fix(typescript): encode non-Latin-1 filenames in Content-Disposition header by fern-api[bot] · Pull Request #17656 · fern-api/fern

fern-api · 2026-09-04T16:36:29Z

Description

Linear ticket: Refs Pylon #23214

Binary uploads in generated TS SDKs (core/file/file.ts → toBinaryUploadRequest) interpolate the raw filename into Content-Disposition. fetch/Headers requires header values to be ISO-8859-1, so any filename with a non-Latin-1 code point throws before the request is sent:

TypeError: Failed to execute 'set' on 'Headers': String contains non ISO-8859-1 code point.

This also hits Latin-1-looking names like été.pdf when they come from an NFD filesystem (macOS): the combining acute accent U+0301 is > 255. Reproduced in Node 22 (Headers.set → Cannot convert argument to a ByteString because the character at index N has a value of 769).

Changes Made

New toContentDisposition(filename) in core/file/file.ts, per RFC 6266 / RFC 5987:

"report.pdf"        -> attachment; filename="report.pdf"
"rapport-été.pdf"   -> attachment; filename="rapport-_t_.pdf"; filename*=UTF-8''rapport-%C3%A9t%C3%A9.pdf
"日本.pdf"           -> attachment; filename="__.pdf"; filename*=UTF-8''%E6%97%A5%E6%9C%AC.pdf

NFC-normalizes first; plain-ASCII names are emitted exactly as before. Non-ASCII (and "/\) chars become _ in the quoted fallback and the exact name is carried in filename*.

toMultipartDataPart is unchanged: FormData.append(name, blob, filename) handles UTF-8 filenames itself and does not throw (checked in Node 22).
Changelog: generators/typescript/sdk/changes/unreleased/fix-content-disposition-non-latin1-filename.yml
Updated README.md generator (if applicable) — N/A

Testing

Unit tests added/updated — tests/unit/file/file.test.ts gains a toContentDisposition suite (ASCII passthrough, non-ASCII, NFD→NFC, quote escaping, lone/paired surrogates, and a new Headers().set(...) acceptance check).
Manual testing completed — ran the new helper + toBinaryUploadRequest through Headers.set in Node 22 for CJK / NFD / quoted filenames; no throw.
Seed snapshots regenerated with pnpm seed test --generator ts-sdk for every fixture that ships core/file (file-upload ×6, bytes-upload, enum, exhaustive incl. local-files-no-source cjs/esm, file-upload-openapi, multiple-request-bodies, openapi-request-body-ref, ts-bytes-upload-path-param). All 39 seed/ts-sdk/**/core/file/file.ts copies now use toContentDisposition. Ran the regenerated tests/unit/file/file.test.ts inside a generated SDK (file-upload/no-custom-config): 36/36 pass, tsc --noEmit clean.
Also handles ill-formed UTF-16: lone surrogates are replaced with U+FFFD before NFC normalization so encodeURIComponent cannot throw (bad\ud800.txt → filename="bad_.txt"; filename*=UTF-8''bad%EF%BF%BD.txt).

…header

nitpickybot

AI Review Summary

Adds toContentDisposition to safely encode non-Latin-1 filenames per RFC 6266/5987. Logic is correct; main gaps are the un-regenerated seed snapshots (39 fixtures) which will fail CI, and a minor line-length/formatting concern for prettier.

🟡 1 warning(s)
🔵 1 suggestion(s)

To request another review, comment /ai-review on this pull request.

nitpickybot · 2026-09-04T16:36:48Z

+    if (asciiFallback === normalized) {
+        return `attachment; filename="${asciiFallback}"`;
+    }
+    const encoded = encodeURIComponent(normalized).replace(/['()*]/g, (c) => `%${c.charCodeAt(0).toString(16).toUpperCase()}`);


🔵 suggestion

This line is ~140 chars; the repo's prettier config (typically 120 for these generators) will likely reformat it. Consider breaking it up so CI lint doesn't complain:

Suggested change

const encoded = encodeURIComponent(normalized).replace(/['()*]/g, (c) => `%${c.charCodeAt(0).toString(16).toUpperCase()}`);

const encoded = encodeURIComponent(normalized).replace(

/['()*]/g,

(c) => `%${c.charCodeAt(0).toString(16).toUpperCase()}`,

);

Done in 366aa47 — broke the encodeURIComponent(...).replace(...) call across lines.

nitpickybot · 2026-09-04T16:36:48Z

+export function toContentDisposition(filename: string): string {
+    const normalized = filename.normalize("NFC");
+    const asciiFallback = normalized.replace(/[^\x20-\x7e]|["\\]/g, "_");
+    if (asciiFallback === normalized) {
+        return `attachment; filename="${asciiFallback}"`;
+    }
+    const encoded = encodeURIComponent(normalized).replace(/['()*]/g, (c) => `%${c.charCodeAt(0).toString(16).toUpperCase()}`);
+    return `attachment; filename="${asciiFallback}"; filename*=UTF-8''${encoded}`;
+}


🟡 warning

Core-utilities source changes must be mirrored into the seed/ts-sdk/*/src/core/file/file.ts snapshots — the PR description acknowledges 39 fixtures weren't regenerated. Without pnpm seed test --generator ts-sdk, seed CI will fail on diff. Worth doing before merge rather than as a follow-up.

Done in 366aa47 — regenerated via pnpm seed test --generator ts-sdk for the fixtures that ship core/file (file-upload ×6 variants, bytes-upload, enum, exhaustive incl. local-files-no-source cjs/esm outputs, file-upload-openapi, multiple-request-bodies, openapi-request-body-ref, ts-bytes-upload-path-param). All 39 seed/ts-sdk/**/core/file/file.ts copies now carry toContentDisposition; no remaining attachment; filename="${filename}".

devin-ai-integration

Devin Review found 1 potential issue.

devin-ai-integration · 2026-09-04T16:38:12Z

+    if (asciiFallback === normalized) {
+        return `attachment; filename="${asciiFallback}"`;
+    }
+    const encoded = encodeURIComponent(normalized).replace(/['()*]/g, (c) => `%${c.charCodeAt(0).toString(16).toUpperCase()}`);


🟡 Malformed Unicode filenames still abort uploads

A filename containing an unpaired UTF-16 surrogate makes encodeURIComponent throw. The upload aborts before sending its request.

Prompt for agents

Handle ill-formed UTF-16 filenames in generators/typescript/utils/core-utilities/src/core/file/file.ts before passing them to encodeURIComponent. JavaScript permits strings containing unpaired surrogates, and encodeURIComponent throws URIError for them. Convert unpaired surrogates to a safe replacement while preserving valid surrogate pairs, then build both the fallback and filename* values from that sanitized string. Add unit coverage for isolated high and low surrogates.

Was this helpful? React with 👍 or 👎 to provide feedback.

Good catch — fixed in 366aa47. toContentDisposition now runs replaceLoneSurrogates() before .normalize("NFC"), replacing unpaired high/low surrogates with U+FFFD (valid pairs are preserved) so encodeURIComponent can't throw. Added unit cases for bad\ud800.txt, \udc00bad.txt, and a valid astral pair.

github-actions · 2026-09-04T17:03:00Z

SDK Generation Benchmark Results

Comparing PR branch against median of 5 nightly run(s) on main (latest: 2026-09-04T04:06:24Z).

Full benchmark table (click to expand)

Generator	Spec	main (generator)	main (E2E)	PR (generator)	Delta
ts-sdk	square	178s (n=5)	184s (n=5)	135s	-43s (-24.2%)

main (generator): generator-only time via --skip-scripts (includes Docker image build, container startup, IR parsing, and code generation — this is the same Docker-based flow customers use via fern generate). main (E2E): full customer-observable time including build/test scripts (nightly baseline, informational). Delta is computed against generator-only baseline.
⚠️ = generation exited with a non-zero exit code (timing may not reflect a successful run).
Baseline from nightly runs on main (latest: 2026-09-04T04:06:24Z). Trigger benchmark-baseline to refresh.
Last updated: 2026-09-04 17:55 UTC

…hots Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

fix(typescript): encode non-Latin-1 filenames in Content-Disposition …

2886d14

…header

nitpickybot Bot reviewed Sep 4, 2026

View reviewed changes

devin-ai-integration Bot reviewed Sep 4, 2026

View reviewed changes

fix(typescript): handle lone surrogates, regenerate ts-sdk seed snaps…

366aa47

…hots Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

Navigation Menu

Uh oh!

fix(typescript): encode non-Latin-1 filenames in Content-Disposition header - #17656

fix(typescript): encode non-Latin-1 filenames in Content-Disposition header#17656
fern-api[bot] wants to merge 2 commits into
mainfrom
devin/1788539479-ts-content-disposition-encoding

fern-api Bot commented Sep 4, 2026 •

edited

Loading

Uh oh!

nitpickybot Bot left a comment

Uh oh!

nitpickybot Bot Sep 4, 2026

Uh oh!

fern-api Bot Sep 4, 2026

Uh oh!

nitpickybot Bot Sep 4, 2026

Uh oh!

fern-api Bot Sep 4, 2026

Uh oh!

devin-ai-integration Bot left a comment

Uh oh!

devin-ai-integration Bot Sep 4, 2026

Uh oh!

fern-api Bot Sep 4, 2026

Uh oh!

github-actions Bot commented Sep 4, 2026 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

0 participants

Uh oh!

Conversation

fern-api Bot commented Sep 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Changes Made

Testing

Uh oh!

nitpickybot Bot left a comment

Choose a reason for hiding this comment

AI Review Summary

Uh oh!

nitpickybot Bot Sep 4, 2026

Choose a reason for hiding this comment

Uh oh!

fern-api Bot Sep 4, 2026

Choose a reason for hiding this comment

Uh oh!

nitpickybot Bot Sep 4, 2026

Choose a reason for hiding this comment

Uh oh!

fern-api Bot Sep 4, 2026

Choose a reason for hiding this comment

Uh oh!

devin-ai-integration Bot left a comment

Choose a reason for hiding this comment

Uh oh!

devin-ai-integration Bot Sep 4, 2026

Choose a reason for hiding this comment

Uh oh!

fern-api Bot Sep 4, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Sep 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

SDK Generation Benchmark Results

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

0 participants

fern-api Bot commented Sep 4, 2026 •

edited

Loading

github-actions Bot commented Sep 4, 2026 •

edited

Loading