fix: #927 修复 CSDN 和知乎公式解析不正常的问题

terwer · Dec 3, 2023 · aee7081 · aee7081
1 parent abc9095
commit aee7081
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 6 deletions.
diff --git a/src/adaptors/web/csdn/csdnUtils.ts b/src/adaptors/web/csdn/csdnUtils.ts
@@ -87,11 +87,11 @@ class CsdnUtils {
 
   public static processCsdnMath(html: string): string {
     // 使用Cheerio加载HTML
-    const $ = cheerio.load(html)
+    const $ = cheerio.load(html, { xmlMode: true })
 
     // 处理两个$符号包裹的公式
     const doubleDollarRegex = /\$\$([^$]+)\$\$/g
-    $("*").each((index, element) => {
+    $("*:not(pre)").each((index, element) => {
       const content = $(element).html()
       const newContent = content.replace(doubleDollarRegex, (match, mathContent) => {
         const mathHtml = KatexUtils.renderToString(mathContent)
@@ -102,7 +102,7 @@ class CsdnUtils {
 
     // 处理一个$符号包裹的公式
     const singleDollarRegex = /\$([^$]+)\$/g
-    $("*").each((index, element) => {
+    $("*:not(pre)").each((index, element) => {
       const content = $(element).html()
       const newContent = content.replace(singleDollarRegex, (match, mathContent) => {
         const mathHtml = KatexUtils.renderToString(mathContent)

diff --git a/src/adaptors/web/zhihu/zhihuUtils.ts b/src/adaptors/web/zhihu/zhihuUtils.ts
@@ -62,7 +62,7 @@ class ZhihuUtils {
 
   public static processZHMath(html: string): string {
     // 使用Cheerio加载HTML
-    const $ = cheerio.load(html)
+    const $ = cheerio.load(html, { xmlMode: true })
 
     // // 选择所有带有类名"language-math"的<span>元素
     // $("span.language-math").each((index, element) => {
@@ -92,7 +92,7 @@ class ZhihuUtils {
 
     // 处理两个$符号包裹的公式
     const doubleDollarRegex = /\$\$([^$]+)\$\$/g
-    $("*").each((index, element) => {
+    $("*:not(pre)").each((index, element) => {
       const content = $(element).html()
       const newContent = content.replace(doubleDollarRegex, (match, mathContent) => {
         return `<img eeimg="1" src="//www.zhihu.com/equation?tex=${encodeURIComponent(mathContent)}" 
@@ -103,7 +103,7 @@ class ZhihuUtils {
 
     // 处理一个$符号包裹的公式
     const singleDollarRegex = /\$([^$]+)\$/g
-    $("*").each((index, element) => {
+    $("*:not(pre)").each((index, element) => {
       const content = $(element).html()
       const newContent = content.replace(singleDollarRegex, (match, mathContent) => {
         return `<img eeimg="1" src="//www.zhihu.com/equation?tex=${encodeURIComponent(mathContent)}" 

diff --git a/src/utils/KatexUtils.ts b/src/utils/KatexUtils.ts
@@ -24,6 +24,7 @@
  */
 
 import katex from "katex"
+import { createAppLogger } from "~/src/utils/appLogger.ts"
 
 /**
  * 公式渲染
@@ -32,12 +33,15 @@ import katex from "katex"
  * @since 1.18.6
  */
 class KatexUtils {
+  private static logger = createAppLogger("katex-utils")
+
   /**
    * 获得要渲染 KaTeX 表达式的 HTML
    *
    * @param mathExpression katex
    */
   public static renderToString(mathExpression: string) {
+    this.logger.debug("准备处理 Katex =>", { mathExpression: mathExpression })
     return katex.renderToString(mathExpression)
   }
 }