fix: Duplicate paths in export ZIP (#12674)

server/queues/tasks/ExportDocumentTreeTask.ts

@@ -150,26 +150,12 @@ export default abstract class ExportDocumentTreeTask extends ExportTask {
     includeAttachments = true
   ) {
     const pathMap = this.createPathMap(collections, format);
-    Logger.debug(
+    await this.addDocumentsToArchive({
-      "task",
+      zip,
-      `Start adding ${Object.values(pathMap).length} documents to archive`
+      pathMap,
-    );
+      format,
-
+      includeAttachments,
-    for (const path of pathMap) {
+    });
-      const documentId = path[0].replace("/doc/", "");
-      const pathInZip = path[1];
-
-      await this.processDocument({
-        zip,
-        pathInZip,
-        documentId,
-        includeAttachments,
-        format,
-        pathMap,
-      });
-    }
-
-    Logger.debug("task", "Completed adding documents to archive");
 
     return await ZipHelper.toTmpFile(zip);
   }
@@ -200,28 +186,58 @@ export default abstract class ExportDocumentTreeTask extends ExportTask {
       format
     );
 
-    Logger.debug(
+    await this.addDocumentsToArchive({
-      "task",
+      zip,
-      `Start adding ${Object.values(pathMap).length} documents to archive`
+      pathMap,
-    );
+      format,
+      includeAttachments: true,
+    });
 
-    for (const entry of pathMap) {
+    return await ZipHelper.toTmpFile(zip);
-      const documentId = entry[0].replace("/doc/", "");
+  }
-      const pathInZip = entry[1];
+
+  /**
+   * Processes each unique document in the path map and adds it to the zip.
+   *
+   * @param zip The yazl ZipFile to add files to
+   * @param pathMap Map of document urls to their path in the zip
+   * @param format The format to export in
+   * @param includeAttachments Whether to include attachments in the export
+   */
+  private async addDocumentsToArchive({
+    zip,
+    pathMap,
+    format,
+    includeAttachments,
+  }: {
+    zip: ZipFile;
+    pathMap: Map<string, string>;
+    format: FileOperationFormat;
+    includeAttachments: boolean;
+  }) {
+    const processedPaths = new Set<string>();
+
+    Logger.debug("task", `Start adding documents to archive`);
+
+    for (const [url, pathInZip] of pathMap) {
+      // A document may be keyed by multiple urls in the path map, only
+      // process each file in the zip once.
+      if (processedPaths.has(pathInZip)) {
+        continue;
+      }
+      processedPaths.add(pathInZip);
 
       await this.processDocument({
         zip,
         pathInZip,
-        documentId,
+        documentId: url.replace("/doc/", ""),
-        includeAttachments: true,
+        includeAttachments,
         format,
         pathMap,
       });
     }
 
     Logger.debug("task", "Completed adding documents to archive");
-
-    return await ZipHelper.toTmpFile(zip);
   }
 
   /**

server/queues/tasks/ExportMarkdownZipTask.test.ts

@@ -0,0 +1,61 @@
+import fs from "fs-extra";
+import ZipHelper from "@server/utils/ZipHelper";
+import {
+  buildCollection,
+  buildDocument,
+  buildFileOperation,
+  buildTeam,
+  buildUser,
+} from "@server/test/factories";
+import ExportMarkdownZipTask from "./ExportMarkdownZipTask";
+
+describe("ExportMarkdownZipTask", () => {
+  it("should not duplicate documents in the zip file", async () => {
+    const team = await buildTeam();
+    const user = await buildUser({ teamId: team.id });
+    const collection = await buildCollection({
+      teamId: team.id,
+      createdById: user.id,
+    });
+    const documents = await Promise.all([
+      buildDocument({
+        teamId: team.id,
+        userId: user.id,
+        collectionId: collection.id,
+        title: "Test1",
+      }),
+      buildDocument({
+        teamId: team.id,
+        userId: user.id,
+        collectionId: collection.id,
+        title: "Test2",
+      }),
+    ]);
+    for (const document of documents) {
+      await collection.addDocumentToStructure(document);
+    }
+    const fileOperation = await buildFileOperation({
+      teamId: team.id,
+      userId: user.id,
+    });
+
+    const task = new ExportMarkdownZipTask();
+    const filePath = await task.exportCollections([collection], fileOperation);
+
+    try {
+      const fileNames: string[] = [];
+      await ZipHelper.walk(filePath, (entry) => {
+        if (!entry.isDirectory) {
+          fileNames.push(entry.fileName);
+        }
+      });
+
+      expect(fileNames.sort()).toEqual([
+        `${collection.name}/Test1.md`,
+        `${collection.name}/Test2.md`,
+      ]);
+    } finally {
+      await fs.remove(filePath);
+    }
+  });
+});