File size: 112,435 Bytes
4d0ec21
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1e24ec7
4d0ec21
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1e24ec7
4d0ec21
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""Hugging Face Space entry point for the CKB prediction package.

The public Space serves two stable programmatic endpoints (``/predict`` and
``/warmup``) as well as a browser-first Gradio experience. Model artifacts are
loaded lazily and never retain a visitor's submitted data.
"""

from __future__ import annotations

import html
import shutil
import tempfile
import threading
import time
import uuid
from pathlib import Path
from typing import Any
from urllib.parse import quote

import gradio as gr
import pandas as pd
import spaces

from predictions.predict_single import CKBSingleSamplePredictor


MODEL_DIR = Path(__file__).resolve().parent / "predictions"
SHAP_CACHE_DIR = Path(tempfile.gettempdir()) / "ckb_space_shap"
SUPPLEMENTARY_IMAGE_DIR = Path(__file__).resolve().parent / "assets" / "supplementary"
SHAP_CACHE_TTL_SECONDS = 60 * 60
SHAP_LIGHT_CANVAS_MARKER = "ckb-force-plot-light-canvas"
SHAP_LIGHT_CANVAS_HEAD = f"""
<meta name="color-scheme" content="light">
<style id="{SHAP_LIGHT_CANVAS_MARKER}">
  html {{ background: #ffffff !important; color-scheme: only light !important; }}
  body {{ min-height: 100%; margin: 0; background: #ffffff !important; color: #111827 !important; color-scheme: only light !important; }}
</style>
"""

FEATURES = [
    "sex",
    "age",
    "edu_level",
    "marital_status",
    "work",
    "retire",
    "hh_size",
    "smoking",
    "alcohol",
    "height_cm",
    "weight_kg",
    "waist_cm",
    "sbp_mmhg",
    "dbp_mmhg",
    "bp_drugs",
    "self_health",
    "chronic_pain",
]

FEATURE_LABELS = {
    "sex": "性别编码",
    "age": "年龄(岁)",
    "edu_level": "教育程度编码",
    "marital_status": "婚姻状态编码",
    "work": "工作状态编码",
    "retire": "退休状态编码",
    "hh_size": "家庭人口数",
    "smoking": "吸烟编码",
    "alcohol": "饮酒编码",
    "height_cm": "身高(cm)",
    "weight_kg": "体重(kg)",
    "waist_cm": "腰围(cm)",
    "sbp_mmhg": "收缩压(mmHg)",
    "dbp_mmhg": "舒张压(mmHg)",
    "bp_drugs": "降压药使用编码",
    "self_health": "自评健康编码",
    "chronic_pain": "慢性疼痛编码",
}

FEATURE_HELP = {
    "sex": "请使用训练数据中的原始分类编码。",
    "age": "以完整岁数填写。",
    "edu_level": "请使用训练数据中的原始分类编码。",
    "marital_status": "请使用训练数据中的原始分类编码。",
    "work": "请使用训练数据中的原始分类编码。",
    "retire": "请使用训练数据中的原始分类编码。",
    "hh_size": "同住家庭成员人数。",
    "smoking": "请使用训练数据中的原始分类编码。",
    "alcohol": "请使用训练数据中的原始分类编码。",
    "height_cm": "连续数值,单位为厘米。",
    "weight_kg": "连续数值,单位为千克。",
    "waist_cm": "连续数值,单位为厘米。",
    "sbp_mmhg": "连续数值,单位为 mmHg。",
    "dbp_mmhg": "连续数值,单位为 mmHg。",
    "bp_drugs": "请使用训练数据中的原始分类编码。",
    "self_health": "请使用训练数据中的原始分类编码。",
    "chronic_pain": "请使用训练数据中的原始分类编码。",
}

FEATURE_GROUPS = [
    (
        "01",
        "基本资料",
        "人口与社会特征",
        ["sex", "age", "edu_level", "marital_status", "work", "retire", "hh_size"],
    ),
    (
        "02",
        "生活与感受",
        "行为及健康自评",
        ["smoking", "alcohol", "self_health", "chronic_pain"],
    ),
    (
        "03",
        "体征与用药",
        "身体测量与血压信息",
        ["height_cm", "weight_kg", "waist_cm", "sbp_mmhg", "dbp_mmhg", "bp_drugs"],
    ),
]

INTEGER_FEATURES = {
    "sex",
    "age",
    "edu_level",
    "marital_status",
    "work",
    "retire",
    "hh_size",
    "smoking",
    "alcohol",
    "sbp_mmhg",
    "dbp_mmhg",
    "bp_drugs",
    "self_health",
    "chronic_pain",
}

# Display text is deliberately human-readable; the paired values remain the
# frozen numeric encodings expected by the trained model.
CATEGORICAL_CHOICES = {
    "sex": [("Male", 0), ("Female", 1)],
    "edu_level": [("Low", 1), ("Intermediate", 2), ("High", 3)],
    "marital_status": [
        ("Other observed status", 0),
        ("Married or partnered", 1),
    ],
    "work": [("No", 0), ("Yes", 1)],
    "retire": [("No", 0), ("Yes", 1)],
    "smoking": [("No", 0), ("Yes", 1)],
    "alcohol": [("No", 0), ("Yes", 1)],
    "bp_drugs": [("No", 0), ("Yes", 1)],
    "self_health": [
        ("Very good or excellent", 1),
        ("Good", 2),
        ("Fair or regular", 3),
        ("Poor or very poor", 4),
    ],
    "chronic_pain": [("No", 0), ("Yes", 1)],
}

EXAMPLE_VALUES = {
    "sex": 0,
    "age": 59,
    "edu_level": 1,
    "marital_status": 1,
    "work": 1,
    "retire": 0,
    "hh_size": 5,
    "smoking": 1,
    "alcohol": 1,
    "height_cm": 163.8,
    "weight_kg": 59.6,
    "waist_cm": 81.1,
    "sbp_mmhg": 125,
    "dbp_mmhg": 67,
    "bp_drugs": 0,
    "self_health": 2,
    "chronic_pain": 0,
}

TABLE_COLUMNS = ["代码", "预测结局", "概率", "冻结阈值", "风险判定"]


DEFAULT_EN_FEATURE_LABELS = {
    "sex": "Sex",
    "age": "Age (years)",
    "edu_level": "Education level",
    "marital_status": "Married or partnered",
    "work": "Currently working",
    "retire": "Retired",
    "hh_size": "Household size",
    "smoking": "Smoking",
    "alcohol": "Alcohol use",
    "height_cm": "Height (cm)",
    "weight_kg": "Weight (kg)",
    "waist_cm": "Waist (cm)",
    "sbp_mmhg": "Systolic BP (mmHg)",
    "dbp_mmhg": "Diastolic BP (mmHg)",
    "bp_drugs": "Blood-pressure medication",
    "self_health": "Self-rated health",
    "chronic_pain": "Chronic pain",
}

DEFAULT_EN_FEATURE_HELP = {feature: "" for feature in FEATURES}

TABLE_COLUMNS_EN = ["Code", "Outcome", "Probability", "Frozen threshold", "Risk decision"]


# Presentation-only labels and cumulative-incidence reference values supplied
# for the public demo. They do not participate in model inference or risk
# classification, which always uses the frozen per-outcome Youden threshold.
OUTCOME_PRESENTATION = {
    "A00_B99": {
        "en": "Infectious and parasitic diseases",
        "zh": "感染性疾病",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 0.7,
        "low_10": 1.9,
        "high_5": 1.5,
        "high_10": 4.6,
    },
    "C00_D48": {
        "en": "Neoplasms",
        "zh": "肿瘤",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 1.9,
        "low_10": 4.8,
        "high_5": 4.3,
        "high_10": 9.5,
    },
    "E00_E90": {
        "en": "Endocrine and metabolic diseases",
        "zh": "内分泌代谢疾病",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 1.2,
        "low_10": 3.6,
        "high_5": 5.2,
        "high_10": 12.4,
    },
    "F00_F99": {
        "en": "Mental and behavioural disorders",
        "zh": "精神和行为障碍",
        "model_en": "Random Forest",
        "model_zh": "随机森林",
        "low_5": 0.2,
        "low_10": 0.5,
        "high_5": 0.6,
        "high_10": 1.3,
    },
    "G00_G99": {
        "en": "Diseases of the nervous system",
        "zh": "神经系统疾病",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 0.6,
        "low_10": 2.7,
        "high_5": 2.5,
        "high_10": 7.7,
    },
    "H00_H95": {
        "en": "Diseases of the eye and ear",
        "zh": "眼耳疾病",
        "model_en": "Random Forest",
        "model_zh": "随机森林",
        "low_5": 0.6,
        "low_10": 2.4,
        "high_5": 2.6,
        "high_10": 8.9,
    },
    "I00_I99": {
        "en": "Diseases of the circulatory system",
        "zh": "循环系统疾病",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 2.6,
        "low_10": 9.9,
        "high_5": 16.1,
        "high_10": 38.1,
    },
    "J00_J99": {
        "en": "Diseases of the respiratory system",
        "zh": "呼吸系统疾病",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 2.0,
        "low_10": 6.4,
        "high_5": 5.9,
        "high_10": 18.6,
    },
    "K00_K93": {
        "en": "Diseases of the digestive system",
        "zh": "消化系统疾病",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 3.1,
        "low_10": 9.2,
        "high_5": 5.6,
        "high_10": 16.0,
    },
    "L00_L99": {
        "en": "Diseases of the skin",
        "zh": "皮肤病",
        "model_en": "Random Forest",
        "model_zh": "随机森林",
        "low_5": 0.1,
        "low_10": 0.5,
        "high_5": 0.3,
        "high_10": 1.2,
    },
    "M00_M99": {
        "en": "Musculoskeletal diseases",
        "zh": "肌肉骨骼疾病",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 1.2,
        "low_10": 5.5,
        "high_5": 2.5,
        "high_10": 11.0,
    },
    "N00_N99": {
        "en": "Diseases of the genitourinary system",
        "zh": "泌尿生殖系统疾病",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 1.6,
        "low_10": 4.6,
        "high_5": 3.2,
        "high_10": 9.2,
    },
    "death": {
        "en": "All-cause death",
        "zh": "全因死亡",
        "model_en": "MLP",
        "model_zh": "MLP",
        "low_5": 0.9,
        "low_10": 2.4,
        "high_5": 8.0,
        "high_10": 20.2,
    },
}

# Keep Chinese presentation copy in unicode escapes so it remains intact when
# the Space is edited from Windows shells with different console code pages.
_OUTCOME_ZH = {
    "A00_B99": "\u611f\u67d3\u6027\u75be\u75c5",
    "C00_D48": "\u80bf\u7624",
    "E00_E90": "\u5185\u5206\u6ccc\u4ee3\u8c22\u75be\u75c5",
    "F00_F99": "\u7cbe\u795e\u548c\u884c\u4e3a\u969c\u788d",
    "G00_G99": "\u795e\u7ecf\u7cfb\u7edf\u75be\u75c5",
    "H00_H95": "\u773c\u8033\u75be\u75c5",
    "I00_I99": "\u5faa\u73af\u7cfb\u7edf\u75be\u75c5",
    "J00_J99": "\u547c\u5438\u7cfb\u7edf\u75be\u75c5",
    "K00_K93": "\u6d88\u5316\u7cfb\u7edf\u75be\u75c5",
    "L00_L99": "\u76ae\u80a4\u75c5",
    "M00_M99": "\u808c\u8089\u9aa8\u9abc\u75be\u75c5",
    "N00_N99": "\u6ccc\u5c3f\u751f\u6b96\u7cfb\u7edf\u75be\u75c5",
    "death": "\u5168\u56e0\u6b7b\u4ea1",
}
for _outcome_code, _outcome_name_zh in _OUTCOME_ZH.items():
    OUTCOME_PRESENTATION[_outcome_code]["zh"] = _outcome_name_zh
OUTCOME_PRESENTATION["F00_F99"]["model_zh"] = "\u968f\u673a\u68ee\u6797"
OUTCOME_PRESENTATION["H00_H95"]["model_zh"] = "\u968f\u673a\u68ee\u6797"
OUTCOME_PRESENTATION["L00_L99"]["model_zh"] = "\u968f\u673a\u68ee\u6797"


CUSTOM_CSS = """
:root {
  --ckb-blue-950: #082f49;
  --ckb-blue-900: #0c4a6e;
  --ckb-blue-700: #0369a1;
  --ckb-blue-600: #0284c7;
  --ckb-blue-100: #e0f2fe;
  --ckb-blue-50: #f0f9ff;
  --ckb-cyan-100: #cffafe;
  --ckb-green-700: #15803d;
  --ckb-green-600: #16a34a;
  --ckb-green-100: #dcfce7;
  --ckb-amber-700: #b45309;
  --ckb-amber-100: #fef3c7;
  --ckb-slate-950: #0f172a;
  --ckb-slate-700: #334155;
  --ckb-slate-600: #475569;
  --ckb-slate-400: #94a3b8;
  --ckb-slate-200: #e2e8f0;
  --ckb-slate-100: #f1f5f9;
  --ckb-white: #ffffff;
  --ckb-shadow-sm: 0 1px 2px rgb(15 23 42 / 0.05);
  --ckb-shadow-md: 0 16px 40px rgb(12 74 110 / 0.10);
  --ckb-radius-lg: 20px;
  --ckb-radius-md: 14px;
  --ckb-radius-sm: 10px;
  --ckb-motion-fast: 180ms;
  --ckb-motion-base: 280ms;
  --ckb-layer-field: 10;
  --ckb-layer-dropdown: 40;
}

.gradio-container {
  background:
    radial-gradient(circle at 10% 0%, rgb(207 250 254 / 0.75), transparent 24rem),
    radial-gradient(circle at 90% 9%, rgb(224 242 254 / 0.95), transparent 25rem),
    var(--ckb-blue-50);
  color: var(--ckb-slate-950);
  color-scheme: light !important;
  font-family: Aptos, "Segoe UI", "Microsoft YaHei UI", "Noto Sans SC", sans-serif;
}

#ckb-shell {
  width: min(1180px, 100%);
  margin: 0 auto;
  padding: 24px 18px 52px;
  color: var(--ckb-slate-950);
}

#ckb-shell * { box-sizing: border-box; }

.ckb-skip {
  position: absolute;
  left: -10000px;
  top: auto;
  overflow: hidden;
  width: 1px;
  height: 1px;
}

.ckb-skip:focus {
  left: 18px;
  top: 18px;
  z-index: 1000;
  width: auto;
  height: auto;
  padding: 10px 14px;
  border-radius: var(--ckb-radius-sm);
  background: var(--ckb-blue-950);
  color: var(--ckb-white);
  font-weight: 700;
}

.ckb-hero {
  position: relative;
  isolation: isolate;
  overflow: hidden;
  min-height: 286px;
  padding: clamp(30px, 6vw, 68px);
  border: 1px solid rgb(255 255 255 / 0.28);
  border-radius: 28px;
  background:
    linear-gradient(123deg, var(--ckb-blue-950) 0%, var(--ckb-blue-900) 56%, var(--ckb-blue-700) 100%);
  box-shadow: var(--ckb-shadow-md);
  color: var(--ckb-white) !important;
}

.ckb-hero::before,
.ckb-hero::after {
  position: absolute;
  z-index: -1;
  display: block;
  border-radius: 999px;
  content: "";
  filter: blur(2px);
}

.ckb-hero::before {
  top: -138px;
  right: -58px;
  width: 322px;
  height: 322px;
  border: 1px solid rgb(207 250 254 / 0.30);
  background: radial-gradient(circle, rgb(34 211 238 / 0.30), transparent 67%);
}

.ckb-hero::after {
  right: 18%;
  bottom: -185px;
  width: 360px;
  height: 360px;
  background: radial-gradient(circle, rgb(22 163 74 / 0.24), transparent 67%);
}

.ckb-hero__content { position: relative; max-width: 720px; }

.ckb-eyebrow,
.ckb-chip,
.ckb-section-kicker,
.ckb-result-kicker {
  display: inline-flex;
  align-items: center;
  gap: 8px;
  font-size: 0.78rem;
  font-weight: 750;
  letter-spacing: 0.08em;
  line-height: 1;
  text-transform: uppercase;
}

.ckb-hero .ckb-eyebrow {
  color: var(--ckb-cyan-100) !important;
  -webkit-text-fill-color: var(--ckb-cyan-100);
}

.ckb-mark {
  display: inline-grid;
  width: 28px;
  height: 28px;
  place-items: center;
  border: 1px solid rgb(207 250 254 / 0.36);
  border-radius: 9px;
  background: rgb(255 255 255 / 0.10);
}

.ckb-mark svg { width: 17px; height: 17px; stroke: currentColor; }

.ckb-hero h1 {
  max-width: 13ch;
  margin: 16px 0 14px;
  color: var(--ckb-white) !important;
  -webkit-text-fill-color: var(--ckb-white);
  font-family: Georgia, "Noto Serif SC", serif;
  font-size: clamp(2.15rem, 5.1vw, 4.25rem);
  font-weight: 650;
  letter-spacing: -0.045em;
  line-height: 1.04;
}

.ckb-hero p {
  max-width: 62ch;
  margin: 0;
  color: #f0f9ff !important;
  -webkit-text-fill-color: #f0f9ff;
  font-size: clamp(1rem, 1.4vw, 1.12rem);
  line-height: 1.7;
}

.ckb-hero__meta {
  display: flex;
  flex-wrap: wrap;
  gap: 10px;
  margin-top: 26px;
}

.ckb-chip {
  min-height: 34px;
  padding: 8px 11px;
  border: 1px solid rgb(255 255 255 / 0.16);
  border-radius: 999px;
  background: rgb(255 255 255 / 0.10);
  color: var(--ckb-white) !important;
  -webkit-text-fill-color: var(--ckb-white);
  letter-spacing: 0.02em;
  text-transform: none;
}

.ckb-chip svg { width: 14px; height: 14px; fill: none; stroke: currentColor; stroke-width: 2; }

.ckb-trust-row {
  display: grid;
  grid-template-columns: repeat(3, minmax(0, 1fr));
  gap: 12px;
  margin: 18px 0 24px;
}

.ckb-trust-card {
  min-width: 0;
  padding: 16px 18px;
  border: 1px solid var(--ckb-blue-100);
  border-radius: var(--ckb-radius-md);
  background: rgb(255 255 255 / 0.78);
  box-shadow: var(--ckb-shadow-sm);
}

.ckb-trust-card strong {
  display: block;
  color: var(--ckb-blue-900);
  font-family: Georgia, "Noto Serif SC", serif;
  font-size: 1.42rem;
  font-variant-numeric: tabular-nums;
  line-height: 1.1;
}

.ckb-trust-card span {
  display: block;
  margin-top: 5px;
  color: var(--ckb-slate-600);
  font-size: 0.86rem;
  line-height: 1.35;
}

.ckb-language-bar {
  display: flex;
  align-items: center;
  flex-wrap: wrap;
  gap: 8px;
  justify-content: flex-end;
  margin: -4px 0 14px;
}

.ckb-language-switch,
.ckb-theme-switch {
  display: inline-flex;
  gap: 4px;
  padding: 4px;
  border: 1px solid var(--ckb-blue-100);
  border-radius: 999px;
  background: rgb(255 255 255 / 0.88);
  box-shadow: var(--ckb-shadow-sm);
}

.ckb-language-switch button,
.ckb-theme-switch button {
  min-width: 74px;
  min-height: 36px;
  padding: 7px 12px;
  border: 0;
  border-radius: 999px;
  background: transparent;
  color: var(--ckb-slate-600);
  cursor: pointer;
  font: inherit;
  font-size: 0.82rem;
  font-weight: 750;
  transition: background var(--ckb-motion-fast) ease, color var(--ckb-motion-fast) ease, transform var(--ckb-motion-fast) ease;
}

.ckb-theme-switch button { min-width: 58px; }
.ckb-language-switch button:hover,
.ckb-theme-switch button:hover { background: var(--ckb-blue-50); color: var(--ckb-blue-900); }
.ckb-language-switch button:active,
.ckb-theme-switch button:active { transform: scale(0.98); }
.ckb-language-switch button[aria-pressed="true"],
.ckb-theme-switch button[aria-pressed="true"] { background: var(--ckb-blue-900); color: var(--ckb-white); }
.ckb-language-switch button:focus-visible,
.ckb-theme-switch button:focus-visible { outline: 3px solid rgb(2 132 199 / 0.32); outline-offset: 2px; }

.ckb-status {
  display: flex;
  align-items: flex-start;
  gap: 10px;
  min-height: 52px;
  margin: 0 0 20px;
  padding: 14px 16px;
  border: 1px solid var(--ckb-blue-100);
  border-radius: var(--ckb-radius-md);
  background: rgb(255 255 255 / 0.82);
  color: var(--ckb-slate-700);
  box-shadow: var(--ckb-shadow-sm);
  line-height: 1.5;
}

.ckb-status__dot {
  flex: 0 0 auto;
  width: 10px;
  height: 10px;
  margin-top: 7px;
  border-radius: 50%;
  background: var(--ckb-blue-600);
  box-shadow: 0 0 0 4px var(--ckb-blue-100);
}

.ckb-status--ready { border-color: var(--ckb-green-100); background: #f6fff8; }
.ckb-status--ready .ckb-status__dot { background: var(--ckb-green-600); box-shadow: 0 0 0 4px var(--ckb-green-100); }
.ckb-status--warning { border-color: var(--ckb-amber-100); background: #fffbeb; }
.ckb-status--warning .ckb-status__dot { background: var(--ckb-amber-700); box-shadow: 0 0 0 4px var(--ckb-amber-100); }

.ckb-panel {
  padding: clamp(22px, 3vw, 34px);
  border: 1px solid var(--ckb-blue-100);
  border-radius: var(--ckb-radius-lg);
  background: rgb(255 255 255 / 0.88);
  box-shadow: var(--ckb-shadow-md);
}

.ckb-section-heading {
  display: flex;
  align-items: flex-start;
  justify-content: space-between;
  gap: 16px;
  margin-bottom: 22px;
}

.ckb-section-kicker { color: var(--ckb-blue-700); }

.ckb-section-heading h2 {
  margin: 7px 0 6px;
  color: var(--ckb-slate-950);
  font-family: Georgia, "Noto Serif SC", serif;
  font-size: clamp(1.55rem, 3vw, 2.1rem);
  letter-spacing: -0.025em;
  line-height: 1.16;
}

.ckb-section-heading p {
  max-width: 58ch;
  margin: 0;
  color: var(--ckb-slate-600);
  font-size: 0.95rem;
  line-height: 1.6;
}

.ckb-coding-note {
  flex: 0 0 auto;
  max-width: 265px;
  padding: 10px 12px;
  border-radius: var(--ckb-radius-sm);
  background: var(--ckb-blue-50);
  color: var(--ckb-blue-900);
  font-size: 0.82rem;
  font-weight: 650;
  line-height: 1.45;
}

.ckb-input-grid { gap: 14px; }

.ckb-form-card {
  height: 100%;
  padding: 17px;
  border: 1px solid var(--ckb-slate-200);
  border-radius: var(--ckb-radius-md);
  background: var(--ckb-white);
  box-shadow: var(--ckb-shadow-sm);
}

.ckb-form-card .form-card__top {
  display: flex;
  align-items: flex-start;
  gap: 10px;
  margin-bottom: 14px;
}

.ckb-form-card .form-card__index {
  display: inline-grid;
  flex: 0 0 auto;
  width: 30px;
  height: 30px;
  place-items: center;
  border-radius: 9px;
  background: var(--ckb-blue-100);
  color: var(--ckb-blue-900);
  font-size: 0.78rem;
  font-weight: 800;
  font-variant-numeric: tabular-nums;
}

.ckb-form-card h3 { margin: 1px 0 3px; color: var(--ckb-slate-950); font-size: 1rem; }
.ckb-form-card p { margin: 0; color: var(--ckb-slate-600); font-size: 0.8rem; line-height: 1.4; }

#ckb-shell .ckb-form-card .block { min-width: 0; margin-bottom: 10px; }
#ckb-shell .ckb-form-card label { color: var(--ckb-slate-700); font-size: 0.86rem; font-weight: 700; }
#ckb-shell .ckb-form-card input {
  min-height: 48px;
  border: 1px solid var(--ckb-slate-200);
  border-radius: var(--ckb-radius-sm);
  background: var(--ckb-white);
  color: var(--ckb-slate-950);
  font-size: 1rem;
  font-variant-numeric: tabular-nums;
  transition: border-color var(--ckb-motion-fast) ease, box-shadow var(--ckb-motion-fast) ease;
}

#ckb-shell .ckb-form-card input:hover { border-color: var(--ckb-blue-600); }
#ckb-shell .ckb-form-card input:focus {
  border-color: var(--ckb-blue-600);
  box-shadow: 0 0 0 3px rgb(2 132 199 / 0.20);
  outline: none;
}

.ckb-actions {
  display: flex;
  flex-wrap: wrap;
  align-items: center;
  gap: 12px;
  margin-top: 24px;
}

#predict-button,
#reset-button {
  min-height: 48px;
  border-radius: var(--ckb-radius-sm);
  font-weight: 750;
  transition: transform var(--ckb-motion-fast) ease, box-shadow var(--ckb-motion-fast) ease, background var(--ckb-motion-fast) ease;
}

#predict-button {
  min-width: 172px;
  border: 0;
  background: var(--ckb-green-600);
  box-shadow: 0 10px 22px rgb(22 163 74 / 0.22);
  color: var(--ckb-white);
}

#predict-button:hover { background: var(--ckb-green-700); box-shadow: 0 13px 26px rgb(22 163 74 / 0.29); transform: translateY(-1px); }
#predict-button:active { transform: translateY(0) scale(0.985); }
#predict-button:focus-visible,
#reset-button:focus-visible { outline: 3px solid rgb(2 132 199 / 0.35); outline-offset: 3px; }

#reset-button {
  border: 1px solid var(--ckb-slate-200);
  background: var(--ckb-white);
  color: var(--ckb-slate-700);
}

#reset-button:hover { border-color: var(--ckb-blue-600); color: var(--ckb-blue-900); transform: translateY(-1px); }

.ckb-privacy-copy {
  display: flex;
  flex: 1 1 250px;
  align-items: center;
  gap: 8px;
  margin: 0;
  color: var(--ckb-slate-600);
  font-size: 0.82rem;
  line-height: 1.45;
}

.ckb-privacy-copy svg { flex: 0 0 auto; width: 17px; height: 17px; stroke: var(--ckb-blue-700); stroke-width: 2; }

.ckb-results {
  margin-top: 26px;
  padding: clamp(22px, 3vw, 34px);
  border: 1px solid var(--ckb-blue-100);
  border-radius: var(--ckb-radius-lg);
  background: rgb(255 255 255 / 0.90);
  box-shadow: var(--ckb-shadow-md);
}

.ckb-results-header { margin-bottom: 17px; }
.ckb-results-header h2 { margin: 7px 0 0; color: var(--ckb-slate-950); font-family: Georgia, "Noto Serif SC", serif; font-size: clamp(1.5rem, 3vw, 2rem); }

.ckb-result-card,
.ckb-result-placeholder {
  min-height: 122px;
  padding: 20px;
  border: 1px solid var(--ckb-blue-100);
  border-radius: var(--ckb-radius-md);
  background: linear-gradient(112deg, var(--ckb-blue-50), var(--ckb-white));
}

.ckb-result-placeholder { display: grid; place-items: center; color: var(--ckb-slate-600); text-align: center; }

.ckb-result-card__header { display: flex; align-items: flex-start; justify-content: space-between; gap: 18px; }
.ckb-result-kicker { color: var(--ckb-blue-700); }
.ckb-result-card h3 { margin: 6px 0 0; color: var(--ckb-blue-950); font-family: Georgia, "Noto Serif SC", serif; font-size: 1.7rem; }
.ckb-result-badge { flex: 0 0 auto; padding: 8px 10px; border-radius: 999px; background: var(--ckb-green-100); color: var(--ckb-green-700); font-size: 0.8rem; font-weight: 800; }
.ckb-result-badge--review { background: var(--ckb-amber-100); color: var(--ckb-amber-700); }

.ckb-result-metrics {
  display: grid;
  grid-template-columns: repeat(2, minmax(0, 1fr));
  gap: 10px;
  margin: 18px 0 0;
}

.ckb-result-metrics div { padding: 11px; border-radius: var(--ckb-radius-sm); background: rgb(255 255 255 / 0.76); }
.ckb-result-metrics dt { color: var(--ckb-slate-600); font-size: 0.76rem; }
.ckb-result-metrics dd { margin: 4px 0 0; color: var(--ckb-slate-950); font-size: 1rem; font-weight: 800; font-variant-numeric: tabular-nums; }

.ckb-risk-visual { margin-top: 18px; padding: 20px; border: 1px solid var(--ckb-slate-200); border-radius: var(--ckb-radius-md); background: var(--ckb-white); }
.ckb-risk-visual__header { display: flex; align-items: flex-start; justify-content: space-between; gap: 14px; margin-bottom: 16px; }
.ckb-risk-visual h3 { margin: 0; color: var(--ckb-slate-950); font-size: 1.05rem; }
.ckb-risk-visual p { margin: 5px 0 0; color: var(--ckb-slate-600); font-size: 0.82rem; line-height: 1.45; }
.ckb-legend { display: flex; flex-wrap: wrap; gap: 10px; color: var(--ckb-slate-600); font-size: 0.76rem; }
.ckb-legend span { display: inline-flex; align-items: center; gap: 5px; }
.ckb-legend i { width: 9px; height: 9px; border-radius: 50%; background: var(--ckb-green-600); }
.ckb-legend .ckb-legend--high { background: var(--ckb-amber-700); }

.ckb-risk-row { display: grid; grid-template-columns: minmax(88px, 0.8fr) minmax(130px, 2fr) 54px; gap: 10px; align-items: center; margin: 10px 0; }
.ckb-risk-label { overflow: hidden; color: var(--ckb-slate-700); font-size: 0.78rem; font-weight: 700; text-overflow: ellipsis; white-space: nowrap; }
.ckb-risk-track { position: relative; height: 11px; overflow: hidden; border-radius: 999px; background: var(--ckb-slate-100); }
.ckb-risk-fill { display: block; width: var(--risk-width); height: 100%; border-radius: inherit; background: var(--ckb-green-600); }
.ckb-risk-row--high .ckb-risk-fill { background: var(--ckb-amber-700); }
.ckb-risk-threshold { position: absolute; top: 0; bottom: 0; left: var(--threshold); width: 2px; background: var(--ckb-slate-950); opacity: 0.58; }
.ckb-risk-value { color: var(--ckb-slate-950); font-size: 0.8rem; font-weight: 800; font-variant-numeric: tabular-nums; text-align: right; }

#outcome-table { margin-top: 18px; overflow-x: auto; }
#outcome-table table { min-width: 680px; border-collapse: separate; border-spacing: 0; }
#outcome-table thead { background: var(--ckb-blue-50); }
#outcome-table th { color: var(--ckb-blue-900); font-size: 0.78rem; font-weight: 800; }
#outcome-table td { color: var(--ckb-slate-700); font-size: 0.84rem; }
#outcome-table th, #outcome-table td { border-color: var(--ckb-slate-200); }

#shap-result { overflow: hidden !important; margin-top: 18px; border: 1px solid var(--ckb-blue-700); border-radius: var(--ckb-radius-md); background: var(--ckb-white); box-shadow: var(--ckb-shadow-sm); }
#shap-result > button {
  position: relative;
  display: flex !important;
  align-items: center;
  width: 100%;
  min-height: 64px;
  padding: 14px 68px 14px 22px !important;
  border: 0 !important;
  border-radius: var(--ckb-radius-md);
  background: linear-gradient(112deg, var(--ckb-blue-950), var(--ckb-blue-700));
  color: var(--ckb-white) !important;
  cursor: pointer;
  font-size: 0.98rem;
  font-weight: 800;
  letter-spacing: 0.01em;
  line-height: 1.35;
  text-align: left;
  touch-action: manipulation;
  transition: background 180ms ease, box-shadow 180ms ease, transform 180ms ease;
}
#shap-result > button::after {
  position: absolute;
  right: 26px;
  width: 11px;
  height: 11px;
  border-right: 2px solid currentColor;
  border-bottom: 2px solid currentColor;
  content: "";
  transform: rotate(45deg) translateY(-3px);
  transform-origin: center;
  transition: transform 180ms ease;
}
#shap-result > button:hover { background: linear-gradient(112deg, var(--ckb-blue-900), var(--ckb-blue-600)); box-shadow: inset 0 0 0 999px rgb(255 255 255 / 0.05); }
#shap-result > button:active { transform: scale(0.995); }
#shap-result > button:focus-visible { outline: 3px solid rgb(2 132 199 / 0.42); outline-offset: -5px; }
#shap-result > button.open { border-bottom: 1px solid var(--ckb-blue-100) !important; border-radius: var(--ckb-radius-md) var(--ckb-radius-md) 0 0; background: var(--ckb-blue-50); color: var(--ckb-blue-950) !important; }
#shap-result > button.open::after { transform: rotate(225deg) translateY(-3px); }
.ckb-shap-placeholder { display: grid; min-height: 108px; place-items: center; color: var(--ckb-slate-600); text-align: center; }
.ckb-shap-gallery { padding: 4px 2px 2px; }
.ckb-shap-gallery__header { margin: 0 0 16px; }
.ckb-shap-gallery__header .ckb-section-kicker { color: var(--ckb-blue-700); }
.ckb-shap-gallery__header h3 { margin: 6px 0 0; color: var(--ckb-slate-950); font-size: 1.2rem; }
.ckb-shap-gallery__header p { margin: 6px 0 0; color: var(--ckb-slate-600); font-size: 0.84rem; line-height: 1.5; }
.ckb-shap-grid { display: grid; grid-template-columns: repeat(2, minmax(0, 1fr)); gap: 14px; }
.ckb-shap-card { min-width: 0; overflow: hidden; border: 1px solid var(--ckb-slate-200); border-radius: var(--ckb-radius-sm); background: var(--ckb-white); box-shadow: var(--ckb-shadow-sm); }
.ckb-shap-card--cluster { grid-column: 1 / -1; }
.ckb-shap-card__header { padding: 13px 14px 10px; border-bottom: 1px solid var(--ckb-slate-100); background: linear-gradient(112deg, var(--ckb-blue-50), var(--ckb-white)); }
.ckb-shap-card__kind { margin: 0; color: var(--ckb-blue-700); font-size: 0.72rem; font-weight: 800; letter-spacing: 0.06em; text-transform: uppercase; }
.ckb-shap-card h4 { margin: 4px 0 0; color: var(--ckb-slate-950); font-size: 0.94rem; line-height: 1.35; }
.ckb-shap-frame { display: block; width: 100%; height: 330px; border: 0; background: var(--ckb-white); }

#raw-result { margin-top: 16px; border: 1px solid var(--ckb-slate-200); border-radius: var(--ckb-radius-sm); background: var(--ckb-white); }
#raw-result button { min-height: 44px; color: var(--ckb-slate-700); font-weight: 750; }

.ckb-footer-note { margin: 20px 0 0; color: var(--ckb-slate-600); font-size: 0.82rem; line-height: 1.6; }
.ckb-footer-note code { padding: 2px 5px; border-radius: 5px; background: var(--ckb-slate-100); color: var(--ckb-blue-900); }

.ckb-hero, .ckb-trust-card, .ckb-panel, .ckb-results { will-change: auto; }

/* The hero now carries one concise model statement; the old labels and count
   tiles remain in the DOM only so legacy locale initialisation stays stable. */
.ckb-eyebrow, .ckb-hero__meta, .ckb-trust-row { display: none !important; }
.ckb-hero { min-height: 230px; display: grid; align-items: center; }
.ckb-hero p { max-width: 66ch; }
.ckb-section-heading > div > p, .ckb-coding-note { display: none !important; }

/* Gradio renders dropdown lists in a popover. Keeping animated ancestors
   transform-free and elevating the portal fixes the offset seen in the form. */
#ckb-shell .ckb-form-card .block,
#ckb-shell .ckb-form-card .wrap { position: relative; overflow: visible !important; }
#ckb-shell .ckb-form-card .block { z-index: 0; }
#ckb-shell .ckb-form-card .block:focus-within { z-index: var(--ckb-layer-field); }
body .secondary-wrap,
body [role="listbox"] { z-index: var(--ckb-layer-dropdown) !important; }
#ckb-shell .ckb-form-card button,
#ckb-shell .ckb-form-card [role="combobox"] { min-height: 48px; border-radius: var(--ckb-radius-sm); }
#ckb-shell,
#outcome-table,
#outcome-table .html-container,
#shap-result,
#shap-result .wrap,
#shap-result .form,
#shap-result .html-container,
#shap-gallery,
#shap-gallery .wrap,
#shap-gallery .html-container,
.ckb-image-gallery,
.ckb-image-card,
.ckb-image-card__force { color-scheme: light !important; }
.ckb-image-frame,
.ckb-reference-figure,
.ckb-cluster-profile-figure { color-scheme: only light !important; }
#shap-result .wrap,
#shap-result .form,
#shap-result .html-container,
#shap-gallery,
#shap-gallery .wrap,
#shap-gallery .html-container { background: var(--ckb-white) !important; color: var(--ckb-slate-950) !important; }
#ckb-shell .ckb-panel .ckb-section-heading,
#ckb-shell .ckb-panel .ckb-section-heading h2,
#ckb-shell .ckb-panel .ckb-section-heading p,
#ckb-shell .ckb-status,
#ckb-shell .ckb-status strong,
#ckb-shell .ckb-results-header,
#ckb-shell .ckb-results-header h2,
#ckb-shell .ckb-result-placeholder,
#ckb-shell .ckb-footer-note { color: var(--ckb-slate-950) !important; }
#ckb-shell .ckb-panel .ckb-section-heading .ckb-section-kicker,
#ckb-shell .ckb-results-header .ckb-section-kicker { color: var(--ckb-blue-700) !important; }
#ckb-shell .ckb-status { color: var(--ckb-slate-700) !important; }
#ckb-shell .ckb-status span,
#ckb-shell .ckb-status strong { color: inherit !important; }
#ckb-shell .ckb-result-placeholder,
#ckb-shell .ckb-footer-note { color: var(--ckb-slate-600) !important; }
#ckb-shell .ckb-footer-note code { color: var(--ckb-blue-900) !important; }
#ckb-shell .ckb-form-card,
#ckb-shell .ckb-form-card > .styler,
#ckb-shell .ckb-form-card .form,
#ckb-shell .ckb-form-card .container { background: var(--ckb-white) !important; color: var(--ckb-slate-950) !important; }
#ckb-shell .ckb-form-card .block { background: transparent !important; border-color: transparent !important; }
#ckb-shell .ckb-form-card .wrap,
#ckb-shell .ckb-form-card .wrap-inner,
#ckb-shell .ckb-form-card .secondary-wrap { background: var(--ckb-white) !important; border-color: var(--ckb-slate-200) !important; color: var(--ckb-slate-950) !important; }
#ckb-shell .ckb-form-card .form-card__index { color: var(--ckb-blue-900) !important; }
#ckb-shell .ckb-form-card h3 { color: var(--ckb-slate-950) !important; }
#ckb-shell .ckb-form-card p { color: var(--ckb-slate-600) !important; }
#ckb-shell .ckb-form-card [data-testid="block-info"],
#ckb-shell .ckb-form-card label { color: var(--ckb-slate-700) !important; }
body .options[role="listbox"] { border: 1px solid var(--ckb-slate-200) !important; border-radius: var(--ckb-radius-sm) !important; background: var(--ckb-white) !important; box-shadow: var(--ckb-shadow-md) !important; }
body .options[role="listbox"] [role="option"] { background: var(--ckb-white) !important; color: var(--ckb-slate-950) !important; }
body .options[role="listbox"] [role="option"]:hover,
body .options[role="listbox"] [role="option"][aria-selected="true"] { background: var(--ckb-blue-50) !important; color: var(--ckb-blue-950) !important; }

.ckb-outcome-table { margin-top: 18px; overflow: hidden; border: 1px solid var(--ckb-slate-200); border-radius: var(--ckb-radius-md); background: var(--ckb-white); box-shadow: var(--ckb-shadow-sm); }
.ckb-outcome-table__header { display: flex; align-items: flex-end; justify-content: space-between; gap: 18px; padding: 18px 20px 15px; border-bottom: 1px solid var(--ckb-slate-100); background: linear-gradient(112deg, var(--ckb-blue-50), var(--ckb-white)); }
.ckb-outcome-table__header h3 { margin: 5px 0 0; color: var(--ckb-slate-950); font-size: 1.08rem; }
.ckb-outcome-table__header > p { max-width: 50ch; margin: 0; color: var(--ckb-slate-600); font-size: 0.8rem; line-height: 1.45; text-align: right; }
.ckb-outcome-table__scroll { overflow-x: auto; }
.ckb-outcome-table table { width: 100%; min-width: 720px; border-collapse: collapse; }
.ckb-outcome-table th, .ckb-outcome-table td { padding: 12px 14px; border-bottom: 1px solid var(--ckb-slate-100); text-align: left; }
.ckb-outcome-table th { color: var(--ckb-blue-900); background: rgb(240 249 255 / 0.72); font-size: 0.76rem; font-weight: 850; letter-spacing: 0.025em; }
.ckb-outcome-table td { color: var(--ckb-slate-700); font-size: 0.84rem; vertical-align: middle; }
.ckb-outcome-table tbody tr:last-child td { border-bottom: 0; }
.ckb-outcome-table tbody tr:hover { background: var(--ckb-blue-50); }
.ckb-outcome-table__code { color: var(--ckb-slate-950) !important; font-weight: 800; font-variant-numeric: tabular-nums; white-space: nowrap; }
.ckb-outcome-table__decision { display: inline-flex; padding: 5px 8px; border-radius: 999px; background: var(--ckb-green-100); color: var(--ckb-green-700); font-size: 0.75rem; font-weight: 800; white-space: nowrap; }
.ckb-outcome-table__decision--high { background: var(--ckb-amber-100); color: var(--ckb-amber-700); }

.ckb-image-placeholder { display: grid; min-height: 108px; place-items: center; padding: 20px; color: var(--ckb-slate-600); text-align: center; }
.ckb-image-gallery { padding: 2px; }
.ckb-image-gallery__header { margin: 0 0 20px; }
.ckb-image-gallery__header h3 { margin: 6px 0 0; color: var(--ckb-slate-950); font-size: 1.2rem; }
.ckb-image-gallery__header p { max-width: 74ch; margin: 6px 0 0; color: var(--ckb-slate-600); font-size: 0.84rem; line-height: 1.5; }
.ckb-image-section + .ckb-image-section { margin-top: 22px; }
.ckb-image-section__header { margin: 0 0 10px; }
.ckb-image-section__header h4 { margin: 0; color: var(--ckb-blue-900); font-size: 0.9rem; font-weight: 850; }
.ckb-image-grid { display: grid; gap: 14px; }
.ckb-image-grid--clusters { grid-template-columns: 1fr; }
.ckb-image-grid--outcomes { grid-template-columns: repeat(2, minmax(0, 1fr)); }
.ckb-image-card { min-width: 0; overflow: hidden; border: 1px solid var(--ckb-slate-200); border-radius: var(--ckb-radius-sm); background: var(--ckb-white); box-shadow: var(--ckb-shadow-sm); }
.ckb-image-card--selected { border-color: var(--ckb-blue-600); box-shadow: 0 0 0 2px rgb(2 132 199 / 0.14), var(--ckb-shadow-sm); }
.ckb-image-card__header { display: flex; align-items: flex-start; justify-content: space-between; gap: 12px; padding: 12px 13px 10px; border-bottom: 1px solid var(--ckb-slate-100); background: linear-gradient(112deg, var(--ckb-blue-50), var(--ckb-white)); }
.ckb-image-card__kind { margin: 0; color: var(--ckb-blue-700); font-size: 0.7rem; font-weight: 850; letter-spacing: 0.06em; text-transform: uppercase; }
.ckb-image-card h4 { margin: 4px 0 0; color: var(--ckb-slate-950); font-size: 0.9rem; line-height: 1.35; }
.ckb-image-card__badge { flex: 0 0 auto; max-width: 50%; padding: 5px 8px; border-radius: 999px; background: var(--ckb-blue-100); color: var(--ckb-blue-900); font-size: 0.7rem; font-weight: 800; line-height: 1.2; text-align: center; }
.ckb-image-card__badge--high { background: var(--ckb-amber-100); color: var(--ckb-amber-700); }
.ckb-image-frame { display: block; width: 100%; height: 218px; border: 0; background: var(--ckb-white); }
.ckb-image-card__context { display: grid; grid-template-columns: minmax(205px, 0.9fr) minmax(0, 1.1fr); gap: 12px; align-items: stretch; padding: 12px; background: #fbfdff; }
.ckb-image-card__curve { min-width: 0; padding: 3px; border: 1px solid var(--ckb-slate-100); border-radius: 8px; background: var(--ckb-white); }
.ckb-reference-figure { display: block; width: 100%; height: 100%; min-height: 184px; object-fit: contain; border-radius: 6px; background: var(--ckb-white); }
.ckb-cluster-profile-figure { display: block; width: 100%; max-height: 230px; object-fit: contain; padding: 10px 12px; background: var(--ckb-white); }
.ckb-image-card__force { border-top: 1px solid var(--ckb-slate-100); background: var(--ckb-white); }
.ckb-risk-curve { display: block; width: 100%; height: auto; }
.ckb-risk-curve__grid path { fill: none; stroke: #dbeafe; stroke-width: 1; }
.ckb-risk-curve__low, .ckb-risk-curve__high { fill: none; stroke-linecap: round; stroke-linejoin: round; stroke-width: 3; }
.ckb-risk-curve__low { stroke: #0d9488; }
.ckb-risk-curve__high { stroke: #d97706; }
.ckb-risk-curve__legend text, .ckb-risk-curve__axis { fill: var(--ckb-slate-600); font-size: 10px; font-family: ui-sans-serif, system-ui, sans-serif; }
.ckb-risk-curve__point--low circle { fill: #0d9488; }.ckb-risk-curve__point--low text { fill: #0f766e; }.ckb-risk-curve__point--high circle { fill: #d97706; }.ckb-risk-curve__point--high text { fill: #b45309; }
.ckb-risk-curve__point text { font-size: 10px; font-weight: 800; font-family: ui-sans-serif, system-ui, sans-serif; }
.ckb-image-card__explanation { display: flex; flex-direction: column; justify-content: center; min-width: 0; }
.ckb-image-card__explanation p { margin: 0; color: var(--ckb-slate-600); font-size: 0.78rem; line-height: 1.55; }
.ckb-image-card__explanation .ckb-image-card__risk { margin-bottom: 7px; color: var(--ckb-blue-900); font-size: 0.76rem; font-weight: 850; }

/* Theme selection is explicit: the page never follows the browser/system theme. */
html[data-ckb-theme="light"] .gradio-container { color-scheme: light !important; }
#ckb-shell[data-theme="dark"] {
  --ckb-blue-950: #082f49;
  --ckb-blue-900: #0c4a6e;
  --ckb-blue-700: #7dd3fc;
  --ckb-blue-600: #38bdf8;
  --ckb-blue-100: #153b58;
  --ckb-blue-50: #0f1d31;
  --ckb-cyan-100: #164e63;
  --ckb-green-700: #bbf7d0;
  --ckb-green-600: #4ade80;
  --ckb-green-100: #14532d;
  --ckb-amber-700: #fcd34d;
  --ckb-amber-100: #78350f;
  --ckb-slate-950: #f8fafc;
  --ckb-slate-700: #e2e8f0;
  --ckb-slate-600: #cbd5e1;
  --ckb-slate-400: #94a3b8;
  --ckb-slate-200: #334155;
  --ckb-slate-100: #1e293b;
  --ckb-white: #111827;
  --ckb-shadow-sm: 0 1px 2px rgb(0 0 0 / 0.32);
  --ckb-shadow-md: 0 16px 40px rgb(0 0 0 / 0.34);
  color-scheme: dark !important;
}
html[data-ckb-theme="dark"] .gradio-container {
  background:
    radial-gradient(circle at 10% 0%, rgb(8 47 73 / 0.76), transparent 24rem),
    radial-gradient(circle at 90% 9%, rgb(15 35 57 / 0.9), transparent 25rem),
    #020617;
  color: #f8fafc;
  color-scheme: dark !important;
}
html[data-ckb-theme="dark"] #ckb-shell,
html[data-ckb-theme="dark"] #outcome-table,
html[data-ckb-theme="dark"] #shap-result,
html[data-ckb-theme="dark"] #shap-gallery,
html[data-ckb-theme="dark"] .ckb-image-gallery,
html[data-ckb-theme="dark"] .ckb-image-card,
html[data-ckb-theme="dark"] .ckb-image-card__force { color-scheme: dark !important; }
html[data-ckb-theme="dark"] #ckb-shell .ckb-panel,
html[data-ckb-theme="dark"] #ckb-shell .ckb-results { background: rgb(15 23 42 / 0.94); border-color: var(--ckb-slate-200); }
html[data-ckb-theme="dark"] #ckb-shell .ckb-status--ready { background: #102a1c; border-color: #166534; }
html[data-ckb-theme="dark"] #ckb-shell .ckb-status--warning { background: #422006; border-color: #92400e; }
html[data-ckb-theme="dark"] #ckb-shell .ckb-result-metrics div { background: #172033; }
html[data-ckb-theme="dark"] #ckb-shell .ckb-image-card__context { background: #111c2d; }
html[data-ckb-theme="dark"] #ckb-shell .ckb-language-switch,
html[data-ckb-theme="dark"] #ckb-shell .ckb-theme-switch { background: #111827; border-color: var(--ckb-slate-200); }
html[data-ckb-theme="dark"] #ckb-shell .ckb-language-switch button[aria-pressed="true"],
html[data-ckb-theme="dark"] #ckb-shell .ckb-theme-switch button[aria-pressed="true"],
html[data-ckb-theme="dark"] #ckb-shell #predict-button,
html[data-ckb-theme="dark"] #ckb-shell #shap-result > button { color: #ffffff !important; }
html[data-ckb-theme="dark"] #ckb-shell .ckb-hero,
html[data-ckb-theme="dark"] #ckb-shell .ckb-hero h1,
html[data-ckb-theme="dark"] #ckb-shell .ckb-hero p {
  color: #ffffff !important;
  -webkit-text-fill-color: #ffffff !important;
}
html[data-ckb-theme="dark"] #ckb-shell #shap-result > button.open { color: #e0f2fe !important; }
html[data-ckb-theme="dark"] #ckb-shell .ckb-image-frame,
html[data-ckb-theme="dark"] #ckb-shell .ckb-reference-figure,
html[data-ckb-theme="dark"] #ckb-shell .ckb-cluster-profile-figure { background: #ffffff !important; color-scheme: only light !important; }
html[data-ckb-theme="dark"] body .options[role="listbox"] { border-color: #334155 !important; background: #111827 !important; }
html[data-ckb-theme="dark"] body .options[role="listbox"] [role="option"] { background: #111827 !important; color: #f8fafc !important; }
html[data-ckb-theme="dark"] body .options[role="listbox"] [role="option"]:hover,
html[data-ckb-theme="dark"] body .options[role="listbox"] [role="option"][aria-selected="true"] { background: #0f2f4a !important; color: #f8fafc !important; }

@media (max-width: 820px) {
  #ckb-shell { padding: 14px 12px 36px; }
  .ckb-hero { min-height: 0; border-radius: var(--ckb-radius-lg); }
  .ckb-trust-row { grid-template-columns: 1fr; }
  .ckb-section-heading { display: block; }
  .ckb-coding-note { max-width: none; margin-top: 14px; }
  .ckb-risk-row { grid-template-columns: 104px minmax(90px, 1fr) 48px; gap: 8px; }
  .ckb-shap-grid { grid-template-columns: 1fr; }
  .ckb-shap-card--cluster { grid-column: auto; }
  .ckb-image-grid--clusters { grid-template-columns: 1fr; }
  .ckb-image-grid--outcomes { grid-template-columns: 1fr; }
}

@media (max-width: 520px) {
  #ckb-shell { padding-inline: 8px; }
  .ckb-hero, .ckb-panel, .ckb-results { border-radius: 16px; }
  .ckb-hero { padding: 28px 22px; }
  .ckb-hero h1 { max-width: 11ch; }
  .ckb-panel, .ckb-results { padding: 18px; }
  .ckb-result-card__header, .ckb-risk-visual__header { display: block; }
  .ckb-result-badge, .ckb-legend { margin-top: 10px; }
  .ckb-shap-frame { height: 360px; }
  .ckb-outcome-table__header { display: block; }
  .ckb-outcome-table__header > p { margin-top: 7px; text-align: left; }
  .ckb-image-grid--clusters { grid-template-columns: 1fr; }
  .ckb-image-frame { height: 210px; }
  .ckb-image-card__context { grid-template-columns: 1fr; }
  .ckb-actions { align-items: stretch; }
  #predict-button, #reset-button { width: 100%; }
  .ckb-privacy-copy { flex-basis: 100%; }
}

@media (prefers-reduced-motion: reduce) {
  *, *::before, *::after { scroll-behavior: auto !important; transition-duration: 0.01ms !important; animation-duration: 0.01ms !important; animation-iteration-count: 1 !important; }
}
"""


HEAD = """
<script>
(() => {
  const startMotion = () => {
    const root = document.querySelector('#ckb-shell');
    if (!root || root.dataset.motionReady === 'true') return Boolean(root);
    root.dataset.motionReady = 'true';

    const run = () => {
      if (!window.gsap) return;
      const media = window.gsap.matchMedia();
      media.add({ reduceMotion: '(prefers-reduced-motion: reduce)' }, (context) => {
        const hero = root.querySelector('.ckb-hero');
        const cards = root.querySelectorAll('.ckb-trust-card');
        const panel = root.querySelector('.ckb-panel');
        const results = root.querySelector('.ckb-results');
        if (context.conditions.reduceMotion) {
          window.gsap.set([hero, cards, panel, results], { autoAlpha: 1, clearProps: 'transform' });
          return;
        }
        const timeline = window.gsap.timeline({ defaults: { duration: 0.42, ease: 'power3.out' } });
        timeline
          .from(hero, { y: 18, autoAlpha: 0 })
          .from(cards, { y: 14, scale: 0.98, autoAlpha: 0, stagger: 0.07, ease: 'back.out(1.35)' }, '-=0.18')
          .from(panel, { y: 16, autoAlpha: 0 }, '-=0.20')
          .from(results, { y: 12, autoAlpha: 0 }, '-=0.24')
          // A transformed ancestor makes Gradio's fixed dropdown portal drift.
          .set([hero, cards, panel, results], { clearProps: 'transform,opacity,visibility' });
        return () => timeline.kill();
      });
    };

    if (window.gsap) {
      run();
      return true;
    }
    const script = document.createElement('script');
    script.src = 'https://cdn.jsdelivr.net/npm/gsap@3/dist/gsap.min.js';
    script.async = true;
    script.onload = run;
    script.onerror = () => { /* Animation is progressive enhancement only. */ };
    document.head.appendChild(script);
    return true;
  };

  if (!startMotion()) {
    const observer = new MutationObserver(() => {
      if (startMotion()) observer.disconnect();
    });
    observer.observe(document.documentElement, { childList: true, subtree: true });
    window.setTimeout(() => observer.disconnect(), 7000);
  }
})();
</script>
<script>
(() => {
  const copy = {
    en: {
      eyebrow: 'CKB RESEARCH INTERFACE',
      title: '17-variable risk prediction',
      hero: 'One complete input returns a frozen K=3 cluster assignment and probabilities for 13 outcomes. The form and API use the same model inference path.',
      chips: ['17 input variables', '13 probability outputs', 'No persistent storage'],
      metrics: ['Complete input variables', 'Frozen nearest-centroid clustering', 'Outcome risk probabilities'],
      ready: '<strong>Model ready.</strong> Predictions can start immediately; submitted data is not persistently stored by this Space.',
      loading: '<strong>Warming the model in the background.</strong> You can complete the form while loading finishes.',
      inputKicker: 'INPUT WORKSPACE',
      inputTitle: 'Enter prediction variables',
      inputBody: 'All 17 variables are required. Categorical values must use the same codes as the training data; this service does not recode, impute, or retain input.',
      coding: 'Research-use note: verify category codes and units before interpreting results.',
      groups: [['Participant profile', 'Demographic and social context'], ['Lifestyle & self-report', 'Behaviour and self-rated health'], ['Measurements & medication', 'Body measures and blood-pressure information']],
      run: 'Run risk prediction',
      reset: 'Restore demo values',
      privacy: 'Temporary files are removed after inference. Do not use this tool for diagnosis or treatment decisions.',
      outputKicker: 'PREDICTION OUTPUT',
      outputTitle: 'Results and probability distribution',
      clusterPlaceholder: 'Submit a complete set of variables to show cluster assignment and distance information here.',
      visualPlaceholder: 'The probability distribution will appear after prediction completes.',
      clusterKicker: 'CLUSTER ASSIGNMENT',
      cluster: 'Risk cluster',
      assigned: 'Assigned',
      review: 'Review suggested',
      nearest: 'Nearest centroid distance',
      margin: 'Separation from next-nearest centroid',
      visualTitle: 'Probability across 13 outcomes',
      visualBody: 'The marker indicates the frozen Youden threshold for each outcome. This is for research interpretation, not clinical diagnosis.',
      low: 'Below threshold',
      high: 'At or above threshold',
      tableLabel: 'Outcome prediction details',
      tableHeaders: ['Code', 'Outcome', 'Probability', 'Frozen threshold', 'Risk decision'],
      json: 'View complete JSON response',
      footer: 'For programmatic use: the stable JSON endpoint is <code>/predict</code>; model warm-up is <code>/warmup</code>. Use the “Use via API” panel for examples.'
    },
    zh: {
      eyebrow: 'CKB 研究界面',
      title: '\u0031\u0037 \u53d8\u91cf\u98ce\u9669\u9884\u6d4b',
      hero: '\u4e00\u6b21\u5b8c\u6574\u8f93\u5165\uff0c\u8fd4\u56de\u51bb\u7ed3 K=3 \u805a\u7c7b\u5206\u914d\u4e0e \u0031\u0033 \u9879\u7ed3\u5c40\u98ce\u9669\u6982\u7387\u3002\u8868\u5355\u4e0e API \u5171\u7528\u540c\u4e00\u5957\u6a21\u578b\u63a8\u7406\u903b\u8f91\u3002',
      chips: ['\u0031\u0037 \u4e2a\u8f93\u5165\u53d8\u91cf', '\u0031\u0033 \u9879\u6982\u7387\u7ed3\u679c', '\u65e0\u6301\u4e45\u5316\u4fdd\u5b58'],
      metrics: ['\u5b8c\u6574\u8f93\u5165\u53d8\u91cf', '\u51bb\u7ed3\u6700\u8fd1\u4e2d\u5fc3\u805a\u7c7b', '\u7ed3\u5c40\u98ce\u9669\u6982\u7387'],
      ready: '<strong>\u6a21\u578b\u5df2\u5c31\u7eea\u3002</strong>\u53ef\u4ee5\u7acb\u5373\u63d0\u4ea4\u9884\u6d4b\uff1b\u8fd9\u4e2a Space \u4e0d\u4f1a\u6301\u4e45\u4fdd\u5b58\u63d0\u4ea4\u6570\u636e\u3002',
      loading: '<strong>\u6b63\u5728\u540e\u53f0\u9884\u70ed\u6a21\u578b\u3002</strong>\u4f60\u53ef\u4ee5\u5148\u586b\u5199\u8868\u5355\uff1b\u52a0\u8f7d\u5b8c\u6210\u540e\u5373\u53ef\u63d0\u4ea4\u3002',
      inputKicker: '\u8f93\u5165\u5de5\u4f5c\u533a',
      inputTitle: '\u586b\u5199\u9884\u6d4b\u53d8\u91cf',
      inputBody: '\u6240\u6709 \u0031\u0037 \u9879\u5747\u4e3a\u5fc5\u586b\u3002\u5206\u7c7b\u53d8\u91cf\u5fc5\u987b\u4e0e\u8bad\u7ec3\u6570\u636e\u4f7f\u7528\u76f8\u540c\u7f16\u7801\uff1b\u672c\u670d\u52a1\u4e0d\u4f1a\u91cd\u65b0\u7f16\u7801\u3001\u63d2\u8865\u6216\u4fdd\u5b58\u8f93\u5165\u3002',
      coding: '\u7814\u7a76\u7528\u9014\u63d0\u793a\uff1a\u8bf7\u5728\u89e3\u91ca\u7ed3\u679c\u524d\u6838\u5bf9\u5206\u7c7b\u7f16\u7801\u4e0e\u91cf\u7eb2\u3002',
      groups: [['\u57fa\u672c\u8d44\u6599', '\u4eba\u53e3\u4e0e\u793e\u4f1a\u7279\u5f81'], ['\u751f\u6d3b\u4e0e\u611f\u53d7', '\u884c\u4e3a\u53ca\u5065\u5eb7\u81ea\u8bc4'], ['\u4f53\u5f81\u4e0e\u7528\u836f', '\u8eab\u4f53\u6d4b\u91cf\u4e0e\u8840\u538b\u4fe1\u606f']],
      run: '\u8fd0\u884c\u98ce\u9669\u9884\u6d4b',
      reset: '\u6062\u590d\u6f14\u793a\u6570\u636e',
      privacy: '\u63a8\u7406\u7ed3\u675f\u540e\u4f1a\u6e05\u9664\u4e34\u65f6\u6587\u4ef6\uff1b\u8bf7\u52ff\u5c06\u672c\u5de5\u5177\u7528\u4e8e\u8bca\u65ad\u6216\u6cbb\u7597\u51b3\u7b56\u3002',
      outputKicker: '\u9884\u6d4b\u8f93\u51fa',
      outputTitle: '\u7ed3\u679c\u4e0e\u6982\u7387\u5206\u5e03',
      clusterPlaceholder: '\u63d0\u4ea4\u4e00\u7ec4\u5b8c\u6574\u53d8\u91cf\u540e\uff0c\u6b64\u5904\u5c06\u663e\u793a\u805a\u7c7b\u5206\u914d\u4e0e\u8ddd\u79bb\u4fe1\u606f\u3002',
      visualPlaceholder: '\u6982\u7387\u5206\u5e03\u56fe\u5c06\u5728\u9884\u6d4b\u5b8c\u6210\u540e\u51fa\u73b0\u3002',
      clusterKicker: '\u805a\u7c7b\u5206\u914d',
      cluster: '\u98ce\u9669\u805a\u7c7b',
      assigned: '\u5df2\u5b8c\u6210\u5206\u914d',
      review: '\u5efa\u8bae\u590d\u6838',
      nearest: '\u6700\u8fd1\u4e2d\u5fc3\u8ddd\u79bb',
      margin: '\u4e0e\u6b21\u8fd1\u4e2d\u5fc3\u7684\u95f4\u9694',
      visualTitle: '\u0031\u0033 \u9879\u7ed3\u5c40\u7684\u9884\u6d4b\u6982\u7387',
      visualBody: '\u6a2a\u7ebf\u6807\u8bb0\u5404\u7ed3\u5c40\u51bb\u7ed3\u7684 Youden \u9608\u503c\uff1b\u6b64\u56fe\u7528\u4e8e\u7814\u7a76\u7ed3\u679c\u9605\u8bfb\uff0c\u5e76\u975e\u4e34\u5e8a\u8bca\u65ad\u3002',
      low: '\u4f4e\u4e8e\u9608\u503c',
      high: '\u8fbe\u5230\u6216\u9ad8\u4e8e\u9608\u503c',
      tableLabel: '\u7ed3\u5c40\u9884\u6d4b\u660e\u7ec6',
      tableHeaders: ['\u4ee3\u7801', '\u9884\u6d4b\u7ed3\u5c40', '\u6982\u7387', '\u51bb\u7ed3\u9608\u503c', '\u98ce\u9669\u5224\u5b9a'],
      json: '\u67e5\u770b\u5b8c\u6574 JSON \u54cd\u5e94',
      footer: '\u7528\u4e8e\u7a0b\u5e8f\u5316\u8c03\u7528\uff1a\u7a33\u5b9a JSON \u63a5\u53e3\u4e3a <code>/predict</code>\uff1b\u6a21\u578b\u9884\u70ed\u63a5\u53e3\u4e3a <code>/warmup</code>\u3002\u53ef\u4ece Space \u7684 “Use via API” \u9762\u677f\u83b7\u53d6\u8c03\u7528\u793a\u4f8b\u3002'
    }
  };

  const fields = {
    en: {
      labels: { sex: 'Sex (training code)', age: 'Age (years)', edu_level: 'Education (training code)', marital_status: 'Marital status (training code)', work: 'Employment (training code)', retire: 'Retirement (training code)', hh_size: 'Household size', smoking: 'Smoking (training code)', alcohol: 'Alcohol use (training code)', height_cm: 'Height (cm)', weight_kg: 'Weight (kg)', waist_cm: 'Waist (cm)', sbp_mmhg: 'Systolic BP (mmHg)', dbp_mmhg: 'Diastolic BP (mmHg)', bp_drugs: 'BP medication (training code)', self_health: 'Self-rated health (training code)', chronic_pain: 'Chronic pain (training code)' },
      help: { sex: 'Use the original categorical code from the training data.', age: 'Enter completed years.', edu_level: 'Use the original categorical code from the training data.', marital_status: 'Use the original categorical code from the training data.', work: 'Use the original categorical code from the training data.', retire: 'Use the original categorical code from the training data.', hh_size: 'Number of people in the household.', smoking: 'Use the original categorical code from the training data.', alcohol: 'Use the original categorical code from the training data.', height_cm: 'Continuous value in centimetres.', weight_kg: 'Continuous value in kilograms.', waist_cm: 'Continuous value in centimetres.', sbp_mmhg: 'Continuous value in mmHg.', dbp_mmhg: 'Continuous value in mmHg.', bp_drugs: 'Use the original categorical code from the training data.', self_health: 'Use the original categorical code from the training data.', chronic_pain: 'Use the original categorical code from the training data.' }
    },
    zh: {
      labels: { sex: '\u6027\u522b\u7f16\u7801', age: '\u5e74\u9f84\uff08\u5c81\uff09', edu_level: '\u6559\u80b2\u7a0b\u5ea6\u7f16\u7801', marital_status: '\u5a5a\u59fb\u72b6\u6001\u7f16\u7801', work: '\u5de5\u4f5c\u72b6\u6001\u7f16\u7801', retire: '\u9000\u4f11\u72b6\u6001\u7f16\u7801', hh_size: '\u5bb6\u5ead\u4eba\u53e3\u6570', smoking: '\u5438\u70df\u7f16\u7801', alcohol: '\u996e\u9152\u7f16\u7801', height_cm: '\u8eab\u9ad8\uff08cm\uff09', weight_kg: '\u4f53\u91cd\uff08kg\uff09', waist_cm: '\u8170\u56f4\uff08cm\uff09', sbp_mmhg: '\u6536\u7f29\u538b\uff08mmHg\uff09', dbp_mmhg: '\u8212\u5f20\u538b\uff08mmHg\uff09', bp_drugs: '\u964d\u538b\u836f\u4f7f\u7528\u7f16\u7801', self_health: '\u81ea\u8bc4\u5065\u5eb7\u7f16\u7801', chronic_pain: '\u6162\u6027\u75bc\u75db\u7f16\u7801' },
      help: { sex: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002', age: '\u4ee5\u5b8c\u6574\u5c81\u6570\u586b\u5199\u3002', edu_level: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002', marital_status: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002', work: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002', retire: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002', hh_size: '\u540c\u4f4f\u5bb6\u5ead\u6210\u5458\u4eba\u6570\u3002', smoking: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002', alcohol: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002', height_cm: '\u8fde\u7eed\u6570\u503c\uff0c\u5355\u4f4d\u4e3a\u5398\u7c73\u3002', weight_kg: '\u8fde\u7eed\u6570\u503c\uff0c\u5355\u4f4d\u4e3a\u5343\u514b\u3002', waist_cm: '\u8fde\u7eed\u6570\u503c\uff0c\u5355\u4f4d\u4e3a\u5398\u7c73\u3002', sbp_mmhg: '\u8fde\u7eed\u6570\u503c\uff0c\u5355\u4f4d\u4e3a mmHg\u3002', dbp_mmhg: '\u8fde\u7eed\u6570\u503c\uff0c\u5355\u4f4d\u4e3a mmHg\u3002', bp_drugs: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002', self_health: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002', chronic_pain: '\u8bf7\u4f7f\u7528\u8bad\u7ec3\u6570\u636e\u4e2d\u7684\u539f\u59cb\u5206\u7c7b\u7f16\u7801\u3002' }
    }
  };

  Object.assign(copy.en, {
    inputBody: 'All 17 variables are required. Choose categorical values by their descriptions; the frozen model encoding is preserved internally and no input is retained.',
    coding: 'Research-use note: verify the selected description and units before interpreting results.',
    privacy: 'Temporary SHAP explanation files expire automatically. Do not use this tool for diagnosis or treatment decisions.',
    shapExpand: 'Show 14 SHAP force plots',
    shapCollapse: 'Hide SHAP force plots',
    shapKicker: 'SHAP EXPLANATIONS',
    shapTitle: '14 force plots',
    shapDescription: 'One force plot explains the assigned cluster; the remaining 13 explain each outcome probability surrogate.',
    shapCluster: 'Cluster assignment',
    shapOutcome: 'Outcome explanation',
    shapPlaceholder: 'Run a prediction to generate 14 SHAP force plots.'
  });
  Object.assign(copy.zh, {
    inputBody: '\u6240\u6709 17 \u9879\u5747\u4e3a\u5fc5\u586b\u3002\u5206\u7c7b\u53d8\u91cf\u901a\u8fc7\u6587\u5b57\u9009\u9879\u9009\u62e9\uff1b\u7cfb\u7edf\u5728\u540e\u53f0\u4fdd\u6301\u4e0e\u8bad\u7ec3\u6570\u636e\u4e00\u81f4\u7684\u51bb\u7ed3\u7f16\u7801\uff0c\u4e0d\u4f1a\u4fdd\u5b58\u8f93\u5165\u3002',
    coding: '\u7814\u7a76\u7528\u9014\u63d0\u793a\uff1a\u8bf7\u5728\u89e3\u91ca\u7ed3\u679c\u524d\u6838\u5bf9\u6240\u9009\u6587\u5b57\u63cf\u8ff0\u4e0e\u91cf\u7eb2\u3002',
    privacy: '\u4e34\u65f6 SHAP \u89e3\u91ca\u6587\u4ef6\u4f1a\u81ea\u52a8\u8fc7\u671f\u6e05\u7406\uff1b\u8bf7\u52ff\u5c06\u672c\u5de5\u5177\u7528\u4e8e\u8bca\u65ad\u6216\u6cbb\u7597\u51b3\u7b56\u3002',
    shapExpand: '\u5c55\u5f00\u67e5\u770b 14 \u5f20 SHAP \u529b\u56fe',
    shapCollapse: '\u6536\u8d77 SHAP \u529b\u56fe',
    shapKicker: 'SHAP \u89e3\u91ca',
    shapTitle: '14 \u5f20\u529b\u56fe',
    shapDescription: '1 \u5f20\u529b\u56fe\u89e3\u91ca\u5df2\u5206\u914d\u7684\u805a\u7c7b\uff1b\u53e6\u5916 13 \u5f20\u5206\u522b\u89e3\u91ca\u5404\u7ed3\u5c40\u6982\u7387\u4ee3\u7406\u6a21\u578b\u3002',
    shapCluster: '\u805a\u7c7b\u5206\u914d',
    shapOutcome: '\u7ed3\u5c40\u89e3\u91ca',
    shapPlaceholder: '\u8fd0\u884c\u9884\u6d4b\u540e\u751f\u6210 14 \u5f20 SHAP \u529b\u56fe\u3002'
  });
  Object.assign(fields.en.labels, {
    sex: 'Sex', edu_level: 'Education level', marital_status: 'Married or partnered',
    work: 'Currently working', retire: 'Retired', smoking: 'Smoking', alcohol: 'Alcohol use',
    bp_drugs: 'Blood-pressure medication', self_health: 'Self-rated health', chronic_pain: 'Chronic pain'
  });
  Object.assign(fields.en.help, {
    sex: 'Choose the description that matches the study definition.', edu_level: 'Choose the description that matches the study definition.',
    marital_status: 'Choose the description that matches the study definition.', work: 'Choose the description that matches the study definition.',
    retire: 'Choose the description that matches the study definition.', smoking: 'Choose the description that matches the study definition.',
    alcohol: 'Choose the description that matches the study definition.', bp_drugs: 'Choose the description that matches the study definition.',
    self_health: 'Choose the description that matches the study definition.', chronic_pain: 'Choose the description that matches the study definition.'
  });
  Object.assign(fields.zh.labels, {
    sex: '\u6027\u522b', edu_level: '\u6559\u80b2\u7a0b\u5ea6', marital_status: '\u5a5a\u59fb\u6216\u4f34\u4fa3\u72b6\u6001',
    work: '\u5f53\u524d\u5de5\u4f5c', retire: '\u5df2\u9000\u4f11', smoking: '\u5438\u70df', alcohol: '\u996e\u9152',
    bp_drugs: '\u964d\u538b\u836f\u4f7f\u7528', self_health: '\u81ea\u8bc4\u5065\u5eb7', chronic_pain: '\u6162\u6027\u75bc\u75db'
  });
  Object.assign(fields.zh.help, {
    sex: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002', edu_level: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002',
    marital_status: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002', work: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002',
    retire: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002', smoking: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002',
    alcohol: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002', bp_drugs: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002',
    self_health: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002', chronic_pain: '\u8bf7\u9009\u62e9\u4e0e\u7814\u7a76\u5b9a\u4e49\u76f8\u7b26\u7684\u6587\u5b57\u9009\u9879\u3002'
  });

  // Keep the visible UI concise. The frozen numeric representation stays in
  // the component value and is never exposed as 0/1/2/3/4 to visitors.
  Object.assign(copy.en, {
    hero: 'One complete input returns a frozen K=3 cluster assignment and probabilities for 13 outcomes. The model was developed using the China Kadoorie Biobank.',
    inputBody: '',
    coding: '',
    outputTitle: 'Prediction results',
    themeLabel: 'Colour theme',
    themeLight: 'Light',
    themeDark: 'Dark',
    shapExpand: 'Show image results',
    shapCollapse: 'Hide image results',
    shapKicker: 'IMAGE RESULTS',
    shapTitle: 'Assigned cluster explanation and 13 outcome panels',
    shapDescription: 'Each outcome panel places its cumulative-risk reference and brief interpretation above the related SHAP force plot.',
    shapCluster: 'Assigned cluster explanation',
    shapOutcome: 'Outcome risk explanations',
    shapPlaceholder: 'Run a prediction to generate image results.'
  });
  Object.assign(copy.zh, {
    hero: '\u4e00\u6b21\u5b8c\u6574\u8f93\u5165\uff0c\u8fd4\u56de\u51bb\u7ed3 K=3 \u805a\u7c7b\u5206\u914d\u4e0e 13 \u9879\u7ed3\u5c40\u98ce\u9669\u6982\u7387\u3002\u6a21\u578b\u57fa\u4e8e China Kadoorie Biobank \u5f00\u53d1\u3002',
    inputBody: '',
    coding: '',
    outputTitle: '\u9884\u6d4b\u7ed3\u679c',
    themeLabel: '\u9875\u9762\u4e3b\u9898',
    themeLight: '\u6d45\u8272',
    themeDark: '\u6df1\u8272',
    shapExpand: '\u5c55\u5f00\u67e5\u770b\u56fe\u7247\u7ed3\u679c',
    shapCollapse: '\u6536\u8d77\u56fe\u7247\u7ed3\u679c',
    shapKicker: '\u56fe\u7247\u7ed3\u679c',
    shapTitle: '\u5df2\u5206\u914d\u805a\u7c7b\u89e3\u91ca\u4e0e 13 \u9879\u7ed3\u5c40\u98ce\u9669\u9762\u677f',
    shapDescription: '\u6bcf\u4e2a\u7ed3\u5c40\u9762\u677f\u5747\u5c06\u7d2f\u79ef\u98ce\u9669\u53c2\u8003\u548c\u7b80\u77ed\u89e3\u8bfb\u7f6e\u4e8e\u76f8\u5173 SHAP \u529b\u56fe\u4e0a\u65b9\u3002',
    shapCluster: '\u5df2\u5206\u914d\u805a\u7c7b\u89e3\u91ca',
    shapOutcome: '\u7ed3\u5c40\u98ce\u9669\u89e3\u91ca',
    shapPlaceholder: '\u8fd0\u884c\u9884\u6d4b\u540e\u751f\u6210\u56fe\u7247\u7ed3\u679c\u3002'
  });

  const dropdownCopy = {
    'Male': { en: 'Male', zh: '\u7537' }, 'Female': { en: 'Female', zh: '\u5973' },
    'Low': { en: 'Low', zh: '\u4f4e' }, 'Intermediate': { en: 'Intermediate', zh: '\u4e2d' }, 'High': { en: 'High', zh: '\u9ad8' },
    'Other observed status': { en: 'Other observed status', zh: '\u5176\u4ed6\u89c2\u5bdf\u72b6\u6001' },
    'Married or partnered': { en: 'Married or partnered', zh: '\u5df2\u5a5a\u6216\u6709\u4f34\u4fa3' },
    'No': { en: 'No', zh: '\u5426' }, 'Yes': { en: 'Yes', zh: '\u662f' },
    'Very good or excellent': { en: 'Very good or excellent', zh: '\u975e\u5e38\u597d\u6216\u6781\u597d' },
    'Good': { en: 'Good', zh: '\u597d' }, 'Fair or regular': { en: 'Fair or regular', zh: '\u4e00\u822c' },
    'Poor or very poor': { en: 'Poor or very poor', zh: '\u5dee\u6216\u5f88\u5dee' }
  };

  const setText = (node, value) => {
    if (node && node.textContent !== value) node.textContent = value;
  };
  const setHtml = (node, value) => {
    if (node && node.innerHTML !== value) node.innerHTML = value;
  };
  const translateDynamicContent = (root, language) => {
    const textKey = language === 'zh' ? 'localeZh' : 'localeEn';
    const ariaKey = language === 'zh' ? 'ariaZh' : 'ariaEn';
    root.querySelectorAll('[data-locale-en]').forEach((node) => setText(node, node.dataset[textKey] || ''));
    root.querySelectorAll('[data-aria-en]').forEach((node) => {
      const value = node.dataset[ariaKey];
      if (value) node.setAttribute('aria-label', value);
    });
  };
  const dropdownEntry = (value) => {
    const normalized = String(value || '').replace(/^[✓✔]\\s*/, '').trim();
    return Object.values(dropdownCopy).find((entry) => entry.en === normalized || entry.zh === normalized);
  };
  const translateDropdownText = (root, language) => {
    root.querySelectorAll('.ckb-form-card input').forEach((input) => {
      const entry = dropdownEntry(input.value);
      if (entry) input.value = entry[language];
    });
    document.querySelectorAll('[role="option"]').forEach((option) => {
      const entry = dropdownEntry(option.textContent);
      if (entry) option.textContent = entry[language];
    });
  };

  const initLocale = () => {
    const root = document.querySelector('#ckb-shell');
    if (!root || root.dataset.localeReady === 'true') return Boolean(root);
    const trustRow = root.querySelector('.ckb-trust-row');
    if (!trustRow) return false;
    root.dataset.localeReady = 'true';
    const syncShapDisclosure = () => {
      const button = root.querySelector('#shap-result > button');
      if (!button) return;
      const text = copy[root.dataset.language || 'en'];
      const expanded = button.classList.contains('open');
      const label = expanded ? text.shapCollapse : text.shapExpand;
      setText(button, label);
      button.setAttribute('aria-label', label);
    };
    const setupShapDisclosure = () => {
      const button = root.querySelector('#shap-result > button');
      if (!button || button.dataset.ckbDisclosureReady === 'true') return;
      button.dataset.ckbDisclosureReady = 'true';
      new MutationObserver(syncShapDisclosure).observe(button, {
        attributes: true,
        attributeFilter: ['class'],
      });
    };
    let language = 'en';
    try { language = localStorage.getItem('ckb-language') === 'zh' ? 'zh' : 'en'; } catch (_) { /* Default to English. */ }
    let theme = 'light';
    try { theme = localStorage.getItem('ckb-theme') === 'dark' ? 'dark' : 'light'; } catch (_) { /* Theme is page-controlled and defaults to light. */ }

    const bar = document.createElement('div');
    bar.className = 'ckb-language-bar';
    bar.innerHTML = '<div class="ckb-language-switch" role="group" aria-label="Language selector"><button type="button" data-language="en">English</button><button type="button" data-language="zh">\u4e2d\u6587</button></div><div class="ckb-theme-switch" role="group" data-theme-switch="true" aria-label="Colour theme"><button type="button" data-theme="light">Light</button><button type="button" data-theme="dark">Dark</button></div>';
    trustRow.insertAdjacentElement('afterend', bar);

    const applyTheme = () => {
      root.dataset.theme = theme;
      document.documentElement.dataset.ckbTheme = theme;
      root.querySelectorAll('[data-theme]').forEach((button) => {
        button.setAttribute('aria-pressed', String(button.dataset.theme === theme));
      });
    };

    const applyLanguage = () => {
      const text = copy[language];
      root.dataset.language = language;
      root.querySelectorAll('[data-language]').forEach((button) => {
        button.setAttribute('aria-pressed', String(button.dataset.language === language));
      });
      const themeSwitch = root.querySelector('[data-theme-switch]');
      if (themeSwitch) themeSwitch.setAttribute('aria-label', text.themeLabel);
      setText(root.querySelector('[data-theme="light"]'), text.themeLight);
      setText(root.querySelector('[data-theme="dark"]'), text.themeDark);
      setText(root.querySelector('.ckb-skip'), language === 'en' ? 'Skip to prediction inputs' : '\u8df3\u8f6c\u5230\u9884\u6d4b\u8f93\u5165');
      const eyebrow = root.querySelector('.ckb-eyebrow');
      if (eyebrow) setText(eyebrow.lastChild, text.eyebrow);
      setText(root.querySelector('.ckb-hero h1'), text.title);
      setText(root.querySelector('.ckb-hero p'), text.hero);
      root.querySelectorAll('.ckb-chip').forEach((chip, index) => setText(chip.lastChild, ` ${text.chips[index] || ''}`));
      root.querySelectorAll('.ckb-trust-card span').forEach((node, index) => setText(node, text.metrics[index] || ''));

      const status = root.querySelector('.ckb-status span:last-child');
      if (status) setHtml(status, root.querySelector('.ckb-status--ready') ? text.ready : text.loading);
      setText(root.querySelector('.ckb-section-heading .ckb-section-kicker'), text.inputKicker);
      setText(root.querySelector('.ckb-section-heading h2'), text.inputTitle);
      setText(root.querySelector('.ckb-section-heading p'), text.inputBody);
      setText(root.querySelector('.ckb-coding-note'), text.coding);
      Array.from(root.querySelectorAll('.ckb-form-card'))
        .filter((card) => !card.parentElement.closest('.ckb-form-card'))
        .forEach((card, index) => {
        setText(card.querySelector('h3'), text.groups[index]?.[0] || '');
        setText(card.querySelector('p'), text.groups[index]?.[1] || '');
        });
      Object.keys(fields[language].labels).forEach((field) => {
        const component = root.querySelector(`#field-${field}`);
        if (!component) return;
        setText(component.querySelector('[data-testid="block-info"]'), fields[language].labels[field]);
        component.querySelector('input')?.setAttribute('aria-label', fields[language].labels[field]);
      });
      translateDropdownText(root, language);
      setText(root.querySelector('#predict-button'), text.run);
      setText(root.querySelector('#reset-button'), text.reset);
      const privacy = root.querySelector('.ckb-privacy-copy');
      if (privacy) {
        const icon = privacy.querySelector('svg')?.outerHTML || '';
        setHtml(privacy, `${icon}${text.privacy}`);
      }
      setText(root.querySelector('.ckb-results-header .ckb-section-kicker'), text.outputKicker);
      setText(root.querySelector('.ckb-results-header h2'), text.outputTitle);
      setText(root.querySelector('.ckb-result-placeholder'), root.querySelector('#cluster-result .ckb-result-placeholder') ? text.clusterPlaceholder : text.visualPlaceholder);
      const placeholders = root.querySelectorAll('.ckb-result-placeholder');
      setText(placeholders[0], text.clusterPlaceholder);
      setText(placeholders[1], text.visualPlaceholder);

      const clusterTitle = root.querySelector('.ckb-result-card h3');
      if (clusterTitle) {
        const label = clusterTitle.textContent.split(/[::]/).pop().trim();
        setText(clusterTitle, `${text.cluster}: ${label}`);
      }
      setText(root.querySelector('.ckb-result-card .ckb-result-kicker'), text.clusterKicker);
      const badge = root.querySelector('.ckb-result-badge');
      if (badge) setText(badge, badge.classList.contains('ckb-result-badge--review') ? text.review : text.assigned);
      const metrics = root.querySelectorAll('.ckb-result-metrics dt');
      setText(metrics[0], text.nearest);
      setText(metrics[1], text.margin);
      setText(root.querySelector('.ckb-risk-visual h3'), text.visualTitle);
      setText(root.querySelector('.ckb-risk-visual p'), text.visualBody);
      root.querySelectorAll('.ckb-legend span').forEach((legend, index) => {
        const marker = legend.querySelector('i')?.outerHTML || '';
        setHtml(legend, `${marker}${index === 0 ? text.low : text.high}`);
      });
      setText(root.querySelector('#outcome-table .label p'), text.tableLabel);
      root.querySelectorAll('#outcome-table th [role="button"]').forEach((node, index) => setText(node, text.tableHeaders[index] || ''));
      root.querySelectorAll('#outcome-table th').forEach((node, index) => node.setAttribute('title', text.tableHeaders[index] || ''));
      root.querySelectorAll('#outcome-table [role="button"]').forEach((node) => {
        const value = node.textContent.trim();
        if (value === '\u8fbe\u5230\u6216\u9ad8\u4e8e\u9608\u503c' || value === 'At or above threshold') setText(node, text.high);
        if (value === '\u4f4e\u4e8e\u9608\u503c' || value === 'Below threshold') setText(node, text.low);
      });
      setupShapDisclosure();
      syncShapDisclosure();
      setText(root.querySelector('[data-shap-kicker]'), text.shapKicker);
      setText(root.querySelector('[data-shap-title]'), text.shapTitle);
      setText(root.querySelector('[data-shap-description]'), text.shapDescription);
      root.querySelectorAll('[data-shap-kind]').forEach((node) => {
        setText(node, node.dataset.shapKind === 'cluster' ? text.shapCluster : text.shapOutcome);
      });
      root.querySelectorAll('.ckb-shap-placeholder').forEach((node) => setText(node, text.shapPlaceholder));
      setText(root.querySelector('[data-image-kicker]'), text.shapKicker);
      setText(root.querySelector('[data-image-title]'), text.shapTitle);
      setText(root.querySelector('[data-image-description]'), text.shapDescription);
      setText(root.querySelector('[data-image-cluster-title]'), text.shapCluster);
      setText(root.querySelector('[data-image-outcome-title]'), text.shapOutcome);
      root.querySelectorAll('.ckb-image-placeholder').forEach((node) => setText(node, text.shapPlaceholder));
      translateDynamicContent(root, language);
      const jsonButton = root.querySelector('#raw-result > button');
      if (jsonButton) setText(jsonButton, text.json);
      setHtml(root.querySelector('.ckb-footer-note'), text.footer);
    };

    bar.addEventListener('click', (event) => {
      const themeButton = event.target.closest('button[data-theme]');
      if (themeButton) {
        theme = themeButton.dataset.theme === 'dark' ? 'dark' : 'light';
        try { localStorage.setItem('ckb-theme', theme); } catch (_) { /* Preference remains for this page. */ }
        applyTheme();
        return;
      }
      const button = event.target.closest('button[data-language]');
      if (!button) return;
      language = button.dataset.language === 'zh' ? 'zh' : 'en';
      try { localStorage.setItem('ckb-language', language); } catch (_) { /* Preference remains for this page. */ }
      applyLanguage();
    });
    document.addEventListener('click', () => {
      requestAnimationFrame(() => translateDropdownText(root, language));
    }, true);
    applyTheme();
    applyLanguage();
    let scheduled = false;
    const observer = new MutationObserver(() => {
      if (scheduled) return;
      scheduled = true;
      requestAnimationFrame(() => { scheduled = false; applyLanguage(); });
    });
    observer.observe(root, { childList: true, subtree: true });
  };

  if (!initLocale()) {
    const observer = new MutationObserver(() => {
      if (initLocale()) observer.disconnect();
    });
    observer.observe(document.documentElement, { childList: true, subtree: true });
    window.setTimeout(() => observer.disconnect(), 7000);
  }
})();
</script>
"""


_predictor: CKBSingleSamplePredictor | None = None
_load_lock = threading.Lock()
_inference_lock = threading.Lock()


def get_predictor() -> CKBSingleSamplePredictor:
    """Load the large model bundle once, on the first prediction request."""
    global _predictor
    if _predictor is None:
        with _load_lock:
            if _predictor is None:
                _predictor = CKBSingleSamplePredictor(MODEL_DIR)
    return _predictor


def _clean_expired_shap_reports() -> None:
    """Remove old, per-request SHAP files from the ephemeral Space filesystem."""
    if not SHAP_CACHE_DIR.exists():
        return
    deadline = time.time() - SHAP_CACHE_TTL_SECONDS
    for report_dir in SHAP_CACHE_DIR.iterdir():
        try:
            if report_dir.is_dir() and report_dir.stat().st_mtime < deadline:
                shutil.rmtree(report_dir, ignore_errors=True)
        except OSError:
            continue


def _ensure_light_shap_canvas(plot_path: str) -> None:
    """Keep generated SHAP HTML legible when the host browser uses dark mode."""
    path = Path(plot_path)
    try:
        document = path.read_text(encoding="utf-8")
        if SHAP_LIGHT_CANVAS_MARKER in document:
            return
        if "</head>" not in document:
            return
        path.write_text(
            document.replace("</head>", SHAP_LIGHT_CANVAS_HEAD + "</head>", 1),
            encoding="utf-8",
        )
    except (OSError, UnicodeError):
        return


def _shap_force_plot_card(kind: str, title: str, plot_path: str, *, cluster: bool = False) -> str:
    """Embed one trusted, locally-generated SHAP HTML force plot by temporary URL."""
    card_class = " ckb-shap-card--cluster" if cluster else ""
    _ensure_light_shap_canvas(plot_path)
    plot_url = "/gradio_api/file=" + quote(Path(plot_path).resolve().as_posix(), safe="/")
    return f"""
    <article class="ckb-shap-card{card_class}">
      <header class="ckb-shap-card__header">
        <p class="ckb-shap-card__kind" data-shap-kind="{kind}">{kind.title()}</p>
        <h4>{html.escape(title)}</h4>
      </header>
      <iframe
        class="ckb-shap-frame"
        title="{html.escape(title)} SHAP force plot"
        loading="lazy"
        sandbox="allow-scripts"
        src="{plot_url}"
      ></iframe>
    </article>
    """


def _shap_force_gallery(result: dict[str, Any]) -> str:
    """Create the one-cluster plus 13-outcome SHAP explanation gallery."""
    try:
        force_plots = result["force_plots"]
        cluster = result["cluster"]
        cluster_label = str(cluster.get("cluster_label") or cluster["provisional_cluster_label"])
        cards = [
            _shap_force_plot_card(
                "cluster",
                f"Frozen cluster explanation · {cluster_label}",
                force_plots["cluster"][cluster_label],
                cluster=True,
            )
        ]
        for outcome_code, outcome in result["outcomes"].items():
            cards.append(
                _shap_force_plot_card(
                    "outcome",
                    f"{outcome_code} · {outcome['outcome']}",
                    force_plots[outcome_code],
                )
            )
    except (KeyError, OSError, TypeError):
        return (
            '<div class="ckb-shap-placeholder">'
            "SHAP force plots could not be rendered for this request."
            "</div>"
        )

    return """
    <section class="ckb-shap-gallery" aria-live="polite">
      <header class="ckb-shap-gallery__header">
        <div class="ckb-section-kicker" data-shap-kicker="true">SHAP EXPLANATIONS</div>
        <h3 data-shap-title="true">14 force plots</h3>
        <p data-shap-description="true">One force plot explains the assigned cluster; the remaining 13 explain each outcome probability surrogate.</p>
      </header>
      <div class="ckb-shap-grid">
    """ + "".join(cards) + "</div></section>"


def _locale_attrs(en: str, zh: str) -> str:
    """Return safe text attributes used by the language switcher."""
    return (
        f'data-locale-en="{html.escape(en, quote=True)}" '
        f'data-locale-zh="{html.escape(zh, quote=True)}"'
    )


def _outcome_presentation(code: str, outcome: dict[str, Any]) -> dict[str, Any]:
    """Get display-only reference data without influencing inference."""
    fallback_name = str(outcome.get("outcome", code))
    return OUTCOME_PRESENTATION.get(
        code,
        {
            "en": fallback_name,
            "zh": fallback_name,
            "model_en": "Model",
            "model_zh": "模型",
            "low_5": 0.0,
            "low_10": 0.0,
            "high_5": 0.0,
            "high_10": 0.0,
        },
    )


def _force_plot_frame(title_en: str, title_zh: str, plot_path: str) -> str:
    """Embed one trusted locally-generated SHAP force plot."""
    _ensure_light_shap_canvas(plot_path)
    plot_url = "/gradio_api/file=" + quote(Path(plot_path).resolve().as_posix(), safe="/")
    aria_en = title_en + " SHAP force plot"
    aria_zh = title_zh + " SHAP 力图"
    return f"""
      <iframe class="ckb-image-frame" title="{html.escape(aria_en, quote=True)}"
        data-aria-en="{html.escape(aria_en, quote=True)}"
        data-aria-zh="{html.escape(aria_zh, quote=True)}"
        loading="lazy" sandbox="allow-scripts" src="{plot_url}"></iframe>
    """


def _supplementary_image_url(path: Path) -> str:
    """Return a temporary Gradio file URL for a bundled static figure."""
    return "/gradio_api/file=" + quote(path.resolve().as_posix(), safe="/")


def _risk_reference_figure(
    code: str,
    presentation: dict[str, Any],
    *,
    is_high_risk: bool,
) -> str:
    """Show the supplied low/high cumulative-risk figure for this outcome."""
    risk_key = "high" if is_high_risk else "low"
    risk_en, risk_zh = ("High risk", "高风险") if is_high_risk else ("Low risk", "低风险")
    figure_path = SUPPLEMENTARY_IMAGE_DIR / f"{code}_{risk_key}.png"
    if not figure_path.is_file():
        return _cumulative_risk_svg(code, presentation)
    display_code = code.replace("_", "–")
    alt_en = f"{display_code} cumulative risk curve with the {risk_en.lower()} trajectory emphasised"
    alt_zh = f"{display_code} 累积风险曲线,突出显示{risk_zh}轨迹"
    return f"""
    <img class="ckb-reference-figure" src="{_supplementary_image_url(figure_path)}"
      loading="lazy" alt="{html.escape(alt_en, quote=True)}"
      data-aria-en="{html.escape(alt_en, quote=True)}"
      data-aria-zh="{html.escape(alt_zh, quote=True)}" />
    """


def _cluster_profile_figure(label: str) -> str:
    """Show the supplied C1/C2/C3 cluster-profile illustration."""
    figure_path = SUPPLEMENTARY_IMAGE_DIR / f"{label}.png"
    title_en, title_zh = f"Cluster {label} risk profile map", f"聚类 {label} 风险概览图"
    if not figure_path.is_file():
        return '<div class="ckb-image-placeholder">Cluster profile illustration is unavailable.</div>'
    return f"""
    <img class="ckb-cluster-profile-figure" src="{_supplementary_image_url(figure_path)}"
      loading="lazy" alt="{html.escape(title_en, quote=True)}"
      data-aria-en="{html.escape(title_en, quote=True)}"
      data-aria-zh="{html.escape(title_zh, quote=True)}" />
    """


def _cumulative_risk_svg(code: str, presentation: dict[str, Any]) -> str:
    """Create a compact native risk curve from the supplied reference values."""
    low_5, low_10 = float(presentation["low_5"]), float(presentation["low_10"])
    high_5, high_10 = float(presentation["high_5"]), float(presentation["high_10"])
    chart_max = max(low_5, low_10, high_5, high_10, 1.0) * 1.18

    def y(value: float) -> float:
        return 137 - (value / chart_max * 90)

    low_y5, low_y10, high_y5, high_y10 = y(low_5), y(low_10), y(high_5), y(high_10)
    label_en = f"{code.replace('_', '–')} cumulative-risk reference curves"
    label_zh = f"{code.replace('_', '–')} 累积风险参考曲线"
    return f"""
    <svg class="ckb-risk-curve" viewBox="0 0 350 178" role="img"
      aria-label="{html.escape(label_en, quote=True)}"
      data-aria-en="{html.escape(label_en, quote=True)}"
      data-aria-zh="{html.escape(label_zh, quote=True)}">
      <g class="ckb-risk-curve__grid" aria-hidden="true"><path d="M40 40H320M40 89H320M40 137H320M40 22V137H320" /></g>
      <g class="ckb-risk-curve__legend" aria-hidden="true">
        <line x1="43" y1="12" x2="65" y2="12" class="ckb-risk-curve__low" />
        <text x="71" y="16" {_locale_attrs('Low-risk reference', '低风险参考')}>Low-risk reference</text>
        <line x1="191" y1="12" x2="213" y2="12" class="ckb-risk-curve__high" />
        <text x="219" y="16" {_locale_attrs('High-risk reference', '高风险参考')}>High-risk reference</text>
      </g>
      <polyline points="40,137 155,{low_y5:.1f} 280,{low_y10:.1f}" class="ckb-risk-curve__low" />
      <polyline points="40,137 155,{high_y5:.1f} 280,{high_y10:.1f}" class="ckb-risk-curve__high" />
      <g class="ckb-risk-curve__point ckb-risk-curve__point--low"><circle cx="155" cy="{low_y5:.1f}" r="4.5" /><circle cx="280" cy="{low_y10:.1f}" r="4.5" /><text x="155" y="{low_y5 - 10:.1f}" text-anchor="middle">{low_5:.1f}%</text><text x="280" y="{low_y10 - 10:.1f}" text-anchor="middle">{low_10:.1f}%</text></g>
      <g class="ckb-risk-curve__point ckb-risk-curve__point--high"><circle cx="155" cy="{high_y5:.1f}" r="4.5" /><circle cx="280" cy="{high_y10:.1f}" r="4.5" /><text x="155" y="{high_y5 - 10:.1f}" text-anchor="middle">{high_5:.1f}%</text><text x="280" y="{high_y10 - 10:.1f}" text-anchor="middle">{high_10:.1f}%</text></g>
      <text x="40" y="163" class="ckb-risk-curve__axis" {_locale_attrs('0 years', '0年')}>0 years</text><text x="140" y="163" class="ckb-risk-curve__axis" {_locale_attrs('5 years', '5年')}>5 years</text><text x="260" y="163" class="ckb-risk-curve__axis" {_locale_attrs('10 years', '10年')}>10 years</text>
    </svg>
    """


def _outcome_image_card(code: str, outcome: dict[str, Any], plot_path: str) -> str:
    """Render a compact outcome card with context above its SHAP force plot."""
    presentation = _outcome_presentation(code, outcome)
    probability = max(0.0, min(1.0, float(outcome["probability"])))
    threshold = max(0.0, min(1.0, float(outcome["youden_threshold"])))
    is_high_risk = probability >= threshold
    risk_en, risk_zh = ("High risk", "\u9ad8\u98ce\u9669") if is_high_risk else ("Low risk", "\u4f4e\u98ce\u9669")
    decision_en = "at or above" if is_high_risk else "below"
    decision_zh = "\u8fbe\u5230\u6216\u9ad8\u4e8e" if is_high_risk else "\u4f4e\u4e8e"
    explanation_en = (
        f"Predicted probability is {probability * 100:.1f}%, "
        f"{decision_en} the frozen threshold of {threshold * 100:.1f}%. Reference cumulative incidence: "
        f"low-risk {presentation['low_5']:.1f}% at 5 years and {presentation['low_10']:.1f}% at 10 years; "
        f"high-risk {presentation['high_5']:.1f}% and {presentation['high_10']:.1f}%."
    )
    explanation_zh = (
        f"\u9884\u6d4b\u6982\u7387\u4e3a {probability * 100:.1f}%\uff0c{decision_zh}\u51bb\u7ed3\u9608\u503c "
        f"{threshold * 100:.1f}%\u3002\u53c2\u8003\u7d2f\u79ef\u53d1\u751f\u7387\uff1a\u4f4e\u98ce\u9669\u7ec4 5 \u5e74 {presentation['low_5']:.1f}%\u300110 \u5e74 "
        f"{presentation['low_10']:.1f}%\uff1b\u9ad8\u98ce\u9669\u7ec4\u5206\u522b\u4e3a {presentation['high_5']:.1f}% \u548c {presentation['high_10']:.1f}%\u3002"
    )
    display_code = code.replace("_", "\u2013")
    title_en, title_zh = str(presentation["en"]), str(presentation["zh"])
    return f"""
    <article class="ckb-image-card ckb-image-card--outcome">
      <header class="ckb-image-card__header"><div><p class="ckb-image-card__kind" {_locale_attrs('Outcome risk panel', '\u7ed3\u5c40\u98ce\u9669\u9762\u677f')}>Outcome risk panel</p><h4>{html.escape(display_code)} · <span {_locale_attrs(title_en, title_zh)}>{html.escape(title_en)}</span></h4></div><span class="ckb-image-card__badge {'ckb-image-card__badge--high' if is_high_risk else ''}" {_locale_attrs(risk_en, risk_zh)}>{risk_en}</span></header>
      <div class="ckb-image-card__context"><div class="ckb-image-card__curve">{_risk_reference_figure(code, presentation, is_high_risk=is_high_risk)}</div><div class="ckb-image-card__explanation"><p class="ckb-image-card__risk" {_locale_attrs(risk_en, risk_zh)}>{risk_en}</p><p {_locale_attrs(explanation_en, explanation_zh)}>{html.escape(explanation_en)}</p></div></div>
      <div class="ckb-image-card__force">{_force_plot_frame(display_code + ' ' + title_en, display_code + ' ' + title_zh, plot_path)}</div>
    </article>
    """


def _cluster_image_card(label: str, plot_path: str) -> str:
    """Render only the assigned cluster profile and its matching SHAP force plot."""
    title_en, title_zh = f"Cluster {label}", f"\u805a\u7c7b {label}"
    state_en, state_zh = "Assigned cluster", "\u672c\u6b21\u5206\u914d"
    return f"""
    <article class="ckb-image-card ckb-image-card--cluster ckb-image-card--selected">
      <header class="ckb-image-card__header"><div><p class="ckb-image-card__kind" {_locale_attrs('Cluster attribution', '\u805a\u7c7b\u5f52\u56e0')}>Cluster attribution</p><h4 {_locale_attrs(title_en, title_zh)}>{title_en}</h4></div><span class="ckb-image-card__badge" {_locale_attrs(state_en, state_zh)}>{state_en}</span></header>
      {_cluster_profile_figure(label)}
      <div class="ckb-image-card__force">{_force_plot_frame(title_en, title_zh, plot_path)}</div>
    </article>
    """


def _image_results_gallery(result: dict[str, Any]) -> str:
    """Create the assigned-cluster and 13-outcome image-result panels."""
    try:
        force_plots = result["force_plots"]
        cluster = result["cluster"]
        selected_label = str(cluster.get("cluster_label") or cluster["provisional_cluster_label"])
        cluster_cards = [_cluster_image_card(selected_label, force_plots["cluster"][selected_label])]
        outcome_cards = [
            _outcome_image_card(code, outcome, force_plots[code])
            for code, outcome in result["outcomes"].items()
        ]
    except (KeyError, OSError, TypeError):
        return '<div class="ckb-image-placeholder" data-locale-en="Image results could not be rendered for this request." data-locale-zh="本次请求无法生成图片结果。">Image results could not be rendered for this request.</div>'

    return """
    <section class="ckb-image-gallery" aria-live="polite">
      <header class="ckb-image-gallery__header"><div class="ckb-section-kicker" data-image-kicker="true">IMAGE RESULTS</div><h3 data-image-title="true">Assigned cluster explanation and 13 outcome panels</h3><p data-image-description="true">Each outcome panel places its cumulative-risk reference and brief interpretation above the related SHAP force plot.</p></header>
      <section class="ckb-image-section" aria-labelledby="cluster-image-title"><div class="ckb-image-section__header"><h4 id="cluster-image-title" data-image-cluster-title="true">Assigned cluster explanation</h4></div><div class="ckb-image-grid ckb-image-grid--clusters">
    """ + "".join(cluster_cards) + """
      </div></section>
      <section class="ckb-image-section" aria-labelledby="outcome-image-title"><div class="ckb-image-section__header"><h4 id="outcome-image-title" data-image-outcome-title="true">Outcome risk explanations</h4></div><div class="ckb-image-grid ckb-image-grid--outcomes">
    """ + "".join(outcome_cards) + "</div></section></section>"


@spaces.GPU
def run_prediction(
    features: dict[str, float],
    sample_id: str | None = None,
    *,
    include_force_plots: bool = False,
) -> dict[str, Any]:
    """Run one prediction; browser-only SHAP files are temporary and auto-expire."""
    predictor = get_predictor()
    with _inference_lock:
        if include_force_plots:
            SHAP_CACHE_DIR.mkdir(parents=True, exist_ok=True)
            _clean_expired_shap_reports()
            output_dir = SHAP_CACHE_DIR / uuid.uuid4().hex
            try:
                result = predictor.predict_one(
                    features,
                    output_dir=output_dir,
                    make_force_plots=True,
                    sample_id=sample_id,
                )
                result["_shap_force_gallery_html"] = _image_results_gallery(result)
            except Exception:
                shutil.rmtree(output_dir, ignore_errors=True)
                raise
        else:
            with tempfile.TemporaryDirectory(prefix="ckb_prediction_") as output_dir:
                result = predictor.predict_one(
                    features,
                    output_dir=output_dir,
                    make_force_plots=False,
                    sample_id=sample_id,
                )

    # The predictor writes an audit JSON to its output directory. The API is
    # intentionally stateless, so that temporary file is removed and its now
    # invalid local path is not returned to callers.
    result.pop("result_json", None)
    result.pop("force_plots", None)
    return result


def warm_up_model() -> str:
    """Load prediction-time artifacts after a visitor opens the demo."""
    try:
        predictor = get_predictor()
        with _inference_lock:
            predictor._ensure_cluster_shap()
            for outcome in predictor.outcome_meta:
                predictor._load_outcome_bundle(outcome)
        return (
            '<div class="ckb-status ckb-status--ready" role="status">'
            '<span class="ckb-status__dot" aria-hidden="true"></span>'
            '<span><strong>模型已就绪。</strong>预测可立即开始;提交数据不会被此 Space 持久保存。</span>'
            "</div>"
        )
    except Exception:
        return (
            '<div class="ckb-status ckb-status--warning" role="status">'
            '<span class="ckb-status__dot" aria-hidden="true"></span>'
            '<span><strong>预热尚未完成。</strong>仍可提交预测,系统会在请求时自动重试加载。</span>'
            "</div>"
        )


def predict_json_api(request: dict[str, Any]) -> dict[str, Any]:
    """Single-object API used by external Python, JavaScript, or curl clients."""
    if not isinstance(request, dict):
        raise gr.Error("Request must be a JSON object.")

    features = request.get("features", request)
    if not isinstance(features, dict):
        raise gr.Error("The 'features' field must be a JSON object.")

    missing = [feature for feature in FEATURES if feature not in features]
    if missing:
        raise gr.Error(f"Missing required input variables: {missing}")

    sample_id = request.get("sample_id")
    if sample_id is not None:
        sample_id = str(sample_id)
        if len(sample_id) > 200:
            raise gr.Error("sample_id must be no longer than 200 characters.")

    return run_prediction(
        {feature: features[feature] for feature in FEATURES},
        sample_id=sample_id,
    )


def _outcome_table(result: dict[str, Any]) -> pd.DataFrame:
    rows: list[dict[str, Any]] = []
    for code, outcome in result["outcomes"].items():
        is_high_risk = outcome["risk_class"] == "high_risk"
        rows.append(
            {
                "代码": code,
                "预测结局": outcome["outcome"],
                "概率": f"{outcome['probability'] * 100:.1f}%",
                "冻结阈值": f"{outcome['youden_threshold'] * 100:.1f}%",
                "风险判定": "At or above threshold" if is_high_risk else "Below threshold",
            }
        )
    return pd.DataFrame(rows).set_axis(TABLE_COLUMNS_EN, axis="columns")


def _outcome_table_html(result: dict[str, Any]) -> str:
    """Render the outcome table as fully bilingual native HTML."""
    rows: list[str] = []
    for code, outcome in result["outcomes"].items():
        presentation = _outcome_presentation(code, outcome)
        probability = max(0.0, min(1.0, float(outcome["probability"])))
        threshold = max(0.0, min(1.0, float(outcome["youden_threshold"])))
        is_high_risk = probability >= threshold
        risk_en, risk_zh = ("At or above threshold", "达到或高于阈值") if is_high_risk else ("Below threshold", "低于阈值")
        rows.append(
            f"""
            <tr>
              <td class="ckb-outcome-table__code">{html.escape(code.replace('_', '–'))}</td>
              <td><span {_locale_attrs(str(presentation['en']), str(presentation['zh']))}>{html.escape(str(presentation['en']))}</span></td>
              <td>{probability * 100:.1f}%</td>
              <td>{threshold * 100:.1f}%</td>
              <td><span class="ckb-outcome-table__decision{' ckb-outcome-table__decision--high' if is_high_risk else ''}" {_locale_attrs(risk_en, risk_zh)}>{risk_en}</span></td>
            </tr>
            """
        )
    return """
    <section class="ckb-outcome-table" aria-live="polite">
      <div class="ckb-outcome-table__header"><div><p class="ckb-section-kicker" data-outcome-table-kicker="true" data-locale-en="OUTCOME PREDICTIONS" data-locale-zh="结局预测">OUTCOME PREDICTIONS</p><h3 data-outcome-table-title="true" data-locale-en="Outcome prediction details" data-locale-zh="结局预测明细">Outcome prediction details</h3></div><p data-outcome-table-description="true" data-locale-en="Risk class is determined by the frozen threshold for each outcome." data-locale-zh="风险判定根据每项结局的冻结阈值确定。">Risk class is determined by the frozen threshold for each outcome.</p></div>
      <div class="ckb-outcome-table__scroll"><table><thead><tr>
        <th data-locale-en="Code" data-locale-zh="代码">Code</th><th data-locale-en="Outcome" data-locale-zh="结局">Outcome</th><th data-locale-en="Probability" data-locale-zh="概率">Probability</th><th data-locale-en="Frozen threshold" data-locale-zh="冻结阈值">Frozen threshold</th><th data-locale-en="Risk decision" data-locale-zh="风险判定">Risk decision</th>
      </tr></thead><tbody>
    """ + "".join(rows) + "</tbody></table></div></section>"


def _cluster_card(cluster: dict[str, Any]) -> str:
    label = str(cluster.get("cluster_label") or cluster["provisional_cluster_label"])
    assigned = cluster["classification_status"] == "assigned"
    badge_class = "" if assigned else " ckb-result-badge--review"
    badge_text = "已完成分配" if assigned else "建议复核"
    return f"""
    <section class="ckb-result-card" aria-live="polite">
      <div class="ckb-result-card__header">
        <div>
          <div class="ckb-result-kicker">Cluster assignment</div>
          <h3>风险聚类:{html.escape(label)}</h3>
        </div>
        <span class="ckb-result-badge{badge_class}">{badge_text}</span>
      </div>
      <dl class="ckb-result-metrics">
        <div><dt>最近中心距离</dt><dd>{cluster['distance_nearest']:.4f}</dd></div>
        <div><dt>与次近中心的间隔</dt><dd>{cluster['margin']:.4f}</dd></div>
      </dl>
    </section>
    """


def _risk_visual(result: dict[str, Any]) -> str:
    """Render a responsive, accessible probability chart without extra dependencies."""
    rows = sorted(
        result["outcomes"].items(),
        key=lambda item: float(item[1]["probability"]),
        reverse=True,
    )
    bars: list[str] = []
    for code, outcome in rows:
        probability = max(0.0, min(1.0, float(outcome["probability"])))
        threshold = max(0.0, min(1.0, float(outcome["youden_threshold"])))
        is_high_risk = outcome["risk_class"] == "high_risk"
        row_class = " ckb-risk-row--high" if is_high_risk else ""
        accessible_label = (
            f"{code}: {outcome['outcome']}, probability {probability * 100:.1f}%, "
            f"threshold {threshold * 100:.1f}%"
        )
        bars.append(
            f'<div class="ckb-risk-row{row_class}" aria-label="{html.escape(accessible_label)}">'
            f'<span class="ckb-risk-label" title="{html.escape(str(outcome["outcome"]))}">{html.escape(code)}</span>'
            f'<span class="ckb-risk-track" aria-hidden="true">'
            f'<span class="ckb-risk-fill" style="--risk-width: {probability * 100:.2f}%"></span>'
            f'<span class="ckb-risk-threshold" style="--threshold: {threshold * 100:.2f}%"></span>'
            "</span>"
            f'<span class="ckb-risk-value">{probability * 100:.1f}%</span>'
            "</div>"
        )
    return """
    <section class="ckb-risk-visual" aria-labelledby="risk-visual-title">
      <div class="ckb-risk-visual__header">
        <div>
          <h3 id="risk-visual-title">13 项结局的预测概率</h3>
          <p>横线标记各结局冻结的 Youden 阈值;此图用于研究结果阅读,并非临床诊断。</p>
        </div>
        <div class="ckb-legend" aria-label="图例">
          <span><i aria-hidden="true"></i>低于阈值</span>
          <span><i class="ckb-legend--high" aria-hidden="true"></i>达到或高于阈值</span>
        </div>
      </div>
    """ + "".join(bars) + "</section>"


def predict_for_demo(
    *values: float | int | None,
) -> tuple[str, str, str, dict[str, Any], str]:
    missing = [
        FEATURE_LABELS[feature]
        for feature, value in zip(FEATURES, values, strict=True)
        if value is None
    ]
    if missing:
        raise gr.Error(f"请补全以下变量后再预测:{', '.join(missing)}")

    features = {
        feature: float(value)
        for feature, value in zip(FEATURES, values, strict=True)
        if value is not None
    }
    result = run_prediction(
        features,
        sample_id="web_demo",
        include_force_plots=True,
    )
    shap_gallery = result.pop(
        "_shap_force_gallery_html",
        '<div class="ckb-image-placeholder" data-locale-en="Image results are unavailable for this request." data-locale-zh="本次请求暂无图片结果。">Image results are unavailable for this request.</div>',
    )
    return (
        _cluster_card(result["cluster"]),
        _outcome_table_html(result),
        _risk_visual(result),
        result,
        shap_gallery,
    )


def restore_example_values() -> list[float]:
    """Restore the prefilled synthetic demonstration values without model work."""
    return [EXAMPLE_VALUES[feature] for feature in FEATURES]


with gr.Blocks(
    title="CKB 17-variable Risk Predictor",
    fill_width=True,
) as demo:
    with gr.Column(elem_id="ckb-shell"):
        gr.HTML(
            """
            <a class="ckb-skip" href="#risk-inputs">跳转到预测输入</a>
            <section class="ckb-hero" aria-labelledby="ckb-title">
              <div class="ckb-hero__content">
                <div class="ckb-eyebrow">
                  <span class="ckb-mark" aria-hidden="true">
                    <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8">
                      <path d="M7 4c5 3 5 13 10 16M17 4C12 7 12 17 7 20M9 6h6M8 12h8M9 18h6"/>
                    </svg>
                  </span>
                  CKB research interface
                </div>
                <h1 id="ckb-title">17 变量风险预测</h1>
                <p>一次完整输入,返回冻结 K=3 聚类分配与 13 项结局风险概率。表单与 API 共用同一套模型推理逻辑。</p>
                <div class="ckb-hero__meta" aria-label="产品特点">
                  <span class="ckb-chip"><svg viewBox="0 0 24 24"><path d="M12 3v18M3 12h18"/></svg>17 个输入变量</span>
                  <span class="ckb-chip"><svg viewBox="0 0 24 24"><path d="M4 19V5m0 14h16M8 15l3-4 3 2 5-7"/></svg>13 项概率结果</span>
                  <span class="ckb-chip"><svg viewBox="0 0 24 24"><rect x="5" y="10" width="14" height="10" rx="2"/><path d="M8 10V7a4 4 0 0 1 8 0v3"/></svg>无持久化保存</span>
                </div>
              </div>
            </section>
            """
        )
        gr.HTML(
            """
            <section class="ckb-trust-row" aria-label="模型概览">
              <article class="ckb-trust-card"><strong>17</strong><span>完整输入变量</span></article>
              <article class="ckb-trust-card"><strong>K = 3</strong><span>冻结最近中心聚类</span></article>
              <article class="ckb-trust-card"><strong>13</strong><span>结局风险概率</span></article>
            </section>
            """
        )
        warm_status = gr.HTML(
            '<div class="ckb-status" role="status">'
            '<span class="ckb-status__dot" aria-hidden="true"></span>'
            '<span><strong>正在后台预热模型。</strong>你可以先填写表单;模型完成加载后即可提交。</span>'
            "</div>",
            elem_id="warm-status",
        )

        with gr.Column(elem_id="risk-inputs", elem_classes="ckb-panel"):
            gr.HTML(
                """
                <header class="ckb-section-heading">
                  <div>
                    <div class="ckb-section-kicker">Input workspace</div>
                    <h2>填写预测变量</h2>
                    <p>所有 17 项均为必填。分类变量必须与训练数据使用相同编码;本服务不会重新编码、插补或保存输入。</p>
                  </div>
                  <aside class="ckb-coding-note">研究用途提示:请在解释结果前核对分类编码与量纲。</aside>
                </header>
                """
            )
            input_by_feature: dict[str, Any] = {}
            with gr.Row(elem_classes="ckb-input-grid"):
                for index, title, subtitle, group_features in FEATURE_GROUPS:
                    with gr.Column(min_width=250):
                        with gr.Group(elem_classes="ckb-form-card"):
                            gr.HTML(
                                f"""
                                <div class="form-card__top">
                                  <span class="form-card__index">{index}</span>
                                  <div><h3>{title}</h3><p>{subtitle}</p></div>
                                </div>
                                """
                            )
                            for feature in group_features:
                                component_kwargs = {
                                    "label": DEFAULT_EN_FEATURE_LABELS[feature],
                                    "value": EXAMPLE_VALUES[feature],
                                    "elem_id": f"field-{feature}",
                                }
                                if feature in CATEGORICAL_CHOICES:
                                    input_by_feature[feature] = gr.Dropdown(
                                        choices=CATEGORICAL_CHOICES[feature],
                                        filterable=False,
                                        **component_kwargs,
                                    )
                                else:
                                    input_by_feature[feature] = gr.Number(
                                        precision=0 if feature in INTEGER_FEATURES else None,
                                        **component_kwargs,
                                    )

            # Components are displayed in user-friendly groups, while the
            # predictor always receives its frozen feature order.
            inputs = [input_by_feature[feature] for feature in FEATURES]

            with gr.Row(elem_classes="ckb-actions"):
                submit = gr.Button("运行风险预测", variant="primary", elem_id="predict-button")
                reset = gr.Button("恢复演示数据", variant="secondary", elem_id="reset-button")
                gr.HTML(
                    """
                    <p class="ckb-privacy-copy">
                      <svg viewBox="0 0 24 24" fill="none"><rect x="5" y="10" width="14" height="10" rx="2"/><path d="M8 10V7a4 4 0 0 1 8 0v3"/></svg>
                      推理结束后会清除临时文件;请勿将本工具用作诊断或治疗决策。
                    </p>
                    """
                )

        with gr.Column(elem_classes="ckb-results"):
            gr.HTML(
                """
                <header class="ckb-results-header">
                  <div class="ckb-section-kicker">Prediction output</div>
                  <h2>结果与概率分布</h2>
                </header>
                """
            )
            cluster_output = gr.HTML(
                '<div class="ckb-result-placeholder">提交一组完整变量后,此处将显示聚类分配与距离信息。</div>',
                elem_id="cluster-result",
            )
            risk_visual = gr.HTML(
                '<div class="ckb-result-placeholder">概率分布图将在预测完成后出现。</div>',
                elem_id="risk-visual",
            )
            outcome_output = gr.HTML(
                '<div class="ckb-image-placeholder" data-locale-en="Run a prediction to generate outcome details." data-locale-zh="运行预测后生成结局明细。">Run a prediction to generate outcome details.</div>',
                elem_id="outcome-table",
            )
            with gr.Accordion("Show image results (16)", open=False, elem_id="shap-result"):
                shap_output = gr.HTML(
                    '<div class="ckb-image-placeholder" data-locale-en="Run a prediction to generate image results." data-locale-zh="运行预测后生成图片结果。">Run a prediction to generate image results.</div>',
                    elem_id="shap-gallery",
                )
            with gr.Accordion("查看完整 JSON 响应", open=False, elem_id="raw-result"):
                raw_output = gr.JSON(label="完整响应")
            gr.HTML(
                '<p class="ckb-footer-note">面向程序化使用:稳定 JSON 接口为 <code>/predict</code>;预热状态接口为 <code>/warmup</code>。可从 Space 的 “Use via API” 获取调用示例。</p>'
            )

        submit.click(
            fn=predict_for_demo,
            inputs=inputs,
            outputs=[cluster_output, outcome_output, risk_visual, raw_output, shap_output],
            api_name="demo_predict",
            show_progress="full",
        )
        reset.click(fn=restore_example_values, outputs=inputs, api_name=False)

        # A compact API-only endpoint. Keeping one JSON input makes it convenient
        # for external Python, JavaScript, and curl clients to call the Space.
        api_request = gr.JSON(label="request", visible=False)
        api_response = gr.JSON(label="response", visible=False)
        api_trigger = gr.Button(visible=False)
        api_trigger.click(
            fn=predict_json_api,
            inputs=api_request,
            outputs=api_response,
            api_name="predict",
        )
        warmup_trigger = gr.Button(visible=False)
        warmup_trigger.click(
            fn=warm_up_model,
            outputs=warm_status,
            api_name="warmup",
            show_progress="hidden",
        )

demo.queue(default_concurrency_limit=1, max_size=16)

if __name__ == "__main__":
    SHAP_CACHE_DIR.mkdir(parents=True, exist_ok=True)
    demo.launch(
        css=CUSTOM_CSS,
        head=HEAD,
        allowed_paths=[str(SHAP_CACHE_DIR), str(SUPPLEMENTARY_IMAGE_DIR)],
    )