feat: add RF4MAP and RF4 Posts research parsers

2026-09-05 07:36:35 +07:00
parent 9cc3d44a49
commit 1a09bd59f3
9 changed files with 230 additions and 7 deletions
@@ -7,6 +7,8 @@ from rf4_research.community_sources import (
    CommunityParseError,
    parse_rf4db_catches,
    parse_rf4db_detail,
+    parse_rf4map_point,
+    parse_rf4posts_spot,
    parse_rf4stat_fishing,
    parse_rf4stat_posts,
 )
@@ -69,7 +71,53 @@ def test_parses_rf4stat_posts_without_using_locked_coordinates() -> None:
    assert row.evidence_urls == ("https://img.example.test/proof.jpg",)


+def test_parses_rf4map_individual_observations_into_common_contract() -> None:
+    rows = parse_rf4map_point(
+        fixture("rf4map_point_sample.html"), source_url="https://rf4map.ru/points/275",
+    )
+
+    assert len(rows) == 2
+    assert (rows[0].source_system, rows[0].source_external_id) == ("rf4map", "37100")
+    assert (rows[0].fish, rows[0].fish_external_id) == ("Линь", "98")
+    assert (rows[0].waterbody, rows[0].waterbody_external_id) == ("оз. Комариное", "16")
+    assert (rows[0].x, rows[0].y, rows[0].weight_g, rows[0].clip) == (53, 87, None, "8")
+    assert rows[1].bait == "Тесто медовое"
+    assert rows[1].evidence_urls == ("https://img.example.test/proof.jpg",)
+
+
+def test_parses_rf4posts_aggregate_spot_without_inventing_weights() -> None:
+    rows = parse_rf4posts_spot(
+        fixture("rf4posts_spot_sample.html"),
+        source_url="https://rf4-posts.com/ru/spots/d0c6d9c6-4ebf-49a7-98a8-9a562553a8ee",
+    )
+
+    assert [row.fish for row in rows] == ["Карп чешуйчатый", "Линь"]
+    assert rows[0].source_external_id.endswith(":carp")
+    assert (rows[0].x, rows[0].y, rows[0].weight_g) == (132, 147, None)
+    assert (rows[0].rig_type, rows[0].fishing_style, rows[0].clip) == ("method_popup", "feeder", "8")
+    assert rows[0].evidence_urls == ("https://img.example.test/spot.png",)
+
+
+def test_new_sources_keep_source_namespaces_for_compatible_records() -> None:
+    rf4map = parse_rf4map_point(
+        fixture("rf4map_point_sample.html"), source_url="https://rf4map.ru/points/275",
+    )[0]
+    rf4posts = parse_rf4posts_spot(
+        fixture("rf4posts_spot_sample.html"), source_url="https://rf4-posts.com/ru/spots/example",
+    )[1]
+
+    assert rf4map.fish == rf4posts.fish == "Линь"
+    assert rf4map.source_system != rf4posts.source_system
+    assert rf4map.source_external_id != rf4posts.source_external_id
+
+
@pytest.mark.parametrize("parser", [parse_rf4db_catches, parse_rf4stat_fishing, parse_rf4stat_posts])
 def test_parsers_reject_unrelated_html(parser) -> None:
    with pytest.raises(CommunityParseError):
        parser("<html><body>not a source page</body></html>")
+
+
+@pytest.mark.parametrize("parser", [parse_rf4map_point, parse_rf4posts_spot])
+def test_detail_parsers_reject_unrelated_html(parser) -> None:
+    with pytest.raises(CommunityParseError):
+        parser("<html><body>not a source page</body></html>", source_url="https://example.test/item/1")