feat: Markdown 工具栏编辑器与服务端预览

共用 md_editor 于发帖/改帖/评论，扩展 ComposeBodyToHTML，并提供 /compose/preview。 Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-30 04:50:07 +08:00
parent 7f048bd9b5
commit 1f777eebb1
17 changed files with 1056 additions and 232 deletions
--- a/services/compose_body.go
+++ b/services/compose_body.go
@@ -7,73 +7,266 @@ import (
 )

 var (
-	mdImageRe = regexp.MustCompile(`!\[([^\]]*)\]\(([^)]+)\)`)
-	mdLinkRe  = regexp.MustCompile(`\[([^\]]+)\]\(([^)]+)\)`)
+	mdImageRe      = regexp.MustCompile(`!\[([^\]]*)\]\(([^)]+)\)`)
+	mdLinkRe       = regexp.MustCompile(`\[([^\]]+)\]\(([^)]+)\)`)
+	mdBoldRe       = regexp.MustCompile(`\*\*([^*]+)\*\*`)
+	mdStrikeRe     = regexp.MustCompile(`~~([^~]+)~~`)
+	mdItalicRe     = regexp.MustCompile(`\*([^*\n]+)\*`)
+	mdInlineCodeRe = regexp.MustCompile("`([^`\n]+)`")
+	mdLangRe       = regexp.MustCompile(`^[a-zA-Z0-9_+-]+$`)
 )

-// ComposeBodyToHTML 将发帖表单正文转为可消毒 HTML。
-// 若已含块级 HTML / 门控标签则原样交 Sanitize；否则按纯文本/轻量 Markdown 转段落。
+// ComposeBodyToHTML 将发帖/评论表单正文转为可消毒 HTML。
+// 若已含块级 HTML / 门控标签则原样交 Sanitize；否则按轻量 Markdown 转换。
 func ComposeBodyToHTML(raw string) string {
 	raw = strings.TrimSpace(raw)
 	if raw == "" {
 		return ""
 	}
-	lower := strings.ToLower(raw)
-	if strings.Contains(lower, "<p") || strings.Contains(lower, "<div") ||
-		strings.Contains(lower, "<h1") || strings.Contains(lower, "<ul") ||
-		strings.Contains(lower, "<ol") || strings.Contains(lower, "<pre") ||
-		strings.Contains(lower, "<blockquote") ||
-		strings.Contains(lower, "<members-only") || strings.Contains(lower, "<reply-only") ||
-		strings.Contains(lower, "<points-only") {
+	if looksLikeRichHTML(raw) {
 		return raw
 	}
+	return mdDocumentToHTML(raw)
+}

-	// 先转义，再恢复轻量 md 图片/链接
-	esc := html.EscapeString(raw)
-	esc = mdImageRe.ReplaceAllStringFunc(esc, func(m string) string {
-		sub := mdImageRe.FindStringSubmatch(m)
-		if len(sub) != 3 {
-			return m
+func looksLikeRichHTML(raw string) bool {
+	lower := strings.ToLower(raw)
+	markers := []string{
+		"<p", "<div", "<h1", "<h2", "<h3", "<ul", "<ol", "<pre", "<blockquote",
+		"<table", "<members-only", "<reply-only", "<points-only",
+	}
+	for _, m := range markers {
+		if strings.Contains(lower, m) {
+			return true
 		}
-		alt, src := sub[1], sub[2]
-		// 仅允许站内 uploads 或 http(s)
-		if !safeComposeURL(src) {
-			return m
-		}
-		return `<p><img src="` + html.EscapeString(src) + `" alt="` + alt + `"></p>`
-	})
-	esc = mdLinkRe.ReplaceAllStringFunc(esc, func(m string) string {
-		sub := mdLinkRe.FindStringSubmatch(m)
-		if len(sub) != 3 {
-			return m
-		}
-		text, href := sub[1], sub[2]
-		if !safeComposeURL(href) {
-			return m
-		}
-		return `<a href="` + html.EscapeString(href) + `" rel="noopener noreferrer">` + text + `</a>`
-	})
+	}
+	return false
+}

-	parts := strings.Split(esc, "\n\n")
+func mdDocumentToHTML(raw string) string {
+	lines := strings.Split(strings.ReplaceAll(raw, "\r\n", "\n"), "\n")
 	var b strings.Builder
-	for _, p := range parts {
-		p = strings.TrimSpace(p)
-		if p == "" {
+	i := 0
+	for i < len(lines) {
+		line := lines[i]
+		trim := strings.TrimSpace(line)
+
+		// 围栏代码块
+		if strings.HasPrefix(trim, "```") {
+			lang := strings.TrimSpace(strings.TrimPrefix(trim, "```"))
+			lang = strings.TrimSpace(strings.Split(lang, " ")[0])
+			i++
+			var code []string
+			for i < len(lines) {
+				if strings.TrimSpace(lines[i]) == "```" {
+					i++
+					break
+				}
+				code = append(code, lines[i])
+				i++
+			}
+			b.WriteString("<pre><code")
+			if lang != "" && mdLangRe.MatchString(lang) {
+				b.WriteString(` class="language-`)
+				b.WriteString(html.EscapeString(lang))
+				b.WriteString(`"`)
+			}
+			b.WriteString(">")
+			b.WriteString(html.EscapeString(strings.Join(code, "\n")))
+			b.WriteString("</code></pre>\n")
 			continue
 		}
-		// 已是独立 img 段
-		if strings.HasPrefix(p, "<p><img ") {
-			b.WriteString(p)
+
+		// 空行跳过
+		if trim == "" {
+			i++
+			continue
+		}
+
+		// 标题：#–###### → h2–h6（禁止 h1）
+		if level, title, ok := parseMDHeading(trim); ok {
+			b.WriteString("<h")
+			b.WriteString(string(rune('0' + level)))
+			b.WriteString(">")
+			b.WriteString(mdInlineToHTML(title))
+			b.WriteString("</h")
+			b.WriteString(string(rune('0' + level)))
+			b.WriteString(">\n")
+			i++
+			continue
+		}
+
+		// 引用
+		if strings.HasPrefix(trim, "> ") || trim == ">" {
+			var quote []string
+			for i < len(lines) {
+				t := strings.TrimSpace(lines[i])
+				if t == ">" {
+					quote = append(quote, "")
+					i++
+					continue
+				}
+				if strings.HasPrefix(t, "> ") {
+					quote = append(quote, strings.TrimPrefix(t, "> "))
+					i++
+					continue
+				}
+				break
+			}
+			b.WriteString("<blockquote><p>")
+			b.WriteString(mdInlineToHTML(strings.Join(quote, "<br>\n")))
+			b.WriteString("</p></blockquote>\n")
+			continue
+		}
+
+		// 无序列表
+		if isMDUL(trim) {
+			b.WriteString("<ul>\n")
+			for i < len(lines) {
+				t := strings.TrimSpace(lines[i])
+				if !isMDUL(t) {
+					break
+				}
+				item := strings.TrimSpace(t[1:])
+				if strings.HasPrefix(t, "- ") || strings.HasPrefix(t, "* ") || strings.HasPrefix(t, "+ ") {
+					item = strings.TrimSpace(t[2:])
+				}
+				b.WriteString("<li>")
+				b.WriteString(mdInlineToHTML(item))
+				b.WriteString("</li>\n")
+				i++
+			}
+			b.WriteString("</ul>\n")
+			continue
+		}
+
+		// 有序列表
+		if isMDOL(trim) {
+			b.WriteString("<ol>\n")
+			for i < len(lines) {
+				t := strings.TrimSpace(lines[i])
+				if !isMDOL(t) {
+					break
+				}
+				idx := strings.Index(t, ". ")
+				item := strings.TrimSpace(t[idx+2:])
+				b.WriteString("<li>")
+				b.WriteString(mdInlineToHTML(item))
+				b.WriteString("</li>\n")
+				i++
+			}
+			b.WriteString("</ol>\n")
+			continue
+		}
+
+		// 普通段落（连续非空、非特殊行）
+		var para []string
+		for i < len(lines) {
+			t := strings.TrimSpace(lines[i])
+			if t == "" || strings.HasPrefix(t, "```") || isMDUL(t) || isMDOL(t) ||
+				strings.HasPrefix(t, "> ") || t == ">" {
+				break
+			}
+			if _, _, ok := parseMDHeading(t); ok {
+				break
+			}
+			para = append(para, lines[i])
+			i++
+		}
+		joined := strings.Join(para, "\n")
+		// 整段已是图片行则单独输出
+		if m := mdImageRe.FindStringSubmatch(strings.TrimSpace(joined)); len(m) == 3 && strings.TrimSpace(joined) == m[0] && safeComposeURL(m[2]) {
+			b.WriteString(`<p><img src="`)
+			b.WriteString(html.EscapeString(m[2]))
+			b.WriteString(`" alt="`)
+			b.WriteString(m[1])
+			b.WriteString(`"></p>` + "\n")
 			continue
 		}
-		p = strings.ReplaceAll(p, "\n", "<br>\n")
 		b.WriteString("<p>")
-		b.WriteString(p)
+		b.WriteString(mdInlineToHTML(strings.ReplaceAll(joined, "\n", "<br>\n")))
 		b.WriteString("</p>\n")
 	}
 	return b.String()
 }

+func parseMDHeading(trim string) (level int, title string, ok bool) {
+	n := 0
+	for n < len(trim) && trim[n] == '#' {
+		n++
+	}
+	if n < 1 || n > 6 {
+		return 0, "", false
+	}
+	if n >= len(trim) || trim[n] != ' ' {
+		return 0, "", false
+	}
+	title = strings.TrimSpace(trim[n+1:])
+	if title == "" {
+		return 0, "", false
+	}
+	level = n
+	if level == 1 {
+		level = 2 // 帖正文禁止 h1
+	}
+	return level, title, true
+}
+
+func isMDUL(trim string) bool {
+	return strings.HasPrefix(trim, "- ") || strings.HasPrefix(trim, "* ") || strings.HasPrefix(trim, "+ ")
+}
+
+var mdOLRe = regexp.MustCompile(`^\d+\. `)
+
+func isMDOL(trim string) bool {
+	return mdOLRe.MatchString(trim)
+}
+
+func mdInlineToHTML(s string) string {
+	// 先抽出行内代码，避免其它规则干扰
+	type ph struct{ key, html string }
+	var phs []ph
+	s = mdInlineCodeRe.ReplaceAllStringFunc(s, func(m string) string {
+		sub := mdInlineCodeRe.FindStringSubmatch(m)
+		if len(sub) != 2 {
+			return m
+		}
+		key := "\x00CODE" + string(rune('A'+len(phs))) + "\x00"
+		phs = append(phs, ph{key: key, html: "<code>" + html.EscapeString(sub[1]) + "</code>"})
+		return key
+	})
+
+	s = html.EscapeString(s)
+	// 占位符被 escape 了，还原 key
+	for i := range phs {
+		escKey := html.EscapeString(phs[i].key)
+		s = strings.ReplaceAll(s, escKey, phs[i].key)
+	}
+
+	s = mdImageRe.ReplaceAllStringFunc(s, func(m string) string {
+		sub := mdImageRe.FindStringSubmatch(m)
+		if len(sub) != 3 || !safeComposeURL(sub[2]) {
+			return m
+		}
+		return `<img src="` + html.EscapeString(sub[2]) + `" alt="` + sub[1] + `">`
+	})
+	s = mdLinkRe.ReplaceAllStringFunc(s, func(m string) string {
+		sub := mdLinkRe.FindStringSubmatch(m)
+		if len(sub) != 3 || !safeComposeURL(sub[2]) {
+			return m
+		}
+		return `<a href="` + html.EscapeString(sub[2]) + `" rel="noopener noreferrer">` + sub[1] + `</a>`
+	})
+	s = mdBoldRe.ReplaceAllString(s, `<strong>$1</strong>`)
+	s = mdStrikeRe.ReplaceAllString(s, `<s>$1</s>`)
+	s = mdItalicRe.ReplaceAllString(s, `<em>$1</em>`)
+
+	for _, p := range phs {
+		s = strings.ReplaceAll(s, p.key, p.html)
+	}
+	return s
+}
+
 func safeComposeURL(u string) bool {
 	u = strings.TrimSpace(u)
 	if strings.HasPrefix(u, "/uploads/") || strings.HasPrefix(u, "/media/") {
@@ -82,24 +275,71 @@ func safeComposeURL(u string) bool {
 	return strings.HasPrefix(u, "https://") || strings.HasPrefix(u, "http://")
 }

-// HTMLToComposePlain 编辑页回显：去掉简单标签便于 textarea 编辑（尽力而为）
+// HTMLToComposePlain 编辑页回显：尽力还原为 Markdown / 纯文本；含门控时保留 HTML。
 func HTMLToComposePlain(htmlBody string) string {
 	s := strings.TrimSpace(htmlBody)
 	if s == "" {
 		return ""
 	}
 	lower := strings.ToLower(s)
-	// 含门控块时保留 HTML，避免编辑丢失
 	if strings.Contains(lower, "<members-only") || strings.Contains(lower, "<reply-only") ||
 		strings.Contains(lower, "<points-only") {
 		return s
 	}
-	// img → markdown
-	imgRe := regexp.MustCompile(`(?i)<img[^>]+src="([^"]+)"[^>]*>`)
-	s = imgRe.ReplaceAllString(s, "![]($1)")
+
+	// 代码块
+	s = regexp.MustCompile(`(?is)<pre[^>]*>\s*<code(?:\s+class="language-([^"]*)")?[^>]*>([\s\S]*?)</code>\s*</pre>`).ReplaceAllStringFunc(s, func(m string) string {
+		sub := regexp.MustCompile(`(?is)<pre[^>]*>\s*<code(?:\s+class="language-([^"]*)")?[^>]*>([\s\S]*?)</code>\s*</pre>`).FindStringSubmatch(m)
+		if len(sub) != 3 {
+			return m
+		}
+		lang, body := sub[1], html.UnescapeString(sub[2])
+		return "```" + lang + "\n" + body + "\n```\n\n"
+	})
+
+	s = regexp.MustCompile(`(?i)<h([2-6])[^>]*>([\s\S]*?)</h[2-6]>`).ReplaceAllStringFunc(s, func(m string) string {
+		sub := regexp.MustCompile(`(?i)<h([2-6])[^>]*>([\s\S]*?)</h[2-6]>`).FindStringSubmatch(m)
+		if len(sub) != 3 {
+			return m
+		}
+		level := sub[1][0] - '0'
+		inner := stripTagsKeepText(sub[2])
+		return strings.Repeat("#", int(level)) + " " + inner + "\n\n"
+	})
+
+	s = regexp.MustCompile(`(?i)<li[^>]*>([\s\S]*?)</li>`).ReplaceAllString(s, "- $1\n")
+	s = regexp.MustCompile(`(?i)</?[uo]l[^>]*>`).ReplaceAllString(s, "\n")
+	s = regexp.MustCompile(`(?i)<blockquote[^>]*>([\s\S]*?)</blockquote>`).ReplaceAllStringFunc(s, func(m string) string {
+		sub := regexp.MustCompile(`(?i)<blockquote[^>]*>([\s\S]*?)</blockquote>`).FindStringSubmatch(m)
+		if len(sub) != 2 {
+			return m
+		}
+		inner := stripTagsKeepText(sub[1])
+		lines := strings.Split(inner, "\n")
+		for i := range lines {
+			lines[i] = "> " + strings.TrimSpace(lines[i])
+		}
+		return strings.Join(lines, "\n") + "\n\n"
+	})
+
+	s = regexp.MustCompile(`(?i)<strong[^>]*>([\s\S]*?)</strong>`).ReplaceAllString(s, "**$1**")
+	s = regexp.MustCompile(`(?i)<b[^>]*>([\s\S]*?)</b>`).ReplaceAllString(s, "**$1**")
+	s = regexp.MustCompile(`(?i)<em[^>]*>([\s\S]*?)</em>`).ReplaceAllString(s, "*$1*")
+	s = regexp.MustCompile(`(?i)<i[^>]*>([\s\S]*?)</i>`).ReplaceAllString(s, "*$1*")
+	s = regexp.MustCompile(`(?i)<s[^>]*>([\s\S]*?)</s>`).ReplaceAllString(s, "~~$1~~")
+	s = regexp.MustCompile(`(?i)<del[^>]*>([\s\S]*?)</del>`).ReplaceAllString(s, "~~$1~~")
+	s = regexp.MustCompile(`(?i)<code[^>]*>([\s\S]*?)</code>`).ReplaceAllString(s, "`$1`")
+	s = regexp.MustCompile(`(?i)<img[^>]+src="([^"]+)"[^>]*>`).ReplaceAllString(s, "![]($1)")
+	s = regexp.MustCompile(`(?i)<a[^>]+href="([^"]+)"[^>]*>([\s\S]*?)</a>`).ReplaceAllString(s, "[$2]($1)")
 	s = regexp.MustCompile(`(?i)</p>\s*<p>`).ReplaceAllString(s, "\n\n")
 	s = regexp.MustCompile(`(?i)<br\s*/?>`).ReplaceAllString(s, "\n")
 	s = regexp.MustCompile(`(?i)</?p[^>]*>`).ReplaceAllString(s, "")
 	s = regexp.MustCompile(`(?i)<[^>]+>`).ReplaceAllString(s, "")
 	return strings.TrimSpace(html.UnescapeString(s))
 }
+
+func stripTagsKeepText(s string) string {
+	s = regexp.MustCompile(`(?i)<br\s*/?>`).ReplaceAllString(s, "\n")
+	s = regexp.MustCompile(`(?i)<[^>]+>`).ReplaceAllString(s, "")
+	return strings.TrimSpace(html.UnescapeString(s))
+}
--- a/services/compose_body_test.go
+++ b/services/compose_body_test.go
@@ -0,0 +1,72 @@
+package services
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestComposeBodyToHTML_MarkdownBasics(t *testing.T) {
+	in := strings.TrimSpace(`
+## 标题
+
+段落 **粗体** 与 *斜体* 和 ~~删除~~ 与 ` + "`code`" + `。
+
+- 甲
+- 乙
+
+1. 一
+2. 二
+
+> 引用行
+
+` + "```go\nfmt.Println(1)\n```" + `
+
+[链接](https://example.com)
+![图](/uploads/a.png)
+`)
+	out := ComposeBodyToHTML(in)
+	checks := []string{
+		"<h2>", "标题", "</h2>",
+		"<strong>粗体</strong>",
+		"<em>斜体</em>",
+		"<s>删除</s>",
+		"<code>code</code>",
+		"<ul>", "<li>",
+		"<ol>",
+		"<blockquote>",
+		`class="language-go"`,
+		"fmt.Println(1)",
+		`href="https://example.com"`,
+		`src="/uploads/a.png"`,
+	}
+	for _, c := range checks {
+		if !strings.Contains(out, c) {
+			t.Fatalf("missing %q in:\n%s", c, out)
+		}
+	}
+	if strings.Contains(out, "<h1>") {
+		t.Fatal("must not emit h1")
+	}
+}
+
+func TestComposeBodyToHTML_PassThroughGates(t *testing.T) {
+	in := `<members-only data-gate="login"><p>隐</p></members-only>`
+	if got := ComposeBodyToHTML(in); got != in {
+		t.Fatalf("gate html should pass through, got %q", got)
+	}
+}
+
+func TestComposeBodyToHTML_H1MapsToH2(t *testing.T) {
+	out := ComposeBodyToHTML("# 顶标题")
+	if !strings.Contains(out, "<h2>") || strings.Contains(out, "<h1>") {
+		t.Fatalf("got %s", out)
+	}
+}
+
+func TestHTMLToComposePlain_RoundTripLite(t *testing.T) {
+	html := ComposeBodyToHTML("## Hi\n\n**x**")
+	plain := HTMLToComposePlain(html)
+	if !strings.Contains(plain, "## Hi") || !strings.Contains(plain, "**x**") {
+		t.Fatalf("plain=%q", plain)
+	}
+}