👷 Make json-schemas sync reliable: authenticate API requests and format output deterministically

[bdibon]

Motivation

Two issues made yarn json-schemas:sync unreliable:

1. Rate limiting — the branch-resolution step resolves the latest rum-events-format commit through an unauthenticated GitHub REST API call, capped at 60/hour per IP. Behind a shared corporate NAT that budget is consumed by everyone on the same address, so the sync fails with HTTP 403:

   API rate limit exceeded for <ip>. ... Authenticated requests get a higher rate limit.
   x-ratelimit-limit: 60   x-ratelimit-remaining: 0
   

2. Non-deterministic formatting — json-schema-to-typescript formats the generated *.types.ts files with whatever prettier it resolves at runtime, which differs between environments (e.g. a nested prettier left by a prior fork build locally vs. a freshly built fork in CI). In CI this collapsed union types onto a single line, so the check-schemas job reported a diff against the committed (repo-prettier-formatted) files and failed.

Changes

Both changes are in scripts/json-schemas.ts:

• Authenticate GitHub requests: resolve the branch using the developer's existing gh CLI token (authenticated limit: 5000/hour). Try gh auth token; on success attach an Authorization: token … header (style matches the existing call in scripts/lib/gitUtils.ts). If gh is missing or the user isn't logged in, print a warning and fall back to the previous unauthenticated behavior rather than failing. No env-var setup is required.
• Deterministic formatting: skip json-schema-to-typescript's internal prettier pass (format: false) and format the compiled output with the repo's pinned prettier (parser: 'typescript') before writing. Generation is now identical across environments and matches prettier --check ., so check-schemas stays green.

(The PR also includes a routine rum-events-format schema bump produced by running the now-reliable sync end-to-end.)

Test instructions

1. Auth — happy path (logged into gh):

   node scripts/json-schemas.ts --update master

   Resolves the latest commit with no 403.
2. Auth — fallback path (simulate gh unavailable):

   d=$(mktemp -d); printf '#!/bin/sh\nexit 1\n' > "$d/gh"; chmod +x "$d/gh"
   PATH="$d:$PATH" node scripts/json-schemas.ts --update master

   Prints the warning and proceeds with an unauthenticated request instead of crashing.
3. Formatting determinism: from a clean tree, regenerate and confirm no diff:

   yarn json-schemas:generate && git diff --stat   # expect: no changes

   Then force a fresh generator build (the original CI failure mode) and repeat:

   rm -rf node_modules/json-schema-to-typescript/dist node_modules/json-schema-to-typescript/node_modules/prettier
   yarn json-schemas:generate && git diff --stat   # still expect: no changes

4. Run the CI check end-to-end (from a clean tree): node scripts/check-schemas.ts exits 0.

Checklist

• Tested locally
• Tested on staging
• Added unit tests for this change.
• Added e2e/integration tests for this change.
• Updated documentation and/or relevant AGENTS.md file

🤖 Generated with Claude Code

[datadog-datadog-prod-us1-2]

🎉 All green!

🧪 All tests passed
❄️ No new flaky tests detected

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 77.20% (+0.00%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 35906e8 | Docs | Datadog PR Page | Give us feedback!}

[cit-pr-commenter-54b7da]

Bundles Sizes Evolution

📦 Bundle Name	Base Size	Local Size	𝚫	𝚫%	Status
Rum	172.75 KiB	172.75 KiB	0 B	0.00%	✅
Rum Profiler	8.22 KiB	8.22 KiB	0 B	0.00%	✅
Rum Recorder	21.14 KiB	21.14 KiB	0 B	0.00%	✅
Logs	54.44 KiB	54.44 KiB	0 B	0.00%	✅
Rum Slim	130.27 KiB	130.27 KiB	0 B	0.00%	✅
Worker	22.96 KiB	22.96 KiB	0 B	0.00%	✅

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a947cff29d

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Pin gh token lookup to github.com

issue: If a developer has GH_HOST set to a GitHub Enterprise host, gh auth token can return that host's token—the GitHub CLI manual says gh auth token chooses a default host without --hostname (https://cli.github.com/manual/gh_auth_token) and GH_HOST supplies that host (https://cli.github.com/manual/gh_help_environment). This code then sends the token to the hard-coded api.github.com URL, so yarn json-schemas:sync can fail instead of falling back and may expose the wrong token; request the token with gh auth token --hostname github.com for this GitHub.com API call.

Useful? React with 👍 / 👎.

[bdibon]

Let's keep it simple and not rely on env variables

[chatgpt-codex-connector]

Normalize unknown quota reasons before using the new schema

issue: This bump brings in a schema that enumerates _dd.profiling.quota_reason, but the SDK still forwards arbitrary backend reason strings: parseQuotaResult casts attrs.reason directly in packages/browser-rum/src/domain/profiling/quotaCheck.ts:23-24, and the existing spec covers unknown_reason in datadogProfiler.spec.ts:1184-1194. If the quota API returns a new reason, startProfilingContext attaches that value to VIEW/LONG_TASK/ACTION/VITAL events, so those events no longer validate against the schema used by our format checks; map unknown reasons to an allowed value (or update the schema) with this bump.

Useful? React with 👍 / 👎.

[bdibon]

Let's drop the middle commit, it should work anyway