@sarmay/kaz-converter 0.1.0
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.
- package/README.md +75 -0
- package/dist/index.cjs +1512 -0
- package/dist/index.d.cts +82 -0
- package/dist/index.d.ts +82 -0
- package/dist/index.js +1480 -0
- package/package.json +52 -0
package/dist/index.js
ADDED
|
@@ -0,0 +1,1480 @@
|
|
|
1
|
+
// src/disambiguation.ts
|
|
2
|
+
var NoopDisambiguator = class {
|
|
3
|
+
disambiguate(rawTokens) {
|
|
4
|
+
return rawTokens.map(([, converted]) => converted);
|
|
5
|
+
}
|
|
6
|
+
};
|
|
7
|
+
|
|
8
|
+
// src/arb2syr.ts
|
|
9
|
+
var CONSONANTS = {
|
|
10
|
+
"\u0628": "\u0431",
|
|
11
|
+
"\u06C6": "\u0432",
|
|
12
|
+
"\u06AF": "\u0433",
|
|
13
|
+
"\u0639": "\u0493",
|
|
14
|
+
"\u062F": "\u0434",
|
|
15
|
+
"\u062C": "\u0436",
|
|
16
|
+
"\u0632": "\u0437",
|
|
17
|
+
"\u0643": "\u043A",
|
|
18
|
+
"\u0642": "\u049B",
|
|
19
|
+
"\u0644": "\u043B",
|
|
20
|
+
"\u0645": "\u043C",
|
|
21
|
+
"\u0646": "\u043D",
|
|
22
|
+
"\u06AD": "\u04A3",
|
|
23
|
+
"\u067E": "\u043F",
|
|
24
|
+
"\u0631": "\u0440",
|
|
25
|
+
"\u0633": "\u0441",
|
|
26
|
+
"\u062A": "\u0442",
|
|
27
|
+
"\u0641": "\u0444",
|
|
28
|
+
"\u062D": "\u0445",
|
|
29
|
+
"\u06BE": "\u04BB",
|
|
30
|
+
"\u0686": "\u0447",
|
|
31
|
+
"\u0634": "\u0448"
|
|
32
|
+
};
|
|
33
|
+
var VOWEL_MAP = {
|
|
34
|
+
"\u0627": { b: "\u0430", f: "\u04D9" },
|
|
35
|
+
"\u0649": { b: "\u044B", f: "\u0456" },
|
|
36
|
+
"\u0648": { b: "\u043E", f: "\u04E9" },
|
|
37
|
+
"\u06C7": { b: "\u04B1", f: "\u04AF" },
|
|
38
|
+
"\u06D5": "\u0435",
|
|
39
|
+
"\u06CB": "\u0443"
|
|
40
|
+
};
|
|
41
|
+
var EXCEPTIONS = {
|
|
42
|
+
"\u0631\u06D5\u0633\u067E\u06CB\u0628\u0644\u064A\u0643\u0627": "\u0440\u0435\u0441\u043F\u0443\u0431\u043B\u0438\u043A\u0430",
|
|
43
|
+
"\u0643\u0648\u0645\u0645\u06CB\u0646\u064A\u0633\u062A\u0649\u0643": "\u043A\u043E\u043C\u043C\u0443\u043D\u0438\u0441\u0442\u0456\u043A",
|
|
44
|
+
"\u06D5\u0643\u0631\u0627\u0646": "\u044D\u043A\u0440\u0430\u043D",
|
|
45
|
+
"\u06D5\u0646\u06D5\u0631\u06AF\u06D5\u062A\u064A\u0643\u0627": "\u044D\u043D\u0435\u0440\u0433\u0435\u0442\u0438\u043A\u0430",
|
|
46
|
+
"\u06D5\u0646\u06D5\u0631\u06AF\u064A\u064A\u0627": "\u044D\u043D\u0435\u0440\u0433\u0438\u044F",
|
|
47
|
+
"\u0643\u0648\u0645\u064A\u062A\u06D5\u062A": "\u043A\u043E\u043C\u0438\u0442\u0435\u0442",
|
|
48
|
+
"\u0643\u0648\u0646\u062A\u0633\u06D5\u0631\u062A": "\u043A\u043E\u043D\u0446\u0435\u0440\u0442",
|
|
49
|
+
"\u0643\u0648\u0633\u0645\u0648\u0633": "\u043A\u043E\u0441\u043C\u043E\u0441",
|
|
50
|
+
"\u0643\u0648\u0644\u0644\u06D5\u0643\u062A\u0649\u064A\u06C6": "\u043A\u043E\u043B\u043B\u0435\u043A\u0442\u0438\u0432",
|
|
51
|
+
"\u0643\u0648\u0644\u0644\u06D5\u06AF\u0627": "\u043A\u043E\u043B\u043B\u0435\u0433\u0430",
|
|
52
|
+
"\u0643\u0648\u0631\u067E\u06CB\u0633": "\u043A\u043E\u0440\u043F\u0443\u0441",
|
|
53
|
+
"\u0643\u0648\u0646\u0633\u062A\u0649\u064A\u062A\u06CB\u062A\u0633\u064A\u064A\u0627": "\u043A\u043E\u043D\u0441\u0442\u0438\u0442\u0443\u0446\u0438\u044F",
|
|
54
|
+
"\u0643\u0648\u0645\u0645\u06CB\u0646\u064A\u0633\u062A": "\u043A\u043E\u043C\u043C\u0443\u043D\u0438\u0441\u0442",
|
|
55
|
+
"\u0631\u0627\u064A\u0648\u0646": "\u0440\u0430\u0439\u043E\u043D",
|
|
56
|
+
"\u0631\u0627\u062F\u064A\u0648": "\u0440\u0430\u0434\u0438\u043E",
|
|
57
|
+
"\u0643\u0627\u0631\u062A\u0627": "\u043A\u0430\u0440\u0442\u0430",
|
|
58
|
+
"\u0627\u0631\u062D\u0649\u064A\u062A\u06D5\u0643\u062A\u06CB\u0631\u0627": "\u0430\u0440\u0445\u0438\u0442\u0435\u043A\u0442\u0443\u0440\u0430",
|
|
59
|
+
"\u0627\u06C6\u062A\u0648\u0646\u0648\u0645\u064A\u0627": "\u0430\u0432\u0442\u043E\u043D\u043E\u043C\u0438\u044F",
|
|
60
|
+
"\u06C6\u0649\u064A\u062F\u064A\u0648": "\u0432\u0438\u0434\u0435\u043E",
|
|
61
|
+
"\u06C6\u0649\u064A\u062F\u06D5\u0648": "\u0432\u0438\u0434\u0435\u043E",
|
|
62
|
+
"\u0628\u0649\u064A\u0648\u0644\u0648\u06AF\u064A\u0627": "\u0431\u0438\u043E\u043B\u043E\u0433\u0438\u044F",
|
|
63
|
+
"\u0643\u0648\u06D5\u0641\u0641\u064A\u062A\u0633\u0649\u064A\u06D5\u0646\u062A": "\u043A\u043E\u044D\u0444\u0444\u0438\u0446\u0438\u0435\u043D\u0442",
|
|
64
|
+
"\u0643\u0648\u06D5\u0641\u0641\u064A\u062A\u0633\u064A\u06D5\u0646\u062A": "\u043A\u043E\u044D\u0444\u0444\u0438\u0446\u0438\u0435\u043D\u0442",
|
|
65
|
+
"\u067E\u0631\u0648\u062A\u0633\u06D5\u0633": "\u043F\u0440\u043E\u0446\u0435\u0441\u0441",
|
|
66
|
+
"\u062A\u06D5\u0644\u06D5\u06C6\u0649\u064A\u0632\u0648\u0631": "\u0442\u0435\u043B\u0435\u0432\u0438\u0437\u043E\u0440",
|
|
67
|
+
"\u062A\u06D5\u0644\u06D5\u06C6\u064A\u0632\u0648\u0631": "\u0442\u0435\u043B\u0435\u0432\u0438\u0437\u043E\u0440",
|
|
68
|
+
"\u0674\u0627\u0631\u062F\u0627\u064A\u0649\u0645": "\u04D9\u0440\u0434\u0430\u0439\u044B\u043C",
|
|
69
|
+
"\u067E\u0627\u0631\u062A\u064A\u0627": "\u043F\u0430\u0440\u0442\u0438\u044F",
|
|
70
|
+
"\u0643\u0648\u0645\u067E\u064A\u06CB\u062A\u06D5\u0631": "\u043A\u043E\u043C\u043F\u044C\u044E\u0442\u0435\u0440",
|
|
71
|
+
"\u062A\u06D5\u0644\u06D5\u0641\u0648\u0646": "\u0442\u0435\u043B\u0435\u0444\u043E\u043D",
|
|
72
|
+
"\u064A\u0646\u062A\u06D5\u0631\u0646\u06D5\u062A": "\u0438\u043D\u0442\u0435\u0440\u043D\u0435\u0442",
|
|
73
|
+
"\u062F\u06D5\u0645\u0648\u0643\u0631\u0627\u062A\u064A\u064A\u0627": "\u0434\u0435\u043C\u043E\u043A\u0440\u0430\u0442\u0438\u044F",
|
|
74
|
+
"\u062F\u06D5\u0645\u0648\u0643\u0631\u0627\u062A\u064A\u0627": "\u0434\u0435\u043C\u043E\u043A\u0440\u0430\u0442\u0438\u044F",
|
|
75
|
+
"\u06D5\u0643\u0648\u0646\u0648\u0645\u064A\u0643\u0627": "\u044D\u043A\u043E\u043D\u043E\u043C\u0438\u043A\u0430",
|
|
76
|
+
"\u067E\u0648\u0644\u064A\u062A\u064A\u0643\u0627": "\u043F\u043E\u043B\u0438\u0442\u0438\u043A\u0430",
|
|
77
|
+
"\u0643\u0648\u0631\u0649\u062F\u0648\u0631": "\u043A\u043E\u0440\u0438\u0434\u043E\u0440",
|
|
78
|
+
"\u0643\u0648\u0646\u06AF\u0631\u06D5\u0633": "\u043A\u043E\u043D\u0433\u0440\u0435\u0441\u0441",
|
|
79
|
+
"\u06D5\u0644\u06D5\u0645\u06D5\u0646\u062A": "\u044D\u043B\u0435\u043C\u0435\u043D\u0442",
|
|
80
|
+
"\u062A\u06D5\u062D\u0646\u0648\u0644\u0648\u06AF\u0649\u064A\u064A\u0627": "\u0442\u0435\u0445\u043D\u043E\u043B\u043E\u0433\u0438\u044F",
|
|
81
|
+
"\u062A\u06D5\u062D\u0646\u0648\u0644\u0648\u06AF\u0649\u064A\u0649\u0627": "\u0442\u0435\u0445\u043D\u043E\u043B\u043E\u0433\u0438\u044F",
|
|
82
|
+
"\u062A\u06D5\u062D\u0646\u0648\u0644\u0648\u06AF\u0649\u064A\u0627": "\u0442\u0435\u0445\u043D\u043E\u043B\u043E\u0433\u0438\u044F",
|
|
83
|
+
"\u0628\u0649\u064A\u0648\u0644\u0648\u06AF\u0649\u064A\u064A\u0627": "\u0431\u0438\u043E\u043B\u043E\u0433\u0438\u044F",
|
|
84
|
+
"\u0628\u064A\u0648\u0644\u0648\u06AF\u064A\u0627": "\u0431\u0438\u043E\u043B\u043E\u0433\u0438\u044F",
|
|
85
|
+
"\u0649\u064A\u0646\u0633\u062A\u0631\u06CB\u0645\u06D5\u0646\u062A": "\u0438\u043D\u0441\u0442\u0440\u0443\u043C\u0435\u043D\u0442",
|
|
86
|
+
"\u064A\u0646\u0633\u062A\u0631\u06CB\u0645\u06D5\u0646\u062A": "\u0438\u043D\u0441\u0442\u0440\u0443\u043C\u0435\u043D\u0442",
|
|
87
|
+
"\u067E\u0648\u062F\u06D5\u0632\u062F": "\u043F\u043E\u0434\u044A\u0435\u0437\u0434",
|
|
88
|
+
"\u0643\u0648\u0646\u0633\u062A\u064A\u062A\u06CB\u062A\u0633\u0649\u064A\u064A\u0627": "\u043A\u043E\u043D\u0441\u0442\u0438\u0442\u0443\u0446\u0438\u044F",
|
|
89
|
+
"\u0633\u062A\u0627\u0646\u062A\u0633\u0649\u064A\u064A\u0627": "\u0441\u0442\u0430\u043D\u0446\u0438\u044F",
|
|
90
|
+
"\u0628\u0649\u0631": "\u0431\u0456\u0440",
|
|
91
|
+
"\u06C7\u0634": "\u04AF\u0448",
|
|
92
|
+
"\u062A\u0648\u0631\u062A": "\u0442\u04E9\u0440\u0442",
|
|
93
|
+
"\u0649\u0633": "\u0456\u0441",
|
|
94
|
+
"\u062F\u064A\u0633\u0633\u06D5\u0631\u062A\u0627\u062A\u0633\u0649\u064A\u064A\u0627": "\u0434\u0438\u0441\u0441\u0435\u0440\u0442\u0430\u0446\u0438\u044F",
|
|
95
|
+
"\u062F\u064A\u0633\u0633\u06D5\u0631\u062A\u0627\u062A\u0633\u064A\u0627": "\u0434\u0438\u0441\u0441\u0435\u0440\u0442\u0430\u0446\u0438\u044F",
|
|
96
|
+
"\u064A\u0646\u062A\u06D5\u06AF\u0631\u0627\u062A\u0633\u0649\u064A\u064A\u0627": "\u0438\u043D\u0442\u0435\u0433\u0440\u0430\u0446\u0438\u044F",
|
|
97
|
+
"\u064A\u0646\u062A\u06D5\u06AF\u0631\u0627\u062A\u0633\u064A\u0627": "\u0438\u043D\u0442\u0435\u0433\u0440\u0430\u0446\u0438\u044F",
|
|
98
|
+
"\u062A\u0631\u0627\u0646\u0633\u0641\u0648\u0631\u0645\u0627\u062A\u0633\u064A\u0627": "\u0442\u0440\u0430\u043D\u0441\u0444\u043E\u0440\u043C\u0430\u0446\u0438\u044F",
|
|
99
|
+
"\u0645\u0627\u062C\u0649\u0644\u0649\u0633": "\u043C\u04D9\u0436\u0456\u043B\u0456\u0441",
|
|
100
|
+
"\u0674\u062A\u0648\u0631\u0627\u0639\u0627": "\u0442\u04E9\u0440\u0430\u0493\u0430",
|
|
101
|
+
"\u0627\u0643\u0627\u062F\u06D5\u0645\u0649\u064A\u064A\u0627": "\u0430\u043A\u0430\u0434\u0435\u043C\u0438\u044F",
|
|
102
|
+
"\u0634\u0649\u0645\u0643\u06D5\u0646\u062A": "\u0428\u044B\u043C\u043A\u0435\u043D\u0442",
|
|
103
|
+
"\u0627\u0644\u0645\u0627\u062A\u0649": "\u0410\u043B\u043C\u0430\u0442\u044B",
|
|
104
|
+
"\u0627\u0633\u062A\u0627\u0646\u0627": "\u0410\u0441\u0442\u0430\u043D\u0430",
|
|
105
|
+
"\u0642\u0627\u0632\u0627\u0642\u0633\u062A\u0627\u0646": "\u049A\u0430\u0437\u0430\u049B\u0441\u0442\u0430\u043D",
|
|
106
|
+
"\u062C\u06C7\u06AD\u06AF\u0648": "\u0416\u04B1\u04A3\u0433\u043E",
|
|
107
|
+
"\u0634\u064A": "\u0421\u0438",
|
|
108
|
+
"\u062C\u064A\u0646\u067E\u064A\u06AD": "\u0426\u0437\u0438\u043D\u044C\u043F\u0438\u043D",
|
|
109
|
+
"\u0643\u0649\u062A\u0627\u067E": "\u043A\u0456\u0442\u0430\u043F",
|
|
110
|
+
"\u0631\u0627\u062D\u0645\u06D5\u062A": "\u0440\u0430\u0445\u043C\u0435\u0442",
|
|
111
|
+
"\u0627\u06CB\u0649\u0644": "\u0430\u0443\u044B\u043B",
|
|
112
|
+
"\u06AF\u0628": "\u0413\u0411",
|
|
113
|
+
"\u067E\u0631\u0648\u06AF\u0631\u06D5\u0633": "\u043F\u0440\u043E\u0433\u0440\u0435\u0441\u0441",
|
|
114
|
+
"\u0674\u0648\u0632\u0627\u0631\u0627": "\u04E9\u0437\u0430\u0440\u0430",
|
|
115
|
+
"\u062C\u0627\u06CB\u0627\u067E\u0643\u06D5\u0631\u0634\u0649\u0644\u0649\u0643": "\u0436\u0430\u0443\u0430\u043F\u043A\u0435\u0440\u0448\u0456\u043B\u0456\u043A",
|
|
116
|
+
"\u06D5\u0644\u06D5\u0643\u062A\u0631\u0644\u06D5\u0646\u062F\u0649\u0631\u06CB": "\u044D\u043B\u0435\u043A\u0442\u0440\u043B\u0435\u043D\u0434\u0456\u0440\u0443",
|
|
117
|
+
"\u0674\u0649\u0632\u0628\u0627\u0633\u0627\u0631": "\u0456\u0437\u0431\u0430\u0441\u0430\u0440",
|
|
118
|
+
"\u0674\u0627\u062F\u0649\u0633-\u062A\u0627\u0633\u0649\u0644": "\u04D9\u0434\u0456\u0441-\u0442\u04D9\u0441\u0456\u043B",
|
|
119
|
+
"\u0674\u062C\u0648\u0646-\u062C\u0648\u0633\u0649\u0642\u0633\u0649\u0632": "\u0436\u04E9\u043D-\u0436\u043E\u0441\u044B\u049B\u0441\u044B\u0437",
|
|
120
|
+
"\u0643\u0648\u0646\u0633\u06D5\u067E\u062A\u0633\u0649\u064A\u064A\u0627": "\u043A\u043E\u043D\u0446\u0435\u043F\u0446\u0438\u044F",
|
|
121
|
+
"\u0643\u0648\u0646\u0633\u06D5\u067E\u062A\u0633\u064A\u064A\u0627": "\u043A\u043E\u043D\u0446\u0435\u043F\u0446\u0438\u044F",
|
|
122
|
+
"\u0643\u0648\u0646\u0633\u06D5\u067E\u062A\u0633\u064A\u0627": "\u043A\u043E\u043D\u0446\u0435\u043F\u0446\u0438\u044F",
|
|
123
|
+
"\u0633\u0649\u064A\u0641\u0631\u0644\u0649\u0642": "\u0446\u0438\u0444\u0440\u043B\u044B\u049B",
|
|
124
|
+
"\u06D5\u06C6\u0648\u0644\u06CB\u062A\u0633\u0649\u064A\u064A\u0627": "\u044D\u0432\u043E\u043B\u044E\u0446\u0438\u044F",
|
|
125
|
+
"\u06D5\u06C6\u0648\u0644\u064A\u06CB\u062A\u0633\u064A\u0627": "\u044D\u0432\u043E\u043B\u044E\u0446\u0438\u044F",
|
|
126
|
+
"\u0645\u06D5\u0674\u062A\u0649\u0644\u0643\u06D5\u0634\u06D5": "\u043C\u04D9\u0442\u0456\u043B\u043A\u0435\u0448\u0435",
|
|
127
|
+
"\u062F\u0649\u064A\u0627\u06AF\u0646\u0648\u0632": "\u0434\u0438\u0430\u0433\u043D\u043E\u0437",
|
|
128
|
+
"\u0628\u06CB\u062F\u062C\u06D5\u062A": "\u0431\u044E\u0434\u0436\u0435\u0442",
|
|
129
|
+
"\u0641\u0649\u064A\u0644\u0645": "\u0444\u0438\u043B\u044C\u043C",
|
|
130
|
+
"\u0627\u0633\u0641\u0627\u0644\u062A": "\u0430\u0441\u0444\u0430\u043B\u044C\u0442",
|
|
131
|
+
"\u06D5\u06CB\u0631\u0648\u067E\u0627": "\u0415\u0443\u0440\u043E\u043F\u0430",
|
|
132
|
+
"\u062A\u0648\u0643\u0649\u064A\u0648": "\u0422\u043E\u043A\u0438\u043E",
|
|
133
|
+
"\u0646\u0649\u064A\u06CB-\u064A\u0648\u0631\u0643": "\u041D\u044C\u044E-\u0419\u043E\u0440\u043A",
|
|
134
|
+
"\u0628\u0627\u0646\u0643": "\u0431\u0430\u043D\u043A",
|
|
135
|
+
"\u0642\u0649\u064A\u0627\u0631": "\u049B\u0438\u044F\u0440",
|
|
136
|
+
"\u0628\u0649\u0631\u0627\u0642": "\u0431\u0456\u0440\u0430\u049B",
|
|
137
|
+
"\u062A\u0648\u0645\u06D5\u0646": "\u0442\u04E9\u043C\u0435\u043D",
|
|
138
|
+
"\u0645\u06D5\u064A\u0631\u0627\u0645": "\u043C\u0435\u0439\u0440\u0430\u043C",
|
|
139
|
+
"\u0645\u06D5\u064A\u0631\u0627\u0645\u062F\u0627\u0631\u0649\u0646\u0649\u06AD": "\u043C\u0435\u0439\u0440\u0430\u043C\u0434\u0430\u0440\u044B\u043D\u044B\u04A3",
|
|
140
|
+
"\u062F\u0627\u0633\u062A\u06C7\u0631": "\u0434\u04D9\u0441\u0442\u04AF\u0440",
|
|
141
|
+
"\u062F\u0627\u0633\u062A\u06C7\u0631\u0644\u06D5\u0631": "\u0434\u04D9\u0441\u0442\u04AF\u0440\u043B\u0435\u0440",
|
|
142
|
+
"\u062F\u06D5\u0646\u0633\u0627\u06CB\u0644\u0649\u0642": "\u0434\u0435\u043D\u0441\u0430\u0443\u043B\u044B\u049B",
|
|
143
|
+
"\u062A\u0627\u0631\u064A\u062D": "\u0442\u0430\u0440\u0438\u0445",
|
|
144
|
+
"\u06AF\u0631\u0627\u0645\u0645\u0627\u062A\u0649\u0643\u0627": "\u0433\u0440\u0430\u043C\u043C\u0430\u0442\u0438\u043A\u0430",
|
|
145
|
+
"\u06AF\u0631\u0627\u0645\u0645\u0627\u062A\u0649\u0643\u0627\u0644\u0649\u0642": "\u0433\u0440\u0430\u043C\u043C\u0430\u0442\u0438\u043A\u0430\u043B\u044B\u049B",
|
|
146
|
+
"\u062C \u0643 \u067E": "\u0416\u041A\u041F",
|
|
147
|
+
"\u0627\u0633\u0643\u06D5\u0631\u064A": "\u04D9\u0441\u043A\u0435\u0440\u0438",
|
|
148
|
+
"\u0628\u0627\u062A\u0649\u0644": "\u0431\u0430\u0442\u044B\u043B",
|
|
149
|
+
"\u0644\u064A \u0686\u064A\u0627\u06AD": "\u041B\u0438 \u0427\u044F\u04A3",
|
|
150
|
+
"\u062C\u0627\u06CB \u0644\u0649\u062C\u064A": "\u0416\u0430\u0443 \u041B\u044B\u0436\u0438",
|
|
151
|
+
"\u06CB\u0627\u06AD \u062D\u06CB\u0646\u064A\u06AD": "\u0423\u0430\u04A3 \u0425\u0443\u043D\u0438\u04A3",
|
|
152
|
+
"\u0633\u0627\u064A \u0686\u064A": "\u0421\u0430\u0439 \u0427\u0438",
|
|
153
|
+
"\u062F\u064A\u06AD \u0634\u06CB\u06D5\u0634\u064A\u0627\u06AD": "\u0414\u0438\u04A3 \u0428\u0443\u0435\u0448\u044F\u04A3",
|
|
154
|
+
"\u0644\u064A \u0634\u064A": "\u041B\u0438 \u0421\u0438",
|
|
155
|
+
"\u0633\u06CB\u0628\u062A\u0649\u064A\u062A\u0631": "\u0441\u0443\u0431\u0442\u0438\u0442\u0440",
|
|
156
|
+
"\u0633\u06CB\u0628\u062A\u0649\u064A\u062A\u0649\u0631\u0644\u06D5\u0631\u062F\u0649": "\u0441\u0443\u0431\u0442\u0438\u0442\u0440\u043B\u0435\u0440\u0434\u0456"
|
|
157
|
+
};
|
|
158
|
+
var LOANWORD_EXACT = /* @__PURE__ */ new Set([
|
|
159
|
+
"\u06CB\u0646\u0649\u064A\u06C6\u06D5\u0631\u0633\u0649\u064A\u062A\u06D5\u062A",
|
|
160
|
+
"\u06CB\u0646\u0649\u064A\u06C6\u06D5\u0631\u0633\u0649\u062A\u06D5\u062A",
|
|
161
|
+
"\u06CB\u0646\u0649\u06C6\u06D5\u0631\u0633\u0649\u062A\u06D5\u062A",
|
|
162
|
+
"\u06CB\u0646\u0649\u06C6\u06D5\u0631\u0633\u0649\u064A\u062A\u06D5\u062A",
|
|
163
|
+
"\u0643\u0648\u0646\u0633\u062A\u0649\u064A\u062A\u06CB\u062A\u0633\u0649\u064A\u0649\u0627",
|
|
164
|
+
"\u0643\u0648\u0646\u0633\u062A\u0649\u064A\u062A\u06CB\u062A\u0633\u0649\u064A\u064A\u0627",
|
|
165
|
+
"\u0643\u0648\u0646\u0633\u062A\u0649\u064A\u062A\u06CB\u062A\u0633\u0649\u064A\u0627",
|
|
166
|
+
"\u0633\u062A\u0627\u0646\u062A\u0633\u0649\u064A\u0649\u0627",
|
|
167
|
+
"\u0633\u062A\u0627\u0646\u062A\u0633\u0649\u064A\u0627",
|
|
168
|
+
"\u064A\u0646\u0641\u0648\u0631\u0645\u0627\u062A\u0633\u0649\u064A\u064A\u0627",
|
|
169
|
+
"\u064A\u0646\u0641\u0648\u0631\u0645\u0627\u062A\u0633\u0649\u064A\u0627",
|
|
170
|
+
"\u0645\u0627\u0634\u064A\u0646\u0627",
|
|
171
|
+
"\u0645\u0627\u0634\u0649\u0646\u0627",
|
|
172
|
+
"\u0627\u062A\u0648\u0645",
|
|
173
|
+
"\u0686\u06D5\u0645\u067E\u0649\u064A\u0648\u0646",
|
|
174
|
+
"\u0686\u06D5\u0645\u067E\u0649\u064A\u06C7\u0646",
|
|
175
|
+
"\u0634\u0634\u06D5\u062A\u0643\u0627"
|
|
176
|
+
]);
|
|
177
|
+
var LOANWORD_E_PREFIXES = [
|
|
178
|
+
"\u06D5\u0643\u0631\u0627\u0646",
|
|
179
|
+
"\u06D5\u0643\u0633\u067E",
|
|
180
|
+
"\u06D5\u0644\u06D5\u0643\u062A\u0631",
|
|
181
|
+
"\u06D5\u0646\u06D5\u0631\u06AF",
|
|
182
|
+
"\u06D5\u0643\u0648\u0644\u0648\u06AF",
|
|
183
|
+
"\u06D5\u062A\u0627\u067E",
|
|
184
|
+
"\u06D5\u0641\u064A\u0631",
|
|
185
|
+
"\u06D5\u0641\u0641\u06D5\u0643\u062A",
|
|
186
|
+
"\u06D5\u0643\u0648\u0646\u0648\u0645",
|
|
187
|
+
"\u06D5\u0644\u06D5\u0645\u06D5\u0646\u062A",
|
|
188
|
+
"\u06D5\u0633\u062A\u0631\u0627\u062F",
|
|
189
|
+
"\u06D5\u067E\u0648\u0633",
|
|
190
|
+
"\u06D5\u067E\u0649\u064A\u0632\u0648\u062F"
|
|
191
|
+
];
|
|
192
|
+
var LOANWORD_PREFIXES = [
|
|
193
|
+
"\u0627\u0631\u062D\u0649",
|
|
194
|
+
"\u0627\u0631\u062D\u064A",
|
|
195
|
+
"\u067E\u0631\u0648",
|
|
196
|
+
"\u067E\u0631\u0648\u06AF",
|
|
197
|
+
"\u067E\u0631\u0648\u062A\u0633",
|
|
198
|
+
"\u0648\u0632\u0627\u0631\u0627",
|
|
199
|
+
"\u0648\u0632\u0627\u0631\u06D5",
|
|
200
|
+
"\u062A\u06D5\u062D\u0646\u0648",
|
|
201
|
+
"\u06D5\u0643\u0648\u0646\u0648\u0645",
|
|
202
|
+
"\u06D5\u0643\u0627\u0646\u0648\u0645",
|
|
203
|
+
"\u06AF\u06CB\u0645\u0627\u0646",
|
|
204
|
+
"\u06AF\u0648\u0645\u0627\u0646",
|
|
205
|
+
"\u0631\u0627\u064A",
|
|
206
|
+
"\u0631\u0627\u062F",
|
|
207
|
+
"\u06C6\u0649\u064A\u062F",
|
|
208
|
+
"\u06C6\u064A\u062F",
|
|
209
|
+
"\u0643\u0648\u06D5\u0641\u0641",
|
|
210
|
+
"\u0643\u0648\u0626\u0641\u0641",
|
|
211
|
+
"\u0643\u0648\u0646\u0633\u06D5\u067E",
|
|
212
|
+
"\u0643\u0648\u0646\u062A\u0633\u06D5\u067E",
|
|
213
|
+
"\u0633\u0649\u064A\u0641\u0631",
|
|
214
|
+
"\u062A\u0633\u0649\u064A\u0641\u0631",
|
|
215
|
+
"\u06AF\u0631\u0627\u0641",
|
|
216
|
+
"\u06D5\u0646\u06D5\u0631\u06AF",
|
|
217
|
+
"\u06D5\u06C6\u0648\u0644",
|
|
218
|
+
"\u0627\u0631\u062D\u0649\u064A\u062A",
|
|
219
|
+
"\u0627\u0631\u062D\u064A\u062A",
|
|
220
|
+
"\u064A\u0646\u062A\u06D5\u0631",
|
|
221
|
+
"\u0643\u0648\u0646\u0633\u062A",
|
|
222
|
+
"\u0633\u062A\u062A",
|
|
223
|
+
"\u067E\u0648\u062F",
|
|
224
|
+
"\u0643\u0627\u0646\u0633\u062A",
|
|
225
|
+
"\u062F\u06D5\u0645\u0648\u0643\u0631",
|
|
226
|
+
"\u062F\u06D5\u0645\u0648",
|
|
227
|
+
"\u067E\u0648\u0644\u064A\u062A",
|
|
228
|
+
"\u062A\u06D5\u0644\u06D5\u0641",
|
|
229
|
+
"\u067E\u0648\u0644\u0649\u064A\u062A\u0633",
|
|
230
|
+
"\u067E\u0648\u0644\u064A\u062A\u0633",
|
|
231
|
+
"\u062F\u06D5\u067E\u0627\u0631\u062A",
|
|
232
|
+
"\u062F\u0649\u067E\u0627\u0631\u062A",
|
|
233
|
+
"\u0648\u067E\u06D5\u0631\u0627\u062A",
|
|
234
|
+
"\u0627\u0648\u067E\u06D5\u0631\u0627\u062A",
|
|
235
|
+
"\u0641\u06D5\u062F\u06D5\u0631\u0627\u0644",
|
|
236
|
+
"\u0641\u0649\u062F\u06D5\u0631\u0627\u0644",
|
|
237
|
+
"\u0643\u0631\u0649\u064A\u0645\u0649\u064A\u0646",
|
|
238
|
+
"\u0643\u0631\u064A\u0645\u064A\u0646",
|
|
239
|
+
"\u067E\u0633\u0649\u064A\u062D\u0648\u0644",
|
|
240
|
+
"\u067E\u0633\u064A\u062D\u0648\u0644",
|
|
241
|
+
"\u06C6\u0627\u0634\u0649\u064A\u0646\u06AF",
|
|
242
|
+
"\u06C6\u0627\u0634\u064A\u0646\u06AF",
|
|
243
|
+
"\u0627\u0645\u06D5\u0631\u0649\u064A\u0643",
|
|
244
|
+
"\u0627\u0645\u06D5\u0631\u064A\u0643",
|
|
245
|
+
"\u0627\u06C6\u062A\u0648\u0645\u0627\u062A",
|
|
246
|
+
"\u06CB\u0646\u0649\u064A\u06C6\u06D5\u0631\u0633",
|
|
247
|
+
"\u06CB\u0646\u064A\u06C6\u06D5\u0631\u0633",
|
|
248
|
+
"\u067E\u0631\u06D5\u0632\u0649\u064A\u062F",
|
|
249
|
+
"\u067E\u0631\u06D5\u0632\u064A\u062F",
|
|
250
|
+
"\u0649\u064A\u0646\u06C6\u06D5\u0633\u062A",
|
|
251
|
+
"\u064A\u0646\u06C6\u06D5\u0633\u062A",
|
|
252
|
+
"\u0631\u06D5\u0641\u0648\u0631\u0645",
|
|
253
|
+
"\u062F\u06D5\u0641\u0649\u064A\u062A\u0633",
|
|
254
|
+
"\u062F\u06D5\u0641\u064A\u062A\u0633",
|
|
255
|
+
"\u0643\u0648\u0648\u067E\u06D5\u0631",
|
|
256
|
+
"\u0643\u0648\u067E\u06D5\u0631",
|
|
257
|
+
"\u0643\u0648\u0646\u0641\u06D5\u0631\u06D5\u0646",
|
|
258
|
+
"\u0649\u064A\u0646\u0641\u0644\u0649\u064A\u0627\u062A",
|
|
259
|
+
"\u064A\u0646\u0641\u0644\u0649\u064A\u0627\u062A",
|
|
260
|
+
"\u0649\u064A\u0646\u0641\u0644\u064A\u0627\u062A",
|
|
261
|
+
"\u064A\u0646\u0641\u0644\u064A\u0627\u062A",
|
|
262
|
+
"\u0643\u0644\u064A\u0645\u0627\u062A",
|
|
263
|
+
"\u064A\u0646\u062A\u06D5\u0644\u0644\u06D5\u0643\u062A",
|
|
264
|
+
"\u06AF\u06D5\u0648\u0633\u0627\u064A\u0627\u0633",
|
|
265
|
+
"\u06AF\u06D5\u0648\u0633\u0627\u064A\u0627",
|
|
266
|
+
"\u0645\u0648\u062F\u06D5\u0644",
|
|
267
|
+
"\u062F\u06D5\u0645\u0648\u0643\u0631\u0627\u062A",
|
|
268
|
+
"\u062A\u0633\u0648\u0646",
|
|
269
|
+
"\u062A\u0633\u0627\u0646",
|
|
270
|
+
"\u062A\u0633\u06D5\u0646",
|
|
271
|
+
"\u062A\u0633\u06D5",
|
|
272
|
+
"\u062A\u0633\u0649\u064A",
|
|
273
|
+
"\u062A\u0633\u064A",
|
|
274
|
+
"\u0634\u0634\u06D5",
|
|
275
|
+
"\u0634\u0634\u0649",
|
|
276
|
+
"\u0634\u0634\u0648",
|
|
277
|
+
"\u0645\u0627\u062A\u0631",
|
|
278
|
+
"\u0645\u0627\u0634",
|
|
279
|
+
"\u0627\u062A\u0648\u0645",
|
|
280
|
+
"\u0633\u062A\u0627\u062A",
|
|
281
|
+
"\u0633\u062A\u0627\u0646\u062A",
|
|
282
|
+
"\u067E\u0648\u062F\u06D5",
|
|
283
|
+
"\u06AF\u0631\u0627\u0645",
|
|
284
|
+
"\u0633\u06CB\u0628\u062A"
|
|
285
|
+
];
|
|
286
|
+
var PROPER_NOUNS = {
|
|
287
|
+
"\u0642\u0627\u0632\u0627\u0642\u0633\u062A\u0627\u0646": "\u049A\u0430\u0437\u0430\u049B\u0441\u0442\u0430\u043D",
|
|
288
|
+
"\u0627\u0644\u0645\u0627\u062A\u0649": "\u0410\u043B\u043C\u0430\u0442\u044B",
|
|
289
|
+
"\u0627\u0633\u062A\u0627\u0646\u0627": "\u0410\u0441\u0442\u0430\u043D\u0430",
|
|
290
|
+
"\u0627\u0631\u0627\u0633\u062A\u0627\u0646\u0627": "\u0410\u0441\u0442\u0430\u043D\u0430"
|
|
291
|
+
};
|
|
292
|
+
var VALID_SUFFIXES = /* @__PURE__ */ new Set([
|
|
293
|
+
"\u0644\u0627\u0631",
|
|
294
|
+
"\u0644\u06D5\u0631",
|
|
295
|
+
"\u062F\u0627\u0631",
|
|
296
|
+
"\u062F\u06D5\u0631",
|
|
297
|
+
"\u062A\u0627\u0631",
|
|
298
|
+
"\u062A\u06D5\u0631",
|
|
299
|
+
"\u0646\u0649\u06AD",
|
|
300
|
+
"\u062F\u0649\u06AD",
|
|
301
|
+
"\u062A\u0649\u06AD",
|
|
302
|
+
"\u0639\u0627",
|
|
303
|
+
"\u06AF\u06D5",
|
|
304
|
+
"\u0642\u0627",
|
|
305
|
+
"\u0643\u06D5",
|
|
306
|
+
"\u0646\u0627",
|
|
307
|
+
"\u0646\u06D5",
|
|
308
|
+
"\u0627",
|
|
309
|
+
"\u06D5",
|
|
310
|
+
"\u0646\u0649",
|
|
311
|
+
"\u062F\u0649",
|
|
312
|
+
"\u062A\u0649",
|
|
313
|
+
"\u0646",
|
|
314
|
+
"\u062F\u0627",
|
|
315
|
+
"\u062F\u06D5",
|
|
316
|
+
"\u062A\u0627",
|
|
317
|
+
"\u062A\u06D5",
|
|
318
|
+
"\u0646\u062F\u0627",
|
|
319
|
+
"\u0646\u062F\u06D5",
|
|
320
|
+
"\u062F\u0627\u0646",
|
|
321
|
+
"\u062F\u06D5\u0646",
|
|
322
|
+
"\u062A\u0627\u0646",
|
|
323
|
+
"\u062A\u06D5\u0646",
|
|
324
|
+
"\u0646\u0627\u0646",
|
|
325
|
+
"\u0646\u06D5\u0646",
|
|
326
|
+
"\u0645\u06D5\u0646",
|
|
327
|
+
"\u0628\u06D5\u0646",
|
|
328
|
+
"\u067E\u06D5\u0646",
|
|
329
|
+
"\u0645",
|
|
330
|
+
"\u0645\u0649\u0632",
|
|
331
|
+
"\u06AD",
|
|
332
|
+
"\u06AD\u0649\u0632",
|
|
333
|
+
"\u0633\u0649",
|
|
334
|
+
"\u0649",
|
|
335
|
+
"\u0649\u0645\u044B\u0437",
|
|
336
|
+
"\u0649\u06AD\u0649\u0632",
|
|
337
|
+
"\u06D5\u0645\u0649\u0632",
|
|
338
|
+
"\u06D5\u06AD\u0649\u0632",
|
|
339
|
+
"\u0649\u0645",
|
|
340
|
+
"\u0649\u06AD",
|
|
341
|
+
"\u06D5\u0645",
|
|
342
|
+
"\u06D5\u06AD",
|
|
343
|
+
"\u0644\u0649\u0642",
|
|
344
|
+
"\u0644\u0649\u0643",
|
|
345
|
+
"\u062F\u0649\u0642",
|
|
346
|
+
"\u062F\u0649\u0643",
|
|
347
|
+
"\u062A\u0649\u0642",
|
|
348
|
+
"\u062A\u0649\u0643",
|
|
349
|
+
"\u0633\u0649\u0632",
|
|
350
|
+
"\u0634\u0649",
|
|
351
|
+
"\u0634\u0649\u0644\u062F\u0649\u0642",
|
|
352
|
+
"\u0634\u0649\u0644\u0649\u0643",
|
|
353
|
+
"\u062F\u0627\u0639\u0649",
|
|
354
|
+
"\u062F\u06D5\u06AF\u0649",
|
|
355
|
+
"\u062A\u0627\u0639\u0649",
|
|
356
|
+
"\u062A\u06D5\u06AF\u0649",
|
|
357
|
+
"\u0646\u062F\u0627\u0639\u0649",
|
|
358
|
+
"\u0646\u062F\u06D5\u06AF\u0649",
|
|
359
|
+
"\u0639\u0627\u0646",
|
|
360
|
+
"\u06AF\u06D5\u0646",
|
|
361
|
+
"\u0642\u0627\u0646",
|
|
362
|
+
"\u0643\u06D5\u0646",
|
|
363
|
+
"\u0645\u0627",
|
|
364
|
+
"\u0645\u06D5",
|
|
365
|
+
"\u0628\u0627",
|
|
366
|
+
"\u0628\u06D5",
|
|
367
|
+
"\u067E\u0627",
|
|
368
|
+
"\u067E\u06D5",
|
|
369
|
+
"\u067E",
|
|
370
|
+
"\u0649\u067E",
|
|
371
|
+
"\u06D5\u067E",
|
|
372
|
+
"\u06CB",
|
|
373
|
+
"\u06CB\u0634\u0649",
|
|
374
|
+
"\u062A\u06CB",
|
|
375
|
+
"\u062F\u06CB",
|
|
376
|
+
"\u0627\u0633\u0649\u06AD",
|
|
377
|
+
"\u06D5\u0633\u0649\u06AD",
|
|
378
|
+
"\u0627\u062F\u0649",
|
|
379
|
+
"\u06D5\u062F\u0649",
|
|
380
|
+
"\u064A\u062F\u0649",
|
|
381
|
+
"\u0645\u0649\u0646",
|
|
382
|
+
"\u0628\u0649\u0646",
|
|
383
|
+
"\u067E\u0649\u0646",
|
|
384
|
+
"\u0627\u0631",
|
|
385
|
+
"\u06D5\u0631",
|
|
386
|
+
"\u0645\u0627\u0633",
|
|
387
|
+
"\u0645\u06D5\u0633",
|
|
388
|
+
"\u0644\u06D5\u0646\u062F\u0649\u0631\u06CB",
|
|
389
|
+
"\u0644\u0627\u0646\u062F\u0649\u0631\u06CB",
|
|
390
|
+
"\u0644\u06D5\u0646\u06CB",
|
|
391
|
+
"\u0644\u0627\u0646\u06CB",
|
|
392
|
+
"\u0644\u06D5\u0646",
|
|
393
|
+
"\u0644\u0627\u0646",
|
|
394
|
+
"\u062F\u0649\u0631\u06CB",
|
|
395
|
+
"\u062F\u0649\u0631"
|
|
396
|
+
]);
|
|
397
|
+
var COMPOUND_PIVOT_ROOTS = [
|
|
398
|
+
"\u0633\u0648\u0632",
|
|
399
|
+
"\u062A\u0649\u0644",
|
|
400
|
+
"\u0628\u0649\u0644\u0649\u0645",
|
|
401
|
+
"\u062D\u0627\u0646\u0627",
|
|
402
|
+
"\u0643\u0648\u0632",
|
|
403
|
+
"\u0648\u0631\u0649\u0646",
|
|
404
|
+
"\u0643\u06D5\u0631",
|
|
405
|
+
"\u0642\u0648\u0631",
|
|
406
|
+
"\u067E\u0627\u0632",
|
|
407
|
+
"\u06AF\u06D5\u0631",
|
|
408
|
+
"\u0634\u0649\u0644\u0649\u0643",
|
|
409
|
+
"\u062A\u06C7\u0633\u062A\u0649\u0643",
|
|
410
|
+
"\u0633\u0627\u06CB",
|
|
411
|
+
"\u0633\u064A\u0627\u0642",
|
|
412
|
+
"\u0648\u064A\u0649\u0646",
|
|
413
|
+
"\u062A\u06C7\u0633\u062A\u0649\u06AF"
|
|
414
|
+
];
|
|
415
|
+
var IMPLICIT_SOFT_ROOTS = /* @__PURE__ */ new Set([
|
|
416
|
+
"\u06C7\u0645\u0649\u062A",
|
|
417
|
+
"\u062A\u06C7\u0628",
|
|
418
|
+
"\u06C7\u0634\u0649\u0646",
|
|
419
|
+
"\u0645\u06C7\u0645\u0643\u0649\u0646",
|
|
420
|
+
"\u0628\u0649\u0631",
|
|
421
|
+
"\u0628\u0649\u0632",
|
|
422
|
+
"\u0633\u0649\u0632",
|
|
423
|
+
"\u0643\u0649\u0645",
|
|
424
|
+
"\u062A\u0649\u0644",
|
|
425
|
+
"\u0643\u06C7\u0646",
|
|
426
|
+
"\u062A\u06C7\u0646",
|
|
427
|
+
"\u062C\u06C7\u0631",
|
|
428
|
+
"\u0649\u0644\u06AF\u06D5\u0631\u0649",
|
|
429
|
+
"\u0628\u06C7\u0644"
|
|
430
|
+
]);
|
|
431
|
+
var ARAB_CONSONANTS_FOR_CLUSTER = "\u0628\u06C6\u06AF\u0639\u062F\u062C\u0632\u0643\u0642\u0644\u0645\u0646\u06AD\u067E\u0631\u0633\u062A\u0641\u062D\u06BE\u0686\u0634";
|
|
432
|
+
var NATIVE_CLUSTERS = /* @__PURE__ */ new Set([
|
|
433
|
+
"\u0642\u062A",
|
|
434
|
+
"\u0642\u0633",
|
|
435
|
+
"\u0642\u0628",
|
|
436
|
+
"\u0642\u062C",
|
|
437
|
+
"\u0642\u062F",
|
|
438
|
+
"\u0642\u0632",
|
|
439
|
+
"\u0644\u062F",
|
|
440
|
+
"\u0644\u062A",
|
|
441
|
+
"\u0644\u0642",
|
|
442
|
+
"\u0644\u062D",
|
|
443
|
+
"\u0644\u0633",
|
|
444
|
+
"\u0644\u0628",
|
|
445
|
+
"\u0644\u062C",
|
|
446
|
+
"\u0646\u062F",
|
|
447
|
+
"\u0646\u062A",
|
|
448
|
+
"\u0646\u0642",
|
|
449
|
+
"\u0646\u0633",
|
|
450
|
+
"\u0646\u062C",
|
|
451
|
+
"\u0646\u0628",
|
|
452
|
+
"\u0646\u0632",
|
|
453
|
+
"\u0633\u062A",
|
|
454
|
+
"\u0633\u0642",
|
|
455
|
+
"\u0633\u062F",
|
|
456
|
+
"\u0633\u0628",
|
|
457
|
+
"\u0633\u062C",
|
|
458
|
+
"\u0633\u0632",
|
|
459
|
+
"\u0634\u062A",
|
|
460
|
+
"\u0634\u0642",
|
|
461
|
+
"\u0634\u062F",
|
|
462
|
+
"\u0634\u0633",
|
|
463
|
+
"\u0631\u062A",
|
|
464
|
+
"\u0631\u062F",
|
|
465
|
+
"\u0631\u0642",
|
|
466
|
+
"\u0631\u0633",
|
|
467
|
+
"\u0631\u062C",
|
|
468
|
+
"\u0631\u0628",
|
|
469
|
+
"\u0631\u0632",
|
|
470
|
+
"\u0632\u062F",
|
|
471
|
+
"\u0632\u0642",
|
|
472
|
+
"\u06AD\u062F",
|
|
473
|
+
"\u06AD\u0642",
|
|
474
|
+
"\u06AD\u062A",
|
|
475
|
+
"\u0645\u062F",
|
|
476
|
+
"\u0645\u0628",
|
|
477
|
+
"\u0645\u0646",
|
|
478
|
+
"\u064A\u062A",
|
|
479
|
+
"\u064A\u0633",
|
|
480
|
+
"\u064A\u0642",
|
|
481
|
+
"\u064A\u062F",
|
|
482
|
+
"\u0649\u0644",
|
|
483
|
+
"\u0649\u0646",
|
|
484
|
+
"\u0649\u0631",
|
|
485
|
+
"\u0649\u0645"
|
|
486
|
+
]);
|
|
487
|
+
function hasKey(map, key) {
|
|
488
|
+
return Object.prototype.hasOwnProperty.call(map, key);
|
|
489
|
+
}
|
|
490
|
+
var PrefixTrieNode = class {
|
|
491
|
+
constructor() {
|
|
492
|
+
this.children = /* @__PURE__ */ new Map();
|
|
493
|
+
this.isEnd = false;
|
|
494
|
+
}
|
|
495
|
+
};
|
|
496
|
+
var PrefixTrie = class {
|
|
497
|
+
constructor() {
|
|
498
|
+
this.root = new PrefixTrieNode();
|
|
499
|
+
}
|
|
500
|
+
insert(word) {
|
|
501
|
+
let node = this.root;
|
|
502
|
+
for (const char of word) {
|
|
503
|
+
let next = node.children.get(char);
|
|
504
|
+
if (!next) {
|
|
505
|
+
next = new PrefixTrieNode();
|
|
506
|
+
node.children.set(char, next);
|
|
507
|
+
}
|
|
508
|
+
node = next;
|
|
509
|
+
}
|
|
510
|
+
node.isEnd = true;
|
|
511
|
+
}
|
|
512
|
+
hasPrefixOf(word) {
|
|
513
|
+
let node = this.root;
|
|
514
|
+
for (const char of word) {
|
|
515
|
+
const next = node.children.get(char);
|
|
516
|
+
if (!next) {
|
|
517
|
+
return false;
|
|
518
|
+
}
|
|
519
|
+
node = next;
|
|
520
|
+
if (node.isEnd) {
|
|
521
|
+
return true;
|
|
522
|
+
}
|
|
523
|
+
}
|
|
524
|
+
return false;
|
|
525
|
+
}
|
|
526
|
+
};
|
|
527
|
+
var ArabicToCyrillicConverter = class {
|
|
528
|
+
constructor(options = {}) {
|
|
529
|
+
this.HAMZA = "\u0674";
|
|
530
|
+
this.loanwordPrefixTrie = new PrefixTrie();
|
|
531
|
+
this.reZwnjEtc = /[\u200B-\u200F\u202A-\u202E\uFEFF]/gu;
|
|
532
|
+
this.reSpaces = /[ \t]+/gu;
|
|
533
|
+
this.reHyphens = /\s*-\s*/gu;
|
|
534
|
+
this.reRedundantYye1 = /ىييە/gu;
|
|
535
|
+
this.reRedundantYye2 = /ييە/gu;
|
|
536
|
+
this.reRedundantYye3 = /يية/gu;
|
|
537
|
+
this.reUndantYa = /ىييا/gu;
|
|
538
|
+
this.reArabicWords = /[\u0600-\u06FF\uFB50-\uFDFF\uFE70-\uFEFF]+(?:[-\s]+[\u0600-\u06FF\uFB50-\uFDFF\uFE70-\uFEFF]+)*/gu;
|
|
539
|
+
this.reCapAfterPunct = /([.。:：?？!！])\s*([a-zа-яәіңғүұқөһ])/giu;
|
|
540
|
+
this.reCapAfterQuote = /([«"'"])\s*([a-zа-яәіңғүұқөһ])/giu;
|
|
541
|
+
this.frontVowelsCyr = /* @__PURE__ */ new Set(["\u04D9", "\u0435", "\u0456", "\u04E9", "\u04AF", "\u044D", "\u0438"]);
|
|
542
|
+
this.backVowelsCyr = /* @__PURE__ */ new Set(["\u0430", "\u043E", "\u04B1", "\u044B", "\u044F", "\u044E"]);
|
|
543
|
+
this.arabicVowels = /* @__PURE__ */ new Set(["\u0627", "\u0649", "\u0648", "\u06C7", "\u06D5", "\u06CB", "\u064A"]);
|
|
544
|
+
if (options.useLm && !options.disambiguator) {
|
|
545
|
+
throw new Error(
|
|
546
|
+
"Built-in LM disambiguation is not bundled with the npm package. Provide options.disambiguator in Node.js or use the pure rule-based converter."
|
|
547
|
+
);
|
|
548
|
+
}
|
|
549
|
+
this.disambiguator = options.disambiguator ?? new NoopDisambiguator();
|
|
550
|
+
for (const prefix of LOANWORD_PREFIXES) {
|
|
551
|
+
this.loanwordPrefixTrie.insert(prefix);
|
|
552
|
+
}
|
|
553
|
+
}
|
|
554
|
+
isLoanword(word) {
|
|
555
|
+
return LOANWORD_EXACT.has(word) || this.loanwordPrefixTrie.hasPrefixOf(word) || this.hasConsonantCluster(word);
|
|
556
|
+
}
|
|
557
|
+
hasConsonantCluster(word) {
|
|
558
|
+
let stem = word;
|
|
559
|
+
for (let length = word.length - 1; length > Math.max(2, word.length - 6); length -= 1) {
|
|
560
|
+
const candidateSuffix = word.slice(length);
|
|
561
|
+
if (candidateSuffix && this.isValidSuffixSequence(candidateSuffix)) {
|
|
562
|
+
stem = word.slice(0, length);
|
|
563
|
+
break;
|
|
564
|
+
}
|
|
565
|
+
}
|
|
566
|
+
let consonantCount = 0;
|
|
567
|
+
const chars = [...stem];
|
|
568
|
+
for (let index = 0; index < chars.length; index += 1) {
|
|
569
|
+
const char = chars[index];
|
|
570
|
+
if (ARAB_CONSONANTS_FOR_CLUSTER.includes(char)) {
|
|
571
|
+
if (consonantCount >= 1 && index >= 1) {
|
|
572
|
+
const pair = `${chars[index - 1]}${char}`;
|
|
573
|
+
if (NATIVE_CLUSTERS.has(pair)) {
|
|
574
|
+
consonantCount = 1;
|
|
575
|
+
continue;
|
|
576
|
+
}
|
|
577
|
+
}
|
|
578
|
+
consonantCount += 1;
|
|
579
|
+
if (consonantCount >= 3) {
|
|
580
|
+
return true;
|
|
581
|
+
}
|
|
582
|
+
} else {
|
|
583
|
+
consonantCount = 0;
|
|
584
|
+
}
|
|
585
|
+
}
|
|
586
|
+
return false;
|
|
587
|
+
}
|
|
588
|
+
isLoanwordWithEPrefix(word) {
|
|
589
|
+
return word.startsWith("\u06D5") && LOANWORD_E_PREFIXES.some((prefix) => word.startsWith(prefix));
|
|
590
|
+
}
|
|
591
|
+
getCyrillicVowelState(cyrillicWord) {
|
|
592
|
+
const cyr = cyrillicWord.toLowerCase();
|
|
593
|
+
if (cyr === "\u043A\u0456\u0442\u0430\u043F") {
|
|
594
|
+
return true;
|
|
595
|
+
}
|
|
596
|
+
for (let index = cyr.length - 1; index >= 0; index -= 1) {
|
|
597
|
+
const char = cyr[index];
|
|
598
|
+
if (this.frontVowelsCyr.has(char)) {
|
|
599
|
+
return true;
|
|
600
|
+
}
|
|
601
|
+
if (this.backVowelsCyr.has(char)) {
|
|
602
|
+
return false;
|
|
603
|
+
}
|
|
604
|
+
}
|
|
605
|
+
return false;
|
|
606
|
+
}
|
|
607
|
+
isValidSuffixSequence(suffix) {
|
|
608
|
+
if (!suffix) {
|
|
609
|
+
return true;
|
|
610
|
+
}
|
|
611
|
+
const dp = Array(suffix.length + 1).fill(false);
|
|
612
|
+
dp[0] = true;
|
|
613
|
+
for (let index = 1; index <= suffix.length; index += 1) {
|
|
614
|
+
for (let start = Math.max(0, index - 10); start < index; start += 1) {
|
|
615
|
+
if (dp[start] && VALID_SUFFIXES.has(suffix.slice(start, index))) {
|
|
616
|
+
dp[index] = true;
|
|
617
|
+
break;
|
|
618
|
+
}
|
|
619
|
+
}
|
|
620
|
+
}
|
|
621
|
+
return dp[suffix.length];
|
|
622
|
+
}
|
|
623
|
+
getHarmonyFromArabicRoot(word) {
|
|
624
|
+
for (const root of IMPLICIT_SOFT_ROOTS) {
|
|
625
|
+
if (word.startsWith(root) && ![...word].some((char) => char === "\u0642" || char === "\u0639")) {
|
|
626
|
+
return "soft";
|
|
627
|
+
}
|
|
628
|
+
}
|
|
629
|
+
let softSignals = 0;
|
|
630
|
+
let hardSignals = 0;
|
|
631
|
+
let hasHamza = false;
|
|
632
|
+
for (const char of word) {
|
|
633
|
+
if (char === this.HAMZA) {
|
|
634
|
+
hasHamza = true;
|
|
635
|
+
} else if (char === "\u0643" || char === "\u06AF") {
|
|
636
|
+
softSignals += 5;
|
|
637
|
+
} else if (char === "\u0642" || char === "\u0639") {
|
|
638
|
+
hardSignals += 5;
|
|
639
|
+
} else if (char === "\u06D5") {
|
|
640
|
+
softSignals += 3;
|
|
641
|
+
} else if ("\u0627\u0648\u06C7".includes(char)) {
|
|
642
|
+
hardSignals += 2;
|
|
643
|
+
}
|
|
644
|
+
}
|
|
645
|
+
if (hasHamza || softSignals > hardSignals) {
|
|
646
|
+
return "soft";
|
|
647
|
+
}
|
|
648
|
+
if (hardSignals > softSignals) {
|
|
649
|
+
return "hard";
|
|
650
|
+
}
|
|
651
|
+
return "hard";
|
|
652
|
+
}
|
|
653
|
+
segmentCompoundWord(word) {
|
|
654
|
+
if (word.includes("-")) {
|
|
655
|
+
return word.split("-");
|
|
656
|
+
}
|
|
657
|
+
if (word.startsWith(this.HAMZA)) {
|
|
658
|
+
return [word];
|
|
659
|
+
}
|
|
660
|
+
for (const pivot of COMPOUND_PIVOT_ROOTS) {
|
|
661
|
+
if (word.includes(pivot) && !word.startsWith(pivot)) {
|
|
662
|
+
const pivotIndex = word.indexOf(pivot);
|
|
663
|
+
if (pivotIndex > 0 && word[pivotIndex - 1] !== this.HAMZA) {
|
|
664
|
+
return [word.slice(0, pivotIndex), word.slice(pivotIndex)];
|
|
665
|
+
}
|
|
666
|
+
}
|
|
667
|
+
}
|
|
668
|
+
const suffixPatterns = [/(تاۋلىق(?:تار)?)$/u, /(زار)$/u, /(ستان)$/u];
|
|
669
|
+
for (const pattern of suffixPatterns) {
|
|
670
|
+
const match = word.match(pattern);
|
|
671
|
+
if (match && match.index !== void 0 && match.index > 0) {
|
|
672
|
+
return [word.slice(0, match.index), word.slice(match.index)];
|
|
673
|
+
}
|
|
674
|
+
}
|
|
675
|
+
return [word];
|
|
676
|
+
}
|
|
677
|
+
extractRootAndSuffix(word) {
|
|
678
|
+
if (!word) {
|
|
679
|
+
return { matchType: null, base: null, suffix: word };
|
|
680
|
+
}
|
|
681
|
+
for (let length = word.length; length > 1; length -= 1) {
|
|
682
|
+
const prefix = word.slice(0, length);
|
|
683
|
+
const suffix = word.slice(length);
|
|
684
|
+
if (!this.isValidSuffixSequence(suffix)) {
|
|
685
|
+
continue;
|
|
686
|
+
}
|
|
687
|
+
if (hasKey(EXCEPTIONS, prefix)) {
|
|
688
|
+
return { matchType: "exception", base: EXCEPTIONS[prefix], suffix };
|
|
689
|
+
}
|
|
690
|
+
if (hasKey(PROPER_NOUNS, prefix)) {
|
|
691
|
+
return { matchType: "proper", base: PROPER_NOUNS[prefix], suffix };
|
|
692
|
+
}
|
|
693
|
+
if (LOANWORD_EXACT.has(prefix)) {
|
|
694
|
+
return { matchType: "loanword", base: prefix, suffix };
|
|
695
|
+
}
|
|
696
|
+
}
|
|
697
|
+
for (let length = word.length - 1; length > 1; length -= 1) {
|
|
698
|
+
const prefix = word.slice(0, length);
|
|
699
|
+
const suffix = word.slice(length);
|
|
700
|
+
if (![...prefix].some((char) => this.arabicVowels.has(char))) {
|
|
701
|
+
continue;
|
|
702
|
+
}
|
|
703
|
+
if (this.isValidSuffixSequence(suffix)) {
|
|
704
|
+
return { matchType: "anonymous", base: prefix, suffix };
|
|
705
|
+
}
|
|
706
|
+
}
|
|
707
|
+
return { matchType: null, base: null, suffix: word };
|
|
708
|
+
}
|
|
709
|
+
convertSuffixOnly(suffix, isFront) {
|
|
710
|
+
if (!suffix) {
|
|
711
|
+
return "";
|
|
712
|
+
}
|
|
713
|
+
const result = [];
|
|
714
|
+
let index = 0;
|
|
715
|
+
while (index < suffix.length) {
|
|
716
|
+
const char = suffix[index];
|
|
717
|
+
const pair = suffix.slice(index, index + 2);
|
|
718
|
+
if (pair === "\u0649\u064A") {
|
|
719
|
+
if (suffix[index + 2] === "\u0627") {
|
|
720
|
+
result.push("\u0438\u044F");
|
|
721
|
+
index += 3;
|
|
722
|
+
} else {
|
|
723
|
+
result.push("\u0438");
|
|
724
|
+
index += 2;
|
|
725
|
+
}
|
|
726
|
+
continue;
|
|
727
|
+
}
|
|
728
|
+
if (char === "\u064A" && index + 1 < suffix.length) {
|
|
729
|
+
const nextChar = suffix[index + 1];
|
|
730
|
+
if (nextChar === "\u0627") {
|
|
731
|
+
result.push("\u044F");
|
|
732
|
+
index += 2;
|
|
733
|
+
continue;
|
|
734
|
+
}
|
|
735
|
+
if (nextChar === "\u06CB") {
|
|
736
|
+
result.push("\u044E");
|
|
737
|
+
index += 2;
|
|
738
|
+
continue;
|
|
739
|
+
}
|
|
740
|
+
}
|
|
741
|
+
if (pair === "\u0634\u0634") {
|
|
742
|
+
result.push("\u0449");
|
|
743
|
+
index += 2;
|
|
744
|
+
continue;
|
|
745
|
+
}
|
|
746
|
+
if (pair === "\u062A\u0633") {
|
|
747
|
+
result.push("\u0446");
|
|
748
|
+
index += 2;
|
|
749
|
+
continue;
|
|
750
|
+
}
|
|
751
|
+
if (char === this.HAMZA && index + 1 < suffix.length) {
|
|
752
|
+
const nextChar = suffix[index + 1];
|
|
753
|
+
if (nextChar === "\u0627") {
|
|
754
|
+
result.push("\u04D9");
|
|
755
|
+
index += 2;
|
|
756
|
+
continue;
|
|
757
|
+
}
|
|
758
|
+
if (nextChar === "\u0649") {
|
|
759
|
+
result.push("\u0456");
|
|
760
|
+
index += 2;
|
|
761
|
+
continue;
|
|
762
|
+
}
|
|
763
|
+
if (nextChar === "\u0648") {
|
|
764
|
+
result.push("\u04E9");
|
|
765
|
+
index += 2;
|
|
766
|
+
continue;
|
|
767
|
+
}
|
|
768
|
+
if (nextChar === "\u06C7") {
|
|
769
|
+
result.push("\u04AF");
|
|
770
|
+
index += 2;
|
|
771
|
+
continue;
|
|
772
|
+
}
|
|
773
|
+
index += 1;
|
|
774
|
+
continue;
|
|
775
|
+
}
|
|
776
|
+
if (hasKey(CONSONANTS, char)) {
|
|
777
|
+
result.push(CONSONANTS[char]);
|
|
778
|
+
index += 1;
|
|
779
|
+
continue;
|
|
780
|
+
}
|
|
781
|
+
if (Object.prototype.hasOwnProperty.call(VOWEL_MAP, char)) {
|
|
782
|
+
const vowel = VOWEL_MAP[char];
|
|
783
|
+
result.push(typeof vowel === "string" ? vowel : isFront ? vowel.f : vowel.b);
|
|
784
|
+
index += 1;
|
|
785
|
+
continue;
|
|
786
|
+
}
|
|
787
|
+
if (char === "\u064A") {
|
|
788
|
+
result.push(index === 0 ? "\u0439" : "\u0627\u0649\u0648\u06C7\u06D5\u06CB".includes(suffix[index - 1]) ? "\u0439" : "\u0438");
|
|
789
|
+
index += 1;
|
|
790
|
+
continue;
|
|
791
|
+
}
|
|
792
|
+
result.push(char);
|
|
793
|
+
index += 1;
|
|
794
|
+
}
|
|
795
|
+
return result.join("");
|
|
796
|
+
}
|
|
797
|
+
convertWord(word) {
|
|
798
|
+
if (!word) {
|
|
799
|
+
return word;
|
|
800
|
+
}
|
|
801
|
+
if (hasKey(EXCEPTIONS, word)) {
|
|
802
|
+
return EXCEPTIONS[word];
|
|
803
|
+
}
|
|
804
|
+
if (hasKey(PROPER_NOUNS, word)) {
|
|
805
|
+
return PROPER_NOUNS[word];
|
|
806
|
+
}
|
|
807
|
+
if (LOANWORD_EXACT.has(word)) {
|
|
808
|
+
return this.convertWordInternal(word);
|
|
809
|
+
}
|
|
810
|
+
const wholeWordIsFront = this.getHarmonyFromArabicRoot(word) === "soft";
|
|
811
|
+
const forcedState = wholeWordIsFront ? "soft" : "hard";
|
|
812
|
+
const { matchType, base, suffix } = this.extractRootAndSuffix(word);
|
|
813
|
+
if ((matchType === "exception" || matchType === "proper") && base) {
|
|
814
|
+
return `${base}${this.convertSuffixOnly(suffix, this.getCyrillicVowelState(base))}`;
|
|
815
|
+
}
|
|
816
|
+
if (matchType === "loanword" && base) {
|
|
817
|
+
const baseCyr = this.convertWordInternal(base);
|
|
818
|
+
return `${baseCyr}${this.convertSuffixOnly(suffix, this.getCyrillicVowelState(baseCyr))}`;
|
|
819
|
+
}
|
|
820
|
+
if (matchType === "anonymous" && base) {
|
|
821
|
+
if (this.isLoanword(base)) {
|
|
822
|
+
const baseCyr2 = this.convertWordInternal(base);
|
|
823
|
+
return `${baseCyr2}${this.convertSuffixOnly(suffix, this.getCyrillicVowelState(baseCyr2))}`;
|
|
824
|
+
}
|
|
825
|
+
const baseCyr = this.convertWordInternal(base, forcedState);
|
|
826
|
+
return `${baseCyr}${this.convertSuffixOnly(suffix, wholeWordIsFront)}`;
|
|
827
|
+
}
|
|
828
|
+
return this.convertWordInternal(word, forcedState);
|
|
829
|
+
}
|
|
830
|
+
convertWordInternal(word, forcedState) {
|
|
831
|
+
const segments = this.segmentCompoundWord(word);
|
|
832
|
+
if (segments.length > 1) {
|
|
833
|
+
const convertedSegments = segments.map((segment) => this.convertWord(segment));
|
|
834
|
+
return word.includes("-") ? convertedSegments.join("-") : convertedSegments.join("");
|
|
835
|
+
}
|
|
836
|
+
const isLoanwordE = this.isLoanwordWithEPrefix(word);
|
|
837
|
+
const isLoanword = this.isLoanword(word);
|
|
838
|
+
let currentState;
|
|
839
|
+
if (isLoanword) {
|
|
840
|
+
currentState = word.includes(this.HAMZA) ? "soft" : "hard";
|
|
841
|
+
} else if (forcedState) {
|
|
842
|
+
currentState = forcedState;
|
|
843
|
+
} else {
|
|
844
|
+
currentState = this.getHarmonyFromArabicRoot(word);
|
|
845
|
+
}
|
|
846
|
+
if (word === "\u062A\u0649\u064A\u0649\u0633") {
|
|
847
|
+
currentState = "soft";
|
|
848
|
+
}
|
|
849
|
+
const result = [];
|
|
850
|
+
let index = 0;
|
|
851
|
+
let isFirstChar = true;
|
|
852
|
+
while (index < word.length) {
|
|
853
|
+
const char = word[index];
|
|
854
|
+
if (isFirstChar && char === "\u06D5" && isLoanwordE) {
|
|
855
|
+
result.push("\u044D");
|
|
856
|
+
index += 1;
|
|
857
|
+
isFirstChar = false;
|
|
858
|
+
continue;
|
|
859
|
+
}
|
|
860
|
+
isFirstChar = false;
|
|
861
|
+
if (!isLoanword) {
|
|
862
|
+
if (char === "\u0642" || char === "\u0639") {
|
|
863
|
+
currentState = "hard";
|
|
864
|
+
} else if (char === "\u0643" || char === "\u06AF" || char === this.HAMZA) {
|
|
865
|
+
currentState = "soft";
|
|
866
|
+
}
|
|
867
|
+
} else if (char === this.HAMZA) {
|
|
868
|
+
currentState = "soft";
|
|
869
|
+
}
|
|
870
|
+
const twoChars = word.slice(index, index + 2);
|
|
871
|
+
const threeChars = word.slice(index, index + 3);
|
|
872
|
+
const fourChars = word.slice(index, index + 4);
|
|
873
|
+
if (twoChars === "\u0649\u064A" || twoChars === "\u064A\u064A") {
|
|
874
|
+
if (fourChars === "\u0649\u064A\u064A\u0627" || fourChars === "\u064A\u064A\u064A\u0627") {
|
|
875
|
+
result.push("\u0438\u044F");
|
|
876
|
+
index += 4;
|
|
877
|
+
continue;
|
|
878
|
+
}
|
|
879
|
+
if (threeChars === "\u0649\u064A\u0649\u0627" || threeChars === "\u064A\u064A\u0649\u0627" || word[index + 2] === "\u0627") {
|
|
880
|
+
result.push("\u0438\u044F");
|
|
881
|
+
index += 3;
|
|
882
|
+
continue;
|
|
883
|
+
}
|
|
884
|
+
result.push("\u0438");
|
|
885
|
+
index += 2;
|
|
886
|
+
continue;
|
|
887
|
+
}
|
|
888
|
+
if (char === "\u064A" && index + 1 < word.length) {
|
|
889
|
+
const nextChar = word[index + 1];
|
|
890
|
+
if (nextChar === "\u0627") {
|
|
891
|
+
if (isLoanword && index > 0) {
|
|
892
|
+
const prevChar = word[index - 1];
|
|
893
|
+
if (!`\u0627\u0649\u0648\u06C7\u06D5\u06CB${this.HAMZA}`.includes(prevChar)) {
|
|
894
|
+
result.push("\u0438\u044F");
|
|
895
|
+
index += 2;
|
|
896
|
+
continue;
|
|
897
|
+
}
|
|
898
|
+
}
|
|
899
|
+
result.push("\u044F");
|
|
900
|
+
index += 2;
|
|
901
|
+
continue;
|
|
902
|
+
}
|
|
903
|
+
if (nextChar === "\u06CB") {
|
|
904
|
+
result.push("\u044E");
|
|
905
|
+
index += 2;
|
|
906
|
+
continue;
|
|
907
|
+
}
|
|
908
|
+
if (nextChar === "\u0648" && index > 0 && "\u0627\u0649\u0648\u06C7\u06D5\u06CB".includes(word[index - 1])) {
|
|
909
|
+
result.push("\u0439\u043E");
|
|
910
|
+
index += 2;
|
|
911
|
+
continue;
|
|
912
|
+
}
|
|
913
|
+
}
|
|
914
|
+
if (twoChars === "\u0634\u0634") {
|
|
915
|
+
result.push("\u0449");
|
|
916
|
+
index += 2;
|
|
917
|
+
continue;
|
|
918
|
+
}
|
|
919
|
+
if (twoChars === "\u062A\u0633") {
|
|
920
|
+
if ((isLoanword || isLoanwordE) && index + 2 < word.length) {
|
|
921
|
+
result.push("\u0446");
|
|
922
|
+
index += 2;
|
|
923
|
+
continue;
|
|
924
|
+
}
|
|
925
|
+
}
|
|
926
|
+
if (isLoanword && word.slice(index, index + 6) === "\u067E\u0648\u062F\u06D5\u0632\u062F") {
|
|
927
|
+
result.push("\u043F\u043E\u0434\u044A\u0435\u0437\u0434");
|
|
928
|
+
index += 6;
|
|
929
|
+
continue;
|
|
930
|
+
}
|
|
931
|
+
if (char === this.HAMZA && index + 1 < word.length) {
|
|
932
|
+
const nextChar = word[index + 1];
|
|
933
|
+
if (nextChar === "\u0627") {
|
|
934
|
+
result.push("\u04D9");
|
|
935
|
+
index += 2;
|
|
936
|
+
continue;
|
|
937
|
+
}
|
|
938
|
+
if (nextChar === "\u0649") {
|
|
939
|
+
result.push("\u0456");
|
|
940
|
+
index += 2;
|
|
941
|
+
continue;
|
|
942
|
+
}
|
|
943
|
+
if (nextChar === "\u0648") {
|
|
944
|
+
result.push("\u04E9");
|
|
945
|
+
index += 2;
|
|
946
|
+
continue;
|
|
947
|
+
}
|
|
948
|
+
if (nextChar === "\u06C7") {
|
|
949
|
+
result.push("\u04AF");
|
|
950
|
+
index += 2;
|
|
951
|
+
continue;
|
|
952
|
+
}
|
|
953
|
+
index += 1;
|
|
954
|
+
continue;
|
|
955
|
+
}
|
|
956
|
+
if (hasKey(CONSONANTS, char)) {
|
|
957
|
+
result.push(CONSONANTS[char]);
|
|
958
|
+
index += 1;
|
|
959
|
+
continue;
|
|
960
|
+
}
|
|
961
|
+
if (Object.prototype.hasOwnProperty.call(VOWEL_MAP, char)) {
|
|
962
|
+
const vowel = VOWEL_MAP[char];
|
|
963
|
+
result.push(typeof vowel === "string" ? vowel : currentState === "soft" ? vowel.f : vowel.b);
|
|
964
|
+
index += 1;
|
|
965
|
+
continue;
|
|
966
|
+
}
|
|
967
|
+
if (char === "\u064A") {
|
|
968
|
+
if (index === 0) {
|
|
969
|
+
if (isLoanword) {
|
|
970
|
+
result.push("\u0438");
|
|
971
|
+
} else if (index + 1 < word.length && this.arabicVowels.has(word[index + 1])) {
|
|
972
|
+
result.push("\u0439");
|
|
973
|
+
} else {
|
|
974
|
+
result.push("\u0438");
|
|
975
|
+
}
|
|
976
|
+
} else {
|
|
977
|
+
const prevChar = word[index - 1];
|
|
978
|
+
result.push(["\u0627", "\u0649", "\u0648", "\u06C7", "\u06D5", "\u06CB"].includes(prevChar) ? "\u0439" : "\u0438");
|
|
979
|
+
}
|
|
980
|
+
index += 1;
|
|
981
|
+
continue;
|
|
982
|
+
}
|
|
983
|
+
if (isLoanword && char === "\u067E" && word.slice(index, index + 3) === "\u067E\u0648\u062F") {
|
|
984
|
+
result.push("\u043F\u043E\u0434");
|
|
985
|
+
index += 3;
|
|
986
|
+
if (index < word.length && word[index] === "\u06D5") {
|
|
987
|
+
result.push("\u044A\u0435");
|
|
988
|
+
index += 1;
|
|
989
|
+
}
|
|
990
|
+
continue;
|
|
991
|
+
}
|
|
992
|
+
result.push(char);
|
|
993
|
+
index += 1;
|
|
994
|
+
}
|
|
995
|
+
return result.join("");
|
|
996
|
+
}
|
|
997
|
+
preprocess(text) {
|
|
998
|
+
let next = text.replace(/ـ/gu, "-").replace(/\u0640/gu, "-");
|
|
999
|
+
next = next.replace(this.reZwnjEtc, "");
|
|
1000
|
+
next = next.replace(/ء/gu, this.HAMZA);
|
|
1001
|
+
next = next.replace(/أ/gu, `${this.HAMZA}\u0627`);
|
|
1002
|
+
next = next.replace(/ؤ/gu, `${this.HAMZA}\u0648`);
|
|
1003
|
+
next = next.replace(/ئ/gu, `${this.HAMZA}\u0649`);
|
|
1004
|
+
next = next.replace(/ٵ/gu, `${this.HAMZA}\u0627`);
|
|
1005
|
+
next = next.replace(/ٶ/gu, `${this.HAMZA}\u0648`);
|
|
1006
|
+
next = next.replace(/ٷ/gu, `${this.HAMZA}\u06C7`);
|
|
1007
|
+
next = next.replace(/ٸ/gu, `${this.HAMZA}\u0649`);
|
|
1008
|
+
next = next.replace(/\u06CC/gu, "\u0649");
|
|
1009
|
+
next = next.replace(/،/gu, ",").replace(/؛/gu, ";").replace(/؟/gu, "?").replace(/۔/gu, ".");
|
|
1010
|
+
next = next.replace(this.reSpaces, " ");
|
|
1011
|
+
next = next.replace(this.reHyphens, "-");
|
|
1012
|
+
next = next.replace(this.reRedundantYye1, "\u06D5");
|
|
1013
|
+
next = next.replace(this.reRedundantYye2, "\u06D5");
|
|
1014
|
+
next = next.replace(this.reRedundantYye3, "\u06D5");
|
|
1015
|
+
next = next.replace(this.reUndantYa, "\u064A\u0627");
|
|
1016
|
+
return next;
|
|
1017
|
+
}
|
|
1018
|
+
postProcessContextFix(rawTokens) {
|
|
1019
|
+
return rawTokens.map(([, cyr]) => cyr);
|
|
1020
|
+
}
|
|
1021
|
+
async postProcessContextFixAsync(rawTokens, contextSentence) {
|
|
1022
|
+
return this.disambiguator.disambiguate(rawTokens, contextSentence);
|
|
1023
|
+
}
|
|
1024
|
+
convertPhrase(phrase) {
|
|
1025
|
+
const words = phrase.split(" ");
|
|
1026
|
+
if (words.length <= 1) {
|
|
1027
|
+
return this.convertWord(phrase);
|
|
1028
|
+
}
|
|
1029
|
+
const rawTokens = words.map((word) => [word, hasKey(EXCEPTIONS, word) ? EXCEPTIONS[word] : this.convertWord(word)]);
|
|
1030
|
+
return this.postProcessContextFix(rawTokens).join(" ");
|
|
1031
|
+
}
|
|
1032
|
+
async convertPhraseAsync(phrase) {
|
|
1033
|
+
const words = phrase.split(" ");
|
|
1034
|
+
if (words.length <= 1) {
|
|
1035
|
+
return this.convertWord(phrase);
|
|
1036
|
+
}
|
|
1037
|
+
const rawTokens = words.map((word) => [word, hasKey(EXCEPTIONS, word) ? EXCEPTIONS[word] : this.convertWord(word)]);
|
|
1038
|
+
const fixed = await this.postProcessContextFixAsync(rawTokens, phrase);
|
|
1039
|
+
return fixed.join(" ");
|
|
1040
|
+
}
|
|
1041
|
+
convert(text) {
|
|
1042
|
+
const normalized = this.preprocess(text);
|
|
1043
|
+
const lines = normalized.split("\n");
|
|
1044
|
+
const convertedLines = [];
|
|
1045
|
+
for (const line of lines) {
|
|
1046
|
+
if (!line.trim()) {
|
|
1047
|
+
convertedLines.push("");
|
|
1048
|
+
continue;
|
|
1049
|
+
}
|
|
1050
|
+
let result = line.replace(this.reArabicWords, (phrase) => phrase.includes(" ") ? this.convertPhrase(phrase) : this.convertWord(phrase));
|
|
1051
|
+
if (result.length > 0) {
|
|
1052
|
+
result = result.replace(/[a-zа-яәіңғүұқөһ]/iu, (match) => match.toUpperCase());
|
|
1053
|
+
}
|
|
1054
|
+
result = result.replace(this.reCapAfterPunct, (_match, punctuation, char) => `${punctuation} ${char.toUpperCase()}`);
|
|
1055
|
+
result = result.replace(this.reCapAfterQuote, (_match, quote, char) => `${quote}${char.toUpperCase()}`);
|
|
1056
|
+
convertedLines.push(result);
|
|
1057
|
+
}
|
|
1058
|
+
return convertedLines.join("\n");
|
|
1059
|
+
}
|
|
1060
|
+
async convertAsync(text) {
|
|
1061
|
+
const normalized = this.preprocess(text);
|
|
1062
|
+
const lines = normalized.split("\n");
|
|
1063
|
+
const convertedLines = [];
|
|
1064
|
+
for (const line of lines) {
|
|
1065
|
+
if (!line.trim()) {
|
|
1066
|
+
convertedLines.push("");
|
|
1067
|
+
continue;
|
|
1068
|
+
}
|
|
1069
|
+
const matches = Array.from(line.matchAll(this.reArabicWords));
|
|
1070
|
+
let result = "";
|
|
1071
|
+
let lastIndex = 0;
|
|
1072
|
+
for (const match of matches) {
|
|
1073
|
+
const phrase = match[0];
|
|
1074
|
+
const matchIndex = match.index ?? 0;
|
|
1075
|
+
result += line.slice(lastIndex, matchIndex);
|
|
1076
|
+
result += phrase.includes(" ") ? await this.convertPhraseAsync(phrase) : this.convertWord(phrase);
|
|
1077
|
+
lastIndex = matchIndex + phrase.length;
|
|
1078
|
+
}
|
|
1079
|
+
result += line.slice(lastIndex);
|
|
1080
|
+
if (result.length > 0) {
|
|
1081
|
+
result = result.replace(/[a-zа-яәіңғүұқөһ]/iu, (char) => char.toUpperCase());
|
|
1082
|
+
}
|
|
1083
|
+
result = result.replace(this.reCapAfterPunct, (_match, punctuation, char) => `${punctuation} ${char.toUpperCase()}`);
|
|
1084
|
+
result = result.replace(this.reCapAfterQuote, (_match, quote, char) => `${quote}${char.toUpperCase()}`);
|
|
1085
|
+
convertedLines.push(result);
|
|
1086
|
+
}
|
|
1087
|
+
return convertedLines.join("\n");
|
|
1088
|
+
}
|
|
1089
|
+
};
|
|
1090
|
+
function arb2syr(text, options) {
|
|
1091
|
+
return new ArabicToCyrillicConverter(options).convert(text);
|
|
1092
|
+
}
|
|
1093
|
+
async function arb2syrAsync(text, options) {
|
|
1094
|
+
return new ArabicToCyrillicConverter(options).convertAsync(text);
|
|
1095
|
+
}
|
|
1096
|
+
|
|
1097
|
+
// src/lexicon.ts
|
|
1098
|
+
var DEFAULT_NATIVE_ROOTS = [
|
|
1099
|
+
"\u0431\u0430\u0441\u043F\u0430",
|
|
1100
|
+
"\u0441\u04E9\u0437",
|
|
1101
|
+
"\u04E9\u043D\u0435\u0440",
|
|
1102
|
+
"\u043A\u04D9\u0441\u0456\u043F",
|
|
1103
|
+
"\u0435\u043C",
|
|
1104
|
+
"\u0445\u0430\u043D\u0430",
|
|
1105
|
+
"\u0435\u04A3\u0431\u0435\u043A",
|
|
1106
|
+
"\u049B\u043E\u0440",
|
|
1107
|
+
"\u0441\u0443\u0440\u0435\u0442",
|
|
1108
|
+
"\u0448\u0456",
|
|
1109
|
+
"\u043E\u0442\u0430\u043D",
|
|
1110
|
+
"\u0430\u0434\u0430\u043C",
|
|
1111
|
+
"\u04E9\u0437\u0435\u043D",
|
|
1112
|
+
"\u04E9\u043C\u0456\u0440",
|
|
1113
|
+
"\u0436\u0430\u0443\u0430\u043F",
|
|
1114
|
+
"\u0436\u04B1\u043C\u044B\u0441"
|
|
1115
|
+
];
|
|
1116
|
+
var DEFAULT_LOAN_ROOTS = [
|
|
1117
|
+
"\u043C\u0430\u0448\u0438\u043D\u0430",
|
|
1118
|
+
"\u0430\u0442\u043E\u043C",
|
|
1119
|
+
"\u0443\u043D\u0438\u0432\u0435\u0440\u0441\u0438\u0442\u0435\u0442",
|
|
1120
|
+
"\u043A\u0430\u0440\u0442\u0430",
|
|
1121
|
+
"\u0431\u044E\u0434\u0436\u0435\u0442",
|
|
1122
|
+
"\u0442\u0435\u043B\u0435\u0444\u043E\u043D",
|
|
1123
|
+
"\u0438\u043D\u0442\u0435\u0440\u043D\u0435\u0442",
|
|
1124
|
+
"\u0447\u0435\u043C\u043F\u0438\u043E\u043D",
|
|
1125
|
+
"\u0446\u0435\u043C\u0435\u043D\u0442",
|
|
1126
|
+
"\u043A\u043E\u043C\u043F\u044C\u044E\u0442\u0435\u0440",
|
|
1127
|
+
"\u0442\u0435\u0445\u043D\u043E\u043B\u043E\u0433\u0438\u044F",
|
|
1128
|
+
"\u0440\u0435\u0441\u0443\u0440\u0441",
|
|
1129
|
+
"\u043A\u043E\u043D\u0441\u0442\u0438\u0442\u0443\u0446\u0438\u044F",
|
|
1130
|
+
"\u0434\u0435\u043C\u043E\u043A\u0440\u0430\u0442\u0438\u044F",
|
|
1131
|
+
"\u0441\u0442\u0430\u043D\u0446\u0438\u044F",
|
|
1132
|
+
"\u0434\u0438\u0430\u0433\u043D\u043E\u0437",
|
|
1133
|
+
"\u0446\u0438\u0440\u043A",
|
|
1134
|
+
"\u0449\u0435\u0442\u043A\u0430",
|
|
1135
|
+
"\u043F\u043E\u0434\u044A\u0435\u0437\u0434"
|
|
1136
|
+
];
|
|
1137
|
+
|
|
1138
|
+
// src/cyr2arb.ts
|
|
1139
|
+
var PROPER_NOUNS2 = {
|
|
1140
|
+
"\u0436\u0456\u04E9": "\u062C\u0649\u064A\u0649\u0648"
|
|
1141
|
+
};
|
|
1142
|
+
var COMMON_WORDS = {
|
|
1143
|
+
"\u0442\u0438\u0456\u0441": "\u062A\u0649\u064A\u0649\u0633",
|
|
1144
|
+
"\u0442\u0438\u0456\u0441\u0442\u0456": "\u062A\u0649\u064A\u0649\u0633\u062A\u0649",
|
|
1145
|
+
"\u0431\u0456\u0440\u0430\u049B": "\u0628\u0649\u0631\u0627\u0642",
|
|
1146
|
+
"\u049B\u043E\u044F\u043D": "\u0642\u0648\u0649\u064A\u0627\u0646",
|
|
1147
|
+
"\u04AF\u0448\u0456\u043D": "\u06C7\u0634\u0649\u0646"
|
|
1148
|
+
};
|
|
1149
|
+
var CONSONANTS2 = {
|
|
1150
|
+
"\u0431": "\u0628",
|
|
1151
|
+
"\u0432": "\u06C6",
|
|
1152
|
+
"\u0433": "\u06AF",
|
|
1153
|
+
"\u0493": "\u0639",
|
|
1154
|
+
"\u0434": "\u062F",
|
|
1155
|
+
"\u0436": "\u062C",
|
|
1156
|
+
"\u0437": "\u0632",
|
|
1157
|
+
"\u0439": "\u064A",
|
|
1158
|
+
"\u043A": "\u0643",
|
|
1159
|
+
"\u049B": "\u0642",
|
|
1160
|
+
"\u043B": "\u0644",
|
|
1161
|
+
"\u043C": "\u0645",
|
|
1162
|
+
"\u043D": "\u0646",
|
|
1163
|
+
"\u04A3": "\u06AD",
|
|
1164
|
+
"\u043F": "\u067E",
|
|
1165
|
+
"\u0440": "\u0631",
|
|
1166
|
+
"\u0441": "\u0633",
|
|
1167
|
+
"\u0442": "\u062A",
|
|
1168
|
+
"\u0444": "\u0641",
|
|
1169
|
+
"\u0445": "\u062D",
|
|
1170
|
+
"\u04BB": "\u06BE",
|
|
1171
|
+
"\u0447": "\u0686",
|
|
1172
|
+
"\u0448": "\u0634"
|
|
1173
|
+
};
|
|
1174
|
+
var VOWELS = {
|
|
1175
|
+
"\u0430": "\u0627",
|
|
1176
|
+
"\u04D9": "\u0627",
|
|
1177
|
+
"\u0435": "\u06D5",
|
|
1178
|
+
"\u043E": "\u0648",
|
|
1179
|
+
"\u04E9": "\u0648",
|
|
1180
|
+
"\u04B1": "\u06C7",
|
|
1181
|
+
"\u04AF": "\u06C7",
|
|
1182
|
+
"\u044B": "\u0649",
|
|
1183
|
+
"\u0456": "\u0649",
|
|
1184
|
+
"\u044D": "\u06D5"
|
|
1185
|
+
};
|
|
1186
|
+
var COMBINATIONS = {
|
|
1187
|
+
"\u0446": "\u062A\u0633",
|
|
1188
|
+
"\u0449": "\u0634\u0634",
|
|
1189
|
+
"\u0451": "\u064A\u0648"
|
|
1190
|
+
};
|
|
1191
|
+
var FRONT_VOWELS = /* @__PURE__ */ new Set(["\u04D9", "\u0435", "\u0456", "\u04E9", "\u04AF"]);
|
|
1192
|
+
var BACK_VOWELS = /* @__PURE__ */ new Set(["\u0430", "\u043E", "\u04B1", "\u044B", "\u0443"]);
|
|
1193
|
+
var I_INITIAL_NATIVE_WORDS = /* @__PURE__ */ new Set(["\u0438\u0456\u0441", "\u0438\u043D\u0435", "\u0438\u0442", "\u0438\u044E", "\u0438\u0456\u0440", "\u0438\u0456\u043B", "\u0438\u0440\u0456", "\u0438\u044B\u049B", "\u0438\u043D"]);
|
|
1194
|
+
var PUNCTUATION = {
|
|
1195
|
+
",": "\u060C",
|
|
1196
|
+
".": ".",
|
|
1197
|
+
":": ":",
|
|
1198
|
+
";": "\u061B",
|
|
1199
|
+
"?": "\u061F",
|
|
1200
|
+
"!": "!"
|
|
1201
|
+
};
|
|
1202
|
+
var TrieNode = class {
|
|
1203
|
+
constructor() {
|
|
1204
|
+
this.children = /* @__PURE__ */ new Map();
|
|
1205
|
+
this.isEndOfWord = false;
|
|
1206
|
+
this.isLoanword = false;
|
|
1207
|
+
this.harmony = null;
|
|
1208
|
+
}
|
|
1209
|
+
};
|
|
1210
|
+
var KazakhTrie = class {
|
|
1211
|
+
constructor() {
|
|
1212
|
+
this.root = new TrieNode();
|
|
1213
|
+
}
|
|
1214
|
+
determineHarmony(word) {
|
|
1215
|
+
const wordLower = word.toLowerCase();
|
|
1216
|
+
if ([...wordLower].some((char) => char === "\u043A" || char === "\u0433")) {
|
|
1217
|
+
return "front";
|
|
1218
|
+
}
|
|
1219
|
+
if ([...wordLower].some((char) => char === "\u049B" || char === "\u0493")) {
|
|
1220
|
+
return "back";
|
|
1221
|
+
}
|
|
1222
|
+
for (const char of wordLower) {
|
|
1223
|
+
if (FRONT_VOWELS.has(char)) {
|
|
1224
|
+
return "front";
|
|
1225
|
+
}
|
|
1226
|
+
if (BACK_VOWELS.has(char)) {
|
|
1227
|
+
return "back";
|
|
1228
|
+
}
|
|
1229
|
+
}
|
|
1230
|
+
return "back";
|
|
1231
|
+
}
|
|
1232
|
+
insert(word, isLoanword = false) {
|
|
1233
|
+
let node = this.root;
|
|
1234
|
+
const wordLower = word.toLowerCase();
|
|
1235
|
+
for (const char of wordLower) {
|
|
1236
|
+
let next = node.children.get(char);
|
|
1237
|
+
if (!next) {
|
|
1238
|
+
next = new TrieNode();
|
|
1239
|
+
node.children.set(char, next);
|
|
1240
|
+
}
|
|
1241
|
+
node = next;
|
|
1242
|
+
}
|
|
1243
|
+
node.isEndOfWord = true;
|
|
1244
|
+
node.isLoanword = isLoanword;
|
|
1245
|
+
node.harmony = this.determineHarmony(wordLower);
|
|
1246
|
+
}
|
|
1247
|
+
loadDictionary(nativeWords, loanWords) {
|
|
1248
|
+
for (const word of nativeWords) {
|
|
1249
|
+
this.insert(word, false);
|
|
1250
|
+
}
|
|
1251
|
+
for (const word of loanWords) {
|
|
1252
|
+
this.insert(word, true);
|
|
1253
|
+
}
|
|
1254
|
+
}
|
|
1255
|
+
};
|
|
1256
|
+
var CompoundSplitter = class {
|
|
1257
|
+
constructor(trie) {
|
|
1258
|
+
this.trie = trie;
|
|
1259
|
+
}
|
|
1260
|
+
splitWord(word) {
|
|
1261
|
+
const wordLower = word.toLowerCase();
|
|
1262
|
+
const parts = [];
|
|
1263
|
+
let index = 0;
|
|
1264
|
+
while (index < wordLower.length) {
|
|
1265
|
+
let node = this.trie.root;
|
|
1266
|
+
let matchLength = 0;
|
|
1267
|
+
let isLoan = false;
|
|
1268
|
+
for (let cursor = index; cursor < wordLower.length; cursor += 1) {
|
|
1269
|
+
const char = wordLower[cursor];
|
|
1270
|
+
const next = node.children.get(char);
|
|
1271
|
+
if (!next) {
|
|
1272
|
+
break;
|
|
1273
|
+
}
|
|
1274
|
+
node = next;
|
|
1275
|
+
if (node.isEndOfWord) {
|
|
1276
|
+
matchLength = cursor - index + 1;
|
|
1277
|
+
isLoan = node.isLoanword;
|
|
1278
|
+
}
|
|
1279
|
+
}
|
|
1280
|
+
if (matchLength > 0) {
|
|
1281
|
+
parts.push([wordLower.slice(index, index + matchLength), isLoan]);
|
|
1282
|
+
index += matchLength;
|
|
1283
|
+
continue;
|
|
1284
|
+
}
|
|
1285
|
+
const remaining = wordLower.slice(index);
|
|
1286
|
+
parts.push([remaining, this.fallbackIsLoanword(remaining)]);
|
|
1287
|
+
break;
|
|
1288
|
+
}
|
|
1289
|
+
return parts;
|
|
1290
|
+
}
|
|
1291
|
+
fallbackIsLoanword(word) {
|
|
1292
|
+
return [...word.toLowerCase()].some((char) => ["\u0444", "\u0432", "\u0446", "\u0447", "\u0449"].includes(char));
|
|
1293
|
+
}
|
|
1294
|
+
};
|
|
1295
|
+
function hasKey2(map, key) {
|
|
1296
|
+
return Object.prototype.hasOwnProperty.call(map, key);
|
|
1297
|
+
}
|
|
1298
|
+
var CyrillicToArabicConverter = class {
|
|
1299
|
+
constructor(options = {}) {
|
|
1300
|
+
this.HAMZA = "\u0674";
|
|
1301
|
+
this.trie = new KazakhTrie();
|
|
1302
|
+
this.splitter = new CompoundSplitter(this.trie);
|
|
1303
|
+
const nativeRoots = [...DEFAULT_NATIVE_ROOTS, ...options.lexicon?.nativeRoots ?? []];
|
|
1304
|
+
const loanRoots = [...DEFAULT_LOAN_ROOTS, ...options.lexicon?.loanRoots ?? []];
|
|
1305
|
+
this.trie.loadDictionary(nativeRoots, loanRoots);
|
|
1306
|
+
}
|
|
1307
|
+
getInitialHarmony(word) {
|
|
1308
|
+
const wordLower = word.toLowerCase();
|
|
1309
|
+
if ([...wordLower].some((char) => char === "\u049B" || char === "\u0493")) {
|
|
1310
|
+
return "back";
|
|
1311
|
+
}
|
|
1312
|
+
if ([...wordLower].some((char) => char === "\u043A" || char === "\u0433")) {
|
|
1313
|
+
return "front";
|
|
1314
|
+
}
|
|
1315
|
+
if (I_INITIAL_NATIVE_WORDS.has(wordLower)) {
|
|
1316
|
+
return "front";
|
|
1317
|
+
}
|
|
1318
|
+
for (const char of wordLower) {
|
|
1319
|
+
if (FRONT_VOWELS.has(char)) {
|
|
1320
|
+
return "front";
|
|
1321
|
+
}
|
|
1322
|
+
if (BACK_VOWELS.has(char)) {
|
|
1323
|
+
return "back";
|
|
1324
|
+
}
|
|
1325
|
+
}
|
|
1326
|
+
return "back";
|
|
1327
|
+
}
|
|
1328
|
+
applyHamzaRule(arabicResult, firstSegText, firstSegIsLoan, isSuffix = false) {
|
|
1329
|
+
if (arabicResult.includes(this.HAMZA) || !firstSegText || isSuffix) {
|
|
1330
|
+
return arabicResult;
|
|
1331
|
+
}
|
|
1332
|
+
const firstSegLower = firstSegText.toLowerCase();
|
|
1333
|
+
if (firstSegIsLoan || this.splitter.fallbackIsLoanword(firstSegLower)) {
|
|
1334
|
+
return arabicResult;
|
|
1335
|
+
}
|
|
1336
|
+
if ([...I_INITIAL_NATIVE_WORDS].some((word) => firstSegLower.startsWith(word))) {
|
|
1337
|
+
return arabicResult.startsWith(this.HAMZA) ? arabicResult : `${this.HAMZA}${arabicResult}`;
|
|
1338
|
+
}
|
|
1339
|
+
if ([...firstSegLower].some((char) => char === "\u043A" || char === "\u0433")) {
|
|
1340
|
+
return arabicResult;
|
|
1341
|
+
}
|
|
1342
|
+
const eHamzaWhitelist = /* @__PURE__ */ new Set(["\u04E9\u0437\u0435\u043D", "\u04E9\u0442\u0435", "\u04E9\u043D\u0435\u0440", "\u0438\u043D\u0435", "\u04D9\u043B\u0435\u043C"]);
|
|
1343
|
+
if (firstSegLower.includes("\u0435") && !eHamzaWhitelist.has(firstSegLower)) {
|
|
1344
|
+
return arabicResult;
|
|
1345
|
+
}
|
|
1346
|
+
if (this.getInitialHarmony(firstSegLower) === "front") {
|
|
1347
|
+
return arabicResult.startsWith(this.HAMZA) ? arabicResult : `${this.HAMZA}${arabicResult}`;
|
|
1348
|
+
}
|
|
1349
|
+
return arabicResult;
|
|
1350
|
+
}
|
|
1351
|
+
convertWord(word, isSuffix = false) {
|
|
1352
|
+
if (!word) {
|
|
1353
|
+
return word;
|
|
1354
|
+
}
|
|
1355
|
+
const wordLower = word.toLowerCase();
|
|
1356
|
+
if (hasKey2(PROPER_NOUNS2, wordLower)) {
|
|
1357
|
+
return PROPER_NOUNS2[wordLower];
|
|
1358
|
+
}
|
|
1359
|
+
if (hasKey2(COMMON_WORDS, wordLower)) {
|
|
1360
|
+
return COMMON_WORDS[wordLower];
|
|
1361
|
+
}
|
|
1362
|
+
const segments = this.splitter.splitWord(wordLower);
|
|
1363
|
+
const isLoanFlags = [];
|
|
1364
|
+
const isFrontFlags = [];
|
|
1365
|
+
const isHardLoanFlags = [];
|
|
1366
|
+
for (const [segText, isLoan] of segments) {
|
|
1367
|
+
let isHardLoan = false;
|
|
1368
|
+
let segFront = false;
|
|
1369
|
+
if (isLoan) {
|
|
1370
|
+
isHardLoan = [...segText].some((char) => "\u0430\u043E\u04B1\u044B".includes(char)) || ![...segText].some((char) => "\u04D9\u0435\u0456\u04E9\u04AF".includes(char));
|
|
1371
|
+
segFront = !isHardLoan;
|
|
1372
|
+
} else {
|
|
1373
|
+
segFront = this.getInitialHarmony(segText) === "front";
|
|
1374
|
+
}
|
|
1375
|
+
isLoanFlags.push(...Array(segText.length).fill(isLoan));
|
|
1376
|
+
isFrontFlags.push(...Array(segText.length).fill(segFront));
|
|
1377
|
+
isHardLoanFlags.push(...Array(segText.length).fill(isHardLoan));
|
|
1378
|
+
}
|
|
1379
|
+
const result = [];
|
|
1380
|
+
for (let index = 0; index < wordLower.length; index += 1) {
|
|
1381
|
+
const char = wordLower[index];
|
|
1382
|
+
const prevChar = index > 0 ? wordLower[index - 1] : "";
|
|
1383
|
+
const isLoanword = isLoanFlags[index];
|
|
1384
|
+
const isHardLoan = isHardLoanFlags[index];
|
|
1385
|
+
if (char === "\u044C" || char === "\u044A") {
|
|
1386
|
+
continue;
|
|
1387
|
+
}
|
|
1388
|
+
if (char === "\u0443") {
|
|
1389
|
+
result.push("\u06CB");
|
|
1390
|
+
continue;
|
|
1391
|
+
}
|
|
1392
|
+
if (char === "\u0438") {
|
|
1393
|
+
if (wordLower.includes("\u043C\u0430\u0448\u0438\u043D\u0430")) {
|
|
1394
|
+
result.push("\u064A");
|
|
1395
|
+
} else if (wordLower.includes("\u043A\u043E\u043D\u0441\u0442\u0438\u0442\u0443\u0446\u0438\u044F") && index < wordLower.indexOf("\u0446")) {
|
|
1396
|
+
result.push("\u064A");
|
|
1397
|
+
} else if (isLoanword) {
|
|
1398
|
+
result.push(isHardLoan ? "\u0649\u064A" : "\u064A");
|
|
1399
|
+
} else if (index === 0) {
|
|
1400
|
+
result.push("\u064A");
|
|
1401
|
+
} else {
|
|
1402
|
+
result.push("\u0649\u064A");
|
|
1403
|
+
}
|
|
1404
|
+
continue;
|
|
1405
|
+
}
|
|
1406
|
+
if (char === "\u044F") {
|
|
1407
|
+
if (prevChar === "\u0438") {
|
|
1408
|
+
if (isLoanword) {
|
|
1409
|
+
const prevPrev = index >= 2 ? wordLower[index - 2] : "";
|
|
1410
|
+
result.push(prevPrev === "\u0433" || prevPrev === "\u043A" ? "\u0627" : "\u064A\u0627");
|
|
1411
|
+
} else {
|
|
1412
|
+
result.push("\u0627");
|
|
1413
|
+
}
|
|
1414
|
+
continue;
|
|
1415
|
+
}
|
|
1416
|
+
result.push("\u064A\u0627");
|
|
1417
|
+
continue;
|
|
1418
|
+
}
|
|
1419
|
+
if (char === "\u044E") {
|
|
1420
|
+
if (prevChar === "\u044C" || prevChar === "\u044A") {
|
|
1421
|
+
result.push("\u064A\u06CB");
|
|
1422
|
+
} else {
|
|
1423
|
+
result.push(isLoanword || prevChar === "\u0438" ? "\u06CB" : "\u064A\u06CB");
|
|
1424
|
+
}
|
|
1425
|
+
continue;
|
|
1426
|
+
}
|
|
1427
|
+
if (hasKey2(COMBINATIONS, char)) {
|
|
1428
|
+
result.push(COMBINATIONS[char]);
|
|
1429
|
+
} else if (hasKey2(CONSONANTS2, char)) {
|
|
1430
|
+
result.push(CONSONANTS2[char]);
|
|
1431
|
+
} else if (hasKey2(VOWELS, char)) {
|
|
1432
|
+
result.push(VOWELS[char]);
|
|
1433
|
+
} else {
|
|
1434
|
+
result.push(char);
|
|
1435
|
+
}
|
|
1436
|
+
}
|
|
1437
|
+
const converted = result.join("");
|
|
1438
|
+
return this.applyHamzaRule(converted, segments[0]?.[0] ?? "", segments[0]?.[1] ?? false, isSuffix);
|
|
1439
|
+
}
|
|
1440
|
+
convertCompoundWord(word) {
|
|
1441
|
+
if (!word) {
|
|
1442
|
+
return word;
|
|
1443
|
+
}
|
|
1444
|
+
const wordLower = word.toLowerCase();
|
|
1445
|
+
if (hasKey2(PROPER_NOUNS2, wordLower)) {
|
|
1446
|
+
return PROPER_NOUNS2[wordLower];
|
|
1447
|
+
}
|
|
1448
|
+
if (hasKey2(COMMON_WORDS, wordLower)) {
|
|
1449
|
+
return COMMON_WORDS[wordLower];
|
|
1450
|
+
}
|
|
1451
|
+
if (!word.includes("-")) {
|
|
1452
|
+
return this.convertWord(wordLower);
|
|
1453
|
+
}
|
|
1454
|
+
const parts = wordLower.split("-");
|
|
1455
|
+
const converted = [this.convertWord(parts[0])];
|
|
1456
|
+
for (const part of parts.slice(1)) {
|
|
1457
|
+
converted.push(this.convertWord(part, true));
|
|
1458
|
+
}
|
|
1459
|
+
return converted.join("-");
|
|
1460
|
+
}
|
|
1461
|
+
convert(text) {
|
|
1462
|
+
let convertedText = text;
|
|
1463
|
+
for (const [cyr, arab] of Object.entries(PUNCTUATION)) {
|
|
1464
|
+
convertedText = convertedText.split(cyr).join(arab);
|
|
1465
|
+
}
|
|
1466
|
+
const pattern = /[а-яәіңғүұқөһёэъь]+(?:-[а-яәіңғүұқөһёэъь]+)*/giu;
|
|
1467
|
+
return convertedText.replace(pattern, (match) => this.convertCompoundWord(match));
|
|
1468
|
+
}
|
|
1469
|
+
};
|
|
1470
|
+
function syr2arb(text, options) {
|
|
1471
|
+
return new CyrillicToArabicConverter(options).convert(text);
|
|
1472
|
+
}
|
|
1473
|
+
export {
|
|
1474
|
+
ArabicToCyrillicConverter,
|
|
1475
|
+
CyrillicToArabicConverter,
|
|
1476
|
+
NoopDisambiguator,
|
|
1477
|
+
arb2syr,
|
|
1478
|
+
arb2syrAsync,
|
|
1479
|
+
syr2arb
|
|
1480
|
+
};
|