|
| 1 | +"""Tests for SqliteCacheRequestsDocumentLoader and HTTP cache behavior.""" |
| 2 | + |
| 3 | +import json |
| 4 | +import threading |
| 5 | +from http.server import BaseHTTPRequestHandler, HTTPServer |
| 6 | +from pathlib import Path |
| 7 | + |
| 8 | +import pytest |
| 9 | + |
| 10 | +from pyld import ( |
| 11 | + DocumentLoader, |
| 12 | + RequestsDocumentLoader, |
| 13 | + SqliteCacheRequestsDocumentLoader, |
| 14 | +) |
| 15 | + |
| 16 | +requests_cache = pytest.importorskip('requests_cache') |
| 17 | +CachedSession = requests_cache.CachedSession |
| 18 | + |
| 19 | + |
| 20 | +class _ContextHandler(BaseHTTPRequestHandler): |
| 21 | + request_count = 0 |
| 22 | + |
| 23 | + def do_GET(self): |
| 24 | + type(self).request_count += 1 |
| 25 | + body = json.dumps({ |
| 26 | + '@context': {'name': 'http://example.org/name'}, |
| 27 | + }).encode() |
| 28 | + self.send_response(200) |
| 29 | + self.send_header('Content-Type', 'application/ld+json') |
| 30 | + self.send_header('Cache-Control', 'max-age=3600') |
| 31 | + self.end_headers() |
| 32 | + self.wfile.write(body) |
| 33 | + |
| 34 | + def log_message(self, format, *args): |
| 35 | + pass |
| 36 | + |
| 37 | + |
| 38 | +@pytest.fixture |
| 39 | +def context_url(): |
| 40 | + """Local HTTP server returning JSON-LD with Cache-Control: max-age=3600.""" |
| 41 | + _ContextHandler.request_count = 0 |
| 42 | + server = HTTPServer(('127.0.0.1', 0), _ContextHandler) |
| 43 | + thread = threading.Thread(target=server.serve_forever, daemon=True) |
| 44 | + thread.start() |
| 45 | + port = server.server_address[1] |
| 46 | + url = f'http://127.0.0.1:{port}/context.jsonld' |
| 47 | + yield url |
| 48 | + server.shutdown() |
| 49 | + |
| 50 | + |
| 51 | +def test_requests_document_loader_accepts_custom_session(): |
| 52 | + """RequestsDocumentLoader accepts a CachedSession via session=.""" |
| 53 | + loader = RequestsDocumentLoader( |
| 54 | + session=CachedSession(backend='memory', cache_control=True)) |
| 55 | + assert isinstance(loader, DocumentLoader) |
| 56 | + assert callable(loader) |
| 57 | + loader.session.close() |
| 58 | + |
| 59 | + |
| 60 | +def test_sqlite_cache_requests_document_loader_is_document_loader(): |
| 61 | + """Sqlite loader is a DocumentLoader composing RequestsDocumentLoader.""" |
| 62 | + loader = SqliteCacheRequestsDocumentLoader() |
| 63 | + assert isinstance(loader, DocumentLoader) |
| 64 | + assert isinstance(loader._loader, RequestsDocumentLoader) |
| 65 | + assert callable(loader) |
| 66 | + loader.session.close() |
| 67 | + |
| 68 | + |
| 69 | +def test_sqlite_cache_requests_document_loader_rejects_relative_sqlite_file_path(): |
| 70 | + """Relative sqlite_file_path is rejected.""" |
| 71 | + with pytest.raises(ValueError, match='absolute path'): |
| 72 | + SqliteCacheRequestsDocumentLoader( |
| 73 | + sqlite_file_path=Path('relative.sqlite')) |
| 74 | + |
| 75 | + |
| 76 | +def test_http_cache_headers_serve_from_cache_with_cache_control(context_url): |
| 77 | + """With cache_control=True, Cache-Control max-age avoids a second HTTP hit.""" |
| 78 | + loader = RequestsDocumentLoader( |
| 79 | + session=CachedSession( |
| 80 | + 'test_memory_cache_control', |
| 81 | + backend='memory', |
| 82 | + cache_control=True, |
| 83 | + )) |
| 84 | + loader(context_url) |
| 85 | + loader(context_url) |
| 86 | + assert _ContextHandler.request_count == 1 |
| 87 | + loader.session.close() |
| 88 | + |
| 89 | + |
| 90 | +def test_http_cache_headers_without_cache_control_hits_server_twice(context_url): |
| 91 | + """With cache_control=False, response Cache-Control headers are ignored.""" |
| 92 | + loader = RequestsDocumentLoader( |
| 93 | + session=CachedSession( |
| 94 | + 'test_memory_no_cache_control', |
| 95 | + backend='memory', |
| 96 | + cache_control=False, |
| 97 | + expire_after=0, |
| 98 | + )) |
| 99 | + loader(context_url) |
| 100 | + loader(context_url) |
| 101 | + assert _ContextHandler.request_count == 2 |
| 102 | + loader.session.close() |
| 103 | + |
| 104 | + |
| 105 | +def test_sqlite_cache_requests_document_loader_persists(context_url, tmp_path): |
| 106 | + """Second loader instance reuses the on-disk SQLite cache.""" |
| 107 | + cache_path = tmp_path / 'contexts.sqlite' |
| 108 | + loader = SqliteCacheRequestsDocumentLoader(sqlite_file_path=cache_path) |
| 109 | + loader(context_url) |
| 110 | + assert _ContextHandler.request_count == 1 |
| 111 | + loader.session.close() |
| 112 | + |
| 113 | + _ContextHandler.request_count = 0 |
| 114 | + loader = SqliteCacheRequestsDocumentLoader(sqlite_file_path=cache_path) |
| 115 | + loader(context_url) |
| 116 | + assert _ContextHandler.request_count == 0 |
| 117 | + loader.session.close() |
0 commit comments