jsmigemo 0.3.1
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 +97 -0
- package/bin/jsmigemo-cli.js +122 -0
- package/bin/jsmigemo-dict.js +37 -0
- package/bin/jsmigemo-skk2migemo.js +61 -0
- package/demo-list.html +116 -0
- package/demo-text.html +65 -0
- package/esm/BitList.d.ts +8 -0
- package/esm/BitList.js +39 -0
- package/esm/BitList.js.map +1 -0
- package/esm/BitVector.d.ts +15 -0
- package/esm/BitVector.js +150 -0
- package/esm/BitVector.js.map +1 -0
- package/esm/CharacterConverter.d.ts +3 -0
- package/esm/CharacterConverter.js +382 -0
- package/esm/CharacterConverter.js.map +1 -0
- package/esm/CompactDictionary.d.ts +17 -0
- package/esm/CompactDictionary.js +124 -0
- package/esm/CompactDictionary.js.map +1 -0
- package/esm/CompactDictionaryBuilder.d.ts +3 -0
- package/esm/CompactDictionaryBuilder.js +118 -0
- package/esm/CompactDictionaryBuilder.js.map +1 -0
- package/esm/CompactHiraganaString.d.ts +6 -0
- package/esm/CompactHiraganaString.js +41 -0
- package/esm/CompactHiraganaString.js.map +1 -0
- package/esm/DoubleArray.d.ts +12 -0
- package/esm/DoubleArray.js +70 -0
- package/esm/DoubleArray.js.map +1 -0
- package/esm/LOUDSTrie.d.ts +13 -0
- package/esm/LOUDSTrie.js +67 -0
- package/esm/LOUDSTrie.js.map +1 -0
- package/esm/LOUDSTrieBuilder.d.ts +4 -0
- package/esm/LOUDSTrieBuilder.js +69 -0
- package/esm/LOUDSTrieBuilder.js.map +1 -0
- package/esm/Migemo.d.ts +13 -0
- package/esm/Migemo.js +71 -0
- package/esm/Migemo.js.map +1 -0
- package/esm/RomajiProcessor.d.ts +18 -0
- package/esm/RomajiProcessor.js +478 -0
- package/esm/RomajiProcessor.js.map +1 -0
- package/esm/RomajiProcessor2.d.ts +15 -0
- package/esm/RomajiProcessor2.js +91 -0
- package/esm/RomajiProcessor2.js.map +1 -0
- package/esm/SimpleDictionary.d.ts +1 -0
- package/esm/SimpleDictionary.js +54 -0
- package/esm/SimpleDictionary.js.map +1 -0
- package/esm/TernaryRegexGenerator.d.ts +27 -0
- package/esm/TernaryRegexGenerator.js +202 -0
- package/esm/TernaryRegexGenerator.js.map +1 -0
- package/esm/index.d.ts +3 -0
- package/esm/index.js +5 -0
- package/esm/index.js.map +1 -0
- package/esm/utils.d.ts +5 -0
- package/esm/utils.js +88 -0
- package/esm/utils.js.map +1 -0
- package/index.html +55 -0
- package/jest.config.js +12 -0
- package/lib/BitList.d.ts +8 -0
- package/lib/BitList.js +43 -0
- package/lib/BitList.js.map +1 -0
- package/lib/BitVector.d.ts +15 -0
- package/lib/BitVector.js +154 -0
- package/lib/BitVector.js.map +1 -0
- package/lib/CharacterConverter.d.ts +3 -0
- package/lib/CharacterConverter.js +388 -0
- package/lib/CharacterConverter.js.map +1 -0
- package/lib/CompactDictionary.d.ts +17 -0
- package/lib/CompactDictionary.js +128 -0
- package/lib/CompactDictionary.js.map +1 -0
- package/lib/CompactDictionaryBuilder.d.ts +3 -0
- package/lib/CompactDictionaryBuilder.js +122 -0
- package/lib/CompactDictionaryBuilder.js.map +1 -0
- package/lib/CompactHiraganaString.d.ts +6 -0
- package/lib/CompactHiraganaString.js +45 -0
- package/lib/CompactHiraganaString.js.map +1 -0
- package/lib/DoubleArray.d.ts +12 -0
- package/lib/DoubleArray.js +74 -0
- package/lib/DoubleArray.js.map +1 -0
- package/lib/LOUDSTrie.d.ts +13 -0
- package/lib/LOUDSTrie.js +71 -0
- package/lib/LOUDSTrie.js.map +1 -0
- package/lib/LOUDSTrieBuilder.d.ts +4 -0
- package/lib/LOUDSTrieBuilder.js +73 -0
- package/lib/LOUDSTrieBuilder.js.map +1 -0
- package/lib/Migemo.d.ts +13 -0
- package/lib/Migemo.js +75 -0
- package/lib/Migemo.js.map +1 -0
- package/lib/RegexGenerator.d.ts +21 -0
- package/lib/RegexGenerator.js +150 -0
- package/lib/RegexGenerator.js.map +1 -0
- package/lib/RomajiProcessor.d.ts +18 -0
- package/lib/RomajiProcessor.js +484 -0
- package/lib/RomajiProcessor.js.map +1 -0
- package/lib/RomajiProcessor2.d.ts +15 -0
- package/lib/RomajiProcessor2.js +96 -0
- package/lib/RomajiProcessor2.js.map +1 -0
- package/lib/RomanEntry.d.ts +1 -0
- package/lib/RomanEntry.js +481 -0
- package/lib/RomanEntry.js.map +1 -0
- package/lib/SimpleDictionary.d.ts +1 -0
- package/lib/SimpleDictionary.js +56 -0
- package/lib/SimpleDictionary.js.map +1 -0
- package/lib/TernaryRegexGenerator.d.ts +27 -0
- package/lib/TernaryRegexGenerator.js +206 -0
- package/lib/TernaryRegexGenerator.js.map +1 -0
- package/lib/index.d.ts +3 -0
- package/lib/index.js +17 -0
- package/lib/index.js.map +1 -0
- package/lib/utils.d.ts +5 -0
- package/lib/utils.js +96 -0
- package/lib/utils.js.map +1 -0
- package/migemo-compact-dict +0 -0
- package/package.json +37 -0
- package/tsconfig.esm.json +9 -0
- package/umd/jsmigemo.js +1 -0
|
@@ -0,0 +1,124 @@
|
|
|
1
|
+
import { LOUDSTrie } from "./LOUDSTrie";
|
|
2
|
+
import { BitVector } from "./BitVector";
|
|
3
|
+
import { BitList } from "./BitList";
|
|
4
|
+
export class CompactDictionary {
|
|
5
|
+
constructor(buffer) {
|
|
6
|
+
const dv = new DataView(buffer);
|
|
7
|
+
let offset = 0;
|
|
8
|
+
[this.keyTrie, offset] = CompactDictionary.readTrie(dv, offset, true);
|
|
9
|
+
[this.valueTrie, offset] = CompactDictionary.readTrie(dv, offset, false);
|
|
10
|
+
const mappingBitVectorSize = dv.getUint32(offset);
|
|
11
|
+
offset += 4;
|
|
12
|
+
const mappingBitVectorWords = new Uint32Array(((mappingBitVectorSize + 63) >> 6) * 2);
|
|
13
|
+
for (let i = 0; i < mappingBitVectorWords.length >> 1; i++) {
|
|
14
|
+
mappingBitVectorWords[i * 2 + 1] = dv.getUint32(offset);
|
|
15
|
+
offset += 4;
|
|
16
|
+
mappingBitVectorWords[i * 2] = dv.getUint32(offset);
|
|
17
|
+
offset += 4;
|
|
18
|
+
}
|
|
19
|
+
this.mappingBitVector = new BitVector(mappingBitVectorWords, mappingBitVectorSize);
|
|
20
|
+
const mappingSize = dv.getUint32(offset);
|
|
21
|
+
offset += 4;
|
|
22
|
+
this.mapping = new Int32Array(mappingSize);
|
|
23
|
+
for (let i = 0; i < mappingSize; i++) {
|
|
24
|
+
this.mapping[i] = dv.getInt32(offset);
|
|
25
|
+
offset += 4;
|
|
26
|
+
}
|
|
27
|
+
if (offset != buffer.byteLength) {
|
|
28
|
+
throw new Error();
|
|
29
|
+
}
|
|
30
|
+
this.hasMappingBitList = CompactDictionary.createHasMappingBitList(this.mappingBitVector);
|
|
31
|
+
}
|
|
32
|
+
static readTrie(dv, offset, compactHiragana) {
|
|
33
|
+
const keyTrieEdgeSize = dv.getInt32(offset);
|
|
34
|
+
offset += 4;
|
|
35
|
+
const keyTrieEdges = new Uint16Array(keyTrieEdgeSize);
|
|
36
|
+
for (let i = 0; i < keyTrieEdgeSize; i++) {
|
|
37
|
+
let c;
|
|
38
|
+
if (compactHiragana) {
|
|
39
|
+
c = this.decode(dv.getUint8(offset));
|
|
40
|
+
offset += 1;
|
|
41
|
+
}
|
|
42
|
+
else {
|
|
43
|
+
c = dv.getUint16(offset);
|
|
44
|
+
offset += 2;
|
|
45
|
+
}
|
|
46
|
+
keyTrieEdges[i] = c;
|
|
47
|
+
}
|
|
48
|
+
const keyTrieBitVectorSize = dv.getUint32(offset);
|
|
49
|
+
offset += 4;
|
|
50
|
+
const keyTrieBitVectorWords = new Uint32Array(((keyTrieBitVectorSize + 63) >> 6) * 2);
|
|
51
|
+
for (let i = 0; i < keyTrieBitVectorWords.length >>> 1; i++) {
|
|
52
|
+
keyTrieBitVectorWords[i * 2 + 1] = dv.getUint32(offset);
|
|
53
|
+
offset += 4;
|
|
54
|
+
keyTrieBitVectorWords[i * 2] = dv.getUint32(offset);
|
|
55
|
+
offset += 4;
|
|
56
|
+
}
|
|
57
|
+
return [new LOUDSTrie(new BitVector(keyTrieBitVectorWords, keyTrieBitVectorSize), keyTrieEdges), offset];
|
|
58
|
+
}
|
|
59
|
+
static decode(c) {
|
|
60
|
+
if (0x20 <= c && c <= 0x7e) {
|
|
61
|
+
return c;
|
|
62
|
+
}
|
|
63
|
+
if (0xa1 <= c && c <= 0xf6) {
|
|
64
|
+
return (c + 0x3040 - 0xa0);
|
|
65
|
+
}
|
|
66
|
+
throw new RangeError();
|
|
67
|
+
}
|
|
68
|
+
static encode(c) {
|
|
69
|
+
if (0x20 <= c && c <= 0x7e) {
|
|
70
|
+
return c;
|
|
71
|
+
}
|
|
72
|
+
if (0x3041 <= c && c <= 0x3096) {
|
|
73
|
+
return (c - 0x3040 + 0xa0);
|
|
74
|
+
}
|
|
75
|
+
if (0x30fc == c) {
|
|
76
|
+
return (c - 0x3040 + 0xa0);
|
|
77
|
+
}
|
|
78
|
+
throw new RangeError();
|
|
79
|
+
}
|
|
80
|
+
static createHasMappingBitList(mappingBitVector) {
|
|
81
|
+
const numOfNodes = mappingBitVector.rank(mappingBitVector.size() + 1, false);
|
|
82
|
+
const bitList = new BitList(numOfNodes);
|
|
83
|
+
let bitPosition = 0;
|
|
84
|
+
for (let node = 1; node < numOfNodes; node++) {
|
|
85
|
+
let hasMapping = mappingBitVector.get(bitPosition + 1);
|
|
86
|
+
bitList.set(node, hasMapping);
|
|
87
|
+
bitPosition = mappingBitVector.nextClearBit(bitPosition + 1);
|
|
88
|
+
}
|
|
89
|
+
return bitList;
|
|
90
|
+
}
|
|
91
|
+
*search(key) {
|
|
92
|
+
const keyIndex = this.keyTrie.lookup(key);
|
|
93
|
+
if (keyIndex != -1 && this.hasMappingBitList.get(keyIndex)) {
|
|
94
|
+
const valueStartPos = this.mappingBitVector.select(keyIndex, false);
|
|
95
|
+
const valueEndPos = this.mappingBitVector.nextClearBit(valueStartPos + 1);
|
|
96
|
+
const size = valueEndPos - valueStartPos - 1;
|
|
97
|
+
if (size > 0) {
|
|
98
|
+
const offset = this.mappingBitVector.rank(valueStartPos, false);
|
|
99
|
+
const result = new Array(size);
|
|
100
|
+
for (let i = 0; i < result.length; i++) {
|
|
101
|
+
yield this.valueTrie.reverseLookup(this.mapping[valueStartPos - offset + i]);
|
|
102
|
+
}
|
|
103
|
+
return result;
|
|
104
|
+
}
|
|
105
|
+
}
|
|
106
|
+
}
|
|
107
|
+
*predictiveSearch(key) {
|
|
108
|
+
const keyIndex = this.keyTrie.lookup(key);
|
|
109
|
+
if (keyIndex > 1) {
|
|
110
|
+
for (let i of this.keyTrie.predictiveSearch(keyIndex)) {
|
|
111
|
+
if (this.hasMappingBitList.get(i)) {
|
|
112
|
+
const valueStartPos = this.mappingBitVector.select(i, false);
|
|
113
|
+
const valueEndPos = this.mappingBitVector.nextClearBit(valueStartPos + 1);
|
|
114
|
+
const size = valueEndPos - valueStartPos - 1;
|
|
115
|
+
const offset = this.mappingBitVector.rank(valueStartPos, false);
|
|
116
|
+
for (let j = 0; j < size; j++) {
|
|
117
|
+
yield this.valueTrie.reverseLookup(this.mapping[valueStartPos - offset + j]);
|
|
118
|
+
}
|
|
119
|
+
}
|
|
120
|
+
}
|
|
121
|
+
}
|
|
122
|
+
}
|
|
123
|
+
}
|
|
124
|
+
//# sourceMappingURL=CompactDictionary.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"CompactDictionary.js","sourceRoot":"","sources":["../src/CompactDictionary.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AACxC,OAAO,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AACxC,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAEpC,MAAM,OAAO,iBAAiB;IAO1B,YAAY,MAAmB;QAC3B,MAAM,EAAE,GAAG,IAAI,QAAQ,CAAC,MAAM,CAAC,CAAC;QAChC,IAAI,MAAM,GAAG,CAAC,CAAC;QACf,CAAC,IAAI,CAAC,OAAO,EAAE,MAAM,CAAC,GAAG,iBAAiB,CAAC,QAAQ,CAAC,EAAE,EAAE,MAAM,EAAE,IAAI,CAAC,CAAC;QACtE,CAAC,IAAI,CAAC,SAAS,EAAE,MAAM,CAAC,GAAG,iBAAiB,CAAC,QAAQ,CAAC,EAAE,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC;QACzE,MAAM,oBAAoB,GAAG,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;QAClD,MAAM,IAAI,CAAC,CAAC;QACZ,MAAM,qBAAqB,GAAG,IAAI,WAAW,CAAC,CAAC,CAAC,oBAAoB,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QACtF,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,qBAAqB,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE;YACxD,qBAAqB,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;YACxD,MAAM,IAAI,CAAC,CAAC;YACZ,qBAAqB,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;YACpD,MAAM,IAAI,CAAC,CAAC;SACf;QACD,IAAI,CAAC,gBAAgB,GAAG,IAAI,SAAS,CAAC,qBAAqB,EAAE,oBAAoB,CAAC,CAAC;QACnF,MAAM,WAAW,GAAG,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;QACzC,MAAM,IAAI,CAAC,CAAC;QACZ,IAAI,CAAC,OAAO,GAAG,IAAI,UAAU,CAAC,WAAW,CAAC,CAAC;QAC3C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,WAAW,EAAE,CAAC,EAAE,EAAE;YAClC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YACtC,MAAM,IAAI,CAAC,CAAC;SACf;QACD,IAAI,MAAM,IAAI,MAAM,CAAC,UAAU,EAAE;YAC7B,MAAM,IAAI,KAAK,EAAE,CAAC;SACrB;QACD,IAAI,CAAC,iBAAiB,GAAG,iBAAiB,CAAC,uBAAuB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC;IAC9F,CAAC;IAEO,MAAM,CAAC,QAAQ,CAAC,EAAY,EAAE,MAAc,EAAE,eAAwB;QAC1E,MAAM,eAAe,GAAG,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;QAC5C,MAAM,IAAI,CAAC,CAAC;QACZ,MAAM,YAAY,GAAG,IAAI,WAAW,CAAC,eAAe,CAAC,CAAC;QACtD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,eAAe,EAAE,CAAC,EAAE,EAAE;YACtC,IAAI,CAAS,CAAC;YACd,IAAI,eAAe,EAAE;gBACjB,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC;gBACrC,MAAM,IAAI,CAAC,CAAC;aACf;iBAAM;gBACH,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;gBACzB,MAAM,IAAI,CAAC,CAAC;aACf;YACD,YAAY,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;SACvB;QACD,MAAM,oBAAoB,GAAG,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;QAClD,MAAM,IAAI,CAAC,CAAC;QACZ,MAAM,qBAAqB,GAAG,IAAI,WAAW,CAAC,CAAC,CAAC,oBAAoB,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QACtF,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,qBAAqB,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC,EAAE,EAAE;YACzD,qBAAqB,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;YACxD,MAAM,IAAI,CAAC,CAAC;YACZ,qBAAqB,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;YACpD,MAAM,IAAI,CAAC,CAAC;SACf;QACD,OAAO,CAAC,IAAI,SAAS,CAAC,IAAI,SAAS,CAAC,qBAAqB,EAAE,oBAAoB,CAAC,EAAE,YAAY,CAAC,EAAE,MAAM,CAAC,CAAC;IAC7G,CAAC;IAEO,MAAM,CAAC,MAAM,CAAC,CAAS;QAC3B,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE;YACxB,OAAO,CAAC,CAAC;SACZ;QACD,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE;YACxB,OAAO,CAAC,CAAC,GAAG,MAAM,GAAG,IAAI,CAAC,CAAC;SAC9B;QACD,MAAM,IAAI,UAAU,EAAE,CAAC;IAC3B,CAAC;IAEO,MAAM,CAAC,MAAM,CAAC,CAAS;QAC3B,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE;YACxB,OAAO,CAAC,CAAC;SACZ;QACD,IAAI,MAAM,IAAI,CAAC,IAAI,CAAC,IAAI,MAAM,EAAE;YAC5B,OAAO,CAAC,CAAC,GAAG,MAAM,GAAG,IAAI,CAAC,CAAC;SAC9B;QACD,IAAI,MAAM,IAAI,CAAC,EAAE;YACb,OAAO,CAAC,CAAC,GAAG,MAAM,GAAG,IAAI,CAAC,CAAC;SAC9B;QACD,MAAM,IAAI,UAAU,EAAE,CAAC;IAC3B,CAAC;IAEO,MAAM,CAAC,uBAAuB,CAAC,gBAA2B;QAC9D,MAAM,UAAU,GAAG,gBAAgB,CAAC,IAAI,CAAC,gBAAgB,CAAC,IAAI,EAAE,GAAG,CAAC,EAAE,KAAK,CAAC,CAAC;QAC7E,MAAM,OAAO,GAAG,IAAI,OAAO,CAAC,UAAU,CAAC,CAAC;QACxC,IAAI,WAAW,GAAG,CAAC,CAAC;QACpB,KAAK,IAAI,IAAI,GAAG,CAAC,EAAE,IAAI,GAAG,UAAU,EAAE,IAAI,EAAE,EAAE;YAC1C,IAAI,UAAU,GAAG,gBAAgB,CAAC,GAAG,CAAC,WAAW,GAAG,CAAC,CAAC,CAAC;YACvD,OAAO,CAAC,GAAG,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;YAC9B,WAAW,GAAG,gBAAgB,CAAC,YAAY,CAAC,WAAW,GAAG,CAAC,CAAC,CAAC;SAChE;QACD,OAAO,OAAO,CAAC;IACnB,CAAC;IAED,CAAC,MAAM,CAAC,GAAW;QACf,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAC1C,IAAI,QAAQ,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,iBAAiB,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE;YACxD,MAAM,aAAa,GAAG,IAAI,CAAC,gBAAgB,CAAC,MAAM,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;YACpE,MAAM,WAAW,GAAG,IAAI,CAAC,gBAAgB,CAAC,YAAY,CAAC,aAAa,GAAG,CAAC,CAAC,CAAC;YAC1E,MAAM,IAAI,GAAG,WAAW,GAAG,aAAa,GAAG,CAAC,CAAC;YAC7C,IAAI,IAAI,GAAG,CAAC,EAAE;gBACV,MAAM,MAAM,GAAG,IAAI,CAAC,gBAAgB,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;gBAChE,MAAM,MAAM,GAAG,IAAI,KAAK,CAAS,IAAI,CAAC,CAAC;gBACvC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;oBACpC,MAAM,IAAI,CAAC,SAAS,CAAC,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,aAAa,GAAG,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC;iBAChF;gBACD,OAAO,MAAM,CAAC;aACjB;SACJ;IACL,CAAC;IAED,CAAC,gBAAgB,CAAC,GAAW;QACzB,MAAM,QAAQ,GAAG,IAAI,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAC1C,IAAI,QAAQ,GAAG,CAAC,EAAE;YACd,KAAK,IAAI,CAAC,IAAI,IAAI,CAAC,OAAO,CAAC,gBAAgB,CAAC,QAAQ,CAAC,EAAE;gBACnD,IAAI,IAAI,CAAC,iBAAiB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE;oBAC/B,MAAM,aAAa,GAAG,IAAI,CAAC,gBAAgB,CAAC,MAAM,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;oBAC7D,MAAM,WAAW,GAAG,IAAI,CAAC,gBAAgB,CAAC,YAAY,CAAC,aAAa,GAAG,CAAC,CAAC,CAAC;oBAC1E,MAAM,IAAI,GAAG,WAAW,GAAG,aAAa,GAAG,CAAC,CAAC;oBAC7C,MAAM,MAAM,GAAG,IAAI,CAAC,gBAAgB,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;oBAChE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,EAAE,CAAC,EAAE,EAAE;wBAC3B,MAAM,IAAI,CAAC,SAAS,CAAC,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,aAAa,GAAG,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC;qBAChF;iBACJ;aACJ;SACJ;IACL,CAAC;CACJ"}
|
|
@@ -0,0 +1,118 @@
|
|
|
1
|
+
import { LOUDSTrieBuilder } from "./LOUDSTrieBuilder";
|
|
2
|
+
import { CompactHiraganaString } from "./CompactHiraganaString";
|
|
3
|
+
import { BitList } from "./BitList";
|
|
4
|
+
export class CompactDictionaryBuilder {
|
|
5
|
+
static build(dict) {
|
|
6
|
+
// remove some keys
|
|
7
|
+
const keysToRemove = new Array();
|
|
8
|
+
for (const key of dict.keys()) {
|
|
9
|
+
try {
|
|
10
|
+
CompactHiraganaString.encodeString(key);
|
|
11
|
+
}
|
|
12
|
+
catch (e) {
|
|
13
|
+
keysToRemove.push(key);
|
|
14
|
+
console.log("skipped the world: " + key);
|
|
15
|
+
}
|
|
16
|
+
}
|
|
17
|
+
for (const key of keysToRemove) {
|
|
18
|
+
dict.delete(key);
|
|
19
|
+
}
|
|
20
|
+
// build key trie
|
|
21
|
+
const keys = Array.from(dict.keys()).sort();
|
|
22
|
+
const keyTrie = LOUDSTrieBuilder.build(keys)[0];
|
|
23
|
+
// build value trie
|
|
24
|
+
const valuesSet = new Set();
|
|
25
|
+
for (const value of dict.values()) {
|
|
26
|
+
for (const v of value) {
|
|
27
|
+
valuesSet.add(v);
|
|
28
|
+
}
|
|
29
|
+
}
|
|
30
|
+
const values = Array.from(valuesSet.values()).sort();
|
|
31
|
+
const valueTrie = LOUDSTrieBuilder.build(values)[0];
|
|
32
|
+
// build trie mapping
|
|
33
|
+
let mappingCount = 0;
|
|
34
|
+
for (const i of dict.values()) {
|
|
35
|
+
mappingCount += i.length;
|
|
36
|
+
}
|
|
37
|
+
const mapping = new Uint32Array(mappingCount);
|
|
38
|
+
let mappingIndex = 0;
|
|
39
|
+
const mappingBitList = new BitList();
|
|
40
|
+
for (let i = 1; i <= keyTrie.size(); i++) {
|
|
41
|
+
let key = keyTrie.reverseLookup(i);
|
|
42
|
+
mappingBitList.add(false);
|
|
43
|
+
let values = dict.get(key);
|
|
44
|
+
if (values != undefined) {
|
|
45
|
+
for (let j = 0; j < values.length; j++) {
|
|
46
|
+
mappingBitList.add(true);
|
|
47
|
+
mapping[mappingIndex] = valueTrie.lookup(values[j]);
|
|
48
|
+
mappingIndex++;
|
|
49
|
+
}
|
|
50
|
+
}
|
|
51
|
+
}
|
|
52
|
+
// calculate output size
|
|
53
|
+
const keyTrieDataSize = 8 + keyTrie.edges.length + ((keyTrie.bitVector.size() + 63) >>> 6) * 8;
|
|
54
|
+
const valueTrieDataSize = 8 + valueTrie.edges.length * 2 + ((valueTrie.bitVector.size() + 63) >>> 6) * 8;
|
|
55
|
+
const mappingDataSize = 8 + ((mappingBitList.size + 63) >>> 6) * 8 + mapping.length * 4;
|
|
56
|
+
const outputDataSize = keyTrieDataSize + valueTrieDataSize + mappingDataSize;
|
|
57
|
+
// ready output
|
|
58
|
+
const arrayBuffer = new ArrayBuffer(outputDataSize);
|
|
59
|
+
const dataView = new DataView(arrayBuffer);
|
|
60
|
+
let dataViewIndex = 0;
|
|
61
|
+
// output key trie
|
|
62
|
+
dataView.setInt32(dataViewIndex, keyTrie.edges.length);
|
|
63
|
+
dataViewIndex += 4;
|
|
64
|
+
for (let i = 0; i < keyTrie.edges.length; i++) {
|
|
65
|
+
const compactChar = CompactHiraganaString.encodeChar(keyTrie.edges[i]);
|
|
66
|
+
dataView.setUint8(dataViewIndex, compactChar);
|
|
67
|
+
dataViewIndex += 1;
|
|
68
|
+
}
|
|
69
|
+
dataView.setInt32(dataViewIndex, keyTrie.bitVector.size());
|
|
70
|
+
dataViewIndex += 4;
|
|
71
|
+
const keyTrieBitVectorWords = keyTrie.bitVector.words;
|
|
72
|
+
for (let i = 0; i < keyTrieBitVectorWords.length >>> 1; i++) {
|
|
73
|
+
dataView.setUint32(dataViewIndex, keyTrieBitVectorWords[i * 2 + 1]);
|
|
74
|
+
dataViewIndex += 4;
|
|
75
|
+
dataView.setUint32(dataViewIndex, keyTrieBitVectorWords[i * 2]);
|
|
76
|
+
dataViewIndex += 4;
|
|
77
|
+
}
|
|
78
|
+
// output value trie
|
|
79
|
+
dataView.setInt32(dataViewIndex, valueTrie.edges.length);
|
|
80
|
+
dataViewIndex += 4;
|
|
81
|
+
for (let i = 0; i < valueTrie.edges.length; i++) {
|
|
82
|
+
dataView.setUint16(dataViewIndex, valueTrie.edges[i]);
|
|
83
|
+
dataViewIndex += 2;
|
|
84
|
+
}
|
|
85
|
+
dataView.setInt32(dataViewIndex, valueTrie.bitVector.size());
|
|
86
|
+
dataViewIndex += 4;
|
|
87
|
+
const valueTrieBitVectorWords = valueTrie.bitVector.words;
|
|
88
|
+
for (let i = 0; i < valueTrieBitVectorWords.length >>> 1; i++) {
|
|
89
|
+
dataView.setUint32(dataViewIndex, valueTrieBitVectorWords[i * 2 + 1]);
|
|
90
|
+
dataViewIndex += 4;
|
|
91
|
+
dataView.setUint32(dataViewIndex, valueTrieBitVectorWords[i * 2]);
|
|
92
|
+
dataViewIndex += 4;
|
|
93
|
+
}
|
|
94
|
+
// output mapping
|
|
95
|
+
dataView.setInt32(dataViewIndex, mappingBitList.size);
|
|
96
|
+
dataViewIndex += 4;
|
|
97
|
+
const mappingWordsLen = (mappingBitList.size + 63) >> 6;
|
|
98
|
+
for (let i = 0; i < mappingWordsLen; i++) {
|
|
99
|
+
dataView.setUint32(dataViewIndex, mappingBitList.words[i * 2 + 1]);
|
|
100
|
+
dataViewIndex += 4;
|
|
101
|
+
dataView.setUint32(dataViewIndex, mappingBitList.words[i * 2]);
|
|
102
|
+
dataViewIndex += 4;
|
|
103
|
+
}
|
|
104
|
+
// TODO: padding to 64bit words
|
|
105
|
+
dataView.setInt32(dataViewIndex, mapping.length);
|
|
106
|
+
dataViewIndex += 4;
|
|
107
|
+
for (let i = 0; i < mapping.length; i++) {
|
|
108
|
+
dataView.setUint32(dataViewIndex, mapping[i]);
|
|
109
|
+
dataViewIndex += 4;
|
|
110
|
+
}
|
|
111
|
+
// check data size
|
|
112
|
+
if (dataViewIndex !== outputDataSize) {
|
|
113
|
+
throw new Error(`file size is not valid: expected=${outputDataSize} actual=${dataViewIndex}`);
|
|
114
|
+
}
|
|
115
|
+
return arrayBuffer;
|
|
116
|
+
}
|
|
117
|
+
}
|
|
118
|
+
//# sourceMappingURL=CompactDictionaryBuilder.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"CompactDictionaryBuilder.js","sourceRoot":"","sources":["../src/CompactDictionaryBuilder.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACtD,OAAO,EAAE,qBAAqB,EAAE,MAAM,yBAAyB,CAAC;AAChE,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAEpC,MAAM,OAAO,wBAAwB;IAC1B,MAAM,CAAC,KAAK,CAAC,IAA2B;QAC3C,mBAAmB;QACnB,MAAM,YAAY,GAAG,IAAI,KAAK,EAAU,CAAC;QACzC,KAAK,MAAM,GAAG,IAAI,IAAI,CAAC,IAAI,EAAE,EAAE;YAC3B,IAAI;gBACA,qBAAqB,CAAC,YAAY,CAAC,GAAG,CAAC,CAAC;aAC3C;YAAC,OAAO,CAAC,EAAE;gBACR,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBACvB,OAAO,CAAC,GAAG,CAAC,qBAAqB,GAAG,GAAG,CAAC,CAAC;aAC5C;SACJ;QACD,KAAK,MAAM,GAAG,IAAI,YAAY,EAAE;YAC5B,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;SACpB;QAED,iBAAiB;QACjB,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;QAC5C,MAAM,OAAO,GAAG,gBAAgB,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;QAEhD,mBAAmB;QACnB,MAAM,SAAS,GAAG,IAAI,GAAG,EAAU,CAAC;QACpC,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,MAAM,EAAE,EAAE;YAC/B,KAAK,MAAM,CAAC,IAAI,KAAK,EAAE;gBACnB,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;aACpB;SACJ;QACD,MAAM,MAAM,GAAG,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;QACrD,MAAM,SAAS,GAAG,gBAAgB,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;QAEpD,qBAAqB;QACrB,IAAI,YAAY,GAAG,CAAC,CAAC;QACrB,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,MAAM,EAAE,EAAE;YAC3B,YAAY,IAAI,CAAC,CAAC,MAAM,CAAC;SAC5B;QACD,MAAM,OAAO,GAAG,IAAI,WAAW,CAAC,YAAY,CAAC,CAAC;QAC9C,IAAI,YAAY,GAAG,CAAC,CAAC;QACrB,MAAM,cAAc,GAAG,IAAI,OAAO,EAAE,CAAC;QACrC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,EAAE,EAAE;YACtC,IAAI,GAAG,GAAG,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC;YACnC,cAAc,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;YAC1B,IAAI,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YAC3B,IAAI,MAAM,IAAI,SAAS,EAAE;gBACrB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;oBACpC,cAAc,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;oBACzB,OAAO,CAAC,YAAY,CAAC,GAAG,SAAS,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;oBACpD,YAAY,EAAE,CAAC;iBAClB;aACJ;SACJ;QAED,wBAAwB;QACxB,MAAM,eAAe,GAAG,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,OAAO,CAAC,SAAS,CAAC,IAAI,EAAE,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC;QAC/F,MAAM,iBAAiB,GAAG,CAAC,GAAG,SAAS,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,GAAG,CAAC,CAAC,SAAS,CAAC,SAAS,CAAC,IAAI,EAAE,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC;QACzG,MAAM,eAAe,GAAG,CAAC,GAAG,CAAC,CAAC,cAAc,CAAC,IAAI,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,GAAG,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC;QACxF,MAAM,cAAc,GAAG,eAAe,GAAG,iBAAiB,GAAG,eAAe,CAAC;QAE7E,eAAe;QACf,MAAM,WAAW,GAAG,IAAI,WAAW,CAAC,cAAc,CAAC,CAAC;QACpD,MAAM,QAAQ,GAAG,IAAI,QAAQ,CAAC,WAAW,CAAC,CAAC;QAC3C,IAAI,aAAa,GAAG,CAAC,CAAC;QAEtB,kBAAkB;QAClB,QAAQ,CAAC,QAAQ,CAAC,aAAa,EAAE,OAAO,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QACvD,aAAa,IAAI,CAAC,CAAC;QACnB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YAC3C,MAAM,WAAW,GAAG,qBAAqB,CAAC,UAAU,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;YACvE,QAAQ,CAAC,QAAQ,CAAC,aAAa,EAAE,WAAW,CAAC,CAAC;YAC9C,aAAa,IAAI,CAAC,CAAC;SACtB;QACD,QAAQ,CAAC,QAAQ,CAAC,aAAa,EAAE,OAAO,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC;QAC3D,aAAa,IAAI,CAAC,CAAC;QACnB,MAAM,qBAAqB,GAAG,OAAO,CAAC,SAAS,CAAC,KAAK,CAAC;QACtD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,qBAAqB,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC,EAAE,EAAE;YACzD,QAAQ,CAAC,SAAS,CAAC,aAAa,EAAE,qBAAqB,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YACpE,aAAa,IAAI,CAAC,CAAC;YACnB,QAAQ,CAAC,SAAS,CAAC,aAAa,EAAE,qBAAqB,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YAChE,aAAa,IAAI,CAAC,CAAC;SACtB;QAED,oBAAoB;QACpB,QAAQ,CAAC,QAAQ,CAAC,aAAa,EAAE,SAAS,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QACzD,aAAa,IAAI,CAAC,CAAC;QACnB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YAC7C,QAAQ,CAAC,SAAS,CAAC,aAAa,EAAE,SAAS,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;YACtD,aAAa,IAAI,CAAC,CAAC;SACtB;QACD,QAAQ,CAAC,QAAQ,CAAC,aAAa,EAAE,SAAS,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,CAAC;QAC7D,aAAa,IAAI,CAAC,CAAC;QACnB,MAAM,uBAAuB,GAAG,SAAS,CAAC,SAAS,CAAC,KAAK,CAAC;QAC1D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,uBAAuB,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC,EAAE,EAAE;YAC3D,QAAQ,CAAC,SAAS,CAAC,aAAa,EAAE,uBAAuB,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YACtE,aAAa,IAAI,CAAC,CAAC;YACnB,QAAQ,CAAC,SAAS,CAAC,aAAa,EAAE,uBAAuB,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YAClE,aAAa,IAAI,CAAC,CAAC;SACtB;QAED,iBAAiB;QACjB,QAAQ,CAAC,QAAQ,CAAC,aAAa,EAAE,cAAc,CAAC,IAAI,CAAC,CAAC;QACtD,aAAa,IAAI,CAAC,CAAC;QACnB,MAAM,eAAe,GAAG,CAAC,cAAc,CAAC,IAAI,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC;QACxD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,eAAe,EAAE,CAAC,EAAE,EAAE;YACtC,QAAQ,CAAC,SAAS,CAAC,aAAa,EAAE,cAAc,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YACnE,aAAa,IAAI,CAAC,CAAC;YACnB,QAAQ,CAAC,SAAS,CAAC,aAAa,EAAE,cAAc,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YAC/D,aAAa,IAAI,CAAC,CAAC;SACtB;QACD,+BAA+B;QAC/B,QAAQ,CAAC,QAAQ,CAAC,aAAa,EAAE,OAAO,CAAC,MAAM,CAAC,CAAC;QACjD,aAAa,IAAI,CAAC,CAAC;QACnB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACrC,QAAQ,CAAC,SAAS,CAAC,aAAa,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;YAC9C,aAAa,IAAI,CAAC,CAAC;SACtB;QAED,kBAAkB;QAClB,IAAI,aAAa,KAAK,cAAc,EAAE;YAClC,MAAM,IAAI,KAAK,CAAC,oCAAoC,cAAc,WAAW,aAAa,EAAE,CAAC,CAAC;SACjG;QAED,OAAO,WAAW,CAAC;IACvB,CAAC;CACJ"}
|
|
@@ -0,0 +1,41 @@
|
|
|
1
|
+
export class CompactHiraganaString {
|
|
2
|
+
static decodeBytes(bytes) {
|
|
3
|
+
let result = "";
|
|
4
|
+
for (let i = 0; i < bytes.length; i++) {
|
|
5
|
+
result += CompactHiraganaString.decodeByte(bytes[i]);
|
|
6
|
+
}
|
|
7
|
+
return result;
|
|
8
|
+
}
|
|
9
|
+
static decodeByte(c) {
|
|
10
|
+
if (0x20 <= c && c <= 0x7e) {
|
|
11
|
+
return String.fromCharCode(c);
|
|
12
|
+
}
|
|
13
|
+
if (0xa1 <= c && c <= 0xf6) {
|
|
14
|
+
return String.fromCharCode(c + 0x3040 - 0xa0);
|
|
15
|
+
}
|
|
16
|
+
throw new RangeError();
|
|
17
|
+
}
|
|
18
|
+
static encodeString(str) {
|
|
19
|
+
const result = new Uint8Array(str.length);
|
|
20
|
+
for (let i = 0; i < str.length; i++) {
|
|
21
|
+
result[i] = CompactHiraganaString.encodeChar(str.charCodeAt(i));
|
|
22
|
+
}
|
|
23
|
+
return result;
|
|
24
|
+
}
|
|
25
|
+
static encodeChar(b) {
|
|
26
|
+
if (b == 0) {
|
|
27
|
+
return 0;
|
|
28
|
+
}
|
|
29
|
+
if (0x20 <= b && b <= 0x7e) {
|
|
30
|
+
return b;
|
|
31
|
+
}
|
|
32
|
+
if (0x3041 <= b && b <= 0x3096) {
|
|
33
|
+
return b - 0x3040 + 0xa0;
|
|
34
|
+
}
|
|
35
|
+
if (0x30fc === b) {
|
|
36
|
+
return b - 0x3040 + 0xa0;
|
|
37
|
+
}
|
|
38
|
+
throw new RangeError('unknown character to encode: ' + b);
|
|
39
|
+
}
|
|
40
|
+
}
|
|
41
|
+
//# sourceMappingURL=CompactHiraganaString.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"CompactHiraganaString.js","sourceRoot":"","sources":["../src/CompactHiraganaString.ts"],"names":[],"mappings":"AACA,MAAM,OAAO,qBAAqB;IACvB,MAAM,CAAC,WAAW,CAAC,KAAiB;QACvC,IAAI,MAAM,GAAG,EAAE,CAAC;QAChB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACnC,MAAM,IAAI,qBAAqB,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;SACxD;QACD,OAAO,MAAM,CAAC;IAClB,CAAC;IAEM,MAAM,CAAC,UAAU,CAAC,CAAS;QAC9B,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE;YACxB,OAAO,MAAM,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC;SACjC;QACD,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE;YACxB,OAAO,MAAM,CAAC,YAAY,CAAC,CAAC,GAAG,MAAM,GAAG,IAAI,CAAC,CAAC;SACjD;QACD,MAAM,IAAI,UAAU,EAAE,CAAC;IAC3B,CAAC;IAEM,MAAM,CAAC,YAAY,CAAC,GAAW;QAClC,MAAM,MAAM,GAAG,IAAI,UAAU,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;QAC1C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACjC,MAAM,CAAC,CAAC,CAAC,GAAG,qBAAqB,CAAC,UAAU,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC;SACnE;QACD,OAAO,MAAM,CAAC;IAClB,CAAC;IAEM,MAAM,CAAC,UAAU,CAAC,CAAS;QAC9B,IAAI,CAAC,IAAI,CAAC,EAAE;YACR,OAAO,CAAC,CAAC;SACZ;QACD,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,EAAE;YACxB,OAAO,CAAC,CAAC;SACZ;QACD,IAAI,MAAM,IAAI,CAAC,IAAI,CAAC,IAAI,MAAM,EAAE;YAC5B,OAAO,CAAC,GAAG,MAAM,GAAG,IAAI,CAAC;SAC5B;QACD,IAAI,MAAM,KAAK,CAAC,EAAE;YACd,OAAO,CAAC,GAAG,MAAM,GAAG,IAAI,CAAC;SAC5B;QACD,MAAM,IAAI,UAAU,CAAC,+BAA+B,GAAG,CAAC,CAAC,CAAC;IAC9D,CAAC;CACJ"}
|
|
@@ -0,0 +1,12 @@
|
|
|
1
|
+
export declare class DoubleArray {
|
|
2
|
+
base: Int16Array;
|
|
3
|
+
check: Int16Array;
|
|
4
|
+
charConverter: (arg: number) => number;
|
|
5
|
+
charSize: number;
|
|
6
|
+
constructor(base: Int16Array, check: Int16Array, charConverter: (arg: number) => number, charSize: number);
|
|
7
|
+
traverse(n: number, k: number): number;
|
|
8
|
+
lookup(str: string): number;
|
|
9
|
+
commonPrefixSearch(key: string): IterableIterator<number>;
|
|
10
|
+
predictiveSearch(key: string): IterableIterator<number>;
|
|
11
|
+
visitRecursive(n: number): IterableIterator<number>;
|
|
12
|
+
}
|
|
@@ -0,0 +1,70 @@
|
|
|
1
|
+
export class DoubleArray {
|
|
2
|
+
constructor(base, check, charConverter, charSize) {
|
|
3
|
+
this.base = base;
|
|
4
|
+
this.check = check;
|
|
5
|
+
this.charConverter = charConverter;
|
|
6
|
+
this.charSize = charSize;
|
|
7
|
+
}
|
|
8
|
+
traverse(n, k) {
|
|
9
|
+
const m = this.base[n] + k;
|
|
10
|
+
if (this.check[m] == n) {
|
|
11
|
+
return m;
|
|
12
|
+
}
|
|
13
|
+
else {
|
|
14
|
+
return -1;
|
|
15
|
+
}
|
|
16
|
+
}
|
|
17
|
+
lookup(str) {
|
|
18
|
+
if (str.length == 0) {
|
|
19
|
+
return 0;
|
|
20
|
+
}
|
|
21
|
+
let n = 0;
|
|
22
|
+
for (let i = 0; i < str.length; i++) {
|
|
23
|
+
const c = this.charConverter(str.charCodeAt(i));
|
|
24
|
+
if (c < 1) {
|
|
25
|
+
throw new Error();
|
|
26
|
+
}
|
|
27
|
+
n = this.traverse(n, c);
|
|
28
|
+
if (n == -1) {
|
|
29
|
+
return -1;
|
|
30
|
+
}
|
|
31
|
+
}
|
|
32
|
+
return n;
|
|
33
|
+
}
|
|
34
|
+
*commonPrefixSearch(key) {
|
|
35
|
+
let index = 0;
|
|
36
|
+
let offset = 0;
|
|
37
|
+
while (index != -1) {
|
|
38
|
+
const lastIndex = index;
|
|
39
|
+
if (offset == key.length) {
|
|
40
|
+
index = -1;
|
|
41
|
+
}
|
|
42
|
+
else {
|
|
43
|
+
const c = this.charConverter(key.charCodeAt(offset));
|
|
44
|
+
index = this.traverse(index, c);
|
|
45
|
+
offset++;
|
|
46
|
+
}
|
|
47
|
+
yield lastIndex;
|
|
48
|
+
}
|
|
49
|
+
}
|
|
50
|
+
*predictiveSearch(key) {
|
|
51
|
+
const n = this.lookup(key);
|
|
52
|
+
if (n == -1) {
|
|
53
|
+
return;
|
|
54
|
+
}
|
|
55
|
+
yield* this.visitRecursive(n);
|
|
56
|
+
}
|
|
57
|
+
*visitRecursive(n) {
|
|
58
|
+
yield n;
|
|
59
|
+
for (let i = 0; i < this.charSize; i++) {
|
|
60
|
+
const m = this.base[n] + i + 1;
|
|
61
|
+
if (m >= this.check.length) {
|
|
62
|
+
return;
|
|
63
|
+
}
|
|
64
|
+
if (this.check[m] == n) {
|
|
65
|
+
yield* this.visitRecursive(m);
|
|
66
|
+
}
|
|
67
|
+
}
|
|
68
|
+
}
|
|
69
|
+
}
|
|
70
|
+
//# sourceMappingURL=DoubleArray.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"DoubleArray.js","sourceRoot":"","sources":["../src/DoubleArray.ts"],"names":[],"mappings":"AAAA,MAAM,OAAO,WAAW;IAMpB,YAAY,IAAgB,EAAE,KAAiB,EAAE,aAAoC,EAAE,QAAgB;QACnG,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC;QACjB,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;QACnB,IAAI,CAAC,aAAa,GAAG,aAAa,CAAC;QACnC,IAAI,CAAC,QAAQ,GAAG,QAAQ,CAAC;IAC7B,CAAC;IAEM,QAAQ,CAAC,CAAS,EAAE,CAAS;QAChC,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;QAC3B,IAAI,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE;YACpB,OAAO,CAAC,CAAC;SACZ;aAAM;YACH,OAAO,CAAC,CAAC,CAAC;SACb;IACL,CAAC;IAEM,MAAM,CAAC,GAAW;QACrB,IAAI,GAAG,CAAC,MAAM,IAAI,CAAC,EAAE;YACjB,OAAO,CAAC,CAAC;SACZ;QACD,IAAI,CAAC,GAAG,CAAC,CAAC;QACV,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACjC,MAAM,CAAC,GAAG,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAA;YAC/C,IAAI,CAAC,GAAG,CAAC,EAAE;gBACP,MAAM,IAAI,KAAK,EAAE,CAAC;aACrB;YACD,CAAC,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;YACxB,IAAI,CAAC,IAAI,CAAC,CAAC,EAAE;gBACT,OAAO,CAAC,CAAC,CAAC;aACb;SACJ;QACD,OAAO,CAAC,CAAC;IACb,CAAC;IAEM,CAAC,kBAAkB,CAAC,GAAW;QAClC,IAAI,KAAK,GAAG,CAAC,CAAC;QACd,IAAI,MAAM,GAAG,CAAC,CAAC;QACf,OAAO,KAAK,IAAI,CAAC,CAAC,EAAE;YAChB,MAAM,SAAS,GAAG,KAAK,CAAC;YACxB,IAAI,MAAM,IAAI,GAAG,CAAC,MAAM,EAAE;gBACtB,KAAK,GAAG,CAAC,CAAC,CAAC;aACd;iBAAM;gBACH,MAAM,CAAC,GAAG,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC;gBACrD,KAAK,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC,CAAC;gBAChC,MAAM,EAAE,CAAC;aACZ;YACD,MAAM,SAAS,CAAC;SACnB;IACL,CAAC;IAEM,CAAC,gBAAgB,CAAC,GAAW;QAChC,MAAM,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAC3B,IAAI,CAAC,IAAI,CAAC,CAAC,EAAE;YACT,OAAO;SACV;QACD,KAAM,CAAC,CAAA,IAAI,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC;IAClC,CAAC;IAEM,CAAC,cAAc,CAAC,CAAS;QAC5B,MAAM,CAAC,CAAC;QACR,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC,EAAE,EAAE;YACpC,MAAM,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YAC/B,IAAI,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE;gBACxB,OAAO;aACV;YACD,IAAI,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE;gBACpB,KAAM,CAAC,CAAA,IAAI,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC;aACjC;SACJ;IACL,CAAC;CAEJ"}
|
|
@@ -0,0 +1,13 @@
|
|
|
1
|
+
import { BitVector } from "./BitVector";
|
|
2
|
+
export declare class LOUDSTrie {
|
|
3
|
+
bitVector: BitVector;
|
|
4
|
+
edges: Uint16Array;
|
|
5
|
+
constructor(bitVector: BitVector, edges: Uint16Array);
|
|
6
|
+
reverseLookup(index: number): string;
|
|
7
|
+
parent(x: number): number;
|
|
8
|
+
firstChild(x: number): number;
|
|
9
|
+
traverse(index: number, c: number): number;
|
|
10
|
+
lookup(key: string): number;
|
|
11
|
+
predictiveSearch(index: number): IterableIterator<number>;
|
|
12
|
+
size(): number;
|
|
13
|
+
}
|
package/esm/LOUDSTrie.js
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
1
|
+
import { binarySearchUint16 } from "./utils";
|
|
2
|
+
export class LOUDSTrie {
|
|
3
|
+
constructor(bitVector, edges) {
|
|
4
|
+
this.bitVector = bitVector;
|
|
5
|
+
this.edges = edges;
|
|
6
|
+
}
|
|
7
|
+
reverseLookup(index) {
|
|
8
|
+
if (index <= 0 || this.edges.length <= index) {
|
|
9
|
+
throw new RangeError();
|
|
10
|
+
}
|
|
11
|
+
let sb = new Array();
|
|
12
|
+
while (index > 1) {
|
|
13
|
+
sb.push(this.edges[index]);
|
|
14
|
+
index = this.parent(index);
|
|
15
|
+
}
|
|
16
|
+
return sb.reverse().map(v => String.fromCharCode(v)).join('');
|
|
17
|
+
}
|
|
18
|
+
parent(x) {
|
|
19
|
+
return this.bitVector.rank(this.bitVector.select(x, true), false);
|
|
20
|
+
}
|
|
21
|
+
firstChild(x) {
|
|
22
|
+
let y = this.bitVector.select(x, false) + 1;
|
|
23
|
+
if (this.bitVector.get(y)) {
|
|
24
|
+
return this.bitVector.rank(y, true) + 1;
|
|
25
|
+
}
|
|
26
|
+
else {
|
|
27
|
+
return -1;
|
|
28
|
+
}
|
|
29
|
+
}
|
|
30
|
+
traverse(index, c) {
|
|
31
|
+
let firstChild = this.firstChild(index);
|
|
32
|
+
if (firstChild == -1) {
|
|
33
|
+
return -1;
|
|
34
|
+
}
|
|
35
|
+
let childStartBit = this.bitVector.select(firstChild, true);
|
|
36
|
+
let childEndBit = this.bitVector.nextClearBit(childStartBit);
|
|
37
|
+
let childSize = childEndBit - childStartBit;
|
|
38
|
+
let result = binarySearchUint16(this.edges, firstChild, firstChild + childSize, c);
|
|
39
|
+
return result >= 0 ? result : -1;
|
|
40
|
+
}
|
|
41
|
+
lookup(key) {
|
|
42
|
+
let nodeIndex = 1;
|
|
43
|
+
for (let i = 0; i < key.length; i++) {
|
|
44
|
+
let c = key.charCodeAt(i);
|
|
45
|
+
nodeIndex = this.traverse(nodeIndex, c);
|
|
46
|
+
if (nodeIndex == -1) {
|
|
47
|
+
break;
|
|
48
|
+
}
|
|
49
|
+
}
|
|
50
|
+
return (nodeIndex >= 0) ? nodeIndex : -1;
|
|
51
|
+
}
|
|
52
|
+
*predictiveSearch(index) {
|
|
53
|
+
let lower = index;
|
|
54
|
+
let upper = index + 1;
|
|
55
|
+
while (upper - lower > 0) {
|
|
56
|
+
for (let i = lower; i < upper; i++) {
|
|
57
|
+
yield i;
|
|
58
|
+
}
|
|
59
|
+
lower = this.bitVector.rank(this.bitVector.select(lower, false) + 1, true) + 1;
|
|
60
|
+
upper = this.bitVector.rank(this.bitVector.select(upper, false) + 1, true) + 1;
|
|
61
|
+
}
|
|
62
|
+
}
|
|
63
|
+
size() {
|
|
64
|
+
return this.edges.length - 2;
|
|
65
|
+
}
|
|
66
|
+
}
|
|
67
|
+
//# sourceMappingURL=LOUDSTrie.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"LOUDSTrie.js","sourceRoot":"","sources":["../src/LOUDSTrie.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,MAAM,SAAS,CAAC;AAG7C,MAAM,OAAO,SAAS;IAIlB,YAAY,SAAoB,EAAE,KAAkB;QAChD,IAAI,CAAC,SAAS,GAAG,SAAS,CAAC;QAC3B,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC;IACvB,CAAC;IAED,aAAa,CAAC,KAAa;QACvB,IAAI,KAAK,IAAI,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,KAAK,EAAE;YAC1C,MAAM,IAAI,UAAU,EAAE,CAAC;SAC1B;QACD,IAAI,EAAE,GAAG,IAAI,KAAK,EAAU,CAAC;QAC7B,OAAO,KAAK,GAAG,CAAC,EAAE;YACd,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC;YAC3B,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;SAC9B;QACD,OAAO,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,MAAM,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAClE,CAAC;IAED,MAAM,CAAC,CAAS;QACZ,OAAO,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,EAAE,IAAI,CAAC,EAAE,KAAK,CAAC,CAAC;IACtE,CAAC;IAED,UAAU,CAAC,CAAS;QAChB,IAAI,CAAC,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC;QAC5C,IAAI,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE;YACvB,OAAO,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC;SAC3C;aAAM;YACH,OAAO,CAAC,CAAC,CAAC;SACb;IACL,CAAC;IAED,QAAQ,CAAC,KAAa,EAAE,CAAS;QAC7B,IAAI,UAAU,GAAG,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;QACxC,IAAI,UAAU,IAAI,CAAC,CAAC,EAAE;YAClB,OAAO,CAAC,CAAC,CAAC;SACb;QACD,IAAI,aAAa,GAAG,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,UAAU,EAAE,IAAI,CAAC,CAAC;QAC5D,IAAI,WAAW,GAAG,IAAI,CAAC,SAAS,CAAC,YAAY,CAAC,aAAa,CAAC,CAAC;QAC7D,IAAI,SAAS,GAAG,WAAW,GAAG,aAAa,CAAC;QAC5C,IAAI,MAAM,GAAG,kBAAkB,CAAC,IAAI,CAAC,KAAK,EAAE,UAAU,EAAE,UAAU,GAAG,SAAS,EAAE,CAAC,CAAC,CAAC;QACnF,OAAO,MAAM,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACrC,CAAC;IAED,MAAM,CAAC,GAAW;QACd,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACjC,IAAI,CAAC,GAAG,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;YAC1B,SAAS,GAAG,IAAI,CAAC,QAAQ,CAAC,SAAS,EAAE,CAAC,CAAC,CAAC;YACxC,IAAI,SAAS,IAAI,CAAC,CAAC,EAAE;gBACjB,MAAM;aACT;SACJ;QACD,OAAO,CAAC,SAAS,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7C,CAAC;IAED,CAAC,gBAAgB,CAAC,KAAa;QAC3B,IAAI,KAAK,GAAG,KAAK,CAAC;QAClB,IAAI,KAAK,GAAG,KAAK,GAAG,CAAC,CAAC;QACtB,OAAO,KAAK,GAAG,KAAK,GAAG,CAAC,EAAE;YACtB,KAAK,IAAI,CAAC,GAAG,KAAK,EAAE,CAAC,GAAG,KAAK,EAAE,CAAC,EAAE,EAAE;gBAChC,MAAM,CAAC,CAAC;aACX;YACD,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC;YAC/E,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC;SAClF;IACL,CAAC;IAED,IAAI;QACA,OAAO,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC;IACjC,CAAC;CACJ"}
|
|
@@ -0,0 +1,69 @@
|
|
|
1
|
+
import { LOUDSTrie } from "./LOUDSTrie";
|
|
2
|
+
import { BitVector } from "./BitVector";
|
|
3
|
+
import { BitList } from "./BitList";
|
|
4
|
+
export class LOUDSTrieBuilder {
|
|
5
|
+
static build(keys) {
|
|
6
|
+
for (let i = 0; i < keys.length; i++) {
|
|
7
|
+
if (keys[i] == null) {
|
|
8
|
+
throw new Error();
|
|
9
|
+
}
|
|
10
|
+
if (i > 0 && keys[i - 1] > keys[i]) {
|
|
11
|
+
throw new Error();
|
|
12
|
+
}
|
|
13
|
+
}
|
|
14
|
+
const nodes = new Uint32Array(keys.length);
|
|
15
|
+
for (let i = 0; i < nodes.length; i++) {
|
|
16
|
+
nodes[i] = 1;
|
|
17
|
+
}
|
|
18
|
+
let cursor = 0;
|
|
19
|
+
let currentNode = 1;
|
|
20
|
+
let edges = " ";
|
|
21
|
+
const louds = new BitList();
|
|
22
|
+
louds.add(true);
|
|
23
|
+
while (true) {
|
|
24
|
+
let lastChar = 0;
|
|
25
|
+
let lastParent = 0;
|
|
26
|
+
let restKeys = 0;
|
|
27
|
+
for (let i = 0; i < keys.length; i++) {
|
|
28
|
+
if (keys[i].length < cursor) {
|
|
29
|
+
continue;
|
|
30
|
+
}
|
|
31
|
+
if (keys[i].length == cursor) {
|
|
32
|
+
louds.add(false);
|
|
33
|
+
lastParent = nodes[i];
|
|
34
|
+
lastChar = 0;
|
|
35
|
+
continue;
|
|
36
|
+
}
|
|
37
|
+
const currentChar = keys[i].charCodeAt(cursor);
|
|
38
|
+
const currentParent = nodes[i];
|
|
39
|
+
if (lastParent != currentParent) {
|
|
40
|
+
louds.add(false);
|
|
41
|
+
louds.add(true);
|
|
42
|
+
edges += String.fromCharCode(currentChar);
|
|
43
|
+
currentNode = currentNode + 1;
|
|
44
|
+
}
|
|
45
|
+
else if (lastChar != currentChar) {
|
|
46
|
+
louds.add(true);
|
|
47
|
+
edges += String.fromCharCode(currentChar);
|
|
48
|
+
currentNode = currentNode + 1;
|
|
49
|
+
}
|
|
50
|
+
nodes[i] = currentNode;
|
|
51
|
+
lastChar = currentChar;
|
|
52
|
+
lastParent = currentParent;
|
|
53
|
+
restKeys++;
|
|
54
|
+
}
|
|
55
|
+
if (restKeys == 0) {
|
|
56
|
+
break;
|
|
57
|
+
}
|
|
58
|
+
cursor++;
|
|
59
|
+
}
|
|
60
|
+
const bitVectorWords = new Uint32Array(louds.words.buffer, 0, ((louds.size + 63) >> 6) * 2);
|
|
61
|
+
const bitVector = new BitVector(bitVectorWords, louds.size);
|
|
62
|
+
const uint16Edges = new Uint16Array(edges.length);
|
|
63
|
+
for (let i = 0; i < edges.length; i++) {
|
|
64
|
+
uint16Edges[i] = edges.charCodeAt(i);
|
|
65
|
+
}
|
|
66
|
+
return [new LOUDSTrie(bitVector, uint16Edges), nodes];
|
|
67
|
+
}
|
|
68
|
+
}
|
|
69
|
+
//# sourceMappingURL=LOUDSTrieBuilder.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"LOUDSTrieBuilder.js","sourceRoot":"","sources":["../src/LOUDSTrieBuilder.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AACxC,OAAO,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AACxC,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAEpC,MAAM,OAAO,gBAAgB;IACzB,MAAM,CAAC,KAAK,CAAC,IAAc;QACvB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YAClC,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,IAAI,EAAE;gBACjB,MAAM,IAAI,KAAK,EAAE,CAAC;aACrB;YACD,IAAI,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,EAAE;gBAChC,MAAM,IAAI,KAAK,EAAE,CAAC;aACrB;SACJ;QACD,MAAM,KAAK,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QAC1C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YACnC,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;SAChB;QACD,IAAI,MAAM,GAAG,CAAC,CAAC;QACf,IAAI,WAAW,GAAG,CAAC,CAAC;QACpB,IAAI,KAAK,GAAG,IAAI,CAAC;QACjB,MAAM,KAAK,GAAG,IAAI,OAAO,EAAE,CAAC;QAC5B,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAChB,OAAO,IAAI,EAAE;YACT,IAAI,QAAQ,GAAG,CAAC,CAAC;YACjB,IAAI,UAAU,GAAG,CAAC,CAAC;YACnB,IAAI,QAAQ,GAAG,CAAC,CAAC;YACjB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;gBAClC,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,GAAG,MAAM,EAAE;oBACzB,SAAS;iBACZ;gBACD,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,IAAI,MAAM,EAAE;oBAC1B,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;oBACjB,UAAU,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;oBACtB,QAAQ,GAAG,CAAC,CAAC;oBACb,SAAS;iBACZ;gBACD,MAAM,WAAW,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;gBAC/C,MAAM,aAAa,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC/B,IAAI,UAAU,IAAI,aAAa,EAAE;oBAC7B,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC;oBACjB,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;oBAChB,KAAK,IAAI,MAAM,CAAC,YAAY,CAAC,WAAW,CAAC,CAAC;oBAC1C,WAAW,GAAG,WAAW,GAAG,CAAC,CAAC;iBACjC;qBAAM,IAAI,QAAQ,IAAI,WAAW,EAAE;oBAChC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;oBAChB,KAAK,IAAI,MAAM,CAAC,YAAY,CAAC,WAAW,CAAC,CAAC;oBAC1C,WAAW,GAAG,WAAW,GAAG,CAAC,CAAC;iBACjC;gBACD,KAAK,CAAC,CAAC,CAAC,GAAG,WAAW,CAAC;gBACvB,QAAQ,GAAG,WAAW,CAAC;gBACvB,UAAU,GAAG,aAAa,CAAC;gBAC3B,QAAQ,EAAE,CAAC;aACd;YACD,IAAI,QAAQ,IAAI,CAAC,EAAE;gBACf,MAAM;aACT;YACD,MAAM,EAAE,CAAC;SACZ;QACD,MAAM,cAAc,GAAG,IAAI,WAAW,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,CAAC,CAAC,KAAK,CAAC,IAAI,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QAC5F,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,cAAc,EAAE,KAAK,CAAC,IAAI,CAAC,CAAC;QAC5D,MAAM,WAAW,GAAG,IAAI,WAAW,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QAClD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAE,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE;YAClC,WAAW,CAAC,CAAC,CAAC,GAAG,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;SACxC;QACD,OAAO,CAAC,IAAI,SAAS,CAAC,SAAS,EAAE,WAAW,CAAC,EAAE,KAAK,CAAC,CAAC;IAC1D,CAAC;CACJ"}
|
package/esm/Migemo.d.ts
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
1
|
+
import { CompactDictionary } from "./CompactDictionary";
|
|
2
|
+
import { RomajiProcessor2 } from "./RomajiProcessor2";
|
|
3
|
+
export declare class Migemo {
|
|
4
|
+
dict: CompactDictionary | null;
|
|
5
|
+
rxop: Array<string> | null;
|
|
6
|
+
processor: RomajiProcessor2;
|
|
7
|
+
constructor();
|
|
8
|
+
queryAWord(word: string): string;
|
|
9
|
+
query(word: string): string;
|
|
10
|
+
setDict(dict: CompactDictionary | null): void;
|
|
11
|
+
setRxop(rxop: [string] | null): void;
|
|
12
|
+
parseQuery(query: string): IterableIterator<string>;
|
|
13
|
+
}
|