make stats count images

web/test_stats.py

@@ -0,0 +1,134 @@
+#!/usr/bin/env python3
+"""
+Tests für den Statserver (web/app.py), ohne Datenbank.
+
+Geprüft werden die DB-freien Teile: Parsen der psql-Zeile, der
+2-Stunden-Rhythmus für den exakten Bild-Count, die Wahl zwischen
+exakter Zählung und Stichprobe sowie die Einbindung der neuen
+Images-Anzeige in die HTML-Seite.
+"""
+
+import importlib.util
+import os
+import subprocess
+import unittest
+from datetime import datetime, timedelta
+from unittest import mock
+
+HERE = os.path.dirname(os.path.abspath(__file__))
+
+# web/app.py unter einem eindeutigen Modulnamen laden, damit es nicht mit
+# priorityweb/app.py (gleicher Dateiname) kollidiert.
+_spec = importlib.util.spec_from_file_location(
+    "jonassearch_stats_app", os.path.join(HERE, "app.py")
+)
+stats_app = importlib.util.module_from_spec(_spec)
+_spec.loader.exec_module(stats_app)
+
+
+class ParseStatsRowTest(unittest.TestCase):
+
+    def test_count_and_images(self):
+        self.assertEqual(
+            stats_app._parse_stats_row("981981|12345"), (981981, 12345)
+        )
+
+    def test_empty_sample_is_none(self):
+        self.assertEqual(stats_app._parse_stats_row("10|"), (10, None))
+        self.assertEqual(stats_app._parse_stats_row("10|\n"), (10, None))
+
+
+class ImagesExactDueTest(unittest.TestCase):
+
+    NOW = datetime(2026, 10, 17, 12, 0, 0)
+
+    def _entry(self, age, exact):
+        return {
+            "timestamp": (self.NOW - age).isoformat(),
+            "count": 1,
+            "images": 2,
+            "images_exact": exact,
+        }
+
+    def test_no_data_means_due(self):
+        self.assertTrue(stats_app.images_exact_due([], now=self.NOW))
+
+    def test_recent_exact_is_not_due(self):
+        data = [self._entry(timedelta(hours=1), True)]
+        self.assertFalse(stats_app.images_exact_due(data, now=self.NOW))
+
+    def test_two_hours_old_is_due(self):
+        data = [self._entry(timedelta(hours=2), True)]
+        self.assertTrue(stats_app.images_exact_due(data, now=self.NOW))
+
+    def test_estimates_in_between_do_not_reset_schedule(self):
+        data = [
+            self._entry(timedelta(hours=1, minutes=30), True),
+            self._entry(timedelta(minutes=30), False),
+        ]
+        self.assertFalse(stats_app.images_exact_due(data, now=self.NOW))
+
+    def test_old_entries_without_flag_are_ignored(self):
+        data = [{"timestamp": self.NOW.isoformat(), "count": 1}]
+        self.assertTrue(stats_app.images_exact_due(data, now=self.NOW))
+
+    def test_malformed_exact_timestamp_is_ignored(self):
+        data = [{"timestamp": "kaputt", "images_exact": True}]
+        self.assertTrue(stats_app.images_exact_due(data, now=self.NOW))
+
+
+class GetStatsTest(unittest.TestCase):
+
+    @mock.patch.object(stats_app, "_psql_scalar")
+    def test_exact_query(self, psql):
+        psql.return_value = "100|250"
+        self.assertEqual(stats_app.get_stats(True), (100, 250, True))
+        self.assertEqual(psql.call_args[0][0], stats_app.EXACT_IMAGE_SQL)
+        self.assertNotIn("TABLESAMPLE", stats_app.EXACT_IMAGE_SQL)
+
+    @mock.patch.object(stats_app, "_psql_scalar")
+    def test_estimate_query(self, psql):
+        psql.return_value = "100|250"
+        self.assertEqual(stats_app.get_stats(False), (100, 250, False))
+        self.assertEqual(psql.call_args[0][0], stats_app.ESTIMATE_IMAGE_SQL)
+        self.assertIn("TABLESAMPLE SYSTEM (1)", stats_app.ESTIMATE_IMAGE_SQL)
+
+    @mock.patch.object(stats_app, "_psql_scalar")
+    def test_empty_sample_on_small_table_falls_back_to_exact(self, psql):
+        psql.side_effect = ["10|", "10|20"]
+        self.assertEqual(stats_app.get_stats(False), (10, 20, True))
+        self.assertEqual(psql.call_count, 2)
+        self.assertEqual(psql.call_args_list[1][0][0],
+                         stats_app.EXACT_IMAGE_SQL)
+
+    @mock.patch.object(stats_app, "_psql_scalar")
+    def test_empty_sample_on_large_table_stays_unknown(self, psql):
+        psql.return_value = "200000|"
+        self.assertEqual(stats_app.get_stats(False), (200000, None, False))
+
+    @mock.patch.object(stats_app, "_psql_scalar")
+    def test_missing_pictures_column_falls_back_to_page_count(self, psql):
+        error = subprocess.CalledProcessError(
+            1, ["psql"], stderr='ERROR: column "pictures" does not exist'
+        )
+        psql.side_effect = [error, "10"]
+        self.assertEqual(stats_app.get_stats(False), (10, None, False))
+        self.assertEqual(psql.call_args_list[1][0][0],
+                         "SELECT COUNT(*) FROM pages;")
+
+
+class StatsPageTest(unittest.TestCase):
+
+    def test_page_contains_image_counter(self):
+        client = stats_app.app.test_client()
+        response = client.get("/")
+        self.assertEqual(response.status_code, 200)
+        html = response.get_data(as_text=True)
+        self.assertIn('id="current-images"', html)
+        self.assertIn('id="current-images-meta"', html)
+        self.assertIn("Estimated · exact every 2 h", html)
+        self.assertIn("imagesExact", html)
+
+
+if __name__ == "__main__":
+    unittest.main(verbosity=2)