fix(error-tracking): search debug IDs progressively

[jhssilva]

What does this change?

Searches built JavaScript artifacts progressively for their injected ddDebugId instead of stopping after the first 1 KiB.

• Reads in 1 KiB chunks.
• Stops immediately after finding a valid debug ID, so the common case still performs one read.
• Retains a 64-character overlap so an ID split across two reads is matched.
• Keeps memory bounded to the current chunk plus overlap.
• If no debug ID exists, accepts reading the artifact to EOF as the rare worst case.

This is intentionally separate from the bundler-specific fixes:

• PR #487 coordinates esbuild post-build injection with sourcemap upload.
• PR #488 keeps Rollup injection after normal chunk transforms.

Tracking issue: RUM-18038

Validation

• 8/8 focused tests pass.
• Covers the one-read common case, a debug ID after the first chunk, a debug ID split across chunks, an absent debug ID that scans to EOF, and unreadable files.
• Focused lint and Error Tracking typechecking pass.

Exact PR package matrix:

Bundler	IDs extracted before upload	Runtime IDs observed	Result
webpack	4/4	2/2	PASS
Vite	6/6	2/4 loaded IDs	PASS
Rollup	4/4	2/2	PASS
esbuild	0/7	2/5 loaded IDs	Expected failure without PR #487
Rspack	4/4	2/2	PASS

The Rollup result uses PR #489 without PR #488, confirming progressive discovery finds the entry debug ID beyond the initial 1 KiB.

The locally stacked PR #487 + PR #489 esbuild package extracted 7/7 debug IDs and observed both runtime-loaded IDs, confirming compatibility once artifact readiness is present.

All five exact PR packages also passed static production builds, deterministic rebuild checks, and browser entry/lazy registration. Staging RUM search returned zero indexed events for all generated targets during the retry window, despite successful browser intake, so this run could not revalidate downstream deobfuscation.

[Copilot]

Pull request overview

This PR updates the Error Tracking sourcemap pipeline to extract injected ddDebugId values by scanning built JavaScript artifacts progressively (in fixed-size chunks with overlap) instead of only inspecting the initial prefix, improving reliability across bundlers/transforms that may move the injected snippet later in the file.

Changes:

• Replaces prefix-only reads with a progressive 1 KiB chunk scan that stops early on match and retains a small overlap for boundary-split literals.
• Exposes the chunk size constant and adds unit tests covering early stop, later-in-file IDs, split-across-chunks IDs, scan-to-EOF, and unreadable files.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
packages/plugins/error-tracking/src/sourcemaps/debugId.ts	Implements progressive chunked scanning with overlap to find ddDebugId without reading entire bundles in the common case.
packages/plugins/error-tracking/src/sourcemaps/debugId.test.ts	Adds coverage for the progressive read behavior and related edge cases.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[jhssilva]

/merge

[gh-worker-devflow-routing-ef8351]

View all feedbacks in Devflow UI.

2026-08-19 15:09:06 UTC ℹ️ Start processing command /merge
Use /merge -c to cancel this operation!

---

2026-08-19 15:09:12 UTC ℹ️ MergeQueue: Pull request is not mergeable yet

It will be processed automatically as soon as GitHub reports it as mergeable. View in MergeQueue UI.

• Run /code blockers to see what is blocking it.
• Run /remove to cancel it.

Use /merge -c to cancel this operation!

---

⏳ Waiting for pull request to become mergeable