domutils: added MoveListItems

MoveListItems moves non-"li" nodes into the previous "li" nodes

internal/domutils/list_items.go

@@ -0,0 +1,53 @@
+package domutils
+
+import (
+	"context"
+	"strings"
+
+	"github.com/JohannesKaufmann/dom"
+	"golang.org/x/net/html"
+	"golang.org/x/net/html/atom"
+)
+
+// MoveListItems moves non-"li" nodes into the previous "li" nodes.
+func MoveListItems(ctx context.Context, n *html.Node) {
+	if n.Type == html.ElementNode && (n.Data == "ol" || n.Data == "ul") {
+		var previousLi *html.Node
+
+		// Collect children to avoid modifying the slice while iterating.
+		children := dom.AllChildNodes(n)
+
+		for _, child := range children {
+			if child.Type == html.ElementNode && child.Data == "li" {
+				previousLi = child
+			} else if child.Type == html.TextNode && strings.TrimSpace(child.Data) == "" {
+				// Skip the node, probably just formatting of code
+			} else {
+				// We expect that inside an "ol"/"ul" there are *only* "li" nodes.
+				// But sometimes that is not the case...
+
+				if previousLi != nil {
+					// There is a previous "li" node,
+					// so we move this content into the other "li" node.
+					n.RemoveChild(child)
+
+					previousLi.AppendChild(child)
+				} else {
+					// There is no previous "li" node,
+					// so we wrap this node with it's own "li" node.
+
+					newNode := &html.Node{
+						Type:     html.ElementNode,
+						DataAtom: atom.Li,
+						Data:     "li",
+					}
+					previousLi = dom.WrapNode(child, newNode)
+				}
+			}
+		}
+	}
+
+	for c := n.FirstChild; c != nil; c = c.NextSibling {
+		MoveListItems(ctx, c)
+	}
+}

internal/domutils/list_items_test.go

@@ -0,0 +1,123 @@
+package domutils
+
+import (
+	"context"
+	"testing"
+
+	"github.com/JohannesKaufmann/html-to-markdown/v2/internal/tester"
+)
+
+func TestMoveListItems(t *testing.T) {
+	runs := []struct {
+		desc     string
+		input    string
+		expected string
+	}{
+		{
+			desc:  "not needed in normal list",
+			input: "<div><ul><li>A</li><li>B</li><li>C</li></ul></div>",
+			expected: `
+├─body
+│ ├─div
+│ │ ├─ul
+│ │ │ ├─li
+│ │ │ │ ├─#text "A"
+│ │ │ ├─li
+│ │ │ │ ├─#text "B"
+│ │ │ ├─li
+│ │ │ │ ├─#text "C"
+			`,
+		},
+		{
+			desc:  "#text moves into the previous li",
+			input: "<ul><li>A</li>B</ul>",
+			expected: `
+├─body
+│ ├─ul
+│ │ ├─li
+│ │ │ ├─#text "A"
+│ │ │ ├─#text "B"
+			`,
+		},
+		{
+			desc:  "div moves into the previous li",
+			input: "<ul><li>A</li><div>B</div></ul>",
+			expected: `
+├─body
+│ ├─ul
+│ │ ├─li
+│ │ │ ├─#text "A"
+│ │ │ ├─div
+│ │ │ │ ├─#text "B"
+			`,
+		},
+		{
+			desc:  "ol moves into the previous li",
+			input: "<ul><li>A</li><ol><li>B</li></ol></ul>",
+			expected: `
+├─body
+│ ├─ul
+│ │ ├─li
+│ │ │ ├─#text "A"
+│ │ │ ├─ol
+│ │ │ │ ├─li
+│ │ │ │ │ ├─#text "B"
+			`,
+		},
+		{
+			desc:  "no existing li",
+			input: "<ul><span>A</span><span>B</span></ul>",
+			expected: `
+├─body
+│ ├─ul
+│ │ ├─li
+│ │ │ ├─span
+│ │ │ │ ├─#text "A"
+│ │ │ ├─span
+│ │ │ │ ├─#text "B"
+			`,
+		},
+		{
+			desc: "basic moved list",
+			input: `
+<ol>
+	<li>One</li>
+	<li>Two</li>
+	<ol>
+		<li>Two point one</li>
+		<li>Two point two</li>
+	</ol>
+</ol>
+			`,
+			expected: `
+├─body
+│ ├─ol
+│ │ ├─#text "\n\t"
+│ │ ├─li
+│ │ │ ├─#text "One"
+│ │ ├─#text "\n\t"
+│ │ ├─li
+│ │ │ ├─#text "Two"
+│ │ │ ├─ol
+│ │ │ │ ├─#text "\n\t\t"
+│ │ │ │ ├─li
+│ │ │ │ │ ├─#text "Two point one"
+│ │ │ │ ├─#text "\n\t\t"
+│ │ │ │ ├─li
+│ │ │ │ │ ├─#text "Two point two"
+│ │ │ │ ├─#text "\n\t"
+│ │ ├─#text "\n\t"
+│ │ ├─#text "\n"
+			`,
+		},
+	}
+	for _, run := range runs {
+		t.Run(run.desc, func(t *testing.T) {
+			doc := tester.Parse(t, run.input, "")
+
+			MoveListItems(context.TODO(), doc)
+
+			tester.ExpectRepresentation(t, doc, "output", run.expected)
+		})
+	}
+}

plugin/commonmark/handle_pre_render.go

@@ -85,7 +85,10 @@ func (c *commonmark) handlePreRender(ctx converter.Context, doc *html.Node) {
 	domutils.RemoveRedundant(doc, nameIsBothLink)
 	domutils.SwapTags(ctx, doc, nameIsBoldOrItalic, nameIsLink)
 
-	// - - - Headings - - - //
+	// - - - Heading - - - //
 	domutils.SwapTags(ctx, doc, nameIsLink, nameIsHeading)
 	domutils.LeafBlockAlternatives(ctx, doc)
+
+	// - - - List - - - //
+	domutils.MoveListItems(ctx, doc)
 }

plugin/commonmark/testdata/GoldenFiles/link.out.md

@@ -79,7 +79,8 @@ Wir freuen uns über eine [Mail](mailto:[email protected]?body=Hello%0AJohannes)!
 <!--list with link-->
 
 - [a(b)\[c\]](/page.html)
-- [a\]](/page.html)
+
+  [a\]](/page.html)
 
 <!--TODO: list with paragraph-->
 

plugin/commonmark/testdata/GoldenFiles/list.out.md

@@ -145,15 +145,20 @@ text between
 <!-- with other elements inside the list -->
 
 1. A 1 (div)
-2. A 2 (#text)
-3. A 3 (li)
-4. A 4 (#text)
-5. 1. B 1 (li)
-   2. 1. C 1 (li)
-      2. C 2 (div)
-      3. C 3 (div)
-   3. B 2 (div)
-   4. B 3 (li)
+
+   A 2 (#text)
+2. A 3 (li) A 4 (#text)
+
+   1. B 1 (li)
+
+      1. C 1 (li)
+
+         C 2 (div)
+
+         C 3 (div)
+
+      B 2 (div)
+   2. B 3 (li)
 
 <!--THE END-->