From be3bcebf6bc8a5909b38abfff12be2a4a975592d Mon Sep 17 00:00:00 2001
From: Tom Moor <tom.moor@gmail.com>
Date: Thu, 25 May 2023 21:34:26 -0400
Subject: [PATCH] fix: Remove empty top-level list items in imported HTML
 content

---
 server/commands/documentImporter.ts           |  5 ++++-
 server/utils/turndown/confluence-task-list.ts | 12 +++++-----
 server/utils/turndown/empty-lists.ts          | 22 +++++++++++++++++++
 server/utils/turndown/images.ts               | 11 +++++-----
 server/utils/turndown/index.ts                |  4 +++-
 5 files changed, 39 insertions(+), 15 deletions(-)
 create mode 100644 server/utils/turndown/empty-lists.ts

diff --git a/server/commands/documentImporter.ts b/server/commands/documentImporter.ts
index e2fbfcabc..4fbc18f35 100644
--- a/server/commands/documentImporter.ts
+++ b/server/commands/documentImporter.ts
@@ -56,7 +56,10 @@ async function fileToMarkdown(content: Buffer | string): Promise<string> {
 
 async function docxToMarkdown(content: Buffer | string): Promise<string> {
   if (content instanceof Buffer) {
-    const { value: html } = await mammoth.convertToHtml({ buffer: content });
+    const { value: html } = await mammoth.convertToHtml({
+      buffer: content,
+    });
+
     return turndownService.turndown(html);
   }
 
diff --git a/server/utils/turndown/confluence-task-list.ts b/server/utils/turndown/confluence-task-list.ts
index 7b5d512ef..8ecaac15a 100644
--- a/server/utils/turndown/confluence-task-list.ts
+++ b/server/utils/turndown/confluence-task-list.ts
@@ -10,16 +10,14 @@ export default function confluenceTaskList(turndownService: TurndownService) {
     filter(node) {
       return (
         node.nodeName === "LI" &&
-        node.parentNode?.nodeName === "UL" &&
-        // @ts-expect-error className exists
-        node.parentNode?.className.includes("inline-task-list")
+        node.parentElement?.nodeName === "UL" &&
+        node.parentElement?.className.includes("inline-task-list")
       );
     },
     replacement(content, node) {
-      return (
-        // @ts-expect-error className exists
-        (node.className === "checked" ? "- [x]" : "- [ ]") + ` ${content} \n`
-      );
+      return "className" in node
+        ? (node.className === "checked" ? "- [x]" : "- [ ]") + ` ${content} \n`
+        : content;
     },
   });
 }
diff --git a/server/utils/turndown/empty-lists.ts b/server/utils/turndown/empty-lists.ts
new file mode 100644
index 000000000..6bceebb8d
--- /dev/null
+++ b/server/utils/turndown/empty-lists.ts
@@ -0,0 +1,22 @@
+import TurndownService from "turndown";
+
+/**
+ * A turndown plugin for unwrapping top-level empty list items.
+ *
+ * @param turndownService The TurndownService instance.
+ */
+export default function emptyLists(turndownService: TurndownService) {
+  turndownService.addRule("empty-lists", {
+    filter(node) {
+      return (
+        node.nodeName === "LI" &&
+        node.childNodes.length === 1 &&
+        (node.firstChild?.nodeName === "OL" ||
+          node.firstChild?.nodeName === "UL")
+      );
+    },
+    replacement(content) {
+      return content;
+    },
+  });
+}
diff --git a/server/utils/turndown/images.ts b/server/utils/turndown/images.ts
index 2fd4eb027..3676d03d7 100644
--- a/server/utils/turndown/images.ts
+++ b/server/utils/turndown/images.ts
@@ -8,14 +8,13 @@ import TurndownService from "turndown";
 export default function images(turndownService: TurndownService) {
   turndownService.addRule("image", {
     filter: "img",
-
     replacement(content, node) {
-      // @ts-expect-error getAttribute exists
-      const alt = cleanAttribute(node.getAttribute("alt"));
-      // @ts-expect-error getAttribute exists
+      if (!("className" in node)) {
+        return content;
+      }
+      const alt = cleanAttribute(node.getAttribute("alt") || "");
       const src = (node.getAttribute("src") || "").replace(/\n+/g, "");
-      // @ts-expect-error getAttribute exists
-      const title = cleanAttribute(node.getAttribute("title"));
+      const title = cleanAttribute(node.getAttribute("title") || "");
       const titlePart = title ? ' "' + title + '"' : "";
       return src ? "![" + alt + "]" + "(" + src + titlePart + ")" : "";
     },
diff --git a/server/utils/turndown/index.ts b/server/utils/turndown/index.ts
index 7b17746af..03eaab61d 100644
--- a/server/utils/turndown/index.ts
+++ b/server/utils/turndown/index.ts
@@ -3,6 +3,7 @@ import TurndownService from "turndown";
 import breaks from "./breaks";
 import confluenceCodeBlock from "./confluence-code-block";
 import confluenceTaskList from "./confluence-task-list";
+import emptyLists from "./empty-lists";
 import images from "./images";
 
 /**
@@ -27,6 +28,7 @@ const service = new TurndownService({
   .use(confluenceTaskList)
   .use(confluenceCodeBlock)
   .use(images)
-  .use(breaks);
+  .use(breaks)
+  .use(emptyLists);
 
 export default service;