langtell 0.4.0 → 0.6.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (40) hide show
  1. package/README.md +39 -13
  2. package/dist/chrome-ai.d.ts +1 -1
  3. package/dist/{chunk-KCK3XWCJ.js → chunk-7TDGJOIJ.js} +3 -3
  4. package/dist/{chunk-KCK3XWCJ.js.map → chunk-7TDGJOIJ.js.map} +1 -1
  5. package/dist/{chunk-U34Z3ZSV.js → chunk-BL627TWI.js} +4 -3
  6. package/dist/chunk-BL627TWI.js.map +1 -0
  7. package/dist/{chunk-3LDE35U2.js → chunk-FEKBPTHK.js} +3 -3
  8. package/dist/{chunk-3LDE35U2.js.map → chunk-FEKBPTHK.js.map} +1 -1
  9. package/dist/{chunk-G44HHVK5.js → chunk-JXGCCDSV.js} +5 -4
  10. package/dist/chunk-JXGCCDSV.js.map +1 -0
  11. package/dist/{chunk-KI4MAI3N.js → chunk-K4MXTIY7.js} +3 -3
  12. package/dist/{chunk-KI4MAI3N.js.map → chunk-K4MXTIY7.js.map} +1 -1
  13. package/dist/{chunk-OVSPOZ5J.js → chunk-YCUSX3GG.js} +17 -5
  14. package/dist/chunk-YCUSX3GG.js.map +1 -0
  15. package/dist/classify.d.ts +9 -2
  16. package/dist/classify.js +1 -1
  17. package/dist/cyrillic.d.ts +60 -0
  18. package/dist/cyrillic.js +69 -0
  19. package/dist/cyrillic.js.map +1 -0
  20. package/dist/franc.d.ts +1 -1
  21. package/dist/franc.js +1 -1
  22. package/dist/fuse.d.ts +1 -1
  23. package/dist/fuse.js +3 -3
  24. package/dist/headers.d.ts +1 -1
  25. package/dist/headers.js +2 -2
  26. package/dist/html.d.ts +1 -1
  27. package/dist/html.js +2 -2
  28. package/dist/index.d.ts +58 -7
  29. package/dist/index.js +47 -13
  30. package/dist/index.js.map +1 -1
  31. package/dist/profiles.d.ts +5 -2
  32. package/dist/profiles.js +431 -2
  33. package/dist/profiles.js.map +1 -1
  34. package/dist/text.d.ts +1 -1
  35. package/dist/text.js +2 -2
  36. package/dist/{types-BIXrkuAr.d.ts → types-DfHZ5hkh.d.ts} +19 -3
  37. package/package.json +5 -1
  38. package/dist/chunk-G44HHVK5.js.map +0 -1
  39. package/dist/chunk-OVSPOZ5J.js.map +0 -1
  40. package/dist/chunk-U34Z3ZSV.js.map +0 -1
package/dist/profiles.js CHANGED
@@ -804,6 +804,300 @@ var FREQUENT_GENERATED = {
804
804
  "\u043F\u0440\u043E\u0438\u0437\u043E\u0448\u043B\u043E",
805
805
  "\u043A\u043E\u0442\u043E\u0440\u0443\u044E"
806
806
  ],
807
+ mk: [
808
+ "\u0434\u0430",
809
+ "\u043D\u0435",
810
+ "\u0441\u0435",
811
+ "\u0435",
812
+ "\u043D\u0430",
813
+ "\u0433\u043E",
814
+ "\u0438",
815
+ "\u0448\u0442\u043E",
816
+ "\u0432\u043E",
817
+ "\u0437\u0430",
818
+ "\u0442\u043E\u0430",
819
+ "\u0442\u0438",
820
+ "\u0434\u0435\u043A\u0430",
821
+ "\u043E\u0434",
822
+ "\u0441\u043E",
823
+ "\u043C\u0438",
824
+ "\u0441\u0438",
825
+ "\u0458\u0430",
826
+ "\u043A\u0430\u043A\u043E",
827
+ "\u0458\u0430\u0441",
828
+ "\u0433\u0438",
829
+ "\u043E\u0432\u0430",
830
+ "\u0441\u0443\u043C",
831
+ "\u0434\u043E\u0431\u0440\u043E",
832
+ "\u043C\u0435",
833
+ "\u0442\u0435",
834
+ "\u0441\u0430\u043C\u043E",
835
+ "\u043B\u0438",
836
+ "\u043D\u0435\u043C\u0430",
837
+ "\u043D\u043E",
838
+ "\u0430\u043A\u043E",
839
+ "\u0430",
840
+ "\u0442\u043E\u0458",
841
+ "\u0431\u0435\u0448\u0435",
842
+ "\u0442\u0440\u0435\u0431\u0430",
843
+ "\u043A\u043E\u0433\u0430",
844
+ "\u0441\u0435\u0433\u0430",
845
+ "\u043C\u043D\u043E\u0433\u0443",
846
+ "\u0437\u043D\u0430\u043C",
847
+ "\u043A\u043E\u0458",
848
+ "\u043C\u0443",
849
+ "\u0438\u043C\u0430",
850
+ "\u043D\u0435\u0448\u0442\u043E",
851
+ "\u043C\u043E\u0436\u0435",
852
+ "\u043C\u0435\u043D\u0435",
853
+ "\u0431\u0438",
854
+ "\u043A\u0430\u0434\u0435",
855
+ "\u0441\u0430\u043A\u0430\u043C",
856
+ "\u0442\u0435\u0431\u0435",
857
+ "\u0442\u0443\u043A\u0430",
858
+ "\u0434\u0430\u043B\u0438",
859
+ "\u0442\u0430\u043A\u0430",
860
+ "\u0437\u043E\u0448\u0442\u043E",
861
+ "\u0441\u0438\u0442\u0435",
862
+ "\u0430\u0458\u0434\u0435",
863
+ "\u0440\u0435\u0434",
864
+ "\u043D\u0438",
865
+ "\u043F\u0430",
866
+ "\u0437\u043D\u0430\u0435\u0448",
867
+ "\u043E\u0432\u0434\u0435",
868
+ "\u0442\u0430\u0430",
869
+ "\u043C\u043E\u0436\u0430\u043C",
870
+ "\u0431\u0438\u0434\u0435",
871
+ "\u043F\u043E",
872
+ "\u043C\u0438\u0441\u043B\u0430\u043C",
873
+ "\u0441\u0430\u043A\u0430\u0448",
874
+ "\u0434\u043E",
875
+ "\u0435\u0434\u0435\u043D",
876
+ "\u0432\u0438",
877
+ "\u0443\u0448\u0442\u0435",
878
+ "\u0438\u043B\u0438",
879
+ "\u043D\u0435\u043A\u043E\u0458",
880
+ "\u0430\u043C\u0430",
881
+ "\u0432\u0440\u0435\u043C\u0435",
882
+ "\u043D\u0438\u0448\u0442\u043E",
883
+ "\u0442\u0438\u0435",
884
+ "\u0442\u0430\u043C\u0443",
885
+ "\u0441\u043C\u0435",
886
+ "\u043C\u043E\u0440\u0430",
887
+ "\u0441\u0442\u0435",
888
+ "\u043F\u0440\u0435\u0434",
889
+ "\u043C\u043E\u0436\u0435\u0431\u0438",
890
+ "\u0438\u043C\u0430\u043C",
891
+ "\u0432\u0435",
892
+ "\u043C\u0430\u043B\u043A\u0443",
893
+ "\u043C\u043E\u043B\u0430\u043C",
894
+ "\u043D\u0438\u0435",
895
+ "\u0435\u0458",
896
+ "\u043E\u0434\u0438\u043C\u0435",
897
+ "\u043D\u0430\u0432\u0438\u0441\u0442\u0438\u043D\u0430",
898
+ "\u043C\u043E\u0436\u0435\u0448",
899
+ "\u043A\u043E\u043B\u043A\u0443",
900
+ "\u043D\u0435\u0433\u043E",
901
+ "\u043C\u043E\u0458\u043E\u0442",
902
+ "\u0440\u0430\u0431\u043E\u0442\u0430",
903
+ "\u0442\u043E\u0433\u0430\u0448",
904
+ "\u043E\u0434\u0438",
905
+ "\u0437\u0434\u0440\u0430\u0432\u043E",
906
+ "\u0442\u043E\u043B\u043A\u0443",
907
+ "\u043D\u0438\u043A\u043E\u0433\u0430\u0448",
908
+ "\u0438\u043C\u0430\u0448",
909
+ "\u0432\u0438\u0435",
910
+ "\u043C\u0438\u0441\u043B\u0438\u0448",
911
+ "\u0438\u043C",
912
+ "\u043A\u043E\u0438",
913
+ "\u0437\u043D\u0430\u0447\u0438",
914
+ "\u0437\u0430\u0442\u043E\u0430",
915
+ "\u0435\u0434\u043D\u0430",
916
+ "\u0431\u043B\u0430\u0433\u043E\u0434\u0430\u0440\u0430\u043C",
917
+ "\u043C\u043E\u0458\u0430\u0442\u0430",
918
+ "\u043E\u0432\u043E\u0458",
919
+ "\u043A\u043E\u0458\u0430",
920
+ "\u0442\u0430\u0442\u043A\u043E",
921
+ "\u0440\u0430\u0431\u043E\u0442\u0438",
922
+ "\u043D\u0435\u043B\u0438",
923
+ "\u0431\u0435\u0437",
924
+ "\u043E\u0434\u0430\u043C",
925
+ "\u043A\u0430\u0436\u0430\u043C",
926
+ "\u0447\u043E\u0432\u0435\u043A",
927
+ "\u0438\u043C\u0430\u043C\u0435",
928
+ "\u0433\u043E\u0434\u0438\u043D\u0438",
929
+ "\u043D\u0430\u0441",
930
+ "\u0434\u0435\u043D",
931
+ "\u043E\u0432\u0430\u0430",
932
+ "\u0444\u0430\u043B\u0430",
933
+ "\u043F\u0440\u0430\u0432\u0438\u0448",
934
+ "\u043A\u0430\u0458",
935
+ "\u0441\u0430\u043A\u0430",
936
+ "\u043D\u0430\u043F\u0440\u0430\u0432\u0438",
937
+ "\u043E",
938
+ "\u0440\u0435\u0447\u0435",
939
+ "\u0432\u0438\u0434\u0438",
940
+ "\u0442\u0432\u043E\u0458\u043E\u0442",
941
+ "\u0431\u0435\u0432",
942
+ "\u0438\u0437\u0433\u043B\u0435\u0434\u0430",
943
+ "\u0434\u043E\u0434\u0435\u043A\u0430",
944
+ "\u0441\u043B\u0443\u0447\u0438",
945
+ "\u043A\u0430\u0436\u0438",
946
+ "\u043F\u0430\u0442",
947
+ "\u0432\u0430\u0441",
948
+ "\u0442\u0432\u043E\u0458\u0430\u0442\u0430",
949
+ "\u043D\u0438\u043A\u043E\u0458",
950
+ "\u043C\u0438\u0441\u043B\u0435\u0432",
951
+ "\u043C\u0435\u0441\u0442\u043E",
952
+ "\u043E\u0432\u0438\u0435",
953
+ "\u0434\u043E\u043C\u0430",
954
+ "\u043C\u0430\u0458\u043A\u0430",
955
+ "\u0441\u0435\u043A\u043E\u0433\u0430\u0448",
956
+ "\u0436\u0430\u043B",
957
+ "\u043C\u043E\u0440\u0430\u043C"
958
+ ],
959
+ kk: [
960
+ "\u043C\u0435\u043D",
961
+ "\u043E\u043B",
962
+ "\u0441\u0435\u043D",
963
+ "\u043D\u0435",
964
+ "\u043A\u0435\u0440\u0435\u043A",
965
+ "\u043E\u043D\u044B",
966
+ "\u0431\u0430",
967
+ "\u043C\u0430",
968
+ "\u0431\u0430\u0440",
969
+ "\u0431\u0435",
970
+ "\u0434\u0435\u043F",
971
+ "\u0435\u043C\u0435\u0441",
972
+ "\u0441\u0456\u0437",
973
+ "\u0431\u0456\u0437",
974
+ "\u0430\u043B",
975
+ "\u043C\u0435\u043D\u0456",
976
+ "\u0434\u0435",
977
+ "\u0436\u0430\u0442\u044B\u0440",
978
+ "\u0435\u043D\u0434\u0456",
979
+ "\u043E\u043B\u0430\u0440",
980
+ "\u0431\u0456\u0440",
981
+ "\u0434\u0430",
982
+ "\u0441\u0435\u043D\u0456",
983
+ "\u043D\u0435\u0433\u0435",
984
+ "\u0430\u0434\u0430\u043C",
985
+ "\u043E\u0441\u044B\u043D\u0434\u0430",
986
+ "\u0431\u043E\u043B\u0430\u0434\u044B",
987
+ "\u043A\u0435\u043B\u0435\u0434\u0456",
988
+ "\u0434\u043E\u043A\u0442\u043E\u0440",
989
+ "\u043C\u044B\u043D\u0430",
990
+ "\u0431\u043E\u043B\u0434\u044B",
991
+ "\u0442\u0435\u043A",
992
+ "\u0440\u0430\u0445\u043C\u0435\u0442",
993
+ "\u0441\u043E\u043B",
994
+ "\u0435\u0434\u0456",
995
+ "\u043C\u0435",
996
+ "\u0431\u043E\u043B\u044B\u043F",
997
+ "\u0435\u0433\u0435\u0440",
998
+ "\u043E\u0441\u044B",
999
+ "\u043C\u044B\u043D\u0434\u0430",
1000
+ "\u0441\u043E\u043B\u0430\u0439",
1001
+ "\u0435\u0434\u0456\u043C",
1002
+ "\u043C\u0456\u043D\u0435",
1003
+ "\u043F\u0440\u0435\u0437\u0438\u0434\u0435\u043D\u0442",
1004
+ "\u0435\u043A\u0435\u043D",
1005
+ "\u043A\u0435\u0442\u0442\u0456",
1006
+ "\u043A\u0456\u043C",
1007
+ "\u0431\u043E\u043B\u0441\u0430",
1008
+ "\u0431\u0456\u043B\u043C\u0435\u0439\u043C\u0456\u043D",
1009
+ "\u0431\u0456\u0437\u0433\u0435",
1010
+ "\u0436\u0430\u0440\u0430\u0439\u0434\u044B",
1011
+ "\u0430\u043B\u044B\u043F",
1012
+ "\u0441\u0456\u0437\u0434\u0456",
1013
+ "\u043F\u0435",
1014
+ "\u0441\u0456\u0437\u0433\u0435",
1015
+ "\u0436\u0435\u0440\u0434\u0435",
1016
+ "\u0448\u0435",
1017
+ "\u043A\u0435\u0442\u0442\u0456\u043A",
1018
+ "\u0431\u043E\u043B\u0430",
1019
+ "\u0448\u044B\u043D",
1020
+ "\u0436\u0435\u0440\u0434\u0435\u043D",
1021
+ "\u0430\u0434\u0430\u043C\u0434\u0430\u0440",
1022
+ "\u0442\u0443\u0440\u0430\u043B\u044B",
1023
+ "\u0436\u0435\u0440\u0433\u0435",
1024
+ "\u0441\u043E\u043D\u0434\u0430",
1025
+ "\u043A\u0435\u043B\u0434\u0456",
1026
+ "\u043C\u044B\u0440\u0437\u0430",
1027
+ "\u0442\u0456\u0440\u0456",
1028
+ "\u0434\u0435\u0433\u0435\u043D",
1029
+ "\u0431\u043E\u043B\u0430\u0440",
1030
+ "\u0431\u0456\u043B\u0435\u043C\u0456\u043D",
1031
+ "\u043E\u043D\u0434\u0430",
1032
+ "\u0430",
1033
+ "\u0431\u0456\u0440\u0433\u0435",
1034
+ "\u0442\u044B\u043C",
1035
+ "\u0441\u0435\u0431\u0435\u0431\u0456",
1036
+ "\u0435\u043A\u0435\u043D\u0456\u043D",
1037
+ "\u043A\u0435\u043B",
1038
+ "\u043A\u0435\u043B\u0433\u0435\u043D",
1039
+ "\u0442\u0430\u0443\u044B\u043F",
1040
+ "\u0434\u0435\u0439\u0434\u0456",
1041
+ "\u0431\u0435\u0440",
1042
+ "\u0442\u0456\u043F\u0442\u0456",
1043
+ "\u043E\u043B\u0430\u0440\u0434\u044B",
1044
+ "\u0430\u043B\u043C\u0430\u0439\u043C\u044B\u043D",
1045
+ "\u043E\u0441\u044B\u043B\u0430\u0439",
1046
+ "\u043E\u043B\u0430\u0439",
1047
+ "\u043E\u0434\u0430\u043D",
1048
+ "\u0431\u0435\u0440\u0456",
1049
+ "\u0436\u0430\u0442\u044B\u0440\u043C\u044B\u043D",
1050
+ "\u043C\u044B\u043D\u0430\u0443",
1051
+ "\u043A\u0435\u043B\u0435",
1052
+ "\u0443",
1053
+ "\u0430\u043F\u0430",
1054
+ "\u0441\u043E\u043D\u0434\u0430\u0439",
1055
+ "\u043A\u0435\u0448\u0456\u0440\u0456\u043C",
1056
+ "\u0435\u0448\u043A\u0456\u043C",
1057
+ "\u0431\u0456\u0437\u0434\u0456",
1058
+ "\u043F\u0430",
1059
+ "\u0456\u0437\u0434\u0435\u043F",
1060
+ "\u0442\u0438\u0456\u0441",
1061
+ "\u0441\u0435\u043D\u0434\u0435\u0440",
1062
+ "\u043C\u0435\u043D\u0434\u0435",
1063
+ "\u0434\u0435\u043D\u0435",
1064
+ "\u043A\u0435\u0439\u0456\u043D",
1065
+ "\u0442\u0435\u0437",
1066
+ "\u0442\u0430\u0440\u0442",
1067
+ "\u0431\u0456\u0440\u0435\u0443",
1068
+ "\u0431\u043E\u043B\u043C\u0430\u0439\u0434\u044B",
1069
+ "\u0430\u043D\u0430",
1070
+ "\u0442\u044B\u043D\u044B\u0448",
1071
+ "\u043E\u0439\u043B\u0430\u043F",
1072
+ "\u0435\u043A\u0456",
1073
+ "\u0431\u043E\u043B\u043C\u0430\u0441\u0430",
1074
+ "\u0430\u0439\u0442\u0442\u044B",
1075
+ "\u043A\u0435\u043B\u0434\u0456\u043C",
1076
+ "\u0430\u0440\u043D\u0430\u0439\u044B",
1077
+ "\u0431\u044B\u043B\u0430\u0439",
1078
+ "\u0448\u044B\u043D\u044B\u043C\u0435\u043D",
1079
+ "\u043A\u0435\u043B\u0456\u043F",
1080
+ "\u0431\u043E\u043B",
1081
+ "\u0431\u0430\u0440\u044B\u043F",
1082
+ "\u0430\u043B\u0442\u044B\u043D",
1083
+ "\u043E\u0442\u044B\u0440",
1084
+ "\u0431\u043E\u043B\u0443",
1085
+ "\u043A\u0456\u0448\u043A\u0435\u043D\u0435",
1086
+ "\u0430\u043B\u043C\u0430\u0439",
1087
+ "\u043A\u0435\u043B\u043C\u0435\u0439\u0434\u0456",
1088
+ "\u043A\u0435\u0440\u0435\u043C\u0435\u0442",
1089
+ "\u0430\u0434\u0430\u043C\u0434\u0430\u0440\u044B",
1090
+ "\u0434\u0435\u0439\u043C\u0456\u043D",
1091
+ "\u0436\u0435\u043F",
1092
+ "\u0431\u043E\u0439\u044B",
1093
+ "\u0441\u044D\u0440",
1094
+ "\u0434\u0435\u0433\u0435\u043D\u043C\u0435\u043D",
1095
+ "\u043A\u0435\u0439",
1096
+ "\u0431\u043E\u043F",
1097
+ "\u0442\u0435\u0437\u0456\u0440\u0435\u043A",
1098
+ "\u0431\u0456\u0440\u0430\u0437",
1099
+ "\u0431\u0430\u0440\u0430"
1100
+ ],
807
1101
  en: []
808
1102
  };
809
1103
 
@@ -835,6 +1129,28 @@ var BG_FREQUENT = [
835
1129
  "\u0438\u0437\u0432\u0435\u0441\u0442\u0435\u043D",
836
1130
  "\u0441\u0442\u0440\u0430\u043D\u0430"
837
1131
  ];
1132
+ var SR_FREQUENT = [
1133
+ "\u0432\u0435\u0441\u0442\u0438",
1134
+ "\u0432\u0438\u0434\u0435\u043E",
1135
+ "\u0433\u0440\u0430\u0434",
1136
+ "\u0434\u0440\u0436\u0430\u0432\u0430",
1137
+ "\u0434\u0430\u043D\u0430\u0441",
1138
+ "\u0436\u0438\u0432\u043E\u0442",
1139
+ "\u0459\u0443\u0434\u0438",
1140
+ "\u043F\u043E\u0441\u0430\u043E",
1141
+ "\u0441\u0440\u0431\u0438\u0458\u0430",
1142
+ "\u0441\u0440\u043F\u0441\u043A\u0438",
1143
+ "\u043F\u043E\u0437\u043D\u0430\u0442",
1144
+ "\u0437\u0435\u043C\u0459\u0430",
1145
+ "\u0432\u0440\u0435\u043C\u0435",
1146
+ "\u0434\u0435\u0446\u0430",
1147
+ "\u0433\u043E\u0434\u0438\u043D\u0435",
1148
+ "\u0447\u043E\u0432\u0435\u043A",
1149
+ "\u043D\u043E\u0432\u0438\u043D\u0435",
1150
+ "\u0441\u0432\u0435\u0442",
1151
+ "\u0434\u0430\u043D\u0430\u0448\u045A\u0438",
1152
+ "\u043D\u0435\u0434\u0435\u0459\u0430"
1153
+ ];
838
1154
  var uk = {
839
1155
  code: "uk",
840
1156
  iso6393: "ukr",
@@ -1016,7 +1332,120 @@ var en = {
1016
1332
  frequent: FREQUENT_GENERATED["en"] ?? []
1017
1333
  }
1018
1334
  };
1019
- var PROFILES = { uk, ru, be, bg, en };
1335
+ var sr = {
1336
+ code: "sr",
1337
+ iso6393: "srp",
1338
+ // Serbian Cyrillic azbuka (Вук), 30 letters. Distinctive vs the other Cyrillic
1339
+ // profiles: ђ ћ (sr-only) and ј љ њ џ (shared with mk, inert between them; the
1340
+ // word rungs decide sr↔mk). No ё й щ ъ ы ь э ю я.
1341
+ alphabet: "\u0430\u0431\u0432\u0433\u0434\u0452\u0435\u0436\u0437\u0438\u0458\u043A\u043B\u0459\u043C\u043D\u045A\u043E\u043F\u0440\u0441\u0442\u045B\u0443\u0444\u0445\u0446\u0447\u045F\u0448",
1342
+ words: {
1343
+ // Serbian-distinctive grammatical markers. `сам` (I am) is omitted: it
1344
+ // collides with Macedonian `сам` ("alone"). Shared forms with mk (и, да, се,
1345
+ // не, на, само, …) are omitted so set-difference is not cancelled to nothing.
1346
+ // The future markers use ћ, a letter no other profile has.
1347
+ function: [
1348
+ "\u0458\u0435",
1349
+ "\u0441\u0443",
1350
+ "\u0430\u043B\u0438",
1351
+ "\u045B\u0435",
1352
+ "\u045B\u0435\u043C\u043E",
1353
+ "\u045B\u0435\u0448",
1354
+ "\u0437\u0431\u043E\u0433",
1355
+ "\u0458\u0435\u0440",
1356
+ "\u043E\u0432\u0430\u0458",
1357
+ "\u043A\u0430\u0434\u0430",
1358
+ "\u0443\u0432\u0435\u043A",
1359
+ "\u0441\u0430\u0434\u0430"
1360
+ ],
1361
+ // Hand-curated Cyrillic fallback — see SR_FREQUENT. Flagged for review.
1362
+ frequent: FREQUENT_GENERATED["sr"] ?? SR_FREQUENT
1363
+ }
1364
+ };
1365
+ var mk = {
1366
+ code: "mk",
1367
+ iso6393: "mkd",
1368
+ // Macedonian alphabet, 31 letters. Distinctive: ѓ ќ ѕ (mk-only) plus ј љ њ џ
1369
+ // (shared with sr). No ё й щ ъ ы ь э ю я, no ђ ћ (those are Serbian).
1370
+ alphabet: "\u0430\u0431\u0432\u0433\u0434\u0453\u0435\u0436\u0437\u0455\u0438\u0458\u043A\u043B\u0459\u043C\u043D\u045A\u043E\u043F\u0440\u0441\u0442\u045C\u0443\u0444\u0445\u0446\u0447\u045F\u0448",
1371
+ words: {
1372
+ // Macedonian-distinctive grammatical markers. The future particle `ќе` and
1373
+ // `зошто`/`затоа` use ќ (mk-only); the clitics `го`/`ги` and `сум` (I am) are
1374
+ // distinctively Macedonian. Forms shared with sr (и, да, се, не, само, …) are
1375
+ // omitted; `вие`/`ние` are omitted as they collide with bg.
1376
+ function: [
1377
+ "\u0433\u043E",
1378
+ "\u0433\u0438",
1379
+ "\u0441\u0443\u043C",
1380
+ "\u045C\u0435",
1381
+ "\u0434\u0435\u043A\u0430",
1382
+ "\u0437\u043E\u0448\u0442\u043E",
1383
+ "\u0437\u0430\u0442\u043E\u0430",
1384
+ "\u0442\u043E\u0430",
1385
+ "\u043E\u0432\u0430\u0430",
1386
+ "\u043E\u0432\u043E\u0458",
1387
+ "\u0442\u043E\u0458",
1388
+ "\u0442\u0430\u0430",
1389
+ "\u0442\u0438\u0435",
1390
+ "\u0441\u0435\u0433\u0430",
1391
+ "\u0441\u0438\u0442\u0435"
1392
+ ],
1393
+ frequent: FREQUENT_GENERATED["mk"] ?? []
1394
+ }
1395
+ };
1396
+ var kk = {
1397
+ code: "kk",
1398
+ iso6393: "kaz",
1399
+ // Kazakh Cyrillic, 42 letters (post-1940). Distinctive Turkic letters ә ғ қ ң
1400
+ // ө ұ ү һ і set it apart from every Slavic profile; it keeps the Russian set
1401
+ // (и й ё ц щ ъ ы ь э ю я) for loanwords. Highly distinctive, so low-risk.
1402
+ alphabet: "\u0430\u04D9\u0431\u0432\u0433\u0493\u0434\u0435\u0451\u0436\u0437\u0438\u0439\u043A\u049B\u043B\u043C\u043D\u04A3\u043E\u04E9\u043F\u0440\u0441\u0442\u0443\u04B1\u04AF\u0444\u0445\u04BB\u0446\u0447\u0448\u0449\u044A\u044B\u0456\u044C\u044D\u044E\u044F",
1403
+ words: {
1404
+ // Kazakh function words — no overlap with any Slavic profile, so every form
1405
+ // is safely distinctive. (`ма`/`ба`/`бе`/`па` interrogative particles are
1406
+ // omitted: too short and ambiguous against Slavic two-letter clitics.)
1407
+ function: [
1408
+ "\u043C\u0435\u043D",
1409
+ "\u0441\u0435\u043D",
1410
+ "\u0431\u04B1\u043B",
1411
+ "\u0436\u04D9\u043D\u0435",
1412
+ "\u0431\u0430\u0440",
1413
+ "\u0436\u043E\u049B",
1414
+ "\u043E\u043B",
1415
+ "\u04AF\u0448\u0456\u043D",
1416
+ "\u043E\u043D\u044B",
1417
+ "\u043C\u0435\u043D\u0456\u04A3",
1418
+ "\u0441\u0435\u043D\u0456\u04A3",
1419
+ "\u043E\u043D\u044B\u04A3",
1420
+ "\u0431\u0456\u0440\u0430\u049B",
1421
+ "\u0435\u0433\u0435\u0440",
1422
+ "\u0441\u043E\u043B",
1423
+ "\u043E\u0441\u044B",
1424
+ "\u0431\u0456\u0437",
1425
+ "\u0441\u0456\u0437",
1426
+ "\u043E\u043B\u0430\u0440",
1427
+ "\u049B\u0430\u0439\u0434\u0430",
1428
+ "\u043D\u0435\u0433\u0435",
1429
+ "\u0441\u0435\u0431\u0435\u0431\u0456",
1430
+ "\u049B\u0430\u043D\u0434\u0430\u0439",
1431
+ "\u043C\u04B1\u043D\u0434\u0430",
1432
+ "\u0434\u0435\u043F",
1433
+ "\u0493\u043E\u0439"
1434
+ ],
1435
+ // Corpus-derived from the smaller `kk_full.txt` (no curated kk_50k exists).
1436
+ frequent: FREQUENT_GENERATED["kk"] ?? []
1437
+ }
1438
+ };
1439
+ var PROFILES = {
1440
+ uk,
1441
+ ru,
1442
+ be,
1443
+ bg,
1444
+ en,
1445
+ sr,
1446
+ mk,
1447
+ kk
1448
+ };
1020
1449
  var PROFILED_CODES = Object.keys(PROFILES);
1021
1450
  function hasProfile(code) {
1022
1451
  return Object.prototype.hasOwnProperty.call(PROFILES, code);
@@ -1025,6 +1454,6 @@ function getProfiles(codes) {
1025
1454
  return codes.map((c) => PROFILES[c]).filter((p) => p !== void 0);
1026
1455
  }
1027
1456
 
1028
- export { PROFILED_CODES, PROFILES, be, bg, en, getProfiles, hasProfile, ru, uk };
1457
+ export { PROFILED_CODES, PROFILES, be, bg, en, getProfiles, hasProfile, kk, mk, ru, sr, uk };
1029
1458
  //# sourceMappingURL=profiles.js.map
1030
1459
  //# sourceMappingURL=profiles.js.map