kb-skills/tests/test_platforms/test_zhihu.py

"""Tests for ZhihuAdapter

Tests cover:
- Answer content extraction
- Zhuanlan (article) content extraction
- Cookie-based authentication
- Login page detection
- AuthError handling

Wave 0 Test Scaffolding (Phase 09 - Pipeline Integration):
- Mock fixtures for ZhihuCrawler CrawlResult
- Test skeletons for integration with new crawler
"""

import pytest
from unittest.mock import MagicMock, patch, PropertyMock
import asyncio
from pathlib import Path

from kb_workflow.pipeline.platforms.zhihu import ZhihuAdapter
from kb_workflow.pipeline.errors import AuthError

# =============================================================================
# Wave 0 Mock Fixtures for ZhihuCrawler Integration (Phase 09)
# =============================================================================


@pytest.fixture
def mock_crawl_answer_result():
    """Mock CrawlResult from ZhihuCrawler.crawl_answer().

    Simulates successful answer crawl with metadata and content.
    Used for testing ZhihuAdapter integration with new crawler.
    """
    from kb_workflow.crawler.models import (
        CrawlResult, AnswerData, AuthorInfo, ContentMetadata, CommentData, CommentAuthor
    )

    return CrawlResult(
        success=True,
        content_type="answer",
        data=AnswerData(
            id=789012,
            question_id=123456,
            question_title="Test Question Title for Pipeline Integration",
            content="<p>This is the <strong>answer</strong> content from ZhihuCrawler.</p>",
            excerpt="This is the answer content from ZhihuCrawler.",
            metadata=ContentMetadata(
                author=AuthorInfo(
                    name="Test Author",
                    url_token="test-author-123",
                    headline="Test Author Headline"
                ),
                voteup_count=1234,
                comment_count=56,
                created_time=1705312800,  # 2024-01-15 10:00:00 UTC
                updated_time=1705399200   # 2024-01-16 10:00:00 UTC
            )
        ),
        comments=[
            CommentData(
                id=111,
                content="Great answer! Very helpful.",
                author=CommentAuthor(name="Commenter1", url_token="commenter-1"),
                created_time=1705316400,
                like_count=42,
                child_count=3,
                is_author=False
            ),
            CommentData(
                id=222,
                content="Thanks for sharing!",
                author=CommentAuthor(name="Commenter2", url_token="commenter-2"),
                created_time=1705320000,
                like_count=15,
                child_count=0,
                is_author=False
            )
        ],
        error=None
    )


@pytest.fixture
def mock_crawl_article_result():
    """Mock CrawlResult from ZhihuCrawler.crawl_article().

    Simulates successful zhuanlan article crawl.
    Used for testing ZhihuAdapter integration with new crawler.
    """
    from kb_workflow.crawler.models import (
        CrawlResult, ArticleData, AuthorInfo, ContentMetadata, CommentData, CommentAuthor
    )

    return CrawlResult(
        success=True,
        content_type="article",
        data=ArticleData(
            id=123456,
            title="Test Article Title for Pipeline Integration",
            content="<h1>Article Content</h1><p>This is the <em>article</em> content.</p>",
            excerpt="Article Content This is the article content.",
            metadata=ContentMetadata(
                author=AuthorInfo(
                    name="Test Article Author",
                    url_token="test-article-author",
                    headline="Writer and Developer"
                ),
                voteup_count=5678,
                comment_count=89,
                created_time=1705226400,  # 2024-01-14 10:00:00 UTC
                updated_time=None
            )
        ),
        comments=[
            CommentData(
                id=333,
                content="Excellent article!",
                author=CommentAuthor(name="Reader1", url_token="reader-1"),
                created_time=1705230000,
                like_count=28,
                child_count=1,
                is_author=False
            )
        ],
        error=None
    )


@pytest.fixture
def mock_comments_list():
    """Mock list of CommentData for testing comment formatting.

    Provides a realistic set of comments with various attributes.
    """
    from kb_workflow.crawler.models import CommentData, CommentAuthor

    return [
        CommentData(
            id=1001,
            content="This is the first comment with some content.",
            author=CommentAuthor(name="User1", url_token="user-1"),
            created_time=1705316400,
            like_count=100,
            child_count=5,
            is_author=False
        ),
        CommentData(
            id=1002,
            content="This is the second comment, from the author.",
            author=CommentAuthor(name="AuthorName", url_token="author-token"),
            created_time=1705320000,
            like_count=50,
            child_count=2,
            is_author=True
        ),
        CommentData(
            id=1003,
            content="Third comment here.",
            author=CommentAuthor(name="User3", url_token="user-3"),
            created_time=1705323600,
            like_count=25,
            child_count=0,
            is_author=False
        )
    ]


@pytest.fixture
def mock_zhihu_login_result():
    """Mock ZhihuLoginResult from ZhihuLogin.login_by_qrcode().

    Simulates successful QR code login result.
    Used for testing cookie refresh flow.
    """
    from kb_workflow.auth.zhihu_login import ZhihuLoginResult
    from pathlib import Path

    return ZhihuLoginResult(
        success=True,
        cookie_str="z_c0=test_cookie_value; _xsrf=test_xsrf; d_c0=test_d_c0",
        qrcode_path=Path("/tmp/test_qrcode.png"),
        error_message=None
    )


@pytest.fixture
def mock_crawl_failed_result():
    """Mock failed CrawlResult for error handling tests."""
    from kb_workflow.crawler.models import CrawlResult

    return CrawlResult(
        success=False,
        content_type="answer",
        data=None,
        comments=None,
        error="Cookie expired or invalid"
    )


class TestZhihuAdapterAnswerContent:
    """Tests for ZhihuAdapter answer extraction"""

    def test_get_content_type_answer(self):
        """Test content type detection for answer URLs"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test"  # Avoid config loading

        assert adapter.get_content_type("https://www.zhihu.com/question/123456/answer/789012") == "answer"
        assert adapter.get_content_type("https://zhihu.com/question/123/answer/456") == "answer"

    def test_get_content_type_zhuanlan(self):
        """Test content type detection for zhuanlan URLs"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test"  # Avoid config loading

        assert adapter.get_content_type("https://zhuanlan.zhihu.com/p/123456") == "zhuanlan"
        assert adapter.get_content_type("https://zhuanlan.zhihu.com/p/abc123") == "zhuanlan"

    def test_get_content_type_unknown(self):
        """Test content type detection for unknown URLs"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test"  # Avoid config loading

        assert adapter.get_content_type("https://www.zhihu.com/question/123456") == "unknown"
        assert adapter.get_content_type("https://www.zhihu.com/") == "unknown"
        assert adapter.get_content_type("https://example.com") == "unknown"

    def test_get_output_filename(self):
        """Test output filename generation"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test"  # Avoid config loading

        assert adapter.get_output_filename("789012", "answer", "metadata.json") == "zhihu_789012_metadata.json"
        assert adapter.get_output_filename("123456", "zhuanlan", "content.md") == "zhihu_123456_content.md"

    @pytest.mark.asyncio
    async def test_fetch_metadata_answer(self, mock_zhihu_answer_result):
        """Test fetching answer metadata"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test_cookie"  # Set cookie directly

        with patch.object(adapter, '_extract_with_playwright', return_value=mock_zhihu_answer_result):
            metadata = await adapter.fetch_metadata("789012", "answer")

        assert metadata["title"] == "Test Question Title"
        assert metadata["author"] == "Test Author"
        assert metadata["type"] == "answer"

    @pytest.mark.asyncio
    async def test_fetch_content_answer(self, mock_zhihu_answer_result):
        """Test fetching answer content as Markdown"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test_cookie"  # Set cookie directly

        with patch.object(adapter, '_extract_with_playwright', return_value=mock_zhihu_answer_result):
            content = await adapter.fetch_content("789012", "answer")

        assert "---" in content  # YAML frontmatter
        assert "title:" in content
        assert "Test Question Title" in content
        assert "This is the answer content" in content

    @pytest.mark.asyncio
    async def test_fetch_metadata_no_cookie_raises_auth_error(self):
        """Test that missing cookie raises AuthError"""
        adapter = ZhihuAdapter()
        adapter._cookie = ""  # Empty cookie

        with pytest.raises(AuthError) as exc_info:
            await adapter.fetch_metadata("789012", "answer")

        assert exc_info.value.auth_type == "zhihu_cookie"


class TestZhihuAdapterZhuanlanContent:
    """Tests for ZhihuAdapter zhuanlan (article) extraction"""

    @pytest.mark.asyncio
    async def test_fetch_metadata_zhuanlan(self, mock_zhihu_zhuanlan_result):
        """Test fetching zhuanlan metadata"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test_cookie"  # Set cookie directly

        with patch.object(adapter, '_extract_with_playwright', return_value=mock_zhihu_zhuanlan_result):
            metadata = await adapter.fetch_metadata("123456", "zhuanlan")

        assert metadata["title"] == "Test Article Title"
        assert metadata["author"] == "Test Author"
        assert metadata["type"] == "zhuanlan"

    @pytest.mark.asyncio
    async def test_fetch_content_zhuanlan(self, mock_zhihu_zhuanlan_result):
        """Test fetching zhuanlan content as Markdown"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test_cookie"  # Set cookie directly

        with patch.object(adapter, '_extract_with_playwright', return_value=mock_zhihu_zhuanlan_result):
            content = await adapter.fetch_content("123456", "zhuanlan")

        assert "---" in content  # YAML frontmatter
        assert "title:" in content
        assert "Test Article Title" in content
        assert "This is the article content" in content

    @pytest.mark.asyncio
    async def test_detect_login_page_raises_auth_error(self):
        """Test that login page detection raises AuthError"""
        adapter = ZhihuAdapter()
        adapter._cookie = "expired_cookie"  # Set cookie directly

        # Simulate login page response (empty content or login title)
        login_page_result = {
            "type": "answer",
            "title": "登录 - 知乎",
            "content": "",
            "content_markdown": "",
        }

        with patch.object(adapter, '_extract_with_playwright', return_value=login_page_result):
            with pytest.raises(AuthError) as exc_info:
                await adapter.fetch_metadata("789012", "answer")

        assert exc_info.value.auth_type == "zhihu_cookie"
        assert "过期" in str(exc_info.value.cause) or "expired" in str(exc_info.value.cause).lower()

    @pytest.mark.asyncio
    async def test_detect_short_content_raises_auth_error(self):
        """Test that very short content (< 100 chars) raises AuthError"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test_cookie"

        # Simulate response with very short content
        short_content_result = {
            "type": "answer",
            "title": "Some Title",
            "content": "x",  # Very short content
            "content_markdown": "",
        }

        with patch.object(adapter, '_extract_with_playwright', return_value=short_content_result):
            with pytest.raises(AuthError) as exc_info:
                await adapter.fetch_metadata("789012", "answer")

        assert exc_info.value.auth_type == "zhihu_cookie"


class TestZhihuAdapterCookieHandling:
    """Tests for cookie handling"""

    def test_cookie_from_settings(self):
        """Test cookie is loaded from settings"""
        with patch('kb_workflow.pipeline.platforms.zhihu.get_settings') as mock_settings:
            mock_secret = MagicMock()
            mock_secret.get_secret_value.return_value = "test_cookie_value"
            mock_settings.return_value.zhihu_cookie = mock_secret

            adapter = ZhihuAdapter()
            assert adapter.cookie == "test_cookie_value"

    def test_platform_constant(self):
        """Test PLATFORM constant"""
        adapter = ZhihuAdapter()
        adapter._cookie = "test"  # Avoid config loading
        assert adapter.PLATFORM == "zhihu"


# =============================================================================
# Wave 0 Test Skeletons for Pipeline Integration (Phase 09)
# These tests are marked skip until ZhihuAdapter is updated to use ZhihuCrawler
# =============================================================================


@pytest.mark.skip(reason="Wave 0 skeleton - awaiting ZhihuCrawler integration")
class TestZhihuAdapterCrawlerIntegration:
    """Test skeletons for ZhihuAdapter integration with ZhihuCrawler.

    These tests will verify the adapter correctly uses the new ZhihuCrawler
    instead of ZhihuPlaywrightExtractor.
    """

    @pytest.mark.asyncio
    async def test_fetch_metadata_answer(self, mock_crawl_answer_result):
        """Test fetching answer metadata via ZhihuCrawler.

        Verifies:
        - ZhihuCrawler.crawl_answer() is called with correct ID
        - Metadata is correctly extracted from CrawlResult
        - Output format matches expected adapter response
        """
        pass

    @pytest.mark.asyncio
    async def test_fetch_metadata_article(self, mock_crawl_article_result):
        """Test fetching zhuanlan article metadata via ZhihuCrawler.

        Verifies:
        - ZhihuCrawler.crawl_article() is called with correct ID
        - Metadata is correctly extracted from CrawlResult
        - Output format matches expected adapter response
        """
        pass

    @pytest.mark.asyncio
    async def test_fetch_content_with_comments(
        self, mock_crawl_answer_result, mock_comments_list
    ):
        """Test fetching content with comments appended.

        Verifies:
        - Content is fetched with include_comments=True
        - Comments are appended to content in correct Markdown format
        - Comment format: numbered list with author mention
        """
        pass

    @pytest.mark.asyncio
    async def test_cookie_expired_triggers_login(
        self, mock_crawl_failed_result, mock_zhihu_login_result
    ):
        """Test that expired cookie triggers QR code login flow.

        Verifies:
        - CrawlResult with success=False triggers login
        - ZhihuLogin.login_by_qrcode() is called
        - New cookie is used to retry the fetch
        - User sees QR code prompt
        """
        pass

    @pytest.mark.asyncio
    async def test_idempotent_execution(self, mock_crawl_answer_result):
        """Test that repeated calls with same content_id are idempotent.

        Verifies:
        - Multiple calls with same ID produce same result
        - No duplicate fetches are made
        - State is correctly managed
        """
        pass

    def test_comment_formatting(self, mock_comments_list):
        """Test comment formatting to Markdown.

        Verifies:
        - Comments are formatted as numbered list
        - Author name is bold/mentioned
        - Like count is included
        - Format matches expected style:
          1. **@AuthorName** (100 likes): Comment content...
        """
        pass


# WAVE_0_COMPLETE: Test scaffolding ready for implementation