[PR #6159] [MERGED] fix(browser): report zero decoded bytes for undecodable gRPC-web text #7173

New issue

Closed

opened 2026-11-01 21:45:54 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-11-01 21:45:54 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/6159
Author: @Abelo9996
Created: 10/21/2026
Status: ✅ Merged
Merged: 10/21/2026
Merged by: @morluto

Base: main ← Head: fix-5942-grpc-web-text-decoded-bytes

📝 Commits (1)

3265c97 fix(browser): report zero decoded bytes for undecodable gRPC-web text

📊 Changes

2 files changed (+13 additions, -7 deletions)

View changed files

📝 src/browser/history/protobuf/ProtobufBody.ts (+1 -1)
📝 src/browser/history/protobuf/ProtobufWire.test.ts (+12 -6)

📄 Description

Summary

Fixes #5942.

When gRPC-web-text protocol text fails base64 preparation, inspect-web-network-capture reported the retained text length as decoded_bytes even though nothing was decoded. For AB== it said 4 decoded bytes, for !!!! also 4, and for a valid chunk followed by AB== it said 16. It now reports 0, which matches the absence of a decoded body. The partial status, error, zero examined bytes and the retained source descriptor are unchanged.

Problem and expected behavior

decodeProtobufBody in src/browser/history/protobuf/ProtobufBody.ts initialized decoded_bytes to body.length and only overwrote it with data.length after preparation succeeded. For gRPC-web-text, body is the retained base64 text, so the early data === undefined return left the encoded length in a field whose coordinates are documented as base64-decoded body bytes. Retained length already lives on the binary sidecar, so decoded_bytes should not fall back to it.

Change and scope

decoded_bytes now starts at 0 and is set from the prepared buffer once one exists. That is the only place the value is produced, so the fix is at the owner rather than in a projection. Successful bodies are unchanged because they still assign data.length. Raw, gRPC and gRPC-web bodies always have prepared data, so their counts are also unchanged.

Not included: recovering frames from a valid prefix before a later bad chunk, and any change to offset_basis. The issue explicitly does not ask for either.

Follow the contribution scope.
Scope decision: not applicable (ordinary fix within existing boundaries).

Contract and boundary impact

Semantic owner and earliest changed stage: decodeProtobufBody (protobuf body owner), at body result construction.
Contribution scope fit and linked maintainer decision: ordinary bug fix, not applicable.
CLI and MCP/tool-catalog contract: no schema change. decoded_bytes keeps its type and is now 0 on this failure path.
Provider, bridge, target-format, or platform compatibility: none.
Evidence, artifact, provenance, or reconstruction contract: decoded-byte accounting on failed gRPC-web-text preparation is corrected. Retained source bytes, hashes and sidecars are unchanged.
Target selection, authority, upstream behavior, and cleanup details (if applicable): not applicable.
Generated metadata (docs/public/product-catalog.json), package, or installation impact (CI generates and validates ignored catalogs and skills/; edit authored instructions in .agents/skills/reverse-engineer-anything/ and never commit build outputs): none.

Expected effects (check known effects; use documentation-only on its own).
Unchecked items do not establish absence; leave uncertainties visible.

Read existing files or captured data.
Launch or attach to a process.
Inject instrumentation or run scripts/code.
Change process memory, registers, files, or configuration.
Contact network services.
Recover protected code or work around protections.
Documentation or usability only; no target interaction.
Not sure yet; investigation or maintainer guidance needed.

Evidence and regression coverage

Tests added or updated: src/browser/history/protobuf/ProtobufWire.test.ts. The existing base64 chunk test only checked status for AB==. It now checks AB==, !!!! and a valid chunk followed by AB==, and expects decoded_bytes: 0, status: "partial", examined_bytes: 0, no frames or fields, and an error at offset 0.

Base reproduction or other evidence (executed): with the test change and without the fix, the test fails on the first case:

FAIL src/browser/history/protobuf/ProtobufWire.test.ts > decodes independently padded base64 chunks with explicit coordinates and rejects noncanonical text
AssertionError: expected { …(13) } to match object { decoded_bytes: +0, …(5) }
-   "decoded_bytes": 0,
+   "decoded_bytes": 4,
Tests  1 failed | 14 passed (15)

With the fix: Tests 15 passed (15).

User-visible CLI/MCP output (if applicable): I ran the issue's five HAR 1.2 fixtures through node scripts/rea.mjs inspect-web-network-capture <file> har --record 0 --json on a build of base 2aadcf7d and on a build with this change, then read normalized_result.records[0].protobuf[0]. All runs exited 0.

Protocol text	`decoded_bytes` before	`decoded_bytes` after	status / examined
`AAAAAAIIAQ==` (control)	7	7	complete / 7
`AB==`	4	0	partial / 0
`!!!!`	4	0	partial / 0
`AAAAAAIIAQ==AB==`	16	0	partial / 0
empty	0	0	complete / 0

The error reason (gRPC-web text is not canonical base64 or independently padded base64 chunks) is the same before and after.

Remaining proof gaps: MCP transport and the separate protobuf body view command were not executed. The view projects the stored count, so I expect it to show the same change, but I did not run it.

For evidence-bearing changes:

Observed, derived, and inferred claims remain distinguishable.
Artifact identity, source provenance, and failed attempts remain preserved.
Unsupported, incomplete, unavailable, or uncertain outcomes remain visible.

Validation performed

Run with the pinned Node 24.18.0 on macOS arm64.

npm run test:focused -- src/browser/history/protobuf/ProtobufWire.test.ts: 15 passed (fails 1 of 15 without the fix, shown above)
npm run test:focused -- src/browser/history/protobuf/ProtobufCapture.test.ts src/browser/history/protobuf/ProtobufBodyTrailer.test.ts: 7 passed
npm run check: exit 0 (typecheck, lint, format:check, knip, metadata:check, architecture guards)
CLI reproduction above, on builds from npm run build:cached with and without the fix

Compatibility, safety, and release

Breaking changes or migration steps: none. Consumers that read decoded_bytes on this failure path now get 0 instead of the encoded text length.
Real Hopper/Ghidra, browser, or OS coverage: not applicable. Offline HAR inspection on macOS only.
Package or release metadata impact: none.
Security, privacy, process, or containment review: no new inputs, execution or I/O.

Author readiness

Human owner: @Abelo9996

I reviewed the final diff and relevant existing code, understand the change and its edge cases, and take responsibility for the complete PR, including AI-assisted work.
I checked the evidence above, confirmed it supports the current change, and recorded meaningful validation gaps.
I declare this PR ready for initial maintainer review.

Before merge

This PR addresses a concrete problem or an agreed enhancement.
Material scope and authority questions are resolved, and any required maintainer scope decision covers the final diff.
The PR has one focused outcome and the title follows type(scope): outcome.
Related issue is linked, or the reason for not linking one is stated above.
Tests cover changed observable behavior and meaningful failure paths.
Owning docs, contracts, and generated metadata are updated where needed.
User-visible CLI/MCP changes include representative output.
I checked the final diff for secrets, unrelated cleanup, and unsupported claims.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/morluto/rea/pull/6159 **Author:** [@Abelo9996](https://github.com/Abelo9996) **Created:** 10/21/2026 **Status:** ✅ Merged **Merged:** 10/21/2026 **Merged by:** [@morluto](https://github.com/morluto) **Base:** `main` ← **Head:** `fix-5942-grpc-web-text-decoded-bytes` --- ### 📝 Commits (1) - [`3265c97`](https://github.com/morluto/rea/commit/3265c977ec8d49e78f4ff2f5a3c5a691618abcbf) fix(browser): report zero decoded bytes for undecodable gRPC-web text ### 📊 Changes **2 files changed** (+13 additions, -7 deletions) <details> <summary>View changed files</summary> 📝 `src/browser/history/protobuf/ProtobufBody.ts` (+1 -1) 📝 `src/browser/history/protobuf/ProtobufWire.test.ts` (+12 -6) </details> ### 📄 Description ## Summary Fixes #5942. When gRPC-web-text protocol text fails base64 preparation, `inspect-web-network-capture` reported the retained text length as `decoded_bytes` even though nothing was decoded. For `AB==` it said 4 decoded bytes, for `!!!!` also 4, and for a valid chunk followed by `AB==` it said 16. It now reports 0, which matches the absence of a decoded body. The partial status, error, zero examined bytes and the retained source descriptor are unchanged. ## Problem and expected behavior `decodeProtobufBody` in `src/browser/history/protobuf/ProtobufBody.ts` initialized `decoded_bytes` to `body.length` and only overwrote it with `data.length` after preparation succeeded. For gRPC-web-text, `body` is the retained base64 text, so the early `data === undefined` return left the encoded length in a field whose coordinates are documented as base64-decoded body bytes. Retained length already lives on the binary sidecar, so `decoded_bytes` should not fall back to it. ## Change and scope `decoded_bytes` now starts at 0 and is set from the prepared buffer once one exists. That is the only place the value is produced, so the fix is at the owner rather than in a projection. Successful bodies are unchanged because they still assign `data.length`. Raw, gRPC and gRPC-web bodies always have prepared data, so their counts are also unchanged. Not included: recovering frames from a valid prefix before a later bad chunk, and any change to `offset_basis`. The issue explicitly does not ask for either. Follow the [contribution scope](https://github.com/morluto/rea/blob/main/CONTRIBUTING.md#contribution-scope). Scope decision: not applicable (ordinary fix within existing boundaries). ## Contract and boundary impact - Semantic owner and earliest changed stage: `decodeProtobufBody` (protobuf body owner), at body result construction. - Contribution scope fit and linked maintainer decision: ordinary bug fix, not applicable. - CLI and MCP/tool-catalog contract: no schema change. `decoded_bytes` keeps its type and is now 0 on this failure path. - Provider, bridge, target-format, or platform compatibility: none. - Evidence, artifact, provenance, or reconstruction contract: decoded-byte accounting on failed gRPC-web-text preparation is corrected. Retained source bytes, hashes and sidecars are unchanged. - Target selection, authority, upstream behavior, and cleanup details (if applicable): not applicable. - Generated metadata (`docs/public/product-catalog.json`), package, or installation impact (CI generates and validates ignored catalogs and `skills/`; edit authored instructions in `.agents/skills/reverse-engineer-anything/` and never commit build outputs): none. Expected effects (check known effects; use documentation-only on its own). Unchecked items do not establish absence; leave uncertainties visible. - [x] Read existing files or captured data. - [ ] Launch or attach to a process. - [ ] Inject instrumentation or run scripts/code. - [ ] Change process memory, registers, files, or configuration. - [ ] Contact network services. - [ ] Recover protected code or work around protections. - [ ] Documentation or usability only; no target interaction. - [ ] Not sure yet; investigation or maintainer guidance needed. ## Evidence and regression coverage - Tests added or updated: `src/browser/history/protobuf/ProtobufWire.test.ts`. The existing base64 chunk test only checked `status` for `AB==`. It now checks `AB==`, `!!!!` and a valid chunk followed by `AB==`, and expects `decoded_bytes: 0`, `status: "partial"`, `examined_bytes: 0`, no frames or fields, and an error at offset 0. - Base reproduction or other evidence (executed): with the test change and without the fix, the test fails on the first case: ``` FAIL src/browser/history/protobuf/ProtobufWire.test.ts > decodes independently padded base64 chunks with explicit coordinates and rejects noncanonical text AssertionError: expected { …(13) } to match object { decoded_bytes: +0, …(5) } - "decoded_bytes": 0, + "decoded_bytes": 4, Tests 1 failed | 14 passed (15) ``` With the fix: `Tests 15 passed (15)`. - User-visible CLI/MCP output (if applicable): I ran the issue's five HAR 1.2 fixtures through `node scripts/rea.mjs inspect-web-network-capture <file> har --record 0 --json` on a build of base `2aadcf7d` and on a build with this change, then read `normalized_result.records[0].protobuf[0]`. All runs exited 0. | Protocol text | `decoded_bytes` before | `decoded_bytes` after | status / examined | | --- | --- | --- | --- | | `AAAAAAIIAQ==` (control) | 7 | 7 | complete / 7 | | `AB==` | 4 | 0 | partial / 0 | | `!!!!` | 4 | 0 | partial / 0 | | `AAAAAAIIAQ==AB==` | 16 | 0 | partial / 0 | | empty | 0 | 0 | complete / 0 | The error reason (`gRPC-web text is not canonical base64 or independently padded base64 chunks`) is the same before and after. - Remaining proof gaps: MCP transport and the separate protobuf body view command were not executed. The view projects the stored count, so I expect it to show the same change, but I did not run it. For evidence-bearing changes: - [x] Observed, derived, and inferred claims remain distinguishable. - [x] Artifact identity, source provenance, and failed attempts remain preserved. - [x] Unsupported, incomplete, unavailable, or uncertain outcomes remain visible. ## Validation performed Run with the pinned Node 24.18.0 on macOS arm64. - `npm run test:focused -- src/browser/history/protobuf/ProtobufWire.test.ts`: 15 passed (fails 1 of 15 without the fix, shown above) - `npm run test:focused -- src/browser/history/protobuf/ProtobufCapture.test.ts src/browser/history/protobuf/ProtobufBodyTrailer.test.ts`: 7 passed - `npm run check`: exit 0 (typecheck, lint, format:check, knip, metadata:check, architecture guards) - CLI reproduction above, on builds from `npm run build:cached` with and without the fix ## Compatibility, safety, and release - Breaking changes or migration steps: none. Consumers that read `decoded_bytes` on this failure path now get 0 instead of the encoded text length. - Real Hopper/Ghidra, browser, or OS coverage: not applicable. Offline HAR inspection on macOS only. - Package or release metadata impact: none. - Security, privacy, process, or containment review: no new inputs, execution or I/O. ## Author readiness Human owner: @Abelo9996 - [x] I reviewed the final diff and relevant existing code, understand the change and its edge cases, and take responsibility for the complete PR, including AI-assisted work. - [x] I checked the evidence above, confirmed it supports the current change, and recorded meaningful validation gaps. - [x] I declare this PR ready for initial maintainer review. ## Before merge - [x] This PR addresses a concrete problem or an agreed enhancement. - [x] Material scope and authority questions are resolved, and any required maintainer scope decision covers the final diff. - [x] The PR has one focused outcome and the title follows `type(scope): outcome`. - [x] Related issue is linked, or the reason for not linking one is stated above. - [x] Tests cover changed observable behavior and meaningful failure paths. - [x] Owning docs, contracts, and generated metadata are updated where needed. - [x] User-visible CLI/MCP changes include representative output. - [x] I checked the final diff for secrets, unrelated cleanup, and unsupported claims. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-11-01 21:45:54 +00:00