fix: normalizer stripping empty list elements

Author: hejsztynx

cpp/parser/GumboNormalizer.c

@@ -546,6 +546,7 @@ typedef struct {
   GumboNode **nested_lists;
   int *nested_count;
   int max_nested;
+  bool has_emitted;
 } li_ctx_t;
 
 static void flatten_li_node(GumboNode *node, buffer_t *ib, buffer_t *out,
@@ -562,6 +563,7 @@ static void flush_li_buffer(buffer_t *ib, buffer_t *out, li_ctx_t *ctx) {
   emit_styles_close(out, ctx->styles);
   buffer_append_str(out, "</li>");
   buffer_clear(ib);
+  ctx->has_emitted = true;
 }
 
 static void flatten_li_children(GumboNode *node, buffer_t *ib, buffer_t *out,
@@ -885,6 +887,14 @@ static void walk_node(GumboNode *node, buffer_t *out) {
       li_ctx_t ctx = {el, es, nested_lists, &nested_count, 16};
       flatten_li_children(node, &li_ib, out, &ctx);
       flush_li_buffer(&li_ib, out, &ctx);
+
+      /* if nothing emitted - the <li> is empty, we add it manually */
+      if (!ctx.has_emitted) {
+        buffer_append_str(out, "<li");
+        emit_attributes(el, "li", out);
+        buffer_append_str(out, "></li>");
+      }
+
       free(li_ib.data);
       for (int k = 0; k < nested_count; k++)
         walk_children(nested_lists[k], out);

cpp/tests/GumboParserTest.cpp

@@ -484,19 +484,22 @@ TEST(GumboParserTest, MSWordCheckboxList) {
 }
 
 TEST(GumboParserTest, EmptyListItems) {
-  EXPECT_EQ(GumboParser::normalizeHtml(
-                "<ul><li></li><li>first</li><li></li><li>second</li><li></li>"
-                "</ul>"),
-            "<ul><li>first</li><li>second</li></ul>");
-  EXPECT_EQ(GumboParser::normalizeHtml(
-                "<ol><li></li><li>first</li><li></li><li>second</li><li></li>"
-                "</ol>"),
-            "<ol><li>first</li><li>second</li></ol>");
-  EXPECT_EQ(GumboParser::normalizeHtml(
-                "<ul data-type=\"checkbox\"><li checked></li><li>first</li><li>"
-                "</li><li checked>second</li><li></li><li></li></ul>"),
-            "<ul data-type=\"checkbox\"><li>first</li><li "
-            "checked>second</li></ul>");
+  EXPECT_EQ(GumboParser::normalizeHtml("<ul><li></li><li>first</li><li></"
+                                       "li><li>second</li><li></li><li></li>"
+                                       "</ul>"),
+            "<ul><li></li><li>first</li><li></li><li>second</li><li></li><li></"
+            "li></ul>");
+  EXPECT_EQ(GumboParser::normalizeHtml("<ol><li></li><li>first</li><li></"
+                                       "li><li>second</li><li></li><li></li>"
+                                       "</ol>"),
+            "<ol><li></li><li>first</li><li></li><li>second</li><li></li><li></"
+            "li></ol>");
+  EXPECT_EQ(
+      GumboParser::normalizeHtml(
+          "<ul data-type=\"checkbox\"><li checked></li><li>first</li><li>"
+          "</li><li checked>second</li><li></li><li></li></ul>"),
+      "<ul data-type=\"checkbox\"><li checked></li><li>first</li><li></li><li "
+      "checked>second</li><li></li><li></li></ul>");
 }
 
 TEST(GumboParserTest, BrRemappings) {

src/web/__tests__/htmlNormalizer.test.ts

@@ -420,16 +420,16 @@ describe('htmlNormalizer', () => {
   describe('EmptyListItems', () => {
     test.each([
       [
-        '<ul><li></li><li>first</li><li></li><li>second</li><li></li></ul>',
-        '<ul><li>first</li><li>second</li></ul>',
+        '<ul><li></li><li>first</li><li></li><li>second</li><li></li><li></li></ul>',
+        '<ul><li></li><li>first</li><li></li><li>second</li><li></li><li></li></ul>',
       ],
       [
-        '<ol><li></li><li>first</li><li></li><li>second</li><li></li></ol>',
-        '<ol><li>first</li><li>second</li></ol>',
+        '<ol><li></li><li>first</li><li></li><li>second</li><li></li><li></li></ol>',
+        '<ol><li></li><li>first</li><li></li><li>second</li><li></li><li></li></ol>',
       ],
       [
         '<ul data-type="checkbox"><li checked></li><li>first</li><li></li><li checked>second</li><li></li><li></li></ul>',
-        '<ul data-type="checkbox"><li>first</li><li checked>second</li></ul>',
+        '<ul data-type="checkbox"><li checked></li><li>first</li><li></li><li checked>second</li><li></li><li></li></ul>',
       ],
     ])('%s → %s', (input, expected) => {
       expect(normalizeHtml(input)).toBe(expected);

src/web/normalization/htmlNormalizer.ts

@@ -363,6 +363,7 @@ type LiCtx = {
   el: Element;
   styles: CssStyles;
   nestedLists: Element[];
+  hasEmitted: boolean;
 };
 
 function flushLiBuffer(
@@ -377,6 +378,7 @@ function flushLiBuffer(
   out.buf += emitStylesClose(ctx.styles);
   out.buf += '</li>';
   ib.buf = '';
+  ctx.hasEmitted = true;
 }
 
 function flattenLiChildren(
@@ -605,9 +607,15 @@ function walkNode(node: Node, out: { buf: string }): void {
   if (outName === 'li') {
     const nestedLists: Element[] = [];
     const liIb = { buf: '' };
-    const ctx: LiCtx = { el: node, styles: es, nestedLists };
+    const ctx: LiCtx = { el: node, styles: es, nestedLists, hasEmitted: false };
     flattenLiChildren(node, liIb, out, ctx);
     flushLiBuffer(liIb, out, ctx);
+
+    // if nothing emitted - the <li> is empty, we add it manually
+    if (!ctx.hasEmitted) {
+      out.buf += `<li${emitAttributes(ctx.el, 'li')}></li>`;
+    }
+
     for (const nl of nestedLists) walkChildren(nl, out);
     return;
   }