1 | ;
|
2 | Object.defineProperty(exports, "__esModule", { value: true });
|
3 | exports.init = exports.WildcardTokenizer = void 0;
|
4 | /**
|
5 | * 通配符识别模块
|
6 | *
|
7 | * @author 老雷<leizongmin@gmail.com>
|
8 | */
|
9 | const mod_1 = require("../mod");
|
10 | class WildcardTokenizer extends mod_1.SubSModuleTokenizer {
|
11 | constructor() {
|
12 | super(...arguments);
|
13 | this.name = 'WildcardTokenizer';
|
14 | }
|
15 | _cache() {
|
16 | super._cache();
|
17 | this._TABLE = this.segment.getDict('WILDCARD');
|
18 | this._TABLE2 = this.segment.getDict('WILDCARD2');
|
19 | }
|
20 | /**
|
21 | * 对未识别的单词进行分词
|
22 | *
|
23 | * @param {array} words 单词数组
|
24 | * @return {array}
|
25 | */
|
26 | split(words) {
|
27 | //return this._splitUnknow(words, this.splitForeign);
|
28 | return this._splitUnknow(words, this.splitWildcard);
|
29 | }
|
30 | createWildcardToken(word, lasttype, attr) {
|
31 | let nw = this.createToken(word, true, attr);
|
32 | return nw;
|
33 | }
|
34 | splitWildcard(text, cur) {
|
35 | //const POSTAG = this._POSTAG;
|
36 | const TABLE = this._TABLE;
|
37 | let ret = [];
|
38 | let self = this;
|
39 | // 分离出已识别的单词
|
40 | let wordinfo = self.matchWord(text);
|
41 | if (wordinfo.length) {
|
42 | let lastc = 0;
|
43 | for (let ui = 0, bw; bw = wordinfo[ui]; ui++) {
|
44 | if (bw.c > lastc) {
|
45 | ret.push({
|
46 | w: text.substr(lastc, bw.c - lastc),
|
47 | });
|
48 | }
|
49 | let nw = self.createWildcardToken({
|
50 | w: bw.w,
|
51 | p: TABLE[bw.w.toLowerCase()].p,
|
52 | });
|
53 | ret.push(nw);
|
54 | lastc = bw.c + bw.w.length;
|
55 | }
|
56 | let lastword = wordinfo[wordinfo.length - 1];
|
57 | if (lastword.c + lastword.w.length < text.length) {
|
58 | ret.push({
|
59 | w: text.substr(lastword.c + lastword.w.length),
|
60 | });
|
61 | }
|
62 | }
|
63 | return ret.length ? ret : undefined;
|
64 | }
|
65 | /**
|
66 | * 匹配单词,返回相关信息
|
67 | *
|
68 | * @param {string} text 文本
|
69 | * @param {int} cur 开始位置
|
70 | * @return {array} 返回格式 {w: '单词', c: 开始位置}
|
71 | */
|
72 | matchWord(text, cur) {
|
73 | //const POSTAG = this._POSTAG;
|
74 | const TABLE = this._TABLE2;
|
75 | if (isNaN(cur))
|
76 | cur = 0;
|
77 | let ret = [];
|
78 | //let self = this;
|
79 | let s = false;
|
80 | // 匹配可能出现的单词,取长度最大的那个
|
81 | let lowertext = text.toLowerCase();
|
82 | while (cur < text.length) {
|
83 | let stopword = null;
|
84 | for (let i in TABLE) {
|
85 | if (lowertext.substr(cur, i) in TABLE[i]) {
|
86 | stopword = {
|
87 | w: text.substr(cur, i),
|
88 | c: cur,
|
89 | };
|
90 | }
|
91 | }
|
92 | if (stopword !== null) {
|
93 | ret.push(stopword);
|
94 | cur += stopword.w.length;
|
95 | }
|
96 | else {
|
97 | cur++;
|
98 | }
|
99 | }
|
100 | return ret;
|
101 | }
|
102 | }
|
103 | exports.WildcardTokenizer = WildcardTokenizer;
|
104 | exports.init = WildcardTokenizer.init.bind(WildcardTokenizer);
|
105 | exports.default = WildcardTokenizer;
|
106 | //# sourceMappingURL=data:application/json;base64,{"version":3,"file":"WildcardTokenizer.js","sourceRoot":"","sources":["WildcardTokenizer.ts"],"names":[],"mappings":"AAAA,YAAY,CAAC;;;AAEb;;;;GAIG;AACH,gCAA8E;AAO9E,MAAa,iBAAkB,SAAQ,yBAAmB;IAA1D;;QAGC,SAAI,GAAG,mBAAmB,CAAC;IA2H5B,CAAC;IAtHA,MAAM;QAEL,KAAK,CAAC,MAAM,EAAE,CAAC;QACf,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC;QAC/C,IAAI,CAAC,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,WAAW,CAAC,CAAC;IAClD,CAAC;IAED;;;;;OAKG;IACH,KAAK,CAAC,KAAc;QAEnB,qDAAqD;QACrD,OAAO,IAAI,CAAC,YAAY,CAAC,KAAK,EAAE,IAAI,CAAC,aAAa,CAAC,CAAC;IACrD,CAAC;IAED,mBAAmB,CAAC,IAAW,EAAE,QAAiB,EAAE,IAAqB;QAExE,IAAI,EAAE,GAAG,IAAI,CAAC,WAAW,CAAQ,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;QAEnD,OAAO,EAAE,CAAC;IACX,CAAC;IAED,aAAa,CAAC,IAAY,EAAE,GAAY;QAEvC,8BAA8B;QAC9B,MAAM,KAAK,GAAG,IAAI,CAAC,MAAM,CAAC;QAE1B,IAAI,GAAG,GAAY,EAAE,CAAC;QACtB,IAAI,IAAI,GAAG,IAAI,CAAC;QAEhB,YAAY;QACZ,IAAI,QAAQ,GAAG,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;QACpC,IAAI,QAAQ,CAAC,MAAM,EACnB;YACC,IAAI,KAAK,GAAG,CAAC,CAAC;YACd,KAAK,IAAI,EAAE,GAAG,CAAC,EAAE,EAAE,EAAE,EAAE,GAAG,QAAQ,CAAC,EAAE,CAAC,EAAE,EAAE,EAAE,EAC5C;gBACC,IAAI,EAAE,CAAC,CAAC,GAAG,KAAK,EAChB;oBACC,GAAG,CAAC,IAAI,CAAC;wBACR,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,GAAG,KAAK,CAAC;qBACnC,CAAC,CAAC;iBACH;gBAED,IAAI,EAAE,GAAG,IAAI,CAAC,mBAAmB,CAAC;oBACjC,CAAC,EAAE,EAAE,CAAC,CAAC;oBACP,CAAC,EAAE,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC;iBAC9B,CAAC,CAAC;gBAEH,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;gBAEb,KAAK,GAAG,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC;aAC3B;YAED,IAAI,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;YAC7C,IAAI,QAAQ,CAAC,CAAC,GAAG,QAAQ,CAAC,CAAC,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM,EAChD;gBACC,GAAG,CAAC,IAAI,CAAC;oBACR,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,GAAG,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC;iBAC9C,CAAC,CAAC;aACH;SACD;QAED,OAAO,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,CAAC;IACrC,CAAC;IAED;;;;;;OAMG;IACH,SAAS,CAAC,IAAY,EAAE,GAAY;QAEnC,8BAA8B;QAC9B,MAAM,KAAK,GAAG,IAAI,CAAC,OAAO,CAAC;QAE3B,IAAI,KAAK,CAAC,GAAG,CAAC;YAAE,GAAG,GAAG,CAAC,CAAC;QAExB,IAAI,GAAG,GAAY,EAAE,CAAC;QACtB,kBAAkB;QAElB,IAAI,CAAC,GAAG,KAAK,CAAC;QAEd,qBAAqB;QACrB,IAAI,SAAS,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;QAEnC,OAAO,GAAG,GAAG,IAAI,CAAC,MAAM,EACxB;YACC,IAAI,QAAQ,GAAU,IAAI,CAAC;YAC3B,KAAK,IAAI,CAAC,IAAI,KAAK,EACnB;gBACC,IAAI,SAAS,CAAC,MAAM,CAAC,GAAG,EAAE,CAAQ,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,EAC/C;oBACC,QAAQ,GAAG;wBACV,CAAC,EAAE,IAAI,CAAC,MAAM,CAAC,GAAG,EAAE,CAAQ,CAAC;wBAC7B,CAAC,EAAE,GAAG;qBACN,CAAC;iBACF;aACD;YACD,IAAI,QAAQ,KAAK,IAAI,EACrB;gBACC,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBACnB,GAAG,IAAI,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC;aACzB;iBAED;gBACC,GAAG,EAAE,CAAC;aACN;SACD;QACD,OAAO,GAAG,CAAC;IACZ,CAAC;CAED;AA9HD,8CA8HC;AAEY,QAAA,IAAI,GAAG,iBAAiB,CAAC,IAAI,CAAC,IAAI,CAAC,iBAAiB,CAA2C,CAAC;AAE7G,kBAAe,iBAAiB,CAAC","sourcesContent":["'use strict';\n\n/**\n * 通配符识别模块\n *\n * @author 老雷<leizongmin@gmail.com>\n */\nimport { SubSModule, SubSModuleTokenizer, ISubTokenizerCreate } from '../mod';\nimport { Segment, IWord, IDICT, IDICT2 } from '../Segment';\nimport { debugToken } from '../util/debug';\nimport UString from 'uni-string';\nimport { debug } from '../util';\nimport { IWordDebugInfo } from '../util/index';\n\nexport class WildcardTokenizer extends SubSModuleTokenizer\n{\n\n\tname = 'WildcardTokenizer';\n\n\tprotected _TABLE: IDICT<IWord>;\n\tprotected _TABLE2: IDICT2<IWord>;\n\n\t_cache()\n\t{\n\t\tsuper._cache();\n\t\tthis._TABLE = this.segment.getDict('WILDCARD');\n\t\tthis._TABLE2 = this.segment.getDict('WILDCARD2');\n\t}\n\n\t/**\n\t * 对未识别的单词进行分词\n\t *\n\t * @param {array} words 单词数组\n\t * @return {array}\n\t */\n\tsplit(words: IWord[]): IWord[]\n\t{\n\t\t//return this._splitUnknow(words, this.splitForeign);\n\t\treturn this._splitUnknow(words, this.splitWildcard);\n\t}\n\n\tcreateWildcardToken(word: IWord, lasttype?: number, attr?: IWordDebugInfo)\n\t{\n\t\tlet nw = this.createToken<IWord>(word, true, attr);\n\n\t\treturn nw;\n\t}\n\n\tsplitWildcard(text: string, cur?: number): IWord[]\n\t{\n\t\t//const POSTAG = this._POSTAG;\n\t\tconst TABLE = this._TABLE;\n\n\t\tlet ret: IWord[] = [];\n\t\tlet self = this;\n\n\t\t// 分离出已识别的单词\n\t\tlet wordinfo = self.matchWord(text);\n\t\tif (wordinfo.length)\n\t\t{\n\t\t\tlet lastc = 0;\n\t\t\tfor (let ui = 0, bw; bw = wordinfo[ui]; ui++)\n\t\t\t{\n\t\t\t\tif (bw.c > lastc)\n\t\t\t\t{\n\t\t\t\t\tret.push({\n\t\t\t\t\t\tw: text.substr(lastc, bw.c - lastc),\n\t\t\t\t\t});\n\t\t\t\t}\n\n\t\t\t\tlet nw = self.createWildcardToken({\n\t\t\t\t\tw: bw.w,\n\t\t\t\t\tp: TABLE[bw.w.toLowerCase()].p,\n\t\t\t\t});\n\n\t\t\t\tret.push(nw);\n\n\t\t\t\tlastc = bw.c + bw.w.length;\n\t\t\t}\n\n\t\t\tlet lastword = wordinfo[wordinfo.length - 1];\n\t\t\tif (lastword.c + lastword.w.length < text.length)\n\t\t\t{\n\t\t\t\tret.push({\n\t\t\t\t\tw: text.substr(lastword.c + lastword.w.length),\n\t\t\t\t});\n\t\t\t}\n\t\t}\n\n\t\treturn ret.length ? ret : undefined;\n\t}\n\n\t/**\n\t * 匹配单词，返回相关信息\n\t *\n\t * @param {string} text 文本\n\t * @param {int} cur 开始位置\n\t * @return {array}  返回格式   {w: '单词', c: 开始位置}\n\t */\n\tmatchWord(text: string, cur?: number)\n\t{\n\t\t//const POSTAG = this._POSTAG;\n\t\tconst TABLE = this._TABLE2;\n\n\t\tif (isNaN(cur)) cur = 0;\n\n\t\tlet ret: IWord[] = [];\n\t\t//let self = this;\n\n\t\tlet s = false;\n\n\t\t// 匹配可能出现的单词，取长度最大的那个\n\t\tlet lowertext = text.toLowerCase();\n\n\t\twhile (cur < text.length)\n\t\t{\n\t\t\tlet stopword: IWord = null;\n\t\t\tfor (let i in TABLE)\n\t\t\t{\n\t\t\t\tif (lowertext.substr(cur, i as any) in TABLE[i])\n\t\t\t\t{\n\t\t\t\t\tstopword = {\n\t\t\t\t\t\tw: text.substr(cur, i as any),\n\t\t\t\t\t\tc: cur,\n\t\t\t\t\t};\n\t\t\t\t}\n\t\t\t}\n\t\t\tif (stopword !== null)\n\t\t\t{\n\t\t\t\tret.push(stopword);\n\t\t\t\tcur += stopword.w.length;\n\t\t\t}\n\t\t\telse\n\t\t\t{\n\t\t\t\tcur++;\n\t\t\t}\n\t\t}\n\t\treturn ret;\n\t}\n\n}\n\nexport const init = WildcardTokenizer.init.bind(WildcardTokenizer) as ISubTokenizerCreate<WildcardTokenizer>;\n\nexport default WildcardTokenizer;\n"]} |
\ | No newline at end of file |