mod:修改抓取文件策略

Author: zhenbushidashen

scripts/LeetCodeProvider.js

@@ -7,7 +7,7 @@ const {
   PROBLEMS_URL,
   QUESTION_DOM_SELECTOR,
   BASE_MARKDWON_DOWNLOAD_URL,
-  ENGLISH_MARKDOWN_SIGN,
+  CRAWL_IGNORE_SUFFIX,
 } = require("./constants");
 
 module.exports = LeetCodeProvider = {
@@ -24,7 +24,7 @@ module.exports = LeetCodeProvider = {
         /** 
          *  由于QUESTION_DOM_SELECTOR 所选择的结构包含非问题���签，获取title会是undefined，在此需将其过滤掉
         */
-        return titles.filter(Boolean).filter((name) => !name.endsWith(ENGLISH_MARKDOWN_SIGN));
+        return titles.filter(Boolean).filter((name) => !CRAWL_IGNORE_SUFFIX.some(suffix => name.endsWith(suffix)));
       })
       .catch((error) => {
         Logger.error("获取问题列表失败", error);

scripts/constants.js

@@ -55,7 +55,7 @@ module.exports = {
   /**
    * 过滤英文文档末尾标识
    */
-  ENGLISH_MARKDOWN_SIGN: ".en.md",
+  CRAWL_IGNORE_SUFFIX: [".en.md", "-en.md"],
 
   /**
    * 爬虫抓取同一文件时的最大失败次数(多为网络原因导致, 切换到gitee源)

scripts/generateleetcode.js

@@ -4,15 +4,15 @@ const {
   SUPPORT_LANGUAGE,
   DB_JSON_OUTPUT_DIR,
   RAW_MARKDOWN_OUTPUT_DIR,
-  ENGLISH_MARKDOWN_SIGN
+  CRAWL_IGNORE_SUFFIX
 } = require("./constants");
 
 const genertateLeetcodeToJson = () => {
   console.time("genertateLeetcodeToJson");
 
   const rawMarkdowns = Utils.getDirsFileNameSync(
     RAW_MARKDOWN_OUTPUT_DIR
-  ).filter(name => !name.endsWith(ENGLISH_MARKDOWN_SIGN));
+  ).filter(name => !CRAWL_IGNORE_SUFFIX.some(suffix => name.endsWith(suffix)));
 
   rawMarkdowns.forEach(filename => {
     let languageResloved = [];