fix(server): bound usage transcript lines

[ifBars]

Summary

• replace readline with a byte-bounded JSONL reader for usage transcripts
• discard any record over 64 MiB and resume scanning at the next newline
• cover oversized records, continued usage extraction, CRLF input, and an unterminated final record

Why

A Codex rollout can contain a very large tool or patch event on one JSONL line. Node's readline constructs the entire line before yielding it, so a line above V8's string limit throws RangeError: Invalid string length from the stream callback and terminates the T3 server when Usage is opened.

The new reader retains at most 64 MiB for a record. Once a record crosses that limit, it drains bytes through the next newline without decoding them, then continues processing later usage records normally.

Validation

• vp test run apps/server/src/usage: 40 tests passed
• targeted lint and formatting checks pass
• reproduced the original Node crash with an untouched 525.98 MiB rollout containing one 516.47 MiB line
• patched reader completed that rollout in 443 ms with 651 usage records and about 142 MiB peak RSS
• extracted records exactly matched a control copy with only the oversized non-usage event removed
• scanned 1.56 GB of active 90-day transcripts in 4.96 s and 688 MB of archived transcripts in 2.24 s, with no failed files

Fixes #7258.

Implemented with GPT-5.6 Sol via Codex.

Note

Bound per-line bytes in readTranscriptRecords to skip oversized JSONL records

• Replaces node:readline with a new readBoundedLines async generator in usageTranscriptReader.ts that accumulates file bytes until a newline, discarding lines that exceed a configurable byte limit before constructing a string.
• Defaults to 64 MiB per line (DEFAULT_MAX_LINE_BYTES), which is well below V8's max string length but above expected valid transcript records.
• Callers can override the limit via the new optional TranscriptReadOptions.maxLineBytes parameter on readTranscriptRecords.
• Also adds CRLF support (trims trailing \r) and yields unterminated final lines.
• Behavioral Change: oversized lines (e.g. large tool results) are now silently skipped rather than causing a potential OOM or parse error.

^{Macroscope summarized 42f1b50.}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 7c27bcfd-dbf1-4813-a7fc-35322c68a3d7

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[macroscopeapp]

Approvability

Verdict: Approved 42f1b50

Defensive bug fix replacing Node's readline with a bounded reader to prevent V8 crashes from oversized transcript lines. The change is self-contained with comprehensive tests and a conservative 64MB default limit.

^{You can customize Macroscope's approvability policy. Learn more.}