🎨 Protyle Improve Paste and Paste As Plain Text siyuan-note/siyua…

…n#8289

ast/node.go

@@ -430,6 +430,7 @@ func (n *Node) Content() (ret string) {
 		case NodeTextMark:
 			if "" != n.TextMarkTextContent {
 				if n.IsTextMarkType("code") {
+					// 搜索代码内容转义问题 https://github.com/siyuan-note/siyuan/issues/5927
 					buf.WriteString(html.UnescapeString(n.TextMarkTextContent))
 				} else {
 					buf.WriteString(n.TextMarkTextContent)
@@ -447,6 +448,13 @@ func (n *Node) Content() (ret string) {
 	return buf.String()
 }
 
+// EscapeMarkerContent 返回 n 及其所有内容子节点的文本值（其中的标记符会被转义），块级节点间通过换行符分隔。
+func (n *Node) EscapeMarkerContent() (ret string) {
+	ret = n.Content()
+	ret = string(lex.EscapeProtyleMarkers([]byte(ret)))
+	return
+}
+
 func (n *Node) Stat() (runeCnt, wordCnt, linkCnt, imgCnt, refCnt int) {
 	buf := make([]byte, 0, 8192)
 	Walk(n, func(n *Node, entering bool) WalkStatus {

h2m.go

@@ -123,7 +123,11 @@ func (lute *Lute) genASTByDOM(n *html.Node, tree *parse.Tree) {
 		}
 		node.Tokens = bytes.ReplaceAll(node.Tokens, []byte{194, 160}, []byte{' '}) // 将   转换为空格
 		node.Tokens = bytes.ReplaceAll(node.Tokens, []byte("\n"), []byte{' '})     // 将 \n 转换为空格 https://github.com/siyuan-note/siyuan/issues/6052
-		node.Tokens = lex.EscapeMarkers(node.Tokens)
+		if lute.ParseOptions.ProtyleWYSIWYG {
+			node.Tokens = lex.EscapeProtyleMarkers(node.Tokens)
+		} else {
+			node.Tokens = lex.EscapeMarkers(node.Tokens)
+		}
 		if nil != tree.Context.Tip && tree.Context.Tip.IsBlock() && nil != n.Parent && atom.Span != n.Parent.DataAtom && 1 > len(lex.TrimWhitespace(node.Tokens)) {
 			// 块级节点下非 span 包裹需要忽略空白
 			// 剪藏时列表下方块缩进不正确 https://github.com/siyuan-note/siyuan/issues/6650

lex/token.go

@@ -423,7 +423,7 @@ func IsCommonInlineMarker(token byte) bool {
 
 func IsProtyleInlineMarker(token byte) bool {
 	switch token {
-	case ItemAsterisk, ItemUnderscore, ItemBackslash, ItemBacktick, ItemTilde, ItemDollar, ItemEqual, ItemCaret:
+	case ItemAsterisk, ItemUnderscore, ItemBackslash, ItemBacktick, ItemTilde, ItemDollar, ItemEqual, ItemCaret, ItemLess, ItemGreater:
 		return true
 	default:
 		return false

protyle.go

@@ -210,6 +210,11 @@ func (lute *Lute) BlockDOM2Content(htmlStr string) (text string) {
 	return tree.Root.Content()
 }
 
+func (lute *Lute) BlockDOM2EscapeMarkerContent(htmlStr string) (text string) {
+	tree := lute.BlockDOM2Tree(htmlStr)
+	return tree.Root.EscapeMarkerContent()
+}
+
 func (lute *Lute) Tree2BlockDOM(tree *parse.Tree, options *render.Options) (vHTML string) {
 	renderer := render.NewProtyleRenderer(tree, options)
 	output := renderer.Render()

test/h2m_test.go

@@ -18,6 +18,7 @@ import (
 
 var html2MdTests = []parseTest{
 
+	{"78", "foo $bar^2$ baz^3", "foo \\$bar^2\\$ baz^3\n"},
 	{"77", "<p>在做拉丁字母间距设计的时候，比较好的方法是从 <code class=\"hljs language-undefined\">n</code> 和 <code class=\"hljs language-undefined\">o</code> 起步。一旦这两个字母的边距设置合理，剩下的就相对简单了。所以，我建议你键入一些类似 <code class=\"hljs language-undefined\">noononno</code> 的组合，然后开始摆弄它们的边距设置，直到你满意为止。你可以在灰色的信息面板（显示 &gt; 显示信息，Cmd-Shift-I）中检查任意字符形的边距和字宽设置：</p>\n<figure><div class=\"RatioContainer\">\n<picture class=\"js-only\"><source data-srcset=\"https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-320x-q80.webp 320w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-640x-q80.webp 640w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-1280x-q80.webp 1280w\" type=\"image/webp\" sizes=\"360px\" srcset=\"https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-320x-q80.webp 320w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-640x-q80.webp 640w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-1280x-q80.webp 1280w\"><source data-srcset=\"https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-320x-q80.png 320w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-640x-q80.png 640w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-1280x-q80.png 1280w\" type=\"image/png\" sizes=\"360px\" srcset=\"https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-320x-q80.png 320w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-640x-q80.png 640w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-1280x-q80.png 1280w\"><img alt=\"\" class=\"lazyautosizes lazyloaded\" data-sizes=\"auto\" data-srcset=\"https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-320x-q80.png 320w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-640x-q80.png 640w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-1280x-q80.png 1280w\" height=\"64\" width=\"360\" sizes=\"360px\" srcset=\"https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-320x-q80.png 320w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-640x-q80.png 640w, https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-1280x-q80.png 1280w\"></picture>\n</div></figure>", "在做拉丁字母间距设计的时候，比较好的方法是从 `n` 和 `o` 起步。一旦这两个字母的边距设置合理，剩下的就相对简单了。所以，我建议你键入一些类似 `noononno` 的组合，然后开始摆弄它们的边距设置，直到你满意为止。你可以在灰色的信息面板（显示 > 显示信息，Cmd-Shift-I）中检查任意字符形的边距和字宽设置：\n\n![](https://cdn2.glyphsapp.com/media/pages/learn/spacing/6e6a8e08ef-1624987024/spacing-1-1280x-q80.png)\n"},
 	{"76", "<div>\n    <ul>\n        <li>foo</li>\n    </ul>\n    <h1>bar</h1>\n</div>", "* foo\n\n# bar\n"},
 	{"75", "foo <mark>bar</mark> baz", "foo ==bar== baz\n"},

test/p2p_test.go

@@ -205,3 +205,35 @@ func TestBlockDOM2Content(t *testing.T) {
 		}
 	}
 }
+
+var blockDOM2EscapeMarkerContent = []parseTest{
+
+	{"1", "<div data-node-id=\"20230517211428-a5q3t4e\" data-node-index=\"1\" data-type=\"NodeParagraph\" class=\"p\" updated=\"20230517211428\"><div contenteditable=\"true\" spellcheck=\"false\">foo <span data-type=\"inline-math\" data-subtype=\"math\" data-content=\"bar^2\" contenteditable=\"false\" class=\"render-node\"></span> baz^3</div><div class=\"protyle-attr\" contenteditable=\"false\">​</div></div>", "foo bar\\^2 baz\\^3"},
+	{"0", "<div data-node-id=\"20230517201616-1i9a98t\" data-node-index=\"1\" data-type=\"NodeParagraph\" class=\"p\" updated=\"20230517201616\"><div contenteditable=\"true\" spellcheck=\"false\">Since <span data-type=\"inline-math\" data-subtype=\"math\" data-content=\"0 \\leqq s-r&lt;d\" contenteditable=\"false\" class=\"render-node\"></span> we must have <span data-type=\"inline-math\" data-subtype=\"math\" data-content=\"s-r=0\" contenteditable=\"false\" class=\"render-node\"></span>. The cyclic subgroup generated by <span data-type=\"inline-math\" data-subtype=\"math\" data-content=\"a\" contenteditable=\"false\" class=\"render-node\"></span> has order <span data-type=\"inline-math\" data-subtype=\"math\" data-content=\"d\" contenteditable=\"false\" class=\"render-node\"></span>.</div><div class=\"protyle-attr\" contenteditable=\"false\">​</div></div>", "Since 0 \\\\leqq s-r\\<d we must have s-r\\=0. The cyclic subgroup generated by a has order d."},
+}
+
+func TestBlockDOM2EscapeMarkerContent(t *testing.T) {
+	luteEngine := lute.New()
+	luteEngine.SetProtyleWYSIWYG(true)
+	luteEngine.ParseOptions.Mark = true
+	luteEngine.ParseOptions.BlockRef = true
+	luteEngine.SetKramdownIAL(true)
+	luteEngine.ParseOptions.SuperBlock = true
+	luteEngine.SetLinkBase("/siyuan/0/测试笔记/")
+	luteEngine.SetTag(true)
+	luteEngine.SetSub(true)
+	luteEngine.SetSup(true)
+	luteEngine.SetGitConflict(true)
+	luteEngine.SetIndentCodeBlock(false)
+	luteEngine.SetEmojiSite("http://127.0.0.1:6806/stage/protyle/images/emoji")
+	luteEngine.SetAutoSpace(true)
+	luteEngine.SetParagraphBeginningSpace(true)
+	luteEngine.SetFileAnnotationRef(true)
+
+	for _, test := range blockDOM2EscapeMarkerContent {
+		result := luteEngine.BlockDOM2EscapeMarkerContent(test.from)
+		if test.to != result {
+			t.Fatalf("test case [%s] failed\nexpected\n\t%q\ngot\n\t%q\noriginal html\n\t%q", test.name, test.to, result, test.from)
+		}
+	}
+}