David Gwilliam
ccbdb84888
- Add pyproject.toml with modern Python packaging (PEP 517/518) - Add uv-based dependency management replacing inline venv bootstrap - Add requirements.txt and requirements-dev.txt for compatibility - Add mise.toml with dev tasks (test, lint, run, sync, ci) - Add .python-version pinned to Python 3.12 - Add comprehensive pytest test suite (73 tests) for: - engine/config, filter, terminal, sources, mic, ntfy modules - Configure pytest with coverage reporting (16% total, 100% on tested modules) - Configure ruff for linting with Python 3.10+ target - Remove redundant venv bootstrap code from mainline.py - Update .gitignore for uv/venv artifacts Run 'uv sync' to install dependencies, 'uv run pytest' to test.
94 lines
3.5 KiB
Python
94 lines
3.5 KiB
Python
"""
|
|
Tests for engine.filter module.
|
|
"""
|
|
|
|
from engine.filter import skip, strip_tags
|
|
|
|
|
|
class TestStripTags:
|
|
"""Tests for strip_tags function."""
|
|
|
|
def test_strips_simple_html(self):
|
|
"""Basic HTML tags are removed."""
|
|
assert strip_tags("<p>Hello</p>") == "Hello"
|
|
assert strip_tags("<b>Bold</b>") == "Bold"
|
|
assert strip_tags("<em>Italic</em>") == "Italic"
|
|
|
|
def test_strips_nested_html(self):
|
|
"""Nested HTML tags are handled."""
|
|
assert strip_tags("<div><p>Nested</p></div>") == "Nested"
|
|
assert strip_tags("<span><strong>Deep</strong></span>") == "Deep"
|
|
|
|
def test_strips_html_with_attributes(self):
|
|
"""HTML with attributes is handled."""
|
|
assert strip_tags('<a href="http://example.com">Link</a>') == "Link"
|
|
assert strip_tags('<img src="test.jpg" alt="test">') == ""
|
|
|
|
def test_handles_empty_string(self):
|
|
"""Empty string returns empty string."""
|
|
assert strip_tags("") == ""
|
|
assert strip_tags(None) == ""
|
|
|
|
def test_handles_plain_text(self):
|
|
"""Plain text without tags passes through."""
|
|
assert strip_tags("Plain text") == "Plain text"
|
|
|
|
def test_unescapes_html_entities(self):
|
|
"""HTML entities are decoded and tags are stripped."""
|
|
assert strip_tags(" test") == "test"
|
|
assert strip_tags("Hello & World") == "Hello & World"
|
|
|
|
def test_handles_malformed_html(self):
|
|
"""Malformed HTML is handled gracefully."""
|
|
assert strip_tags("<p>Unclosed") == "Unclosed"
|
|
assert strip_tags("</p>No start") == "No start"
|
|
|
|
|
|
class TestSkip:
|
|
"""Tests for skip function - content filtering."""
|
|
|
|
def test_skips_sports_content(self):
|
|
"""Sports-related headlines are skipped."""
|
|
assert skip("Football: Team wins championship") is True
|
|
assert skip("NBA Finals Game 7 results") is True
|
|
assert skip("Soccer match ends in draw") is True
|
|
assert skip("Premier League transfer news") is True
|
|
assert skip("Super Bowl halftime show") is True
|
|
|
|
def test_skips_vapid_content(self):
|
|
"""Vapid/celebrity content is skipped."""
|
|
assert skip("Kim Kardashian's new look") is True
|
|
assert skip("Influencer goes viral") is True
|
|
assert skip("Red carpet best dressed") is True
|
|
assert skip("Celebrity couple splits") is True
|
|
|
|
def test_allows_real_news(self):
|
|
"""Legitimate news headlines are allowed."""
|
|
assert skip("Scientists discover new planet") is False
|
|
assert skip("Economy grows by 3%") is False
|
|
assert skip("World leaders meet for summit") is False
|
|
assert skip("New technology breakthrough") is False
|
|
|
|
def test_case_insensitive(self):
|
|
"""Filter is case insensitive."""
|
|
assert skip("FOOTBALL scores") is True
|
|
assert skip("Football SCORES") is True
|
|
assert skip("Kardashian") is True
|
|
|
|
def test_word_boundary_matching(self):
|
|
"""Word boundary matching works correctly."""
|
|
assert skip("The football stadium") is True
|
|
assert skip("Footballer scores") is False
|
|
assert skip("Footballs on sale") is False
|
|
|
|
|
|
class TestIntegration:
|
|
"""Integration tests combining filter functions."""
|
|
|
|
def test_full_pipeline(self):
|
|
"""Test strip_tags followed by skip."""
|
|
html = '<p><a href="#">Breaking: Football championship final</a></p>'
|
|
text = strip_tags(html)
|
|
assert text == "Breaking: Football championship final"
|
|
assert skip(text) is True
|