refactor: use sets to avoid costly loops

This commit is contained in:
Felipe Marinho 2024-11-14 19:41:50 -03:00 committed by Zoe Roux
parent 8f4aecb236
commit 6992de5e2f
No known key found for this signature in database

View File

@ -11,12 +11,11 @@ logger = getLogger(__name__)
def get_ignore_pattern():
"""Compile ignore pattern from environment variable."""
try:
pattern = os.environ.get("LIBRARY_IGNORE_PATTERN")
if pattern:
return re.compile(pattern)
return None
except Exception as e:
return re.compile(pattern) if pattern else None
except re.error as e:
logger.error(f"Invalid ignore pattern. Ignoring. Error: {e}")
return None
@ -25,11 +24,14 @@ async def scan(
path_: Optional[str], publisher: Publisher, client: KyooClient, remove_deleted=False
):
path = path_ or os.environ.get("SCANNER_LIBRARY_ROOT", "/video")
logger.info("Starting the scan. It can take some time...")
ignore_pattern = get_ignore_pattern()
logger.info("Starting scan at %s. This may take some time...", path)
registered = await client.get_registered_paths()
videos = []
ignore_pattern = get_ignore_pattern()
if ignore_pattern:
logger.info(f"Applying ignore pattern: {ignore_pattern}")
registered = set(await client.get_registered_paths())
videos = set()
for dirpath, dirnames, files in os.walk(path):
# Skip directories with a `.ignore` file
@ -42,23 +44,26 @@ async def scan(
# Apply ignore pattern, if any
if ignore_pattern and ignore_pattern.match(file_path):
continue
videos.append(file_path)
videos.add(file_path)
to_register = [p for p in videos if p not in registered]
to_register = videos - registered
to_delete = registered - videos if remove_deleted else set()
if to_register:
logger.info("Found %d new files to register.", len(to_register))
await asyncio.gather(*[publisher.add(path) for path in to_register])
if to_delete:
logger.info("Removing %d stale files.", len(to_delete))
await asyncio.gather(*[publisher.delete(path) for path in to_delete])
if remove_deleted:
deleted = [x for x in registered if x not in videos]
logger.info("Found %d stale files to remove.", len(deleted))
if len(deleted) != len(registered):
await asyncio.gather(*map(publisher.delete, deleted))
elif len(deleted) > 0:
logger.warning("All video files are unavailable. Check your disks.")
issues = set(await client.get_issues())
issues_to_delete = issues - videos
if issues_to_delete:
logger.info("Removing %d stale issues.", len(issues_to_delete))
await asyncio.gather(
*[client.delete_issue(issue) for issue in issues_to_delete]
)
issues = await client.get_issues()
for x in issues:
if x not in videos:
await client.delete_issue(x)
logger.info("Found %d new files (counting non-video files)", len(to_register))
await asyncio.gather(*map(publisher.add, to_register))
logger.info("Scan finished for %s.", path)