[adarshdigievo]

AI disclosure

• Model: GPT-6; Astra xhigh
• Harness: Codex desktop.
• Used for: implementation, documentation, test review and consolidation, live validation, and this PR draft.

Summary

Add client.extract() (and serpapi.extract()) for scraping pages, discovering URLs, and crawling websites. Both accept parameter dictionaries or keyword arguments and set engine="extract" automatically.

• Document Markdown, JSON, and HTML output, crawl limits, and API versus HTTP timeouts.
• Use output="md" directly for page Markdown, without a JSON restrictor.
• Consolidate Extract unit and live coverage into one test file.
• Handle Google's successful empty final pagination response in documentation validation while continuing to reject genuine failures.
• Refresh the README's reverse-image URL and obtain current Immersive Product tokens from Google Shopping.
• Add Python 3.15 to CI, release testing, and package classifiers.

No breaking changes or new runtime dependencies.

Type

• Bug fix
• Feature
• Refactor / performance
• Documentation / examples
• Tests
• Maintenance / packaging / CI

Related issue

None.

Affected SDK functionality

• Package version: 1.1.2
• Full-suite environment: Python 3.13.12, macOS 27.0.1, ARM64
• Methods: Client.extract() and module-level serpapi.extract()
• Endpoint: /search, with engine="extract"
• CI matrix: Python 3.8–3.15

Testing

• AI-executed testing:
  • Latest full pytest suite with live documentation checks enabled: 196 passed, 0 failed.
  • Includes all nine Extract cases and all live documentation examples, including README and Client Usage pagination.
  • Sphinx HTML and EPUB builds passed; uv build produced the wheel and source distribution.
• CI: Not run for this commit; the full version matrix remains unverified.

[Copilot]

🟢 Approval recommended

No active review issues block approval of the Extract API additions and related documentation and compatibility updates.

No open issues 🎉

_{Review effort: Balanced}

[strzibny]

LGTM. I haven't found any blockers. One small suggestion could be to use sentence case for titles since we are adopting it across the official SerpApi text and materials.

[adarshdigievo]

Thanks @strzibny

Will fix the sentence casing when I merge #44 (currently working on it locally)