1 | 'use strict';
|
2 | Object.defineProperty(exports, "__esModule", { value: true });
|
3 | exports.init = exports.ZhuyinTokenizer = void 0;
|
4 | const mod_1 = require("../mod");
|
5 |
|
6 |
|
7 |
|
8 | class ZhuyinTokenizer extends mod_1.SubSModuleTokenizer {
|
9 | constructor() {
|
10 | super(...arguments);
|
11 | this.name = 'ZhuyinTokenizer';
|
12 | }
|
13 | _cache(...argv) {
|
14 | super._cache(...argv);
|
15 | }
|
16 | split(words) {
|
17 | return this._splitUnset(words, this.splitZhuyin);
|
18 | }
|
19 | splitZhuyin(text, cur) {
|
20 | let ret = [];
|
21 | let self = this;
|
22 | let _r = /[\u31A0-\u31BA\u3105-\u312E]/u;
|
23 | if (!_r.test(text)) {
|
24 | return null;
|
25 | }
|
26 | text
|
27 | .split(/([\u31A0-\u31BA\u3105-\u312E]+)/u)
|
28 | .forEach(function (w, i) {
|
29 | if (w !== '') {
|
30 | if (_r.test(w)) {
|
31 | ret.push(self.debugToken({
|
32 | w,
|
33 | }, {
|
34 | [self.name]: true,
|
35 | }, true));
|
36 | }
|
37 | else {
|
38 | ret.push({
|
39 | w,
|
40 | });
|
41 | }
|
42 | }
|
43 | });
|
44 | return ret.length ? ret : null;
|
45 | }
|
46 | }
|
47 | exports.ZhuyinTokenizer = ZhuyinTokenizer;
|
48 | exports.init = ZhuyinTokenizer.init.bind(ZhuyinTokenizer);
|
49 | exports.default = ZhuyinTokenizer;
|
50 |
|
\ | No newline at end of file |