[WIP] Add the test that we can actually download habr article (markdown and image): https://habr.com/ru/articles/895896 using all our support engines

[konard]

Summary

This PR adds comprehensive integration tests to verify that the web-capture service can successfully download real-world content from Habr.com (a Russian tech publication) using both Puppeteer and Playwright browser engines.

What Was Tested

The tests verify that we can download the Habr article at https://habr.com/ru/articles/895896 in both supported formats:

1. Markdown conversion - Verify HTML can be fetched and converted to markdown
2. PNG screenshots - Verify screenshots can be captured with proper image format

Implementation Details

New Files

• tests/integration/habr-article.test.js - Integration tests for Habr article downloads (5 test cases)
  • Puppeteer markdown download test
  • Puppeteer image screenshot test
  • Playwright markdown download test
  • Playwright image screenshot test
  • Engine comparison test

Modified Files

• jest.config.mjs - Added **/tests/integration/**/*.test.js to testMatch patterns
• src/browser.js - Fixed Playwright adapter to properly handle browser context creation and setUserAgent limitation

Test Results

All 5 new tests pass successfully:

PASS tests/integration/habr-article.test.js (76.873 s)
  Habr Article Download Tests
    Puppeteer Engine
      ✓ can download Habr article as markdown (11966 ms)
      ✓ can download Habr article as image screenshot (8658 ms)
    Playwright Engine
      ✓ can download Habr article as markdown (6918 ms)
      ✓ can download Habr article as image screenshot (11104 ms)
    Engine Comparison for Habr Article
      ✓ both engines can successfully download the same Habr article (11244 ms)

Technical Notes

Playwright User Agent Limitation

During implementation, I discovered that Playwright doesn't support setUserAgent() after page creation (unlike Puppeteer). The user agent must be set during browser context creation. I updated the browser abstraction layer to:

1. Create a browser context during Playwright browser initialization
2. Add a warning when setUserAgent() is called on Playwright pages (as it has no effect)
3. This maintains API compatibility while documenting the limitation

This is acceptable because:

• Most use cases don't require custom user agents
• When needed, the user agent can be set via browser context options
• The warning helps developers understand the limitation

Test Strategy

The tests use domcontentloaded instead of networkidle0 to avoid timeouts on complex pages. The tests verify:

• HTML content is fetched (> 1000 characters)
• Markdown conversion produces valid output (> 100 characters with markdown syntax)
• Screenshots are valid PNG images (correct signature and size)
• Both engines produce similar results for the same URL

Fixes

Fixes #7

---

🤖 Generated with Claude Code

[konard]

🤖 Solution Draft Log

This log file contains the complete execution trace of the AI solution draft process.

📎 Log file uploaded as GitHub Gist (275KB)
🔗 View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.