fix(blob): normalize contentDisposition to use original filename when addRandomSuffix is enabled

[matingathani]

Summary

Fixes #903

When addRandomSuffix: true is configured for client uploads (upload()), the Vercel Blob API returns a contentDisposition header that includes the random suffix in the filename (e.g. attachment; filename="img-abc123.jpg"). Server-side put() with addRandomSuffix: true correctly returns the original filename (e.g. attachment; filename="img.jpg").

The inconsistency is in how the API handles the two request types:

• Server put(): sends x-add-random-suffix: 1 header → API returns original filename in Content-Disposition ✓
• Client upload(): sends addRandomSuffix: true inside the signed client token → API returns suffixed filename in Content-Disposition ✗

Fix: Added a normalizeContentDisposition() helper in put.ts that replaces the response filename with the original pathname filename when they differ. Applied to both the regular and multipart code paths in createPutMethod.

Changes

• packages/blob/src/put.ts — add normalizeContentDisposition() and apply it to both regular and multipart upload responses
• packages/blob/src/index.node.test.ts — add regression tests for the fix; update one pre-existing snapshot that used a mismatched mock (the fix now correctly derives contentDisposition from the original pathname)
• .changeset/fix-blob-client-content-disposition.md — patch changeset

Test plan

• pnpm test in packages/blob — all 137 tests pass
• put('foo.txt', body, { addRandomSuffix: true }) where API returns pathname: 'foo-abc123.txt' → result.contentDisposition is attachment; filename="foo.txt"
• put('foo.txt', body, { addRandomSuffix: false }) → result.contentDisposition unchanged

[vercel]

@matingathani is attempting to deploy a commit to the Curated Tests - Permanent E2E Team on Vercel.

A member of the Team first needs to authorize it.

[changeset-bot]

🦋 Changeset detected

Latest commit: bc768a4

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 2 packages

Name	Type
@vercel/blob	Patch
vercel-storage-integration-test-suite	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[Copilot]

Pull request overview

Fixes an inconsistency in the SDK’s contentDisposition value when the Blob API returns a suffixed pathname (e.g. with addRandomSuffix: true), ensuring the returned contentDisposition filename matches the original requested pathname.

Changes:

• Add a normalizeContentDisposition() helper and apply it to both single-part and multipart put() result mapping.
• Add regression tests for the normalization behavior and update an existing inline snapshot impacted by the normalization.
• Add a patch changeset for @vercel/blob.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
packages/blob/src/put.ts	Normalizes contentDisposition filename to the original requested pathname (single + multipart paths).
packages/blob/src/index.node.test.ts	Adds regression tests and updates an inline snapshot affected by filename normalization.
.changeset/fix-blob-client-content-disposition.md	Declares a patch release and documents the behavior change.

Comments suppressed due to low confidence (1)

packages/blob/src/index.node.test.ts:487

• The has an onUploadProgress option test now returns a snapshot where contentDisposition is normalized to progress.txt, but the mocked API response still returns pathname: "foo.txt". This makes the snapshot internally inconsistent and couples this progress test to the new normalization behavior. Consider using a dedicated mock response for this test where pathname/contentDisposition match the requested pathname (or assert pathname explicitly if the mismatch is intentional).

    it('has an onUploadProgress option', async () => {
      mockClient
        .intercept({
          path: () => true,
          method: 'PUT',
        })
        .reply(200, () => {
          return mockedFileMetaPut;
        });

      const onUploadProgress = jest.fn();

      await expect(
        put('progress.txt', 'Test Body', {
          access: 'public',
          onUploadProgress,
        }),
      ).resolves.toMatchInlineSnapshot(`
        {
          "contentDisposition": "attachment; filename="progress.txt"",
          "contentType": "text/plain",
          "downloadUrl": "https://storeId.public.blob.vercel-storage.com/foo-id.txt?download=1",
          "etag": ""abc123"",
          "pathname": "foo.txt",
          "url": "https://storeId.public.blob.vercel-storage.com/foo-id.txt",
        }

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[matingathani]

Addressed this by adding a regression that exercises the actual multipart put() path with a suffixed completion pathname and verifies contentDisposition is normalized back to the original requested filename. I also kept the completion-path regression so both the high-level multipart flow and the shared completion behavior are covered.