simplify caching

2025-08-29 16:31:10 +02:00
parent d5f686bf1e
commit 2dac771d47
4 changed files with 102 additions and 17 deletions
--- a/tests/test_cache_paths.py
+++ b/tests/test_cache_paths.py
@@ -0,0 +1,66 @@
+import os
+import tempfile
+import pytest
+
+import web.db as db
+from web.utils import get_cache_dir
+
+
+# Skip unless explicitly enabled (MySQL integration expected)
+if not os.getenv("RUN_DB_TESTS"):
+    pytest.skip("Set RUN_DB_TESTS=1 to run cache path integration tests",
+                allow_module_level=True)
+
+
+def test_db_sync_inserts_relative_paths(tmp_path, monkeypatch):
+    # arrange: create a temporary cache dir and a fake html file
+    cache_dir = tmp_path / "cache"
+    cache_dir.mkdir()
+    f = cache_dir / "example.org_path_to_page_123.html"
+    f.write_text("<html>ok</html>")
+
+    # point app at this cache dir
+    monkeypatch.setenv("PYTEST_CACHE_DIR", str(cache_dir))
+    # monkeypatch get_cache_dir used by db functions
+    monkeypatch.setattr('web.utils.get_cache_dir', lambda: str(cache_dir))
+
+    # ensure DB initialized
+    db.db_init()
+
+    # act
+    db.db_sync_cached_pages(str(cache_dir))
+
+    # assert: DB contains relative path, not absolute
+    rows = db.db_get_all_cached_pages()
+    assert any(r['file_path'] == os.path.relpath(
+        str(f), start=str(cache_dir)) for r in rows)
+
+
+def test_normalize_cached_page_paths_converts_absolute(tmp_path, monkeypatch):
+    cache_dir = tmp_path / "cache"
+    cache_dir.mkdir()
+    # create an actual file
+    f = cache_dir / "site_example_page_1.html"
+    f.write_text("<html>ok</html>")
+
+    monkeypatch.setattr('web.utils.get_cache_dir', lambda: str(cache_dir))
+    db.db_init()
+
+    abs_fp = str(f)
+    rel_fp = os.path.relpath(abs_fp, start=str(cache_dir))
+
+    # Insert an absolute path row directly (simulate legacy data)
+    with db._ensure_session() as session:
+        session.execute(
+            db.text("INSERT INTO cached_pages(file_path, url_guess, last_modified, size_bytes, job_id) VALUES(:fp, :ug, :lm, :sz, :jid)"),
+            {"fp": abs_fp, "ug": "https://example.org/page1.html",
+                "lm": None, "sz": 10, "jid": None}
+        )
+        session.commit()
+
+    # normalize should convert absolute to relative
+    changed = db.normalize_cached_page_paths()
+    assert changed >= 1
+
+    rows = db.db_get_all_cached_pages()
+    assert any(r['file_path'] == rel_fp for r in rows)