添加字幕搜索功能

src/util/pinyin_util.ts

@@ -0,0 +1,125 @@
+import pinyin from 'tiny-pinyin'
+import {uniq} from 'lodash-es'
+
+/**
+ * pinyin的返回结果
+ */
+interface Ret {
+  type: 1 | 2 | 3
+  source: string
+  target: string
+}
+
+interface Phase {
+  pinyin: boolean
+  list: Ret[]
+}
+
+/**
+ * 获取Phase列表(中英文分离列表)
+ */
+export const getPhases = (str: string) => {
+  const rets = pinyin.parse(str)
+
+  const phases: Phase[] = []
+  let curPinyin_ = false
+  let curPhase_: Ret[] = []
+  const addCurrentPhase = () => {
+    if (curPhase_.length > 0) {
+      phases.push({
+        pinyin: curPinyin_,
+        list: curPhase_,
+      })
+    }
+  }
+
+  // 遍历rets
+  for (const ret of rets) {
+    const newPinyin = ret.type === 2
+    // 如果跟旧的pinyin类型不同，先保存旧的
+    if (newPinyin !== curPinyin_) {
+      addCurrentPhase()
+      // 重置
+      curPinyin_ = newPinyin
+      curPhase_ = []
+    }
+    // 添加新的
+    curPhase_.push(ret)
+  }
+  // 最后一个
+  addCurrentPhase()
+
+  return phases
+}
+
+/**
+ * 获取原子字符列表，如 tool tab 汉 字
+ */
+export const getAtoms = (str: string) => {
+  const phases = getPhases(str)
+
+  const atoms = []
+  for (const phase of phases) {
+    if (phase.pinyin) { // all words
+      atoms.push(...phase.list.map(e => e.source).filter(e => e))
+    } else { // split
+      atoms.push(...(phase.list.map((e: any) => e.source).join('').match(/\w+/g)??[]).filter((e: string) => e))
+    }
+  }
+
+  return atoms
+}
+
+const fixStrs = (atoms: string[]) => {
+  // 小写
+  atoms = atoms.map(e => e.toLowerCase())
+
+  // 去重
+  atoms = uniq(atoms)
+
+  // 返回
+  return atoms
+}
+
+export const getWords = (str: string) => {
+  // 获取全部原子字符
+  const atoms = getAtoms(str)
+  // fix
+  return fixStrs(atoms)
+}
+
+/**
+ * 我的世界Minecraft => ['wodeshijie', 'deshijie', 'shijie', 'jie'] + ['wdsj', 'dsj', 'sj', 'j']
+ *
+ * 1. only handle pinyin, other is ignored
+ */
+export const getWordsPinyin = (str: string) => {
+  let result: string[] = []
+
+  for (const phase of getPhases(str)) {
+    // only handle pinyin
+    if (phase.pinyin) { // 我的世界
+      // 获取全部原子字符
+      // 我的世界 => [我, 的, 世, 界]
+      const atoms: string[] = []
+      atoms.push(...phase.list.map(e => e.source).filter(e => e))
+      // 获取全部子串
+      // [我, 的, 世, 界] => [我的世界, 的世界, 世界, 界]
+      const allSubStr = []
+      for (let i = 0; i < atoms.length; i++) {
+        allSubStr.push(atoms.slice(i).join(''))
+      }
+      // pinyin version
+      const pinyinList = allSubStr.map((e: string) => pinyin.convertToPinyin(e))
+      result.push(...pinyinList)
+      // pinyin first version
+      const pinyinFirstList = allSubStr.map((e: string) => pinyin.parse(e).map((e: any) => e.type === 2?e.target[0]:null).filter(e => !!e).join(''))
+      result.push(...pinyinFirstList)
+    }
+  }
+
+  // fix
+  result = fixStrs(result)
+
+  return result
+}

src/util/search.ts

@@ -0,0 +1,65 @@
+import * as JsSearch from 'js-search'
+import {uniq} from 'lodash-es'
+import {getWords, getWordsPinyin} from './pinyin_util'
+
+const tokenize = (maxLength: number, content: string, options?: SearchOptions) => {
+  const result: string[] = []
+
+  // 最大长度
+  if (content.length > maxLength) {
+    content = content.substring(0, maxLength)
+  }
+  result.push(...getWords(content))
+  // check cn
+  if (options?.cnSearchEnabled) {
+    result.push(...getWordsPinyin(content))
+  }
+
+  // console.debug('[Search] tokenize:', str, '=>', result)
+
+  return uniq(result)
+}
+
+export interface SearchOptions {
+  cnSearchEnabled?: boolean
+}
+
+export const Search = (uidFieldName: string, index: string, maxLength: number, options?: SearchOptions) => {
+  let searchRef: JsSearch.Search | undefined// 搜索器
+
+  /**
+   * 重置索引
+   */
+  const reset = (documents?: Object[]) => {
+    // 搜索器
+    searchRef = new JsSearch.Search(uidFieldName)
+    searchRef.tokenizer = {
+      tokenize: (str) => {
+        return tokenize(maxLength, str, options)
+      }
+    }
+    searchRef.addIndex(index)
+
+    // 检测添加文档
+    if (documents != null) {
+      searchRef.addDocuments(documents)
+    }
+  }
+
+  /**
+   * 添加文档
+   */
+  const add = (document: Object) => {
+    searchRef?.addDocument(document)
+  }
+
+  /**
+   * 搜索
+   * @return 未去重
+   */
+  const search = (text: string) => {
+    return searchRef?.search(text.toLowerCase())
+  }
+
+  return {reset, add, search}
+}