Complete stage 0 source research

2026-09-02 19:52:32 +07:00
commit a6f91a1329
9 changed files with 1096 additions and 0 deletions
@@ -0,0 +1,40 @@
+<!doctype html>
+<html lang="de">
+  <body>
+    <!-- Reduced and anonymized from the public RF4 records DOM contract. -->
+    <div class="records flex_table">
+      <div class="row header">
+        <div class="col fish">Fisch</div>
+        <div class="col weight">Gewicht</div>
+        <div class="col location">Gewässer</div>
+        <div class="col bait">Kunstköder</div>
+        <div class="col gamername">Spieler</div>
+        <div class="col data">Datum</div>
+      </div>
+      <div class="rows">
+        <div class="row">
+          <div class="records_subtable flex_table">
+            <div class="row header">
+              <div class="col fish"><div class="text">Hecht</div></div>
+              <div class="col weight">27.902&nbsp;kg</div>
+              <div class="col location">Testsee</div>
+              <div class="col bait"><div class="bait_icon" title="Testköder 01"></div></div>
+              <div class="col gamername">Spieler A</div>
+              <div class="col data">11.02.26</div>
+            </div>
+            <div class="rows">
+              <div class="row">
+                <div class="col fish"></div>
+                <div class="col weight">2 519.264&nbsp;kg</div>
+                <div class="col location">Testmeer</div>
+                <div class="col bait"><div class="bait_icon" title="Testköder 02"></div></div>
+                <div class="col gamername">Spieler B</div>
+                <div class="col data">03.05.26</div>
+              </div>
+            </div>
+          </div>
+        </div>
+      </div>
+    </div>
+  </body>
+</html>
@@ -0,0 +1,47 @@
+from datetime import date
+from pathlib import Path
+import unittest
+
+from rf4_research.records import RecordsParseError, parse_records_html, parse_weight_g
+
+
+FIXTURE = Path(__file__).parent / "fixtures" / "records_ru_sample.html"
+
+
+class RecordsParserTests(unittest.TestCase):
+    def test_parses_group_header_and_nested_records(self) -> None:
+        records = parse_records_html(
+            FIXTURE.read_text(encoding="utf-8"),
+            region="ru",
+            category="records",
+            source_url="https://rf4game.de/records/region/RU/",
+            today=date(2026, 9, 2),
+        )
+
+        self.assertEqual(len(records), 2)
+        self.assertEqual(records[0].fish, "Hecht")
+        self.assertEqual(records[0].weight_g, 27_902)
+        self.assertEqual(records[0].bait, "Testköder 01")
+        self.assertEqual(records[1].fish, "Hecht")
+        self.assertEqual(records[1].weight_g, 2_519_264)
+        self.assertEqual(records[1].record_date, date(2026, 5, 3))
+
+    def test_rejects_changed_column_contract(self) -> None:
+        html = FIXTURE.read_text(encoding="utf-8").replace(
+            'class="col data"', 'class="col changed"', 1
+        )
+        with self.assertRaisesRegex(RecordsParseError, "records columns changed"):
+            parse_records_html(
+                html,
+                region="RU",
+                category="records",
+                source_url="fixture://changed",
+            )
+
+    def test_weight_units_are_normalized_to_grams(self) -> None:
+        self.assertEqual(parse_weight_g("423 g"), 423)
+        self.assertEqual(parse_weight_g("8.023 kg"), 8_023)
+
+
+if __name__ == "__main__":
+    unittest.main()