File size: 123,136 Bytes
5bfd639 2eeb21e 5bfd639 2eeb21e 5bfd639 2eeb21e 5bfd639 d16dd39 5bfd639 d16dd39 5bfd639 2eeb21e 5bfd639 2eeb21e 5bfd639 2eeb21e 5bfd639 2eeb21e 5bfd639 2eeb21e 5bfd639 2eeb21e 5bfd639 2eeb21e 5bfd639 d16dd39 5bfd639 2eeb21e 5bfd639 2eeb21e 5bfd639 d16dd39 5bfd639 d16dd39 5bfd639 d16dd39 5bfd639 9258c0d 5bfd639 2eeb21e 5bfd639 9258c0d 5bfd639 d16dd39 5bfd639 d16dd39 5bfd639 d16dd39 5bfd639 d16dd39 5bfd639 d16dd39 5bfd639 d16dd39 2eeb21e 5bfd639 d16dd39 9258c0d d16dd39 5bfd639 d16dd39 5bfd639 d16dd39 5bfd639 d16dd39 2eeb21e d16dd39 5bfd639 71f3a82 5bfd639 71f3a82 5bfd639 71f3a82 5bfd639 | 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 | # models_manager.py
import copy
import json
import os
import importlib
import sys
import time
import threading
import traceback
import tkinter as tk
from tkinter import ttk, messagebox, filedialog
from tkinter.scrolledtext import ScrolledText
import urllib.request
import urllib.error
from pathlib import Path
import subprocess
import tempfile
import pandas as pd
from datasets import Dataset, Features, Image as HFImage, Value
from huggingface_hub import HfApi, DatasetCard, CommitOperationDelete
from PIL import Image
import os
import re
import shutil
for k in ["HTTP_PROXY", "HTTPS_PROXY", "http_proxy", "https_proxy", "ALL_PROXY", "all_proxy"]:
os.environ.pop(k, None)
# ================= 配置区 =================
SITE_CONFIG_FILE = "site_config.json"
SPACE_ID = "waveshaker/loras_gallery"
MODEL_REPO_ID = "waveshaker/loras"
DATASET_REPO_ID = "waveshaker/dataset_collections"
# 🌟 统一管理的基础 URL
SPACE_BASE = f"https://huggingface.co/{SPACE_ID}"
DATASET_BASE = f"https://huggingface.co/datasets/{DATASET_REPO_ID}/tree/main"
MODEL_BASE = f"https://huggingface.co/{MODEL_REPO_ID}"
GUIDE_BASE = f"https://huggingface.co/{MODEL_REPO_ID}/tree/main/usage_guide"
# 🌟 本地固定文件
OUTPUT_FILE = "models.json"
TOKEN_FILE = "hf_token.txt"
LOG_FILE = "models_manager.log"
# 🌟 核心目录名与后缀
SHOWCASE_DIR = "comfyui_showcase" # 案例上传的远端根目录
MODELS_LOCAL_DIR = "models_upload_folder" # 默认本地模型上传文件夹
TEMP_UPLOAD_JSON = "models.upload.json"
IMG_EXTS = {".jpg", ".jpeg", ".png", ".webp"}
HTTP_TIMEOUT = 20
# ==========================================
# 🌟 动态加载的排序规则
TOP_ORDER = {}
SUB_ORDER = {}
IMG_EXTS = {".jpg", ".jpeg", ".png", ".webp"}
def custom_sort_key(item_tuple):
file_path, meta = item_tuple
parts = file_path.split('/')
top_dir = parts[0] if len(parts) > 0 else ""
sub_dir = parts[1] if len(parts) > 1 else ""
ts = 0
if isinstance(meta, dict):
try:
ts = int(meta.get("update_time", 0))
except (ValueError, TypeError):
ts = 0
return (
TOP_ORDER.get(top_dir, 999),
SUB_ORDER.get(sub_dir, 999),
-ts,
file_path
)
def expand_urls(items_dict):
"""自动为精简版数据补齐完整的 HF 前缀（仅对不包含 http 的相对路径生效）"""
for meta in items_dict.values():
if not isinstance(meta, dict): continue
if meta.get("dataset"):
val = str(meta["dataset"])
if not val.startswith("http"):
meta["dataset"] = f"{DATASET_BASE}/{val}"
if meta.get("showcase"):
val = str(meta["showcase"])
if not val.startswith("http"):
meta["showcase"] = f"{MODEL_BASE}/tree/main/{SHOWCASE_DIR}/{val}"
if meta.get("guide"):
val = str(meta["guide"])
if not val.startswith("http"):
meta["guide"] = f"{GUIDE_BASE}/{val}"
# ---------- 可选拖拽 ----------
try:
from tkinterdnd2 import DND_FILES, TkinterDnD
_HAS_DND = True
except Exception:
DND_FILES = None
TkinterDnD = None
_HAS_DND = False
class ModelsManager(tk.Tk):
def __init__(self):
super().__init__()
self.title("LoRA 仓库管理工具")
self.geometry("1250x850")
# 全局 Token
self.token_var = tk.StringVar(value=self._load_token())
# 🌟 新增：保存 site_config 数据
self.site_config_data = {"models": [], "categories": []}
# JSON 相关状态
self.items = {}
self.remote_files = []
self.new_paths = set()
self.stale_paths = set()
self._search_job = None
self._loading = False
self.api = None
# 🌟 Model 上传相关状态：自动获取当前脚本路径，并指向同级的 models 文件夹
if getattr(sys, 'frozen', False):
current_dir = os.path.dirname(sys.executable)
else:
current_dir = os.path.dirname(os.path.abspath(__file__))
default_models_dir = os.path.join(current_dir, MODELS_LOCAL_DIR)
if not os.path.exists(default_models_dir):
os.makedirs(default_models_dir, exist_ok=True)
self.local_folder_var = tk.StringVar(value=default_models_dir)
self.remote_dir_var = tk.StringVar()
# 全局 Token
self.token_var = tk.StringVar(value=self._load_token())
# ============ 初始化 UI ============
self._build_global_ui()
self._build_json_tab()
self._build_model_tab()
self._build_dataset_tab() # <--- 新增这一行
self._build_showcase_tab() # 🌟 加上这一行
self._refresh_api()
self._setup_dnd()
self.log_json("程序启动")
# ==========================================================
# ===================== 全局基础函数 =======================
# ==========================================================
def log_json(self, msg):
self._log_to_tab(msg, getattr(self, 'log_json_text', None))
def log_model(self, msg):
self._log_to_tab(msg, getattr(self, 'log_model_text', None))
def _log_to_tab(self, msg, target_text_widget):
if not target_text_widget: return
ts = time.strftime("%H:%M:%S")
line = f"[{ts}] {msg}"
try:
with open(LOG_FILE, "a", encoding="utf-8") as f:
f.write(line + "\n")
except Exception: pass
def _append():
try:
target_text_widget.configure(state="normal")
target_text_widget.insert("end", line + "\n")
target_text_widget.see("end")
target_text_widget.configure(state="disabled")
except Exception: pass
if threading.current_thread() is threading.main_thread():
_append()
else:
self.after(0, _append)
def log_exc(self, prefix="", is_model=False):
tb = traceback.format_exc()
msg = f"{prefix}异常:\n{tb}"
if is_model:
self.log_model(msg)
else:
self.log_json(msg)
def _load_token(self):
env = os.environ.get("HF_TOKEN", "").strip()
if env: return env
try:
with open(TOKEN_FILE, "r", encoding="utf-8") as f:
return f.read().strip()
except FileNotFoundError:
return ""
def _save_token(self, token):
if token:
with open(TOKEN_FILE, "w", encoding="utf-8") as f: f.write(token)
else:
try: os.remove(TOKEN_FILE)
except FileNotFoundError: pass
def _refresh_api(self):
token = self.token_var.get().strip()
self.api = HfApi(token=token) if token else HfApi()
def _ensure_token(self):
token = self.token_var.get().strip()
if not token:
messagebox.showwarning("需要 Token", "请先在上方填写 HF Token 再操作。")
self.log_json("操作中止：未填写 Token")
return None
self._refresh_api()
return token
def _build_global_ui(self):
# 1. 顶部 Token 行
token_row = ttk.Frame(self)
token_row.pack(fill="x", padx=8, pady=(8, 2))
ttk.Label(token_row, text="HF Token:").pack(side="left")
ttk.Entry(token_row, textvariable=self.token_var, width=60, show="*").pack(side="left", padx=6)
ttk.Button(token_row, text="📥 保存 Token", command=self._on_save_token).pack(side="left")
ttk.Button(token_row, text="❌ 清除", command=self._on_clear_token).pack(side="left", padx=4)
ttk.Button(token_row, text="📄 打开日志", command=self.open_log_file).pack(side="right", padx=6)
# 1. 恢复 Windows 默认主题（看起来最舒服）
style = ttk.Style()
# 如果你在 Windows 11 上，可以用 'vista' 或 'winnative'，如果想要更现代可以试试 'xpnative'
if sys.platform.startswith("win"):
style.theme_use('vista')
# 2. 强制扁平化 Tab 的渲染结构（屏蔽掉系统的边框和虚线焦点）
style.layout("TNotebook.Tab", [
('Notebook.tab', {'sticky': 'nswe', 'children': [
('Notebook.padding', {'side': 'top', 'sticky': 'nswe', 'children': [
('Notebook.label', {'side': 'top', 'sticky': ''})
]})
]})
])
# 3. 基础设置
style.configure("TNotebook.Tab",
padding=[12, 6],
font=("Microsoft YaHei", 9),
borderwidth=0,
background="#f0f0f0",
foreground="#333333"
)
# 4. 锁定所有状态，防止任何像素级跳动
style.map("TNotebook.Tab",
padding=[("selected", [12, 6]), ("active", [12, 6]), ("!selected", [12, 6])],
borderwidth=[("selected", 0), ("active", 0), ("!selected", 0)],
# 焦点框设为空，去掉点击时的虚线框
focuscolor=[("selected", "none"), ("active", "none"), ("!selected", "none")],
# 选中和悬停时的颜色
background=[("selected", "#e1f5fe"), ("active", "#f5f5f5")],
foreground=[("selected", "#0277bd")]
)
# 3. 创建 Notebook 和 Tab 页面
self.notebook = ttk.Notebook(self)
self.notebook.pack(fill="both", expand=True, padx=8, pady=4)
self.tab_json = ttk.Frame(self.notebook)
self.tab_model = ttk.Frame(self.notebook)
self.tab_dataset = ttk.Frame(self.notebook)
# 4. 🌟 添加 Tab（Emoji + 文字）
self.notebook.add(self.tab_json, text=" 🚀 Space - JSON 管理 ")
self.notebook.add(self.tab_model, text=" 🧊 Models - 模型上传 ")
self.notebook.add(self.tab_dataset, text=" 📚 Dataset - 数据集上传 ")
self.tab_showcase = ttk.Frame(self.notebook)
self.notebook.add(self.tab_showcase, text=" ✨ Showcase - 案例上传 ")
def _on_save_token(self):
self._save_token(self.token_var.get().strip())
self._refresh_api()
self.log_json("Token 已保存")
messagebox.showinfo("完成", "Token 已保存到本地")
def _on_clear_token(self):
self.token_var.set("")
self._save_token("")
self._refresh_api()
self.log_json("Token 已清除")
messagebox.showinfo("完成", "Token 已清除")
def open_log_file(self):
try: os.startfile(LOG_FILE)
except Exception:
try:
import subprocess
subprocess.Popen(["notepad", LOG_FILE])
except Exception as e:
messagebox.showerror("错误", f"无法打开日志: {e}")
def _setup_dnd(self):
if not _HAS_DND: return
try:
TkinterDnD._require(self)
self.tree.drop_target_register(DND_FILES)
self.tree.dnd_bind("<<Drop>>", self._on_drop_json)
except Exception as e:
self.log_json(f"DnD 初始化失败: {e}")
# ==========================================================
# ==================== 区域 1: JSON 管理 ===================
# ==========================================================
def _build_json_tab(self):
parent = self.tab_json
top = ttk.Frame(parent)
top.pack(fill="x", padx=8, pady=4)
# 1. 拉取
self.btn_load_remote = ttk.Button(top, text="🔄 拉取 models.json", command=self.load_remote)
self.btn_load_remote.pack(side="left")
# 2. 上传
ttk.Button(top, text="🢁 上传 models.json", command=self.upload_models_json).pack(side="left", padx=6)
# 3. 管理全局 KV
ttk.Button(top, text="🔧 管理全局 KV", command=self.open_fields_manager).pack(side="left", padx=6)
# 🌟 新增：删除选中条目
ttk.Button(top, text="❌ 删除选中", command=self.delete_selected_item).pack(side="left", padx=6)
# 4. 加载本地
ttk.Button(top, text="📂 加载本地 models.json", command=self.load_local).pack(side="left", padx=6)
# 5. 保存到本地
ttk.Button(top, text="💾 保存 models.json 到本地", command=self.save_local).pack(side="left")
# 6. 🌟 新增：保存精简版到本地
ttk.Button(top, text="📦 保存精简版到本地", command=self.save_upload_format_local).pack(side="left", padx=6)
# 右侧的配置区保持现状即可
ttk.Button(top, text="🌐 打开 Dataset Repo", compound="left", command=self.open_dataset_repo).pack(side="right", padx=6)
ttk.Button(top, text="⚙️ 编辑 site_config", command=self.open_site_config_editor).pack(side="right", padx=12)
filt = ttk.Frame(parent)
filt.pack(fill="x", padx=8)
ttk.Label(filt, text="🔍 搜索:").pack(side="left")
self.search_var = tk.StringVar()
self.search_var.trace_add("write", lambda *_: self._debounced_refresh())
ttk.Entry(filt, textvariable=self.search_var, width=30).pack(side="left", padx=4)
self.only_new = tk.BooleanVar(value=False)
ttk.Checkbutton(filt, text="只看新增", variable=self.only_new, command=self.refresh_tree).pack(side="left", padx=6)
self.only_empty_trigger = tk.BooleanVar(value=False)
ttk.Checkbutton(filt, text="只看 trigger 为空", variable=self.only_empty_trigger, command=self.refresh_tree).pack(side="left")
body = ttk.Panedwindow(parent, orient="vertical")
body.pack(fill="both", expand=True, padx=8, pady=6)
tree_frame = ttk.Frame(body)
# 🌟 修改为：初始化时先塞入 4 个基础列，防止报错
base_cols = ("title", "trigger", "dataset", "showcase")
self.tree = ttk.Treeview(tree_frame, columns=base_cols, show="tree headings", height=20)
# 锁定前四列的基础宽度配置
self.tree.column("#0", width=380, stretch=True)
self.tree.column("title", width=180)
self.tree.column("trigger", width=260)
self.tree.column("dataset", width=240)
self.tree.column("showcase", width=200)
# 设置基础表头
self.tree.heading("#0", text="路径")
self.tree.heading("title", text="title")
self.tree.heading("trigger", text="trigger")
self.tree.heading("dataset", text="dataset")
self.tree.heading("showcase", text="showcase")
vsb = ttk.Scrollbar(tree_frame, orient="vertical", command=self.tree.yview)
self.tree.configure(yscrollcommand=vsb.set)
self.tree.pack(side="left", fill="both", expand=True)
vsb.pack(side="left", fill="y")
self.tree.tag_configure("new", background="#d4f5d4")
self.tree.tag_configure("empty_trigger", background="#fff3c4")
self.tree.tag_configure("stale", foreground="#999999")
self.tree.bind("<Double-1>", self.on_double_click)
log_frame = ttk.LabelFrame(body, text="JSON 操作日志")
self.log_json_text = ScrolledText(log_frame, height=10, state="disabled", wrap="word")
self.log_json_text.pack(fill="both", expand=True)
body.add(tree_frame, weight=3)
body.add(log_frame, weight=1)
def open_fields_manager(self):
"""打开全局字段管理器，用于统一添加/删除所有 JSON 的 Key"""
if not self.items:
messagebox.showwarning("提示", "当前没有加载任何数据，请先拉取或加载 models.json")
return
# 提取当前所有的 Key（也就是已有的列）
# 锁定的四大金刚
locked_keys = {"title", "trigger", "dataset", "showcase"}
# 获取所有条目中出现过的所有自定义 Key
all_keys = set()
for meta in self.items.values():
if isinstance(meta, dict):
all_keys.update(meta.keys())
custom_keys = list(all_keys - locked_keys)
win = tk.Toplevel(self)
win.title("管理全局字段 (全局生效)")
win.geometry("400x450")
win.transient(self); win.grab_set()
ttk.Label(win, text="💡 内置字段(锁定)：title, trigger, dataset, showcase", foreground="gray").pack(pady=10)
# 1. 列出当前已有的自定义字段
ttk.Label(win, text="当前自定义字段：").pack(anchor="w", padx=10)
listbox = tk.Listbox(win, height=8)
listbox.pack(fill="x", padx=10, pady=5)
for k in custom_keys:
listbox.insert("end", k)
# 2. 添加新字段的输入框
input_frame = ttk.Frame(win)
input_frame.pack(fill="x", padx=10, pady=10)
ttk.Label(input_frame, text="新字段名:").pack(side="left")
new_key_var = tk.StringVar()
ttk.Entry(input_frame, textvariable=new_key_var, width=20).pack(side="left", padx=6)
def add_key():
key = new_key_var.get().strip()
if not key:
return
if key in locked_keys:
messagebox.showerror("错误", "不能覆盖内置锁定字段！")
return
if key in custom_keys:
messagebox.showwarning("提示", "该字段已经存在！")
return
# 🌟 核心：遍历所有数据，强制添加该 key，默认值为 ""
for path, meta in self.items.items():
if isinstance(meta, dict):
meta[key] = ""
# 更新 UI
listbox.insert("end", key)
custom_keys.append(key)
new_key_var.set("")
self.log_json(f"✅ 全局添加了字段: {key}")
# 刷新主树形列表（虽然不显示，但保持内存同步）
self.refresh_tree()
ttk.Button(input_frame, text="➕ 全局添加", command=add_key).pack(side="left")
# 3. 删除逻辑
def del_key():
sel = listbox.curselection()
if not sel:
return
key = listbox.get(sel[0])
if not messagebox.askyesno("确认", f"确定要全局删除字段 [{key}] 吗？\n这将会从所有数据中彻底移除它。"):
return
# 🌟 核心：遍历所有数据，强制弹出该 key
for path, meta in self.items.items():
if isinstance(meta, dict):
meta.pop(key, None)
listbox.delete(sel[0])
custom_keys.remove(key)
self.log_json(f"🗑️ 全局删除了字段: {key}")
self.refresh_tree()
ttk.Button(win, text="❌ 删除选中字段", command=del_key).pack(pady=10)
ttk.Label(win, text="注意：操作会直接修改内存中的数据，完成后再去点【保存本地】", foreground="red").pack(pady=10)
def delete_selected_item(self):
"""从内存中删除选中的条目"""
sel = self.tree.focus()
if not sel or sel not in self.items:
messagebox.showwarning("提示", "请先在表格中选中要删除的条目！")
return
if not messagebox.askyesno("确认删除", f"确定要删除这条记录吗？\n\n{sel}\n\n注意：这只会从本地 JSON 中移除，不会删除云端实际的模型文件。"):
return
# 从内存中移除
self.items.pop(sel, None)
# 如果它恰好是新增的，也从 new_paths 里移除
if sel in self.new_paths:
self.new_paths.remove(sel)
self.log_json(f"🗑️ 已在本地删除条目: {sel}")
self.refresh_tree()
def open_dataset_repo(self):
import webbrowser
webbrowser.open(f"https://huggingface.co/datasets/{DATASET_REPO_ID}/tree/main")
def _debounced_refresh(self):
if self._search_job: self.after_cancel(self._search_job)
self._search_job = self.after(250, self.refresh_tree)
def load_local(self):
self.log_json(f"加载本地 {OUTPUT_FILE}")
try:
with open(OUTPUT_FILE, "r", encoding="utf-8") as f:
data = json.load(f)
except FileNotFoundError:
messagebox.showwarning("提示", f"本地没有 {OUTPUT_FILE}")
return
except Exception as e:
self.log_json(f"读取失败: {e}")
return
items = data.get("items", {})
if not items and "titles" in data:
items = {k: {"title": v, "trigger": "", "dataset": ""} for k, v in data["titles"].items()}
# 🌟 补齐前缀
expand_urls(items)
# 🌟 核心修改：强制按分类 + 时间戳倒序重排
sorted_items = dict(sorted(items.items(), key=custom_sort_key))
self.items = sorted_items
self.new_paths = set()
self.stale_paths = set()
self.refresh_tree()
self.log_json(f"本地加载完成，{len(self.items)} 条")
def load_remote(self):
if self._loading: return
self._loading = True
self.btn_load_remote.configure(state="disabled")
token = self.token_var.get().strip()
self.log_json(f"开始拉取远端…")
threading.Thread(target=self._load_remote_worker, args=(token,), daemon=True).start()
def _load_remote_worker(self, token):
global TOP_ORDER, SUB_ORDER
try:
api = HfApi(token=token) if token else HfApi()
# 🌟 核心新增：先拉取 site_config.json，动态生成排序规则
try:
self.log_json("正在拉取 site_config.json 以同步排序规则...")
config_path = api.hf_hub_download(repo_id=SPACE_ID, filename=SITE_CONFIG_FILE, repo_type="space")
with open(config_path, "r", encoding="utf-8") as f:
site_config = json.load(f)
# 根据数组顺序生成 { key: index } 的字典
top_list = site_config.get("models", [])
top_dict = {item["key"]: idx + 1 for idx, item in enumerate(top_list) if "key" in item}
sub_list = site_config.get("categories", [])
sub_dict = {item["key"]: idx + 1 for idx, item in enumerate(sub_list) if "key" in item}
if top_dict:
TOP_ORDER = top_dict
if sub_dict:
SUB_ORDER = sub_dict
self.log_json(f"✅ 排序规则已更新: Models有{len(top_dict)}个, Categories有{len(sub_dict)}个")
except Exception as e:
self.log_json(f"⚠️ 拉取 site_config 失败，将使用默认空排序: {e}")
t0 = time.time()
all_files = list(api.list_repo_files(repo_id=MODEL_REPO_ID, repo_type="model"))
remote_files = [f for f in all_files if f.endswith(".safetensors")]
self.log_json(f"步骤1: 远端文件 {len(all_files)}，safetensors {len(remote_files)}，{time.time()-t0:.2f}s")
space_subdomain = SPACE_ID.replace("/", "-").replace("_", "-").lower()
url = f"https://{space_subdomain}.static.hf.space/models.json"
remote_data = None
try:
req = urllib.request.Request(url, headers={'User-Agent': 'Mozilla/5.0'})
with urllib.request.urlopen(req, timeout=HTTP_TIMEOUT) as response:
remote_data = json.loads(response.read().decode())
self.log_json(f"步骤2: JSON 获取成功")
except Exception as e:
self.log_json(f"步骤2 失败: {type(e).__name__}: {e}")
existing_items = {}
if remote_data:
existing_items = remote_data.get("items", {})
if not existing_items and "titles" in remote_data:
existing_items = {k: {"title": v, "trigger": "", "dataset": ""} for k, v in remote_data["titles"].items()}
merged, new_paths = {}, set()
remote_set = set(remote_files)
for path in remote_files:
if path in existing_items:
merged[path] = existing_items[path]
else:
merged[path] = {"title": "", "trigger": "", "dataset": ""}
new_paths.add(path)
stale_paths = set(existing_items.keys()) - remote_set
for p in stale_paths:
merged[p] = existing_items[p]
items = dict(sorted(merged.items(), key=custom_sort_key))
self.log_json(f"步骤3: 合并 {len(items)} 条，新增 {len(new_paths)}，失效 {len(stale_paths)}")
self.after(0, lambda: self._apply_remote(items, remote_files, new_paths, stale_paths))
except Exception:
self.log_exc("拉取失败")
self.after(0, lambda: messagebox.showerror("错误", "拉取失败，详见日志"))
finally:
self.after(0, self._finish_loading)
# def _apply_remote(self, items, remote_files, new_paths, stale_paths):
# # 补齐前缀
# expand_urls(items)
# self.items = items
# self.remote_files = remote_files
# self.new_paths = new_paths
# self.stale_paths = stale_paths
# self.refresh_tree()
def _apply_remote(self, items, remote_files, new_paths, stale_paths):
# 补齐前缀
expand_urls(items)
# 🌟 核心修改：强制按分类 + 时间戳倒序重排
sorted_items = dict(sorted(items.items(), key=custom_sort_key))
self.items = sorted_items
self.remote_files = remote_files
self.new_paths = new_paths
self.stale_paths = stale_paths
self.refresh_tree()
def _finish_loading(self):
self._loading = False
self.btn_load_remote.configure(state="normal")
def refresh_tree(self):
self.tree.delete(*self.tree.get_children())
kw = self.search_var.get().strip().lower()
only_new = self.only_new.get()
only_empty = self.only_empty_trigger.get()
# 🌟 核心：动态收集所有出现过的 key（除了前四个）
base_cols = ["title", "trigger", "dataset", "showcase"]
dynamic_cols = set()
for meta in self.items.values():
if isinstance(meta, dict):
for k in meta.keys():
if k not in base_cols:
dynamic_cols.add(k)
# 排序以确保列的顺序稳定
dynamic_cols = sorted(list(dynamic_cols))
all_cols = base_cols + dynamic_cols
# 1. 动态设置表头和列宽
current_cols = self.tree["columns"]
if list(current_cols) != all_cols:
self.tree["columns"] = all_cols
for col in all_cols:
self.tree.heading(col, text=col)
if col == "title": self.tree.column(col, width=180)
elif col == "trigger": self.tree.column(col, width=260)
elif col == "dataset": self.tree.column(col, width=240)
elif col == "showcase": self.tree.column(col, width=200)
else: self.tree.column(col, width=100) # 自定义列默认稍微窄一点
# 2. 插入数据
for path, meta in self.items.items():
title = meta.get("title", "")
trigger = meta.get("trigger", "")
if kw and kw not in path.lower() and kw not in str(title).lower(): continue
if only_new and path not in self.new_paths: continue
if only_empty and trigger not in ("", [], None): continue
tags = []
if path in self.new_paths: tags.append("new")
elif path in self.stale_paths: tags.append("stale")
if trigger in ("", [], None): tags.append("empty_trigger")
trig_display = trigger if isinstance(trigger, str) else " | ".join(map(str, trigger))
# 动态拼装每一列的值
row_values = [
title,
trig_display,
meta.get("dataset", ""),
meta.get("showcase", "")
]
for col in dynamic_cols:
val = meta.get(col, "")
# 🌟 新增：如果列名是 update_time，把毫秒时间戳转成可读日期展示
if col == "update_time" and val:
try:
# 转换成本地时间的 YYYY-MM-DD 格式
import datetime
dt = datetime.datetime.fromtimestamp(int(val) / 1000.0)
val = dt.strftime('%Y-%m-%d')
except Exception:
pass
row_values.append(val)
self.tree.insert("", "end", iid=path, text=path, values=row_values, tags=tags)
def on_double_click(self, event):
sel = self.tree.focus()
if sel and sel in self.items:
self.open_manager(sel)
def open_manager(self, path):
meta = self.items[path]
win = tk.Toplevel(self)
win.title(f"编辑: {path}")
win.geometry("720x560")
win.transient(self); win.grab_set()
ttk.Label(win, text="路径").pack(anchor="w", padx=10, pady=(10, 0))
p_entry = ttk.Entry(win, width=90)
p_entry.insert(0, path); p_entry.configure(state="readonly"); p_entry.pack(padx=10, fill="x")
ttk.Label(win, text="title").pack(anchor="w", padx=10, pady=(8, 0))
title_var = tk.StringVar(value=meta.get("title", ""))
ttk.Entry(win, textvariable=title_var, width=90).pack(padx=10, fill="x")
ttk.Label(win, text="trigger").pack(anchor="w", padx=10, pady=(8, 0))
trig_text = tk.Text(win, height=8, wrap="word"); trig_text.pack(padx=10, fill="both")
trig = meta.get("trigger", "")
trig_text.insert("1.0", "\n".join(map(str, trig)) if isinstance(trig, list) else trig)
ttk.Label(win, text="dataset").pack(anchor="w", padx=10, pady=(8, 0))
ds_var = tk.StringVar(value=meta.get("dataset", ""))
ttk.Entry(win, textvariable=ds_var, width=90).pack(padx=10, fill="x")
# 🌟 新增：showcase 输入框
ttk.Label(win, text="showcase (URL)").pack(anchor="w", padx=10, pady=(8, 0))
sc_var = tk.StringVar(value=meta.get("showcase", ""))
ttk.Entry(win, textvariable=sc_var, width=90).pack(padx=10, fill="x")
def auto_dataset():
parts = path.split("/")
if len(parts) >= 3:
top, sub = parts[0], parts[1]
base = Path(parts[2]).stem
for noise in ("_onetrainer", "_Flux_512", "_Flux", "_512"):
base = base.replace(noise, "")
ds_var.set(f"{DATASET_BASE}/{top}/{sub}/{base}")
ttk.Button(win, text="自动生成 dataset", command=auto_dataset).pack(anchor="w", padx=10, pady=4)
# ========================================================
# 🌟 动态渲染自定义字段（兼容空数据）
# ========================================================
custom_vars = {}
locked_keys = ["title", "trigger", "dataset", "showcase"]
# 1. 当前数据自身的 key
current_custom_keys = set([k for k in meta.keys() if k not in locked_keys])
# 2. 全局所有数据里出现过的 key
global_custom_keys = set()
for m in self.items.values():
if isinstance(m, dict):
global_custom_keys.update([k for k in m.keys() if k not in locked_keys])
# 3. 合并去重
all_custom_keys = list(current_custom_keys) + [k for k in global_custom_keys if k not in current_custom_keys]
if all_custom_keys:
ext_frame = ttk.LabelFrame(win, text="自定义扩展字段")
ext_frame.pack(fill="x", padx=10, pady=5)
for k in all_custom_keys:
row = ttk.Frame(ext_frame)
row.pack(fill="x", pady=2)
ttk.Label(row, text=f"{k}:", width=12).pack(side="left")
var = tk.StringVar(value=str(meta.get(k, "")))
custom_vars[k] = var
ttk.Entry(row, textvariable=var).pack(side="left", fill="x", expand=True, padx=5)
def save_and_close():
title = title_var.get().strip()
raw = trig_text.get("1.0", "end").strip()
trigger_val = ([l.strip() for l in raw.splitlines() if l.strip()] if "\n" in raw else raw)
# 保存基础字段
new_meta = {
"title": title,
"trigger": trigger_val,
"dataset": ds_var.get().strip(),
"showcase": sc_var.get().strip()
}
# 保存所有自定义字段
for k, var in custom_vars.items():
new_meta[k] = var.get().strip()
self.items[path] = new_meta
self.refresh_tree()
win.destroy()
btn = ttk.Frame(win); btn.pack(fill="x", pady=10, padx=10)
ttk.Button(btn, text="保存", command=save_and_close).pack(side="right")
ttk.Button(btn, text="取消", command=win.destroy).pack(side="right", padx=6)
def save_local(self):
data_items = {k: v for k, v in self.items.items() if k not in self.stale_paths}
sorted_items = dict(sorted(data_items.items(), key=custom_sort_key))
with open(OUTPUT_FILE, "w", encoding="utf-8") as f:
json.dump({"items": sorted_items}, f, indent=2, ensure_ascii=False)
self.log_json(f"已保存本地 {OUTPUT_FILE}，{len(sorted_items)} 条")
messagebox.showinfo("完成", f"已保存到 {OUTPUT_FILE}")
def upload_models_json(self):
if not self._ensure_token(): return
if not messagebox.askyesno("确认", f"上传 models.json 到 {SPACE_ID}？"): return
import copy
data_items = {k: v for k, v in self.items.items() if k not in self.stale_paths}
sorted_items = dict(sorted(data_items.items(), key=custom_sort_key))
# 🌟 核心修改：深拷贝一份用于上传
upload_items = copy.deepcopy(sorted_items)
# 🌟 精准时间戳策略：只为从未有过时间戳的新文件打上当前时间，老文件保留原本时间
import time
current_ts = int(time.time() * 1000)
for path, meta in upload_items.items():
if not isinstance(meta, dict): continue
# 只给全新文件（没有 update_time）打时间戳，绝不覆盖历史时间
if not meta.get("update_time"):
meta["update_time"] = current_ts
# 剥离前缀
if meta.get("dataset") and meta["dataset"].startswith(DATASET_BASE):
stripped = meta["dataset"][len(DATASET_BASE):].lstrip("/")
if stripped: meta["dataset"] = stripped
if meta.get("showcase"):
prefix = f"{MODEL_BASE}/tree/main/{SHOWCASE_DIR}"
if meta["showcase"].startswith(prefix):
stripped = meta["showcase"][len(prefix):].lstrip("/")
if stripped: meta["showcase"] = stripped
if meta.get("guide"):
meta["guide"] = re.sub(r'^https?://huggingface\.co/waveshaker/loras/(?:blob|tree)/main/usage_guide/?', '', meta["guide"]).strip('/')
tmp = TEMP_UPLOAD_JSON
with open(tmp, "w", encoding="utf-8") as f:
json.dump({"items": upload_items}, f, indent=2, ensure_ascii=False)
self.log_json(f"开始上传精简版 JSON ({len(upload_items)} 条)")
api = self.api
def work():
try:
api.upload_file(path_or_fileobj=tmp, path_in_repo=OUTPUT_FILE, repo_id=SPACE_ID, repo_type="space")
self.log_json("精简版 models.json 上传成功")
self.after(0, lambda: messagebox.showinfo("完成", "上传成功"))
except Exception:
self.log_exc("JSON 上传失败")
self.after(0, lambda: messagebox.showerror("失败", "详见日志"))
finally:
try: os.remove(tmp)
except OSError: pass
threading.Thread(target=work, daemon=True).start()
# ==========================================================
# ================= 新增：site_config 管理 =================
# ==========================================================
def open_site_config_editor(self):
if not self._ensure_token(): return
self.log_json("正在从 Space 拉取 site_config.json...")
def fetch_worker():
try:
# 直接从 Space 文件仓库下载
file_path = self.api.hf_hub_download(
repo_id=SPACE_ID,
filename=SITE_CONFIG_FILE,
repo_type="space"
)
with open(file_path, "r", encoding="utf-8") as f:
config_data = json.load(f)
self.after(0, lambda: self._show_config_window(config_data))
except Exception as e:
self.log_json(f"拉取 site_config 失败: {e}")
# 如果远端没有，就初始化一个空的
self.after(0, lambda: self._show_config_window({"models": [], "categories": []}))
threading.Thread(target=fetch_worker, daemon=True).start()
def _show_config_window(self, config_data):
win = tk.Toplevel(self)
win.title("编辑 site_config.json (保存后自动上传)")
win.geometry("600x550")
win.transient(self); win.grab_set()
ttk.Label(win, text="💡 提示：修改顺序即可改变前端左侧导航栏的展示顺序。", foreground="gray").pack(pady=5)
# 分成两个框：Models 和 Categories
frame_models = ttk.LabelFrame(win, text="Base Models (按顺序)")
frame_models.pack(fill="both", expand=True, padx=10, pady=5)
text_models = tk.Text(frame_models, height=10, wrap="none")
text_models.pack(fill="both", expand=True, padx=5, pady=5)
# 将 JSON 格式化为易读的字符串填入
text_models.insert("1.0", json.dumps(config_data.get("models", []), indent=2, ensure_ascii=False))
frame_cats = ttk.LabelFrame(win, text="Categories (按顺序)")
frame_cats.pack(fill="both", expand=True, padx=10, pady=5)
text_cats = tk.Text(frame_cats, height=10, wrap="none")
text_cats.pack(fill="both", expand=True, padx=5, pady=5)
text_cats.insert("1.0", json.dumps(config_data.get("categories", []), indent=2, ensure_ascii=False))
def save_and_upload():
try:
# 解析用户修改后的文本
new_models = json.loads(text_models.get("1.0", "end").strip())
new_cats = json.loads(text_cats.get("1.0", "end").strip())
new_config = {
"models": new_models,
"categories": new_cats
}
# 保存到本地临时文件
with open(SITE_CONFIG_FILE, "w", encoding="utf-8") as f:
json.dump(new_config, f, indent=2, ensure_ascii=False)
self.log_json("正在上传 site_config.json 到 Space...")
self.api.upload_file(
path_or_fileobj=SITE_CONFIG_FILE,
path_in_repo=SITE_CONFIG_FILE,
repo_id=SPACE_ID,
repo_type="space"
)
self.log_json("🎉 site_config.json 上传成功！")
messagebox.showinfo("成功", "配置已更新并上传到 Space！")
win.destroy()
except json.JSONDecodeError as e:
messagebox.showerror("JSON 格式错误", f"请检查括号、逗号是否漏了。\n报错信息: {e}")
except Exception as e:
self.log_json(f"上传失败: {e}")
messagebox.showerror("错误", f"上传失败: {e}")
btn_frame = ttk.Frame(win)
btn_frame.pack(fill="x", pady=10, padx=10)
ttk.Button(btn_frame, text="保存并上传", command=save_and_upload).pack(side="right")
ttk.Button(btn_frame, text="取消", command=win.destroy).pack(side="right", padx=6)
def _on_drop_json(self, event):
raw = event.data.strip("{}")
p = Path(raw)
if p.is_dir():
self.notebook.select(self.tab_model)
self.local_folder_var.set(str(p))
def save_upload_format_local(self):
"""剥离绝对前缀，生成精简版，绝不污染内存"""
import copy
data_items = {k: v for k, v in self.items.items() if k not in self.stale_paths}
sorted_items = dict(sorted(data_items.items(), key=custom_sort_key))
# 🌟 核心修改：深拷贝一份，只对拷贝体动手
upload_items = copy.deepcopy(sorted_items)
for path, meta in upload_items.items():
if not isinstance(meta, dict): continue
if meta.get("dataset"):
meta["dataset"] = meta["dataset"].replace(f"{DATASET_BASE}/", "").replace(DATASET_BASE, "")
if meta.get("showcase"):
prefix = f"{MODEL_BASE}/tree/main/{SHOWCASE_DIR}/"
meta["showcase"] = meta["showcase"].replace(prefix, "")
if meta.get("guide"):
meta["guide"] = re.sub(r'^https?://huggingface\.co/waveshaker/loras/(?:blob|tree)/main/usage_guide/?', '', meta["guide"]).strip('/')
out_path = "models.upload_preview.json"
with open(out_path, "w", encoding="utf-8") as f:
json.dump({"items": upload_items}, f, indent=2, ensure_ascii=False)
self.log_json(f"已保存精简版到本地 {out_path}，共 {len(upload_items)} 条")
messagebox.showinfo("完成", f"精简版已保存到 {out_path}")
# ==========================================================
# ==================== 区域 2: 模型上传 ====================
# ==========================================================
def _build_model_tab(self):
parent = self.tab_model
# 1. 输入区
frame = ttk.Frame(parent)
frame.pack(fill="x", padx=10, pady=10)
# 本地文件夹
ttk.Label(frame, text="本地文件夹:").grid(row=0, column=0, sticky="w", pady=4)
ttk.Entry(frame, textvariable=self.local_folder_var, width=70).grid(row=0, column=1, padx=6, sticky="we")
ttk.Button(frame, text="📁 浏览...", width=18, command=self._browse_model_local).grid(row=0, column=2)
# 远端目录 (Treeview 改造)
ttk.Label(frame, text="上传到 HF 目录:").grid(row=1, column=0, sticky="nw", pady=4)
tree_frame = ttk.Frame(frame)
tree_frame.grid(row=1, column=1, padx=6, sticky="nsew")
# 滚动条
vsb = ttk.Scrollbar(tree_frame, orient="vertical")
vsb.pack(side="right", fill="y")
# 树形列表
self.remote_tree = ttk.Treeview(tree_frame, height=6, show="tree",
yscrollcommand=vsb.set, selectmode="browse")
self.remote_tree.pack(side="left", fill="both", expand=True)
vsb.config(command=self.remote_tree.yview)
# 绑定树形点击和双击事件
self.remote_tree.bind("<<TreeviewSelect>>", self._on_remote_tree_select)
# self.remote_tree.bind("<Double-1>", lambda e: self._start_model_upload())
# 刷新按钮
ttk.Button(frame, text="🔄 拉取 Models 目录", width=18, command=self._fetch_hf_dirs).grid(row=1, column=2, sticky="n")
# 按钮区
btn_row = ttk.Frame(frame)
btn_row.grid(row=2, column=0, columnspan=3, pady=15, sticky="w")
# 🌟 开始上传按钮 (绑定 self.btn_upload，方便后面禁用/恢复)
self.btn_upload = ttk.Button(btn_row, text="🢁 开始上传", command=self._start_model_upload)
self.btn_upload.pack(side="left")
# 状态文字
self.status_model_var = tk.StringVar(value="就绪")
ttk.Label(btn_row, textvariable=self.status_model_var).pack(side="left", padx=10)
# 🌟 进度条
self.progress_var = tk.DoubleVar()
self.progress_bar = ttk.Progressbar(btn_row, variable=self.progress_var, maximum=100, length=200, mode='determinate')
self.progress_bar.pack(side="left", padx=10)
frame.columnconfigure(1, weight=1)
# 让 Treeview 高度自适应
frame.rowconfigure(1, weight=1)
# 2. 日志区
log_frame = ttk.LabelFrame(parent, text="模型上传日志")
log_frame.pack(fill="both", expand=True, padx=10, pady=6)
self.log_model_text = ScrolledText(log_frame, height=10, state="disabled", wrap="word")
self.log_model_text.pack(fill="both", expand=True)
def _browse_model_local(self):
# 🌟 强制从当前脚本所在目录开始（即 ws_huggingface 这一层）
if getattr(sys, 'frozen', False):
current_dir = os.path.dirname(sys.executable)
else:
current_dir = os.path.dirname(os.path.abspath(__file__))
folder = filedialog.askdirectory(title="选择包含 .safetensors 的文件夹", initialdir=current_dir)
if folder:
self.local_folder_var.set(folder)
def _fetch_hf_dirs(self):
if not self._ensure_token(): return
self.status_model_var.set("正在读取远端目录...")
self.log_model("正在读取远端仓库目录列表...")
def work():
try:
api = self.api
all_files = list(api.list_repo_files(repo_id=MODEL_REPO_ID, repo_type="model"))
dirs = set()
for f in all_files:
# 🌟 核心修改：如果是被过滤的目录下的文件，直接跳过
if f.startswith(f"{SHOWCASE_DIR}/") or f.startswith("preview/"):
continue
d = str(Path(f).parent)
if d != ".":
dirs.add(d)
sorted_dirs = sorted(list(dirs))
self.after(0, lambda: self._update_remote_tree(sorted_dirs))
except Exception:
self.log_exc("读取远端目录失败", is_model=True)
finally:
self.after(0, lambda: self.status_model_var.set("就绪"))
threading.Thread(target=work, daemon=True).start()
def _update_remote_tree(self, dirs):
# 清空旧的树
for item in self.remote_tree.get_children():
self.remote_tree.delete(item)
tree_nodes = {} # 用于记录路径对应的 treeview item id
for d in dirs:
# 把 Windows 反斜杠统一替换成正斜杠，方便拆分
safe_d = d.replace("\\", "/")
parts = safe_d.split("/")
parent_id = ""
current_path = ""
for i, part in enumerate(parts):
current_path = f"{current_path}/{part}" if current_path else part
# 如果这个节点还没创建，就创建它
if current_path not in tree_nodes:
# 判断是不是最后一级目录（如果是最后一级，加上图标和展开符）
is_leaf = (i == len(parts) - 1)
display_text = f"📁 {part}" if is_leaf else part
node_id = self.remote_tree.insert(
parent_id, "end",
text=display_text,
values=(current_path,) # 把真实路径藏在 values 里
)
tree_nodes[current_path] = node_id
# 下一级的父节点变成了当前节点
parent_id = tree_nodes[current_path]
# 默认展开所有根节点
for root_node in self.remote_tree.get_children(""):
self.remote_tree.item(root_node, open=True)
self.log_model(f"已加载 {len(dirs)} 个远端目录结构")
def _on_remote_tree_select(self, event):
# 当用户在树上点击时，提取路径存到 self.remote_dir_var
selection = self.remote_tree.selection()
if selection:
item_id = selection[0]
# 从 values 里拿我们存的真实路径
values = self.remote_tree.item(item_id, "values")
if values:
self.remote_dir_var.set(values[0])
def _start_model_upload(self):
local_dir = self.local_folder_var.get().strip()
# 如果用户没点选树，默认上传到根目录；如果点了，就是选中的路径
remote_dir = self.remote_dir_var.get().strip().strip("/")
if not local_dir or not os.path.exists(local_dir):
messagebox.showwarning("错误", "请选择有效的本地文件夹。")
return
if not remote_dir:
if not messagebox.askyesno("确认", "远端目录为空，将上传到仓库根目录，继续？"):
return
if not self._ensure_token(): return
self.status_model_var.set("正在扫描并清理预览图...")
self.log_model(f"开始扫描: {local_dir} -> {MODEL_REPO_ID}/{remote_dir}")
def scan_worker():
try:
api = self.api
# 1. 清洗本地图片：全部转为纯净的 WebP 并剥离元数据
local_files = [f for f in Path(local_dir).iterdir() if f.is_file()]
for f in local_files:
# 拦截 jpg/jpeg/png/webp 四种格式
if f.suffix.lower() in [".jpg", ".jpeg", ".png", ".webp"]:
webp_path = f.with_suffix(".webp")
# 如果当前已经是 .webp，且没有别的修改，只重新保存一遍剥离数据
if f.suffix.lower() == ".webp" and f == webp_path:
pass # 继续走下面的覆盖流程
try:
self.log_model(f"清理并转换预览图: {f.name} -> {webp_path.name}")
with Image.open(f) as img:
if img.mode in ("RGBA", "P"):
img = img.convert("RGB")
# 核心：不传 exif，彻底剥离元数据，quality 设为 85 足够清晰且体积小
img.save(webp_path, format="WEBP", quality=85, method=6)
# 如果原图不是 webp (说明后缀不一致，是 jpg/png 等)，删掉原图
if f.suffix.lower() != ".webp" or f != webp_path:
os.remove(f)
except Exception as e:
self.log_model(f"转码失败 {f.name}: {e}")
# 2. 重新读取清洗后的文件列表
local_files = [f for f in Path(local_dir).iterdir() if f.is_file()]
if not local_files:
self.after(0, lambda: messagebox.showwarning("提示", "本地文件夹是空的。"))
return
# 3. 拉取远端文件用于比对
remote_all = set(api.list_repo_files(repo_id=MODEL_REPO_ID, repo_type="model"))
# 4. 过滤待上传文件（只允许 safetensors 和 webp）
to_upload = []
for f in local_files:
if f.suffix.lower() not in [".safetensors", ".webp"]:
continue
remote_path = f"{remote_dir}/{f.name}" if remote_dir else f.name
if remote_path not in remote_all:
to_upload.append((f, remote_path))
if not to_upload:
self.after(0, lambda: messagebox.showinfo("完成", "远端已存在所有文件，无需上传。"))
self.after(0, lambda: self.status_model_var.set("无需上传"))
self.log_model("扫描完毕，无新增文件。")
return
msg = f"预览图清理完毕。找到 {len(to_upload)} 个新文件待上传。\n\n目标目录: {remote_dir}\n\n继续？"
self.after(0, lambda: self._do_model_upload_confirm(msg, to_upload, api))
except Exception:
self.log_exc("扫描失败", is_model=True)
self.after(0, lambda: self.status_model_var.set("扫描失败"))
threading.Thread(target=scan_worker, daemon=True).start()
def _do_model_upload_confirm(self, msg, to_upload, api):
if not messagebox.askyesno("确认上传", msg):
self.status_model_var.set("已取消")
return
def upload_worker():
try:
from huggingface_hub import CommitOperationAdd
self.after(0, lambda: self.btn_upload.configure(state="disabled"))
self.after(0, lambda: self.progress_var.set(0))
self.after(0, lambda: self.status_model_var.set("正在上传模型文件..."))
# ========== 步骤 1：上传物理文件 ==========
operations = []
for local_file, remote_path in to_upload:
safe_remote_path = remote_path.replace("\\", "/")
operations.append(
CommitOperationAdd(
path_or_fileobj=str(local_file),
path_in_repo=safe_remote_path
)
)
remote_dir = self.remote_dir_var.get().strip().replace("\\", "/").strip("/")
commit_msg = f"Upload files to {remote_dir}" if remote_dir else "Upload files to root"
is_uploading = [True]
def update_progress():
if not is_uploading[0]: return
current = self.progress_var.get()
if current < 95:
self.progress_var.set(current + 2)
self.after(300, update_progress)
self.after(100, update_progress)
self.log_model(f"开始打包 {len(operations)} 个文件，一次性 Commit...")
api.create_commit(
repo_id=MODEL_REPO_ID,
repo_type="model",
operations=operations,
commit_message=commit_msg
)
self.log_model("模型文件上传成功！正在同步 models.json...")
# ========== 步骤 2：拉取云端最新 JSON ==========
self.after(0, lambda: self.status_model_var.set("正在拉取云端最新 JSON..."))
space_subdomain = SPACE_ID.replace("/", "-").replace("_", "-").lower()
url = f"https://{space_subdomain}.static.hf.space/models.json"
latest_remote_items = {}
try:
req = urllib.request.Request(url, headers={'User-Agent': 'Mozilla/5.0'})
with urllib.request.urlopen(req, timeout=HTTP_TIMEOUT) as response:
remote_data = json.loads(response.read().decode())
latest_remote_items = remote_data.get("items", {})
self.log_model(f"成功拉取云端 JSON，包含 {len(latest_remote_items)} 条记录。")
except Exception as e:
self.log_model(f"拉取云端 JSON 失败，放弃本次同步以防止覆盖: {e}")
raise e # 如果拉取失败，直接中断，绝不用空数据覆盖！
# ========== 步骤 3：精准追加新条目 ==========
import time
current_ts = int(time.time() * 1000)
added_count = 0
for local_file, remote_path in to_upload:
safe_remote_path = remote_path.replace("\\", "/")
# 🚨 绝对严格过滤：只有 .safetensors 才配进入 models.json
if not safe_remote_path.endswith(".safetensors"):
continue
# 如果云端已经有这个模型了，更新它的时间；如果没有，则新增
if safe_remote_path in latest_remote_items:
latest_remote_items[safe_remote_path]["update_time"] = current_ts
else:
latest_remote_items[safe_remote_path] = {
"title": "", "trigger": "", "dataset": "", "showcase": "", "update_time": current_ts
}
added_count += 1
self.log_model(f"已为 {added_count} 个新增模型写入时间戳。")
# 按原有逻辑重新排序
# sorted_items = {k: latest_remote_items[k] for k in sorted(latest_remote_items.keys(), key=custom_sort_key)}
sorted_items = dict(sorted(latest_remote_items.items(), key=custom_sort_key))
# ========== 步骤 4：剥离前缀并推送 ==========
if added_count > 0:
self.after(0, lambda: self.status_model_var.set("正在推送到云端..."))
# 补全 URL 后存入本地（保持本地数据完整）
self.items = sorted_items
expand_urls(self.items)
self.save_local_silent()
self.after(0, self.refresh_tree)
# 深拷贝用于剥离上传
import copy
upload_items = copy.deepcopy(sorted_items)
for p, meta in upload_items.items():
if not isinstance(meta, dict): continue
if meta.get("dataset") and meta["dataset"].startswith(DATASET_BASE):
stripped = meta["dataset"][len(DATASET_BASE):].lstrip("/")
if stripped: meta["dataset"] = stripped
if meta.get("showcase"):
prefix = f"{MODEL_BASE}/tree/main/{SHOWCASE_DIR}"
if meta["showcase"].startswith(prefix):
stripped = meta["showcase"][len(prefix):].lstrip("/")
if stripped: meta["showcase"] = stripped
if meta.get("guide"):
meta["guide"] = re.sub(r'^https?://huggingface\.co/waveshaker/loras/(?:blob|tree)/main/usage_guide/?', '', meta["guide"]).strip('/')
tmp_json = TEMP_UPLOAD_JSON
with open(tmp_json, "w", encoding="utf-8") as f:
json.dump({"items": upload_items}, f, indent=2, ensure_ascii=False)
api.upload_file(
path_or_fileobj=tmp_json,
path_in_repo=OUTPUT_FILE,
repo_id=SPACE_ID,
repo_type="space",
commit_message=f"Update JSON for new model(s)"
)
try: os.remove(tmp_json)
except OSError: pass
self.log_model("models.json 已同步更新到云端！")
# ========== 步骤 5：收尾 ==========
is_uploading[0] = False
self.after(0, lambda: self.progress_var.set(100))
self.after(0, lambda: self.status_model_var.set("上传完成 ✅"))
self.after(0, lambda: self.btn_upload.configure(state="normal"))
self.after(0, lambda: messagebox.showinfo("完成", "模型与 JSON 均已安全同步！"))
except Exception as e:
is_uploading[0] = False
self.after(0, lambda: self.progress_var.set(0))
self.after(0, lambda: self.btn_upload.configure(state="normal"))
self.after(0, lambda: self.status_model_var.set("执行失败 ❌"))
self.log_exc("上传或同步失败", is_model=True)
self.after(0, lambda: messagebox.showerror("失败", "详见日志"))
threading.Thread(target=upload_worker, daemon=True).start()
def save_local_silent(self):
"""静默保存本地 JSON，不弹出提示框"""
data_items = {k: v for k, v in self.items.items() if k not in self.stale_paths}
sorted_items = dict(sorted(data_items.items(), key=custom_sort_key))
with open(OUTPUT_FILE, "w", encoding="utf-8") as f:
json.dump({"items": sorted_items}, f, indent=2, ensure_ascii=False)
self.log_json(f"已静默保存本地 {OUTPUT_FILE}，{len(sorted_items)} 条")
# ==========================================================
# ==================== 区域 3: Dataset 上传 =================
# ==========================================================
def _build_dataset_tab(self):
parent = self.tab_dataset
frame = ttk.Frame(parent)
frame.pack(fill="x", padx=10, pady=10)
# 1. 本地母文件夹
ttk.Label(frame, text="本地母文件夹:").grid(row=0, column=0, sticky="w", pady=4)
self.ds_local_folder_var = tk.StringVar()
ttk.Entry(frame, textvariable=self.ds_local_folder_var, width=70).grid(row=0, column=1, columnspan=2, padx=6, sticky="we")
ttk.Button(frame, text="📁 浏览...", width=18, command=self._browse_ds_local).grid(row=0, column=3)
# 2. 地区/分类前缀 (动态读取改造)
ttk.Label(frame, text="Region 前缀:").grid(row=1, column=0, sticky="w", pady=8)
prefix_frame = ttk.Frame(frame)
prefix_frame.grid(row=1, column=1, columnspan=3, sticky="w", padx=6, pady=8)
# 核心变量初始化
self.ds_region_var = tk.StringVar(value="KR")
self.ds_custom_region_var = tk.StringVar() # 必须保留这个变量用于输入框
# 按钮：刷新 Dataset 目录
ttk.Button(prefix_frame, text="🔄 拉取 datasets 目录", command=self._fetch_dataset_dirs).pack(side="left", padx=(0, 10))
# 用子 Frame 专门放单选框，刷新时只清空这里的控件
self.ds_radio_container = ttk.Frame(prefix_frame)
self.ds_radio_container.pack(side="left")
# 初始默认放几个，防止还没拉取时界面太难看
self._render_dataset_radios(["CN", "KR", "JP", "Fictional", "WE", "Heels"])
# 3. 按钮区
btn_row = ttk.Frame(frame)
btn_row.grid(row=2, column=0, columnspan=4, pady=15, sticky="w")
self.btn_ds_upload = ttk.Button(btn_row, text="🢁 开始 5合1 处理并上传", command=self._start_dataset_pipeline)
self.btn_ds_upload.pack(side="left")
self.status_ds_var = tk.StringVar(value="就绪")
ttk.Label(btn_row, textvariable=self.status_ds_var).pack(side="left", padx=10)
# 🌟 新增：Dataset 链接展示区 (放在按钮下面)
result_row = ttk.Frame(frame)
result_row.grid(row=3, column=0, columnspan=4, pady=(0, 10), sticky="we")
ttk.Label(result_row, text="Dataset 链接:").pack(side="left")
self.ds_result_var = tk.StringVar()
self.ds_result_entry = ttk.Entry(result_row, textvariable=self.ds_result_var, state="readonly")
self.ds_result_entry.pack(side="left", fill="x", expand=True, padx=6)
ttk.Button(result_row, text="📋 复制", command=lambda: self._copy_ds_url()).pack(side="left")
frame.columnconfigure(1, weight=1)
frame.columnconfigure(2, weight=1)
# 4. 日志区
log_frame = ttk.LabelFrame(parent, text="Dataset 自动化流水线日志")
log_frame.pack(fill="both", expand=True, padx=10, pady=6)
self.log_dataset_text = ScrolledText(log_frame, height=15, state="disabled", wrap="word")
self.log_dataset_text.pack(fill="both", expand=True)
def _fetch_dataset_dirs(self):
if not self._ensure_token(): return
self.log_dataset("正在读取 Dataset 仓库根目录...")
def work():
try:
files = self.api.list_repo_files(repo_id=DATASET_REPO_ID, repo_type="dataset")
dirs = set()
for f in files:
parts = f.split("/")
# 过滤掉 tool 和根目录文件，只保留文件夹
if len(parts) > 1 and parts[0] not in ["tool", ".git"]:
dirs.add(parts[0])
sorted_dirs = sorted(list(dirs))
# 🌟 新增：在主线程更新 UI 和打印日志
def update_ui():
self._render_dataset_radios(sorted_dirs)
self.log_dataset(f"✅ 远端目录刷新成功！共发现 {len(sorted_dirs)} 个分类: {', '.join(sorted_dirs)}")
self.after(0, update_ui)
except Exception as e:
self.log_exc("读取 Dataset 目录失败", is_model=True)
# 🌟 新增：报错时更新状态栏
self.after(0, lambda: self.status_ds_var.set("读取目录失败"))
threading.Thread(target=work, daemon=True).start()
def _render_dataset_radios(self, dirs):
# 清空旧的单选框和输入框
for widget in self.ds_radio_container.winfo_children():
widget.destroy()
# 保持原本选中的值，如果不存在了，默认选第一个
current_val = self.ds_region_var.get()
if current_val not in dirs and current_val != "Custom" and dirs:
self.ds_region_var.set(dirs[0])
# 动态生成分类单选框
for region in dirs:
ttk.Radiobutton(self.ds_radio_container, text=region, variable=self.ds_region_var, value=region).pack(side="left", padx=(0, 10))
# 加上自定义部分 (注意：父级是 self.ds_radio_container)
ttk.Radiobutton(self.ds_radio_container, text="自定义:", variable=self.ds_region_var, value="Custom").pack(side="left", padx=(10, 2))
self.ds_custom_entry = ttk.Entry(self.ds_radio_container, textvariable=self.ds_custom_region_var, width=15, state="disabled")
self.ds_custom_entry.pack(side="left")
# 🌟 重新绑定输入框激活逻辑 (因为输入框是刚刚重新创建的)
def on_region_change(*args):
if self.ds_region_var.get() == "Custom":
self.ds_custom_entry.configure(state="normal")
else:
self.ds_custom_entry.configure(state="disabled")
# 清理之前的 trace，防止多次刷新后绑定重复导致卡顿或报错
if hasattr(self, '_region_trace_id'):
try:
self.ds_region_var.trace_remove("write", self._region_trace_id)
except Exception:
pass
self._region_trace_id = self.ds_region_var.trace_add("write", on_region_change)
def log_dataset(self, msg):
self._log_to_tab(msg, getattr(self, 'log_dataset_text', None))
def _browse_ds_local(self):
# 🌟 初始目录设置为当前脚本所在目录，方便用户找
current_dir = os.path.dirname(os.path.abspath(__file__))
folder = filedialog.askdirectory(title="选择包含子文件夹的数据集母目录", initialdir=current_dir)
if folder:
self.ds_local_folder_var.set(folder)
def _start_dataset_pipeline(self):
local_dir = self.ds_local_folder_var.get().strip()
# ==== 修改这里：获取前缀逻辑 ====
selected_region = self.ds_region_var.get()
if selected_region == "Custom":
region_prefix = self.ds_custom_region_var.get().strip().replace(" ", "_").replace("-", "_")
if not region_prefix:
messagebox.showwarning("错误", "您选择了自定义前缀，但输入框为空。")
return
else:
region_prefix = selected_region
# ===============================
if not local_dir or not os.path.exists(local_dir):
messagebox.showwarning("错误", "请选择有效的本地母文件夹。")
return
if not self._ensure_token(): return
if not messagebox.askyesno("确认", f"即将对以下目录执行 5合1 流水线：\n{local_dir}\n前缀：{region_prefix}\n\n继续？"):
return
self.btn_ds_upload.configure(state="disabled")
self.status_ds_var.set("处理中...")
self.log_dataset(f"================ 开始 5合1 流水线 ================")
self.log_dataset(f"目标目录: {local_dir} | 前缀: {region_prefix}")
def worker():
api = self.api
try:
# 1. 清理 txt 换行符
self._clean_txt_newlines(local_dir)
# 2. 清理 Mask 元数据
self._remove_mask_metadata(local_dir)
# 3. 转换 PNG -> JPG
self._convert_png_to_jpg(local_dir)
# 4. 打包 Parquet 并更新 YAML
temp_readme_path, valid_folders = self._generate_parquet_and_update(api, local_dir, region_prefix)
# 5. 上传
self._upload_files_and_readme(api, local_dir, region_prefix, temp_readme_path, valid_folders)
self.log_dataset("🎉 5合1 全流程处理完毕！")
self.after(0, lambda: messagebox.showinfo("完成", "Dataset 流水线执行成功！"))
# 🌟 必须先在本地算出文件夹名（和 _upload_files_and_readme 里的逻辑保持一致）
folder_name = os.path.basename(local_dir).replace(" ", "_").replace("-", "_")
# 🌟 生成专属 Dataset 链接
dataset_url = f"{DATASET_BASE}/{region_prefix}/{folder_name}"
self.log_dataset(f"🔗 Dataset 页面: {dataset_url}")
# 🌟 安全地在主线程更新 UI
def update_ui():
self.status_ds_var.set("上传完成")
self.ds_result_var.set(dataset_url)
self.after(0, update_ui)
# 刷新 JSON 的拉取（可选）
# self.after(0, self.load_remote)
except Exception as e:
self.log_exc("Dataset 流水线异常", is_model=True)
self.after(0, lambda: self.status_ds_var.set("执行失败"))
finally:
self.after(0, lambda: self.btn_ds_upload.configure(state="normal"))
threading.Thread(target=worker, daemon=True).start()
# ---------------- 原有脚本的函数改造 ----------------
def _clean_txt_newlines(self, current_dir):
import glob
self.log_dataset("🧹 [任务 1/5] 正在清理所有 TXT 文件中的换行符...")
txt_files = glob.glob(os.path.join(current_dir, "**", "*.txt"), recursive=True)
if not txt_files:
self.log_dataset(" ✅ 没有找到需要处理的 TXT 文件。")
return
cleaned_count = 0
for txt_path in txt_files:
if os.path.dirname(txt_path) == current_dir:
continue
try:
with open(txt_path, 'r', encoding='utf-8') as f:
content = ''.join(line.strip() for line in f)
with open(txt_path, 'w', encoding='utf-8') as f:
f.write(content)
cleaned_count += 1
except Exception as e:
self.log_dataset(f" ⚠️ 处理 {os.path.basename(txt_path)} 失败: {e}")
self.log_dataset(f" 🎉 TXT 清理完成！共处理了 {cleaned_count} 个文件。")
def _remove_mask_metadata(self, current_dir):
import glob
self.log_dataset("🛡️ [任务 2/5] 正在剥离 Mask 蒙版图的元数据...")
png_files = glob.glob(os.path.join(current_dir, "**", "*.png"), recursive=True)
mask_files = [f for f in png_files if "masklabel" in os.path.basename(f).lower()]
if not mask_files:
self.log_dataset(" ✅ 没有找到 masklabel.png 文件，跳过此步骤。")
return
success_count = 0
for file_path in mask_files:
try:
with Image.open(file_path) as img:
data = list(img.getdata())
clean_img = Image.new(img.mode, img.size)
clean_img.putdata(data)
clean_img.save(file_path, "PNG")
success_count += 1
except Exception as e:
self.log_dataset(f" ⚠️ 处理失败 [{os.path.basename(file_path)}]: {e}")
self.log_dataset(f" 🎉 元数据清理完成！共处理了 {success_count} 张 Mask 图。")
def _convert_png_to_jpg(self, current_dir):
import glob
self.log_dataset("🎬 [任务 3/5] 正在批量转换常规 PNG 为 JPG...")
png_files = glob.glob(os.path.join(current_dir, "**", "*.png"), recursive=True)
if not png_files:
self.log_dataset(" ✅ 没有找到需要转换的 PNG 文件。")
return
converted_count = 0
for png_path in png_files:
if "masklabel" in png_path.lower():
continue
jpg_path = os.path.splitext(png_path)[0] + ".jpg"
try:
subprocess.run(
["ffmpeg", "-y", "-i", png_path, "-q:v", "2", jpg_path],
stdout=subprocess.DEVNULL,
stderr=subprocess.DEVNULL,
check=True
)
os.remove(png_path)
converted_count += 1
self.log_dataset(f" 🔄 转换: {os.path.basename(png_path)} -> {os.path.basename(jpg_path)}")
except subprocess.CalledProcessError:
self.log_dataset(f" ⚠️ 转换失败: {png_path}")
except FileNotFoundError:
self.log_dataset(" ❌ 错误: 找不到 ffmpeg！请确保已安装并配置了环境变量。")
return
self.log_dataset(f" 🎉 转换完成！共转换 {converted_count} 张图片。")
def _generate_parquet_and_update(self, api, current_dir, region_prefix):
import glob, pandas as pd
from datasets import Dataset, Features, Image as HFImage, Value
self.log_dataset("📦 [任务 4/5] 正在本地打包 Parquet (已启用极简聚合模式)...")
valid_folders = []
for sub_folder_name in os.listdir(current_dir):
sub_folder_path = os.path.join(current_dir, sub_folder_name)
if os.path.isdir(sub_folder_path) and not sub_folder_name.startswith('.'):
jpg_files = glob.glob(os.path.join(sub_folder_path, "*.jpg")) + glob.glob(os.path.join(sub_folder_path, "*.jpeg"))
if not jpg_files:
self.log_dataset(f" ⚠️ 跳过子文件夹 [{sub_folder_name}]: 未找到任何 .jpg 图片。")
continue
self.log_dataset(f" 📦 正在打包: {sub_folder_name}")
data_rows = []
for jpg_path in jpg_files:
file_name = os.path.basename(jpg_path)
base_name = os.path.splitext(file_name)[0]
txt_path = os.path.join(sub_folder_path, f"{base_name}.txt")
text_content = ""
if os.path.exists(txt_path):
with open(txt_path, "r", encoding="utf-8") as f:
text_content = f.read().strip()
try:
with open(jpg_path, "rb") as f:
image_bytes = {"bytes": f.read(), "path": None}
except Exception as e:
self.log_dataset(f" ⚠️ 跳过损坏图片 {file_name}: {e}")
continue
data_rows.append({
"image": image_bytes,
"file_name": file_name,
"text": text_content
})
if data_rows:
df = pd.DataFrame(data_rows)
features = Features({
"image": HFImage(),
"file_name": Value("string"),
"text": Value("string")
})
ds = Dataset.from_pandas(df, features=features)
ds.to_parquet(os.path.join(sub_folder_path, "train.parquet"))
valid_folders.append(sub_folder_name)
self.log_dataset(f" ✅ 生成 train.parquet ({len(data_rows)} 条数据)")
return None, valid_folders
def _upload_files_and_readme(self, api, current_dir, region_prefix, temp_readme_path, valid_folders):
self.log_dataset("🚀 [任务 5/5] 正在上传文件到 Hugging Face...")
local_folder_name = os.path.basename(current_dir).replace(" ", "_").replace("-", "_")
for sub_folder_name in valid_folders:
sub_folder_path = os.path.join(current_dir, sub_folder_name)
path_in_repo = f"{region_prefix}/{local_folder_name}/{sub_folder_name}"
self.log_dataset(f" 📦 开始上传: {sub_folder_name} -> {path_in_repo}")
try:
api.upload_folder(
folder_path=sub_folder_path,
repo_id=DATASET_REPO_ID,
repo_type="dataset",
path_in_repo=path_in_repo,
commit_message=f"Update files for {path_in_repo}",
allow_patterns=["*.jpg", "*.jpeg", "*.txt", "*.png", "*.parquet"]
)
self.log_dataset(f" 🎉 {sub_folder_name} 上传完成！")
except Exception as e:
self.log_dataset(f" ❌ {sub_folder_name} 上传失败: {e}")
# ==========================================================
# ==================== 区域 4: Showcase 管理 ================
# ==========================================================
def _build_showcase_tab(self):
parent = self.tab_showcase
# 1. 输入区
frame = ttk.Frame(parent)
frame.pack(fill="x", padx=10, pady=10)
# 本地文件夹
ttk.Label(frame, text="本地文件夹:").grid(row=0, column=0, sticky="w", pady=4)
self.sc_local_folder_var = tk.StringVar()
ttk.Entry(frame, textvariable=self.sc_local_folder_var, width=70).grid(row=0, column=1, padx=6, sticky="we")
ttk.Button(frame, text="📁 浏览...", width=18, command=self._browse_sc_local).grid(row=0, column=2)
# 远端目录 (Treeview 改造，完全复用 Models 的样式)
ttk.Label(frame, text="上传到 Showcase 目录:").grid(row=1, column=0, sticky="nw", pady=4)
tree_frame = ttk.Frame(frame)
tree_frame.grid(row=1, column=1, padx=6, sticky="nsew")
# 滚动条
vsb = ttk.Scrollbar(tree_frame, orient="vertical")
vsb.pack(side="right", fill="y")
# 树形列表
self.sc_remote_tree = ttk.Treeview(tree_frame, height=6, show="tree",yscrollcommand=vsb.set, selectmode="browse")
self.sc_remote_tree.pack(side="left", fill="both", expand=True)
vsb.config(command=self.sc_remote_tree.yview)
# 绑定树形点击事件
self.sc_remote_tree.bind("<<TreeviewSelect>>", self._on_sc_remote_tree_select)
# 刷新按钮
ttk.Button(frame, text="🔄 拉取案例目录", width=18, command=self._fetch_showcase_dirs).grid(row=1, column=2, sticky="n")
# 按钮区
btn_row = ttk.Frame(frame)
btn_row.grid(row=2, column=0, columnspan=3, pady=15, sticky="w")
# 🌟 分离为两个独立的按钮
self.btn_sc_process = ttk.Button(btn_row, text="🔄 仅本地处理并生成 JSON", command=lambda: self._start_showcase_pipeline(upload=False))
self.btn_sc_process.pack(side="left")
self.btn_sc_upload = ttk.Button(btn_row, text="🢁 上传 WebP 和 JSON 到 HF", command=lambda: self._start_showcase_pipeline(upload=True))
self.btn_sc_upload.pack(side="left", padx=10)
# 🌟 新增：远端洗数据按钮
self.status_sc_var = tk.StringVar(value="就绪")
ttk.Label(btn_row, textvariable=self.status_sc_var).pack(side="left", padx=10)
# Showcase 链接展示区
result_row = ttk.Frame(frame)
result_row.grid(row=3, column=0, columnspan=4, pady=(0, 10), sticky="we")
ttk.Label(result_row, text="Showcase 链接:").pack(side="left")
self.sc_result_var = tk.StringVar()
self.sc_result_entry = ttk.Entry(result_row, textvariable=self.sc_result_var, state="readonly")
self.sc_result_entry.pack(side="left", fill="x", expand=True, padx=6)
# 复制按钮
ttk.Button(result_row, text="📋 复制", command=self._copy_sc_url).pack(side="left")
frame.columnconfigure(1, weight=1)
frame.rowconfigure(1, weight=1)
# 2. 日志区
log_frame = ttk.LabelFrame(parent, text="Showcase 操作日志")
log_frame.pack(fill="both", expand=True, padx=10, pady=6)
self.log_sc_text = ScrolledText(log_frame, height=10, state="disabled", wrap="word")
self.log_sc_text.pack(fill="both", expand=True)
def _copy_sc_url(self):
url = self.sc_result_var.get()
if url:
self.clipboard_clear()
self.clipboard_append(url)
messagebox.showinfo("成功", "链接已复制到剪贴板！")
def _copy_ds_url(self):
url = self.ds_result_var.get()
if url:
self.clipboard_clear()
self.clipboard_append(url)
messagebox.showinfo("成功", "Dataset 链接已复制到剪贴板！")
def log_showcase(self, msg):
self._log_to_tab(msg, getattr(self, 'log_sc_text', None))
def _browse_sc_local(self):
if getattr(sys, 'frozen', False):
current_dir = os.path.dirname(sys.executable)
else:
current_dir = os.path.dirname(os.path.abspath(__file__))
folder = filedialog.askdirectory(title="选择包含 PNG 的 Showcase 文件夹", initialdir=current_dir)
if folder:
self.sc_local_folder_var.set(folder)
def _fetch_showcase_dirs(self):
if not self._ensure_token(): return
self.status_sc_var.set("正在读取远端目录...")
self.log_showcase("正在读取 Showcase 仓库目录列表...")
def work():
try:
api = self.api
all_files = list(api.list_repo_files(repo_id=MODEL_REPO_ID, repo_type="model"))
dirs = set()
for f in all_files:
if f.startswith(f"{SHOWCASE_DIR}/"):
d = str(Path(f).parent)
if d != "comfyui_showcase":
dirs.add(d)
sorted_dirs = sorted(list(dirs))
self.after(0, lambda: self._update_sc_remote_tree(sorted_dirs))
except Exception:
self.log_exc("读取 Showcase 目录失败", is_model=True)
finally:
self.after(0, lambda: self.status_sc_var.set("就绪"))
threading.Thread(target=work, daemon=True).start()
def _update_sc_remote_tree(self, dirs):
for item in self.sc_remote_tree.get_children():
self.sc_remote_tree.delete(item)
tree_nodes = {}
for d in dirs:
safe_d = d.replace("\\", "/")
parts = safe_d.split("/")
parent_id = ""
current_path = ""
for i, part in enumerate(parts):
current_path = f"{current_path}/{part}" if current_path else part
if current_path not in tree_nodes:
is_leaf = (i == len(parts) - 1)
display_text = f"📁 {part}" if is_leaf else part
node_id = self.sc_remote_tree.insert(parent_id, "end", text=display_text, values=(current_path,))
tree_nodes[current_path] = node_id
parent_id = tree_nodes[current_path]
for root_node in self.sc_remote_tree.get_children(""):
self.sc_remote_tree.item(root_node, open=True)
self.log_showcase(f"已加载 {len(dirs)} 个远端目录结构")
def _on_sc_remote_tree_select(self, event):
selection = self.sc_remote_tree.selection()
if selection:
item_id = selection[0]
values = self.sc_remote_tree.item(item_id, "values")
if values:
self.sc_remote_dir = values[0]
def _start_showcase_pipeline(self, upload=False):
local_dir = self.sc_local_folder_var.get().strip()
remote_dir = getattr(self, 'sc_remote_dir', "").strip().strip("/")
if not local_dir or not os.path.exists(local_dir):
messagebox.showwarning("错误", "请选择有效的本地文件夹。")
return
if upload and not remote_dir:
messagebox.showwarning("错误", "上传模式下，请先在右侧选择要上传到的远端目录。")
return
if not self._ensure_token(): return
local_folder_name = os.path.basename(os.path.normpath(local_dir))
final_target_path = f"{remote_dir}/{local_folder_name}" if remote_dir else ""
action_name = "上传" if upload else "本地处理"
if not messagebox.askyesno("确认", f"即将执行【{action_name}】流水线。\n目标目录: {final_target_path if upload else local_dir}\n\n继续？"):
return
# 禁用按钮防止连点
self.btn_sc_process.configure(state="disabled")
if upload:
self.btn_sc_upload.configure(state="disabled")
self.status_sc_var.set(f"{action_name}中...")
self.log_showcase(f"================ 开始 Showcase {action_name} 流水线 ================")
def worker():
api = self.api
try:
png_archive_dir = os.path.join(local_dir, "png_input")
if not os.path.exists(png_archive_dir):
os.makedirs(png_archive_dir, exist_ok=True)
# 1. 拉取云端旧 JSON (仅在 upload=True 时去拉取合并)
existing_data = []
if upload:
remote_json_path = f"{final_target_path}/gallery.json"
self.log_showcase(f"☁️ 正在检查云端旧 gallery.json...")
try:
local_tmp_json = api.hf_hub_download(repo_id=MODEL_REPO_ID, repo_type="model", filename=remote_json_path)
with open(local_tmp_json, "r", encoding="utf-8") as f:
existing_data = json.load(f)
self.log_showcase(f" ✅ 发现旧数据 {len(existing_data)} 条。")
except Exception:
self.log_showcase(" ℹ️ 无旧数据，将新建。")
# 2. 扫描本地 PNG（只要有 PNG，就处理）
image_files = [f for f in Path(local_dir).iterdir() if f.suffix.lower() == ".png"]
new_gallery_data = []
if image_files:
self.log_showcase(f"📦 发现 {len(image_files)} 张 PNG，开始提取数据并转换...")
for img_file in image_files:
data = self._extract_single_png(img_file)
if not data:
self.log_showcase(f" ⚠️ 数据提取失败: {img_file.name}")
continue
webp_name = img_file.stem + ".webp"
webp_path = os.path.join(local_dir, webp_name)
ok, msg = self._convert_png_to_webp(img_file, webp_path, quality=90)
if not ok:
self.log_showcase(f" ❌ 转换 WebP 失败 {img_file.name}: {msg}")
continue
try:
shutil.move(str(img_file), os.path.join(png_archive_dir, img_file.name))
except Exception as e:
self.log_showcase(f" ⚠️ 原图归档失败: {e}")
data["file_name"] = webp_name
new_gallery_data.append(data)
self.log_showcase(f" ✅ 解析并转换完成: {webp_name}")
else:
self.log_showcase(" ℹ️ 本地没有新的 PNG 需要处理，准备直接打包 JSON。")
# 3. 处理 JSON 数据
local_json_path = os.path.join(local_dir, "gallery.json")
# 如果是上传模式，且拉到过旧数据，就合并
if upload:
merged_dict = {item["file_name"]: item for item in existing_data if "file_name" in item}
for item in new_gallery_data:
merged_dict[item["file_name"]] = item
final_gallery_data = list(merged_dict.values())
else:
# 如果只是本地处理，就把当前本地生成的直接写入
final_gallery_data = new_gallery_data
if final_gallery_data:
with open(local_json_path, "w", encoding="utf-8") as f:
json.dump(final_gallery_data, f, indent=2, ensure_ascii=False)
self.log_showcase(f"📄 本地 gallery.json 已生成，包含 {len(final_gallery_data)} 条数据。")
else:
self.log_showcase("❌ 没有任何数据可以生成 JSON。")
return
# 4. 如果指定了上传，则开始上传
if upload:
self.log_showcase(f"🚀 准备上传 WebP 和 JSON 到云端...")
api.upload_folder(
folder_path=local_dir,
repo_id=MODEL_REPO_ID,
repo_type="model",
path_in_repo=final_target_path,
commit_message=f"Add new images and update gallery.json at {final_target_path}",
allow_patterns=["*.webp", "*.json"]
)
self.log_showcase("🎉 上传完毕！")
showcase_url = f"{MODEL_BASE}/tree/main/{final_target_path}"
self.after(0, lambda: self.sc_result_var.set(showcase_url))
self.after(0, lambda: messagebox.showinfo("完成", "Showcase 上传成功！"))
else:
self.log_showcase("🎉 本地处理完毕！请检查本地文件后，再点击上传。")
self.after(0, lambda: messagebox.showinfo("完成", "本地处理成功！"))
self.after(0, lambda: self.status_sc_var.set("处理完成"))
except Exception as e:
self.log_exc("Showcase 流水线异常", is_model=True)
self.after(0, lambda: self.status_sc_var.set("执行失败"))
finally:
self.after(0, lambda: self.btn_sc_process.configure(state="normal"))
self.after(0, lambda: self.btn_sc_upload.configure(state="normal"))
threading.Thread(target=worker, daemon=True).start()
# ==========================================================
# ==================== Showcase 转换与提取核心 ================
# ==========================================================
def _convert_png_to_webp(self, src_path, dst_path, quality=90):
"""将 PNG 转 WebP，并将原始 metadata 完整写入 EXIF"""
try:
with Image.open(src_path) as img:
# 1. 抓取原生的 JSON 数据
raw_prompt = img.info.get('prompt', '')
raw_workflow = img.info.get('workflow', '')
# 2. 构建 EXIF 写入
exif = img.getexif()
# 271: Make (对应 Prompt), 270: ImageDescription (对应 Workflow)
if raw_prompt:
exif[0x010f] = "Prompt: " + raw_prompt
if raw_workflow:
exif[0x010e] = "Workflow: " + raw_workflow
if img.mode in ("RGBA", "P"):
img = img.convert("RGB")
# 3. 保存 WebP
save_kwargs = {"quality": int(quality), "method": 6}
if raw_prompt or raw_workflow:
save_kwargs["exif"] = exif.tobytes()
img.save(dst_path, "WEBP", **save_kwargs)
return True, "OK"
except Exception as e:
return False, str(e)
def _clean_showcase_name(self, name):
if not name or not isinstance(name, str): return "-"
base = str(name).split('\\')[-1].split('/')[-1]
return re.sub(r'\.(safetensors|pt|ckpt|sft|pth|bin|gguf)$', '', base, flags=re.IGNORECASE).strip()
def _resolve_value(self, val, prompt_data, visited=None, depth=0):
"""连线穿透：解决 sampler/scheduler/seed 返回 ["79", 0] 数组的问题"""
if visited is None: visited = set()
if depth > 10: return "-"
if not isinstance(val, list) or len(val) != 2: return val
node_id = str(val[0])
if node_id in visited or node_id not in prompt_data: return "-"
visited.add(node_id)
node = prompt_data[node_id]
if not isinstance(node, dict): return "-"
if node.get("mode") == 4: return "-"
inputs = node.get("inputs", {})
for key in ["value", "sampler_name", "scheduler", "int", "float", "string", "text", "seed"]:
if key in inputs:
res = self._resolve_value(inputs[key], prompt_data, visited, depth+1)
if res != "-" and not isinstance(res, list):
return res
return "-"
# def _deep_extract_text(self, node_id, prompt_data, visited=None, depth=0):
# """深度溯源：允许获取所有字符串"""
# if visited is None: visited = set()
# if depth > 30 or node_id not in prompt_data or node_id in visited: return ""
# visited.add(node_id)
# node = prompt_data[node_id]
# if not isinstance(node, dict): return ""
# if node.get("mode") == 4: return ""
# inputs = node.get("inputs", {})
# for key in ["text", "string", "value", "text_a", "Prompt", "positive"]:
# if key in inputs and isinstance(inputs[key], str) and len(inputs[key].strip()) > 0:
# return inputs[key].strip()
# parts = []
# for key in ["text", "string", "positive", "conditioning", "clip"]:
# if key in inputs and isinstance(inputs[key], list) and len(inputs[key]) == 2:
# res = self._deep_extract_text(str(inputs[key][0]), prompt_data, visited.copy(), depth + 1)
# if res and res not in parts: parts.append(res)
# if not parts:
# for val in inputs.values():
# if isinstance(val, list) and len(val) == 2 and isinstance(val[0], str):
# res = self._deep_extract_text(val[0], prompt_data, visited.copy(), depth + 1)
# if res and res not in parts: parts.append(res)
# if len(parts) > 0:
# return ", ".join(parts)
# return ""
def _deep_extract_text(self, node_id, prompt_data, visited=None, depth=0):
"""深度溯源：允许获取所有字符串，但屏蔽容易造成污染的条件分支节点"""
if visited is None: visited = set()
if depth > 30 or node_id not in prompt_data or node_id in visited: return ""
visited.add(node_id)
node = prompt_data[node_id]
if not isinstance(node, dict): return ""
if node.get("mode") == 4: return ""
ctype = node.get("class_type", "")
# 🌟 核心防污染：遇到 Switch、Random 等分支/随机节点，直接斩断，不给它收集多个分支的机会
if "Random" in ctype or "Switch" in ctype or "Function" in ctype:
return ""
inputs = node.get("inputs", {})
for key in ["text", "string", "value", "text_a", "Prompt", "positive"]:
if key in inputs and isinstance(inputs[key], str) and len(inputs[key].strip()) > 0:
return inputs[key].strip()
parts = []
for key in ["text", "string", "positive", "conditioning", "clip"]:
if key in inputs and isinstance(inputs[key], list) and len(inputs[key]) == 2:
res = self._deep_extract_text(str(inputs[key][0]), prompt_data, visited.copy(), depth + 1)
if res and res not in parts: parts.append(res)
if not parts:
for val in inputs.values():
if isinstance(val, list) and len(val) == 2 and isinstance(val[0], str):
res = self._deep_extract_text(val[0], prompt_data, visited.copy(), depth + 1)
if res and res not in parts: parts.append(res)
if len(parts) > 0:
return ", ".join(parts)
return ""
def _parse_comfy_data(self, prompt_data, img_width, img_height, dasiwa_text):
res = {
"size": f"{img_width}x{img_height}", "loras": [], "model": "-", "vae": "-", "clip": "-",
"sampler": "-", "scheduler": "-", "seed": "-", "steps": "-", "cfg": "-", "denoise": "-",
"pos": "", "neg": ""
}
if not isinstance(prompt_data, dict):
return res
seen_loras = set()
all_text_for_lora = "" # 🌟 用来收集所有可能包含 <lora:xxx> 的文本
# 1. 提取基础参数
for nid, node in prompt_data.items():
if not isinstance(node, dict) or node.get("mode") == 4: continue
ctype = node.get("class_type", "")
inputs = node.get("inputs", {})
if "UNETLoader" in ctype or "CheckpointLoader" in ctype:
m = self._clean_showcase_name(inputs.get("unet_name") or inputs.get("ckpt_name") or "")
if m != "-": res["model"] = m
if "VAELoader" in ctype:
res["vae"] = self._clean_showcase_name(inputs.get("vae_name", ""))
if "CLIPLoader" in ctype:
res["clip"] = self._clean_showcase_name(inputs.get("clip_name", ""))
if "LoraLoader" in ctype or "lora" in ctype.lower():
l_name = self._clean_showcase_name(inputs.get("lora_name", ""))
if l_name != "-" and l_name not in seen_loras:
seen_loras.add(l_name)
res["loras"].append(l_name)
if "KSampler" in ctype:
res["seed"] = self._resolve_value(inputs.get("seed"), prompt_data)
res["steps"] = self._resolve_value(inputs.get("steps"), prompt_data)
res["cfg"] = self._resolve_value(inputs.get("cfg"), prompt_data)
res["denoise"] = self._resolve_value(inputs.get("denoise"), prompt_data)
res["sampler"] = self._resolve_value(inputs.get("sampler_name"), prompt_data)
res["scheduler"] = self._resolve_value(inputs.get("scheduler"), prompt_data)
# 🌟 新增：把节点里所有的 text/string/positive 等文本字段收集起来
for k, v in inputs.items():
if isinstance(v, str) and ("<lora:" in v or "lora_name" in v):
all_text_for_lora += "\n" + v
if all_text_for_lora:
# 正则匹配 <lora: 后面直到 : 或 > 之间的内容
found_tags = re.findall(r'<lora:([^:>]+)(?::[^>]+)?>', all_text_for_lora)
for tag in found_tags:
l_name = self._clean_showcase_name(tag)
if l_name != "-" and l_name not in seen_loras:
seen_loras.add(l_name)
res["loras"].append(l_name)
# 2. 兜底找 seed
if res["seed"] == "-" or res["seed"] == "":
for nid, node in prompt_data.items():
if not isinstance(node, dict) or node.get("mode") == 4: continue
ctype = node.get("class_type", "")
inputs = node.get("inputs", {})
if "KSampler" in ctype and "seed" in inputs:
seed_val = self._resolve_value(inputs["seed"], prompt_data)
if seed_val != "-" and not isinstance(seed_val, list):
res["seed"] = seed_val
break
if "Seed" in ctype or "seed" in inputs:
seed_val = inputs.get("seed") or inputs.get("value")
if isinstance(seed_val, (int, str)) and str(seed_val).isdigit():
res["seed"] = seed_val
break
# 3. 溯源正负文本
for nid, node in prompt_data.items():
if not isinstance(node, dict): continue
if "KSampler" in node.get("class_type", ""):
pos_input = node.get("inputs", {}).get("positive")
if isinstance(pos_input, list):
p_text = self._deep_extract_text(str(pos_input[0]), prompt_data)
if len(p_text) > len(res["pos"]): res["pos"] = p_text
neg_input = node.get("inputs", {}).get("negative")
if isinstance(neg_input, list):
n_text = self._deep_extract_text(str(neg_input[0]), prompt_data)
if len(n_text) > len(res["neg"]): res["neg"] = n_text
# 4. 文本兜底 (dasiwa_text)
if dasiwa_text:
pos_match = re.search(r"^(.*?)(?:Negative prompt:|$)", dasiwa_text, re.DOTALL | re.IGNORECASE)
if pos_match:
text_val = pos_match.group(1).strip()
if len(text_val) > len(res["pos"]):
res["pos"] = text_val
neg_match = re.search(r"Negative prompt:\s*(.*?)(?:Steps:|$)", dasiwa_text, re.DOTALL | re.IGNORECASE)
if neg_match:
neg_text = neg_match.group(1).strip()
if neg_text and "negative_prompt_not_found" not in neg_text:
if len(neg_text) > len(res["neg"]): res["neg"] = neg_text
# 🌟 顺手也从 dasiwa_text 里扫一遍 <lora:...>
found_tags = re.findall(r'<lora:([^:>]+)(?::[^>]+)?>', dasiwa_text)
for tag in found_tags:
l_name = self._clean_showcase_name(tag)
if l_name != "-" and l_name not in seen_loras:
seen_loras.add(l_name)
res["loras"].append(l_name)
if res["seed"] == "-" or res["seed"] == "":
seed_match = re.search(r"Seed:\s*(\d+)", dasiwa_text, re.IGNORECASE)
if seed_match:
res["seed"] = seed_match.group(1)
# 5. 清洗文本
def clean_prompt_text(text):
if not text: return ""
text = re.sub(r'<lora:[^>]+>', '', text)
text = re.sub(r'\{[^{}]*__.*?\}', '', text)
text = re.sub(r'__[a-zA-Z0-9_-]+__', '', text)
text = re.sub(r',\s*,', ',', text)
return text.strip(', ')
res["pos"] = clean_prompt_text(res["pos"])
res["neg"] = clean_prompt_text(res["neg"])
if dasiwa_text and not res["loras"]:
lora_matches = re.findall(r"<lora:([^:>]+)(?::[^>]+)?>", dasiwa_text)
res["loras"] = list(dict.fromkeys([self._clean_showcase_name(l) for l in lora_matches]))
return res
def _extract_single_png(self, png_path):
"""纯净版提取逻辑：强制提取汇总节点文本，屏蔽任何乱序分支"""
try:
with Image.open(png_path) as img:
real_w, real_h = img.size
raw_prompt = None
dasiwa_text = img.info.get('parameters', '') or img.info.get('Description', '')
# 1. 获取底层 JSON
if "prompt" in img.info:
raw_prompt = img.info["prompt"]
elif "workflow" in img.info:
raw_prompt = img.info["workflow"]
if not raw_prompt:
exif = img.getexif()
if exif:
for tag in [270, 271, 37510, 0x010e, 0x010f]:
val = exif.get(tag, "")
if isinstance(val, bytes): val = val.decode('utf-8', errors='ignore')
if isinstance(val, str):
val = val.strip().replace('\x00', '')
if val.startswith("Prompt: "): raw_prompt = val[8:]
elif val.startswith("Workflow: "): raw_prompt = val[10:]
elif "{" in val: raw_prompt = val
if raw_prompt and "{" in raw_prompt: break
parsed = None
if raw_prompt and isinstance(raw_prompt, str):
json_match = re.search(r'\{.*\}', raw_prompt, re.DOTALL)
if json_match:
try:
parsed = json.loads(json_match.group(0))
except Exception: pass
if not parsed and not dasiwa_text:
return None
# 2. 先解析基础参数（model, lora, seed 等）
report = self._parse_comfy_data(parsed, real_w, real_h, dasiwa_text)
# 3. 🌟 终极提取：直接暴力遍历找汇总节点，无视 KSampler 溯源
if isinstance(parsed, dict):
pos_text = ""
neg_text = ""
for nid, node in parsed.items():
if not isinstance(node, dict): continue
ctype = node.get("class_type", "")
inputs = node.get("inputs", {})
# 找正向文本
if "WS_Positive" in ctype or "String Function" in ctype:
val = inputs.get("text")
# 只要比当前长就覆盖（防止被短文本抢先）
if isinstance(val, str) and len(val.strip()) > len(pos_text):
pos_text = val.strip()
# 找负向文本（如果有负向汇总节点的话）
if "WS_Negative" in ctype or "Negative" in ctype:
val = inputs.get("text") or inputs.get("value")
if isinstance(val, str) and len(val.strip()) > len(neg_text):
neg_text = val.strip()
# 如果找到了汇总文本，直接覆盖 report 里的内容
if pos_text:
report["pos"] = pos_text
if neg_text:
report["neg"] = neg_text
# 4. 如果 JSON 里没拿到，或者 dasiwa_text 里明显更长（比如包含中英混排），用纯文本兜底
if dasiwa_text:
pos_match = re.search(r"^(.*?)(?:Negative prompt:|$)", dasiwa_text, re.DOTALL | re.IGNORECASE)
if pos_match:
text_val = pos_match.group(1).strip()
if len(text_val) > len(report.get("pos", "")):
report["pos"] = text_val
neg_match = re.search(r"Negative prompt:\s*(.*?)(?:Steps:|$)", dasiwa_text, re.DOTALL | re.IGNORECASE)
if neg_match:
neg_text_val = neg_match.group(1).strip()
if neg_text_val and "negative_prompt_not_found" not in neg_text_val:
if len(neg_text_val) > len(report.get("neg", "")):
report["neg"] = neg_text_val
# 5. 终极清洗（去 LoRA 标签，去动态模板）
def clean_prompt_text(text):
if not text: return ""
text = re.sub(r'<lora:[^>]+>', '', text) # 去 LoRA
text = re.sub(r'\{[^{}]*__.*?\}', '', text) # 去动态模板
text = re.sub(r'__[a-zA-Z0-9_-]+__', '', text)
text = re.sub(r',\s*,', ',', text) # 去重复逗号
return text.strip(', ')
report["pos"] = clean_prompt_text(report["pos"])
report["neg"] = clean_prompt_text(report["neg"])
report["file_name"] = os.path.basename(png_path)
return report
except Exception as e:
self.log_showcase(f" ⚠️ 提取失败 {os.path.basename(png_path)}: {e}")
return None
def _start_showcase_pipeline(self, upload=False):
local_dir = self.sc_local_folder_var.get().strip()
remote_dir = getattr(self, 'sc_remote_dir', "").strip().strip("/")
if not local_dir or not os.path.exists(local_dir):
messagebox.showwarning("错误", "请选择有效的本地文件夹。")
return
if upload and not remote_dir:
messagebox.showwarning("错误", "上传模式下，请先在右侧选择要上传到的远端目录。")
return
if not self._ensure_token(): return
local_folder_name = os.path.basename(os.path.normpath(local_dir))
final_target_path = f"{remote_dir}/{local_folder_name}" if remote_dir else ""
action_name = "上传" if upload else "本地处理"
if not messagebox.askyesno("确认", f"即将执行【{action_name}】流水线。\n目标目录: {final_target_path if upload else local_dir}\n\n继续？"):
return
# 禁用按钮
self.btn_sc_process.configure(state="disabled")
if upload:
self.btn_sc_upload.configure(state="disabled")
self.status_sc_var.set(f"{action_name}中...")
self.log_showcase(f"================ 开始 Showcase {action_name} 流水线 ================")
def worker():
api = self.api
try:
local_json_path = os.path.join(local_dir, "gallery.json")
# ================= 分支 A：仅本地处理 =================
if not upload:
png_archive_dir = os.path.join(local_dir, "png_input")
if not os.path.exists(png_archive_dir):
os.makedirs(png_archive_dir, exist_ok=True)
image_files = [f for f in Path(local_dir).iterdir() if f.suffix.lower() == ".png"]
new_gallery_data = []
if image_files:
self.log_showcase(f"📦 发现 {len(image_files)} 张 PNG，开始提取数据并转换...")
for img_file in image_files:
data = self._extract_single_png(img_file)
if not data:
self.log_showcase(f" ⚠️ 数据提取失败: {img_file.name}")
continue
webp_name = img_file.stem + ".webp"
webp_path = os.path.join(local_dir, webp_name)
ok, msg = self._convert_png_to_webp(img_file, webp_path, quality=90)
if not ok:
self.log_showcase(f" ❌ 转换 WebP 失败 {img_file.name}: {msg}")
continue
try:
shutil.move(str(img_file), os.path.join(png_archive_dir, img_file.name))
except Exception as e:
self.log_showcase(f" ⚠️ 原图归档失败: {e}")
data["file_name"] = webp_name
new_gallery_data.append(data)
self.log_showcase(f" ✅ 解析并转换完成: {webp_name}")
else:
self.log_showcase(" ℹ️ 本地没有新的 PNG 需要处理。")
# 🌟 核心修复：读取本地已有的 JSON，进行增量合并
existing_local_data = []
if os.path.exists(local_json_path):
try:
with open(local_json_path, "r", encoding="utf-8") as f:
existing_local_data = json.load(f)
self.log_showcase(f" ✅ 发现本地已有 {len(existing_local_data)} 条旧 JSON。")
except Exception:
pass
# 合并旧数据和新数据
merged_dict = {item["file_name"]: item for item in existing_local_data if "file_name" in item}
for item in new_gallery_data:
merged_dict[item["file_name"]] = item
final_gallery_data = list(merged_dict.values())
if final_gallery_data:
with open(local_json_path, "w", encoding="utf-8") as f:
json.dump(final_gallery_data, f, indent=2, ensure_ascii=False)
self.log_showcase(f"📄 本地 gallery.json 已更新，包含 {len(final_gallery_data)} 条数据（本地旧+新）。")
self.after(0, lambda: messagebox.showinfo("完成", "本地处理成功！"))
else:
self.log_showcase("❌ 没有生成任何数据。")
self.after(0, lambda: self.status_sc_var.set("处理完成"))
# ================= 分支 B：直接上传 =================
else:
if not os.path.exists(local_json_path):
self.log_showcase("❌ 本地找不到 gallery.json，请先点击【仅本地处理】生成！")
self.after(0, lambda: self.status_sc_var.set("缺少 JSON"))
return
# 🌟 核心修复：上传前，从云端拉取旧 JSON 再次合并（防止本地漏传）
self.log_showcase(f"☁️ 正在拉取云端旧 JSON 进行最终合并...")
remote_json_path = f"{final_target_path}/gallery.json"
remote_existing_data = []
try:
local_tmp_json = api.hf_hub_download(repo_id=MODEL_REPO_ID, repo_type="model", filename=remote_json_path)
with open(local_tmp_json, "r", encoding="utf-8") as f:
remote_existing_data = json.load(f)
self.log_showcase(f" ✅ 云端旧数据 {len(remote_existing_data)} 条。")
except Exception:
self.log_showcase(" ℹ️ 云端无旧 JSON。")
# 读取本地的 JSON
with open(local_json_path, "r", encoding="utf-8") as f:
local_gallery_data = json.load(f)
# 云端 + 本地 彻底合并
final_merged_dict = {item["file_name"]: item for item in remote_existing_data if "file_name" in item}
for item in local_gallery_data:
final_merged_dict[item["file_name"]] = item
final_upload_data = list(final_merged_dict.values())
# 重写本地 JSON，保证上传的是合并后的终极版
with open(local_json_path, "w", encoding="utf-8") as f:
json.dump(final_upload_data, f, indent=2, ensure_ascii=False)
self.log_showcase(f"📄 最终合并 JSON 包含 {len(final_upload_data)} 条数据。开始上传...")
api.upload_folder(
folder_path=local_dir,
repo_id=MODEL_REPO_ID,
repo_type="model",
path_in_repo=final_target_path,
commit_message=f"Add new images and update gallery.json at {final_target_path}",
allow_patterns=["*.webp", "*.json"]
)
self.log_showcase("🎉 上传完毕！")
showcase_url = f"https://huggingface.co/{MODEL_REPO_ID}/tree/main/{final_target_path}"
self.after(0, lambda: self.sc_result_var.set(showcase_url))
self.after(0, lambda: messagebox.showinfo("完成", "Showcase 上传成功！"))
self.after(0, lambda: self.status_sc_var.set("上传完成"))
except Exception as e:
self.log_exc("Showcase 流水线异常", is_model=True)
self.after(0, lambda: self.status_sc_var.set("执行失败"))
finally:
self.after(0, lambda: self.btn_sc_process.configure(state="normal"))
self.after(0, lambda: self.btn_sc_upload.configure(state="normal"))
threading.Thread(target=worker, daemon=True).start()
if __name__ == "__main__":
app = ModelsManager()
app.mainloop()
|