🎨 Protyle Markdown syntax switch no longer affects clip/paste HTML pa…

…rsing siyuan-note/siyuan#11604

ast/node.go

@@ -1128,5 +1128,11 @@ const (
 
 	NodeCustomBlock NodeType = 560 // 自定义块
 
+	// HTML 标签，在无法使用 Markdown 标记符的情况下直接使用 HTML 标签
+
+	NodeHTMLTag      NodeType = 570 // HTML 标签
+	NodeHTMLTagOpen  NodeType = 571 // 开始 HTML 标签
+	NodeHTMLTagClose NodeType = 572 // 结束 HTML 标签
+
 	NodeTypeMaxVal NodeType = 1024 // 节点类型最大值
 )

h2m.go

@@ -599,9 +599,13 @@ func (lute *Lute) genASTByDOM(n *html.Node, tree *parse.Tree) {
 			tree.Context.Tip.AppendChild(&ast.Node{Type: ast.NodeText, Tokens: util.StrToBytes(editor.Zwsp)})
 		}
 
-		node.Type = ast.NodeEmphasis
-		marker := "*"
-		node.AppendChild(&ast.Node{Type: ast.NodeEmA6kOpenMarker, Tokens: util.StrToBytes(marker)})
+		if !lute.ParseOptions.InlineAsterisk && !lute.ParseOptions.InlineUnderscore {
+			node.Type = ast.NodeHTMLTag
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagOpen, Tokens: util.StrToBytes("<em>")})
+		} else {
+			node.Type = ast.NodeEmphasis
+			node.AppendChild(&ast.Node{Type: ast.NodeEmA6kOpenMarker, Tokens: util.StrToBytes("*")})
+		}
 		tree.Context.Tip.AppendChild(node)
 		tree.Context.Tip = node
 		defer tree.Context.ParentTip()
@@ -619,9 +623,13 @@ func (lute *Lute) genASTByDOM(n *html.Node, tree *parse.Tree) {
 			tree.Context.Tip.AppendChild(&ast.Node{Type: ast.NodeText, Tokens: util.StrToBytes(editor.Zwsp)})
 		}
 
-		node.Type = ast.NodeStrong
-		marker := "**"
-		node.AppendChild(&ast.Node{Type: ast.NodeStrongA6kOpenMarker, Tokens: util.StrToBytes(marker)})
+		if !lute.ParseOptions.InlineAsterisk && !lute.ParseOptions.InlineUnderscore {
+			node.Type = ast.NodeHTMLTag
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagOpen, Tokens: util.StrToBytes("<strong>")})
+		} else {
+			node.Type = ast.NodeStrong
+			node.AppendChild(&ast.Node{Type: ast.NodeStrongA6kOpenMarker, Tokens: util.StrToBytes("**")})
+		}
 		tree.Context.Tip.AppendChild(node)
 		tree.Context.Tip = node
 		defer tree.Context.ParentTip()
@@ -813,28 +821,52 @@ func (lute *Lute) genASTByDOM(n *html.Node, tree *parse.Tree) {
 			}
 		}
 	case atom.Del, atom.S, atom.Strike:
-		node.Type = ast.NodeStrikethrough
-		marker := "~~"
-		node.AppendChild(&ast.Node{Type: ast.NodeStrikethrough2OpenMarker, Tokens: util.StrToBytes(marker)})
+		if !lute.ParseOptions.GFMStrikethrough {
+			node.Type = ast.NodeHTMLTag
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagOpen, Tokens: util.StrToBytes("<s>")})
+		} else {
+			node.Type = ast.NodeStrikethrough
+			node.AppendChild(&ast.Node{Type: ast.NodeStrikethrough2OpenMarker, Tokens: util.StrToBytes("~~")})
+		}
+		tree.Context.Tip.AppendChild(node)
+		tree.Context.Tip = node
+		defer tree.Context.ParentTip()
+	case atom.U:
+		node.Type = ast.NodeHTMLTag
+		node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagOpen, Tokens: util.StrToBytes("<u>")})
 		tree.Context.Tip.AppendChild(node)
 		tree.Context.Tip = node
 		defer tree.Context.ParentTip()
 	case atom.Mark:
-		node.Type = ast.NodeMark
-		marker := "=="
-		node.AppendChild(&ast.Node{Type: ast.NodeMark2OpenMarker, Tokens: util.StrToBytes(marker)})
+		if !lute.ParseOptions.Mark {
+			node.Type = ast.NodeHTMLTag
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagOpen, Tokens: util.StrToBytes("<mark>")})
+		} else {
+			node.Type = ast.NodeMark
+			node.AppendChild(&ast.Node{Type: ast.NodeMark2OpenMarker, Tokens: util.StrToBytes("==")})
+		}
 		tree.Context.Tip.AppendChild(node)
 		tree.Context.Tip = node
 		defer tree.Context.ParentTip()
 	case atom.Sup:
-		node.Type = ast.NodeSup
-		node.AppendChild(&ast.Node{Type: ast.NodeSupOpenMarker})
+		if !lute.ParseOptions.Sup {
+			node.Type = ast.NodeHTMLTag
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagOpen, Tokens: util.StrToBytes("<sup>")})
+		} else {
+			node.Type = ast.NodeSup
+			node.AppendChild(&ast.Node{Type: ast.NodeSupOpenMarker})
+		}
 		tree.Context.Tip.AppendChild(node)
 		tree.Context.Tip = node
 		defer tree.Context.ParentTip()
 	case atom.Sub:
-		node.Type = ast.NodeSub
-		node.AppendChild(&ast.Node{Type: ast.NodeSubOpenMarker})
+		if !lute.ParseOptions.Sub {
+			node.Type = ast.NodeHTMLTag
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagOpen, Tokens: util.StrToBytes("<sub>")})
+		} else {
+			node.Type = ast.NodeSub
+			node.AppendChild(&ast.Node{Type: ast.NodeSubOpenMarker})
+		}
 		tree.Context.Tip.AppendChild(node)
 		tree.Context.Tip = node
 		defer tree.Context.ParentTip()
@@ -1214,12 +1246,19 @@ func (lute *Lute) genASTByDOM(n *html.Node, tree *parse.Tree) {
 
 	switch n.DataAtom {
 	case atom.Em, atom.I:
-		marker := "*"
-		node.AppendChild(&ast.Node{Type: ast.NodeEmA6kCloseMarker, Tokens: util.StrToBytes(marker)})
+		if !lute.ParseOptions.InlineAsterisk && !lute.ParseOptions.InlineUnderscore {
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagClose, Tokens: util.StrToBytes("</em>")})
+		} else {
+			node.Type = ast.NodeEmphasis
+			node.AppendChild(&ast.Node{Type: ast.NodeEmA6kCloseMarker, Tokens: util.StrToBytes("*")})
+		}
 		appendSpace(n, tree, lute)
 	case atom.Strong, atom.B:
-		marker := "**"
-		node.AppendChild(&ast.Node{Type: ast.NodeStrongA6kCloseMarker, Tokens: util.StrToBytes(marker)})
+		if !lute.ParseOptions.InlineAsterisk && !lute.ParseOptions.InlineUnderscore {
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagClose, Tokens: util.StrToBytes("</strong>")})
+		} else {
+			node.AppendChild(&ast.Node{Type: ast.NodeStrongA6kCloseMarker, Tokens: util.StrToBytes("**")})
+		}
 		appendSpace(n, tree, lute)
 	case atom.A:
 		node.AppendChild(&ast.Node{Type: ast.NodeCloseBracket})
@@ -1232,18 +1271,34 @@ func (lute *Lute) genASTByDOM(n *html.Node, tree *parse.Tree) {
 		}
 		node.AppendChild(&ast.Node{Type: ast.NodeCloseParen})
 	case atom.Del, atom.S, atom.Strike:
-		marker := "~~"
-		node.AppendChild(&ast.Node{Type: ast.NodeStrikethrough2CloseMarker, Tokens: util.StrToBytes(marker)})
+		if !lute.ParseOptions.GFMStrikethrough {
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagClose, Tokens: util.StrToBytes("</s>")})
+		} else {
+			node.AppendChild(&ast.Node{Type: ast.NodeStrikethrough2CloseMarker, Tokens: util.StrToBytes("~~")})
+		}
 		appendSpace(n, tree, lute)
+	case atom.U:
+		node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagClose, Tokens: util.StrToBytes("</u>")})
 	case atom.Mark:
-		marker := "=="
-		node.AppendChild(&ast.Node{Type: ast.NodeMark2CloseMarker, Tokens: util.StrToBytes(marker)})
+		if !lute.ParseOptions.Mark {
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagClose, Tokens: util.StrToBytes("</mark>")})
+		} else {
+			node.AppendChild(&ast.Node{Type: ast.NodeMark2CloseMarker, Tokens: util.StrToBytes("==")})
+		}
 		appendSpace(n, tree, lute)
 	case atom.Sup:
-		node.AppendChild(&ast.Node{Type: ast.NodeSupCloseMarker})
+		if !lute.ParseOptions.Sup {
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagClose, Tokens: util.StrToBytes("</sup>")})
+		} else {
+			node.AppendChild(&ast.Node{Type: ast.NodeSupCloseMarker})
+		}
 		appendSpace(n, tree, lute)
 	case atom.Sub:
-		node.AppendChild(&ast.Node{Type: ast.NodeSubCloseMarker})
+		if !lute.ParseOptions.Sub {
+			node.AppendChild(&ast.Node{Type: ast.NodeHTMLTagClose, Tokens: util.StrToBytes("</sub>")})
+		} else {
+			node.AppendChild(&ast.Node{Type: ast.NodeSubCloseMarker})
+		}
 		appendSpace(n, tree, lute)
 	case atom.Details:
 		tree.Context.ParentTip()

parse/inline_html.go

@@ -121,25 +121,49 @@ func (t *Tree) parseInlineHTML(ctx *InlineContext) (ret *ast.Node) {
 		}
 		ctx.pos += len(tags)
 
-		if t.Context.ParseOption.ProtyleWYSIWYG {
+		if t.Context.ParseOption.ProtyleWYSIWYG && nil == ctx.brackets /* 方括号后跟标签的情况不处理 [<strong> */ {
 			if bytes.EqualFold(tags, []byte("<br />")) || bytes.EqualFold(tags, []byte("<br/>")) || bytes.EqualFold(tags, []byte("<br>")) {
 				ret = &ast.Node{Type: ast.NodeBr}
 				return
 			} else if bytes.HasPrefix(tags, []byte("<span data-type=")) {
 				ret = t.processSpanTag(tags, "<span data-type=", "</span>", ctx)
 				return
-			} else if bytes.Equal(tags, []byte("<kbd>")) {
+			} else if bytes.EqualFold(tags, []byte("<kbd>")) {
 				ret = t.processSpanTag(tags, "<kbd>", "</kbd>", ctx)
 				return
-			} else if bytes.Equal(tags, []byte("<u>")) {
+			} else if bytes.EqualFold(tags, []byte("<u>")) {
 				ret = t.processSpanTag(tags, "<u>", "</u>", ctx)
 				return
-			} else if bytes.Equal(tags, []byte("<sup>")) {
+			} else if bytes.EqualFold(tags, []byte("<sup>")) {
 				ret = t.processSpanTag(tags, "<sup>", "</sup>", ctx)
 				return
-			} else if bytes.Equal(tags, []byte("<sub>")) {
+			} else if bytes.EqualFold(tags, []byte("<sub>")) {
 				ret = t.processSpanTag(tags, "<sub>", "</sub>", ctx)
 				return
+			} else if bytes.EqualFold(tags, []byte("<mark>")) {
+				ret = t.processSpanTag(tags, "<mark>", "</mark>", ctx)
+				return
+			} else if bytes.EqualFold(tags, []byte("<s>")) {
+				ret = t.processSpanTag(tags, "<s>", "</s>", ctx)
+				return
+			} else if bytes.EqualFold(tags, []byte("<del>")) {
+				ret = t.processSpanTag(tags, "<del>", "</del>", ctx)
+				return
+			} else if bytes.EqualFold(tags, []byte("<strike>")) {
+				ret = t.processSpanTag(tags, "<strike>", "</strike>", ctx)
+				return
+			} else if bytes.EqualFold(tags, []byte("<em>")) {
+				ret = t.processSpanTag(tags, "<em>", "</em>", ctx)
+				return
+			} else if bytes.EqualFold(tags, []byte("<i>")) {
+				ret = t.processSpanTag(tags, "<i>", "</i>", ctx)
+				return
+			} else if bytes.EqualFold(tags, []byte("<strong>")) {
+				ret = t.processSpanTag(tags, "<strong>", "</strong>", ctx)
+				return
+			} else if bytes.EqualFold(tags, []byte("<b>")) {
+				ret = t.processSpanTag(tags, "<b>", "</b>", ctx)
+				return
 			}
 		}
 		ret = &ast.Node{Type: ast.NodeInlineHTML, Tokens: tags}
@@ -176,12 +200,19 @@ func (t *Tree) processSpanTag(tags []byte, startTag, endTag string, ctx *InlineC
 
 	var typ string
 	startTagLen := len(startTag)
-	if "<kbd>" == startTag || "<u>" == startTag || "<sup>" == startTag || "<sub>" == startTag {
+	if "<kbd>" == startTag || "<u>" == startTag || "<sup>" == startTag || "<sub>" == startTag || "<mark>" == startTag || "<s>" == startTag || "<del>" == startTag || "<strike>" == startTag || "<em>" == startTag || "<i>" == startTag || "<strong>" == startTag || "<b>" == startTag {
 		if !t.Context.ParseOption.HTMLTag2TextMark {
 			ret = &ast.Node{Type: ast.NodeInlineHTML, Tokens: tags}
 			return
 		}
 		typ = node.Data
+		if "b" == typ {
+			typ = "strong"
+		} else if "i" == typ {
+			typ = "em"
+		} else if "del" == typ || "strike" == typ {
+			typ = "s"
+		}
 	} else { // <span data-type="a">
 		typ = string(tags[startTagLen+1:])
 		typ = typ[:strings.Index(typ, "\"")]
@@ -612,7 +643,18 @@ func SetTextMarkNode(node *ast.Node, n *html.Node, options *Options) {
 		if n.DataAtom == atom.Span {
 			dataType = "text"
 		} else {
-			dataType = n.DataAtom.String()
+			if "" != node.TextMarkType {
+				dataType = node.TextMarkType
+			} else {
+				dataType = n.DataAtom.String()
+				if "b" == dataType {
+					dataType = "strong"
+				} else if "i" == dataType {
+					dataType = "em"
+				} else if "del" == dataType || "strike" == dataType {
+					dataType = "s"
+				}
+			}
 		}
 	}
 	node.TextMarkType = dataType

parse/parse.go

@@ -405,9 +405,8 @@ type Options struct {
 	DataImage bool
 	// TextMark 设置是否打开通用行级节点解析支持。
 	TextMark bool
-	// HTMLTag2TextMark 设置是否打开 HTML 某些标签解析为 TextMark 节点支持。目前仅支持 <u>、<kbd>、<sub> 和 <sup> 标签。
-	// 这个开关主要用于兼容 Markdown 输入 API 上 https://github.com/siyuan-note/siyuan/issues/6039
-	// 不用于 Protyle 自旋过程 https://github.com/siyuan-note/siyuan/issues/5877
+	// HTMLTag2TextMark 设置是否打开 HTML 某些标签解析为 TextMark 节点支持。
+	// 目前仅支持 <u>、<kbd>、<sub>、<sup>、<strong>/<b>、<em>/<i>、<s>/<del>/<strike> 和 <mark>。
 	HTMLTag2TextMark bool
 	// Spin 设置是否打开自旋解析支持，该选项仅用于 Spin 内部过程，设置时请注意使用场景。
 	//

render/format_renderer.go

@@ -167,9 +167,30 @@ func NewFormatRenderer(tree *parse.Tree, options *Options) *FormatRenderer {
 	ret.RendererFuncs[ast.NodeTextMark] = ret.renderTextMark
 	ret.RendererFuncs[ast.NodeAttributeView] = ret.renderAttributeView
 	ret.RendererFuncs[ast.NodeCustomBlock] = ret.renderCustomBlock
+	ret.RendererFuncs[ast.NodeHTMLTag] = ret.renderHTMLTag
+	ret.RendererFuncs[ast.NodeHTMLTagOpen] = ret.renderHTMLTagOpen
+	ret.RendererFuncs[ast.NodeHTMLTagClose] = ret.renderHTMLTagClose
 	return ret
 }
 
+func (r *FormatRenderer) renderHTMLTag(node *ast.Node, entering bool) ast.WalkStatus {
+	return ast.WalkContinue
+}
+
+func (r *FormatRenderer) renderHTMLTagOpen(node *ast.Node, entering bool) ast.WalkStatus {
+	if entering {
+		r.Write(node.Tokens)
+	}
+	return ast.WalkContinue
+}
+
+func (r *FormatRenderer) renderHTMLTagClose(node *ast.Node, entering bool) ast.WalkStatus {
+	if entering {
+		r.Write(node.Tokens)
+	}
+	return ast.WalkContinue
+}
+
 func (r *FormatRenderer) renderCustomBlock(node *ast.Node, entering bool) ast.WalkStatus {
 	if entering {
 		r.Newline()

test/h2m_test.go

@@ -257,10 +257,46 @@ var html2MdTests = []parseTest{
 func TestHTML2Md(t *testing.T) {
 	luteEngine := lute.New()
 	luteEngine.SetAutoSpace(true)
+	luteEngine.SetSup(true)
+	luteEngine.SetSub(true)
+	luteEngine.SetMark(true)
+	luteEngine.SetGFMStrikethrough(true)
+	luteEngine.SetInlineAsterisk(true)
+	luteEngine.SetInlineUnderscore(true)
 	for _, test := range html2MdTests {
 		md := luteEngine.HTML2Md(test.from)
 		if test.to != md {
 			t.Fatalf("test case [%s] failed\nexpected\n\t%q\ngot\n\t%q\noriginal html\n\t%q", test.name, test.to, md, test.from)
 		}
 	}
 }
+
+var html2MdDisableSyntaxTests = []parseTest{
+
+	{"7", "<strong><em>foo</em></strong> <em><strong>bar</strong></em>", "<strong><em>foo</em></strong> <em><strong>bar</strong></em>\n"},
+	{"6", "<kbd>foo</kbd>", "<kbd>foo</kbd>\n"},
+	{"5", "<mark>foo</mark>", "<mark>foo</mark>\n"},
+	{"4", "<u>foo</u>", "<u>foo</u>\n"},
+	{"3", "<s>foo</s> <del>bar</del> <strike>baz</strike>", "<s>foo</s> <s>bar</s> <s>baz</s>\n"},
+	{"2", "foo<strong>bar</strong> <b>baz</b>", "foo<strong>bar</strong> <strong>baz</strong>\n"},
+	{"1", "foo<em>bar</em> <i>baz</i>", "foo<em>bar</em> <em>baz</em>\n"},
+	{"0", "foo<sup>bar</sup></span><sub>baz</sub>", "foo<sup>bar</sup><sub>baz</sub>\n"},
+}
+
+func TestHTML2MdDisableSyntax(t *testing.T) {
+	luteEngine := lute.New()
+	luteEngine.SetAutoSpace(true)
+	luteEngine.SetSup(false)
+	luteEngine.SetSub(false)
+	luteEngine.SetMark(false)
+	luteEngine.SetGFMStrikethrough(false)
+	luteEngine.SetInlineAsterisk(false)
+	luteEngine.SetInlineUnderscore(false)
+	luteEngine.SetHTMLTag2TextMark(true)
+	for _, test := range html2MdDisableSyntaxTests {
+		md := luteEngine.HTML2Md(test.from)
+		if test.to != md {
+			t.Fatalf("test case [%s] failed\nexpected\n\t%q\ngot\n\t%q\noriginal html\n\t%q", test.name, test.to, md, test.from)
+		}
+	}
+}