refactor: cleanup

Author: hejsztynx

apps/example/ios/Podfile.lock

@@ -2026,7 +2026,7 @@ EXTERNAL SOURCES:
 
 SPEC CHECKSUMS:
   FBLazyVector: c00c20551d40126351a6783c47ce75f5b374851b
-  hermes-engine: d80056af9dcb5ba3db80f37943f96751037c7381
+  hermes-engine: b4dad6ba67535bb03c8ff1006b337cba14db16cb
   RCTDeprecation: 3bb167081b134461cfeb875ff7ae1945f8635257
   RCTRequired: 74839f55d5058a133a0bc4569b0afec750957f64
   RCTSwiftUI: 87a316382f3eab4dd13d2a0d0fd2adcce917361a
@@ -2035,7 +2035,7 @@ SPEC CHECKSUMS:
   React: 1b1536b9099195944034e65b1830f463caaa8390
   React-callinvoker: 6dff6d17d1d6cc8fdf85468a649bafed473c65f5
   React-Core: 39ee05b5798296f433dd3c3624c57a187c1510e3
-  React-Core-prebuilt: ff86a9cfffdf1ac6deca527352be5e6d0290354a
+  React-Core-prebuilt: 69556f895326f23c007f3a6869340045d7dca106
   React-CoreModules: e78bfd2617075bc0e50c689df4a29232bd72ad82
   React-cxxreact: 3fe21801d46097cf74c3dff6953677bebc4a3c2a
   React-debug: e1f00fcd2cef58a2897471a6d76a4ef5f5f90c74
@@ -2097,7 +2097,7 @@ SPEC CHECKSUMS:
   ReactAppDependencyProvider: 706b65371b90b5cc797b6639e8979f2e5cecd6da
   ReactCodegen: ab01ebfffac5cda9140204eb872ed97c15df225f
   ReactCommon: 47ef95b0920948a0b54d7439f7452501eeeac071
-  ReactNativeDependencies: 6e31d9f8b60229e795cddfbb4d99db3858730bbf
+  ReactNativeDependencies: 8a208df374583424130645685d86306befc275cf
   ReactNativeEnrichedHtml: 7d90df4aced7f533c7bd15ac296879b214413361
   Yoga: e83c3121d079541e69f3c5c623faaaf933fb5812
 

cpp/parser/GumboNormalizer.c

@@ -411,6 +411,34 @@ static void emit_one_attr(buffer_t *out, GumboElement *el,
   }
 }
 
+static bool is_checkbox_list(GumboElement *el) {
+  const char *val = get_attr(el, "data-type");
+  if (val && (strcmp(val, "checkbox") == 0 || strcmp(val, "checkboxList") == 0)) {
+    return true;
+  }
+
+  // In Google Docs and MS Word the <li> elements define if it is a checkbox
+  // list. We only need to check the first <li>.
+  GumboVector *children = &el->children;
+  for (unsigned int i = 0; i < children->length; i++) {
+    GumboNode *child = children->data[i];
+    if (is_element(child)) {
+      char child_tag[64];
+      if (get_tag_name(child, child_tag, sizeof(child_tag)) && strcmp(child_tag, "li") == 0) {
+        GumboElement *child_el = &child->v.element;
+        const char *role = get_attr(child_el, "role");
+        const char *cls = get_attr(child_el, "class");
+
+        // Matches Google Docs (role="checkbox") OR MS Word (class includes "checklist")
+        return (role && strcmp(role, "checkbox") == 0) ||
+               (cls && strstr(cls, "checklist") != NULL);
+      }
+    }
+  }
+
+  return false;
+}
+
 static void emit_attributes(GumboElement *el, const char *tag_name,
                             buffer_t *out) {
   if (strcmp(tag_name, "a") == 0) {
@@ -421,33 +449,7 @@ static void emit_attributes(GumboElement *el, const char *tag_name,
     emit_one_attr(out, el, "width");
     emit_one_attr(out, el, "height");
   } else if (strcmp(tag_name, "ul") == 0) {
-
-    const char *val = get_attr(el, "data-type");
-    bool is_checkbox = (val && (strcmp(val, "checkbox") == 0 || strcmp(val, "checkboxList") == 0));
-
-    // In Google Docs and MS Word the <li> elements define if it is a checkbox list
-    if (!is_checkbox) {
-      GumboVector *children = &el->children;
-      for (unsigned int i = 0; i < children->length; i++) {
-        GumboNode *child = children->data[i];
-        if (is_element(child)) {
-          char child_tag[64];
-          if (get_tag_name(child, child_tag, sizeof(child_tag)) && strcmp(child_tag, "li") == 0) {
-            GumboElement *child_el = &child->v.element;
-            const char *role = get_attr(child_el, "role");
-            const char *cls = get_attr(child_el, "class");
-
-            if ((role && strcmp(role, "checkbox") == 0) ||
-                (cls && strstr(cls, "checklist") != NULL)) {
-              is_checkbox = true;
-            }
-            break; // We only need to check the first <li>
-          }
-        }
-      }
-    }
-
-    if (is_checkbox) {
+    if (is_checkbox_list(el)) {
       buffer_append_str(out, " data-type=\"checkbox\"");
     }
   } else if (strcmp(tag_name, "li") == 0) {

src/web/normalization/htmlNormalizer.ts

@@ -249,29 +249,10 @@ function emitAttributes(el: Element, name: string): string {
         emitOneAttr(el, 'width') +
         emitOneAttr(el, 'height')
       );
-    case 'ul': {
-      let isCheckbox =
-        el.getAttribute('data-type') === 'checkbox' ||
-        el.getAttribute('data-type') === 'checkboxList';
-
-      if (!isCheckbox) {
-        const firstLi = Array.from(el.children).find(
-          (c) => c.tagName.toLowerCase() === 'li'
-        );
-        if (firstLi) {
-          const role = firstLi.getAttribute('role');
-          const className = firstLi.getAttribute('class') || '';
-
-          // Matches Google Docs (role="checkbox") OR MS Word (class includes "checklist")
-          if (role === 'checkbox' || className.includes('checklist')) {
-            isCheckbox = true;
-          }
-        }
-      }
-
-      return isCheckbox ? ' data-type="checkbox"' : '';
-    }
+    case 'ul':
+      return isCheckboxList(el) ? ' data-type="checkbox"' : '';
     case 'li':
+      // "" is the UTF-8 hex encoding for U+F0FE (MS Word Checked Box)
       const isChecked =
         el.hasAttribute('checked') ||
         el.getAttribute('data-checked') === 'true' ||
@@ -289,6 +270,32 @@ function emitAttributes(el: Element, name: string): string {
   }
 }
 
+function isCheckboxList(el: Element): boolean {
+  if (
+    el.getAttribute('data-type') === 'checkbox' ||
+    el.getAttribute('data-type') === 'checkboxList'
+  ) {
+    return true;
+  }
+
+  // In Google Docs and MS Word the <li> elements define if it is a checkbox
+  // list. We only need to check the first <li>.
+  const firstLi = Array.from(el.children).find(
+    (c) => c.tagName.toLowerCase() === 'li'
+  );
+  if (firstLi) {
+    const role = firstLi.getAttribute('role');
+    const className = firstLi.getAttribute('class') || '';
+
+    // Matches Google Docs (role="checkbox") OR MS Word (class includes "checklist")
+    if (role === 'checkbox' || className.includes('checklist')) {
+      return true;
+    }
+  }
+
+  return false;
+}
+
 function isGoogleDocsWrapper(el: Element, tag: string): boolean {
   if (tag !== 'b') return false;
   const id = el.getAttribute('id');