mirror of
https://wget.la/https://github.com/Wxw-Gu/WechatExplorer
synced 2026-08-18 03:57:02 +08:00
104 lines
3.0 KiB
Go
104 lines
3.0 KiB
Go
package messaging
|
|
|
|
import (
|
|
"regexp"
|
|
"strings"
|
|
)
|
|
|
|
var (
|
|
// Code blocks: strip fences, keep code content
|
|
reCodeBlock = regexp.MustCompile("(?s)```[^\n]*\n?(.*?)```")
|
|
// Inline code: strip backticks, keep content
|
|
reInlineCode = regexp.MustCompile("`([^`]+)`")
|
|
// Images: remove entirely
|
|
reImage = regexp.MustCompile(`!\[[^\]]*\]\([^)]*\)`)
|
|
// Links: keep display text only
|
|
reLink = regexp.MustCompile(`\[([^\]]+)\]\([^)]*\)`)
|
|
// Table separator rows: remove
|
|
reTableSep = regexp.MustCompile(`(?m)^\|[\s:|\-]+\|$`)
|
|
// Table rows: convert pipe-delimited to space-delimited
|
|
reTableRow = regexp.MustCompile(`(?m)^\|(.+)\|$`)
|
|
// Headers: remove # prefix
|
|
reHeader = regexp.MustCompile(`(?m)^#{1,6}\s+`)
|
|
// Bold: **text** or __text__
|
|
reBold = regexp.MustCompile(`\*\*(.+?)\*\*|__(.+?)__`)
|
|
// Italic: *text* or _text_
|
|
reItalic = regexp.MustCompile(`(?:^|[^*])\*([^*]+)\*(?:[^*]|$)|(?:^|[^_])_([^_]+)_(?:[^_]|$)`)
|
|
// Strikethrough: ~~text~~
|
|
reStrike = regexp.MustCompile(`~~(.+?)~~`)
|
|
// Blockquote: > prefix
|
|
reBlockquote = regexp.MustCompile(`(?m)^>\s?`)
|
|
// Horizontal rule
|
|
reHR = regexp.MustCompile(`(?m)^[-*_]{3,}\s*$`)
|
|
// Unordered list markers: -, *, +
|
|
reUL = regexp.MustCompile(`(?m)^(\s*)[-*+]\s+`)
|
|
)
|
|
|
|
// MarkdownToPlainText converts markdown to readable plain text for WeChat.
|
|
func MarkdownToPlainText(text string) string {
|
|
result := text
|
|
|
|
// Code blocks: strip fences, keep code content
|
|
result = reCodeBlock.ReplaceAllStringFunc(result, func(match string) string {
|
|
parts := reCodeBlock.FindStringSubmatch(match)
|
|
if len(parts) > 1 {
|
|
return strings.TrimSpace(parts[1])
|
|
}
|
|
return match
|
|
})
|
|
|
|
// Images: remove entirely
|
|
result = reImage.ReplaceAllString(result, "")
|
|
|
|
// Links: keep display text only
|
|
result = reLink.ReplaceAllString(result, "$1")
|
|
|
|
// Table separator rows: remove
|
|
result = reTableSep.ReplaceAllString(result, "")
|
|
|
|
// Table rows: pipe-delimited to space-delimited
|
|
result = reTableRow.ReplaceAllStringFunc(result, func(match string) string {
|
|
parts := reTableRow.FindStringSubmatch(match)
|
|
if len(parts) > 1 {
|
|
cells := strings.Split(parts[1], "|")
|
|
for i := range cells {
|
|
cells[i] = strings.TrimSpace(cells[i])
|
|
}
|
|
return strings.Join(cells, " ")
|
|
}
|
|
return match
|
|
})
|
|
|
|
// Headers: remove # prefix
|
|
result = reHeader.ReplaceAllString(result, "")
|
|
|
|
// Bold
|
|
result = reBold.ReplaceAllStringFunc(result, func(match string) string {
|
|
parts := reBold.FindStringSubmatch(match)
|
|
if parts[1] != "" {
|
|
return parts[1]
|
|
}
|
|
return parts[2]
|
|
})
|
|
|
|
// Strikethrough
|
|
result = reStrike.ReplaceAllString(result, "$1")
|
|
|
|
// Blockquote
|
|
result = reBlockquote.ReplaceAllString(result, "")
|
|
|
|
// Horizontal rule -> empty line
|
|
result = reHR.ReplaceAllString(result, "")
|
|
|
|
// Unordered list: replace markers with "• "
|
|
result = reUL.ReplaceAllString(result, "${1}• ")
|
|
|
|
// Inline code: strip backticks (do after code blocks)
|
|
result = reInlineCode.ReplaceAllString(result, "$1")
|
|
|
|
// Clean up excessive blank lines
|
|
result = regexp.MustCompile(`\n{3,}`).ReplaceAllString(result, "\n\n")
|
|
|
|
return strings.TrimSpace(result)
|
|
}
|