Merge branch 'master' into ru

pionxzh · Sep 9, 2024 · a45ee8f · a45ee8f
2 parents 12f8dfa + dd52ffc
commit a45ee8f
Show file tree

Hide file tree

Showing 6 changed files with 97 additions and 23 deletions.
diff --git a/.release-please-manifest.json b/.release-please-manifest.json
@@ -1 +1 @@
-{".":"2.24.5"}
+{".":"2.24.6"}
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,12 @@
 # Changelog
 
+## [2.24.6](https://github.com/pionxzh/chatgpt-exporter/compare/userscript-v2.24.5...userscript-v2.24.6) (2024-09-08)
+
+
+### Bug Fixes
+
+* fix unexpected latex formula escape ([30f0b62](https://github.com/pionxzh/chatgpt-exporter/commit/30f0b623b306aff6ec5a2278207cf2481a609b85))
+
 ## [2.24.5](https://github.com/pionxzh/chatgpt-exporter/compare/userscript-v2.24.4...userscript-v2.24.5) (2024-09-07)
 
 

diff --git a/dist/chatgpt.user.js b/dist/chatgpt.user.js
@@ -3,7 +3,7 @@
 // @name:zh-CN         ChatGPT Exporter
 // @name:zh-TW         ChatGPT Exporter
 // @namespace          pionxzh
-// @version            2.24.5
+// @version            2.24.6
 // @author             pionxzh
 // @description        Easily export the whole ChatGPT conversation history for further analysis or sharing.
 // @description:zh-CN  轻松导出 ChatGPT 聊天记录，以便进一步分析或分享。
@@ -20903,6 +20903,7 @@ ${content2.text}
     downloadFile("chatgpt-export.zip", "application/zip", blob);
     return true;
   }
+  const LatexRegex$1 = /(\s\$\$.+\$\$\s|\s\$.+\$\s|\\\[.+\\\]|\\\(.+\\\))|(^\$$[\S\s]+^\$$)|(^\$\$[\S\s]+^\$\$$)/gm;
   function conversationToMarkdown(conversation, metaList) {
     const { id, title: title2, model, modelSlug, createTime, updateTime, conversationNodes } = conversation;
     const source = `${baseUrl}/c/${id}`;
@@ -20942,17 +20943,29 @@ ${_metaList.join("\n")}
 `;
       }
       const author = transformAuthor$1(message.author);
-      let postSteps = [];
+      const postSteps = [];
       if (message.author.role === "assistant") {
-        postSteps = [...postSteps, (input) => transformFootNotes$1(input, message.metadata)];
+        postSteps.push((input) => transformFootNotes$1(input, message.metadata));
       }
       if (message.author.role === "assistant") {
-        postSteps = [...postSteps, (input) => {
-          if (!/```/.test(input)) {
-            input = input.replace(/^\\\[(.+)\\\]$/gm, "$$$$$1$$$$").replace(/\\\[/g, "$").replace(/\\\]/g, "$").replace(/\\\(/g, "$").replace(/\\\)/g, "$");
+        postSteps.push((input) => {
+          input = input.replace(/^\\\[(.+)\\\]$/gm, "$$$$$1$$$$").replace(/\\\[/g, "$").replace(/\\\]/g, "$").replace(/\\\(/g, "$").replace(/\\\)/g, "$");
+          const matches = input.match(LatexRegex$1);
+          const isCodeBlock = /```/.test(input);
+          if (!isCodeBlock && matches) {
+            let index2 = 0;
+            input = input.replace(LatexRegex$1, () => {
+              return `╬${index2++}╬`;
+            });
           }
-          return toMarkdown(fromMarkdown(input));
-        }];
+          let transformed = toMarkdown(fromMarkdown(input));
+          if (!isCodeBlock && matches) {
+            transformed = transformed.replace(/╬(\d+)╬/g, (_24, index2) => {
+              return matches[+index2];
+            });
+          }
+          return transformed;
+        });
       }
       const postProcess = (input) => postSteps.reduce((acc, fn2) => fn2(acc), input);
       const content22 = transformContent$1(message.content, message.metadata, postProcess);
@@ -21065,6 +21078,7 @@ ${content2.text}
     copyToClipboard(standardizeLineBreaks(text2));
     return true;
   }
+  const LatexRegex = /(\s\$\$.+\$\$\s|\s\$.+\$\s|\\\[.+\\\]|\\\(.+\\\))|(^\$$[\S\s]+^\$$)|(^\$\$[\S\s]+^\$\$$)/gm;
   function transformMessage(message) {
     var _a, _b, _c;
     if (!message || !message.content) return null;
@@ -21080,12 +21094,24 @@ ${content2.text}
     }
     const author = transformAuthor(message.author);
     let content2 = transformContent(message.content, message.metadata);
+    const matches = content2.match(LatexRegex);
+    if (matches) {
+      let index2 = 0;
+      content2 = content2.replace(LatexRegex, () => {
+        return `╬${index2++}╬`;
+      });
+    }
     if (message.author.role === "assistant") {
       content2 = transformFootNotes(content2, message.metadata);
     }
     if (message.author.role === "assistant" && content2) {
       content2 = reformatContent(content2);
     }
+    if (matches) {
+      content2 = content2.replace(/╬(\d+)╬/g, (_24, index2) => {
+        return matches[+index2];
+      });
+    }
     return `${author}:
 ${content2}`;
   }

diff --git a/package.json b/package.json
@@ -1,7 +1,7 @@
 {
   "name": "@pionxzh/chatgpt-exporter",
   "type": "module",
-  "version": "2.24.5",
+  "version": "2.24.6",
   "private": true,
   "packageManager": "[email protected]",
   "title": "ChatGPT Exporter",

diff --git a/src/exporter/markdown.ts b/src/exporter/markdown.ts
@@ -63,6 +63,8 @@ export async function exportAllToMarkdown(fileNameFormat: string, apiConversatio
     return true
 }
 
+const LatexRegex = /(\s\$\$.+\$\$\s|\s\$.+\$\s|\\\[.+\\\]|\\\(.+\\\))|(^\$$[\S\s]+^\$$)|(^\$\$[\S\s]+^\$\$$)/gm
+
 function conversationToMarkdown(conversation: ConversationResult, metaList?: ExportMeta[]) {
     const { id, title, model, modelSlug, createTime, updateTime, conversationNodes } = conversation
     const source = `${baseUrl}/c/${id}`
@@ -125,25 +127,44 @@ function conversationToMarkdown(conversation: ConversationResult, metaList?: Exp
 
         const author = transformAuthor(message.author)
 
-        let postSteps: Array<(input: string) => string> = []
+        const postSteps: Array<(input: string) => string> = []
         if (message.author.role === 'assistant') {
-            postSteps = [...postSteps, input => transformFootNotes(input, message.metadata)]
+            postSteps.push(input => transformFootNotes(input, message.metadata))
         }
         // Only message from assistant will be reformatted
         if (message.author.role === 'assistant') {
-            postSteps = [...postSteps, (input) => {
+            postSteps.push((input) => {
+                // Replace mathematical formula annotation
+                input = input
+                    .replace(/^\\\[(.+)\\\]$/gm, '$$$$$1$$$$')
+                    .replace(/\\\[/g, '$')
+                    .replace(/\\\]/g, '$')
+                    .replace(/\\\(/g, '$')
+                    .replace(/\\\)/g, '$')
+
+                const matches = input.match(LatexRegex)
+
                 // Skip code block as the following steps can potentially break the code
-                if (!(/```/.test(input))) {
-                    // Replace mathematical formula annotation
-                    input = input
-                        .replace(/^\\\[(.+)\\\]$/gm, '$$$$$1$$$$')
-                        .replace(/\\\[/g, '$')
-                        .replace(/\\\]/g, '$')
-                        .replace(/\\\(/g, '$')
-                        .replace(/\\\)/g, '$')
+                const isCodeBlock = /```/.test(input)
+                if (!isCodeBlock && matches) {
+                    let index = 0
+                    input = input.replace(LatexRegex, () => {
+                        // Replace it with `╬${index}╬` to avoid markdown processor ruin the formula
+                        return `╬${index++}╬`
+                    })
                 }
-                return toMarkdown(fromMarkdown(input))
-            }]
+
+                let transformed = toMarkdown(fromMarkdown(input))
+
+                if (!isCodeBlock && matches) {
+                    // Replace `╬${index}╬` back to the original latex
+                    transformed = transformed.replace(/╬(\d+)╬/g, (_, index) => {
+                        return matches[+index]
+                    })
+                }
+
+                return transformed
+            })
         }
         const postProcess = (input: string) => postSteps.reduce((acc, fn) => fn(acc), input)
         const content = transformContent(message.content, message.metadata, postProcess)

diff --git a/src/exporter/text.ts b/src/exporter/text.ts
@@ -29,6 +29,8 @@ export async function exportToText() {
     return true
 }
 
+const LatexRegex = /(\s\$\$.+\$\$\s|\s\$.+\$\s|\\\[.+\\\]|\\\(.+\\\))|(^\$$[\S\s]+^\$$)|(^\$\$[\S\s]+^\$\$$)/gm
+
 function transformMessage(message?: ConversationNodeMessage) {
     if (!message || !message.content) return null
 
@@ -53,6 +55,16 @@ function transformMessage(message?: ConversationNodeMessage) {
 
     const author = transformAuthor(message.author)
     let content = transformContent(message.content, message.metadata)
+
+    const matches = content.match(LatexRegex)
+    if (matches) {
+        let index = 0
+        content = content.replace(LatexRegex, () => {
+            // Replace it with `╬${index}╬` to avoid markdown processor ruin the formula
+            return `╬${index++}╬`
+        })
+    }
+
     if (message.author.role === 'assistant') {
         content = transformFootNotes(content, message.metadata)
     }
@@ -61,6 +73,14 @@ function transformMessage(message?: ConversationNodeMessage) {
     if (message.author.role === 'assistant' && content) {
         content = reformatContent(content)
     }
+
+    if (matches) {
+        // Replace `╬${index}╬` back to the original latex
+        content = content.replace(/╬(\d+)╬/g, (_, index) => {
+            return matches[+index]
+        })
+    }
+
     return `${author}:\n${content}`
 }