US01-01 follow-up: replace donor characterization iteration with YAML ledger + full suite (#46)

tests/characterization/test_webapp_query.py

@@ -0,0 +1,105 @@
+"""Characterize webapp read layer (donor: query._fts_query, _where, search,
+facets, stats, photo, all_paths) against a seeded temp DB."""
+from pathlib import Path
+
+import pytest
+
+import photo_analyzer as pa
+from webapp import query
+from conftest import seed_analyzed
+
+
+def test_fts_query_sanitization_goldens():
+    assert query._fts_query("beach sun") == '"beach"* "sun"*'
+    assert query._fts_query('AND OR "quoted"') == '"AND"* "OR"* "quoted"*'
+    assert query._fts_query("!!! ???") is None, "pure punctuation can't crash MATCH"
+    assert query._fts_query("") is None
+    assert query._fts_query("Straße") == '"Straße"*', "unicode words survive"
+
+
+def test_where_clause_goldens():
+    clauses, params = query._where({
+        "setting": "indoor", "people": "3+", "year_min": "2010",
+        "has_location": "1"})
+    assert clauses == [
+        "p.setting = ?", "p.people_count >= 3", "p.approx_year >= ?",
+        "p.location_hint IS NOT NULL AND TRIM(p.location_hint) != '' "
+        "AND LOWER(p.location_hint) != 'null'"]
+    assert params == ["indoor", 2010]
+    assert query._where({}) == ([], [])
+
+
+@pytest.fixture
+def seeded(db):
+    seed_analyzed(db, "/lib/album_a/beach.jpg",
+                  description="A sunny beach with palm trees.",
+                  tags=["beach", "palm"], setting="outdoor", people_count=2,
+                  approx_year=2015, location_hint="Rome, Italy")
+    seed_analyzed(db, "/lib/album_a/city.jpg",
+                  description="A night city street.", tags=["city"],
+                  setting="outdoor", time_of_day="night", people_count=0,
+                  approx_year=2011)
+    seed_analyzed(db, "/lib/album_b/dinner.jpg",
+                  description="Family dinner at home.", tags=["family", "dinner"],
+                  setting="indoor", people_count=4)
+    pa.mark_exif_written(db, "/lib/album_b/dinner.jpg")
+    pa.upsert_pending(db, "/lib/album_b/err.jpg")
+    pa.mark_error(db, "/lib/album_b/err.jpg", "timeout")
+    return db
+
+
+def test_search_fts_prefix_match(seeded):
+    out = query.search(seeded, q="beach")
+    assert out["total"] == 1
+    assert out["rows"][0]["path"] == "/lib/album_a/beach.jpg"
+    assert out["rows"][0]["tags"] == ["beach", "palm"], "tags decoded from JSON"
+    assert query.search(seeded, q="bea")["total"] == 1, "prefix term"
+
+
+def test_search_filters_and_browse(seeded):
+    assert query.search(seeded)["total"] == 4, "empty query = browse all"
+    assert query.search(seeded, filters={"setting": "indoor"})["total"] == 1
+    assert query.search(seeded, filters={"people": "3+"})["total"] == 1
+    assert query.search(seeded, filters={"year_min": 2012})["total"] == 1
+    assert query.search(seeded, filters={"has_location": "1"})["total"] == 1
+    assert query.search(seeded, q="city", filters={"setting": "indoor"})["total"] == 0, \
+        "search ANDs with filters"
+
+
+def test_search_paging(seeded):
+    page = query.search(seeded, limit=2, offset=2)
+    assert page["total"] == 4
+    assert len(page["rows"]) == 2
+    assert page["offset"] == 2
+
+
+def test_photo_omits_raw_response(seeded):
+    d = query.photo(seeded, "/lib/album_a/beach.jpg")
+    assert d["description"] == "A sunny beach with palm trees."
+    assert "raw_response" not in d
+    assert query.photo(seeded, "/nope.jpg") is None
+
+
+def test_facets_and_albums(seeded):
+    f = query.facets(seeded, Path("/lib"))
+    assert (f["year_min"], f["year_max"]) == (2011, 2015)
+    assert {a["album"]: a["count"] for a in f["albums"]} == {
+        "album_a": 2, "album_b": 2}
+    assert {v["value"]: v["count"] for v in f["setting"]} == {
+        "outdoor": 2, "indoor": 1}
+
+
+def test_stats_album_progress_and_errors(seeded):
+    s = query.stats(seeded, Path("/lib"))
+    assert s["total"] == 4
+    assert s["status"] == {"analyzed": 2, "exif_written": 1, "error": 1}
+    albums = {a["album"]: (a["done"], a["total"]) for a in s["albums"]}
+    assert albums == {"album_a": (2, 2), "album_b": (1, 2)}, \
+        "done counts analyzed + exif_written"
+    assert s["errors"] == [{"path": "/lib/album_b/err.jpg", "error": "timeout"}]
+
+
+def test_all_paths_is_the_image_endpoint_allowlist(seeded):
+    assert query.all_paths(seeded) == {
+        "/lib/album_a/beach.jpg", "/lib/album_a/city.jpg",
+        "/lib/album_b/dinner.jpg", "/lib/album_b/err.jpg"}