Skip to content

~95% on SimpleQA (e.g. Qwen3.6-27B on a 3090). Supports all local and cloud LLMs (llama.cpp, Ollama, Google, ...). 10+ search engines - arXiv, PubMed, your private documents. Everything Local & Encrypted.

5.8 KB
Raw Blame
"""
Test for Wikipedia URL validation security fix.

This addresses the security issue mentioned in PR #361 review.
"""

from tests.fixtures.search_engine_mocks import validate_wikipedia_url


class TestWikipediaURLSecurity:
    """Test Wikipedia URL validation for security."""

    def test_valid_wikipedia_urls(self):
        """Test that valid Wikipedia URLs are accepted."""
        valid_urls = [
            "https://en.wikipedia.org/wiki/Python",
            "https://wikipedia.org/wiki/Main_Page",
            "https://fr.wikipedia.org/wiki/Accueil",
            "https://de.wikipedia.org/wiki/Deutschland",
            "https://es.wikipedia.org/wiki/España",
            "https://ja.wikipedia.org/wiki/日本",
            "https://zh.wikipedia.org/wiki/中国",
            "https://simple.wikipedia.org/wiki/Computer",
            "https://meta.wikipedia.org/wiki/Main_Page",
        ]

        for url in valid_urls:
            assert validate_wikipedia_url(url) is True, (
                f"Valid URL rejected: {url}"
            )

    def test_invalid_wikipedia_urls(self):
        """Test that invalid/malicious URLs are rejected."""
        invalid_urls = [
            # Non-Wikipedia domains
            "https://fakepedia.org/wiki/Test",
            "https://wikipedia.com/wiki/Test",  # .com not .org
            "https://wikipediaorg.com/wiki/Test",
            "https://en-wikipedia.org/wiki/Test",  # Hyphenated domain
            # Malicious attempts
            "https://wikipedia.org.evil.com/wiki/Test",
            "https://evil.com/wikipedia.org/wiki/Test",
            "https://[email protected]/wiki/Test",
            "https://wikipedia.org.evil.com",
            # Invalid URLs
            "not-a-url",
            "ftp://wikipedia.org/wiki/Test",
            "javascript:alert('xss')",
            "file:///etc/passwd",
            "",
            None,
            # Edge cases
            "https://wikipediaXorg/wiki/Test",
            "https://wikipedia_org/wiki/Test",
            "https://wikipedia..org/wiki/Test",
        ]

        for url in invalid_urls:
            assert validate_wikipedia_url(url) is False, (
                f"Invalid URL accepted: {url}"
            )

    def test_url_validation_handles_exceptions(self):
        """Test that URL validation handles exceptions gracefully."""
        # Test with various problematic inputs
        problematic_inputs = [
            123,  # Integer
            [],  # List
            {},  # Dict
            object(),  # Generic object
            float("inf"),  # Infinity
            float("nan"),  # NaN
        ]

        for input_val in problematic_inputs:
            # Should return False without raising exception
            assert validate_wikipedia_url(input_val) is False

    def test_wikipedia_search_uses_validation(self, monkeypatch):
        """Test that Wikipedia search engine uses URL validation."""
        from tests.test_utils import add_src_to_path

        add_src_to_path()

        # Import after adding src to path
        from local_deep_research.web_search_engines.engines.search_engine_wikipedia import (
            WikipediaSearchEngine,
        )

        # The WikipediaSearchEngine should construct safe URLs internally
        # It should not use user-provided URLs directly
        search = WikipediaSearchEngine(max_results=1)

        # Mock the wikipedia module's search method
        import wikipedia

        def mock_search(*args, **kwargs):
            return ["Test Article"]

        monkeypatch.setattr(wikipedia, "search", mock_search)

        # Mock wikipedia.summary to avoid actual API calls
        def mock_summary(title, *args, **kwargs):
            return "Test summary for the article."

        monkeypatch.setattr(wikipedia, "summary", mock_summary)

        # Get previews (this is where URLs are constructed)
        previews = search._get_previews("test query")

        # Verify that the URL is properly constructed
        if previews:
            for preview in previews:
                url = preview.get("link", "")
                # URL should be a valid Wikipedia URL
                assert validate_wikipedia_url(url), (
                    f"Invalid URL generated: {url}"
                )
                # URL should follow the expected pattern
                assert url.startswith("https://en.wikipedia.org/wiki/"), (
                    f"Unexpected URL format: {url}"
                )


def test_wikipedia_url_in_response_validation():
    """Test that we validate URLs in Wikipedia API responses."""
    # This is a conceptual test showing how URL validation should be used
    # when processing Wikipedia API responses

    # audit: PUNCHLIST reviewed 2026-05 — issue resolved by prior PR (recommendation: delete).
    def process_wikipedia_response(response_data):
        """Process Wikipedia response with URL validation."""
        results = []

        for item in response_data.get("results", []):
            # If the response contains URLs, validate them
            if "url" in item:
                if not validate_wikipedia_url(item["url"]):
                    # Skip items with invalid URLs
                    continue

            results.append(item)

        return results

    # Test with mixed valid/invalid URLs
    test_response = {
        "results": [
            {
                "title": "Valid Article",
                "url": "https://en.wikipedia.org/wiki/Valid",
            },
            {
                "title": "Invalid Article",
                "url": "https://fake-wikipedia.org/wiki/Invalid",
            },
            {"title": "No URL Article"},  # No URL field
        ]
    }

    processed = process_wikipedia_response(test_response)

    # Should only include valid items
    assert len(processed) == 2
    assert processed[0]["title"] == "Valid Article"
    assert processed[1]["title"] == "No URL Article"