbacktracking') {
$msg = q{Back-tracking past } . lc($msg) . q{ without rematching};
}
elsif ($backtrack && $construct_type ne '_named_subpattern_call') {
$msg = q{Back-tracked within regex and re} . lc($msg);
}
# Track trying and matching...
my $is_match = index($msg, 'matched') >= 0 || index($msg, 'Matched') >= 0;
my $is_rematch = index($msg, 'rematched') >= 0 || index($msg, 'Rematched') >= 0;
my $is_trying = index($msg, 'trying') >= 0 || index($msg, 'Trying') >= 0;
my $is_skip = index($msg, 'skipping') >= 0 || index($msg, 'Skipping') >= 0;
my $is_fail = index($msg, 'failed') >= 0 || index($msg, 'Failed') >= 0;
# Track string heatmap...
if ($forward_step) {
my @str_range = $str_pos-$forward_step+1 .. $str_pos-1;
$_++ for @{$history_of{string_heatmap}}[@str_range];
}
elsif ($is_trying) {
$history_of{string_heatmap}[$str_pos]++;
}
# Trace regex heatmap...
if ($is_rematch || !$is_match && !$is_fail && !$is_skip) {
my @regex_range = $regex_pos..$regex_pos+length($construct)-1;
$_++ for @{$history_of{match_heatmap}}[@regex_range];
}
# Track start and end positions for each capture...
if ($construct_type eq '_capture_group') {
if ($event_type eq 'pre') {
$capture{$capture_name}{from} = $str_pos;
$capture{$capture_name}{start_pos} = $regex_pos;
}
elsif ($event_type eq 'post') {
$capture{$capture_name}{to} = $str_pos;
}
}
# Remember when a match/fail is pending...
my $is_pending = $matchable
&& $event_type eq 'pre'
&& $construct_type ne '_named_subpattern_call';
if ($is_pending) {
# Pre- and post- events have adjacent IDs so add 1 to get post ID...
$pre_is_pending = $event_ID + 1;
}
# Send starting position to corresponding post- event...
if ($shared_str_pos_ref && $event_type eq 'pre' && $construct ne '|') {
${$shared_str_pos_ref} = $str_pos;
}
# Compute indent for message (from paren depth + subcall depth)...
my $indent = $INDENT x ($event_ref->{depth} + $subpattern_depth);
# Indicate any backtracking...
if (length($event_str) > 1 || length($event_regex) > 1) {
$event_str = q{} if length($event_str) == 1;
$event_regex = q{} if length($event_regex) == 1;
my $backtrack_msg
= $event_str && $event_regex ? 'Back-tracking in both regex and string'
: $event_str ? 'Back-tracking ' . $str_backtrack_len
. ' character'
. ($str_backtrack_len == 1 ? q{} : 's')
. ' in string'
: "Back-tracking in regex"
;
$backtrack_msg = _info_colourer($backtrack_msg);
$event_regex .= q{ } x ($EVENT_COL_WIDTH - length $event_regex);
$event_str .= q{ } x ($EVENT_COL_WIDTH - length $event_str);
_show_event $display_mode,
sprintf("%s | %s | %s",
_info_colourer($event_str),
_info_colourer($event_regex),
$indent . $backtrack_msg);
}
# Colour the message...
my $colourer = _colourer_for($msg);
# Log (and perhaps display) event...
_show_event $display_mode,
sprintf("%-s | %-${EVENT_COL_WIDTH}s | %s",
_ws_colourer(substr($str_src . (q{ } x $EVENT_COL_WIDTH), $str_pos, $EVENT_COL_WIDTH)),
substr($regex_src, $regex_pos, $EVENT_COL_WIDTH),
$indent . $colourer->($msg));
# Display event mode line, if appropriate...
if ($display_mode eq 'events' && !$lexical_config->{save_to_fh}) {
say _info_colourer( qq{\n[Events of regex at $state{$regex_ID}{location}]} );
}
# Generate (and perhaps display) the JSON...
{
# The data we're encoding...
my $data = {
regex_pos => $regex_pos,
str_pos => $str_pos,
event => { %{$event_ref}, msg => $msg },
};
# But sanitize any procedural msg...
if (ref $data->{event}{msg} eq 'CODE') {
delete $data->{event}{msg};
}
# And sanitize any reference to internal communications channel...
my $starting_str_pos = delete $data->{event}{shared_str_pos};
if (ref $starting_str_pos eq 'SCALAR' && ${$starting_str_pos} && ${$starting_str_pos} ne $str_pos) {
$data->{starting_str_pos} = ${$starting_str_pos};
}
my $json_rep = $JSON_encoder->($data);
# Display opening delimiter at start...
if ($construct_type eq '_START' && $str_pos == 0) {
_show_JSON $display_mode, '[';
}
# Display event data (with comma, if needed)...
my $comma = $construct_type eq '_END' ? q{} : q{,};
_show_JSON $display_mode, qq{ $json_rep$comma};
# Display closing delimiter at end...
if ($construct_type eq '_END') {
_show_JSON $display_mode, ']';
}
# Display mode line...
if ($display_mode eq 'JSON' && !$lexical_config->{save_to_fh}) {
say _info_colourer( qq{\n[JSON data of regex at $state{$regex_ID}{location}]} );
}
}
# Build and display (if appropriate) the "2D" visualizations...
my %data = (
regex_ID => $regex_ID,
regex_src => $regex_src,
regex_pos => $regex_pos,
construct_len => $construct_len,
str_src => $str_src,
str_pos => $str_pos,
is_match => $is_match,
is_fail => $is_fail,
is_trying => $is_trying,
is_capture => $is_capture,
backtrack => $backtrack,
forward_step => $forward_step,
nested_because => $nested_because,
msg => $msg,
colourer => $colourer,
step => scalar @{$history_of{visual}||[]},
);
_build_visualization('visual', \%data);
_build_visualization('heatmap', \%data);
$data{no_window} = 1;
_build_visualization('full_visual', \%data);
_build_visualization('full_heatmap', \%data);
# Do any interaction...
my $input;
INPUT:
while (!$non_interactive) {
# Adaptive rate of display when skipping interactions...
state $skip_rate = 0.1;
state $MIN_SKIP_RATE = 0.001;
$skip_rate = max($MIN_SKIP_RATE, $skip_rate * 0.98);
_pause($skip_rate);
# Skip interactions if current mode does not require them...
last INPUT if $event_type ne 'break' && (
# Skip-to-match mode...
$interaction_mode eq 'm'
&& !$is_match
&& index($msg,'Back-tracked') < 0
||
# Skip-to-fail mode...
$interaction_mode eq 'f'
&& !$is_fail
||
# Skip-to-end mode...
$interaction_mode eq 'c'
&& $construct_type ne '_END'
);
# Reset adaptive skip rate on any interaction...
$skip_rate = 0.1;
# Reset to step mode on a break...
if ($event_type eq 'break') {
$interaction_mode = 's';
}
# Do what, John???
$input = _interact();
# A <CTRL-C> terminates the process...
if ($input eq "\cC") {
kill 9, $$;
}
# An 'x' exits the process...
elsif ($input eq 'x') {
exit(0);
}
# A <CTRL-L> redraws the screen...
elsif ($input eq "\cL") {
_print $history_of{$display_mode}[-1]{display};
if ($display_mode eq 'events' || $display_mode eq 'JSON') {
say _info_colourer( qq{\n\n[\u$display_mode of regex at $state{$regex_ID}{location}]} );
}
next INPUT;
}
# Display explanation of regex...
elsif ($input eq 'd') {
_show_regex_description($regex_ID);
next INPUT;
}
# Help!
elsif ($input eq '?') {
_show_help();
next INPUT;
}
# Quit all debugging???
elsif ($input eq 'q' || $input eq "\cD") {
$interaction_quit = 1;
last INPUT;
}
# Step backwards...
elsif ($input eq '-') {
($input) = _revisualize($regex_ID);
if ($input eq 'q' || $input eq "\cD") {
$interaction_quit = 1;
last INPUT;
}
next INPUT;
}
# Switch between visualizer/event/heatmap/JSON modes...
elsif ($input eq 'v') {
$display_mode = 'visual';
_print $history_of{'visual'}[-1]{display};
next INPUT;
}
elsif ($input eq 'h') {
# Can we use heatmap mode?
if ($heatmaps_invisible) {
say 'Cannot show heatmaps (Term::ANSIColor unavailable)';
say "Try 'H' instead";
$input = '?';
}
# If heatmaps available, check for misuse of 'h' instead of '?'...
else {
my $prompt_help = $display_mode eq 'heatmap';
$display_mode = 'heatmap';
_print $CLEAR_SCREEN;
_print $history_of{'heatmap'}[-1]{display};
if ($prompt_help) {
say "(Type '?' for help)";
}
}
next INPUT;
}
elsif ($input eq 'e') {
$display_mode = 'events';
_print $CLEAR_SCREEN;
_print $history_of{'events'}[-1]{display};
say _info_colourer( qq{\n\n[Events of regex at $state{$regex_ID}{location}]} );
next INPUT;
}
elsif ($input eq 'j') {
$display_mode = 'JSON';
_print $CLEAR_SCREEN;
_print $history_of{'JSON'}[-1]{display};
say _info_colourer( qq{\n\n[JSON data of regex at $state{$regex_ID}{location}]} );
next INPUT;
}
# Take a snapshot...
elsif ($input eq 'V') { _save_snapshot('full_visual') ; next INPUT; }
elsif ($input eq 'H') { _save_snapshot('full_heatmap') ; next INPUT; }
elsif ($input eq 'E') { _save_snapshot('events') ; next INPUT; }
elsif ($input eq 'J') { _save_snapshot('JSON') ; next INPUT; }
elsif ($input eq 'D') { _show_regex_description($regex_ID,'save') ; next INPUT; }
# Require an explicit quit request at end of match...
# elsif ($construct_type eq '_END') {
# print $history_of{$display_mode}[-1]{display};
# say Term::ANSIColor::colored("(type 'q' to leave visualization)", $ERR_COL);
# next INPUT;
# }
# Change of interaction mode???
elsif ($input ~~ ['f','m','s','c']) {
$interaction_mode = $input;
last INPUT;
}
# Otherwise, move on...
else {
last INPUT;
}
}
# At end of debugging, save data to file (if requested), and clean up...
if ($construct_type eq '_END') {
_save_to_fh($regex_ID, $str_src);
%history_of = ();
$history_of{match_heatmap} = [];
$history_of{string_heatmap} = [];
}
return $input;
}
# Dump all history and config data to a stream...
sub _save_to_fh {
my ($regex_ID, $str_src) = @_;
# No-op if not saving to file...
my $fh = delete $lexical_config->{save_to_fh}
or return;
# Extract data to correct level...
my $match_heatmap = delete $history_of{match_heatmap};
my $string_heatmap = delete $history_of{string_heatmap};
my $location = $state{$regex_ID}{location};
my $regex_display = $state{$regex_ID}{regex_src};
my $regex_original = $state{$regex_ID}{raw_regex};
# Ensure print prints everything...
my $prev_select = select $fh;
local $|=1;
# Encode and print...
print {$fh} $JSON_encoder->({
regex_ID => $regex_ID,
regex_location => $location,
regex_original => $regex_original,
regex_display => $regex_display,
string_display => $str_src,
config => $lexical_config,
match_data => $JSON_decoder->($history_of{JSON}[-1]{display}),
match_heatmap => $match_heatmap,
string_heatmap => $string_heatmap,
visualization => \%history_of,
}), "\n";
# Restore filehandles...
select $prev_select;
$lexical_config->{save_to_fh} = $fh;
}
sub _show_regex_description {
my ($regex_ID, $save) = @_;
# How wide to display regex components...
my $MAX_DISPLAY = 20;
# The info we're displaying...
my $info = $state{$regex_ID};
# Coloured separator...
my $separator = $save ? q{}
: Term::ANSIColor::colored(
q{ } x $MAX_WIDTH . "\n",
$lexical_config->{desc_sep_col}
);
# Direct the output...
my $STDOUT;
if ($save) {
$STDOUT = _prompt_for_file('description');
}
else {
my $pager = $ENV{PAGER} // 'more';
if ($pager eq 'less') {
$pager .= ' -R';
}
open $STDOUT, '|-', $pager or return;
}
# Build the display...
say {$STDOUT}
$separator
. join q{},
map {
my $indent = $info->{$_}{indent};
my $construct = sprintf('%-*s', $MAX_DISPLAY, $indent . $info->{$_}{construct});
my $desc = $indent . $info->{$_}{desc};
# Decorate according to destination...
if ($save) {
$desc = '#' . $desc
}
else {
$construct = Term::ANSIColor::colored($construct, $lexical_config->{desc_regex_col});
$desc = Term::ANSIColor::colored($desc, $lexical_config->{desc_text_col});
}
# Format and return...
if (length($indent . $info->{$_}{construct}) > 20) {
$construct . "\n"
. q{ } x ($MAX_DISPLAY+2) . "$desc\n"
. $separator
}
else {
"$construct $desc\n"
. $separator
}
}
sort { $a <=> $b }
grep { /^\d+$/ && exists $info->{$_}{desc} }
keys %$info;
}
sub _show_help {
say <<'END_HELP';
________________________________________________/ Help \______
Motion: s : step forwards
- : step backwards
m : continue to next partial match
f : continue to next partial failure
c : continue to end of full match
<RET> : repeat last motion
Display: v : change to visualization
e : change to event log
h : change to heatmaps
j : change to JSON representation
d : describe the regex in detail
Snapshot: V : take snapshot of current visualization
E : take snapshot of current event log
H : take snapshot of current heatmaps
J : take snapshot of current JSON representation
D : take snapshot of regex description
Control: q : quit debugger and continue program
x : exit debugger and terminate program
______________________________________________________________
END_HELP
}
# Take a snapshot of the current debugger state...
my @ERR_MODE = ( -timeout => 10, -style => $ERR_COL, -single);
sub _prompt_for_file {
my ($data_mode) = @_;
if (!eval { require Time::HiRes; }) {
*Time::HiRes::time = sub { time };
}
# Default target for save...
my $open_mode = '>';
my $filename = 'rxrx_' . $data_mode . '_' . Time::HiRes::time();
# Request a filename...
print "Save $data_mode snapshot as: ";
my $input = _interact();
# Default to paged-to-screen...
if ($input eq "\n") {
say '<screen>';
$open_mode = '|-';
$filename = $ENV{PAGER} // 'more';
if ($filename eq 'less') {
$filename .= ' -R';
}
}
# <TAB> selects precomputed filename...
elsif ($input eq "\t") {
say $filename;
_pause(2);
}
# Otherwise, use whatever they type...
else {
$filename = $input;
print $input;
$filename .= readline *STDIN;
chomp $filename;
}
# Set up the output stream...
open my $fh, $open_mode, $filename or do {
say Term::ANSIColor::colored("Can't open $filename: $!", $ERR_COL);
say Term::ANSIColor::colored("(Hit any key to continue)", $ERR_COL);
_interact();
return;
};
return $fh;
}
sub _save_snapshot {
my ($data_mode, $step) = @_;
$step //= -1;
# Open the save target...
my $fh = _prompt_for_file($data_mode);
# Output current state (appropriately trimmed)...
my $state = $history_of{$data_mode}[$step]{display};
while (substr($state, 0, 1) eq "\n") {
substr($state, 0, 1, q{});
}
print {$fh} $state;
# JSON output may be partial...
if ($data_mode eq 'JSON' && substr($state, -2) eq ",\n") {
print {$fh} " { MATCH_INCOMPLETE => 1 }\n]\n";
}
# Clean up...
close $fh;
# Restore previous visuals...
_print $history_of{$display_mode}[-1]{display};
return;
}
sub _build_heatmap {
my ($str, $count_ref) = @_;
# Determine colours to be used...
my @HEAT_COLOUR = @{$lexical_config->{heatmap_col}};
# Normalize counts to match @HEAT_COLOUR entries...
my $max = max 1, map { $_ // 0 } @{$count_ref};
my @count = map { int( 0.5 + $#HEAT_COLOUR * ($_//0) / $max ) } @{$count_ref};
# Colour each character...
my $heatmap = q{};
for my $n (0..length($str)-1) {
my $heat = $HEAT_COLOUR[$count[$n] // 0];
$heatmap .= _ws_colourer(substr($str,$n,1), $heat);
}
return $heatmap;
}
# Extract a window-into-string to fit it on screen...
sub _make_window {
my %arg = @_;
my $src = $arg{text} // q{};
my $pos = $arg{pos} // 0;
my $start_pos = $arg{start} // 0;
my @heatmap = @{ $arg{heat} // [] };
my $ws_colour = $arg{ws_colour};
my $window = !$arg{no_window};
my $marker = $arg{marker};
# Extend heatmap and marker to length of text...
if (@heatmap) {
push @heatmap, (0) x (length($src) - @heatmap);
}
if ($marker) {
$marker .= q{ } x (length($src) - length($marker));
}
# Crop to window, if necessary...
if ($window) {
# How big is the space we have to fill???
my $window_width = $MAX_WIDTH - 2; # ...allow 2 chars for delimiters
my $mid_window = $MAX_WIDTH/2;
# Only modify values if content longer than window...
if (length($src) > $window_width) {
# At the start of the string, chop off the end...
if ($pos <= $mid_window) {
if ($marker) {
$marker = substr($marker, 0, $window_width);
}
$src = substr($src, 0, $window_width);
substr($src,-3,3,q{...});
}
# At the end of the string, chop off the start...
elsif (length($src) - $pos < $mid_window) {
$pos = $window_width - length($src) + $pos;
$start_pos = $window_width - length($src) + $start_pos;
if (@heatmap) {
@heatmap = @heatmap[length($src)-$window_width..$#heatmap];
}
if ($marker) {
$marker = substr($marker, length($src)-$window_width, $window_width);
}
$src = substr($src, -$window_width);
substr($src,0,3,q{...});
}
# In the middle of the string, centre the window on the position...
else {
$src = substr($src, $pos-$mid_window+1, $window_width);
if (@heatmap) {
@heatmap= splice(@heatmap, $pos-$mid_window+1, $window_width);
}
if ($marker) {
$marker = substr($marker, $pos-$mid_window+1, $window_width);
}
$start_pos -= $pos;
$pos = $window_width/2;
$start_pos += $pos;
substr($src,0,3,q{...});
substr($src,-3,3,q{...});
}
}
}
# Convert to heatmap, if requested...
if (@heatmap) {
$src = _build_heatmap($src, \@heatmap);
}
elsif ($ws_colour) {
$src = _ws_colourer($src);
}
# Trim trailing whitespace from marker...
while ($marker && substr($marker,-1) eq q{ }) {
substr($marker, -1) = q{};
}
return ($src, $pos, max($start_pos,0), $marker);
}
# Colour message appropriately...
sub _fail_colourer {
my ($str, $ws_colouring) = @_;
my $colourer = $ws_colouring ? \&_ws_colourer : \&Term::ANSIColor::colored;
return $colourer->($str, $lexical_config->{fail_col});
}
sub _info_colourer {
my ($str, $ws_colouring) = @_;
my $colourer = $ws_colouring ? \&_ws_colourer : \&Term::ANSIColor::colored;
return $colourer->($str, $lexical_config->{info_col});
}
sub _try_colourer {
my ($str, $extras, $ws_colouring) = @_;
$extras //= q{};
my $colourer = $ws_colouring ? \&_ws_colourer : \&Term::ANSIColor::colored;
return $colourer->($str, "$lexical_config->{try_col} $extras");
}
sub _match_colourer {
my ($str, $extras, $ws_colouring) = @_;
$extras //= q{};
my $colourer = $ws_colouring ? \&_ws_colourer : \&Term::ANSIColor::colored;
return $colourer->($str, "$lexical_config->{match_col} $extras");
}
my %DISPLAY_FOR = (
"\n" => 'n',
"\t" => 't',
"\r" => 'r',
"\f" => 'f',
"\b" => 'b',
"\a" => 'a',
"\e" => 'e',
"\0" => '0',
);
sub _ws_colourer {
my ($str, $colour_scheme) = @_;
# How to colour the text...
$colour_scheme //= 'clear';
my $ws_colour_scheme = "$colour_scheme $lexical_config->{ws_col}";
# Accumulate the text...
my $coloured_str = q{};
my $prefix = q{};
# Step through char-by-char...
CHAR:
for my $n (0..length($str)-1) {
my $char = substr($str, $n, 1);
# If it's special, handle it...
for my $special_char (keys %DISPLAY_FOR) {
if ($char eq $special_char) {
if (length($prefix)) {
$coloured_str .= Term::ANSIColor::colored($prefix, $colour_scheme);
$prefix = q{};
}
$coloured_str .= Term::ANSIColor::colored($DISPLAY_FOR{$special_char}, $ws_colour_scheme);
next CHAR;
}
}
# Otherwise, accumulate it...
$prefix .= $char;
}
# Clean up any remaining text...
if (length($prefix)) {
$coloured_str .= Term::ANSIColor::colored($prefix, $colour_scheme);
}
return $coloured_str;
}
sub _colourer_for {
my $msg = shift;
if (index($msg,'forgetting') >= 0 || index($msg,'Forgetting') >= 0) {
return \&_info_colourer;
}
if (index($msg,'try') >= 0 || index($msg,'Try') >= 0) {
return \&_try_colourer;
}
if (index($msg,'failed') >= 0 || index($msg,'Failed') >= 0) {
return \&_fail_colourer;
}
if (index($msg,'matched') >= 0 || index($msg,'Matched') >= 0) {
return \&_match_colourer;
}
return \&_info_colourer;
}
# Set up interaction as spiffily as possible...
if (eval{ require Term::ReadKey }) {
*_interact = sub {
# No interactions when piping output to a filehandle...
return 'c' if $lexical_config->{save_to_fh};
# Otherwise grab a single key and return it...
Term::ReadKey::ReadMode('raw');
my $input = Term::ReadKey::ReadKey(0);
Term::ReadKey::ReadMode('restore');
return $input;
}
}
else {
*_interact = sub {
# No interactions when piping output to a filehandle...
return 'c' if $lexical_config->{save_to_fh};
# Otherwise return the first letter typed...
my $input = readline;
return substr($input, 0, 1);
}
}
#====[ REPL (a.k.a. rxrx) ]=======================
# Deal with v5.16 weirdness...
BEGIN {
if ($] >= 5.016) {
require feature;
feature->import('evalbytes');
*evaluate = \&CORE::evalbytes;
}
else {
*evaluate = sub{ eval shift };
}
}
my $FROM_START = 0;
sub rxrx {
# Handle: rxrx <filename>
if (@_) {
local @ARGV = @_;
# If file is a debugger dump, decode and step through it...
my $filetext = do { local $/; <> };
my $dumped_data = eval { $JSON_decoder->($filetext) };
if (ref($dumped_data) eq 'HASH' && defined $dumped_data->{regex_ID} ) {
# Reconstruct internal state...
my $regex_ID = $dumped_data->{regex_ID};
%history_of = %{ $dumped_data->{visualization} };
$history_of{match_heatmap} = $dumped_data->{match_heatmap};
$history_of{string_heatmap} = $dumped_data->{string_heatmap};
$display_mode = $dumped_data->{config}{display_mode};
$state{$regex_ID}{location} = $dumped_data->{regex_location};
# Display...
my $step = $FROM_START;
my $cmd;
while (1) {
($cmd, $step) = _revisualize($regex_ID, $step);
last if $cmd =~ /[q]/i;
$step = min($step, @{$history_of{visual}}-1);
}
exit;
}
# Otherwise, assume it's a perl source file and debug it...
else {
exec $^X, '-MRegexp::Debugger', @_
or die "Couldn't invoke perl: $!";
}
}
# Otherwise, be interactive...
# Track input history...
my $str_history = [];
my $regex_history = [];
# Start with empty data...
my $input_regex = '';
my $regex = '';
my $regex_flags = '';
my $string = '';
# And display it...
_display($string, $input_regex,q{});
INPUT:
while (1) {
my $input = _prompt('>');
# String history mode?
if ($input =~ /^['"]$/) {
$input = _rxrx_history($str_history);
}
# Regex history mode?
elsif ($input eq '/') {
$input = _rxrx_history($regex_history);
}
# Are we updating the regex or string???
if ($input =~ m{^ (?<cmd> [+]\s*[/]|[/"']) (?<data> .*?) (?<endcmd> \k<cmd> (?<flags> [imsxlaud]*) )? \s* \z }x) {
my ($cmd, $data, $endcmd, $flags) = @+{qw< cmd data endcmd flags >};
# Load the rest of the regex (if any)...
if ($cmd =~ m{[+]\s*[/]}xms) {
$cmd = '/';
while (my $input = _prompt(' +')) {
last if $input eq q{};
if ($input =~ m{\A (?<data>.*) [/][imsxlaud]*\Z}xms) {
$data .= "\n$+{data}";
last;
}
else {
$data .= "\n$input";
}
}
}
# Compile and save the new regex...
if ($cmd eq q{/}) {
if ($data eq q{}) {
state $NULL_REGEX = eval q{use Regexp::Debugger; qr{(?#NULL)}; };
$regex = $NULL_REGEX;
}
else {
$input_regex = $data;
$regex_flags = $flags // 'x';
$regex = evaluate qq{\n# line 0 rxrx\nuse Regexp::Debugger; qr/$data/$regex_flags;};
}
# Report any errors...
if (!defined $regex) {
say '>', eval qq{\n# line 0 rxrx\n qr/$data/$regex_flags;};
$input_regex = "Invalid regex:\n$@";
}
else { # Remember it...
push @{$regex_history}, $input;
}
}
# Otherwise compile the string (interpolated or not)...
elsif ($+{cmd} eq q{"}) {
$string = evaluate qq{"$+{data}"};
# Report any errors...
print "$@\n" if $@;
print "Invalid input\n" if !defined $string;
# Remember it...
push @{$str_history}, $input;
}
elsif ($+{cmd} eq q{'}) {
$string = evaluate qq{'$+{data}'};
# Report any errors...
print "$@\n" if $@;
print "Invalid input\n" if !defined $string;
# Remember it...
push @{$str_history}, $input;
}
}
# Quit if quitting requested...
elsif ($input =~ /^ \s* [xXqQ]/x) {
say q{};
last INPUT;
}
# Help...
elsif ($input =~ /^ \s* [?hH]/x) {
print "\n" x 2;
say '____________________________________________/ Help \____';
say ' ';
say ' / : Enter a pattern in a single line';
say ' +/ : Enter first line of a multi-line pattern';
say " ' : Enter a new literal string";
say ' " : Enter a new double-quoted string';
if (eval { require IO::Prompter }) {
say '';
say 'CTRL-R : History completion - move backwards one input';
say 'CTRL-N : History completion - move forwards one input';
say '';
say 'CTRL-B : Cursor motion - move back one character';
say 'CTRL-F : Cursor motion - move forwards one character';
say 'CTRL-A : Cursor motion - move to start of input';
say 'CTRL-E : Cursor motion - move to end of input';
}
say '';
say ' m : Match current string against current pattern';
say '';
say ' d : Deconstruct and explain the current regex';
say '';
say 'q or x : quit debugger and exit';
next INPUT;
}
# Visualize the match...
elsif ($input =~ /m/i) {
$string =~ $regex;
}
# Explain the regex...
elsif ($input =~ /d/i) {
_show_regex_description($next_regex_ID-1);
}
# Redisplay the new regex and/or string...
if (defined $string && defined $input_regex) {
_display($string, $input_regex, $regex_flags);
}
}
}
# Lay out the regex and string as does Regexp::Debugger...
sub _display {
my ($string, $regex, $flags) = @_;
say "\n" x 100;
say Term::ANSIColor::colored('regex:', 'white');
say qq{/$regex/$flags\n\n\n};
say Term::ANSIColor::colored('string:', 'white');
say q{'} . _ws_colourer($string) . qq{'\n\n\n};
}
# Make whitespace characters visible (without using a regex)...
sub _quote_ws {
my $str = shift;
my $index;
for my $ws_char ( ["\n"=>'\n'], ["\t"=>'\n'] ) {
SEARCH:
while (1) {
$index = index($str, $ws_char->[0]);
last SEARCH if $index < 0;
substr($str, $index, 1, $ws_char->[1]);
}
}
return $str;
}
# Hi-res sleep...
sub _pause {
select undef, undef, undef, shift;
}
# Simple prompter...
*_prompt = eval { require IO::Prompter }
? sub {
return IO::Prompter::prompt(@_)
}
: sub {
my ($prompt) = @_;
print "$prompt ";
my $input = readline *STDIN;
chomp $input;
return $input;
};
1; # Magic true value required at end of module
__END__
=head1 NAME
Regexp::Debugger - Visually debug regexes in-place
=head1 VERSION
This document describes Regexp::Debugger version 0.001020
=head1 SYNOPSIS
use Regexp::Debugger;
=head1 DESCRIPTION
When you load this module, any regex in the same lexical scope will be visually
(and interactively) debugged as it matches.
=head1 INTERFACE
The module itself provides no API.
You load it and the debugger is automatically
activated in that lexical scope.
The debugger offers the following commands:
=over
=item C<?>
: Print a help message listing these commands
=item C<s>
: Step forwards
=item C<->
: Step backwards
=item C<m>
: Continue forward to the next event that matches something
=item C<f>
: Continue forward to the next event that fails to match something
=item C<c>
: Continue forward until the regex matches or completely backtracks
=item C<< <RETURN>/<ENTER> >>
: Repeat the previous command
=item C<v>
: Switch to visualization mode
=item C<h>
: Switch to heatmapped visualization log
=item C<e>
: Switch to the event log
=item C<j>
: Switch to the underlying JSON data
=item C<d>
: Describe each component of the regex in detail
=item C<V>
=item C<H>
=item C<E>
=item C<J>
: Take a snapshot of the corresponding display mode.
When prompted for a filename:
=over
=item C<< <RET> >>
...prints the snapshot to the terminal
=item C<< <TAB> >>
...prints the snapshot to a file named "./rxrx_I<DISPLAY_MODE>_I<TIMESTAMP>"
=item Anything else
...prints the snapshot to that file
=back
=item C<q>
: Quit the debugger and finish matching this regex without any further
visualization. The program continues to execute and other regexes may
still be debugged.
=item C<x>
: Exit the debugger and the entire program immediately.
=back
=head1 CONFIGURATION
You can configure the debugger by setting up a F<.rxrx> file in
in the current directory or in your home directory. This configuration
consists of I<key>:I<value> pairs
(everything else in the file is silently ignored).
=head2 Display mode configuration
If the C<C<'display'>> key is specified, the debugger starts in that
mode. The four available modes are:
# Show dynamic visualization of matching (the default)...
display : visual
# Show dynamic heatmap visualization of matching...
display : heatmap
# Show multi-line matching event log...
display : events
# Show JSON encoding of matching process...
display : JSON
=head2 Whitespace display configuration
Normally, the debugger compacts whitespaces in the regex down to a
single space character, but you can configure that with the
C<show_ws> key:
# Compact whitespace and comments to a single space (the default)...
show_ws : compact
# Compact whitespace, but show comments, newlines (\n), and tabs (\t)...
show_ws : visible
# Don't compact whitespace, and show newlines and tabs as \n and \t...
show_ws : original
=head2 Colour configuration
The following keys reconfigure the colours with which the debugger
displays various information:
=head3 Colours for debugging information
=over
=item *
C<try_col>
The colour in which attempts to match part of the regex are reported
=item *
C<match_col>
The colour in which successful matches of part of the regex are reported
=item *
C<fail_col>
The colour in which unsuccessful matches of part of the regex are reported
=item *
C<ws_col>
The colour in which special characters (such as "\n", "\t", "\e", etc.)
are reported (as single letters: 'n', 't', 'e', etc.)
=item *
C<info_col>
The colour in which other information is reported
=back
=head3 Colours for regex descriptions
=over
=item *
C<desc_regex_col>
The colour in which components of the regex are displayed
=item *
C<desc_text_col>
The colour in which descriptions of regex components are displayed
=item *
C<desc_sep_col>
The colour in which separators between component descriptions are displayed.
=back
=head3 Colours for heatmaps
Any key that starts with C<heatmap>... is treated as a specifier for an
equal part of the total range of each heatmap.
These names are sorted (numerically, if possible; otherwise
alphabetically) and the corresponding values are then used to display
equal percentiles from the heatmap.
For example (using numeric sorting):
heatmap_0_colour : cyan on_black # 0-33rd percentile
heatmap_50_colour : yellow on_black # 34-66th percentile
heatmap_100_colour : red on_black # 67-100th percentile
Or, equivalently (using alphabetic sorting):
heatmap_infrequent : cyan on_black # 0-33rd percentile
heatmap_more_frequent : yellow on_black # 34-66th percentile
heatmap_very_frequent : red on_black # 67-100th percentile
=head3 Colour specifications
The colour values that may be used in any of the above colour
specifications are any combination of the following (i.e. the
colour specifiers supported by the Term::ANSIColor module):
clear reset bold dark
faint underline underscore blink
reverse concealed
black red green yellow
blue magenta cyan white
bright_black bright_red bright_green bright_yellow
bright_blue bright_magenta bright_cyan bright_white
on_black on_red on_green on_yellow
on_blue on_magenta on_cyan on_white
on_bright_black on_bright_red on_bright_green on_bright_yellow
on_bright_blue on_bright_magenta on_bright_cyan on_bright_white
The default colour configurations are:
try_col : bold magenta on_black
match_col : bold cyan on_black
fail_col : yellow on_red
ws_col : bold blue underline
info_col : white on_black
desc_regex_col : white on_black
desc_text_col : cyan on_black
desc_sep_col : blue on_black underline
heatmap__20th_percentile : white on_black
heatmap__40th_percentile : cyan on_blue
heatmap__60th_percentile : blue on_cyan
heatmap__80th_percentile : red on_yellow
heatmap_100th_percentile : yellow on_red
=head2 Output configuration
Normally Regexp::Debugger sends its visualizations to the terminal
and expects input from the same device.
However, you can configure the module to output its information
(in standard JSON format) to a nominated file instead, using the
C<'save_to'> option:
save_to : filename_to_save_data_to.json
Data saved in this way may be re-animated using the C<rxrx> utility,
or by calling C<Regexp::Debugger::rxrx()> directly. (See: L<"Command-line
debugging"> for details).
=head2 Configuration API
You can also configure the debugger on a program-by-program basis, by
passing any of the above key/value pairs when the module is loaded.
For example:
use Regexp::Debugger fail => 'bold red', whitespace => 'compact';
Note that any configuration specified in the user's F<.rxrx> file
is overridden by an explicit specification of this type.
The commonest use of this mechanism is to dump regex debugging
information from an non-interactive program:
use Regexp::Debugger save_to => 'regex_debugged.json';
Note that, when C<'save_to'> is specified within a program, the value
supplied does not have to be a string specifying the filename. You can
also provide an actual filehandle (or equivalent). For example:
use Regexp::Debugger save_to => IO::Socket::INET->new(
Proto => "tcp",
PeerAddr => 'localhost:666',
);
=head1 COMMAND-LINE DEBUGGING
The module provides a non-exported subroutine (C<rxrx()>) that
implements a useful command-line regex debugging utility.
The utility can be invoked with:
perl -MRegexp::Debugger -E 'Regexp::Debugger::rxrx\(@ARGV\)'
which is usually aliased in the shell to C<rxrx> (and will be referred
to by that name hereafter).
=head2 Regex debugging REPL
When called without any arguments, C<rxrx> initiates a simple REPL
that allows the user to type in regexes and strings and debug matches
between them:
=over
=item *
Any line starting with a C</> is treated as a new regex to match with.
The closing C</> may be omitted. If the closing C</> is supplied, any
one or more of the following flags may be specified immediately after
it: C<x>, C<i>, C<m>, C<s>, C<a>, C<u>, C<d>, C<l>.
=item *
Any line starting with a C<+/> is treated as the first line of a new multi-
line regex to match with. Subsequent lines are added to the regex until
the closing C</> is encountered. Any one or more of the following flags
may be specified immediately after the closing C</>: C<x>, C<i>, C<m>,
C<s>, C<a>, C<u>, C<d>, C<l>.
=item *
Any line starting with a C<'> or C<"> is treated as a new string to match
against. The corresponding closing delimiter may be omitted.
=item *
Any line beginning with C<m> causes the REPL to match the current regex
against the current string, visualizing the match in the usual way.
=item *
Any line beginning with C<d> causes the REPL to display a detailed
decomposition and explanation of the current regex.
=item *
Any line beginning with C<q> or C<x> causes the REPL to quit and exit.
=item *
Any line beginning with C<?> invokes the help listing for the REPL.
=back
If the IO::Prompter module (version 0.004 or later) is available, the
input process remembers its history, which you can recall by typing
C<CTRL-R>. Repeated C<CTRL-R>'s step successively backwards through earlier
inputs. C<CTRL-N> steps successfully forward again.
You can then use C<CTRL-B>/C<CTRL-F>/C<CTRL-A>/C<CTRL-E> to move the
cursor around the line of recalled input, to delete or insert
characters. This is useful for modifying and retrying a recently entered
regex or string.
=head2 Debugging regexes from a dumped session
When called with a filename, C<rxrx> first checks whether the file
contains a JSON dump of a previous debugging, in which case it replays
the visualization of that regex match interactively.
This is useful for debugging non-interactive programs where the
C<'save_to'> option was used (see L<"Output configuration"> and
L<"Configuration API">).
In this mode, all the features of the interactive debugger (as listed
under L<"INTERFACE">) are fully available: you can step forwards and
backwards through the match, skip to the successful submatch or a
breakpoint, swap visualization modes, and take snapshots.
=head2 Wrap-around regex debugging
When called with the name of a file that does I<not> contain a JSON
dump, C<rxrx> attempts to execute the file as a Perl program, with
Regexp::Debugger enabled at the top level. In other words:
rxrx prog.pl
is a convenient shorthand for:
perl -MRegexp::Debugger prog.pl
=head1 LIMITATIONS
=head2 C</x>-mode comments
Due to limitations in the Perl C<overload::constant()> mechanism, the
current implementation cannot always distinguish whether a regex has an
external /x modifier (and hence, what whitespace and comment characters
mean). Whitespace is handled correctly in almost all cases, but
comments are not.
When processing a C<# comment to end of line> within a regex, the module
currently assumes a C</x> is in effect at start of the regex. This will cause
erroneous behaviour if an unescaped C<#> is used in a non-C</x> regex.
Note that this limitation is likely to be corrected in a future release.
This limitation does not affect the handling of comments in
C<(?x:...)> and C<(?-x:...)> blocks within the regex. These are
always correctly handled, so explicitly using one of these blocks
is a reliable workaround...as is always using the C</x> modifier
on every debugged regex.
As regards whitespace, the one case where the current implementation
does not always correctly infer behaviour is where whitespace is used to
separate a repetition qualifier from the atom it qualifies in a non-C</x>
regex, such as:
/ x + /
Because the module defaults to assuming that regexes always have C</x> applied,
this is always interpreted as:
/\ x+\ /x
rather than what it really is, namely:
/\ x\ +\ /
The most reliable workaround for the time being is either to always use
C</x> on any regex, or never to separate repetition qualifiers from
their preceding atoms.
=head2 Multiple 'save_to' with the same target
At present, making the same file the target of two successive C<save_to> requests
causes the second JSON data structure to overwrite the first.
This limitation will be removed in a subsequent release (but this will
certainly involve a small change to the structure of the JSON data that
is written, even when only one C<save_to> is specified).
=head2 Variable interpolations
The module handles the interpolation of strings correctly,
expanding them in-place before debugging begins.
However, it currently does not correctly handle the interpolation
of C<qr>'d regexes. That is, this:
use Regexp::Debugger;
my $ident = qr{ [^\W\d]\w* }x; # a qr'd regex...
$str =~ m{ ($ident) : (.*) }xms; # ...interpolated into another regex
does not work correctly...and usually will not even compile.
It is expected that this limitation will be removed in a future
release, however it may only be possible to fix the problem for more
recent versions of Perl (i.e. 5.14 and later) in which the regex engine
is re-entrant.
=head1 DIAGNOSTICS
=over
=item C<< Odd number of configuration args after "use Regexp::Debugger" >>
The module expects configuration arguments (see L<"Configuration API">)
to be passed as C<< key => value >> pairs. You passed something else.
=item C<< Unknown 'show_ws' option: %s >>
The only valid options for the C<'show_ws'> configuration option are
C<'compact'>, C<'visible'>, or C<'original'>.
You specified something else (or misspelled one of the above).
=item C<< Unknown 'display' option: %s >>
The only valid options for the C<'display'> configuration option are
C<'visual'> or C<'heatmap'> or C<'events'> or C<'JSON'>.
You specified something else (or misspelled one of the above).
=item C<< Invalid 'save_to' option: %s (%s) >>
The value associated with the C<'save_to'> option is expected
to be a filehandle opened for writing, or else a string containing
the name of a file that can be opened for writing. You either passed
an unopened filehandle, an unwritable filename, or something that
wasn't a plausible file. Alternatively, if you passed a filepath,
was the directory not accessible to, or writeable by, you?
=back
=head1 DEPENDENCIES
This module only works with Perl 5.10.1 and later.
The following modules are used when available:
=over
=item Term::ANSIColor
Text colouring only works if this module can be loaded.
Otherwise, all output will be monochromatic.
=item Win32::Console::ANSI
Under Windows, text colouring also requires that this module can be loaded.
Otherwise, all output will be monochromatic.
=item File::HomeDir
If it can't find a useful value for C<$ENV{HOME}>, Regexp::Debugger
attempts to use this module to determine the user's home directory,
in order to search for a F<.rxrx> config file.
=item JSON::XS
=item JSON
=item JSON::DWIW
=item JSON::Syck
JSON output (i.e. for the C<'save_to'> option) is only
possible if one of these modules can be loaded.
Otherwise, all JSON output will default to an empty C<{}>.
=item Term::ReadKey
Single-character interactions only work if this module can be loaded.
Otherwise, all command interactions will require a C<< <RETURN> >>
after them.
=item Time::HiRes
Autogenerated timestamps (e.g. for snapshots) will only be sub-second
accurate if this module can be loaded. Otherwise, all timestamps will
only be to the nearest second.
=back
=head1 INCOMPATIBILITIES
None reported, but this module will almost certainly not play nicely
with any other that modifies regexes using C<overload::constant>.
=head1 BUGS AND LIMITATIONS
No bugs have been reported.
Please report any bugs or feature requests to
C<bug-regexp-debugger@rt.cpan.org>, or through the web interface at
L<http://rt.cpan.org>.
=head1 AUTHOR
Damian Conway C<< <DCONWAY@CPAN.org> >>
=head1 LICENCE AND COPYRIGHT
Copyright (c) 2011-2012, Damian Conway C<< <DCONWAY@CPAN.org> >>. All rights reserved.
This module is free software; you can redistribute it and/or
modify it under the same terms as Perl itself. See L<perlartistic>.
=head1 DISCLAIMER OF WARRANTY
BECAUSE THIS SOFTWARE IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
FOR THE SOFTWARE, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
PROVIDE THE SOFTWARE "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER
EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE
ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE SOFTWARE IS WITH
YOU. SHOULD THE SOFTWARE PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL
NECESSARY SERVICING, REPAIR, OR CORRECTION.
IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
REDISTRIBUTE THE SOFTWARE AS PERMITTED BY THE ABOVE LICENCE, BE
LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL,
OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE
THE SOFTWARE (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING
RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A
FAILURE OF THE SOFTWARE TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF
SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
SUCH DAMAGES.
|