feat: 补全通用汉字规范表字符

lib/common/utils.ts

@@ -28,15 +28,6 @@ export function splitString(text: string): string[] {
   return result;
 }
 
-// todo: 双 unicode 编码字符的适配
-export function isZhChar(char: string) {
-  if (typeof char !== "string") {
-    return false;
-  }
-  let code = char.charCodeAt(0);
-  return code >= 19968 && code <= 40869;
-}
-
 export class FastDictFactory {
   NumberDICT: string[];
   StringDICT: Map<string, string>;
@@ -55,10 +46,8 @@ export class FastDictFactory {
     }
   }
 
-  set(word: string | number, pinyin: string) {
-    if (typeof word === 'number') {
-      this.NumberDICT[word] = pinyin;
-    } else if (word.length > 1) {
+  set(word: string, pinyin: string) {
+    if (word.length > 1) {
       this.StringDICT.set(word, pinyin);
     } else {
       const code = word.charCodeAt(0);

lib/core/pinyin/middlewares.ts

@@ -1,9 +1,5 @@
-import { stringLength, isZhChar } from "@/common/utils";
+import { stringLength } from "@/common/utils";
 import type { SingleWordResult } from "../../common/type";
-import {
-  DoubleUnicodePrefixReg,
-  DoubleUnicodeSuffixReg,
-} from "@/common/constant";
 import { getAllPinyin, getMultiplePinyin } from "./handle";
 import { CompleteOptions } from "./index";
 import {
@@ -14,6 +10,7 @@ import {
   getPinyinWithoutTone,
   getPinyinWithNum,
 } from "./handle";
+import DICT1 from "@/data/dict1";
 
 // 验证输入是否为字符串
 export const validateType = (word: unknown) => {
@@ -202,7 +199,7 @@ export const middlewareType = (
         num: Number(getNumOfTone(item.originPinyin)),
         isZh: item.isZh,
         polyphonic,
-        inZhRange: isZhChar(item.origin),
+        inZhRange: !!DICT1.get(item.origin),
       };
     });
   }

lib/core/polyphonic/index.ts

@@ -14,7 +14,7 @@ import {
   getFinalParts,
 } from '@/core/pinyin/handle';
 import { getCustomPolyphonicDict } from '../custom';
-import { isZhChar, splitString } from '@/common/utils';
+import { splitString } from '@/common/utils';
 
 interface BasicOptions {
   /**
@@ -255,7 +255,7 @@ export const handleType = (
         finalTail: tail,
         num: Number(getNumOfTone(item.originPinyin)),
         isZh: item.isZh,
-        inZhRange: isZhChar(item.origin),
+        inZhRange: !!DICT1.get(item.origin),
       };
     });
   }

lib/data/special.ts

@@ -5,7 +5,7 @@ import {
   getNumOfTone,
   getPinyinWithoutTone,
 } from '@/core/pinyin/handle';
-import { isZhChar } from '@/common/utils';
+import DICT1 from './dict1';
 
 export const InitialList = [
   'zh',
@@ -225,7 +225,7 @@ export function processToneSandhi(cur: string, pre: string, next: string) {
 
 // 处理「了」字的变调
 export function processToneSandhiLiao(cur: string, pre: string) {
-  if (cur === '了' && !isZhChar(pre)) {
+  if (cur === '了' && (!pre || !DICT1.get(pre))) {
     return 'liǎo';
   }
 }

test/double-unicode.test.js

@@ -28,4 +28,10 @@ describe('double unicode', () => {
     });
     expect(result5).to.be.equal('cè shì a𧒽𧒽a cè shì a𧒽𧒽a cè shì');
   });
+
+  it('[double unicode]dp consecutive', () => {
+    const result4 = pinyin('测试𬭬𬭬测试𬭬测试');
+    expect(result4).to.be.equal('cè shì huì huì cè shì huì cè shì');
+  });
+  
 });

types/common/utils.d.ts

@@ -1,11 +1,10 @@
 export declare function stringLength(text: string): number;
 export declare function splitString(text: string): string[];
-export declare function isZhChar(char: string): boolean;
 export declare class FastDictFactory {
     NumberDICT: string[];
     StringDICT: Map<string, string>;
     constructor();
     get(word: string): string;
-    set(word: string | number, pinyin: string): void;
+    set(word: string, pinyin: string): void;
     clear(): void;
 }