Improve conference importer date handling

tests/test_add_new_conference.py

@@ -26,6 +26,19 @@ def test_extract_google_maps_latlon_at_pattern() -> None:
     assert result == (51.5242464, -0.0997024)
 
 
+def test_url_has_year_component() -> None:
+    """Only actual year or edition components should count as year-specific."""
+    cases = [
+        ("https://www.foss4gna.org/", False),
+        ("https://foss4g.asia/2026/", True),
+        ("https://2027.fossy.ca/", True),
+        ("https://www.socallinuxexpo.org/scale/24x/", True),
+        ("https://2026.stateofthebrowser.com/", True),
+    ]
+    for url, expected in cases:
+        assert add_new_conference.url_has_year_component(url) is expected
+
+
 def test_insert_sorted_allows_same_url_different_year_without_year_component() -> None:
     """The same non-year-specific URL can be reused for a different year."""
     conferences: list[dict[str, typing.Any]] = [
@@ -74,6 +87,86 @@ def test_insert_sorted_supports_nested_dates() -> None:
     assert [conf["name"] for conf in updated] == ["FOSDEM", "PyCascades"]
 
 
+def test_insert_sorted_updates_inexact_existing_entry() -> None:
+    """Exact dates should replace an existing inexact series entry."""
+    conferences: list[dict[str, typing.Any]] = [
+        {
+            "name": "PyCascades",
+            "series": "pycascades",
+            "topic": "Python",
+            "location": "Seattle, Washington",
+            "dates": {
+                "status": "approximate",
+                "label": "March 2027",
+                "earliest": date(2027, 3, 1),
+                "latest": date(2027, 3, 31),
+            },
+            "url": "https://2027.pycascades.com/",
+        }
+    ]
+    new_conf: dict[str, typing.Any] = {
+        "name": "PyCascades",
+        "series": "pycascades",
+        "topic": "Python",
+        "location": "Seattle, Washington",
+        "dates": {
+            "status": "exact",
+            "start": date(2027, 3, 12),
+            "end": date(2027, 3, 14),
+        },
+        "url": "https://2027.pycascades.com/",
+        "venue": "Example Hall",
+    }
+
+    updated = add_new_conference.insert_sorted(conferences, new_conf)
+
+    assert len(updated) == 1
+    assert updated[0]["dates"]["status"] == "exact"
+    assert updated[0]["dates"]["start"] == date(2027, 3, 12)
+    assert updated[0]["venue"] == "Example Hall"
+
+
+def test_normalize_dates_field_moves_legacy_dates() -> None:
+    """Legacy start/end model output should be converted before writing YAML."""
+    conf: dict[str, typing.Any] = {
+        "name": "PyCon",
+        "start": date(2026, 4, 10),
+        "end": date(2026, 4, 12),
+    }
+
+    add_new_conference.normalize_dates_field(conf)
+
+    assert "start" not in conf
+    assert "end" not in conf
+    assert conf["dates"] == {
+        "status": "exact",
+        "start": date(2026, 4, 10),
+        "end": date(2026, 4, 12),
+    }
+
+
+def test_build_prompt_includes_nested_dates_and_series() -> None:
+    """The prompt should describe nested dates and known series IDs."""
+    prompt = add_new_conference.build_prompt(
+        "https://example.com",
+        "Conference details",
+        None,
+        {
+            "pycascades": {
+                "name": "PyCascades",
+                "topic": "Python",
+                "usual_location": "Seattle, Washington",
+                "country": "us",
+            }
+        },
+    )
+
+    assert "Do not output legacy top-level `start`, `end`, or `date_status`" in prompt
+    assert "dates.status" in prompt
+    assert "- pycascades: PyCascades" in prompt
+    assert "March 2027" in prompt
+
+
 def test_validate_country_normalises_name() -> None:
     """Country names should be normalised to alpha-2 codes."""
     conf: dict[str, typing.Any] = {"country": "United Kingdom"}
@@ -197,6 +290,76 @@ def test_add_new_conference_updates_yaml(
     assert len(written) == 2
     assert written[1]["name"] == "NewConf"
     assert written[1]["country"] == "us"
-    assert written[1]["end"] == date(2026, 5, 3)
+    assert written[1]["dates"] == {
+        "status": "exact",
+        "start": date(2026, 5, 3),
+        "end": date(2026, 5, 3),
+    }
     assert written[1]["latitude"] == 40.0
     assert written[1]["longitude"] == -74.0
+
+
+def test_add_new_conference_reuses_generic_url_for_new_year(
+    tmp_path: typing.Any, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    """Generic URLs with digits in the domain should not be skipped early."""
+    yaml_path = tmp_path / "conferences.yaml"
+    yaml_path.write_text(
+        yaml.dump(
+            [
+                {
+                    "name": "FOSS4G North America",
+                    "series": "foss4g-north-america",
+                    "dates": {
+                        "status": "exact",
+                        "start": date(2025, 11, 3),
+                        "end": date(2025, 11, 5),
+                    },
+                    "url": "https://www.foss4gna.org/",
+                }
+            ],
+            sort_keys=False,
+        )
+    )
+
+    root = lxml.html.fromstring("<html><body>Conference details</body></html>")
+    monkeypatch.setattr(add_new_conference, "fetch_webpage", lambda url: root)
+    monkeypatch.setattr(
+        add_new_conference,
+        "webpage_to_text",
+        lambda parsed: "FOSS4G North America 2026",
+    )
+    monkeypatch.setattr(
+        add_new_conference, "detect_page_coordinates", lambda parsed: None
+    )
+    monkeypatch.setattr(
+        add_new_conference,
+        "get_from_open_ai",
+        lambda prompt: {
+            "yaml": yaml.dump(
+                {
+                    "name": "FOSS4G North America",
+                    "series": "foss4g-north-america",
+                    "topic": "Geospatial",
+                    "location": "St. Louis, Missouri",
+                    "country": "us",
+                    "dates": {
+                        "status": "exact",
+                        "start": date(2026, 10, 26),
+                        "end": date(2026, 10, 29),
+                    },
+                    "url": "https://www.foss4gna.org/",
+                },
+                sort_keys=False,
+            )
+        },
+    )
+
+    added = add_new_conference.add_new_conference(
+        "https://www.foss4gna.org/", str(yaml_path)
+    )
+
+    assert added is True
+    written = yaml.safe_load(yaml_path.read_text())
+    assert len(written) == 2
+    assert [conf["dates"]["start"].year for conf in written] == [2025, 2026]