[JOY (JOY)]

Closes #1542

Phase P1 of the #1531 design (Docs/Proposals/2026-10-31-criteria-stars-pros-cons.html, sections 2.1, 2.2, 4.1, 4.2). JOY approved decisions D1 and D5 on 2026-10-31.

Change (packages/shared-data/src/rules/)

• rating-criteria.ts
  • RATING_CRITERIA_VERSION = 1.
  • Criteria lists for smartphones, tablets, laptops, smartwatches, earbuds and services, as in design 2.1. Each criterion has a key, a vi/en label, a weight (each category sums to 100), the related spec group and a required flag. The required criterion is value for products and price for services.
  • getRatingCriteria(slug): an unknown or missing category rates as a phone.
  • weightedReviewStar(slug, ratings): the weighted average over the rated criteria, at 0.1. Unknown keys and bad stars are ignored; it returns null when nothing valid is rated.
  • validateCriteriaRatings(slug, ratings) follows D5. It flags:
    • unknown keys;
    • stars outside 1 to 5 in half steps;
    • a missing required criterion;
    • more than 2 skipped criteria.
  • mapLegacyCriteriaRatings(slug, legacy) maps the five legacy form keys (screen, performance, camera, battery, value) per category. For earbuds, screen becomes sound and camera becomes anc, matching the labels the product page uses today. A key with no counterpart is dropped, and nothing is inferred.
• review-form.ts holds the D5 text rules:
  • a verdict of 20 to 300 characters;
  • 1 to 5 pros and 1 to 5 cons, each 5 to 150 characters;
  • evasive cons ("không có", "chưa thấy", "không có gì để chê", "ko co", "none") are rejected, while real cons such as "Không có củ sạc trong hộp" pass;
  • duplicate lines and a line given as both a pro and a con are rejected.
  • Lines are compared without diacritics or punctuation. The stored text is not touched.
• Both modules are exported from @repo/shared-data/rules. They are browser-safe: they import only each other.

Not in this PR

• Web wiring. The form, the API and the product page are P3 (feat(web): review form v2 with weighted criteria, verdict and mandatory pros and cons #1544) and P4 (feat(web): one rating everywhere from the stored bayesian score #1545), owned by the web side.
• Storage. The P2 (chore(db): dos-me migration for review criteria, pros and cons and rating columns #1543) DOS-Me migration covers it.
• No database change.

Notes

• The earbuds legacy mapping (screen to sound, camera to anc) follows the product page labels (critSoundstage, critAncMic). The issue text said "camera to sound". The old write form showed phone labels for every category, so any mapping is a best guess.
• Legacy stars that have no counterpart are dropped: camera on tablets and laptops, performance and camera on watches, performance on earbuds. A mapped legacy review may rate fewer criteria than validateCriteriaRatings asks of a new review. That check is for new writes only.
• The constant is named RATING_CRITERIA_VERSION. The issue called it CRITERIA_VERSION.
• The modules are exported from @repo/shared-data/rules only, which is the browser-safe entry, and not from the root.
• hasRatingCriteria(slug) lets an API refuse a category that has no list instead of rating it as a phone.

Tests

packages/shared-data test step 11 covers:

• weights sum to 100, unique keys, both labels, one required criterion;
• weighted star math, including skipped criteria, legacy 0.1-step stars and ignored bad input;
• every D5 criteria issue, with null counted as skipped;
• prototype keys used as a slug or a rating;
• the legacy mapping;
• the verdict and pros/cons rules: no-content lines, lengths in code points, diacritics kept when comparing, malformed bodies;
• 40+ dodging and real cons;
• a linear-time guard on an oversized con.

npm test in packages/shared-data passes (tsc and the suite), and prettier is clean.

Review

Codex is not available in this session, so two independent fresh-context reviews ran in its place, followed by a round-2 review.

• Round 1 (890d2d0):
  • prototype-key slugs crashed;
  • legacy 0.1 stars were dropped;
  • null stars counted as bad stars;
  • many dodging cons were missed;
  • zero-width and fullwidth characters slipped through;
  • lengths were counted in UTF-16 units;
  • punctuation-only pros and verdicts passed;
  • stripping diacritics caused false pro_and_con hits.
• Round 2 (d0e6694):
  • a quadratic regex on oversized cons;
  • real cons read as dodges once diacritics were stripped ("hỏng luôn", "mọi thứ đều ồn");
  • more dodges ("mình không thấy gì để chê", "không có ý kiến").
• All P1 and P2 findings are fixed.

🤖 Generated with Claude Code

[claude]

⚠️ Code review skipped — your organization has no extra usage available to pay for this review.

If your organization's extra usage balance is empty, an organization admin can add extra usage credits at claude.ai/admin-settings/usage. If its monthly spend limit was reached, an admin can raise it on the same page. If neither applies, contact Anthropic support.

Once extra usage is available, someone with write access to this repository can comment @claude review on this pull request to trigger a review.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
rate	Ready	Preview	Oct 31, 2026 1:39pm UTC

[coderabbitai]

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 46 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Learn how review limits work.

Review configuration:

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 16e00ceb-2616-4df8-896e-6d72abc804a4

Commits

Reviewing files that changed from the base of the PR and between 76a033e and d0e6694.

Files selected for processing: 4

• packages/shared-data/src/__tests__/shared-data.test.ts
• packages/shared-data/src/rules/index.ts
• packages/shared-data/src/rules/rating-criteria.ts
• packages/shared-data/src/rules/review-form.ts

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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

[dos]

⏱️ Code Review completed (4 eligible files · 21,143 chars · 1 model call(s) · 1 PR unit(s))

Whole diff with bounded context at base 76a033e / head dd417c5.

✅ CLEAN_PASS: Complete diff reviewed

[dos]

⏱️ Code Review completed (4 eligible files · 27,815 chars · 1 model call(s) · 1 PR unit(s))

Whole diff with bounded context at base 76a033e / head 890d2d0.

✅ CLEAN_PASS: Complete diff reviewed

[dos]

⏱️ Code Review completed (4 eligible files · 29,667 chars · 1 model call(s) · 1 PR unit(s))

Whole diff with bounded context at base 76a033e / head d0e6694.

✅ CLEAN_PASS: Complete diff reviewed