增加内容可搜索字段,屏蔽各种标记。

This commit is contained in:
2026-08-29 01:40:52 +08:00
parent cfe5cf53aa
commit f216ef4d1b
5 changed files with 125 additions and 34 deletions
+85
View File
@@ -3,6 +3,8 @@ package utils
import (
"bytes"
"html/template"
"regexp"
"strings"
"github.com/yuin/goldmark"
"github.com/yuin/goldmark/extension"
@@ -29,3 +31,86 @@ func RenderMarkdown(source string) template.HTML {
}
return template.HTML(buf.String())
}
// 全局预编译正则(保持不变,但移除冗余)
var (
// 标题
reHeading = regexp.MustCompile(`(?m)^\s*#{1,6}\s+`)
// 无序列表
reUnorderedList = regexp.MustCompile(`(?m)^\s*[-*+]\s+`)
// 有序列表
reOrderedList = regexp.MustCompile(`(?m)^\s*\d+[.)]\s+`)
// 块引用
reBlockquote = regexp.MustCompile(`(?m)^\s*>\s+`)
// 水平线
reHr = regexp.MustCompile(`(?m)^\s*[-*_]{3,}\s*$`)
// 脚注定义
reFootnoteDef = regexp.MustCompile(`(?m)^\[\^.+?\]:\s.*$`)
// 引用定义
reRefDef = regexp.MustCompile(`(?m)^\s*\[.+?\]:\s*(.*)$`)
// 图片
reImage = regexp.MustCompile(`(?s)!\[.*?\]\(.*?\)`)
// 引用图片
reRefImage = regexp.MustCompile(`!\[.*?\](\[.*?\])?`)
// 链接
reLink = regexp.MustCompile(`(?s)\[(.*?)\]\(.*?\)`)
// 引用式链接
reRefLink = regexp.MustCompile(`\[([^\s=\[\]/]+)\](\[.*?\])?`)
// 粗体
reBoldStar = regexp.MustCompile(`(?s)\*\*(.*?)\*\*`)
// 斜体
reItalicStar = regexp.MustCompile(`(?s)\*(.*?)\*`)
// 下划线粗体
reBoldUnderscore = regexp.MustCompile(`(?s)_(.*?)_`)
// 下划线斜体
reItalicUnderscore = regexp.MustCompile(`(?s)_(.*?)_`)
// HTML 标签
reHTMLTag = regexp.MustCompile(`<[^>]+>`)
// 多余空白
reMultiSpace = regexp.MustCompile(`\s+`)
)
// GenerateSearchableText 提取 Markdown 中的纯文本
func GenerateSearchableText(raw string) string {
if raw == "" {
return ""
}
text := raw
// 全角空格转半角
text = strings.ReplaceAll(text, "\u3000", " ")
// 移除块级标记(一次性替换多个)
text = reHr.ReplaceAllString(text, "")
text = reBlockquote.ReplaceAllString(text, "")
text = reUnorderedList.ReplaceAllString(text, "")
text = reOrderedList.ReplaceAllString(text, "")
text = reHeading.ReplaceAllString(text, "")
// 移除引用定义
text = reFootnoteDef.ReplaceAllString(text, "")
text = reRefDef.ReplaceAllString(text, "")
// 移除图片
text = reImage.ReplaceAllString(text, "")
text = reRefImage.ReplaceAllString(text, "")
// 处理链接:保留可见文字(先引用式后内联)
text = reRefLink.ReplaceAllString(text, "$1")
text = reLink.ReplaceAllString(text, "$1")
// 移除粗体
text = reBoldStar.ReplaceAllString(text, "")
text = reBoldUnderscore.ReplaceAllString(text, "")
// 移除斜体
text = reItalicStar.ReplaceAllString(text, "")
text = reItalicUnderscore.ReplaceAllString(text, "")
// 移除 HTML 标签
text = reHTMLTag.ReplaceAllString(text, "")
// 压缩空白
text = reMultiSpace.ReplaceAllString(text, " ")
return strings.TrimSpace(text)
}