opencc-wasm 0.9.0 → 0.10.0

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 (62) hide show
  1. package/README.md +10 -1
  2. package/README.zh.md +10 -1
  3. package/dist/cjs/index.cjs +6 -0
  4. package/dist/cjs/opencc-wasm.cjs +1 -1
  5. package/dist/cjs/opencc-wasm.wasm +0 -0
  6. package/dist/data/config/hk2s.json +5 -0
  7. package/dist/data/config/hk2sp.json +5 -0
  8. package/dist/data/config/hk2sp_jieba.json +30 -5
  9. package/dist/data/config/hk2t.json +4 -4
  10. package/dist/data/config/jp2t.json +4 -4
  11. package/dist/data/config/opencc_config.schema.json +9 -4
  12. package/dist/data/config/s2hk.json +21 -2
  13. package/dist/data/config/s2hk_jieba.json +36 -4
  14. package/dist/data/config/s2hkp.json +21 -2
  15. package/dist/data/config/s2hkp_jieba.json +40 -5
  16. package/dist/data/config/s2t.json +12 -5
  17. package/dist/data/config/s2t_cngov.json +16 -11
  18. package/dist/data/config/s2t_jieba.json +27 -2
  19. package/dist/data/config/s2tw.json +21 -2
  20. package/dist/data/config/s2tw_jieba.json +36 -4
  21. package/dist/data/config/s2twp.json +21 -2
  22. package/dist/data/config/s2twp_jieba.json +40 -5
  23. package/dist/data/config/t2cngov.json +16 -11
  24. package/dist/data/config/t2cngov_keep_simp.json +16 -11
  25. package/dist/data/config/t2hk.json +4 -10
  26. package/dist/data/config/t2jp.json +3 -4
  27. package/dist/data/config/t2s.json +4 -4
  28. package/dist/data/config/t2s_cngov.json +16 -11
  29. package/dist/data/config/t2tw.json +4 -10
  30. package/dist/data/config/tw2s.json +5 -0
  31. package/dist/data/config/tw2sp.json +5 -0
  32. package/dist/data/config/tw2sp_jieba.json +30 -5
  33. package/dist/data/config/tw2t.json +4 -4
  34. package/dist/data/dict/CJK_Compatibility_Ideographs.ocd2 +0 -0
  35. package/dist/data/dict/HKPhrases.ocd2 +0 -0
  36. package/dist/data/dict/HKPhrasesRev.ocd2 +0 -0
  37. package/dist/data/dict/HKVariants.ocd2 +0 -0
  38. package/dist/data/dict/HKVariantsRev.ocd2 +0 -0
  39. package/dist/data/dict/HKVariantsRevPhrases.ocd2 +0 -0
  40. package/dist/data/dict/JPShinjitaiPhrases.ocd2 +0 -0
  41. package/dist/data/dict/STPhrases.ocd2 +0 -0
  42. package/dist/data/dict/STPhrases_GeneratedFromRegionalPhrases.ocd2 +0 -0
  43. package/dist/data/dict/TSCharacters.ocd2 +0 -0
  44. package/dist/data/dict/TSPhrases.ocd2 +0 -0
  45. package/dist/data/dict/TWPhrases.ocd2 +0 -0
  46. package/dist/data/dict/TWPhrasesRev.ocd2 +0 -0
  47. package/dist/data/dict/cngov/STCharacters.ocd2 +0 -0
  48. package/dist/data/dict/cngov/STPhrases.ocd2 +0 -0
  49. package/dist/data/dict/cngov/TGCharacters.ocd2 +0 -0
  50. package/dist/data/dict/cngov/TGCharacters_keep_simp.ocd2 +0 -0
  51. package/dist/data/dict/cngov/TGPhrases.ocd2 +0 -0
  52. package/dist/data/dict/cngov/TSCharacters.ocd2 +0 -0
  53. package/dist/data/dict/cngov/TSPhrases.ocd2 +0 -0
  54. package/dist/data/jieba_dict/jieba.dict.utf8 +1 -1
  55. package/dist/data/jieba_dict/jieba_merged.ocd2 +0 -0
  56. package/dist/data/jieba_dict/user.dict.utf8 +0 -1
  57. package/dist/esm/index.js +6 -0
  58. package/dist/esm/opencc-wasm.js +1 -1
  59. package/dist/esm/opencc-wasm.wasm +0 -0
  60. package/dist/opencc-wasm.wasm +0 -0
  61. package/index.d.ts +1 -0
  62. package/package.json +1 -1
package/dist/data/config/opencc_config.schema.json CHANGED
@@ -3,10 +3,15 @@
3
3
  "id": "https://opencc.byvoid.com/schema/opencc_config.schema.json",
4
4
  "title": "OpenCC configuration",
5
5
  "type": "object",
6
- "required": ["segmentation", "conversion_chain"],
6
+ "required": ["name", "conversion_chain"],
7
7
  "additionalProperties": false,
8
8
  "properties": {
9
9
  "name": { "type": "string" },
10
+ "normalization": {
11
+ "type": "array",
12
+ "minItems": 1,
13
+ "items": { "$ref": "#/definitions/conversion" }
14
+ },
10
15
  "segmentation": { "$ref": "#/definitions/segmentation" },
11
16
  "conversion_chain": {
12
17
  "type": "array",
@@ -77,17 +82,17 @@
77
82
  "type": { "enum": ["inline"] },
78
83
  "entries": {
79
84
  "type": "object",
80
- "patternProperties": { ".+": { "type": "string", "minLength": 1 } },
81
- "additionalProperties": false
85
+ "additionalProperties": { "type": "string", "minLength": 1 }
82
86
  }
83
87
  }
84
88
  },
85
89
  "group_dict": {
86
90
  "type": "object",
87
- "required": ["type", "dicts"],
91
+ "required": ["type", "dicts", "match_policy"],
88
92
  "additionalProperties": false,
89
93
  "properties": {
90
94
  "type": { "enum": ["group"] },
95
+ "match_policy": { "enum": ["short_circuit", "union"] },
91
96
  "dicts": {
92
97
  "type": "array",
93
98
  "minItems": 1,
package/dist/data/config/s2hk.json CHANGED
@@ -1,15 +1,33 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (Hong Kong variant)",
3
+ "normalization": [
4
+ { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
5
+ ],
3
6
  "segmentation": {
4
7
  "type": "mmseg",
5
- "dict": { "type": "ocd2", "file": "STPhrases.ocd2" }
8
+ "dict": {
9
+ "type": "group",
10
+ "match_policy": "union",
11
+ "dicts": [
12
+ { "type": "ocd2", "file": "STPhrases.ocd2" },
13
+ { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
14
+ ]
15
+ }
6
16
  },
7
17
  "conversion_chain": [
8
18
  {
9
19
  "dict": {
10
20
  "type": "group",
21
+ "match_policy": "short_circuit",
11
22
  "dicts": [
12
- { "type": "ocd2", "file": "STPhrases.ocd2" },
23
+ {
24
+ "type": "group",
25
+ "match_policy": "union",
26
+ "dicts": [
27
+ { "type": "ocd2", "file": "STPhrases.ocd2" },
28
+ { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
29
+ ]
30
+ },
13
31
  { "type": "ocd2", "file": "STCharacters.ocd2" }
14
32
  ]
15
33
  }
@@ -17,6 +35,7 @@
17
35
  {
18
36
  "dict": {
19
37
  "type": "group",
38
+ "match_policy": "short_circuit",
20
39
  "dicts": [
21
40
  { "type": "ocd2", "file": "HKVariantsPhrases.ocd2" },
22
41
  { "type": "ocd2", "file": "HKVariants.ocd2" }
package/dist/data/config/s2hk_jieba.json CHANGED
@@ -1,5 +1,13 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (Hong Kong variant, Jieba Segmentation - Experimental)",
3
+ "normalization": [
4
+ {
5
+ "dict": {
6
+ "type": "ocd2",
7
+ "file": "CJK_Compatibility_Ideographs.ocd2"
8
+ }
9
+ }
10
+ ],
3
11
  "segmentation": {
4
12
  "type": "jieba",
5
13
  "resources": {
@@ -11,18 +19,42 @@
11
19
  {
12
20
  "dict": {
13
21
  "type": "group",
22
+ "match_policy": "short_circuit",
14
23
  "dicts": [
15
- { "type": "ocd2", "file": "STPhrases.ocd2" },
16
- { "type": "ocd2", "file": "STCharacters.ocd2" }
24
+ {
25
+ "type": "group",
26
+ "match_policy": "union",
27
+ "dicts": [
28
+ {
29
+ "type": "ocd2",
30
+ "file": "STPhrases.ocd2"
31
+ },
32
+ {
33
+ "type": "ocd2",
34
+ "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2"
35
+ }
36
+ ]
37
+ },
38
+ {
39
+ "type": "ocd2",
40
+ "file": "STCharacters.ocd2"
41
+ }
17
42
  ]
18
43
  }
19
44
  },
20
45
  {
21
46
  "dict": {
22
47
  "type": "group",
48
+ "match_policy": "short_circuit",
23
49
  "dicts": [
24
- { "type": "ocd2", "file": "HKVariantsPhrases.ocd2" },
25
- { "type": "ocd2", "file": "HKVariants.ocd2" }
50
+ {
51
+ "type": "ocd2",
52
+ "file": "HKVariantsPhrases.ocd2"
53
+ },
54
+ {
55
+ "type": "ocd2",
56
+ "file": "HKVariants.ocd2"
57
+ }
26
58
  ]
27
59
  }
28
60
  }
package/dist/data/config/s2hkp.json CHANGED
@@ -1,15 +1,33 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (Hong Kong variant, with phrases)",
3
+ "normalization": [
4
+ { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
5
+ ],
3
6
  "segmentation": {
4
7
  "type": "mmseg",
5
- "dict": { "type": "ocd2", "file": "STPhrases.ocd2" }
8
+ "dict": {
9
+ "type": "group",
10
+ "match_policy": "union",
11
+ "dicts": [
12
+ { "type": "ocd2", "file": "STPhrases.ocd2" },
13
+ { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
14
+ ]
15
+ }
6
16
  },
7
17
  "conversion_chain": [
8
18
  {
9
19
  "dict": {
10
20
  "type": "group",
21
+ "match_policy": "short_circuit",
11
22
  "dicts": [
12
- { "type": "ocd2", "file": "STPhrases.ocd2" },
23
+ {
24
+ "type": "group",
25
+ "match_policy": "union",
26
+ "dicts": [
27
+ { "type": "ocd2", "file": "STPhrases.ocd2" },
28
+ { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
29
+ ]
30
+ },
13
31
  { "type": "ocd2", "file": "STCharacters.ocd2" }
14
32
  ]
15
33
  }
@@ -17,6 +35,7 @@
17
35
  {
18
36
  "dict": {
19
37
  "type": "group",
38
+ "match_policy": "short_circuit",
20
39
  "dicts": [
21
40
  { "type": "ocd2", "file": "HKPhrases.ocd2" },
22
41
  { "type": "ocd2", "file": "HKVariantsPhrases.ocd2" },
package/dist/data/config/s2hkp_jieba.json CHANGED
@@ -1,5 +1,13 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (Hong Kong variant, with phrases, Jieba Segmentation - Experimental)",
3
+ "normalization": [
4
+ {
5
+ "dict": {
6
+ "type": "ocd2",
7
+ "file": "CJK_Compatibility_Ideographs.ocd2"
8
+ }
9
+ }
10
+ ],
3
11
  "segmentation": {
4
12
  "type": "jieba",
5
13
  "resources": {
@@ -11,19 +19,46 @@
11
19
  {
12
20
  "dict": {
13
21
  "type": "group",
22
+ "match_policy": "short_circuit",
14
23
  "dicts": [
15
- { "type": "ocd2", "file": "STPhrases.ocd2" },
16
- { "type": "ocd2", "file": "STCharacters.ocd2" }
24
+ {
25
+ "type": "group",
26
+ "match_policy": "union",
27
+ "dicts": [
28
+ {
29
+ "type": "ocd2",
30
+ "file": "STPhrases.ocd2"
31
+ },
32
+ {
33
+ "type": "ocd2",
34
+ "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2"
35
+ }
36
+ ]
37
+ },
38
+ {
39
+ "type": "ocd2",
40
+ "file": "STCharacters.ocd2"
41
+ }
17
42
  ]
18
43
  }
19
44
  },
20
45
  {
21
46
  "dict": {
22
47
  "type": "group",
48
+ "match_policy": "short_circuit",
23
49
  "dicts": [
24
- { "type": "ocd2", "file": "HKPhrases.ocd2" },
25
- { "type": "ocd2", "file": "HKVariantsPhrases.ocd2" },
26
- { "type": "ocd2", "file": "HKVariants.ocd2" }
50
+ {
51
+ "type": "ocd2",
52
+ "file": "HKPhrases.ocd2"
53
+ },
54
+ {
55
+ "type": "ocd2",
56
+ "file": "HKVariantsPhrases.ocd2"
57
+ },
58
+ {
59
+ "type": "ocd2",
60
+ "file": "HKVariants.ocd2"
61
+ }
27
62
  ]
28
63
  }
29
64
  }
package/dist/data/config/s2t.json CHANGED
@@ -1,15 +1,22 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (OpenCC Standard)",
3
- "segmentation": {
4
- "type": "mmseg",
5
- "dict": { "type": "ocd2", "file": "STPhrases.ocd2" }
6
- },
3
+ "normalization": [
4
+ { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
5
+ ],
7
6
  "conversion_chain": [
8
7
  {
9
8
  "dict": {
10
9
  "type": "group",
10
+ "match_policy": "short_circuit",
11
11
  "dicts": [
12
- { "type": "ocd2", "file": "STPhrases.ocd2" },
12
+ {
13
+ "type": "group",
14
+ "match_policy": "union",
15
+ "dicts": [
16
+ { "type": "ocd2", "file": "STPhrases.ocd2" },
17
+ { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
18
+ ]
19
+ },
13
20
  { "type": "ocd2", "file": "STCharacters.ocd2" }
14
21
  ]
15
22
  }
package/dist/data/config/s2t_cngov.json CHANGED
@@ -7,16 +7,21 @@
7
7
  "file": "cngov/STPhrases.ocd2"
8
8
  }
9
9
  },
10
- "conversion_chain": [{
11
- "dict": {
12
- "type": "group",
13
- "dicts": [{
14
- "type": "ocd2",
15
- "file": "cngov/STPhrases.ocd2"
16
- }, {
17
- "type": "ocd2",
18
- "file": "cngov/STCharacters.ocd2"
19
- }]
10
+ "conversion_chain": [
11
+ {
12
+ "dict": {
13
+ "type": "group",
14
+ "dicts": [
15
+ {
16
+ "type": "ocd2",
17
+ "file": "cngov/STPhrases.ocd2"
18
+ },
19
+ {
20
+ "type": "ocd2",
21
+ "file": "cngov/STCharacters.ocd2"
22
+ }
23
+ ]
24
+ }
20
25
  }
21
- }]
26
+ ]
22
27
  }
package/dist/data/config/s2t_jieba.json CHANGED
@@ -1,5 +1,13 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (Jieba Segmentation - Experimental)",
3
+ "normalization": [
4
+ {
5
+ "dict": {
6
+ "type": "ocd2",
7
+ "file": "CJK_Compatibility_Ideographs.ocd2"
8
+ }
9
+ }
10
+ ],
3
11
  "segmentation": {
4
12
  "type": "jieba",
5
13
  "resources": {
@@ -11,9 +19,26 @@
11
19
  {
12
20
  "dict": {
13
21
  "type": "group",
22
+ "match_policy": "short_circuit",
14
23
  "dicts": [
15
- { "type": "ocd2", "file": "STPhrases.ocd2" },
16
- { "type": "ocd2", "file": "STCharacters.ocd2" }
24
+ {
25
+ "type": "group",
26
+ "match_policy": "union",
27
+ "dicts": [
28
+ {
29
+ "type": "ocd2",
30
+ "file": "STPhrases.ocd2"
31
+ },
32
+ {
33
+ "type": "ocd2",
34
+ "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2"
35
+ }
36
+ ]
37
+ },
38
+ {
39
+ "type": "ocd2",
40
+ "file": "STCharacters.ocd2"
41
+ }
17
42
  ]
18
43
  }
19
44
  }
package/dist/data/config/s2tw.json CHANGED
@@ -1,15 +1,33 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (Taiwan Standard)",
3
+ "normalization": [
4
+ { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
5
+ ],
3
6
  "segmentation": {
4
7
  "type": "mmseg",
5
- "dict": { "type": "ocd2", "file": "STPhrases.ocd2" }
8
+ "dict": {
9
+ "type": "group",
10
+ "match_policy": "union",
11
+ "dicts": [
12
+ { "type": "ocd2", "file": "STPhrases.ocd2" },
13
+ { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
14
+ ]
15
+ }
6
16
  },
7
17
  "conversion_chain": [
8
18
  {
9
19
  "dict": {
10
20
  "type": "group",
21
+ "match_policy": "short_circuit",
11
22
  "dicts": [
12
- { "type": "ocd2", "file": "STPhrases.ocd2" },
23
+ {
24
+ "type": "group",
25
+ "match_policy": "union",
26
+ "dicts": [
27
+ { "type": "ocd2", "file": "STPhrases.ocd2" },
28
+ { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
29
+ ]
30
+ },
13
31
  { "type": "ocd2", "file": "STCharacters.ocd2" }
14
32
  ]
15
33
  }
@@ -17,6 +35,7 @@
17
35
  {
18
36
  "dict": {
19
37
  "type": "group",
38
+ "match_policy": "short_circuit",
20
39
  "dicts": [
21
40
  { "type": "ocd2", "file": "TWVariantsPhrases.ocd2" },
22
41
  { "type": "ocd2", "file": "TWVariants.ocd2" }
package/dist/data/config/s2tw_jieba.json CHANGED
@@ -1,5 +1,13 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (Taiwan standard, Jieba Segmentation - Experimental)",
3
+ "normalization": [
4
+ {
5
+ "dict": {
6
+ "type": "ocd2",
7
+ "file": "CJK_Compatibility_Ideographs.ocd2"
8
+ }
9
+ }
10
+ ],
3
11
  "segmentation": {
4
12
  "type": "jieba",
5
13
  "resources": {
@@ -11,18 +19,42 @@
11
19
  {
12
20
  "dict": {
13
21
  "type": "group",
22
+ "match_policy": "short_circuit",
14
23
  "dicts": [
15
- { "type": "ocd2", "file": "STPhrases.ocd2" },
16
- { "type": "ocd2", "file": "STCharacters.ocd2" }
24
+ {
25
+ "type": "group",
26
+ "match_policy": "union",
27
+ "dicts": [
28
+ {
29
+ "type": "ocd2",
30
+ "file": "STPhrases.ocd2"
31
+ },
32
+ {
33
+ "type": "ocd2",
34
+ "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2"
35
+ }
36
+ ]
37
+ },
38
+ {
39
+ "type": "ocd2",
40
+ "file": "STCharacters.ocd2"
41
+ }
17
42
  ]
18
43
  }
19
44
  },
20
45
  {
21
46
  "dict": {
22
47
  "type": "group",
48
+ "match_policy": "short_circuit",
23
49
  "dicts": [
24
- { "type": "ocd2", "file": "TWVariantsPhrases.ocd2" },
25
- { "type": "ocd2", "file": "TWVariants.ocd2" }
50
+ {
51
+ "type": "ocd2",
52
+ "file": "TWVariantsPhrases.ocd2"
53
+ },
54
+ {
55
+ "type": "ocd2",
56
+ "file": "TWVariants.ocd2"
57
+ }
26
58
  ]
27
59
  }
28
60
  }
package/dist/data/config/s2twp.json CHANGED
@@ -1,15 +1,33 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (Taiwan Standard, with Taiwan Phrases)",
3
+ "normalization": [
4
+ { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
5
+ ],
3
6
  "segmentation": {
4
7
  "type": "mmseg",
5
- "dict": { "type": "ocd2", "file": "STPhrases.ocd2" }
8
+ "dict": {
9
+ "type": "group",
10
+ "match_policy": "union",
11
+ "dicts": [
12
+ { "type": "ocd2", "file": "STPhrases.ocd2" },
13
+ { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
14
+ ]
15
+ }
6
16
  },
7
17
  "conversion_chain": [
8
18
  {
9
19
  "dict": {
10
20
  "type": "group",
21
+ "match_policy": "short_circuit",
11
22
  "dicts": [
12
- { "type": "ocd2", "file": "STPhrases.ocd2" },
23
+ {
24
+ "type": "group",
25
+ "match_policy": "union",
26
+ "dicts": [
27
+ { "type": "ocd2", "file": "STPhrases.ocd2" },
28
+ { "type": "ocd2", "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2" }
29
+ ]
30
+ },
13
31
  { "type": "ocd2", "file": "STCharacters.ocd2" }
14
32
  ]
15
33
  }
@@ -17,6 +35,7 @@
17
35
  {
18
36
  "dict": {
19
37
  "type": "group",
38
+ "match_policy": "short_circuit",
20
39
  "dicts": [
21
40
  { "type": "ocd2", "file": "TWPhrases.ocd2" },
22
41
  { "type": "ocd2", "file": "TWVariantsPhrases.ocd2" },
package/dist/data/config/s2twp_jieba.json CHANGED
@@ -1,5 +1,13 @@
1
1
  {
2
2
  "name": "Simplified Chinese to Traditional Chinese (Taiwan standard, with phrases, Jieba Segmentation - Experimental)",
3
+ "normalization": [
4
+ {
5
+ "dict": {
6
+ "type": "ocd2",
7
+ "file": "CJK_Compatibility_Ideographs.ocd2"
8
+ }
9
+ }
10
+ ],
3
11
  "segmentation": {
4
12
  "type": "jieba",
5
13
  "resources": {
@@ -11,19 +19,46 @@
11
19
  {
12
20
  "dict": {
13
21
  "type": "group",
22
+ "match_policy": "short_circuit",
14
23
  "dicts": [
15
- { "type": "ocd2", "file": "STPhrases.ocd2" },
16
- { "type": "ocd2", "file": "STCharacters.ocd2" }
24
+ {
25
+ "type": "group",
26
+ "match_policy": "union",
27
+ "dicts": [
28
+ {
29
+ "type": "ocd2",
30
+ "file": "STPhrases.ocd2"
31
+ },
32
+ {
33
+ "type": "ocd2",
34
+ "file": "STPhrases_GeneratedFromRegionalPhrases.ocd2"
35
+ }
36
+ ]
37
+ },
38
+ {
39
+ "type": "ocd2",
40
+ "file": "STCharacters.ocd2"
41
+ }
17
42
  ]
18
43
  }
19
44
  },
20
45
  {
21
46
  "dict": {
22
47
  "type": "group",
48
+ "match_policy": "short_circuit",
23
49
  "dicts": [
24
- { "type": "ocd2", "file": "TWPhrases.ocd2" },
25
- { "type": "ocd2", "file": "TWVariantsPhrases.ocd2" },
26
- { "type": "ocd2", "file": "TWVariants.ocd2" }
50
+ {
51
+ "type": "ocd2",
52
+ "file": "TWPhrases.ocd2"
53
+ },
54
+ {
55
+ "type": "ocd2",
56
+ "file": "TWVariantsPhrases.ocd2"
57
+ },
58
+ {
59
+ "type": "ocd2",
60
+ "file": "TWVariants.ocd2"
61
+ }
27
62
  ]
28
63
  }
29
64
  }
package/dist/data/config/t2cngov.json CHANGED
@@ -7,16 +7,21 @@
7
7
  "file": "cngov/TGPhrases.ocd2"
8
8
  }
9
9
  },
10
- "conversion_chain": [{
11
- "dict": {
12
- "type": "group",
13
- "dicts": [{
14
- "type": "ocd2",
15
- "file": "cngov/TGPhrases.ocd2"
16
- }, {
17
- "type": "ocd2",
18
- "file": "cngov/TGCharacters.ocd2"
19
- }]
10
+ "conversion_chain": [
11
+ {
12
+ "dict": {
13
+ "type": "group",
14
+ "dicts": [
15
+ {
16
+ "type": "ocd2",
17
+ "file": "cngov/TGPhrases.ocd2"
18
+ },
19
+ {
20
+ "type": "ocd2",
21
+ "file": "cngov/TGCharacters.ocd2"
22
+ }
23
+ ]
24
+ }
20
25
  }
21
- }]
26
+ ]
22
27
  }
package/dist/data/config/t2cngov_keep_simp.json CHANGED
@@ -7,16 +7,21 @@
7
7
  "file": "cngov/TGPhrases.ocd2"
8
8
  }
9
9
  },
10
- "conversion_chain": [{
11
- "dict": {
12
- "type": "group",
13
- "dicts": [{
14
- "type": "ocd2",
15
- "file": "cngov/TGPhrases.ocd2"
16
- }, {
17
- "type": "ocd2",
18
- "file": "cngov/TGCharacters_keep_simp.ocd2"
19
- }]
10
+ "conversion_chain": [
11
+ {
12
+ "dict": {
13
+ "type": "group",
14
+ "dicts": [
15
+ {
16
+ "type": "ocd2",
17
+ "file": "cngov/TGPhrases.ocd2"
18
+ },
19
+ {
20
+ "type": "ocd2",
21
+ "file": "cngov/TGCharacters_keep_simp.ocd2"
22
+ }
23
+ ]
24
+ }
20
25
  }
21
- }]
26
+ ]
22
27
  }
package/dist/data/config/t2hk.json CHANGED
@@ -1,19 +1,13 @@
1
1
  {
2
2
  "name": "Traditional Chinese (OpenCC Standard) to Traditional Chinese (Hong Kong variant)",
3
- "segmentation": {
4
- "type": "mmseg",
5
- "dict": {
6
- "type": "group",
7
- "dicts": [
8
- { "type": "ocd2", "file": "HKVariantsPhrases.ocd2" },
9
- { "type": "ocd2", "file": "HKVariants.ocd2" }
10
- ]
11
- }
12
- },
3
+ "normalization": [
4
+ { "dict": { "type": "ocd2", "file": "CJK_Compatibility_Ideographs.ocd2" } }
5
+ ],
13
6
  "conversion_chain": [
14
7
  {
15
8
  "dict": {
16
9
  "type": "group",
10
+ "match_policy": "short_circuit",
17
11
  "dicts": [
18
12
  { "type": "ocd2", "file": "HKVariantsPhrases.ocd2" },
19
13
  { "type": "ocd2", "file": "HKVariants.ocd2" }