fix: Handle emoji field in imported documents

closes #5810

server/commands/documentImporter.ts

@@ -1,5 +1,6 @@
 import path from "path";
 import emojiRegex from "emoji-regex";
+import escapeRegExp from "lodash/escapeRegExp";
 import truncate from "lodash/truncate";
 import mammoth from "mammoth";
 import quotedPrintable from "quoted-printable";
@@ -131,6 +132,15 @@ async function confluenceToMarkdown(value: Buffer | string): Promise<string> {
   return html.replace(/<br>/g, " \\n ");
 }
 
+type Props = {
+  user: User;
+  mimeType: string;
+  fileName: string;
+  content: Buffer | string;
+  ip?: string;
+  transaction?: Transaction;
+};
+
 async function documentImporter({
   mimeType,
   fileName,
@@ -138,14 +148,8 @@ async function documentImporter({
   user,
   ip,
   transaction,
-}: {
-  user: User;
-  mimeType: string;
-  fileName: string;
-  content: Buffer | string;
-  ip?: string;
-  transaction?: Transaction;
-}): Promise<{
+}: Props): Promise<{
+  emoji?: string;
   text: string;
   title: string;
   state: Buffer;
@@ -177,27 +181,22 @@ async function documentImporter({
   let text = await fileInfo.getMarkdown(content);
   text = text.trim();
 
-  // find and extract first emoji, in the case of some imports it can be outside
-  // of the title, at the top of the document.
+  // find and extract emoji near the beginning of the document.
   const regex = emojiRegex();
-  const matches = regex.exec(text);
-  const firstEmoji = matches ? matches[0] : undefined;
-  const textStartsWithEmoji = firstEmoji && text.startsWith(firstEmoji);
-  if (textStartsWithEmoji) {
-    text = text.replace(firstEmoji, "").trim();
+  const matches = regex.exec(text.slice(0, 10));
+  const emoji = matches ? matches[0] : undefined;
+  if (emoji) {
+    text = text.replace(emoji, "");
   }
 
   // If the first line of the imported text looks like a markdown heading
-  // then we can use this as the document title
+  // then we can use this as the document title rather than the file name.
   if (text.startsWith("# ")) {
     const result = parseTitle(text);
     title = result.title;
-    text = text.replace(`# ${title}`, "").trimStart();
-  }
-
-  // If we parsed an emoji from _above_ the title then add it back at prefixing
-  if (textStartsWithEmoji) {
-    title = `${firstEmoji} ${title}`;
+    text = text
+      .replace(new RegExp(`#\\s+${escapeRegExp(title)}`), "")
+      .trimStart();
   }
 
   // Replace any <br> generated by the turndown plugin with escaped newlines
@@ -227,6 +226,7 @@ async function documentImporter({
     text,
     state,
     title,
+    emoji,
   };
 }
 

server/queues/tasks/ImportMarkdownZipTask.ts

@@ -81,7 +81,7 @@ export default class ImportMarkdownZipTask extends ImportTask {
             return;
           }
 
-          const { title, text } = await documentImporter({
+          const { title, emoji, text } = await documentImporter({
             mimeType: "text/markdown",
             fileName: child.name,
             content: await zipObject.async("string"),
@@ -123,6 +123,7 @@ export default class ImportMarkdownZipTask extends ImportTask {
             output.documents.push({
               id,
               title,
+              emoji,
               text,
               updatedAt,
               createdAt,

server/queues/tasks/ImportNotionTask.ts

@@ -86,7 +86,7 @@ export default class ImportNotionTask extends ImportTask {
 
           Logger.debug("task", `Processing ${name} as ${mimeType}`);
 
-          const { title, text } = await documentImporter({
+          const { title, emoji, text } = await documentImporter({
             mimeType: mimeType || "text/markdown",
             fileName: name,
             content: zipObject ? await zipObject.async("string") : "",
@@ -117,6 +117,7 @@ export default class ImportNotionTask extends ImportTask {
             output.documents.push({
               id,
               title,
+              emoji,
               text,
               collectionId,
               parentDocumentId,

server/queues/tasks/ImportTask.ts

@@ -55,6 +55,7 @@ export type StructuredImportData = {
     id: string;
     urlId?: string;
     title: string;
+    emoji?: string;
     /**
      * The document text. To reference an attachment or image use the special
      * formatting <<attachmentId>>. It will be replaced with a reference to the

server/routes/api/documents/documents.ts

@@ -1249,7 +1249,7 @@ router.post(
     }
 
     const content = await fs.readFile(file.filepath);
-    const { text, state, title } = await documentImporter({
+    const { text, state, title, emoji } = await documentImporter({
       user,
       fileName: file.originalFilename ?? file.newFilename,
       mimeType: file.mimetype ?? "",
@@ -1261,6 +1261,7 @@ router.post(
     const document = await documentCreator({
       source: "import",
       title,
+      emoji,
       text,
       state,
       publish,