feat: add modular problem-fetching workflow with persistent storage

[khoahd-anz]

Summary

Refactors the LeetCode problem-fetching logic into a reusable, modular architecture that stores problem metadata persistently in problem.md files.

What's Changed

New Components

1. Composite Action (.github/actions/fetch-leetcode-problem)

   • ✅ Reusable action for fetching and storing LeetCode problems
   • ✅ Idempotent: skips API call if problem.md already exists
   • ✅ Outputs all necessary metadata for downstream workflows
   • ✅ Handles premium problems, duplicates, and force-refetch

2. Python Scripts

   • scripts/convert-problem-json-to-md.py: Converts API JSON → markdown with YAML frontmatter
   • scripts/parse-problem-md.py: Parses markdown → JSON for build-prompt.py

3. Manual Workflow (.github/workflows/fetch-problem.yml)

   • Allows ad-hoc problem fetching via workflow_dispatch
   • Commits problem.md to repository
   • Shows summary in workflow UI

4. Complete Specification (specs/modular-problem-fetching-workflow.md)

   • Architecture design and implementation guide (1,879 lines)
   • Testing strategy and migration path

Modified Workflows

• daily-leetcode.yml: Updated to use new composite action
  • Reads from problem.md when it exists (no API call needed!)
  • Maintains full backward compatibility

Key Benefits

🎯 Persistent Storage: Problem metadata stored in git with timestamps
⚡ Efficiency: Eliminates redundant API calls for duplicate problems
👀 Transparency: Problem data visible in GitHub UI and PRs
🧩 Modularity: Reusable components for future workflows
📊 Auditability: Track when problems were fetched and changes over time

Technical Details

problem.md Format

Each problem directory will contain a problem.md file with:

---
number: "0001"
frontend_id: "1"
title: "Two Sum"
slug: "two-sum"
difficulty: "Easy"
topics:
  - "Array"
  - "Hash Table"
acceptance_rate: 49.2
is_premium: false
created_at: "2026-02-12T08:30:19Z"
fetched_at: "2026-02-12T08:30:19Z"
link: "https://leetcode.com/problems/two-sum/"
date: "2026-02-12"
---

# 0001. Two Sum

[Clean markdown content converted from HTML]

Dependencies

• Python 3.11+
• Optional: html2text, pyyaml (fallback implementations included)
• GitHub Actions automatically installs these

Testing

✅ Round-trip conversion (JSON→MD→JSON) verified
✅ Scripts work with both libraries and regex fallback
✅ Compatible with existing build-prompt.py
✅ No breaking changes to existing workflows

Migration

No breaking changes. Existing workflows continue to work as-is.

• First daily run: Fetches from API, creates problem.md
• Subsequent runs: Reads from problem.md (no API call!)
• Existing 107 problems can optionally have problem.md backfilled later

How to Test

Test Manual Workflow

1. Go to Actions → Fetch LeetCode Problem
2. Click Run workflow
3. Enter a problem URL (e.g., https://leetcode.com/problems/two-sum/)
4. Watch it create problems/0001-two-sum/problem.md

Test Daily Workflow

The daily workflow will automatically use the new system:

• If problem.md exists: Read from file ⚡
• If doesn't exist: Fetch from API and create file 🆕

Files Changed

New Files

• .github/actions/fetch-leetcode-problem/action.yml
• .github/workflows/fetch-problem.yml
• scripts/convert-problem-json-to-md.py
• scripts/parse-problem-md.py
• specs/modular-problem-fetching-workflow.md

Modified Files

• .github/workflows/daily-leetcode.yml

---

Implementation follows spec: specs/modular-problem-fetching-workflow.md

Ready for review! 🚀

[khoahuynhdev]

@claude help me review this PR

[claude]

Claude encountered an error —— View job

Failed with exit code 128

---

I'll analyze this and get back to you.