fix(core): refine Office HTML normalization and simplify parsing tests

Author: nperez0111

…pshots__/officeHTML.browser.test.ts.snap …rd/__snapshots__/officeHTML.test.ts.snappackages/core/src/api/clipboard/fromClipboard/__snapshots__/officeHTML.browser.test.ts.snap renamed to packages/core/src/api/clipboard/fromClipboard/__snapshots__/officeHTML.test.ts.snap packages/core/src/api/clipboard/fromClipboard/__snapshots__/officeHTML.browser.test.ts.snap renamed to packages/core/src/api/clipboard/fromClipboard/__snapshots__/officeHTML.test.ts.snap

@@ -425,43 +425,44 @@ exports[`Office clipboard HTML > parses captured clipboard HTML: Word mixed list
     "type": "bulletListItem",
   },
   {
-    "children": [],
-    "content": [
+    "children": [
       {
-        "styles": {},
-        "text": "Three Alskdfj alksdfj lkasj df",
-        "type": "text",
+        "children": [],
+        "content": [
+          {
+            "styles": {},
+            "text": "Three sub one Asldkfjalksdfj laksjdf",
+            "type": "text",
+          },
+        ],
+        "props": {
+          "backgroundColor": "default",
+          "textAlignment": "left",
+          "textColor": "default",
+        },
+        "type": "bulletListItem",
       },
-    ],
-    "props": {
-      "backgroundColor": "default",
-      "textAlignment": "left",
-      "textColor": "default",
-    },
-    "type": "bulletListItem",
-  },
-  {
-    "children": [],
-    "content": [
       {
-        "styles": {},
-        "text": "Three sub one Asldkfjalksdfj laksjdf",
-        "type": "text",
+        "children": [],
+        "content": [
+          {
+            "styles": {},
+            "text": "Three sub two Alskdfjlaksjfd lkasd f",
+            "type": "text",
+          },
+        ],
+        "props": {
+          "backgroundColor": "default",
+          "textAlignment": "left",
+          "textColor": "default",
+        },
+        "type": "bulletListItem",
       },
     ],
-    "props": {
-      "backgroundColor": "default",
-      "textAlignment": "left",
-      "textColor": "default",
-    },
-    "type": "bulletListItem",
-  },
-  {
-    "children": [],
     "content": [
       {
         "styles": {},
-        "text": "Three sub two Alskdfjlaksjfd lkasd f",
+        "text": "Three Alskdfj alksdfj lkasj df",
         "type": "text",
       },
     ],

packages/core/src/api/clipboard/fromClipboard/normalizeOfficeHTML.ts

@@ -13,39 +13,121 @@ function getMSOList(element: Element): string | undefined {
     .trim();
 }
 
+function getWordListInfo(element: Element) {
+  const match = getMSOList(element)?.match(/^l(\d+)\s+level(\d+)\s+lfo(\d+)$/i);
+  return match
+    ? { level: Number(match[2]), id: `${match[1]}:${match[3]}` }
+    : undefined;
+}
+
+function isWordListWhitespace(node: Node) {
+  return (
+    node.nodeType === 8 || (node.nodeType === 3 && !node.textContent?.trim())
+  );
+}
+
+// Word sometimes wraps a native level-two list in an otherwise empty list.
+// Only peel those wrappers when they contain nothing else (including text).
+function getWordNativeList(outer: Element) {
+  let list = outer;
+  while (true) {
+    const children = Array.from(list.childNodes).filter(
+      (node) => !isWordListWhitespace(node),
+    );
+    const child = children[0];
+    if (
+      children.length !== 1 ||
+      !(child instanceof Element) ||
+      !/^(UL|OL)$/.test(child.tagName)
+    ) {
+      break;
+    }
+    list = child;
+  }
+  const items = Array.from(list.childNodes).filter(
+    (node) => !isWordListWhitespace(node),
+  );
+  if (!items.every((item) => item instanceof HTMLLIElement)) {
+    return undefined;
+  }
+  const first = items[0];
+  const info = first && getWordListInfo(first);
+  if (
+    !info ||
+    !items.every((item) => {
+      const itemInfo = getWordListInfo(item);
+      return itemInfo?.id === info.id && itemInfo.level === info.level;
+    })
+  ) {
+    return undefined;
+  }
+  return { list, items, ...info };
+}
+
 function normalizeWordLists(root: HTMLElement) {
   const stack: {
-    element: HTMLElement;
+    element: Element;
     level: number;
     lastItem: HTMLLIElement;
   }[] = [];
   let listID: string | undefined;
 
-  for (const paragraph of Array.from(root.querySelectorAll("p[style]"))) {
-    const declaration = getMSOList(paragraph);
-    const match = declaration?.match(/^l(\d+)\s+level(\d+)\s+lfo(\d+)$/i);
-    if (!match) {
+  for (const paragraph of root.querySelectorAll("p[style], ul, ol")) {
+    // Native list content already has its own structure. In particular, do not
+    // turn a styled paragraph inside an li into a second list item.
+    if (paragraph.parentElement?.closest("ul, ol, li")) {
       continue;
     }
-
-    const level = Number(match[2]);
-    const id = `${match[1]}:${match[3]}`;
+    const native = paragraph.tagName !== "P";
+    const nativeList = native ? getWordNativeList(paragraph) : undefined;
+    const info = native ? nativeList : getWordListInfo(paragraph);
+    if (!info) {
+      continue;
+    }
+    const { level, id } = info;
     let previous = paragraph.previousSibling;
-    while (
-      previous &&
-      (previous.nodeType === 8 ||
-        (previous.nodeType === 3 && !previous.textContent?.trim()))
-    ) {
+    while (previous && isWordListWhitespace(previous)) {
       previous = previous.previousSibling;
     }
     if (id !== listID || previous !== stack[0]?.element) {
       stack.length = 0;
     }
     listID = id;
 
+    while (stack.length && stack[stack.length - 1].level > level) {
+      stack.pop();
+    }
+
+    if (nativeList) {
+      const current = stack[stack.length - 1];
+      const lastItem = nativeList.items[nativeList.items.length - 1];
+      if (current?.level === level) {
+        if (current.element.tagName !== nativeList.list.tagName) {
+          stack.length = 0;
+          continue;
+        }
+        current.element.append(...nativeList.list.childNodes);
+        current.lastItem = lastItem;
+        paragraph.remove();
+      } else if (current && level === current.level + 1) {
+        current.lastItem.append(nativeList.list);
+        if (paragraph !== nativeList.list) {
+          paragraph.remove();
+        }
+        stack.push({ element: nativeList.list, level, lastItem });
+      } else if (!current && paragraph === nativeList.list) {
+        stack.push({ element: nativeList.list, level, lastItem });
+      } else {
+        // A missing parent or a skipped level is not enough evidence to move
+        // an existing list under an unrelated preceding item.
+        stack.length = 0;
+      }
+      continue;
+    }
+
     let marker = "";
     const markers: Element[] = [];
-    for (const span of Array.from(paragraph.querySelectorAll("span[style]"))) {
+    for (const span of paragraph.querySelectorAll("span[style]")) {
       if (
         getMSOList(span)?.toLowerCase() === "ignore" &&
         !markers.some((parent) => parent.contains(span))
@@ -67,9 +149,6 @@ function normalizeWordLists(root: HTMLElement) {
       span.remove();
     }
 
-    while (stack.length && stack[stack.length - 1].level > level) {
-      stack.pop();
-    }
     if (
       stack[stack.length - 1]?.level === level &&
       stack[stack.length - 1].element.tagName.toLowerCase() !== tag
@@ -132,6 +211,15 @@ export function normalizeOfficeHTML(html: string): string {
   }
 
   for (const wrapper of root.querySelectorAll(googleDocsWrapper)) {
+    // Firefox can put the source ID on the selected paragraph itself. Its BRs
+    // are hard breaks, not sibling paragraphs. Only fragment containers can
+    // expand direct BRs into paragraphs (never textblocks or inline spans).
+    if (
+      !wrapper.matches("b, div") ||
+      wrapper.parentElement?.closest("p, h1, h2, h3, h4, h5, h6, pre")
+    ) {
+      continue;
+    }
     // Direct BRs represent blank paragraphs only in block-level fragments.
     if (
       Array.from(wrapper.childNodes).some(