Match more videos

Author: marcoacierno

backend/conferences/admin/conference.py

@@ -18,7 +18,7 @@
     OrderedTabularInline,
 )
 from itertools import permutations
-from unicodedata import normalize
+from unicodedata import combining, normalize
 from conferences.models import ConferenceVoucher
 from schedule.models import ScheduleItem
 from sponsors.models import SponsorLevel
@@ -307,14 +307,30 @@ def run_video_uploaded_path_matcher(self, request, object_id, ignore_cache):
         conference = Conference.objects.get(pk=object_id)
         all_events = (
             conference.schedule_items.select_related(
-                "submission__speaker", "keynote", "language"
+                "submission__speaker", "keynote", "language", "slot__day"
             )
             .prefetch_related(
                 "additional_speakers__user",
                 "keynote__speakers__user",
             )
             .all()
         )
+        # files are numbered by talk day (D1 = first day with talks),
+        # days with only workshops/trainings don't count
+        talk_days = (
+            conference.days.filter(
+                slots__items__type__in=[
+                    ScheduleItem.TYPES.submission,
+                    ScheduleItem.TYPES.talk,
+                    ScheduleItem.TYPES.keynote,
+                    ScheduleItem.TYPES.panel,
+                ]
+            )
+            .order_by("day")
+            .values_list("day", flat=True)
+            .distinct()
+        )
+        day_numbers = {day: number for number, day in enumerate(talk_days, start=1)}
 
         cache_key = f"{conference.code}:video-upload-files-cache"
         files = cache.get(cache_key)
@@ -332,7 +348,9 @@ def run_video_uploaded_path_matcher(self, request, object_id, ignore_cache):
         used_files = set()
 
         for event in all_events:
-            video_uploaded_path = self.match_event_to_video_file(event, files)
+            video_uploaded_path = self.match_event_to_video_file(
+                event, files, day_numbers
+            )
             event.video_uploaded_path = video_uploaded_path
             event.save(update_fields=["video_uploaded_path"])
 
@@ -361,15 +379,27 @@ def run_video_uploaded_path_matcher(self, request, object_id, ignore_cache):
                 messages.WARNING,
             )
 
-    def match_event_to_video_file(self, event, files):
+    def match_event_to_video_file(self, event, files, day_numbers):
         possible_file_names = []
 
         def best_name(speaker):
             return cleanup_string(speaker.full_name.strip() or speaker.name.strip())
 
-        normalized_files = [
-            (cleanup_string(video_file), video_file) for video_file in files
-        ]
+        event_day = None
+        if event.slot_id:
+            event_day = day_numbers.get(event.slot.day.day)
+
+        # when both the event and the file have a known conference day,
+        # a mismatch means the file belongs to another day's event
+        normalized_files = []
+        for video_file in files:
+            normalized_video_file = cleanup_string(video_file)
+            file_day = extract_day_from_video_path(normalized_video_file)
+
+            if event_day and file_day and event_day != file_day:
+                continue
+
+            normalized_files.append((normalized_video_file, video_file))
 
         all_speakers_names = [best_name(speaker) for speaker in event.speakers]
 
@@ -418,9 +448,26 @@ def best_name(speaker):
                 if possible_file_name in video_file:
                     return original_video_file
 
+        # multi-speaker talks are sometimes uploaded with the name
+        # of only one of the speakers
+        if count_speakers > 1:
+            for video_file, original_video_file in normalized_files:
+                if "," in video_file:
+                    continue
+
+                if any(name in video_file for name in all_speakers_names):
+                    return original_video_file
+
         return ""
 
 
+def extract_day_from_video_path(normalized_video_path: str) -> int | None:
+    match = re.search(r"/d(\d+)/", normalized_video_path) or re.search(
+        r"\bday (\d+)\b", normalized_video_path
+    )
+    return int(match.group(1)) if match else None
+
+
 def walk_conference_videos_folder(storage, base_path):
     folders, files = storage.listdir(base_path)
     all_files = [f"{base_path}{file_}" for file_ in files]
@@ -566,6 +613,7 @@ def cleanup_string(string: str) -> str:
     new_string = normalize(
         "NFKD", "".join(char for char in string if char.isprintable())
     ).lower()
-    new_string = new_string.replace("-", " ")
+    new_string = "".join(char for char in new_string if not combining(char))
+    new_string = new_string.replace("-", " ").replace("_", " ")
     new_string = re.sub(r"\s+", " ", new_string)
     return new_string.strip()

backend/conferences/tests/test_admin.py

@@ -1,6 +1,10 @@
+from datetime import date, time
+
 from schedule.tests.factories import (
+    DayFactory,
     ScheduleItemAdditionalSpeakerFactory,
     ScheduleItemFactory,
+    SlotFactory,
 )
 from submissions.tests.factories import SubmissionFactory
 from conferences.tests.factories import (
@@ -645,6 +649,130 @@ def test_video_uploaded_path_matcher(
     )
 
 
+def test_video_uploaded_path_matcher_with_day_signal(rf, mocker):
+    conference = ConferenceFactory(code="pycon2026")
+    # workshops only on the first day: videos are numbered by talk day,
+    # so D1 in the uploaded files is the second conference day
+    workshops_day = DayFactory(conference=conference, day=date(2026, 5, 27))
+    day_1 = DayFactory(conference=conference, day=date(2026, 5, 28))
+    day_2 = DayFactory(conference=conference, day=date(2026, 5, 29))
+    day_3 = DayFactory(conference=conference, day=date(2026, 5, 30))
+
+    marco = UserFactory(name="Marco", full_name="Marco Santoni")
+    igor = UserFactory(name="Igor", full_name="Igor Saggese")
+    marc_andre = UserFactory(name="Marc-André", full_name="Marc-André Lemburg")
+    trainer = UserFactory(name="Trainer", full_name="Some Trainer")
+
+    mocker.patch(
+        "conferences.admin.conference.walk_conference_videos_folder",
+        return_value=[
+            "conference-videos/pycon2026/SPAGHETTI/D1/02 - igor-saggese.mp4",
+            "conference-videos/pycon2026/SPAGHETTI/D2/03 - igor-saggese.mp4",
+            "conference-videos/pycon2026/TORTELLINI/D1/05 - lightning_talks_day_1.mp4",
+            "conference-videos/pycon2026/TORTELLINI/D2/04 - lightning_talks_day_2.mp4",
+            "conference-videos/pycon2026/TORTELLINI/D3/05 - lightning_talks_day_3.mp4",
+            "conference-videos/pycon2026/TORTELLINI/D3/01 - marc-andre-lemburg.mp4",
+        ],
+    )
+
+    def slot_for(day):
+        return SlotFactory(day=day, hour=time(10, 0), duration=45)
+
+    workshop_event = ScheduleItemFactory(
+        conference=conference,
+        title="Workshop",
+        type=ScheduleItem.TYPES.training,
+        submission__speaker=trainer,
+        slot=slot_for(workshops_day),
+    )
+
+    multi_speaker_event = ScheduleItemFactory(
+        conference=conference,
+        title="Talk with two speakers",
+        type=ScheduleItem.TYPES.talk,
+        submission__speaker=marco,
+        slot=slot_for(day_1),
+    )
+    ScheduleItemAdditionalSpeakerFactory(scheduleitem=multi_speaker_event, user=igor)
+
+    igor_solo_event = ScheduleItemFactory(
+        conference=conference,
+        title="Igor solo talk",
+        type=ScheduleItem.TYPES.talk,
+        submission__speaker=igor,
+        slot=slot_for(day_2),
+    )
+
+    lightning_talks_events = [
+        ScheduleItemFactory(
+            conference=conference,
+            title="Lightning Talks",
+            type=ScheduleItem.TYPES.custom,
+            submission=None,
+            slot=slot_for(day),
+        )
+        for day in (day_1, day_2, day_3)
+    ]
+
+    marc_andre_event = ScheduleItemFactory(
+        conference=conference,
+        title="Accented speaker talk",
+        type=ScheduleItem.TYPES.talk,
+        submission__speaker=marc_andre,
+        slot=slot_for(day_3),
+    )
+
+    admin = ConferenceAdmin(
+        model=conference.__class__,
+        admin_site=AdminSite(),
+    )
+    admin.message_user = mocker.Mock()
+
+    request = rf.post("/", data={"run_matcher": "1"})
+    ret = admin.map_videos(request, conference.id)
+
+    assert ret.status_code == 302
+
+    workshop_event.refresh_from_db()
+    multi_speaker_event.refresh_from_db()
+    igor_solo_event.refresh_from_db()
+    marc_andre_event.refresh_from_db()
+    for event in lightning_talks_events:
+        event.refresh_from_db()
+
+    assert workshop_event.video_uploaded_path == ""
+    # the file is named after only one of the two speakers,
+    # the day disambiguates it from igor's solo talk
+    assert (
+        multi_speaker_event.video_uploaded_path
+        == "conference-videos/pycon2026/SPAGHETTI/D1/02 - igor-saggese.mp4"
+    )
+    assert (
+        igor_solo_event.video_uploaded_path
+        == "conference-videos/pycon2026/SPAGHETTI/D2/03 - igor-saggese.mp4"
+    )
+    assert (
+        lightning_talks_events[0].video_uploaded_path
+        == "conference-videos/pycon2026/TORTELLINI/D1/05 - lightning_talks_day_1.mp4"
+    )
+    assert (
+        lightning_talks_events[1].video_uploaded_path
+        == "conference-videos/pycon2026/TORTELLINI/D2/04 - lightning_talks_day_2.mp4"
+    )
+    assert (
+        lightning_talks_events[2].video_uploaded_path
+        == "conference-videos/pycon2026/TORTELLINI/D3/05 - lightning_talks_day_3.mp4"
+    )
+    assert (
+        marc_andre_event.video_uploaded_path
+        == "conference-videos/pycon2026/TORTELLINI/D3/01 - marc-andre-lemburg.mp4"
+    )
+
+    assert admin.message_user.mock_calls[0].args[1] == "Matched 6 videos to events."
+    # every file was matched exactly once: no reuse or unused warnings
+    assert admin.message_user.call_count == 1
+
+
 def test_storage_walk_conference_videos_folder(mocker):
     mock_storage = mocker.Mock()
     mock_storage.listdir.side_effect = [