Handle missing GraphQL aliases as non-blocking repository skips

[Copilot]

The TypeScript repository-details workflow was treating GraphQL batch responses that silently omit a repo alias as hard hydration failures. That blocked production runs when repositories disappeared between search indexing and detail hydration (for example, deleted or newly private repos).

• Reclassify absent GraphQL aliases

  • Tag the providers.ts error emitted for repoN aliases missing from the GraphQL response with isMissingAlias: true.
  • This preserves the existing error surface while making the condition distinguishable from real hydration failures.

• Track skipped repos separately in pipeline metrics

  • Extend PipelineMetrics with:
    • missingRepoSkips
    • missingRepoSkipNames
  • Add a narrow type guard for missing-alias errors so generator logic can branch without weakening other error handling.

• Route missing aliases away from detailFailures

  • In the batch hydration path, handle isMissingAlias before PAT-policy classification.
  • Missing aliases now:
    • increment missingRepoSkips
    • record the repo name
    • emit a warning
    • skip the repo from output
  • They no longer inflate detailFailures.

• Keep the production guard strict on real failures only

  • Update the workflow guard to emit a ::warning:: for missingRepoSkips instead of adding them to blockingFailures.
  • detailFailures remains the production-blocking signal for true structural hydration problems.

• Add focused regression coverage

  • Verify provider-side missing-alias errors carry isMissingAlias: true.
  • Verify generator-side handling records a missing-repo skip, leaves detailFailures at zero, and excludes the vanished repo from output.

Example of the new provider tagging:

result.set(
  repo.fullName,
  Object.assign(
    new Error(`GraphQL response missing alias '${alias}' for ${repo.fullName}`),
    { isMissingAlias: true as const }
  )
);

Original prompt

Background

During the 1-update-github-repositories-details workflow, the TypeScript pipeline makes batched GraphQL calls (20 repos per batch). Each repo gets an alias (repo0…repo19). When a repo has been deleted or made private after the search API indexed it, GitHub's GraphQL API silently omits the alias from the response — with no accompanying error. This causes the code in Pipeline/src/providers.ts (~L306) to record a "GraphQL response missing alias 'repoN' for owner/repo" error.

These errors are counted as detailFailures in PipelineMetrics. The production guard in the workflow (Guard TypeScript repository-count delta step) blocks the entire run whenever detailFailures > 0:

if (failures > 0) {
  blockingFailures.push(`TypeScript pipeline had ${failures} repository detail hydration failure(s); all must succeed in production live mode`);
}

This causes valid runs to fail when repos simply disappear from GitHub between the search and the detail-hydration phase.

---

Goal

Implement Option A: classify "missing alias" errors as a separate non-blocking metric (missingRepoSkips) so the production guard does not block on repos that have simply vanished.

---

Required Changes

1. Pipeline/src/providers.ts

When recording the "missing alias" error (the else if (node === undefined) branch, ~L306), tag the error so callers can distinguish it from a true hydration failure:

result.set(
  repo.fullName,
  Object.assign(
    new Error(`GraphQL response missing alias '${alias}' for ${repo.fullName}`),
    { isMissingAlias: true as const }
  )
);

2. Pipeline/src/types.ts

Add a new counter to PipelineMetrics:

missingRepoSkips: number;       // repos silently dropped from GraphQL response (deleted/private)
missingRepoSkipNames: string[]; // names of those repos

Also add a helper type / type-guard if appropriate.

3. Pipeline/src/generator.ts

In the batch hydration path (where result instanceof Error is checked), add a branch before the isPatPolicyError check to handle isMissingAlias:

if ((result as any).isMissingAlias) {
  metrics.missingRepoSkips += 1;
  metrics.missingRepoSkipNames.push(repository.fullName);
  metrics.warnings.push(`Skipping '${repository.fullName}' because it is no longer accessible (GraphQL alias absent): ${result.message}`);
  return null;
}

Make sure missingRepoSkips and missingRepoSkipNames are initialised to 0 / [] in createMetrics().

4. .github/workflows/update-github-repositories-details.yml

In the "Guard TypeScript repository-count delta" step, add a non-blocking informational log for missingRepoSkips, and ensure the hard-block only fires on detailFailures (true structural failures), not missingRepoSkips. Something like:

const missingRepoSkips = metrics.missingRepoSkips ?? 0;
if (missingRepoSkips > 0) {
  const names = (metrics.missingRepoSkipNames ?? []).join(', ');
  console.log(`::warning::TypeScript pipeline skipped ${missingRepoSkips} repo(s) whose GraphQL alias was absent (likely deleted/private). Affected: ${names}`);
}

The existing detailFailures block should remain unchanged — it still blocks if there are genuine hydration errors.

5. Pipeline/tests/providers.test.ts

The existing test "records a missing-alias error for an absent alias in the GraphQL response with no errors" should still pass — the error value it checks just now carries the extra isMissingAlias property, so no test changes are needed there. However, add a test that verifies the tagged property is present:

expect((r as any).isMissingAlias).toBe(true);

6. Pipeline/tests/generator.test.ts

Add a test for the new branch:

• A batch provider that returns an error with isMissingAlias: true for one repo
• Assert metrics.missingRepoSkips === 1
• Assert metrics.detailFailures === 0
• Assert the repo is absent from the output

---

Acceptance Criteria

• isMissingAlias: true is set on the error recorded in providers.ts for the absent-alias case
• PipelineMetrics has missingRepoSkips: number and missingRepoSkipNames: string[]
• Generator routes isMissingAlias errors to missingRepoSkips, not detailFailures
• Workflow guard emits a ::warning:: (not ::error::) for missingRepoSkips > 0 and does not add it to blockingFailures
• All existing tests pass
• New unit tests cover the isMissingAlias path in both provider and generator

The following is the prior conversation context from the user's chat exploration (may be truncated):

User: Investigate, explain and propose a plan to fix the error in the following github workflow run: https://github.com/rpothin/PowerPlatform-OpenSource-Hub/actions/runs/26032659964/job/76522433000

Assistant: Here's a full breakdown of what happened, why it failed, and how to fix it.

---

🔍 What Failed

The wor...

This pull request was created from Copilot chat.