Spaces:
Running
Running
File size: 158,468 Bytes
7b9da4a 805099a e2366f7 805099a e2366f7 805099a e2366f7 805099a e2366f7 805099a e2366f7 7b9da4a e2366f7 805099a e2366f7 805099a a01c107 e2366f7 7b9da4a e2366f7 7b9da4a 83730d1 35c0681 0c961d6 e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a 616dd44 7b9da4a d6b96dc 7b9da4a d6b96dc 7b9da4a e2366f7 0c961d6 d6b96dc 0c961d6 e2366f7 7b9da4a e2366f7 7b9da4a b927143 7b9da4a b927143 7b9da4a b927143 7b9da4a b927143 7b9da4a b927143 7b9da4a e2366f7 7b9da4a b927143 7b9da4a e2366f7 7b9da4a 0c961d6 7b9da4a 805099a 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a 0c961d6 7b9da4a e2366f7 e3cd24c 7b9da4a 0c961d6 7b9da4a 805099a 7b9da4a c8eaa51 7b9da4a 805099a e3cd24c 7b9da4a 805099a e3cd24c 7b9da4a e2366f7 7b9da4a 805099a 7b9da4a 805099a 7b9da4a 805099a 7b9da4a 805099a 7b9da4a 805099a 7b9da4a 805099a 7b9da4a 805099a 0c961d6 e2366f7 e3cd24c e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a 0c961d6 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a 805099a 7b9da4a 805099a e2366f7 7b9da4a e3cd24c e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 012529d e2366f7 012529d 267a582 e2366f7 012529d 267a582 e2366f7 267a582 35c0681 e2366f7 267a582 e2366f7 35c0681 e2366f7 7b9da4a e2366f7 35c0681 7b9da4a 805099a 7b9da4a e2366f7 35c0681 7b9da4a 805099a 35c0681 7b9da4a 805099a 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a c8eaa51 7b9da4a 805099a 7b9da4a 0c961d6 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a e3cd24c 7b9da4a e2366f7 7b9da4a e2366f7 0c961d6 e3cd24c 7b9da4a e2366f7 0c961d6 7b9da4a 805099a 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a 805099a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a e3cd24c 7b9da4a 805099a e2366f7 805099a 7b9da4a e2366f7 7b9da4a e3cd24c 0c961d6 7b9da4a e2366f7 7b9da4a b927143 7b9da4a b927143 7b9da4a e2366f7 7b9da4a 805099a e2366f7 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a d6b96dc 7b9da4a e2366f7 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a 805099a 7b9da4a 805099a e2366f7 7b9da4a 805099a 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a 805099a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 805099a 7b9da4a d6b96dc 7b9da4a e2366f7 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a e2366f7 805099a 7b9da4a d6b96dc 7b9da4a e2366f7 805099a 7b9da4a 805099a 7b9da4a 805099a 7b9da4a d6b96dc 7b9da4a 805099a 7b9da4a 0c961d6 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a 805099a 7b9da4a d6b96dc 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e3cd24c e2366f7 805099a 7b9da4a 805099a 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a 805099a e2366f7 7b9da4a e2366f7 7b9da4a e3cd24c 7b9da4a e3cd24c 7b9da4a a01c107 b927143 a01c107 b927143 e3cd24c b927143 a01c107 e3cd24c b927143 a01c107 b927143 e3cd24c 7b9da4a e3cd24c 7b9da4a 805099a 7b9da4a e3cd24c b927143 7b9da4a a01c107 7b9da4a a01c107 7b9da4a e3cd24c e2366f7 e3cd24c 7b9da4a 6881cac 012529d 805099a 7b9da4a e2366f7 35c0681 e2366f7 0c961d6 a01c107 35c0681 e2366f7 35c0681 e2366f7 35c0681 e2366f7 35c0681 e2366f7 35c0681 e2366f7 35c0681 e2366f7 a01c107 e2366f7 35c0681 e2366f7 35c0681 e2366f7 35c0681 e2366f7 35c0681 e2366f7 35c0681 e2366f7 e3cd24c 35c0681 0c961d6 e2366f7 2db874e e2366f7 0c961d6 e2366f7 0c961d6 a01c107 e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a d6b96dc e2366f7 7b9da4a 805099a e3cd24c 7b9da4a e2366f7 7b9da4a e3cd24c 0c961d6 e2366f7 0c961d6 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a e2366f7 a01c107 e2366f7 a01c107 e2366f7 7b9da4a e2366f7 7b9da4a 805099a e2366f7 e3cd24c 805099a a01c107 7b9da4a 805099a a01c107 7b9da4a 805099a e2366f7 7b9da4a 805099a 35c0681 7b9da4a 805099a 35c0681 7b9da4a 805099a 35c0681 7b9da4a 805099a 35c0681 805099a 7b9da4a 805099a 35c0681 7b9da4a 805099a 35c0681 805099a 7b9da4a 805099a 35c0681 7b9da4a 805099a 35c0681 805099a 7b9da4a 805099a 35c0681 7b9da4a 805099a e2366f7 7b9da4a e2366f7 7b9da4a e3cd24c e2366f7 7b9da4a e3cd24c 35c0681 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a 805099a 7b9da4a e2366f7 7b9da4a 805099a 7b9da4a c8eaa51 e2366f7 7b9da4a e3abb9a e2366f7 c8eaa51 e2366f7 38ed81a e2366f7 adada7e e2366f7 adada7e e2366f7 adada7e e2366f7 7b9da4a 8e7296e e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 805099a 7b9da4a e2366f7 7b9da4a e2366f7 7b9da4a e2366f7 |
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 |
#!/usr/bin/env python3
import argparse
import asyncio
import atexit
import configparser
import hashlib
import json
import logging
import os
import platform
import re
import shutil
import signal
import sqlite3
import subprocess
import sys
import time
from multiprocessing import process
from typing import List, Tuple, Optional, Dict, Callable
import zipfile
from datetime import datetime
from typing import List, Tuple
from typing import Optional
import webbrowser
from bs4 import BeautifulSoup
import gradio as gr
from huggingface_hub import InferenceClient
from playwright.async_api import async_playwright
import requests
from requests.exceptions import RequestException
from SQLite_DB import *
import tiktoken
import trafilatura
import unicodedata
import yt_dlp
# OpenAI Tokenizer support
from openai import OpenAI
from tqdm import tqdm
import tiktoken
#######################
log_level = "DEBUG"
logging.basicConfig(level=getattr(logging, log_level), format='%(asctime)s - %(levelname)s - %(message)s')
os.environ["GRADIO_ANALYTICS_ENABLED"] = "False"
#######
# Function Sections
#
# Database Setup
# Config Loading
# System Checks
# DataBase Functions
# Processing Paths and local file handling
# Video Download/Handling
# Audio Transcription
# Diarization
# Chunking-related Techniques & Functions
# Tokenization-related Techniques & Functions
# Summarizers
# Gradio UI
# Main
#
#######
# To Do
# Offline diarization - https://github.com/pyannote/pyannote-audio/blob/develop/tutorials/community/offline_usage_speaker_diarization.ipynb
####
#
# TL/DW: Too Long Didn't Watch
#
# Project originally created by https://github.com/the-crypt-keeper
# Modifications made by https://github.com/rmusser01
# All credit to the original authors, I've just glued shit together.
#
#
# Usage:
#
# Download Audio only from URL -> Transcribe audio:
# python summarize.py https://www.youtube.com/watch?v=4nd1CDZP21s`
#
# Download Audio+Video from URL -> Transcribe audio from Video:**
# python summarize.py -v https://www.youtube.com/watch?v=4nd1CDZP21s`
#
# Download Audio only from URL -> Transcribe audio -> Summarize using (`anthropic`/`cohere`/`openai`/`llama` (llama.cpp)/`ooba` (oobabooga/text-gen-webui)/`kobold` (kobold.cpp)/`tabby` (Tabbyapi)) API:**
# python summarize.py -v https://www.youtube.com/watch?v=4nd1CDZP21s -api <your choice of API>` - Make sure to put your API key into `config.txt` under the appropriate API variable
#
# Download Audio+Video from a list of videos in a text file (can be file paths or URLs) and have them all summarized:**
# python summarize.py ./local/file_on_your/system --api_name <API_name>`
#
# Run it as a WebApp**
# python summarize.py -gui` - This requires you to either stuff your API keys into the `config.txt` file, or pass them into the app every time you want to use it.
# Can be helpful for setting up a shared instance, but not wanting people to perform inference on your server.
#
###
#######################
# Random issues I've encountered and how I solved them:
# 1. Something about cuda nn library missing, even though cuda is installed...
# https://github.com/tensorflow/tensorflow/issues/54784 - Basically, installing zlib made it go away. idk.
#
# 2. ERROR: Could not install packages due to an OSError: [WinError 2] The system cannot find the file specified: 'C:\\Python312\\Scripts\\dateparser-download.exe' -> 'C:\\Python312\\Scripts\\dateparser-download.exe.deleteme'
# Resolved through adding --user to the pip install command
#
#
#######################
#######################
# DB Setup
# Handled by SQLite_DB.py
#######################
######################
# Global Variables
global local_llm_model, \
userOS, \
processing_choice, \
segments, \
detail_level_number, \
summary, \
audio_file, \
detail_level
process = None
#######################
# Config loading
#
# Read configuration from file
config = configparser.ConfigParser()
config.read('config.txt')
# API Keys
anthropic_api_key = config.get('API', 'anthropic_api_key', fallback=None)
logging.debug(f"Loaded Anthropic API Key: {anthropic_api_key}")
cohere_api_key = config.get('API', 'cohere_api_key', fallback=None)
logging.debug(f"Loaded cohere API Key: {cohere_api_key}")
groq_api_key = config.get('API', 'groq_api_key', fallback=None)
logging.debug(f"Loaded groq API Key: {groq_api_key}")
openai_api_key = config.get('API', 'openai_api_key', fallback=None)
logging.debug(f"Loaded openAI Face API Key: {openai_api_key}")
huggingface_api_key = config.get('API', 'huggingface_api_key', fallback=None)
logging.debug(f"Loaded HuggingFace Face API Key: {huggingface_api_key}")
# Models
anthropic_model = config.get('API', 'anthropic_model', fallback='claude-3-sonnet-20240229')
cohere_model = config.get('API', 'cohere_model', fallback='command-r-plus')
groq_model = config.get('API', 'groq_model', fallback='llama3-70b-8192')
openai_model = config.get('API', 'openai_model', fallback='gpt-4-turbo')
huggingface_model = config.get('API', 'huggingface_model', fallback='CohereForAI/c4ai-command-r-plus')
# Local-Models
kobold_api_IP = config.get('Local-API', 'kobold_api_IP', fallback='http://127.0.0.1:5000/api/v1/generate')
kobold_api_key = config.get('Local-API', 'kobold_api_key', fallback='')
llama_api_IP = config.get('Local-API', 'llama_api_IP', fallback='http://127.0.0.1:8080/v1/chat/completions')
llama_api_key = config.get('Local-API', 'llama_api_key', fallback='')
ooba_api_IP = config.get('Local-API', 'ooba_api_IP', fallback='http://127.0.0.1:5000/v1/chat/completions')
ooba_api_key = config.get('Local-API', 'ooba_api_key', fallback='')
tabby_api_IP = config.get('Local-API', 'tabby_api_IP', fallback='http://127.0.0.1:5000/api/v1/generate')
tabby_api_key = config.get('Local-API', 'tabby_api_key', fallback=None)
vllm_api_url = config.get('Local-API', 'vllm_api_IP', fallback='http://127.0.0.1:500/api/v1/chat/completions')
vllm_api_key = config.get('Local-API', 'vllm_api_key', fallback=None)
# Chunk settings for timed chunking summarization
DEFAULT_CHUNK_DURATION = config.getint('Settings', 'chunk_duration', fallback='30')
WORDS_PER_SECOND = config.getint('Settings', 'words_per_second', fallback='3')
# Retrieve output paths from the configuration file
output_path = config.get('Paths', 'output_path', fallback='results')
# Retrieve processing choice from the configuration file
processing_choice = config.get('Processing', 'processing_choice', fallback='cpu')
# Log file
# logging.basicConfig(filename='debug-runtime.log', encoding='utf-8', level=logging.DEBUG)
#
#
#######################
# Dirty hack - sue me.
os.environ['KMP_DUPLICATE_LIB_OK'] = 'True'
whisper_models = ["small", "medium", "small.en", "medium.en"]
source_languages = {
"en": "English",
"zh": "Chinese",
"de": "German",
"es": "Spanish",
"ru": "Russian",
"ko": "Korean",
"fr": "French"
}
source_language_list = [key[0] for key in source_languages.items()]
print(r"""_____ _ ________ _ _
|_ _|| | / /| _ \| | | | _
| | | | / / | | | || | | |(_)
| | | | / / | | | || |/\| |
| | | |____ / / | |/ / \ /\ / _
\_/ \_____//_/ |___/ \/ \/ (_)
_ _
| | | |
| |_ ___ ___ | | ___ _ __ __ _
| __| / _ \ / _ \ | | / _ \ | '_ \ / _` |
| |_ | (_) || (_) | | || (_) || | | || (_| | _
\__| \___/ \___/ |_| \___/ |_| |_| \__, |( )
__/ ||/
|___/
_ _ _ _ _ _ _
| |(_) | | ( )| | | | | |
__| | _ __| | _ __ |/ | |_ __ __ __ _ | |_ ___ | |__
/ _` || | / _` || '_ \ | __| \ \ /\ / / / _` || __| / __|| '_ \
| (_| || || (_| || | | | | |_ \ V V / | (_| || |_ | (__ | | | |
\__,_||_| \__,_||_| |_| \__| \_/\_/ \__,_| \__| \___||_| |_|
""")
time.sleep(1)
#######################################################################################################################
# System Checks
#
#
# Perform Platform Check
userOS = ""
def platform_check():
global userOS
if platform.system() == "Linux":
print("Linux OS detected \n Running Linux appropriate commands")
userOS = "Linux"
elif platform.system() == "Windows":
print("Windows OS detected \n Running Windows appropriate commands")
userOS = "Windows"
else:
print("Other OS detected \n Maybe try running things manually?")
exit()
# Check for NVIDIA GPU and CUDA availability
def cuda_check():
global processing_choice
try:
nvidia_smi = subprocess.check_output("nvidia-smi", shell=True).decode()
if "NVIDIA-SMI" in nvidia_smi:
print("NVIDIA GPU with CUDA is available.")
processing_choice = "cuda" # Set processing_choice to gpu if NVIDIA GPU with CUDA is available
else:
print("NVIDIA GPU with CUDA is not available.\nYou either have an AMD GPU, or you're stuck with CPU only.")
processing_choice = "cpu" # Set processing_choice to cpu if NVIDIA GPU with CUDA is not available
except subprocess.CalledProcessError:
print("NVIDIA GPU with CUDA is not available.\nYou either have an AMD GPU, or you're stuck with CPU only.")
processing_choice = "cpu" # Set processing_choice to cpu if nvidia-smi command fails
# Ask user if they would like to use either their GPU or their CPU for transcription
def decide_cpugpu():
global processing_choice
processing_input = input("Would you like to use your GPU or CPU for transcription? (1/cuda)GPU/(2/cpu)CPU): ")
if processing_choice == "cuda" and (processing_input.lower() == "cuda" or processing_input == "1"):
print("You've chosen to use the GPU.")
logging.debug("GPU is being used for processing")
processing_choice = "cuda"
elif processing_input.lower() == "cpu" or processing_input == "2":
print("You've chosen to use the CPU.")
logging.debug("CPU is being used for processing")
processing_choice = "cpu"
else:
print("Invalid choice. Please select either GPU or CPU.")
# check for existence of ffmpeg
def check_ffmpeg():
if shutil.which("ffmpeg") or (os.path.exists("Bin") and os.path.isfile(".\\Bin\\ffmpeg.exe")):
logging.debug("ffmpeg found installed on the local system, in the local PATH, or in the './Bin' folder")
pass
else:
logging.debug("ffmpeg not installed on the local system/in local PATH")
print(
"ffmpeg is not installed.\n\n You can either install it manually, or through your package manager of "
"choice.\n Windows users, builds are here: https://www.gyan.dev/ffmpeg/builds/")
if userOS == "Windows":
download_ffmpeg()
elif userOS == "Linux":
print(
"You should install ffmpeg using your platform's appropriate package manager, 'apt install ffmpeg',"
"'dnf install ffmpeg' or 'pacman', etc.")
else:
logging.debug("running an unsupported OS")
print("You're running an unspported/Un-tested OS")
exit_script = input("Let's exit the script, unless you're feeling lucky? (y/n)")
if exit_script == "y" or "yes" or "1":
exit()
# Download ffmpeg
def download_ffmpeg():
user_choice = input("Do you want to download ffmpeg? (y)Yes/(n)No: ")
if user_choice.lower() == 'yes' or 'y' or '1':
print("Downloading ffmpeg")
url = "https://www.gyan.dev/ffmpeg/builds/ffmpeg-release-essentials.zip"
response = requests.get(url)
if response.status_code == 200:
print("Saving ffmpeg zip file")
logging.debug("Saving ffmpeg zip file")
zip_path = "ffmpeg-release-essentials.zip"
with open(zip_path, 'wb') as file:
file.write(response.content)
logging.debug("Extracting the 'ffmpeg.exe' file from the zip")
print("Extracting ffmpeg.exe from zip file to '/Bin' folder")
with zipfile.ZipFile(zip_path, 'r') as zip_ref:
ffmpeg_path = "ffmpeg-7.0-essentials_build/bin/ffmpeg.exe"
logging.debug("checking if the './Bin' folder exists, creating if not")
bin_folder = "Bin"
if not os.path.exists(bin_folder):
logging.debug("Creating a folder for './Bin', it didn't previously exist")
os.makedirs(bin_folder)
logging.debug("Extracting 'ffmpeg.exe' to the './Bin' folder")
zip_ref.extract(ffmpeg_path, path=bin_folder)
logging.debug("Moving 'ffmpeg.exe' to the './Bin' folder")
src_path = os.path.join(bin_folder, ffmpeg_path)
dst_path = os.path.join(bin_folder, "ffmpeg.exe")
shutil.move(src_path, dst_path)
logging.debug("Removing ffmpeg zip file")
print("Deleting zip file (we've already extracted ffmpeg.exe, no worries)")
os.remove(zip_path)
logging.debug("ffmpeg.exe has been downloaded and extracted to the './Bin' folder.")
print("ffmpeg.exe has been successfully downloaded and extracted to the './Bin' folder.")
else:
logging.error("Failed to download the zip file.")
print("Failed to download the zip file.")
else:
logging.debug("User chose to not download ffmpeg")
print("ffmpeg will not be downloaded.")
#
#
#######################################################################################################################
########################################################################################################################
# DB Setup
#
#
# FIXME
# DB Functions
# create_tables()
# add_keyword()
# delete_keyword()
# add_keyword()
# add_media_with_keywords()
# search_db()
# format_results()
# search_and_display()
# export_to_csv()
# is_valid_url()
# is_valid_date()
#
#
########################################################################################################################
########################################################################################################################
# Processing Paths and local file handling
#
#
def read_paths_from_file(file_path):
""" Reads a file containing URLs or local file paths and returns them as a list. """
paths = [] # Initialize paths as an empty list
with open(file_path, 'r') as file:
paths = [line.strip() for line in file]
return paths
def process_path(path):
""" Decides whether the path is a URL or a local file and processes accordingly. """
if path.startswith('http'):
logging.debug("file is a URL")
# For YouTube URLs, modify to download and extract info
return get_youtube(path)
elif os.path.exists(path):
logging.debug("File is a path")
# For local files, define a function to handle them
return process_local_file(path)
else:
logging.error(f"Path does not exist: {path}")
return None
# FIXME
def process_local_file(file_path):
logging.info(f"Processing local file: {file_path}")
title = normalize_title(os.path.splitext(os.path.basename(file_path))[0])
info_dict = {'title': title}
logging.debug(f"Creating {title} directory...")
download_path = create_download_directory(title)
logging.debug(f"Converting '{title}' to an audio file (wav).")
audio_file = convert_to_wav(file_path) # Assumes input files are videos needing audio extraction
logging.debug(f"'{title}' successfully converted to an audio file (wav).")
return download_path, info_dict, audio_file
#
#
#######################################################################################################################
#######################################################################################################################
# Online Article Extraction / Handling
#
def get_page_title(url: str) -> str:
try:
response = requests.get(url)
response.raise_for_status()
soup = BeautifulSoup(response.text, 'html.parser')
title_tag = soup.find('title')
return title_tag.string.strip() if title_tag else "Untitled"
except requests.RequestException as e:
logging.error(f"Error fetching page title: {e}")
return "Untitled"
def get_article_text(url: str) -> str:
pass
def get_artice_title(article_url_arg: str) -> str:
# Use beautifulsoup to get the page title - Really should be using ytdlp for this....
article_title = get_page_title(article_url_arg)
#
#
#######################################################################################################################
#######################################################################################################################
# Video Download/Handling
#
def sanitize_filename(filename):
return re.sub(r'[<>:"/\\|?*]', '_', filename)
def get_video_info(url: str) -> dict:
ydl_opts = {
'quiet': True,
'no_warnings': True,
'skip_download': True,
}
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
try:
info_dict = ydl.extract_info(url, download=False)
return info_dict
except Exception as e:
logging.error(f"Error extracting video info: {e}")
return None
def process_url(url,
num_speakers,
whisper_model,
custom_prompt,
offset,
api_name,
api_key,
vad_filter,
download_video,
download_audio,
rolling_summarization,
detail_level,
question_box,
keywords,
chunk_summarization,
chunk_duration_input,
words_per_second_input,
):
# Validate input
if not url:
return "No URL provided.", "No URL provided.", None, None, None, None, None, None
if not is_valid_url(url):
return "Invalid URL format.", "Invalid URL format.", None, None, None, None, None, None
print("API Name received:", api_name) # Debugging line
logging.info(f"Processing URL: {url}")
video_file_path = None
try:
# Instantiate the database, db as a instance of the Database class
db = Database()
media_url = url
info_dict = get_youtube(url) # Extract video information using yt_dlp
media_title = info_dict['title'] if 'title' in info_dict else 'Untitled'
results = main(url, api_name=api_name, api_key=api_key,
num_speakers=num_speakers,
whisper_model=whisper_model,
offset=offset,
vad_filter=vad_filter,
download_video_flag=download_video,
custom_prompt=custom_prompt,
overwrite=args.overwrite,
rolling_summarization=rolling_summarization,
detail=detail_level,
keywords=keywords,
chunk_summarization=chunk_summarization,
chunk_duration=chunk_duration_input,
words_per_second=words_per_second_input,
)
if not results:
return "No URL provided.", "No URL provided.", None, None, None, None, None, None
transcription_result = results[0]
transcription_text = json.dumps(transcription_result['transcription'], indent=2)
summary_text = transcription_result.get('summary', 'Summary not available')
# Prepare file paths for transcription and summary
# Sanitize filenames
audio_file_sanitized = sanitize_filename(transcription_result['audio_file'])
json_file_path = audio_file_sanitized.replace('.wav', '.segments_pretty.json')
summary_file_path = audio_file_sanitized.replace('.wav', '_summary.txt')
logging.debug(f"Transcription result: {transcription_result}")
logging.debug(f"Audio file path: {transcription_result['audio_file']}")
# Write the transcription to the JSON File
try:
with open(json_file_path, 'w') as json_file:
json.dump(transcription_result['transcription'], json_file, indent=2)
except IOError as e:
logging.error(f"Error writing transcription to JSON file: {e}")
# Write the summary to the summary file
with open(summary_file_path, 'w') as summary_file:
summary_file.write(summary_text)
if download_video:
video_file_path = transcription_result['video_path'] if 'video_path' in transcription_result else None
# Check if files exist before returning paths
if not os.path.exists(json_file_path):
raise FileNotFoundError(f"File not found: {json_file_path}")
if not os.path.exists(summary_file_path):
raise FileNotFoundError(f"File not found: {summary_file_path}")
formatted_transcription = format_transcription(transcription_result)
# Check for chunk summarization
if chunk_summarization:
chunk_duration = chunk_duration_input if chunk_duration_input else DEFAULT_CHUNK_DURATION
words_per_second = words_per_second_input if words_per_second_input else WORDS_PER_SECOND
summary_text = summarize_chunks(api_name, api_key, transcription_result['transcription'], chunk_duration,
words_per_second)
# FIXME - This is a mess
# # Check for time-based chunking summarization
# if time_based_summarization:
# logging.info("MAIN: Time-based Summarization")
#
# # Set the json_file_path
# json_file_path = audio_file.replace('.wav', '.segments.json')
#
# # Perform time-based summarization
# summary = time_chunk_summarize(api_name, api_key, json_file_path, time_chunk_duration, custom_prompt)
#
# # Handle the summarized output
# if summary:
# transcription_result['summary'] = summary
# logging.info("MAIN: Time-based Summarization successful.")
# save_summary_to_file(summary, json_file_path)
# else:
# logging.warning("MAIN: Time-based Summarization failed.")
# Add media to the database
try:
# Ensure these variables are correctly populated
custom_prompt = args.custom_prompt if args.custom_prompt else ("\n\nabove is the transcript of a video "
"Please read through the transcript carefully. Identify the main topics that are discussed over the "
"course of the transcript. Then, summarize the key points about each main topic in a concise bullet "
"point. The bullet points should cover the key information conveyed about each topic in the video, "
"but should be much shorter than the full transcript. Please output your bullet point summary inside "
"<bulletpoints> tags.")
db = Database()
create_tables()
media_url = url
# FIXME - IDK?
video_info = get_video_info(media_url)
media_title = get_page_title(media_url)
media_type = "video"
media_content = transcription_text
keyword_list = keywords.split(',') if keywords else ["default"]
media_keywords = ', '.join(keyword_list)
media_author = "auto_generated"
media_ingestion_date = datetime.now().strftime('%Y-%m-%d')
transcription_model = whisper_model # Add the transcription model used
# Log the values before calling the function
logging.info(f"Media URL: {media_url}")
logging.info(f"Media Title: {media_title}")
logging.info(f"Media Type: {media_type}")
logging.info(f"Media Content: {media_content}")
logging.info(f"Media Keywords: {media_keywords}")
logging.info(f"Media Author: {media_author}")
logging.info(f"Ingestion Date: {media_ingestion_date}")
logging.info(f"Custom Prompt: {custom_prompt}")
logging.info(f"Summary Text: {summary_text}")
logging.info(f"Transcription Model: {transcription_model}")
# Check if any required field is empty
if not media_url or not media_title or not media_type or not media_content or not media_keywords or not custom_prompt or not summary_text:
raise InputError("Please provide all required fields.")
add_media_with_keywords(
url=media_url,
title=media_title,
media_type=media_type,
content=media_content,
keywords=media_keywords,
prompt=custom_prompt,
summary=summary_text,
transcription_model=transcription_model, # Pass the transcription model
author=media_author,
ingestion_date=media_ingestion_date
)
except Exception as e:
logging.error(f"Failed to add media to the database: {e}")
if summary_file_path and os.path.exists(summary_file_path):
return transcription_text, summary_text, json_file_path, summary_file_path, video_file_path, None # audio_file_path
else:
return transcription_text, summary_text, json_file_path, None, video_file_path, None # audio_file_path
except Exception as e:
logging.error(f"Error processing URL: {e}")
return str(e), 'Error processing the request.', None, None, None, None
def create_download_directory(title):
base_dir = "Results"
# Remove characters that are illegal in Windows filenames and normalize
safe_title = normalize_title(title)
logging.debug(f"{title} successfully normalized")
session_path = os.path.join(base_dir, safe_title)
if not os.path.exists(session_path):
os.makedirs(session_path, exist_ok=True)
logging.debug(f"Created directory for downloaded video: {session_path}")
else:
logging.debug(f"Directory already exists for downloaded video: {session_path}")
return session_path
def normalize_title(title):
# Normalize the string to 'NFKD' form and encode to 'ascii' ignoring non-ascii characters
title = unicodedata.normalize('NFKD', title).encode('ascii', 'ignore').decode('ascii')
title = title.replace('/', '_').replace('\\', '_').replace(':', '_').replace('"', '').replace('*', '').replace('?',
'').replace(
'<', '').replace('>', '').replace('|', '')
return title
def get_youtube(video_url):
ydl_opts = {
'format': 'bestaudio[ext=m4a]',
'noplaylist': False,
'quiet': True,
'extract_flat': True
}
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
logging.debug("About to extract youtube info")
info_dict = ydl.extract_info(video_url, download=False)
logging.debug("Youtube info successfully extracted")
return info_dict
def get_playlist_videos(playlist_url):
ydl_opts = {
'extract_flat': True,
'skip_download': True,
'quiet': True
}
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
info = ydl.extract_info(playlist_url, download=False)
if 'entries' in info:
video_urls = [entry['url'] for entry in info['entries']]
playlist_title = info['title']
return video_urls, playlist_title
else:
print("No videos found in the playlist.")
return [], None
def save_to_file(video_urls, filename):
with open(filename, 'w') as file:
file.write('\n'.join(video_urls))
print(f"Video URLs saved to {filename}")
def download_video(video_url, download_path, info_dict, download_video_flag):
logging.debug("About to normalize downloaded video title")
title = normalize_title(info_dict['title'])
if not download_video_flag:
file_path = os.path.join(download_path, f"{title}.m4a")
ydl_opts = {
'format': 'bestaudio[ext=m4a]',
'outtmpl': file_path,
}
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
logging.debug("yt_dlp: About to download audio with youtube-dl")
ydl.download([video_url])
logging.debug("yt_dlp: Audio successfully downloaded with youtube-dl")
return file_path
else:
video_file_path = os.path.join(download_path, f"{title}_video.mp4")
audio_file_path = os.path.join(download_path, f"{title}_audio.m4a")
ydl_opts_video = {
'format': 'bestvideo[ext=mp4]',
'outtmpl': video_file_path,
}
ydl_opts_audio = {
'format': 'bestaudio[ext=m4a]',
'outtmpl': audio_file_path,
}
with yt_dlp.YoutubeDL(ydl_opts_video) as ydl:
logging.debug("yt_dlp: About to download video with youtube-dl")
ydl.download([video_url])
logging.debug("yt_dlp: Video successfully downloaded with youtube-dl")
with yt_dlp.YoutubeDL(ydl_opts_audio) as ydl:
logging.debug("yt_dlp: About to download audio with youtube-dl")
ydl.download([video_url])
logging.debug("yt_dlp: Audio successfully downloaded with youtube-dl")
output_file_path = os.path.join(download_path, f"{title}.mp4")
if sys.platform.startswith('win'):
logging.debug("Running ffmpeg on Windows...")
ffmpeg_command = [
'.\\Bin\\ffmpeg.exe',
'-i', video_file_path,
'-i', audio_file_path,
'-c:v', 'copy',
'-c:a', 'copy',
output_file_path
]
subprocess.run(ffmpeg_command, check=True)
elif userOS == "Linux":
logging.debug("Running ffmpeg on Linux...")
ffmpeg_command = [
'ffmpeg',
'-i', video_file_path,
'-i', audio_file_path,
'-c:v', 'copy',
'-c:a', 'copy',
output_file_path
]
subprocess.run(ffmpeg_command, check=True)
else:
logging.error("ffmpeg: Unsupported operating system for video download and merging.")
raise RuntimeError("ffmpeg: Unsupported operating system for video download and merging.")
os.remove(video_file_path)
os.remove(audio_file_path)
return output_file_path
def read_paths_from_file(file_path: str) -> List[str]:
"""Read paths from a text file."""
with open(file_path, 'r') as file:
paths = file.readlines()
return [path.strip() for path in paths]
def save_summary_to_file(summary: str, file_path: str):
"""Save summary to a JSON file."""
summary_data = {'summary': summary, 'generated_at': datetime.now().isoformat()}
with open(file_path, 'w') as file:
json.dump(summary_data, file, indent=4)
def extract_text_from_segments(segments: List[Dict]) -> str:
"""Extract text from segments."""
return " ".join([segment['text'] for segment in segments])
#
#
#######################################################################################################################
#######################################################################################################################
# Audio Transcription
#
# Convert video .m4a into .wav using ffmpeg
# ffmpeg -i "example.mp4" -ar 16000 -ac 1 -c:a pcm_s16le "output.wav"
# https://www.gyan.dev/ffmpeg/builds/
#
# os.system(r'.\Bin\ffmpeg.exe -ss 00:00:00 -i "{video_file_path}" -ar 16000 -ac 1 -c:a pcm_s16le "{out_path}"')
def convert_to_wav(video_file_path, offset=0, overwrite=False):
out_path = os.path.splitext(video_file_path)[0] + ".wav"
if os.path.exists(out_path) and not overwrite:
print(f"File '{out_path}' already exists. Skipping conversion.")
logging.info(f"Skipping conversion as file already exists: {out_path}")
return out_path
print("Starting conversion process of .m4a to .WAV")
out_path = os.path.splitext(video_file_path)[0] + ".wav"
try:
if os.name == "nt":
logging.debug("ffmpeg being ran on windows")
if sys.platform.startswith('win'):
ffmpeg_cmd = ".\\Bin\\ffmpeg.exe"
logging.debug(f"ffmpeg_cmd: {ffmpeg_cmd}")
else:
ffmpeg_cmd = 'ffmpeg' # Assume 'ffmpeg' is in PATH for non-Windows systems
command = [
ffmpeg_cmd, # Assuming the working directory is correctly set where .\Bin exists
"-ss", "00:00:00", # Start at the beginning of the video
"-i", video_file_path,
"-ar", "16000", # Audio sample rate
"-ac", "1", # Number of audio channels
"-c:a", "pcm_s16le", # Audio codec
out_path
]
try:
# Redirect stdin from null device to prevent ffmpeg from waiting for input
with open(os.devnull, 'rb') as null_file:
result = subprocess.run(command, stdin=null_file, text=True, capture_output=True)
if result.returncode == 0:
logging.info("FFmpeg executed successfully")
logging.debug("FFmpeg output: %s", result.stdout)
else:
logging.error("Error in running FFmpeg")
logging.error("FFmpeg stderr: %s", result.stderr)
raise RuntimeError(f"FFmpeg error: {result.stderr}")
except Exception as e:
logging.error("Error occurred - ffmpeg doesn't like windows")
raise RuntimeError("ffmpeg failed")
elif os.name == "posix":
os.system(f'ffmpeg -ss 00:00:00 -i "{video_file_path}" -ar 16000 -ac 1 -c:a pcm_s16le "{out_path}"')
else:
raise RuntimeError("Unsupported operating system")
logging.info("Conversion to WAV completed: %s", out_path)
except subprocess.CalledProcessError as e:
logging.error("Error executing FFmpeg command: %s", str(e))
raise RuntimeError("Error converting video file to WAV")
except Exception as e:
logging.error("Unexpected error occurred: %s", str(e))
raise RuntimeError("Error converting video file to WAV")
return out_path
# Transcribe .wav into .segments.json
def speech_to_text(audio_file_path, selected_source_lang='en', whisper_model='small.en', vad_filter=False):
logging.info('speech-to-text: Loading faster_whisper model: %s', whisper_model)
from faster_whisper import WhisperModel
model = WhisperModel(whisper_model, device=f"{processing_choice}")
time_start = time.time()
if audio_file_path is None:
raise ValueError("speech-to-text: No audio file provided")
logging.info("speech-to-text: Audio file path: %s", audio_file_path)
try:
_, file_ending = os.path.splitext(audio_file_path)
out_file = audio_file_path.replace(file_ending, ".segments.json")
prettified_out_file = audio_file_path.replace(file_ending, ".segments_pretty.json")
if os.path.exists(out_file):
logging.info("speech-to-text: Segments file already exists: %s", out_file)
with open(out_file) as f:
global segments
segments = json.load(f)
return segments
logging.info('speech-to-text: Starting transcription...')
options = dict(language=selected_source_lang, beam_size=5, best_of=5, vad_filter=vad_filter)
transcribe_options = dict(task="transcribe", **options)
segments_raw, info = model.transcribe(audio_file_path, **transcribe_options)
segments = []
for segment_chunk in segments_raw:
chunk = {
"start": segment_chunk.start,
"end": segment_chunk.end,
"text": segment_chunk.text
}
logging.debug("Segment: %s", chunk)
segments.append(chunk)
logging.info("speech-to-text: Transcription completed with faster_whisper")
# Save prettified JSON
with open(prettified_out_file, 'w') as f:
json.dump(segments, f, indent=2)
# Save non-prettified JSON
with open(out_file, 'w') as f:
json.dump(segments, f)
except Exception as e:
logging.error("speech-to-text: Error transcribing audio: %s", str(e))
raise RuntimeError("speech-to-text: Error transcribing audio")
return segments
#
#
#######################################################################################################################
#######################################################################################################################
# Diarization
#
# TODO: https://huggingface.co/pyannote/speaker-diarization-3.1
# embedding_model = "pyannote/embedding", embedding_size=512
# embedding_model = "speechbrain/spkrec-ecapa-voxceleb", embedding_size=192
# def speaker_diarize(video_file_path, segments, embedding_model = "pyannote/embedding", embedding_size=512, num_speakers=0):
# """
# 1. Generating speaker embeddings for each segments.
# 2. Applying agglomerative clustering on the embeddings to identify the speaker for each segment.
# """
# try:
# from pyannote.audio import Audio
# from pyannote.core import Segment
# from pyannote.audio.pipelines.speaker_verification import PretrainedSpeakerEmbedding
# import numpy as np
# import pandas as pd
# from sklearn.cluster import AgglomerativeClustering
# from sklearn.metrics import silhouette_score
# import tqdm
# import wave
#
# embedding_model = PretrainedSpeakerEmbedding( embedding_model, device=torch.device("cuda" if torch.cuda.is_available() else "cpu"))
#
#
# _,file_ending = os.path.splitext(f'{video_file_path}')
# audio_file = video_file_path.replace(file_ending, ".wav")
# out_file = video_file_path.replace(file_ending, ".diarize.json")
#
# logging.debug("getting duration of audio file")
# with contextlib.closing(wave.open(audio_file,'r')) as f:
# frames = f.getnframes()
# rate = f.getframerate()
# duration = frames / float(rate)
# logging.debug("duration of audio file obtained")
# print(f"duration of audio file: {duration}")
#
# def segment_embedding(segment):
# logging.debug("Creating embedding")
# audio = Audio()
# start = segment["start"]
# end = segment["end"]
#
# # Enforcing a minimum segment length
# if end-start < 0.3:
# padding = 0.3-(end-start)
# start -= padding/2
# end += padding/2
# print('Padded segment because it was too short:',segment)
#
# # Whisper overshoots the end timestamp in the last segment
# end = min(duration, end)
# # clip audio and embed
# clip = Segment(start, end)
# waveform, sample_rate = audio.crop(audio_file, clip)
# return embedding_model(waveform[None])
#
# embeddings = np.zeros(shape=(len(segments), embedding_size))
# for i, segment in enumerate(tqdm.tqdm(segments)):
# embeddings[i] = segment_embedding(segment)
# embeddings = np.nan_to_num(embeddings)
# print(f'Embedding shape: {embeddings.shape}')
#
# if num_speakers == 0:
# # Find the best number of speakers
# score_num_speakers = {}
#
# for num_speakers in range(2, 10+1):
# clustering = AgglomerativeClustering(num_speakers).fit(embeddings)
# score = silhouette_score(embeddings, clustering.labels_, metric='euclidean')
# score_num_speakers[num_speakers] = score
# best_num_speaker = max(score_num_speakers, key=lambda x:score_num_speakers[x])
# print(f"The best number of speakers: {best_num_speaker} with {score_num_speakers[best_num_speaker]} score")
# else:
# best_num_speaker = num_speakers
#
# # Assign speaker label
# clustering = AgglomerativeClustering(best_num_speaker).fit(embeddings)
# labels = clustering.labels_
# for i in range(len(segments)):
# segments[i]["speaker"] = 'SPEAKER ' + str(labels[i] + 1)
#
# with open(out_file,'w') as f:
# f.write(json.dumps(segments, indent=2))
#
# # Make CSV output
# def convert_time(secs):
# return datetime.timedelta(seconds=round(secs))
#
# objects = {
# 'Start' : [],
# 'End': [],
# 'Speaker': [],
# 'Text': []
# }
# text = ''
# for (i, segment) in enumerate(segments):
# if i == 0 or segments[i - 1]["speaker"] != segment["speaker"]:
# objects['Start'].append(str(convert_time(segment["start"])))
# objects['Speaker'].append(segment["speaker"])
# if i != 0:
# objects['End'].append(str(convert_time(segments[i - 1]["end"])))
# objects['Text'].append(text)
# text = ''
# text += segment["text"] + ' '
# objects['End'].append(str(convert_time(segments[i - 1]["end"])))
# objects['Text'].append(text)
#
# save_path = video_file_path.replace(file_ending, ".csv")
# df_results = pd.DataFrame(objects)
# df_results.to_csv(save_path)
# return df_results, save_path
#
# except Exception as e:
# raise RuntimeError("Error Running inference with local model", e)
#
#
#######################################################################################################################
#######################################################################################################################
# Chunking-related Techniques & Functions
#
#
######### Words-per-second Chunking #########
def chunk_transcript(transcript: str, chunk_duration: int, words_per_second) -> List[str]:
words = transcript.split()
words_per_chunk = chunk_duration * words_per_second
chunks = [' '.join(words[i:i + words_per_chunk]) for i in range(0, len(words), words_per_chunk)]
return chunks
def summarize_chunks(api_name: str, api_key: str, transcript: List[dict], chunk_duration: int,
words_per_second: int) -> str:
if api_name not in summarizers: # See 'summarizers' dict in the main script
return f"Unsupported API: {api_name}"
summarizer = summarizers[api_name]
text = extract_text_from_segments(transcript)
chunks = chunk_transcript(text, chunk_duration, words_per_second)
summaries = []
for chunk in chunks:
if api_name == 'openai':
# Ensure the correct model and prompt are passed
summaries.append(summarizer(api_key, chunk, custom_prompt))
else:
summaries.append(summarizer(api_key, chunk))
return "\n\n".join(summaries)
################## ####################
######### Token-size Chunking ######### FIXME - OpenAI only currently
# This is dirty and shameful and terrible. It should be replaced with a proper implementation.
# anyways lets get to it....
client = OpenAI(api_key=openai_api_key)
def get_chat_completion(messages, model='gpt-4-turbo'):
response = client.chat.completions.create(
model=model,
messages=messages,
temperature=0,
)
return response.choices[0].message.content
# This function chunks a text into smaller pieces based on a maximum token count and a delimiter
def chunk_on_delimiter(input_string: str,
max_tokens: int,
delimiter: str) -> List[str]:
chunks = input_string.split(delimiter)
combined_chunks, _, dropped_chunk_count = combine_chunks_with_no_minimum(
chunks, max_tokens, chunk_delimiter=delimiter, add_ellipsis_for_overflow=True)
if dropped_chunk_count > 0:
print(f"Warning: {dropped_chunk_count} chunks were dropped due to exceeding the token limit.")
combined_chunks = [f"{chunk}{delimiter}" for chunk in combined_chunks]
return combined_chunks
# This function combines text chunks into larger blocks without exceeding a specified token count.
# It returns the combined chunks, their original indices, and the number of dropped chunks due to overflow.
def combine_chunks_with_no_minimum(
chunks: List[str],
max_tokens: int,
chunk_delimiter="\n\n",
header: Optional[str] = None,
add_ellipsis_for_overflow=False,
) -> Tuple[List[str], List[int]]:
dropped_chunk_count = 0
output = [] # list to hold the final combined chunks
output_indices = [] # list to hold the indices of the final combined chunks
candidate = (
[] if header is None else [header]
) # list to hold the current combined chunk candidate
candidate_indices = []
for chunk_i, chunk in enumerate(chunks):
chunk_with_header = [chunk] if header is None else [header, chunk]
# FIXME MAKE NOT OPENAI SPECIFIC
if len(openai_tokenize(chunk_delimiter.join(chunk_with_header))) > max_tokens:
print(f"warning: chunk overflow")
if (
add_ellipsis_for_overflow
# FIXME MAKE NOT OPENAI SPECIFIC
and len(openai_tokenize(chunk_delimiter.join(candidate + ["..."]))) <= max_tokens
):
candidate.append("...")
dropped_chunk_count += 1
continue # this case would break downstream assumptions
# estimate token count with the current chunk added
# FIXME MAKE NOT OPENAI SPECIFIC
extended_candidate_token_count = len(openai_tokenize(chunk_delimiter.join(candidate + [chunk])))
# If the token count exceeds max_tokens, add the current candidate to output and start a new candidate
if extended_candidate_token_count > max_tokens:
output.append(chunk_delimiter.join(candidate))
output_indices.append(candidate_indices)
candidate = chunk_with_header # re-initialize candidate
candidate_indices = [chunk_i]
# otherwise keep extending the candidate
else:
candidate.append(chunk)
candidate_indices.append(chunk_i)
# add the remaining candidate to output if it's not empty
if (header is not None and len(candidate) > 1) or (header is None and len(candidate) > 0):
output.append(chunk_delimiter.join(candidate))
output_indices.append(candidate_indices)
return output, output_indices, dropped_chunk_count
def rolling_summarize(text: str,
detail: float = 0,
model: str = 'gpt-4-turbo',
additional_instructions: Optional[str] = None,
minimum_chunk_size: Optional[int] = 500,
chunk_delimiter: str = ".",
summarize_recursively=False,
verbose=False):
"""
Summarizes a given text by splitting it into chunks, each of which is summarized individually.
The level of detail in the summary can be adjusted, and the process can optionally be made recursive.
Parameters: - text (str): The text to be summarized. - detail (float, optional): A value between 0 and 1
indicating the desired level of detail in the summary. 0 leads to a higher level summary, and 1 results in a more
detailed summary. Defaults to 0. - model (str, optional): The model to use for generating summaries. Defaults to
'gpt-3.5-turbo'. - additional_instructions (Optional[str], optional): Additional instructions to provide to the
model for customizing summaries. - minimum_chunk_size (Optional[int], optional): The minimum size for text
chunks. Defaults to 500. - chunk_delimiter (str, optional): The delimiter used to split the text into chunks.
Defaults to ".". - summarize_recursively (bool, optional): If True, summaries are generated recursively,
using previous summaries for context. - verbose (bool, optional): If True, prints detailed information about the
chunking process.
Returns:
- str: The final compiled summary of the text.
The function first determines the number of chunks by interpolating between a minimum and a maximum chunk count
based on the `detail` parameter. It then splits the text into chunks and summarizes each chunk. If
`summarize_recursively` is True, each summary is based on the previous summaries, adding more context to the
summarization process. The function returns a compiled summary of all chunks.
"""
# check detail is set correctly
assert 0 <= detail <= 1
# interpolate the number of chunks based to get specified level of detail
max_chunks = len(chunk_on_delimiter(text, minimum_chunk_size, chunk_delimiter))
min_chunks = 1
num_chunks = int(min_chunks + detail * (max_chunks - min_chunks))
# adjust chunk_size based on interpolated number of chunks
# FIXME MAKE NOT OPENAI SPECIFIC
document_length = len(openai_tokenize(text))
chunk_size = max(minimum_chunk_size, document_length // num_chunks)
text_chunks = chunk_on_delimiter(text, chunk_size, chunk_delimiter)
if verbose:
print(f"Splitting the text into {len(text_chunks)} chunks to be summarized.")
# FIXME MAKE NOT OPENAI SPECIFIC
print(f"Chunk lengths are {[len(openai_tokenize(x)) for x in text_chunks]}")
# set system message
system_message_content = "Rewrite this text in summarized form."
if additional_instructions is not None:
system_message_content += f"\n\n{additional_instructions}"
accumulated_summaries = []
for chunk in tqdm(text_chunks):
if summarize_recursively and accumulated_summaries:
# Creating a structured prompt for recursive summarization
accumulated_summaries_string = '\n\n'.join(accumulated_summaries)
user_message_content = f"Previous summaries:\n\n{accumulated_summaries_string}\n\nText to summarize next:\n\n{chunk}"
else:
# Directly passing the chunk for summarization without recursive context
user_message_content = chunk
# Constructing messages based on whether recursive summarization is applied
messages = [
{"role": "system", "content": system_message_content},
{"role": "user", "content": user_message_content}
]
# Assuming this function gets the completion and works as expected
response = get_chat_completion(messages, model=model)
accumulated_summaries.append(response)
# Compile final summary from partial summaries
global final_summary
final_summary = '\n\n'.join(accumulated_summaries)
return final_summary
#######################################
######### Words-per-second Chunking #########
# FIXME - WHole section needs to be re-written
def chunk_transcript(transcript: str, chunk_duration: int, words_per_second) -> List[str]:
words = transcript.split()
words_per_chunk = chunk_duration * words_per_second
chunks = [' '.join(words[i:i + words_per_chunk]) for i in range(0, len(words), words_per_chunk)]
return chunks
def summarize_chunks(api_name: str, api_key: str, transcript: List[dict], chunk_duration: int,
words_per_second: int) -> str:
if api_name not in summarizers: # See 'summarizers' dict in the main script
return f"Unsupported API: {api_name}"
if not transcript:
logging.error("Empty or None transcript provided to summarize_chunks")
return "Error: Empty or None transcript provided"
text = extract_text_from_segments(transcript)
chunks = chunk_transcript(text, chunk_duration, words_per_second)
custom_prompt = args.custom_prompt
summaries = []
for chunk in chunks:
if api_name == 'openai':
# Ensure the correct model and prompt are passed
summaries.append(summarize_with_openai(api_key, chunk, custom_prompt))
elif api_name == 'anthropic':
summaries.append(summarize_with_cohere(api_key, chunk, anthropic_model, custom_prompt))
elif api_name == 'cohere':
summaries.append(summarize_with_claude(api_key, chunk, cohere_model, custom_prompt))
elif api_name == 'groq':
summaries.append(summarize_with_groq(api_key, chunk, groq_model, custom_prompt))
elif api_name == 'llama':
summaries.append(summarize_with_llama(llama_api_IP, chunk, api_key, custom_prompt))
elif api_name == 'kobold':
summaries.append(summarize_with_kobold(kobold_api_IP, chunk, api_key, custom_prompt))
elif api_name == 'ooba':
summaries.append(summarize_with_oobabooga(ooba_api_IP, chunk, api_key, custom_prompt))
elif api_name == 'tabbyapi':
summaries.append(summarize_with_vllm(api_key, tabby_api_IP, chunk, llm_model, custom_prompt))
elif api_name == 'local-llm':
summaries.append(summarize_with_local_llm(chunk, custom_prompt))
else:
return f"Unsupported API: {api_name}"
return "\n\n".join(summaries)
#######################################
#
#
#######################################################################################################################
#######################################################################################################################
# Tokenization-related Techniques & Functions
#
#
def openai_tokenize(text: str) -> List[str]:
encoding = tiktoken.encoding_for_model('gpt-4-turbo')
return encoding.encode(text)
# openai summarize chunks
#
#
#######################################################################################################################
#######################################################################################################################
# Website-related Techniques & Functions
#
#
def scrape_article(url):
async def fetch_html(url: str) -> str:
async with async_playwright() as p:
browser = await p.chromium.launch(headless=True)
context = await browser.new_context(
user_agent="Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3")
page = await context.new_page()
await page.goto(url)
await page.wait_for_load_state("networkidle") # Wait for the network to be idle
content = await page.content()
await browser.close()
return content
def extract_article_data(html: str) -> dict:
downloaded = trafilatura.extract(html, include_comments=False, include_tables=False, include_images=False)
if downloaded:
metadata = trafilatura.extract_metadata(html)
if metadata:
return {
'title': metadata.title if metadata.title else 'N/A',
'author': metadata.author if metadata.author else 'N/A',
'content': downloaded,
'date': metadata.date if metadata.date else 'N/A',
}
else:
print("Metadata extraction failed.")
return None
else:
print("Content extraction failed.")
return None
def convert_html_to_markdown(html: str) -> str:
soup = BeautifulSoup(html, 'html.parser')
# Convert each paragraph to markdown
for para in soup.find_all('p'):
para.append('\n') # Add a newline at the end of each paragraph for markdown separation
# Use .get_text() with separator to keep paragraph separation
text = soup.get_text(separator='\n\n')
return text
async def fetch_and_extract_article(url: str):
html = await fetch_html(url)
print("HTML Content:", html[:500]) # Print first 500 characters of the HTML for inspection
article_data = extract_article_data(html)
if article_data:
article_data['content'] = convert_html_to_markdown(article_data['content'])
return article_data
else:
return None
# Using asyncio.run to handle event loop creation and execution
article_data = asyncio.run(fetch_and_extract_article(url))
return article_data
def ingest_article_to_db(url, title, author, content, keywords, summary, ingestion_date, custom_prompt):
try:
# Check if content is not empty or whitespace
if not content.strip():
raise ValueError("Content is empty.")
db = Database()
create_tables()
keyword_list = keywords.split(',') if keywords else ["default"]
keyword_str = ', '.join(keyword_list)
# Set default values for missing fields
url = url or 'Unknown'
title = title or 'Unknown'
author = author or 'Unknown'
keywords = keywords or 'default'
summary = summary or 'No summary available'
ingestion_date = ingestion_date or datetime.now().strftime('%Y-%m-%d')
# Log the values of all fields before calling add_media_with_keywords
logging.debug(f"URL: {url}")
logging.debug(f"Title: {title}")
logging.debug(f"Author: {author}")
logging.debug(f"Content: {content[:50]}... (length: {len(content)})") # Log first 50 characters of content
logging.debug(f"Keywords: {keywords}")
logging.debug(f"Summary: {summary}")
logging.debug(f"Ingestion Date: {ingestion_date}")
logging.debug(f"Custom Prompt: {custom_prompt}")
# Check if any required field is empty and log the specific missing field
if not url:
logging.error("URL is missing.")
raise ValueError("URL is missing.")
if not title:
logging.error("Title is missing.")
raise ValueError("Title is missing.")
if not content:
logging.error("Content is missing.")
raise ValueError("Content is missing.")
if not keywords:
logging.error("Keywords are missing.")
raise ValueError("Keywords are missing.")
if not summary:
logging.error("Summary is missing.")
raise ValueError("Summary is missing.")
if not ingestion_date:
logging.error("Ingestion date is missing.")
raise ValueError("Ingestion date is missing.")
if not custom_prompt:
logging.error("Custom prompt is missing.")
raise ValueError("Custom prompt is missing.")
# Add media with keywords to the database
result = add_media_with_keywords(
url=url,
title=title,
media_type='article',
content=content,
keywords=keyword_str or "article_default",
prompt=custom_prompt or None,
summary=summary or "No summary generated",
transcription_model=None, # or some default value if applicable
author=author or 'Unknown',
ingestion_date=ingestion_date
)
return result
except Exception as e:
logging.error(f"Failed to ingest article to the database: {e}")
return str(e)
def scrape_and_summarize(url, custom_prompt_arg, api_name, api_key, keywords, custom_article_title):
# Step 1: Scrape the article
article_data = scrape_article(url)
print(f"Scraped Article Data: {article_data}") # Debugging statement
if not article_data:
return "Failed to scrape the article."
# Use the custom title if provided, otherwise use the scraped title
title = custom_article_title.strip() if custom_article_title else article_data.get('title', 'Untitled')
author = article_data.get('author', 'Unknown')
content = article_data.get('content', '')
ingestion_date = datetime.now().strftime('%Y-%m-%d')
print(f"Title: {title}, Author: {author}, Content Length: {len(content)}") # Debugging statement
# Custom prompt for the article
article_custom_prompt = custom_prompt_arg or "Summarize this article."
# Step 2: Summarize the article
summary = None
if api_name:
logging.debug(f"Article_Summarizer: Summarization being performed by {api_name}")
# Sanitize filename for saving the JSON file
sanitized_title = sanitize_filename(title)
json_file_path = os.path.join("Results", f"{sanitized_title}_segments.json")
with open(json_file_path, 'w') as json_file:
json.dump([{'text': content}], json_file, indent=2)
try:
if api_name.lower() == 'openai':
openai_api_key = api_key if api_key else config.get('API', 'openai_api_key', fallback=None)
logging.debug(f"Article_Summarizer: trying to summarize with openAI")
summary = summarize_with_openai(openai_api_key, json_file_path, article_custom_prompt)
elif api_name.lower() == "anthropic":
anthropic_api_key = api_key if api_key else config.get('API', 'anthropic_api_key', fallback=None)
logging.debug(f"Article_Summarizer: Trying to summarize with anthropic")
summary = summarize_with_claude(anthropic_api_key, json_file_path, anthropic_model,
custom_prompt_arg=article_custom_prompt)
elif api_name.lower() == "cohere":
cohere_api_key = api_key if api_key else config.get('API', 'cohere_api_key', fallback=None)
logging.debug(f"Article_Summarizer: Trying to summarize with cohere")
summary = summarize_with_cohere(cohere_api_key, json_file_path, cohere_model,
custom_prompt_arg=article_custom_prompt)
elif api_name.lower() == "groq":
groq_api_key = api_key if api_key else config.get('API', 'groq_api_key', fallback=None)
logging.debug(f"Article_Summarizer: Trying to summarize with Groq")
summary = summarize_with_groq(groq_api_key, json_file_path, groq_model,
custom_prompt_arg=article_custom_prompt)
elif api_name.lower() == "llama":
llama_token = api_key if api_key else config.get('API', 'llama_api_key', fallback=None)
llama_ip = llama_api_IP
logging.debug(f"Article_Summarizer: Trying to summarize with Llama.cpp")
summary = summarize_with_llama(llama_ip, json_file_path, llama_token, article_custom_prompt)
elif api_name.lower() == "kobold":
kobold_token = api_key if api_key else config.get('API', 'kobold_api_key', fallback=None)
kobold_ip = kobold_api_IP
logging.debug(f"Article_Summarizer: Trying to summarize with kobold.cpp")
summary = summarize_with_kobold(kobold_ip, json_file_path, kobold_token, article_custom_prompt)
elif api_name.lower() == "ooba":
ooba_token = api_key if api_key else config.get('API', 'ooba_api_key', fallback=None)
ooba_ip = ooba_api_IP
logging.debug(f"Article_Summarizer: Trying to summarize with oobabooga")
summary = summarize_with_oobabooga(ooba_ip, json_file_path, ooba_token, article_custom_prompt)
elif api_name.lower() == "tabbyapi":
tabbyapi_key = api_key if api_key else config.get('API', 'tabby_api_key', fallback=None)
tabbyapi_ip = tabby_api_IP
logging.debug(f"Article_Summarizer: Trying to summarize with tabbyapi")
tabby_model = llm_model
summary = summarize_with_tabbyapi(tabbyapi_key, tabbyapi_ip, json_file_path, tabby_model,
article_custom_prompt)
elif api_name.lower() == "vllm":
logging.debug(f"Article_Summarizer: Trying to summarize with VLLM")
summary = summarize_with_vllm(vllm_api_url, vllm_api_key, llm_model, json_file_path,
article_custom_prompt)
elif api_name.lower() == "huggingface":
huggingface_api_key = api_key if api_key else config.get('API', 'huggingface_api_key', fallback=None)
logging.debug(f"Article_Summarizer: Trying to summarize with huggingface")
summary = summarize_with_huggingface(huggingface_api_key, json_file_path, article_custom_prompt)
except requests.exceptions.ConnectionError as e:
logging.error(f"Connection error while trying to summarize with {api_name}: {str(e)}")
if summary:
logging.info(f"Article_Summarizer: Summary generated using {api_name} API")
save_summary_to_file(summary, json_file_path)
else:
summary = "Summary not available"
logging.warning(f"Failed to generate summary using {api_name} API")
else:
summary = "Article Summarization: No API provided for summarization."
print(f"Summary: {summary}") # Debugging statement
# Step 3: Ingest the article into the database
ingestion_result = ingest_article_to_db(url, title, author, content, keywords, summary, ingestion_date,
article_custom_prompt)
return f"Title: {title}\nAuthor: {author}\nSummary: {summary}\nIngestion Result: {ingestion_result}"
def ingest_unstructured_text(text, custom_prompt, api_name, api_key, keywords, custom_article_title):
title = custom_article_title.strip() if custom_article_title else "Unstructured Text"
author = "Unknown"
ingestion_date = datetime.now().strftime('%Y-%m-%d')
# Summarize the unstructured text
if api_name:
json_file_path = f"Results/{title.replace(' ', '_')}_segments.json"
with open(json_file_path, 'w') as json_file:
json.dump([{'text': text}], json_file, indent=2)
if api_name.lower() == 'openai':
summary = summarize_with_openai(api_key, json_file_path, custom_prompt)
# Add other APIs as needed
else:
summary = "Unsupported API."
else:
summary = "No API provided for summarization."
# Ingest the unstructured text into the database
ingestion_result = ingest_article_to_db('Unstructured Text', title, author, text, keywords, summary, ingestion_date,
custom_prompt)
return f"Title: {title}\nSummary: {summary}\nIngestion Result: {ingestion_result}"
#
#
#######################################################################################################################
#######################################################################################################################
# Summarizers
#
#
# Fixme , function is replicated....
def extract_text_from_segments(segments):
logging.debug(f"Main: extracting text from {segments}")
text = ' '.join([segment['text'] for segment in segments])
logging.debug(f"Main: Successfully extracted text from {segments}")
return text
def summarize_with_openai(api_key, file_path, custom_prompt_arg):
try:
logging.debug("openai: Loading json data for summarization")
with open(file_path, 'r') as file:
segments = json.load(file)
open_ai_model = openai_model or 'gpt-4-turbo'
logging.debug("openai: Extracting text from the segments")
text = extract_text_from_segments(segments)
headers = {
'Authorization': f'Bearer {api_key}',
'Content-Type': 'application/json'
}
logging.debug(f"openai: API Key is: {api_key}")
logging.debug("openai: Preparing data + prompt for submittal")
openai_prompt = f"{text} \n\n\n\n{custom_prompt_arg}"
data = {
"model": open_ai_model,
"messages": [
{
"role": "system",
"content": "You are a professional summarizer."
},
{
"role": "user",
"content": openai_prompt
}
],
"max_tokens": 8192, # Adjust tokens as needed
"temperature": 0.1
}
logging.debug("openai: Posting request")
response = requests.post('https://api.openai.com/v1/chat/completions', headers=headers, json=data)
if response.status_code == 200:
response_data = response.json()
if 'choices' in response_data and len(response_data['choices']) > 0:
summary = response_data['choices'][0]['message']['content'].strip()
logging.debug("openai: Summarization successful")
print("openai: Summarization successful.")
return summary
else:
logging.warning("openai: Summary not found in the response data")
return "openai: Summary not available"
else:
logging.debug("openai: Summarization failed")
print("openai: Failed to process summary:", response.text)
return "openai: Failed to process summary"
except Exception as e:
logging.debug("openai: Error in processing: %s", str(e))
print("openai: Error occurred while processing summary with openai:", str(e))
return "openai: Error occurred while processing summary"
def summarize_with_claude(api_key, file_path, model, custom_prompt_arg, max_retries=3, retry_delay=5):
try:
logging.debug("anthropic: Loading JSON data")
with open(file_path, 'r') as file:
segments = json.load(file)
logging.debug("anthropic: Extracting text from the segments file")
text = extract_text_from_segments(segments)
headers = {
'x-api-key': api_key,
'anthropic-version': '2023-06-01',
'Content-Type': 'application/json'
}
anthropic_prompt = custom_prompt_arg # Sanitize the custom prompt
logging.debug(f"anthropic: Prompt is {anthropic_prompt}")
user_message = {
"role": "user",
"content": f"{text} \n\n\n\n{anthropic_prompt}"
}
data = {
"model": model,
"max_tokens": 4096, # max _possible_ tokens to return
"messages": [user_message],
"stop_sequences": ["\n\nHuman:"],
"temperature": 0.1,
"top_k": 0,
"top_p": 1.0,
"metadata": {
"user_id": "example_user_id",
},
"stream": False,
"system": "You are a professional summarizer."
}
for attempt in range(max_retries):
try:
logging.debug("anthropic: Posting request to API")
response = requests.post('https://api.anthropic.com/v1/messages', headers=headers, json=data)
# Check if the status code indicates success
if response.status_code == 200:
logging.debug("anthropic: Post submittal successful")
response_data = response.json()
try:
summary = response_data['content'][0]['text'].strip()
logging.debug("anthropic: Summarization successful")
print("Summary processed successfully.")
return summary
except (IndexError, KeyError) as e:
logging.debug("anthropic: Unexpected data in response")
print("Unexpected response format from Claude API:", response.text)
return None
elif response.status_code == 500: # Handle internal server error specifically
logging.debug("anthropic: Internal server error")
print("Internal server error from API. Retrying may be necessary.")
time.sleep(retry_delay)
else:
logging.debug(
f"anthropic: Failed to summarize, status code {response.status_code}: {response.text}")
print(f"Failed to process summary, status code {response.status_code}: {response.text}")
return None
except RequestException as e:
logging.error(f"anthropic: Network error during attempt {attempt + 1}/{max_retries}: {str(e)}")
if attempt < max_retries - 1:
time.sleep(retry_delay)
else:
return f"anthropic: Network error: {str(e)}"
except FileNotFoundError as e:
logging.error(f"anthropic: File not found: {file_path}")
return f"anthropic: File not found: {file_path}"
except json.JSONDecodeError as e:
logging.error(f"anthropic: Invalid JSON format in file: {file_path}")
return f"anthropic: Invalid JSON format in file: {file_path}"
except Exception as e:
logging.error(f"anthropic: Error in processing: {str(e)}")
return f"anthropic: Error occurred while processing summary with Anthropic: {str(e)}"
# Summarize with Cohere
def summarize_with_cohere(api_key, file_path, model, custom_prompt_arg):
try:
logging.debug("cohere: Loading JSON data")
with open(file_path, 'r') as file:
segments = json.load(file)
logging.debug(f"cohere: Extracting text from segments file")
text = extract_text_from_segments(segments)
headers = {
'accept': 'application/json',
'content-type': 'application/json',
'Authorization': f'Bearer {api_key}'
}
cohere_prompt = f"{text} \n\n\n\n{custom_prompt_arg}"
logging.debug("cohere: Prompt being sent is {cohere_prompt}")
data = {
"chat_history": [
{"role": "USER", "message": cohere_prompt}
],
"message": "Please provide a summary.",
"model": model,
"connectors": [{"id": "web-search"}]
}
logging.debug("cohere: Submitting request to API endpoint")
print("cohere: Submitting request to API endpoint")
response = requests.post('https://api.cohere.ai/v1/chat', headers=headers, json=data)
response_data = response.json()
logging.debug("API Response Data: %s", response_data)
if response.status_code == 200:
if 'text' in response_data:
summary = response_data['text'].strip()
logging.debug("cohere: Summarization successful")
print("Summary processed successfully.")
return summary
else:
logging.error("Expected data not found in API response.")
return "Expected data not found in API response."
else:
logging.error(f"cohere: API request failed with status code {response.status_code}: {response.text}")
print(f"Failed to process summary, status code {response.status_code}: {response.text}")
return f"cohere: API request failed: {response.text}"
except Exception as e:
logging.error("cohere: Error in processing: %s", str(e))
return f"cohere: Error occurred while processing summary with Cohere: {str(e)}"
# https://console.groq.com/docs/quickstart
def summarize_with_groq(api_key, file_path, model, custom_prompt_arg):
try:
logging.debug("groq: Loading JSON data")
with open(file_path, 'r') as file:
segments = json.load(file)
logging.debug(f"groq: Extracting text from segments file")
text = extract_text_from_segments(segments)
headers = {
'Authorization': f'Bearer {api_key}',
'Content-Type': 'application/json'
}
groq_prompt = f"{text} \n\n\n\n{custom_prompt_arg}"
logging.debug("groq: Prompt being sent is {groq_prompt}")
data = {
"messages": [
{
"role": "user",
"content": groq_prompt
}
],
"model": model
}
logging.debug("groq: Submitting request to API endpoint")
print("groq: Submitting request to API endpoint")
response = requests.post('https://api.groq.com/openai/v1/chat/completions', headers=headers, json=data)
response_data = response.json()
logging.debug("API Response Data: %s", response_data)
if response.status_code == 200:
if 'choices' in response_data and len(response_data['choices']) > 0:
summary = response_data['choices'][0]['message']['content'].strip()
logging.debug("groq: Summarization successful")
print("Summarization successful.")
return summary
else:
logging.error("Expected data not found in API response.")
return "Expected data not found in API response."
else:
logging.error(f"groq: API request failed with status code {response.status_code}: {response.text}")
return f"groq: API request failed: {response.text}"
except Exception as e:
logging.error("groq: Error in processing: %s", str(e))
return f"groq: Error occurred while processing summary with groq: {str(e)}"
#################################
#
# Local Summarization
def summarize_with_local_llm(file_path, custom_prompt_arg):
try:
logging.debug("Local LLM: Loading json data for summarization")
with open(file_path, 'r') as file:
segments = json.load(file)
logging.debug("Local LLM: Extracting text from the segments")
text = extract_text_from_segments(segments)
headers = {
'Content-Type': 'application/json'
}
logging.debug("Local LLM: Preparing data + prompt for submittal")
local_llm_prompt = f"{text} \n\n\n\n{custom_prompt_arg}"
data = {
"messages": [
{
"role": "system",
"content": "You are a professional summarizer."
},
{
"role": "user",
"content": local_llm_prompt
}
],
"max_tokens": 28000, # Adjust tokens as needed
}
logging.debug("Local LLM: Posting request")
response = requests.post('http://127.0.0.1:8080/v1/chat/completions', headers=headers, json=data)
if response.status_code == 200:
response_data = response.json()
if 'choices' in response_data and len(response_data['choices']) > 0:
summary = response_data['choices'][0]['message']['content'].strip()
logging.debug("Local LLM: Summarization successful")
print("Local LLM: Summarization successful.")
return summary
else:
logging.warning("Local LLM: Summary not found in the response data")
return "Local LLM: Summary not available"
else:
logging.debug("Local LLM: Summarization failed")
print("Local LLM: Failed to process summary:", response.text)
return "Local LLM: Failed to process summary"
except Exception as e:
logging.debug("Local LLM: Error in processing: %s", str(e))
print("Error occurred while processing summary with Local LLM:", str(e))
return "Local LLM: Error occurred while processing summary"
def summarize_with_llama(api_url, file_path, token, custom_prompt):
try:
logging.debug("llama: Loading JSON data")
with open(file_path, 'r') as file:
segments = json.load(file)
logging.debug(f"llama: Extracting text from segments file")
text = extract_text_from_segments(segments) # Define this function to extract text properly
headers = {
'accept': 'application/json',
'content-type': 'application/json',
}
if len(token) > 5:
headers['Authorization'] = f'Bearer {token}'
llama_prompt = f"{text} \n\n\n\n{custom_prompt}"
logging.debug("llama: Prompt being sent is {llama_prompt}")
data = {
"prompt": llama_prompt
}
logging.debug("llama: Submitting request to API endpoint")
print("llama: Submitting request to API endpoint")
response = requests.post(api_url, headers=headers, json=data)
response_data = response.json()
logging.debug("API Response Data: %s", response_data)
if response.status_code == 200:
# if 'X' in response_data:
logging.debug(response_data)
summary = response_data['content'].strip()
logging.debug("llama: Summarization successful")
print("Summarization successful.")
return summary
else:
logging.error(f"llama: API request failed with status code {response.status_code}: {response.text}")
return f"llama: API request failed: {response.text}"
except Exception as e:
logging.error("llama: Error in processing: %s", str(e))
return f"llama: Error occurred while processing summary with llama: {str(e)}"
# https://lite.koboldai.net/koboldcpp_api#/api%2Fv1/post_api_v1_generate
def summarize_with_kobold(api_url, file_path, kobold_api_token, custom_prompt):
try:
logging.debug("kobold: Loading JSON data")
with open(file_path, 'r') as file:
segments = json.load(file)
logging.debug(f"kobold: Extracting text from segments file")
text = extract_text_from_segments(segments)
headers = {
'accept': 'application/json',
'content-type': 'application/json',
}
kobold_prompt = f"{text} \n\n\n\n{custom_prompt}"
logging.debug("kobold: Prompt being sent is {kobold_prompt}")
# FIXME
# Values literally c/p from the api docs....
data = {
"max_context_length": 8096,
"max_length": 4096,
"prompt": kobold_prompt,
}
logging.debug("kobold: Submitting request to API endpoint")
print("kobold: Submitting request to API endpoint")
response = requests.post(api_url, headers=headers, json=data)
response_data = response.json()
logging.debug("kobold: API Response Data: %s", response_data)
if response.status_code == 200:
if 'results' in response_data and len(response_data['results']) > 0:
summary = response_data['results'][0]['text'].strip()
logging.debug("kobold: Summarization successful")
print("Summarization successful.")
return summary
else:
logging.error("Expected data not found in API response.")
return "Expected data not found in API response."
else:
logging.error(f"kobold: API request failed with status code {response.status_code}: {response.text}")
return f"kobold: API request failed: {response.text}"
except Exception as e:
logging.error("kobold: Error in processing: %s", str(e))
return f"kobold: Error occurred while processing summary with kobold: {str(e)}"
# https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API
def summarize_with_oobabooga(api_url, file_path, ooba_api_token, custom_prompt):
try:
logging.debug("ooba: Loading JSON data")
with open(file_path, 'r') as file:
segments = json.load(file)
logging.debug(f"ooba: Extracting text from segments file\n\n\n")
text = extract_text_from_segments(segments)
logging.debug(f"ooba: Finished extracting text from segments file")
headers = {
'accept': 'application/json',
'content-type': 'application/json',
}
# prompt_text = "I like to eat cake and bake cakes. I am a baker. I work in a French bakery baking cakes. It
# is a fun job. I have been baking cakes for ten years. I also bake lots of other baked goods, but cakes are
# my favorite." prompt_text += f"\n\n{text}" # Uncomment this line if you want to include the text variable
ooba_prompt = "{text}\n\n\n\n{custom_prompt}"
logging.debug("ooba: Prompt being sent is {ooba_prompt}")
data = {
"mode": "chat",
"character": "Example",
"messages": [{"role": "user", "content": ooba_prompt}]
}
logging.debug("ooba: Submitting request to API endpoint")
print("ooba: Submitting request to API endpoint")
response = requests.post(api_url, headers=headers, json=data, verify=False)
logging.debug("ooba: API Response Data: %s", response)
if response.status_code == 200:
response_data = response.json()
summary = response.json()['choices'][0]['message']['content']
logging.debug("ooba: Summarization successful")
print("Summarization successful.")
return summary
else:
logging.error(f"oobabooga: API request failed with status code {response.status_code}: {response.text}")
return f"ooba: API request failed with status code {response.status_code}: {response.text}"
except Exception as e:
logging.error("ooba: Error in processing: %s", str(e))
return f"ooba: Error occurred while processing summary with oobabooga: {str(e)}"
# FIXME - https://docs.vllm.ai/en/latest/getting_started/quickstart.html .... Great docs.
def summarize_with_vllm(vllm_api_url, vllm_api_key_function_arg, llm_model, text, vllm_custom_prompt_function_arg):
vllm_client = OpenAI(
base_url=vllm_api_url,
api_key=vllm_api_key_function_arg
)
custom_prompt = vllm_custom_prompt_function_arg
completion = client.chat.completions.create(
model=llm_model,
messages=[
{"role": "system", "content": "You are a professional summarizer."},
{"role": "user", "content": f"{text} \n\n\n\n{custom_prompt}"}
]
)
vllm_summary = completion.choices[0].message.content
return vllm_summary
# FIXME - Install is more trouble than care to deal with right now.
def summarize_with_tabbyapi(tabby_api_key, tabby_api_IP, text, tabby_model, custom_prompt):
model = tabby_model
headers = {
'Authorization': f'Bearer {tabby_api_key}',
'Content-Type': 'application/json'
}
data = {
'text': text,
'model': 'tabby' # Specify the model if needed
}
try:
response = requests.post('https://api.tabbyapi.com/summarize', headers=headers, json=data)
response.raise_for_status()
summary = response.json().get('summary', '')
return summary
except requests.exceptions.RequestException as e:
logger.error(f"Error summarizing with TabbyAPI: {e}")
return "Error summarizing with TabbyAPI."
def save_summary_to_file(summary, file_path):
logging.debug("Now saving summary to file...")
summary_file_path = file_path.replace('.segments.json', '_summary.txt')
logging.debug("Opening summary file for writing, *segments.json with *_summary.txt")
with open(summary_file_path, 'w') as file:
file.write(summary)
logging.info(f"Summary saved to file: {summary_file_path}")
summarizers: Dict[str, Callable[[str, str], str]] = {
'tabbyapi': summarize_with_tabbyapi,
'openai': summarize_with_openai,
'anthropic': summarize_with_claude,
'cohere': summarize_with_cohere,
'groq': summarize_with_groq,
'llama': summarize_with_llama,
'kobold': summarize_with_kobold,
'oobabooga': summarize_with_oobabooga
# Add more APIs here as needed
}
#
#
#######################################################################################################################
#######################################################################################################################
# Summarization with Detail
#
def summarize_with_detail_openai(text, detail, verbose=False):
summary_with_detail_variable = rolling_summarize(text, detail=detail, verbose=True)
print(len(openai_tokenize(summary_with_detail_variable)))
return summary_with_detail_variable
def summarize_with_detail_recursive_openai(text, detail, verbose=False):
summary_with_recursive_summarization = rolling_summarize(text, detail=detail, summarize_recursively=True)
print(summary_with_recursive_summarization)
#
#
#######################################################################################################################
#######################################################################################################################
# Gradio UI
#
# Only to be used when configured with Gradio for HF Space
def summarize_with_huggingface(huggingface_api_key, json_file_path, custom_prompt):
logging.debug(f"huggingface: Summarization process starting...")
client = InferenceClient()
#model = "microsoft/Phi-3-mini-128k-instruct"
model = "CohereForAI/c4ai-command-r-plus"
API_URL = f"https://api-inference.huggingface.co/models/{model}"
headers = {"Authorization": f"Bearer {huggingface_api_key}"}
client = InferenceClient(model=f"{model}", token=f"{huggingface_api_key}")
response = client.post(json={"inputs": "The goal of life is [MASK]."}, model="bert-base-uncased")
with open(json_file_path, 'r') as file:
segments = json.load(file)
text = ''.join([segment['text'] for segment in segments])
hf_prompt = text + "\n\n\n\n" + custom_prompt
if huggingface_api_key == "":
api_key = os.getenv(HF_TOKEN)
logging.debug("HUGGINGFACE API KEY CHECK: " + huggingface_api_key)
try:
logging.debug("huggingface: Loading json data for summarization")
with open(json_file_path, 'r') as file:
segments = json.load(file)
logging.debug("huggingface: Extracting text from the segments")
text = ' '.join([segment['text'] for segment in segments])
#api_key = os.getenv('HF_TOKEN').replace('"', '')
logging.debug("HUGGINGFACE API KEY CHECK #2: " + huggingface_api_key)
logging.debug("huggingface: Submitting request...")
response = client.text_generation(prompt=hf_prompt, max_new_tokens=4096)
if response is not None:
return response
#if response == FIXME:
#logging.debug("huggingface: Summarization successful")
#print("Summarization successful.")
#return response
#elif Bad Stuff:
# logging.debug(f"huggingface: Model is currently loading...{response.status_code}: {response.text}")
# global waiting_summary
# pretty_json = json.dumps(json.loads(response.text), indent=4) # Prettify JSON
# waiting_summary = f" {pretty_json} " # Use prettified JSON
# return waiting_summary
else:
logging.error(f"huggingface: Summarization failed with status code {response}")
return f"Failed to process summary, huggingface library error: {response}"
except Exception as e:
logging.error("huggingface: Error in processing: %s", str(e))
print(f"Error occurred while processing summary with huggingface: {str(e)}")
return None
# FIXME
# This is here for gradio authentication
# Its just not setup.
# def same_auth(username, password):
# return username == password
def format_transcription(transcription_result):
if transcription_result:
json_data = transcription_result['transcription']
return json.dumps(json_data, indent=2)
else:
return ""
def format_file_path(file_path, fallback_path=None):
if file_path and os.path.exists(file_path):
logging.debug(f"File exists: {file_path}")
return file_path
elif fallback_path and os.path.exists(fallback_path):
logging.debug(f"File does not exist: {file_path}. Returning fallback path: {fallback_path}")
return fallback_path
else:
logging.debug(f"File does not exist: {file_path}. No fallback path available.")
return None
def search_media(query, fields, keyword, page):
try:
results = search_and_display(query, fields, keyword, page)
return results
except Exception as e:
logger.error(f"Error searching media: {e}")
return str(e)
# FIXME - Change to use 'check_api()' function - also, create 'check_api()' function
def ask_question(transcription, question, api_name, api_key):
if not question.strip():
return "Please enter a question."
prompt = f"""Transcription:\n{transcription}
Given the above transcription, please answer the following:\n\n{question}"""
# FIXME - Refactor main API checks so they're their own function - api_check()
# Call api_check() function here
if api_name.lower() == "openai":
openai_api_key = api_key if api_key else config.get('API', 'openai_api_key', fallback=None)
headers = {
'Authorization': f'Bearer {openai_api_key}',
'Content-Type': 'application/json'
}
if openai_model:
pass
else:
openai_model = 'gpt-4-turbo'
data = {
"model": openai_model,
"messages": [
{
"role": "system",
"content": "You are a helpful assistant that answers questions based on the given "
"transcription and summary."
},
{
"role": "user",
"content": prompt
}
],
"max_tokens": 150000,
"temperature": 0.1
}
response = requests.post('https://api.openai.com/v1/chat/completions', headers=headers, json=data)
if response.status_code == 200:
answer = response.json()['choices'][0]['message']['content'].strip()
return answer
else:
return "Failed to process the question."
else:
return "Question answering is currently only supported with the OpenAI API."
import gradio as gr
def launch_ui(demo_mode=False):
whisper_models = ["small.en", "medium.en", "large"]
with gr.Blocks() as iface:
# Tab 1: Audio Transcription + Summarization
with gr.Tab("Audio Transcription + Summarization"):
with gr.Row():
# Light/Dark mode toggle switch
theme_toggle = gr.Radio(choices=["Light", "Dark"], value="Light",
label="Light/Dark Mode Toggle (Toggle to change UI color scheme)")
# UI Mode toggle switch
ui_mode_toggle = gr.Radio(choices=["Simple", "Advanced"], value="Simple",
label="UI Mode (Toggle to show all options)")
# URL input is always visible
url_input = gr.Textbox(label="URL (Mandatory)", placeholder="Enter the video URL here")
# Inputs to be shown or hidden
num_speakers_input = gr.Number(value=2, label="Number of Speakers(Optional - Currently has no effect)",
visible=False)
whisper_model_input = gr.Dropdown(choices=whisper_models, value="small.en",
label="Whisper Model(This is the ML model used for transcription.)",
visible=False)
custom_prompt_input = gr.Textbox(
label="Custom Prompt (Customize your summarization, or ask a question about the video and have it "
"answered)",
placeholder="Above is the transcript of a video. Please read "
"through the transcript carefully. Identify the main topics that are discussed over the "
"course of the transcript. Then, summarize the key points about each main topic in a "
"concise bullet point. The bullet points should cover the key information conveyed about "
"each topic in the video, but should be much shorter than the full transcript. Please "
"output your bullet point summary inside <bulletpoints> tags.",
lines=3, visible=True)
offset_input = gr.Number(value=0, label="Offset (Seconds into the video to start transcribing at)",
visible=False)
api_name_input = gr.Dropdown(
choices=[None, "Local-LLM", "OpenAI", "Anthropic", "Cohere", "Groq", "Llama.cpp", "Kobold", "Ooba", "HuggingFace"],
value=None,
label="(Optional) The LLM endpoint to have summarize your request. If you're running a local model, select 'Local-LLM'",
visible=True)
api_key_input = gr.Textbox(label="API Key (Mandatory unless you're running a local model/server/no API selected)",
placeholder="Enter your API key here; Ignore if using Local API or Built-in API('Local-LLM')",
visible=True)
vad_filter_input = gr.Checkbox(label="VAD Filter (WIP)", value=False,
visible=False)
rolling_summarization_input = gr.Checkbox(label="Enable Rolling Summarization", value=False,
visible=False)
download_video_input = gr.components.Checkbox(label="Download Video(Select to allow for file download of "
"selected video)", value=False, visible=False)
download_audio_input = gr.components.Checkbox(label="Download Audio(Select to allow for file download of "
"selected Video's Audio)", value=False, visible=False)
detail_level_input = gr.Slider(minimum=0.01, maximum=1.0, value=0.01, step=0.01, interactive=True,
label="Summary Detail Level (Slide me) (Only OpenAI currently supported)",
visible=False)
keywords_input = gr.Textbox(label="Keywords", placeholder="Enter keywords here (comma-separated Example: "
"tag_one,tag_two,tag_three)",
value="default,no_keyword_set",
visible=True)
question_box_input = gr.Textbox(label="Question",
placeholder="Enter a question to ask about the transcription",
visible=False)
chunk_summarization_input = gr.Checkbox(label="Time-based Chunk Summarization",
value=False,
visible=False)
chunk_duration_input = gr.Number(label="Chunk Duration (seconds)", value=DEFAULT_CHUNK_DURATION,
visible=False)
words_per_second_input = gr.Number(label="Words per Second", value=WORDS_PER_SECOND,
visible=False)
# time_based_summarization_input = gr.Checkbox(label="Enable Time-based Summarization", value=False,
# visible=False) time_chunk_duration_input = gr.Number(label="Time Chunk Duration (seconds)", value=60,
# visible=False) llm_model_input = gr.Dropdown(label="LLM Model", choices=["gpt-4o", "gpt-4-turbo",
# "claude-3-sonnet-20240229", "command-r-plus", "CohereForAI/c4ai-command-r-plus", "llama3-70b-8192"],
# value="gpt-4o", visible=False)
inputs = [
num_speakers_input, whisper_model_input, custom_prompt_input, offset_input, api_name_input,
api_key_input, vad_filter_input, download_video_input, download_audio_input,
rolling_summarization_input, detail_level_input, question_box_input, keywords_input,
chunk_summarization_input, chunk_duration_input, words_per_second_input
]
# inputs_1 = [
# url_input_1,
# num_speakers_input, whisper_model_input, custom_prompt_input_1, offset_input, api_name_input_1,
# api_key_input_1, vad_filter_input, download_video_input, download_audio_input,
# rolling_summarization_input, detail_level_input, question_box_input, keywords_input_1,
# chunk_summarization_input, chunk_duration_input, words_per_second_input,
# time_based_summarization_input, time_chunk_duration_input, llm_model_input
# ]
outputs = [
gr.Textbox(label="Transcription (Resulting Transcription from your input URL)"),
gr.Textbox(label="Summary or Status Message (Current status of Summary or Summary itself)"),
gr.File(label="Download Transcription as JSON (Download the Transcription as a file)"),
gr.File(label="Download Summary as Text (Download the Summary as a file)"),
gr.File(label="Download Video (Download the Video as a file)", visible=False),
gr.File(label="Download Audio (Download the Audio as a file)", visible=False),
]
def toggle_light(mode):
if mode == "Dark":
return """
<style>
body {
background-color: #1c1c1c;
color: #ffffff;
}
.gradio-container {
background-color: #1c1c1c;
color: #ffffff;
}
.gradio-button {
background-color: #4c4c4c;
color: #ffffff;
}
.gradio-input {
background-color: #4c4c4c;
color: #ffffff;
}
.gradio-dropdown {
background-color: #4c4c4c;
color: #ffffff;
}
.gradio-slider {
background-color: #4c4c4c;
}
.gradio-checkbox {
background-color: #4c4c4c;
}
.gradio-radio {
background-color: #4c4c4c;
}
.gradio-textbox {
background-color: #4c4c4c;
color: #ffffff;
}
.gradio-label {
color: #ffffff;
}
</style>
"""
else:
return """
<style>
body {
background-color: #ffffff;
color: #000000;
}
.gradio-container {
background-color: #ffffff;
color: #000000;
}
.gradio-button {
background-color: #f0f0f0;
color: #000000;
}
.gradio-input {
background-color: #f0f0f0;
color: #000000;
}
.gradio-dropdown {
background-color: #f0f0f0;
color: #000000;
}
.gradio-slider {
background-color: #f0f0f0;
}
.gradio-checkbox {
background-color: #f0f0f0;
}
.gradio-radio {
background-color: #f0f0f0;
}
.gradio-textbox {
background-color: #f0f0f0;
color: #000000;
}
.gradio-label {
color: #000000;
}
</style>
"""
# Set the event listener for the Light/Dark mode toggle switch
theme_toggle.change(fn=toggle_light, inputs=theme_toggle, outputs=gr.HTML())
# Function to toggle visibility of advanced inputs
def toggle_ui(mode):
visible = (mode == "Advanced")
return [
gr.update(visible=True) if i in [0, 3, 5, 6, 13] else gr.update(visible=visible)
for i in range(len(inputs))
]
# Set the event listener for the UI Mode toggle switch
ui_mode_toggle.change(fn=toggle_ui, inputs=ui_mode_toggle, outputs=inputs)
# Combine URL input and inputs lists
all_inputs = [url_input] + inputs
gr.Interface(
fn=process_url,
inputs=all_inputs,
outputs=outputs,
title="Video Transcription and Summarization",
description="Submit a video URL for transcription and summarization. Ensure you input all necessary "
"information including API keys."
)
# Tab 2: Scrape & Summarize Articles/Websites
with gr.Tab("Scrape & Summarize Articles/Websites"):
url_input = gr.Textbox(label="Article URL", placeholder="Enter the article URL here")
custom_article_title_input = gr.Textbox(label="Custom Article Title (Optional)",
placeholder="Enter a custom title for the article")
custom_prompt_input = gr.Textbox(
label="Custom Prompt (Optional)",
placeholder="Provide a custom prompt for summarization",
lines=3
)
api_name_input = gr.Dropdown(
choices=[None, "huggingface", "openai", "anthropic", "cohere", "groq", "llama", "kobold", "ooba"],
value=None,
label="API Name (Mandatory for Summarization)"
)
api_key_input = gr.Textbox(label="API Key (Mandatory if API Name is specified)",
placeholder="Enter your API key here; Ignore if using Local API or Built-in API")
keywords_input = gr.Textbox(label="Keywords", placeholder="Enter keywords here (comma-separated)",
value="default,no_keyword_set", visible=True)
scrape_button = gr.Button("Scrape and Summarize")
result_output = gr.Textbox(label="Result")
scrape_button.click(scrape_and_summarize, inputs=[url_input, custom_prompt_input, api_name_input,
api_key_input, keywords_input,
custom_article_title_input], outputs=result_output)
gr.Markdown("### Or Paste Unstructured Text Below (Will use settings from above)")
text_input = gr.Textbox(label="Unstructured Text", placeholder="Paste unstructured text here", lines=10)
text_ingest_button = gr.Button("Ingest Unstructured Text")
text_ingest_result = gr.Textbox(label="Result")
text_ingest_button.click(ingest_unstructured_text,
inputs=[text_input, custom_prompt_input, api_name_input, api_key_input,
keywords_input, custom_article_title_input], outputs=text_ingest_result)
with gr.Tab("Ingest & Summarize Documents"):
gr.Markdown("Plan to put ingestion form for documents here")
gr.Markdown("Will ingest documents and store into SQLite DB")
gr.Markdown("RAG here we come....:/")
with gr.Tab("Sample Prompts/Questions"):
gr.Markdown("Plan to put Sample prompts/questions here")
gr.Markdown("Fabric prompts/live UI?")
# Searchable list
with gr.Row():
search_box = gr.Textbox(label="Search prompts", placeholder="Type to filter prompts")
search_result = gr.Textbox(label="Matching prompts", interactive=False)
search_box.change(search_prompts, inputs=search_box, outputs=search_result)
# Interactive list
with gr.Row():
prompt_selector = gr.Radio(choices=all_prompts, label="Select a prompt")
selected_output = gr.Textbox(label="Selected prompt")
prompt_selector.change(handle_prompt_selection, inputs=prompt_selector, outputs=selected_output)
# Categorized display
with gr.Accordion("Category 1"):
gr.Markdown("\n".join(prompts_category_1))
with gr.Accordion("Category 2"):
gr.Markdown("\n".join(prompts_category_2))
# Gradio interface setup with tabs
search_tab = gr.Interface(
fn=search_and_display,
inputs=[
gr.Textbox(label="Search Query", placeholder="Enter your search query here..."),
gr.CheckboxGroup(label="Search Fields", choices=["Title", "Content", "URL", "Type", "Author"],
value=["Title"]),
gr.Textbox(label="Keyword", placeholder="Enter keywords here..."),
gr.Number(label="Page", value=1, precision=0),
gr.Checkbox(visible=False) # Dummy input to match the expected number of arguments
],
outputs=[
gr.Dataframe(label="Search Results"),
gr.Textbox(label="Message", visible=False)
],
title="Search Media Summaries",
description="Search for media (documents, videos, articles) and their summaries in the database. Use keywords for better filtering.",
allow_flagging="never"
)
export_tab = gr.Interface(
fn=export_to_csv,
inputs=[
gr.Textbox(label="Search Query", placeholder="Enter your search query here..."),
gr.CheckboxGroup(label="Search Fields", choices=["Title", "Content"], value=["Title"]),
gr.Textbox(label="Keyword (Match ALL, can use multiple keywords, separated by ',' (comma) )",
placeholder="Enter keywords here..."),
gr.Number(label="Page", value=1, precision=0),
gr.Number(label="Results per File", value=1000, precision=0)
],
outputs="text",
title="Export Search Results to CSV",
description="Export the search results to a CSV file."
)
keyword_add_interface = gr.Interface(
fn=add_keyword,
inputs=gr.Textbox(label="Add Keywords (comma-separated)", placeholder="Enter keywords here..."),
outputs="text",
title="Add Keywords",
description="Add one, or multiple keywords to the database.",
allow_flagging="never"
)
keyword_delete_interface = gr.Interface(
fn=delete_keyword,
inputs=gr.Textbox(label="Delete Keyword", placeholder="Enter keyword to delete here..."),
outputs="text",
title="Delete Keyword",
description="Delete a keyword from the database.",
allow_flagging="never"
)
keyword_tab = gr.TabbedInterface(
[keyword_add_interface, keyword_delete_interface],
["Add Keywords", "Delete Keywords"]
)
# Combine interfaces into a tabbed interface
tabbed_interface = gr.TabbedInterface([iface, search_tab, export_tab, keyword_tab],
["Transcription + Summarization", "Search", "Export", "Keywords"])
# Launch the interface
server_port_variable = 7860
if server_mode:
tabbed_interface.launch(share=True, server_port=server_port_variable, server_name="http://0.0.0.0")
elif share_public:
tabbed_interface.launch(share=True,)
else:
tabbed_interface.launch(share=False,)
#
#
#######################################################################################################################
#######################################################################################################################
# Prompt Sample Box
#
# Sample data
prompts_category_1 = [
"What are the key points discussed in the video?",
"Summarize the main arguments made by the speaker.",
"Describe the conclusions of the study presented."
]
prompts_category_2 = [
"How does the proposed solution address the problem?",
"What are the implications of the findings?",
"Can you explain the theory behind the observed phenomenon?"
]
all_prompts = prompts_category_1 + prompts_category_2
# Search function
def search_prompts(query):
filtered_prompts = [prompt for prompt in all_prompts if query.lower() in prompt.lower()]
return "\n".join(filtered_prompts)
# Handle prompt selection
def handle_prompt_selection(prompt):
return f"You selected: {prompt}"
#
#
#######################################################################################################################
#######################################################################################################################
# Local LLM Setup / Running
#
# Download latest llamafile from Github
# Example usage
#repo = "Mozilla-Ocho/llamafile"
#asset_name_prefix = "llamafile-"
#output_filename = "llamafile"
#download_latest_llamafile(repo, asset_name_prefix, output_filename)
def download_latest_llamafile(repo, asset_name_prefix, output_filename):
# Globals
global local_llm_model, llamafile
# Check if the file already exists
print("Checking for and downloading Llamafile it it doesn't already exist...")
if os.path.exists(output_filename):
time.sleep(1)
print("Llamafile already exists. Skipping download.")
logging.debug(f"{output_filename} already exists. Skipping download.")
time.sleep(1)
llamafile = output_filename
llamafile_exists = True
else:
llamafile_exists = False
if llamafile_exists == True:
pass
else:
# Get the latest release information
latest_release_url = f"https://api.github.com/repos/{repo}/releases/latest"
response = requests.get(latest_release_url)
if response.status_code != 200:
raise Exception(f"Failed to fetch latest release info: {response.status_code}")
latest_release_data = response.json()
tag_name = latest_release_data['tag_name']
# Get the release details using the tag name
release_details_url = f"https://api.github.com/repos/{repo}/releases/tags/{tag_name}"
response = requests.get(release_details_url)
if response.status_code != 200:
raise Exception(f"Failed to fetch release details for tag {tag_name}: {response.status_code}")
release_data = response.json()
assets = release_data.get('assets', [])
# Find the asset with the specified prefix
asset_url = None
for asset in assets:
if re.match(f"{asset_name_prefix}.*", asset['name']):
asset_url = asset['browser_download_url']
break
if not asset_url:
raise Exception(f"No asset found with prefix {asset_name_prefix}")
# Download the asset
response = requests.get(asset_url)
if response.status_code != 200:
raise Exception(f"Failed to download asset: {response.status_code}")
print("Llamafile downloaded successfully.")
logging.debug("Main: Llamafile downloaded successfully.")
# Save the file
with open(output_filename, 'wb') as file:
file.write(response.content)
logging.debug(f"Downloaded {output_filename} from {asset_url}")
print(f"Downloaded {output_filename} from {asset_url}")
# Check to see if the LLM already exists, and if not, download the LLM
print("Checking for and downloading LLM from Huggingface if needed...")
logging.debug("Main: Checking and downloading LLM from Huggingface if needed...")
mistral_7b_instruct_v0_2_q8_0_llamafile = "mistral-7b-instruct-v0.2.Q8_0.llamafile"
Samantha_Mistral_Instruct_7B_Bulleted_Notes_Q8 = "samantha-mistral-instruct-7b-bulleted-notes.Q8_0.gguf"
Phi_3_mini_4k_instruct_Q8_0_llamafile = "Phi-3-mini-4k-instruct.Q8_0.llamafile"
meta_Llama_3_8B_Instruct_Q8_0_llamafile = 'Meta-Llama-3-8B-Instruct.Q8_0.llamafile'
available_models = []
# Check for existence of model files
if os.path.exists(mistral_7b_instruct_v0_2_q8_0_llamafile):
available_models.append(mistral_7b_instruct_v0_2_q8_0_llamafile)
print("Mistral-7B-Instruct-v0.2.Q8_0.llamafile already exists. Skipping download.")
if os.path.exists(Samantha_Mistral_Instruct_7B_Bulleted_Notes_Q8):
available_models.append(Samantha_Mistral_Instruct_7B_Bulleted_Notes_Q8)
print("Samantha-Mistral-Instruct-7B-Bulleted-Notes-Q8_0.gguf already exists. Skipping download.")
if os.path.exists(Phi_3_mini_4k_instruct_Q8_0_llamafile):
available_models.append(Phi_3_mini_4k_instruct_Q8_0_llamafile)
print("Phi-3-mini-4k-instruct-Q8_0.llamafile already exists. Skipping download.")
if os.path.exists(meta_Llama_3_8B_Instruct_Q8_0_llamafile):
available_models.append(meta_Llama_3_8B_Instruct_Q8_0_llamafile)
print("Meta-Llama-3-8B-Instruct.Q8_0.llamafile already exists. Skipping download.")
# If no models are available, download the models
if not available_models:
user_choice_main = input("Would you like to download an LLM model? (Y/N): ")
elif available_models:
user_choice_main = input("\nSeems you already have a model available, would you like to download another LLM model? (Y/N): ")
if user_choice_main.lower() == "y":
logging.debug("Main: Checking and downloading LLM from Huggingface if needed...")
time.sleep(1)
dl_check = input("Final chance to back out, hit 'N'/'n' to cancel, or 'Y'/'y' to continue: ")
if dl_check.lower == "n" or "2":
exit()
else:
llm_choice = input("\nWhich LLM model would you like to download?\n\n1. Mistral-7B-Instruct-v0.2-GGUF \n2. Samantha-Mistral-Instruct-7B-Bulleted-Notes) \n3. Microsoft Phi3-Mini-128k 3.8B): \n\nPress '1', '2', or '3' to specify:\n\n ")
while llm_choice != "1" and llm_choice != "2" and llm_choice != "3":
print("Invalid choice. Please try again.")
if llm_choice == "1":
print("Downloading the Mistral-7B-Instruct-v0.2 LLM from Huggingface...")
print("Gonna be a bit...")
print("Like seriously, an 8GB file...(don't say I didn't warn you...)")
time.sleep(2)
mistral_7b_instruct_v0_2_q8_0_llamafile_sha256 = "1ee6114517d2f770425c880e5abc443da36b193c82abec8e2885dd7ce3b9bfa6"
llm_download_model_hash = mistral_7b_instruct_v0_2_q8_0_llamafile_sha256
llamafile_llm_url = "https://huggingface.co/Mozilla/Mistral-7B-Instruct-v0.2-llamafile/resolve/main/mistral-7b-instruct-v0.2.Q8_0.llamafile?download=true"
llamafile_llm_output_filename = "mistral-7b-instruct-v0.2.Q8_0.llamafile"
download_file(llamafile_llm_url, llamafile_llm_output_filename, llm_download_model_hash)
local_llm_model = "mistral-7b-instruct-v0.2.Q8_0.llamafile"
elif llm_choice == "2":
print("Downloading the samantha-mistra-instruct-7b-bulleted-notes LLM from Huggingface...")
print("Gonna be a bit...")
print("Like seriously, an 8GB file...(don't say I didn't warn you...)")
time.sleep(2)
samantha_mistral_instruct_7b_bulleted_notes_q8_0_gguf_sha256 = "6334c1ab56c565afd86535271fab52b03e67a5e31376946bce7bf5c144e847e4"
llm_download_model_hash = samantha_mistral_instruct_7b_bulleted_notes_q8_0_gguf_sha256
llamafile_llm_output_filename = "samantha-mistral-instruct-7b-bulleted-notes.Q8_0.gguf"
llamafile_llm_url = "https://huggingface.co/cognitivetech/samantha-mistral-instruct-7b-bulleted-notes-GGUF/resolve/main/samantha-mistral-instruct-7b-bulleted-notes.Q8_0.gguf?download=true"
download_file(llamafile_llm_url, llamafile_llm_output_filename, llm_download_model_hash)
local_llm_model = "samantha-mistral-instruct-7b-bulleted-notes.Q8_0.gguf"
elif llm_choice == "3":
print("Downloading MS Phi-3-4k-3.8B LLM from Huggingface...")
print("Gonna be a bit...")
print("Like seriously, a 4GB file...(don't say I didn't warn you...)")
time.sleep(2)
Phi_3_mini_4k_instruct_Q8_0_gguf_sha256 = "1b51fc72fda221dd7b4d3e84603db37fbb1ce53c17f2e7583b7026d181b8d20f"
llm_download_model_hash = Phi_3_mini_4k_instruct_Q8_0_gguf_sha256
llamafile_llm_output_filename = "Phi-3-mini-4k-instruct.Q8_0.llamafile"
llamafile_llm_url = "https://huggingface.co/Mozilla/Phi-3-mini-4k-instruct-llamafile/resolve/main/Phi-3-mini-4k-instruct.Q8_0.llamafile?download=true"
download_file(llamafile_llm_url, llamafile_llm_output_filename, llm_download_model_hash)
local_llm_model = "Phi-3-mini-4k-instruct-Q8_0.llamafile"
elif llm_choice == "4":
print("Downloading the Llama-3-8B LLM from Huggingface...")
print("Gonna be a bit...")
print("Like seriously, a 8GB file...(don't say I didn't warn you...)")
time.sleep(2)
meta_Llama_3_8B_Instruct_Q8_0_lamafile_sha256 = "406868a97f02f57183716c7e4441d427f223fdbc7fa42964ef10c4d60dd8ed37"
llm_download_model_hash = meta_Llama_3_8B_Instruct_Q8_0_lamafile_sha256
llamafile_llm_output_filename = "Meta-Llama-3-8B-Instruct.Q8_0.llamafile"
llamafile_llm_url = "https://huggingface.co/Mozilla/Meta-Llama-3-8B-Instruct-llamafile/resolve/main/Meta-Llama-3-8B-Instruct.Q8_0.llamafile?download=true"
download_file(llamafile_llm_url, llamafile_llm_output_filename, llm_download_model_hash)
local_llm_model = "Meta-Llama-3-8B-Instruct.Q8_0.llamafile"
else:
print("Invalid choice. Please try again.")
else:
pass
if available_models:
print("\n\nAvailable models:")
for idx, model in enumerate(available_models, start=1):
print(f"{idx}. {model}")
user_choice = input("\nWhich model would you like to use? Please enter the corresponding number: ")
while not user_choice.isdigit() or int(user_choice) not in range(1, len(available_models) + 1):
print("Invalid choice. Please try again.")
user_choice = input("Which model would you like to use? Please enter the corresponding number: ")
user_answer = available_models[int(user_choice) - 1]
local_llm_model = user_answer
print(f"You have chosen to use: {user_answer}")
else:
print("No models available/Found.")
print("Please run the script again and select a model, or download one. Exiting...")
exit()
return llamafile, user_answer
def download_file(url, dest_path, expected_checksum=None, max_retries=3, delay=5):
temp_path = dest_path + '.tmp'
for attempt in range(max_retries):
try:
# Check if a partial download exists and get its size
resume_header = {}
if os.path.exists(temp_path):
resume_header = {'Range': f'bytes={os.path.getsize(temp_path)}-'}
response = requests.get(url, stream=True, headers=resume_header)
response.raise_for_status()
# Get the total file size from headers
total_size = int(response.headers.get('content-length', 0))
initial_pos = os.path.getsize(temp_path) if os.path.exists(temp_path) else 0
mode = 'ab' if 'Range' in response.headers else 'wb'
with open(temp_path, mode) as temp_file, tqdm(
total=total_size, unit='B', unit_scale=True, desc=dest_path, initial=initial_pos, ascii=True
) as pbar:
for chunk in response.iter_content(chunk_size=8192):
if chunk: # filter out keep-alive new chunks
temp_file.write(chunk)
pbar.update(len(chunk))
# Verify the checksum if provided
if expected_checksum:
if not verify_checksum(temp_path, expected_checksum):
os.remove(temp_path)
raise ValueError("Downloaded file's checksum does not match the expected checksum")
# Move the file to the final destination
os.rename(temp_path, dest_path)
print("Download complete and verified!")
return dest_path
except Exception as e:
print(f"Attempt {attempt + 1} failed: {e}")
if attempt < max_retries - 1:
print(f"Retrying in {delay} seconds...")
time.sleep(delay)
else:
print("Max retries reached. Download failed.")
raise
def verify_checksum(file_path, expected_checksum):
sha256_hash = hashlib.sha256()
with open(file_path, 'rb') as f:
for byte_block in iter(lambda: f.read(4096), b''):
sha256_hash.update(byte_block)
return sha256_hash.hexdigest() == expected_checksum
# FIXME - Doesn't work...
# Function to close out llamafile process on script exit.
def cleanup_process():
global process
if process is not None:
process.terminate()
process = None
print("Terminated the external process")
def signal_handler(sig, frame):
logging.info('Signal handler called with signal: %s', sig)
cleanup_process()
sys.exit(0)
# Function to launch the llamafile in an external terminal window
# local_llm_model = Whatever the local model is
def local_llm_function():
repo = "Mozilla-Ocho/llamafile"
asset_name_prefix = "llamafile-"
useros = os.name
if useros == "nt":
output_filename = "llamafile.exe"
else:
output_filename = "llamafile"
print(
"WARNING - Checking for existence of llamafile and HuggingFace model, downloading if needed...This could be a while")
print("WARNING - and I mean a while. We're talking an 8 Gigabyte model here...")
print("WARNING - Hope you're comfy. Or it's already downloaded.")
time.sleep(6)
logging.debug("Main: Checking and downloading Llamafile from Github if needed...")
llamafile, user_answer = download_latest_llamafile(repo, asset_name_prefix, output_filename)
logging.debug("Main: Llamafile downloaded successfully.")
# Launch the llamafile in an external process with the specified argument
arguments = ["-m", user_answer]
try:
logging.info("Main: Launching the LLM (llamafile) in an external terminal window...")
if useros == "nt":
launch_in_new_terminal_windows(llamafile, arguments)
elif useros == "posix":
launch_in_new_terminal_linux(llamafile, arguments)
else:
launch_in_new_terminal_mac(llamafile, arguments)
# FIXME - pid doesn't exist in this context
#logging.info(f"Main: Launched the {llamafile_path} with PID {process.pid}")
atexit.register(cleanup_process)
except Exception as e:
logging.error(f"Failed to launch the process: {e}")
print(f"Failed to launch the process: {e}")
def launch_in_new_terminal_windows(executable, args):
command = f'start cmd /k "{executable} {" ".join(args)}"'
process = subprocess.run(command, shell=True)
# FIXME
def launch_in_new_terminal_linux(executable, args):
command = f'gnome-terminal -- {executable} {" ".join(args)}'
process = subprocess.run(command, shell=True)
# FIXME
def launch_in_new_terminal_mac(executable, args):
command = f'open -a Terminal.app {executable} {" ".join(args)}'
process = subprocess.run(command, shell=True)
#
#
#######################################################################################################################
#######################################################################################################################
# Main()
#
def main(input_path, api_name=None, api_key=None,
num_speakers=2,
whisper_model="small.en",
offset=0,
vad_filter=False,
download_video_flag=False,
custom_prompt=None,
overwrite=False,
rolling_summarization=False,
detail=0.01,
keywords=None,
chunk_summarization=False,
chunk_duration=None,
words_per_second=None,
llm_model=None,
time_based=False):
global detail_level_number, summary, audio_file, detail_level, summary
detail_level = detail
print(f"Keywords: {keywords}")
if input_path is None and args.user_interface:
return []
start_time = time.monotonic()
paths = [] # Initialize paths as an empty list
if os.path.isfile(input_path) and input_path.endswith('.txt'):
logging.debug("MAIN: User passed in a text file, processing text file...")
paths = read_paths_from_file(input_path)
elif os.path.exists(input_path):
logging.debug("MAIN: Local file path detected")
paths = [input_path]
elif (info_dict := get_youtube(input_path)) and 'entries' in info_dict:
logging.debug("MAIN: YouTube playlist detected")
print(
"\n\nSorry, but playlists aren't currently supported. You can run the following command to generate a "
"text file that you can then pass into this script though! (It may not work... playlist support seems "
"spotty)" + """\n\n\tpython Get_Playlist_URLs.py <Youtube Playlist URL>\n\n\tThen,\n\n\tpython
diarizer.py <playlist text file name>\n\n""")
return
else:
paths = [input_path]
results = []
for path in paths:
try:
if path.startswith('http'):
logging.debug("MAIN: URL Detected")
info_dict = get_youtube(path)
json_file_path = None
if info_dict:
logging.debug("MAIN: Creating path for video file...")
download_path = create_download_directory(info_dict['title'])
logging.debug("MAIN: Path created successfully\n MAIN: Now Downloading video from yt_dlp...")
try:
video_path = download_video(path, download_path, info_dict, download_video_flag)
except RuntimeError as e:
logging.error(f"Error downloading video: {str(e)}")
# FIXME - figure something out for handling this situation....
continue
logging.debug("MAIN: Video downloaded successfully")
logging.debug("MAIN: Converting video file to WAV...")
audio_file = convert_to_wav(video_path, offset)
logging.debug("MAIN: Audio file converted successfully")
else:
if os.path.exists(path):
logging.debug("MAIN: Local file path detected")
download_path, info_dict, audio_file = process_local_file(path)
else:
logging.error(f"File does not exist: {path}")
continue
if info_dict:
logging.debug("MAIN: Creating transcription file from WAV")
segments = speech_to_text(audio_file, whisper_model=whisper_model, vad_filter=vad_filter)
transcription_result = {
'video_path': path,
'audio_file': audio_file,
'transcription': segments
}
results.append(transcription_result)
logging.info(f"MAIN: Transcription complete: {audio_file}")
# Perform rolling summarization based on API Name, detail level, and if an API key exists
# Will remove the API key once rolling is added for llama.cpp
# FIXME - Add input for model name for tabby and vllm
if rolling_summarization:
logging.info("MAIN: Rolling Summarization")
# Extract the text from the segments
text = extract_text_from_segments(segments)
# Set the json_file_path
json_file_path = audio_file.replace('.wav', '.segments.json')
# Perform rolling summarization
summary = summarize_with_detail_openai(text, detail=detail_level, verbose=False)
# Handle the summarized output
if summary:
transcription_result['summary'] = summary
logging.info("MAIN: Rolling Summarization successful.")
save_summary_to_file(summary, json_file_path)
else:
logging.warning("MAIN: Rolling Summarization failed.")
# FIXME - fucking mess of a function.
# # Time-based Summarization
# elif args.time_based:
# logging.info("MAIN: Time-based Summarization")
# global time_based_value
# time_based_value = args.time_based
# # Set the json_file_path
# json_file_path = audio_file.replace('.wav', '.segments.json')
#
# # Perform time-based summarization
# summary = time_chunk_summarize(api_name, api_key, segments, args.time_based, custom_prompt,
# llm_model)
#
# # Handle the summarized output
# if summary:
# transcription_result['summary'] = summary
# logging.info("MAIN: Time-based Summarization successful.")
# save_summary_to_file(summary, json_file_path)
# else:
# logging.warning("MAIN: Time-based Summarization failed.")
# Perform chunk summarization - FIXME
elif chunk_summarization:
logging.info("MAIN: Chunk Summarization")
# Set the json_file_path
json_file_path = audio_file.replace('.wav', '.segments.json')
# Perform chunk summarization
summary = summarize_chunks(api_name, api_key, segments, chunk_duration, words_per_second)
# Handle the summarized output
if summary:
transcription_result['summary'] = summary
logging.info("MAIN: Chunk Summarization successful.")
save_summary_to_file(summary, json_file_path)
else:
logging.warning("MAIN: Chunk Summarization failed.")
# Perform summarization based on the specified API
elif api_name:
logging.debug(f"MAIN: Summarization being performed by {api_name}")
json_file_path = audio_file.replace('.wav', '.segments.json')
if api_name.lower() == 'openai':
openai_api_key = api_key if api_key else config.get('API', 'openai_api_key',
fallback=None)
try:
logging.debug(f"MAIN: trying to summarize with openAI")
summary = summarize_with_openai(openai_api_key, json_file_path, custom_prompt)
except requests.exceptions.ConnectionError:
requests.status_code = "Connection: "
elif api_name.lower() == "anthropic":
anthropic_api_key = api_key if api_key else config.get('API', 'anthropic_api_key',
fallback=None)
try:
logging.debug(f"MAIN: Trying to summarize with anthropic")
summary = summarize_with_claude(anthropic_api_key, json_file_path, anthropic_model,
custom_prompt)
except requests.exceptions.ConnectionError:
requests.status_code = "Connection: "
elif api_name.lower() == "cohere":
cohere_api_key = os.getenv('COHERE_TOKEN').replace('"', '') if api_key is None else api_key
try:
logging.debug(f"MAIN: Trying to summarize with cohere")
summary = summarize_with_cohere(cohere_api_key, json_file_path, cohere_model, custom_prompt)
except requests.exceptions.ConnectionError:
requests.status_code = "Connection: "
elif api_name.lower() == "groq":
groq_api_key = api_key if api_key else config.get('API', 'groq_api_key', fallback=None)
try:
logging.debug(f"MAIN: Trying to summarize with Groq")
summary = summarize_with_groq(groq_api_key, json_file_path, groq_model, custom_prompt)
except requests.exceptions.ConnectionError:
requests.status_code = "Connection: "
elif api_name.lower() == "llama":
llama_token = api_key if api_key else config.get('API', 'llama_api_key', fallback=None)
llama_ip = llama_api_IP
try:
logging.debug(f"MAIN: Trying to summarize with Llama.cpp")
summary = summarize_with_llama(llama_ip, json_file_path, llama_token, custom_prompt)
except requests.exceptions.ConnectionError:
requests.status_code = "Connection: "
elif api_name.lower() == "kobold":
kobold_token = api_key if api_key else config.get('API', 'kobold_api_key', fallback=None)
kobold_ip = kobold_api_IP
try:
logging.debug(f"MAIN: Trying to summarize with kobold.cpp")
summary = summarize_with_kobold(kobold_ip, json_file_path, kobold_token, custom_prompt)
except requests.exceptions.ConnectionError:
requests.status_code = "Connection: "
elif api_name.lower() == "ooba":
ooba_token = api_key if api_key else config.get('API', 'ooba_api_key', fallback=None)
ooba_ip = ooba_api_IP
try:
logging.debug(f"MAIN: Trying to summarize with oobabooga")
summary = summarize_with_oobabooga(ooba_ip, json_file_path, ooba_token, custom_prompt)
except requests.exceptions.ConnectionError:
requests.status_code = "Connection: "
elif api_name.lower() == "tabbyapi":
tabbyapi_key = api_key if api_key else config.get('API', 'tabby_api_key', fallback=None)
tabbyapi_ip = tabby_api_IP
try:
logging.debug(f"MAIN: Trying to summarize with tabbyapi")
tabby_model = llm_model
summary = summarize_with_tabbyapi(tabby_api_key, tabby_api_IP, json_file_path, tabby_model,
custom_prompt)
except requests.exceptions.ConnectionError:
requests.status_code = "Connection: "
elif api_name.lower() == "vllm":
logging.debug(f"MAIN: Trying to summarize with VLLM")
summary = summarize_with_vllm(vllm_api_url, vllm_api_key, llm_model, json_file_path,
custom_prompt)
elif api_name.lower() == "local-llm":
logging.debug(f"MAIN: Trying to summarize with the local LLM, Mistral Instruct v0.2")
local_llm_url = "http://127.0.0.1:8080"
summary = summarize_with_local_llm(json_file_path, custom_prompt)
elif api_name.lower() == "huggingface":
huggingface_api_key = api_key if api_key else config.get('API', 'huggingface_api_key',
fallback=None)
try:
logging.debug(f"MAIN: Trying to summarize with huggingface")
summarize_with_huggingface(huggingface_api_key, json_file_path, custom_prompt)
except requests.exceptions.ConnectionError:
requests.status_code = "Connection: "
else:
logging.warning(f"Unsupported API: {api_name}")
summary = None
if summary:
transcription_result['summary'] = summary
logging.info(f"Summary generated using {api_name} API")
save_summary_to_file(summary, json_file_path)
elif final_summary:
logging.info(f"Rolling summary generated using {api_name} API")
logging.info(f"Final Rolling summary is {final_summary}\n\n")
save_summary_to_file(final_summary, json_file_path)
else:
logging.warning(f"Failed to generate summary using {api_name} API")
else:
logging.info("MAIN: #2 - No API specified. Summarization will not be performed")
# Add media to the database
add_media_with_keywords(
url=path,
title=info_dict.get('title', 'Untitled'),
media_type='video',
content=' '.join([segment['text'] for segment in segments]),
keywords=','.join(keywords),
prompt=custom_prompt or 'No prompt provided',
summary=summary or 'No summary provided',
transcription_model=whisper_model,
author=info_dict.get('uploader', 'Unknown'),
ingestion_date=datetime.now().strftime('%Y-%m-%d')
)
except Exception as e:
logging.error(f"Error processing {path}: {str(e)}")
continue
except Exception as e:
logging.error(f"Error processing path: {path}")
logging.error(str(e))
continue
# end_time = time.monotonic()
# print("Total program execution time: " + timedelta(seconds=end_time - start_time))
return results
def signal_handler(signal, frame):
logging.info('Signal received, exiting...')
sys.exit(0)
############################## MAIN ##############################
#
#
if __name__ == "__main__":
# Register signal handlers
signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGTERM, signal_handler)
# Establish logging baseline
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
parser = argparse.ArgumentParser(
description='Transcribe and summarize videos.',
epilog='''
Sample commands:
1. Simple Sample command structure:
summarize.py <path_to_video> -api openai -k tag_one tag_two tag_three
2. Rolling Summary Sample command structure:
summarize.py <path_to_video> -api openai -prompt "custom_prompt_goes_here-is-appended-after-transcription" -roll -detail 0.01 -k tag_one tag_two tag_three
3. FULL Sample command structure:
summarize.py <path_to_video> -api openai -ns 2 -wm small.en -off 0 -vad -log INFO -prompt "custom_prompt" -overwrite -roll -detail 0.01 -k tag_one tag_two tag_three
4. Sample command structure for UI:
summarize.py -gui -log DEBUG
''',
formatter_class=argparse.RawTextHelpFormatter
)
parser.add_argument('input_path', type=str, help='Path or URL of the video', nargs='?')
parser.add_argument('-v', '--video', action='store_true', help='Download the video instead of just the audio')
parser.add_argument('-api', '--api_name', type=str, help='API name for summarization (optional)')
parser.add_argument('-key', '--api_key', type=str, help='API key for summarization (optional)')
parser.add_argument('-ns', '--num_speakers', type=int, default=2, help='Number of speakers (default: 2)')
parser.add_argument('-wm', '--whisper_model', type=str, default='small.en',
help='Whisper model (default: small.en)')
parser.add_argument('-off', '--offset', type=int, default=0, help='Offset in seconds (default: 0)')
parser.add_argument('-vad', '--vad_filter', action='store_true', help='Enable VAD filter')
parser.add_argument('-log', '--log_level', type=str, default='INFO',
choices=['DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL'], help='Log level (default: INFO)')
parser.add_argument('-gui', '--user_interface', action='store_true', help="Launch the Gradio user interface")
parser.add_argument('-demo', '--demo_mode', action='store_true', help='Enable demo mode')
parser.add_argument('-prompt', '--custom_prompt', type=str,
help='Pass in a custom prompt to be used in place of the existing one.\n (Probably should just '
'modify the script itself...)')
parser.add_argument('-overwrite', '--overwrite', action='store_true', help='Overwrite existing files')
parser.add_argument('-roll', '--rolling_summarization', action='store_true', help='Enable rolling summarization')
parser.add_argument('-detail', '--detail_level', type=float, help='Mandatory if rolling summarization is enabled, '
'defines the chunk size.\n Default is 0.01(lots '
'of chunks) -> 1.00 (few chunks)\n Currently '
'only OpenAI works. ',
default=0.01, )
# FIXME - This or time based...
parser.add_argument('--chunk_duration', type=int, default=DEFAULT_CHUNK_DURATION,
help='Duration of each chunk in seconds')
# FIXME - This or chunk_duration.... -> Maybe both???
parser.add_argument('-time', '--time_based', type=int,
help='Enable time-based summarization and specify the chunk duration in seconds (minimum 60 seconds, increments of 30 seconds)')
parser.add_argument('-model', '--llm_model', type=str, default='',
help='Model to use for LLM summarization (only used for vLLM/TabbyAPI)')
parser.add_argument('-k', '--keywords', nargs='+', default=['cli_ingest_no_tag'],
help='Keywords for tagging the media, can use multiple separated by spaces (default: cli_ingest_no_tag)')
parser.add_argument('--log_file', type=str, help='Where to save logfile (non-default)')
parser.add_argument('--local_llm', action='store_true', help="Use a local LLM from the script(Downloads llamafile from github and 'mistral-7b-instruct-v0.2.Q8' - 8GB model from Huggingface)")
parser.add_argument('--server_mode', action='store_true', help='Run in server mode (This exposes the GUI/Server to the network)')
parser.add_argument('--share_public', type=int, default=7860, help="This will use Gradio's built-in ngrok tunneling to share the server publicly on the internet. Specify the port to use (default: 7860)")
parser.add_argument('--port', type=int, default=7860, help='Port to run the server on')
# parser.add_argument('-o', '--output_path', type=str, help='Path to save the output file')
args = parser.parse_args()
share_public = args.share_public
server_mode = args.server_mode
server_port = args.port
########## Logging setup
logger = logging.getLogger()
logger.setLevel(getattr(logging, args.log_level))
# Create console handler
console_handler = logging.StreamHandler()
console_handler.setLevel(getattr(logging, args.log_level))
console_formatter = logging.Formatter('%(asctime)s - %(levelname)s - %(message)s')
console_handler.setFormatter(console_formatter)
logger.addHandler(console_handler)
if args.log_file:
# Create file handler
file_handler = logging.FileHandler(args.log_file)
file_handler.setLevel(getattr(logging, args.log_level))
file_formatter = logging.Formatter('%(asctime)s - %(levelname)s - %(message)s')
file_handler.setFormatter(file_formatter)
logger.addHandler(file_handler)
logger.info(f"Log file created at: {args.log_file}")
########## Custom Prompt setup
custom_prompt = args.custom_prompt
if custom_prompt is None or custom_prompt == "":
logging.debug("No custom prompt defined, will use default")
args.custom_prompt = ("\n\nabove is the transcript of a video "
"Please read through the transcript carefully. Identify the main topics that are "
"discussed over the course of the transcript. Then, summarize the key points about each "
"main topic in a concise bullet point. The bullet points should cover the key "
"information conveyed about each topic in the video, but should be much shorter than "
"the full transcript. Please output your bullet point summary inside <bulletpoints> "
"tags.")
custom_prompt = args.custom_prompt
print("No custom prompt defined, will use default")
else:
logging.debug(f"Custom prompt defined, will use \n\nf{custom_prompt} \n\nas the prompt")
print(f"Custom Prompt has been defined. Custom prompt: \n\n {args.custom_prompt}")
# Check if the user wants to use the local LLM from the script
local_llm = args.local_llm
logging.info(f'Local LLM flag: {local_llm}')
if args.user_interface:
# if local_llm:
# local_llm_function()
# time.sleep(3)
# webbrowser.open_new_tab('http://127.0.0.1:7860')
launch_ui(demo_mode=False)
else:
if not args.input_path:
parser.print_help()
sys.exit(1)
logging.info('Starting the transcription and summarization process.')
logging.info(f'Input path: {args.input_path}')
logging.info(f'API Name: {args.api_name}')
logging.info(f'Number of speakers: {args.num_speakers}')
logging.info(f'Whisper model: {args.whisper_model}')
logging.info(f'Offset: {args.offset}')
logging.info(f'VAD filter: {args.vad_filter}')
logging.info(f'Log Level: {args.log_level}')
logging.info(f'Demo Mode: {args.demo_mode}')
logging.info(f'Custom Prompt: {args.custom_prompt}')
logging.info(f'Overwrite: {args.overwrite}')
logging.info(f'Rolling Summarization: {args.rolling_summarization}')
logging.info(f'User Interface: {args.user_interface}')
logging.info(f'Video Download: {args.video}')
# logging.info(f'Save File location: {args.output_path}')
# logging.info(f'Log File location: {args.log_file}')
# Get all API keys from the config
api_keys = {key: value for key, value in config.items('API') if key.endswith('_api_key')}
api_name = args.api_name
# Rolling Summarization will only be performed if an API is specified and the API key is available
# and the rolling summarization flag is set
#
summary = None # Initialize to ensure it's always defined
if args.detail_level == None:
args.detail_level = 0.01
if args.api_name and args.rolling_summarization and any(
key.startswith(args.api_name) and value is not None for key, value in api_keys.items()):
logging.info(f'MAIN: API used: {args.api_name}')
logging.info('MAIN: Rolling Summarization will be performed.')
elif args.api_name:
logging.info(f'MAIN: API used: {args.api_name}')
logging.info('MAIN: Summarization (not rolling) will be performed.')
else:
logging.info('No API specified. Summarization will not be performed.')
logging.debug("Platform check being performed...")
platform_check()
logging.debug("CUDA check being performed...")
cuda_check()
logging.debug("ffmpeg check being performed...")
check_ffmpeg()
llm_model = args.llm_model or None
try:
results = main(args.input_path, api_name=args.api_name,
api_key=args.api_key,
num_speakers=args.num_speakers,
whisper_model=args.whisper_model,
offset=args.offset,
vad_filter=args.vad_filter,
download_video_flag=args.video,
custom_prompt=args.custom_prompt,
overwrite=args.overwrite,
rolling_summarization=args.rolling_summarization,
detail=args.detail_level,
keywords=args.keywords,
chunk_summarization=False,
chunk_duration=None,
words_per_second=None,
llm_model=args.llm_model,
time_based=args.time_based)
logging.info('Transcription process completed.')
atexit.register(cleanup_process)
except Exception as e:
logging.error('An error occurred during the transcription process.')
logging.error(str(e))
sys.exit(1)
finally:
cleanup_process()
|