Add the desk worklist and fix cfp_candidates series matching

tests/test_cfp_candidates.py

@@ -0,0 +1,220 @@
+"""Tests for cfp_candidates: archived series that have no entry in conferences.yml.
+
+A series counts as tracked when its name or any alt_name matches the name or
+alt_name of an entry in conferences.yml, after normalize_conference_name and
+the titles.yml mappings, ignoring case, accents and whitespace.
+"""
+
+import json
+import sys
+from datetime import date
+from pathlib import Path
+
+import pytest
+import yaml
+from freezegun import freeze_time
+
+sys.path.append(str(Path(__file__).parent.parent / "utils"))
+
+from cfp_candidates import entry_names
+from cfp_candidates import find_candidates
+from cfp_candidates import load_data
+from cfp_candidates import main
+from cfp_candidates import series_key
+
+# Real entries: archive.yml at 3d34de3, conferences.yml before 405fc07 and 219d0b0,
+# when the upcoming editions were spelt differently from their archived ones.
+ARCHIVE_XTREME_PYTHON = {
+    "conference": "Xtreme Python",
+    "year": 2025,
+    "link": "https://xtremepython.dev/2025/",
+    "cfp": "2025-09-20 23:59:00",
+    "place": "Online",
+    "start": date(2025, 11, 18),
+    "end": date(2025, 11, 18),
+    "sponsor": "https://xtremepython.dev/2025/sponsors/",
+    "sub": "PY",
+}
+CURRENT_XTREMEPYTHON = {
+    "conference": "XtremePython",
+    "year": 2026,
+    "link": "https://xtremepython.dev/",
+    "cfp": "TBA",
+    "place": "Online",
+    "start": date(2026, 11, 17),
+    "end": date(2026, 11, 17),
+    "sub": "PY",
+}
+ARCHIVE_PYHEP_DEV = {
+    "conference": "PyHEP.dev",
+    "alt_name": "PyHEP",
+    "year": 2025,
+    "link": "https://indico.cern.ch/event/1515852/",
+    "cfp": "2025-07-08 23:59:00",
+    "place": "Seattle, USA",
+    "start": date(2025, 7, 14),
+    "end": date(2025, 7, 17),
+    "sub": "SCIPY",
+}
+CURRENT_PYHEP = {
+    "conference": "PyHEP",
+    "year": 2026,
+    "link": "https://indico.cern.ch/category/7971/",
+    "cfp": "TBA",
+    "timezone": "Europe/Zurich",
+    "place": "Geneva, Switzerland",
+    "start": date(2026, 9, 7),
+    "end": date(2026, 9, 9),
+    "sub": "PY",
+}
+
+
+def edition(conference, year, **fields):
+    """Build a minimal archived edition."""
+    return {"conference": conference, "year": year, "link": f"https://example.org/{year}/", **fields}
+
+
+def names(candidates):
+    return [c["conference"] for c in candidates]
+
+
+class TestSeriesKey:
+    @pytest.mark.parametrize(
+        ("left", "right"),
+        [
+            ("Xtreme Python", "XtremePython"),
+            ("PyDay Copiapó", "PyDay Copiapo"),
+            ("PyCon mini Tokai", "PyCon Mini Tokai"),
+            ("PyCon NZ", "Kiwi PyCon"),
+            ("EuroPython Conference", "EuroPython"),
+            ("PyCon Greece 2027", "PyCon Greece"),
+        ],
+    )
+    def test_same_series(self, left, right):
+        assert series_key(left) == series_key(right)
+
+    @pytest.mark.parametrize(("left", "right"), [("PyHEP.dev", "PyHEP"), ("PyData Berlin", "PyData Paris")])
+    def test_different_series(self, left, right):
+        assert series_key(left) != series_key(right)
+
+
+class TestEntryNames:
+    def test_conference_then_alt_name(self):
+        assert entry_names({"conference": "PyHEP.dev", "alt_name": "PyHEP"}) == ["PyHEP.dev", "PyHEP"]
+
+    def test_skips_missing_and_blank(self):
+        assert entry_names({"conference": "PyCon", "alt_name": " "}) == ["PyCon"]
+
+
+class TestFindCandidates:
+    def test_xtreme_python_spelling_is_tracked(self):
+        """Archive "Xtreme Python" is tracked by an upcoming "XtremePython"."""
+        assert find_candidates([CURRENT_XTREMEPYTHON], [ARCHIVE_XTREME_PYTHON], current_year=2026) == []
+
+    def test_xtreme_python_spelling_reversed(self):
+        archive = [{**CURRENT_XTREMEPYTHON, "year": 2025}]
+        current = [{**ARCHIVE_XTREME_PYTHON, "year": 2026}]
+        assert find_candidates(current, archive, current_year=2026) == []
+
+    def test_pyhep_dev_alt_name_is_tracked(self):
+        """Archive "PyHEP.dev" (alt_name PyHEP) is tracked by an upcoming "PyHEP"."""
+        assert find_candidates([CURRENT_PYHEP], [ARCHIVE_PYHEP_DEV], current_year=2026) == []
+
+    def test_pyhep_tracked_through_upcoming_alt_name(self):
+        """Archive "PyHEP" is tracked by an upcoming "PyHEP.dev" whose alt_name is PyHEP."""
+        current = [{**ARCHIVE_PYHEP_DEV, "year": 2026}]
+        archive = [{**CURRENT_PYHEP, "year": 2025}]
+        assert find_candidates(current, archive, current_year=2026) == []
+
+    def test_reported_without_the_other_spelling(self):
+        candidates = find_candidates([], [ARCHIVE_XTREME_PYTHON, ARCHIVE_PYHEP_DEV], current_year=2026)
+        assert names(candidates) == ["PyHEP.dev", "Xtreme Python"]
+
+    def test_punctuation_is_not_folded(self):
+        """Without the alt_name, "PyHEP.dev" and "PyHEP" stay separate series."""
+        archive = [{k: v for k, v in ARCHIVE_PYHEP_DEV.items() if k != "alt_name"}]
+        assert names(find_candidates([CURRENT_PYHEP], archive, current_year=2026)) == ["PyHEP.dev"]
+
+    def test_accents_and_case(self):
+        current = [edition("PyDay Copiapo", 2026)]
+        archive = [edition("PyDay Copiapó", 2025), edition("pycon mini tokai", 2025)]
+        assert names(find_candidates(current, archive, current_year=2026)) == ["pycon mini tokai"]
+
+    def test_titles_mapping(self):
+        """titles.yml maps "PyCon NZ" to "Kiwi PyCon"."""
+        assert find_candidates([edition("Kiwi PyCon", 2026)], [edition("PyCon NZ", 2025)], current_year=2026) == []
+
+    def test_any_archived_alt_name_counts(self):
+        """The alt_name of an older edition still tracks the series, so a rename is not lost."""
+        archive = [edition("PyCon Old Name", 2024, alt_name="PyCon New Name"), edition("PyCon Old Name", 2025)]
+        current = [edition("PyCon New Name", 2026)]
+        assert find_candidates(current, archive, current_year=2026) == []
+
+    def test_host_alt_name_hides_the_series(self):
+        """Known cost of the rule above: an old edition named after its host hides the series.
+
+        The 2023 edition was hosted by PyCon Japan and carries it as alt_name, so the
+        upcoming PyCon Japan tracks PyCon Asia Pacific although it has no entry of its own.
+        """
+        archive = [
+            edition("PyCon Asia Pacific", 2025, alt_name="PyCon Philippines"),
+            edition("PyCon Asia Pacific", 2023, alt_name="PyCon Japan"),
+        ]
+        current = [edition("PyCon Japan", 2027)]
+        assert find_candidates(current, archive, current_year=2026) == []
+        assert names(find_candidates(current, archive[:1], current_year=2026)) == ["PyCon Asia Pacific"]
+
+    def test_spellings_grouped_into_one_series(self):
+        archive = [edition("PyCon Mini Tokai", 2024), edition("PyCon mini Tokai", 2025)]
+        candidates = find_candidates([], archive, current_year=2026)
+        assert [(c["conference"], c["latest_year"]) for c in candidates] == [("PyCon mini Tokai", 2025)]
+
+    def test_lookback(self):
+        archive = [edition("Old Conf", 2023), edition("Recent Conf", 2024)]
+        assert names(find_candidates([], archive, current_year=2026)) == ["Recent Conf"]
+        assert names(find_candidates([], archive, current_year=2026, lookback=3)) == ["Recent Conf", "Old Conf"]
+
+    def test_carries_fields_of_latest_edition(self):
+        archive = [
+            edition("PyCon Example", 2024, place="Old Town", sub="PY"),
+            edition("PyCon Example", 2025, place="New Town", sub="PY", cfp_link="https://example.org/cfp"),
+        ]
+        assert find_candidates([], archive, current_year=2026) == [
+            {
+                "conference": "PyCon Example",
+                "latest_year": 2025,
+                "link": "https://example.org/2025/",
+                "place": "New Town",
+                "sub": "PY",
+                "cfp_link": "https://example.org/cfp",
+            },
+        ]
+
+    def test_order_newest_first_then_name(self):
+        archive = [edition("B Conf", 2025), edition("A Conf", 2025), edition("C Conf", 2026)]
+        assert names(find_candidates([], archive, current_year=2026)) == ["C Conf", "A Conf", "B Conf"]
+
+
+class TestCli:
+    @freeze_time("2026-10-14")
+    def test_output_shape(self, tmp_path, monkeypatch, capsys):
+        data = tmp_path / "_data"
+        data.mkdir()
+        (data / "conferences.yml").write_text(yaml.safe_dump([CURRENT_XTREMEPYTHON]), encoding="utf-8")
+        (data / "archive.yml").write_text(
+            yaml.safe_dump([ARCHIVE_XTREME_PYTHON, edition("PyCon Example", 2025)]),
+            encoding="utf-8",
+        )
+        monkeypatch.setattr(sys, "argv", ["cfp_candidates.py", "--base", str(tmp_path)])
+
+        main()
+
+        output = json.loads(capsys.readouterr().out)
+        assert output == {
+            "generated": "2026",
+            "count": 1,
+            "candidates": [{"conference": "PyCon Example", "latest_year": 2025, "link": "https://example.org/2025/"}],
+        }
+
+    def test_load_data_missing_files(self, tmp_path):
+        assert load_data(tmp_path) == {"conferences": [], "archive": [], "legacy": []}