hvac-kia-content/tests/test_wordpress_scraper.py
Ben Reed 95e0499791 feat: Implement WordPress scraper with comprehensive tests
- Created WordPressScraper class extending BaseScraper
- Fetches posts with pagination support
- Enriches posts with author, category, and tag information
- Implements incremental updates via state management
- Word count calculation for content
- All 11 tests passing

🤖 Generated with Claude Code

Co-Authored-By: Claude <noreply@anthropic.com>
2025-08-18 12:19:56 -03:00

265 lines
No EOL
9.5 KiB
Python

import pytest
from unittest.mock import Mock, patch, MagicMock
from datetime import datetime
import json
from pathlib import Path
from src.wordpress_scraper import WordPressScraper
from src.base_scraper import ScraperConfig
class TestWordPressScraper:
@pytest.fixture
def config(self):
return ScraperConfig(
source_name="wordpress",
brand_name="hvacknowitall",
data_dir=Path("data"),
logs_dir=Path("logs"),
timezone="America/Halifax"
)
@pytest.fixture
def mock_env(self):
with patch.dict('os.environ', {
'WORDPRESS_URL': 'https://hvacknowitall.com/',
'WORDPRESS_USERNAME': 'test@example.com',
'WORDPRESS_API_KEY': 'test_api_key'
}):
yield
def test_initialization(self, config, mock_env):
scraper = WordPressScraper(config)
assert scraper.config == config
assert scraper.base_url == 'https://hvacknowitall.com/'
assert scraper.auth == ('test@example.com', 'test_api_key')
@patch('requests.get')
def test_fetch_posts_success(self, mock_get, config, mock_env):
# Mock successful API response
mock_response = Mock()
mock_response.status_code = 200
mock_response.json.return_value = [
{
'id': 1,
'title': {'rendered': 'Test Post 1'},
'author': 1,
'date': '2024-01-01T12:00:00',
'content': {'rendered': '<p>Test content 1</p>'},
'excerpt': {'rendered': '<p>Test excerpt 1</p>'},
'link': 'https://hvacknowitall.com/test-post-1',
'categories': [1, 2],
'tags': [3, 4]
},
{
'id': 2,
'title': {'rendered': 'Test Post 2'},
'author': 1,
'date': '2024-01-02T12:00:00',
'content': {'rendered': '<p>Test content 2</p>'},
'excerpt': {'rendered': '<p>Test excerpt 2</p>'},
'link': 'https://hvacknowitall.com/test-post-2',
'categories': [1],
'tags': [3]
}
]
mock_get.return_value = mock_response
scraper = WordPressScraper(config)
posts = scraper.fetch_posts()
assert len(posts) == 2
assert posts[0]['id'] == 1
assert posts[0]['title']['rendered'] == 'Test Post 1'
mock_get.assert_called_with(
'https://hvacknowitall.com/wp-json/wp/v2/posts',
params={'per_page': 100, 'page': 1},
auth=('test@example.com', 'test_api_key'),
timeout=30
)
@patch('requests.get')
def test_fetch_posts_with_pagination(self, mock_get, config, mock_env):
# First page response
mock_response1 = Mock()
mock_response1.status_code = 200
mock_response1.headers = {'X-WP-TotalPages': '2'}
mock_response1.json.return_value = [{'id': 1, 'title': {'rendered': 'Post 1'}}]
# Second page response
mock_response2 = Mock()
mock_response2.status_code = 200
mock_response2.headers = {'X-WP-TotalPages': '2'}
mock_response2.json.return_value = [{'id': 2, 'title': {'rendered': 'Post 2'}}]
mock_get.side_effect = [mock_response1, mock_response2]
scraper = WordPressScraper(config)
posts = scraper.fetch_posts()
assert len(posts) == 2
assert posts[0]['id'] == 1
assert posts[1]['id'] == 2
assert mock_get.call_count == 2
@patch('requests.get')
def test_fetch_posts_error_handling(self, mock_get, config, mock_env):
mock_get.side_effect = Exception("Connection error")
scraper = WordPressScraper(config)
posts = scraper.fetch_posts()
assert posts == []
@patch('requests.get')
def test_fetch_author_info(self, mock_get, config, mock_env):
mock_response = Mock()
mock_response.status_code = 200
mock_response.json.return_value = {
'id': 1,
'name': 'John Doe',
'slug': 'john-doe'
}
mock_get.return_value = mock_response
scraper = WordPressScraper(config)
author = scraper.fetch_author(1)
assert author['name'] == 'John Doe'
mock_get.assert_called_with(
'https://hvacknowitall.com/wp-json/wp/v2/users/1',
auth=('test@example.com', 'test_api_key'),
timeout=30
)
@patch('requests.get')
def test_fetch_categories(self, mock_get, config, mock_env):
# Mock individual category responses
mock_response1 = Mock()
mock_response1.status_code = 200
mock_response1.json.return_value = {'id': 1, 'name': 'Category 1'}
mock_response2 = Mock()
mock_response2.status_code = 200
mock_response2.json.return_value = {'id': 2, 'name': 'Category 2'}
mock_get.side_effect = [mock_response1, mock_response2]
scraper = WordPressScraper(config)
categories = scraper.fetch_categories([1, 2])
assert len(categories) == 2
assert categories[0]['name'] == 'Category 1'
assert categories[1]['name'] == 'Category 2'
@patch('requests.get')
def test_fetch_tags(self, mock_get, config, mock_env):
# Mock individual tag responses
mock_response1 = Mock()
mock_response1.status_code = 200
mock_response1.json.return_value = {'id': 3, 'name': 'Tag 1'}
mock_response2 = Mock()
mock_response2.status_code = 200
mock_response2.json.return_value = {'id': 4, 'name': 'Tag 2'}
mock_get.side_effect = [mock_response1, mock_response2]
scraper = WordPressScraper(config)
tags = scraper.fetch_tags([3, 4])
assert len(tags) == 2
assert tags[0]['name'] == 'Tag 1'
assert tags[1]['name'] == 'Tag 2'
def test_format_markdown(self, config, mock_env):
scraper = WordPressScraper(config)
# Mock enriched posts data
posts = [
{
'id': 1,
'title': {'rendered': 'Test Post 1'},
'author_name': 'John Doe',
'date': '2024-01-01T12:00:00',
'content': {'rendered': '<p>Test content 1</p>'},
'link': 'https://hvacknowitall.com/test-post-1',
'category_names': ['HVAC', 'Tips'],
'tag_names': ['maintenance', 'diy'],
'word_count': 500
}
]
markdown = scraper.format_markdown(posts)
assert '# ID: 1' in markdown
assert '## Title: Test Post 1' in markdown
assert '## Author: John Doe' in markdown
assert '## Publish Date: 2024-01-01T12:00:00' in markdown
assert '## Word Count: 500' in markdown
assert '## Categories: HVAC, Tips' in markdown
assert '## Tags: maintenance, diy' in markdown
assert '## Permalink: https://hvacknowitall.com/test-post-1' in markdown
def test_get_incremental_items(self, config, mock_env):
scraper = WordPressScraper(config)
posts = [
{'id': 3, 'date': '2024-01-03T12:00:00'},
{'id': 2, 'date': '2024-01-02T12:00:00'},
{'id': 1, 'date': '2024-01-01T12:00:00'}
]
# Test with no previous state
state = {}
new_posts = scraper.get_incremental_items(posts, state)
assert len(new_posts) == 3
# Test with existing state
state = {'last_post_id': 2, 'last_post_date': '2024-01-02T12:00:00'}
new_posts = scraper.get_incremental_items(posts, state)
assert len(new_posts) == 1
assert new_posts[0]['id'] == 3
def test_update_state(self, config, mock_env):
scraper = WordPressScraper(config)
state = {}
posts = [
{'id': 3, 'date': '2024-01-03T12:00:00'},
{'id': 2, 'date': '2024-01-02T12:00:00'}
]
updated_state = scraper.update_state(state, posts)
assert updated_state['last_post_id'] == 3
assert updated_state['last_post_date'] == '2024-01-03T12:00:00'
@patch('src.wordpress_scraper.WordPressScraper.fetch_tags')
@patch('src.wordpress_scraper.WordPressScraper.fetch_categories')
@patch('src.wordpress_scraper.WordPressScraper.fetch_author')
@patch('src.wordpress_scraper.WordPressScraper.fetch_posts')
def test_fetch_content(self, mock_fetch_posts, mock_fetch_author,
mock_fetch_categories, mock_fetch_tags, config, mock_env):
# Setup mock returns
mock_fetch_posts.return_value = [
{
'id': 1,
'title': {'rendered': 'Test Post'},
'author': 1,
'categories': [1],
'tags': [2],
'content': {'rendered': '<p>Content</p>'}
}
]
mock_fetch_author.return_value = {'name': 'John Doe'}
mock_fetch_categories.return_value = [{'name': 'HVAC'}]
mock_fetch_tags.return_value = [{'name': 'tips'}]
scraper = WordPressScraper(config)
content = scraper.fetch_content()
assert len(content) == 1
assert content[0]['author_name'] == 'John Doe'
assert content[0]['category_names'] == ['HVAC']
assert content[0]['tag_names'] == ['tips']
assert content[0]['word_count'] > 0