fix: Add application/octet-stream as a valid mimetype for docx uploads (#2105)

* fix: Add application/octet-stream as a valid mimetype for docx uploads

* fix: Include application/octet-stream in frontend filter
fix: Add file size and file type guards

* Validate .docx extension in files with application/octet-stream mimetype

* refactor: Move MAXIMUM_IMPORT_SIZE to an optional environment config
fix: Add file size check on server too

Co-authored-by: Saumya Pandey <sp160899@gmail.com>

server/commands/documentImporter.js

@@ -44,6 +44,10 @@ const importMapping: ImportableFile[] = [
     type: "application/msword",
     getMarkdown: confluenceToMarkdown,
   },
+  {
+    type: "application/octet-stream",
+    getMarkdown: docxToMarkdown,
+  },
   {
     type:
       "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
@@ -142,6 +146,12 @@ export default async function documentImporter({
 }): Promise<{ text: string, title: string }> {
   const fileInfo = importMapping.filter((item) => {
     if (item.type === file.type) {
+      if (
+        file.type === "application/octet-stream" &&
+        path.extname(file.name) !== ".docx"
+      ) {
+        return false;
+      }
       return true;
     }
     if (item.type === "text/markdown" && path.extname(file.name) === ".md") {

server/commands/documentImporter.test.js

@@ -37,6 +37,75 @@ describe("documentImporter", () => {
     expect(response.title).toEqual("images");
   });
 
+  it("should convert Word Document to markdown for application/octet-stream mimetype", async () => {
+    const user = await buildUser();
+    const name = "images.docx";
+    const file = new File({
+      name,
+      type: "application/octet-stream",
+      path: path.resolve(__dirname, "..", "test", "fixtures", name),
+    });
+
+    const response = await documentImporter({
+      user,
+      file,
+      ip,
+    });
+
+    const attachments = await Attachment.count();
+    expect(attachments).toEqual(1);
+
+    expect(response.text).toContain("This is a test document for images");
+    expect(response.text).toContain("![](/api/attachments.redirect?id=");
+    expect(response.title).toEqual("images");
+  });
+
+  it("should error when a file with application/octet-stream mimetype doesn't have .docx extension", async () => {
+    const user = await buildUser();
+    const name = "normal.docx.txt";
+    const file = new File({
+      name,
+      type: "application/octet-stream",
+      path: path.resolve(__dirname, "..", "test", "fixtures", name),
+    });
+
+    let error;
+    try {
+      await documentImporter({
+        user,
+        file,
+        ip,
+      });
+    } catch (err) {
+      error = err.message;
+    }
+
+    expect(error).toEqual("File type application/octet-stream not supported");
+  });
+
+  it("should convert Word Document on Windows to markdown", async () => {
+    const user = await buildUser();
+    const name = "images.docx";
+    const file = new File({
+      name,
+      type: "application/octet-stream",
+      path: path.resolve(__dirname, "..", "test", "fixtures", name),
+    });
+
+    const response = await documentImporter({
+      user,
+      file,
+      ip,
+    });
+
+    const attachments = await Attachment.count();
+    expect(attachments).toEqual(1);
+
+    expect(response.text).toContain("This is a test document for images");
+    expect(response.text).toContain("![](/api/attachments.redirect?id=");
+    expect(response.title).toEqual("images");
+  });
+
   it("should convert HTML Document to markdown", async () => {
     const user = await buildUser();
     const name = "webpage.html";
@@ -118,7 +187,7 @@ describe("documentImporter", () => {
     const name = "markdown.md";
     const file = new File({
       name,
-      type: "application/octet-stream",
+      type: "application/lol",
       path: path.resolve(__dirname, "..", "test", "fixtures", name),
     });