recently_viewed.yaml: the no-delete merge covers files outside the window

Without a database version, newer history in the database wins (its
progress is kept; a newer plain view keeps its `viewed` and takes the
config's position).  That applied only to files among the newest views;
a file listed in the config that had fallen out of them was "added",
and its stale entry replaced the newer history.  prepare now reads the
database's history of added files and applies the same rule.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

wrolpi/files/progress.py

@@ -278,28 +278,33 @@ class RecentlyViewedConfig(ConfigFile):
             diff.removed = {i for i in diff.removed
                             if oldest and db_rows[i]['viewed'] >= oldest and i not in malformed}
 
-        # Without deletes (no database version), progress written after the config is kept; a newer plain view keeps
-        # its `viewed`, but it is not a newer place to resume.
-        if not self.import_deletes:
-            for key, row in list(diff.changed.items()):
-                if db_rows[key]['viewed'] >= row['viewed']:
-                    if db_rows[key]['progress'] is not None:
-                        del diff.changed[key]
-                    else:
-                        diff.changed[key] = dict(row, viewed=db_rows[key]['viewed'])
-
-        # The files of new entries which are not yet in the database; a deleted file's history goes with it.
+        # New entries which are in the database (outside the newest views), and which are not.
         from wrolpi.files.lib import glob_shared_stem
         from wrolpi.files.models import FileGroup
         media_directory = get_media_directory()
-        known = set()
+        known = dict()
         with get_db_session() as session:
             for paths in chunks([str(media_directory / i) for i in diff.added], 500):
-                known.update(str(i) for i, in
-                             session.query(FileGroup.primary_path).filter(FileGroup.primary_path.in_(paths)))
+                known.update((str(path.relative_to(media_directory)), _viewed_row(*values)) for path, *values in
+                             session.query(FileGroup.primary_path, FileGroup.viewed, FileGroup.progress,
+                                           FileGroup.position).filter(FileGroup.primary_path.in_(paths)))
+
+        # Without deletes (no database version), progress written after the config is kept; a newer plain view keeps
+        # its `viewed`, but it is not a newer place to resume.  This applies to a file that fell out of the newest
+        # views too.
+        if not self.import_deletes:
+            for rows, current in ((diff.changed, db_rows), (diff.added, known)):
+                for key, row in list(rows.items()):
+                    if key in current and current[key]['viewed'] and current[key]['viewed'] >= row['viewed']:
+                        if current[key]['progress'] is not None:
+                            del rows[key]
+                        else:
+                            rows[key] = dict(row, viewed=current[key]['viewed'])
+
+        # The files of new entries which are not yet in the database; a deleted file's history goes with it.
         for key, row in list(diff.added.items()):
             absolute_path = media_directory / key
-            if str(absolute_path) in known:
+            if key in known:
                 continue
             if absolute_path.is_file():
                 row['files'] = glob_shared_stem(absolute_path)

wrolpi/files/test/test_recently_viewed_import.py

@@ -156,3 +156,32 @@ async def test_refresh_changing_primary_file_counts(test_session, test_directory
     test_session.expire_all()
     assert test_session.query(FileGroup).one().primary_path != test_directory / 'video.mp4'
     assert get_config_version(test_session, 'recently_viewed.yaml') > 0
+
+
+@pytest.mark.asyncio
+async def test_without_deletes_newer_history_outside_the_window_is_kept(test_session, test_directory, async_client,
+                                                                        make_files_structure, monkeypatch):
+    """Without a database version, a newer view in the database wins even when it fell out of the newest views: its
+    progress is kept, or (a plain view) its `viewed` is kept and the config's position is taken."""
+    monkeypatch.setattr('wrolpi.files.progress.RECENTLY_VIEWED_LIMIT', 2)
+    now = dates.now()
+    file_groups = {name: FileGroup.from_paths(test_session, *make_files_structure([name]))
+                   for name in ('progress.txt', 'plain.txt', 'b.txt', 'c.txt')}
+    file_groups['progress.txt'].viewed, file_groups['progress.txt'].progress = now - timedelta(hours=3), 0.8
+    file_groups['plain.txt'].viewed = now - timedelta(hours=3)
+    file_groups['b.txt'].viewed, file_groups['c.txt'].viewed = now - timedelta(hours=2), now - timedelta(hours=1)
+    test_session.commit()
+    config = get_recently_viewed_config()
+    config.get_file().parent.mkdir(exist_ok=True)
+    config.get_file().write_text(yaml.dump(dict(version=1, files=[
+        dict(path='progress.txt', viewed=(now - timedelta(days=1)).isoformat(), progress=0.1),
+        dict(path='plain.txt', viewed=(now - timedelta(days=1)).isoformat(), progress=0.5,
+             position=dict(kind='time', seconds=60)),
+    ])))
+
+    config.import_config()
+
+    found = viewed(test_session)
+    assert (found['progress.txt'].viewed, found['progress.txt'].progress) == (now - timedelta(hours=3), 0.8)
+    assert (found['plain.txt'].viewed, found['plain.txt'].progress, found['plain.txt'].position) == \
+           (now - timedelta(hours=3), 0.5, dict(kind='time', seconds=60.0))