Files
vrobbler/vrobbler/apps/books/management/commands/merge_koreader_scrobbles.py
Colin Powell c71b51fdb8
Some checks failed
build / test (push) Has been cancelled
[books] Fix one-extra-scrobble bug on Koreader imports
2026-06-12 09:52:33 -04:00

131 lines
4.3 KiB
Python

import logging
from datetime import timedelta
from django.core.management.base import BaseCommand
from django.db import transaction
from books.koreader import SESSION_GAP_SECONDS, fix_long_play_stats_for_scrobbles
logger = logging.getLogger(__name__)
SESSION_GAP = timedelta(seconds=SESSION_GAP_SECONDS)
def _page_data_keys(pages):
return sorted(int(k) for k in (pages or {}))
class Command(BaseCommand):
help = "Merge orphaned 1-page KOReader scrobbles into the preceding scrobble"
def add_arguments(self, parser):
parser.add_argument(
"--commit",
action="store_true",
help="Commit changes to the database",
)
def handle(self, *args, **options):
from scrobbles.models import Scrobble
commit = options["commit"]
qs = Scrobble.objects.filter(
media_type="Book", source="KOReader"
).order_by("book_id", "timestamp")
if not qs.exists():
self.stdout.write("No KOReader book scrobbles found.")
return
merged = 0
affected_books = set()
# Group by book_id manually since we're iterating in order
book_scrobbles = {}
for s in qs:
book_scrobbles.setdefault(s.book_id, []).append(s)
if not commit:
self.stdout.write("Dry run — no changes will be saved. Use --commit to apply.")
for book_id, scrobbles in book_scrobbles.items():
batch_merged = 0
i = 0
while i < len(scrobbles) - 1:
current = scrobbles[i]
orphan = scrobbles[i + 1]
orphan_pages = orphan.logdata.page_data if orphan.logdata else {}
orphan_keys = _page_data_keys(orphan_pages)
if len(orphan_keys) != 1:
i += 1
continue
current_pages = current.logdata.page_data if current.logdata else {}
current_keys = _page_data_keys(current_pages)
if not current_keys:
i += 1
continue
orphan_page_num = orphan_keys[0]
current_last_page = current_keys[-1]
if orphan_page_num != current_last_page + 1:
i += 1
continue
# Check that the orphan is close enough in time
gap = orphan.timestamp - current.stop_timestamp
if gap > SESSION_GAP:
i += 1
continue
# Merge orphan into current
current_pages[str(orphan_page_num)] = orphan_pages[str(orphan_page_num)]
current.log["page_data"] = current_pages
current.log["pages_read"] = len(current_pages)
current.stop_timestamp = orphan.stop_timestamp
current.playback_position_seconds += orphan.playback_position_seconds
affected_books.add(book_id)
if commit:
with transaction.atomic():
current.save(
update_fields=[
"log",
"stop_timestamp",
"playback_position_seconds",
]
)
orphan.delete()
merged += 1
batch_merged += 1
scrobbles.pop(i + 1)
if batch_merged:
self.stdout.write(
f" Book {book_id}: merged {batch_merged} orphan scrobble(s)"
)
self.stdout.write(f"\nTotal orphans merged: {merged}")
if commit and affected_books:
self.stdout.write("Recalculating long_play_stats for affected books...")
for book_id in affected_books:
scrobbles_to_fix = (
Scrobble.objects.filter(book_id=book_id, source="KOReader")
.order_by("timestamp")
)
fix_long_play_stats_for_scrobbles(list(scrobbles_to_fix))
self.stdout.write(f"Fixed stats for {len(affected_books)} books.")
if not commit:
self.stdout.write(
f"\nWould merge {merged} orphan scrobble(s) across "
f"{len(affected_books)} book(s)."
)