upstream-archive/chatgpt-exporter
1
0
Fork 0
mirror of https://github.com/pionxzh/chatgpt-exporter.git synced 2026-07-23 09:00:51 -05:00
Code Issues Packages Projects Releases Wiki Activity

fix: fix unexpected latex formula escape

Browse Source
This commit is contained in:
Pionxzh
2024-09-09 03:28:45 +08:00
parent 67af2bf530
commit 30f0b623b3
2 changed files with 54 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

47
src/exporter/markdown.ts
View File

@@ -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)

20
src/exporter/text.ts
View File

@@ -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}`
}