fix: imgs with empty src and no src are properly handled

Author: hejsztynx

cpp/parser/GumboNormalizer.c

@@ -491,7 +491,12 @@ static void emit_attributes(GumboElement *el, const char *tag_name,
   if (strcmp(tag_name, "a") == 0) {
     emit_one_attr(out, el, "href");
   } else if (strcmp(tag_name, "img") == 0) {
-    emit_one_attr(out, el, "src");
+    const char *src_val = get_attr(el, "src");
+    if (src_val && src_val[0]) {
+      emit_one_attr(out, el, "src");
+    } else {
+      buffer_append_str(out, " src=\"\"");
+    }
     emit_one_attr(out, el, "alt");
     emit_one_attr(out, el, "width");
     emit_one_attr(out, el, "height");

cpp/tests/GumboParserTest.cpp

@@ -282,6 +282,11 @@ TEST(GumboParserTest, EnrichedTagRemappings) {
   EXPECT_EQ(
       GumboParser::normalizeHtml("<img src='x' width='100' height='100' />"),
       "<img src=\"x\" width=\"100\" height=\"100\" />");
+  EXPECT_EQ(GumboParser::normalizeHtml("<img width=\"100\" height=\"100\" />"),
+            "<img src=\"\" width=\"100\" height=\"100\" />");
+  EXPECT_EQ(GumboParser::normalizeHtml(
+                "<img src=\"\" width=\"100\" height=\"100\" />"),
+            "<img src=\"\" width=\"100\" height=\"100\" />");
 
   // Lists
   EXPECT_EQ(GumboParser::normalizeHtml("<ul><li>x</li></ul>"),

ios/htmlParser/HtmlParser.mm

@@ -670,7 +670,7 @@ + (NSArray *_Nonnull)getTextAndStylesFromHtml:(NSString *_Nonnull)fixedHtml
       [styleArr addObject:@([ItalicStyle getType])];
     } else if ([tagName isEqualToString:@"img"]) {
       NSRegularExpression *srcRegex =
-          [NSRegularExpression regularExpressionWithPattern:@"src=\"([^\"]+)\""
+          [NSRegularExpression regularExpressionWithPattern:@"src=\"([^\"]*)\""
                                                     options:0
                                                       error:nullptr];
       NSTextCheckingResult *match =

src/web/__tests__/htmlNormalizer.test.ts

@@ -262,6 +262,14 @@ describe('htmlNormalizer', () => {
         "<img src='x' width='100' height='100' />",
         '<img src="x" width="100" height="100" />',
       ],
+      [
+        '<img width="100" height="100" />',
+        '<img src="" width="100" height="100" />',
+      ],
+      [
+        '<img src="" width="100" height="100" />',
+        '<img src="" width="100" height="100" />',
+      ],
 
       // Lists
       ['<ul><li>x</li></ul>', '<ul><li>x</li></ul>'],

src/web/normalization/htmlNormalizer.ts

@@ -267,7 +267,7 @@ function emitAttributes(el: Element, name: string): string {
       return emitOneAttr(el, 'href');
     case 'img':
       return (
-        emitOneAttr(el, 'src') +
+        (el.getAttribute('src') ? emitOneAttr(el, 'src') : ' src=""') +
         emitOneAttr(el, 'alt') +
         emitOneAttr(el, 'width') +
         emitOneAttr(el, 'height')