Require GFM adjacency for translate table detection

[feruzm]

Parity port of the tightening that shipped in the mobile compose-translation review (ecency/ecency-mobile#3325): translateMarkdown classified a block as a table when any line had a pipe and any other line looked like a |---| separator, so prose containing pipe characters near a separator-like line was silently skipped by translation. The separator row must now sit directly below a pipe-containing row (the GFM shape).

Two regression tests added (real table still skipped; pipes-in-prose now translates); suite 30/30.

Replaces #1087 (branch renamed so the build workflow triggers).

[coderabbitai]

Warning

Review limit reached

@feruzm, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 19 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: ff980bb6-2503-4223-b32c-bb32f6445c02

📥 Commits

Reviewing files that changed from the base of the PR and between 071f467 and 3f8733b.

📒 Files selected for processing (2)

• apps/web/src/api/translation.ts
• apps/web/src/specs/api/translate-markdown.spec.ts

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch bugfix/translate-table-adjacency

---

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.}

[greptile-apps]

Greptile Summary

This PR tightens markdown table detection for translated content. The main changes are:

• Requires a separator row to sit directly below a pipe row.
• Adds matching cell-count checks before skipping a block as a table.
• Adds tests for real tables, stray separator-like lines, and pipe prose.

Confidence Score: 4/5

This is close, but the table heuristic can still skip translatable prose.

• Adjacent prose with matching pipe counts can still be classified as a table.
• That path leaves user text untranslated without an error.

apps/web/src/api/translation.ts

Important Files Changed

Filename	Overview
apps/web/src/api/translation.ts	Updates table detection to require adjacent separator rows with matching cell counts.
apps/web/src/specs/api/translate-markdown.spec.ts	Adds tests for table skipping and pipe-prose translation behavior.

_{Reviews (2): Last reviewed commit: "fix(publish): match table cell counts in..." | Re-trigger Greptile}

[greptile-apps]

Matching Counts Still Skip

When adjacent prose has the same pipe-delimited count as a separator-looking line, this still treats the whole block as a table. For example, Valores: min | max | defecto\n-- | -- | -- gives both lines a count of 3, so translateMarkdown skips the block instead of translating the prose. The count check fixes mismatched examples, but matching-count prose still produces the silent untranslated output this change is meant to prevent.

Context Used: CLAUDE.md (source)