prosemirror html conversion unwrapping + table rows collection

Signed-off-by: Émile Ré <emile@getprobo.com>

pkg/prosemirror/html_block.go

@@ -225,6 +225,16 @@ func (c *htmlBlockConverter) convertBlockElement(n *html.Node) ([]Node, error) {
 }
 
 func (c *htmlBlockConverter) unwrapBlockElement(n *html.Node) ([]Node, error) {
+	if !hasBlockElementChild(n) {
+		inlines, err := c.convertInlineFragments(n)
+		if err != nil {
+			return nil, err
+		}
+		if len(inlines) == 0 {
+			return nil, nil
+		}
+		return []Node{{Type: NodeParagraph, Content: inlines}}, nil
+	}
 	return c.convertBlockChildren(n)
 }
 
@@ -395,26 +405,46 @@ func hasBlockElementChild(n *html.Node) bool {
 func blockTagName(name string) bool {
 	switch name {
 	case "p", "div", "blockquote", "pre", "ul", "ol", "table",
-		"h1", "h2", "h3", "h4", "h5", "h6", "hr":
+		"h1", "h2", "h3", "h4", "h5", "h6", "hr",
+		"section", "article", "aside", "main", "header", "footer", "nav",
+		"figure", "center",
+		"html", "body", "head":
 		return true
 	default:
 		return false
 	}
 }
 
-func (c *htmlBlockConverter) convertTable(n *html.Node) ([]Node, error) {
+// collectTableRows returns <tr> nodes that belong to this table only: direct
+// children of thead/tbody/tfoot, or direct <tr> children of the table element
+// (implicit tbody). It does not descend into cells or nested tables, so inner
+// tables cannot contribute rows to the outer table.
+func collectTableRows(table *html.Node) []*html.Node {
 	var rows []*html.Node
-	var collect func(*html.Node)
-	collect = func(x *html.Node) {
-		if x.Type == html.ElementNode && x.Data == "tr" {
-			rows = append(rows, x)
-			return
+	for ch := table.FirstChild; ch != nil; ch = ch.NextSibling {
+		if ch.Type != html.ElementNode {
+			continue
 		}
-		for ch := x.FirstChild; ch != nil; ch = ch.NextSibling {
-			collect(ch)
+		switch ch.Data {
+		case "thead", "tbody", "tfoot":
+			for tr := ch.FirstChild; tr != nil; tr = tr.NextSibling {
+				if tr.Type == html.ElementNode && tr.Data == "tr" {
+					rows = append(rows, tr)
+				}
+			}
+		case "tr":
+			rows = append(rows, ch)
+		case "caption", "colgroup", "col":
+			// Table metadata / columns; not row containers.
+		default:
+			// Ignore other direct children (e.g. invalid markup).
 		}
 	}
-	collect(n)
+	return rows
+}
+
+func (c *htmlBlockConverter) convertTable(n *html.Node) ([]Node, error) {
+	rows := collectTableRows(n)
 
 	var rowNodes []Node
 	for _, tr := range rows {

pkg/prosemirror/markdown_test.go

@@ -406,6 +406,42 @@ func TestParseMarkdown_BlockHTML(t *testing.T) {
 	assert.Equal(t, "block", *doc.Content[0].Content[0].Text)
 }
 
+func TestParseMarkdown_BlockHTMLDivPreservesInlineMarks(t *testing.T) {
+	t.Parallel()
+
+	doc, err := ParseMarkdown("<div>hello <strong>world</strong></div>\n")
+	require.NoError(t, err)
+	require.Len(t, doc.Content, 1)
+	p := doc.Content[0]
+	require.Equal(t, NodeParagraph, p.Type)
+
+	var foundStrong bool
+	for _, ch := range p.Content {
+		if ch.Type != NodeText || ch.Text == nil {
+			continue
+		}
+		if *ch.Text != "world" {
+			continue
+		}
+		require.Len(t, ch.Marks, 1)
+		assert.Equal(t, MarkStrong, ch.Marks[0].Type)
+		foundStrong = true
+	}
+	assert.True(t, foundStrong, "expected bold mark on 'world' inside a single paragraph")
+}
+
+func TestParseMarkdown_BlockHTMLDivAroundSectionWithParagraphs(t *testing.T) {
+	t.Parallel()
+
+	doc, err := ParseMarkdown("<div><section><p>a</p><p>b</p></section></div>\n")
+	require.NoError(t, err)
+	require.Len(t, doc.Content, 2)
+	assert.Equal(t, NodeParagraph, doc.Content[0].Type)
+	assert.Equal(t, NodeParagraph, doc.Content[1].Type)
+	assert.Equal(t, "a", *doc.Content[0].Content[0].Text)
+	assert.Equal(t, "b", *doc.Content[1].Content[0].Text)
+}
+
 func TestParseMarkdown_BlockHTMLWithClosureLine(t *testing.T) {
 	t.Parallel()
 
@@ -467,6 +503,19 @@ func TestParseMarkdown_BlockHTMLTable(t *testing.T) {
 	assert.Equal(t, 1, tdAttrs.Rowspan)
 }
 
+func TestParseMarkdown_BlockHTMLNestedTableDoesNotHoistInnerRows(t *testing.T) {
+	t.Parallel()
+
+	md := "<table><tr><td><table><tr><td>inner</td></tr></table></td></tr></table>\n"
+	doc, err := ParseMarkdown(md)
+	require.NoError(t, err)
+	require.Len(t, doc.Content, 1)
+	assert.Equal(t, NodeTable, doc.Content[0].Type)
+	// Outer table must have exactly one row; inner <tr> must not become a second outer row.
+	require.Len(t, doc.Content[0].Content, 1)
+	assert.Equal(t, NodeTableRow, doc.Content[0].Content[0].Type)
+}
+
 func TestParseMarkdown_BlockHTMLTableCellSpans(t *testing.T) {
 	t.Parallel()