Add podgrab featureset
This commit is contained in:
Cody Cook
parent
095bf52a2f
commit
233dd5b5c0
33 changed files with 2315 additions and 125 deletions
|
|
@ -173,6 +173,8 @@ def format_filename(format_string, podcast, episode):
|
|||
# If episode_number exists but is not a digit, format as S01E{episode_number}
|
||||
else f"S{episode.season or 1:02d}E{episode.episode_number}"
|
||||
if episode.episode_number
|
||||
# If neither season nor episode_number are available, use published date
|
||||
else episode.published_date.strftime('%Y-%m-%d') if episode.published_date
|
||||
# Otherwise, return empty string
|
||||
else ''
|
||||
),
|
||||
|
|
@ -195,10 +197,23 @@ def format_filename(format_string, podcast, episode):
|
|||
|
||||
# Handle empty path segments by removing them
|
||||
path_parts = formatted_path.split(os.path.sep)
|
||||
path_parts = [part for part in path_parts if part.strip()]
|
||||
|
||||
# Remove empty segments and segments that would be just placeholders without values
|
||||
cleaned_parts = []
|
||||
for part in path_parts:
|
||||
part = part.strip()
|
||||
if not part:
|
||||
continue
|
||||
# Check for common placeholders without values
|
||||
if part in ["Season ", "Season", "Episode ", "Episode", "E", "S"]:
|
||||
continue
|
||||
# Check for patterns like "S01E" without an episode number
|
||||
if part.startswith("S") and part.endswith("E") and len(part) > 2:
|
||||
continue
|
||||
cleaned_parts.append(part)
|
||||
|
||||
# Rejoin the path with proper separators
|
||||
return os.path.sep.join(path_parts)
|
||||
return os.path.sep.join(cleaned_parts)
|
||||
|
||||
def sanitize_filename(filename):
|
||||
"""
|
||||
|
|
@ -277,6 +292,7 @@ def delete_old_episodes(days=30):
|
|||
def verify_downloaded_episodes(podcast_id=None, progress_callback=None):
|
||||
"""
|
||||
Verify that downloaded episodes still exist on disk and update their status.
|
||||
Also checks for existing files for episodes that aren't marked as downloaded.
|
||||
|
||||
Args:
|
||||
podcast_id (int, optional): ID of the podcast to check. If None, check all podcasts.
|
||||
|
|
@ -286,23 +302,24 @@ def verify_downloaded_episodes(podcast_id=None, progress_callback=None):
|
|||
dict: Statistics about the verification process.
|
||||
"""
|
||||
from app.models.podcast import Episode, Podcast
|
||||
from app.models.settings import Settings
|
||||
|
||||
# Get episodes to check
|
||||
# First, verify episodes that are marked as downloaded
|
||||
query = Episode.query.filter(Episode.downloaded == True)
|
||||
if podcast_id:
|
||||
query = query.filter(Episode.podcast_id == podcast_id)
|
||||
|
||||
episodes = query.all()
|
||||
total = len(episodes)
|
||||
downloaded_episodes = query.all()
|
||||
total_downloaded = len(downloaded_episodes)
|
||||
|
||||
if progress_callback:
|
||||
progress_callback(0, f"Verifying {total} downloaded episodes")
|
||||
progress_callback(0, f"Verifying {total_downloaded} downloaded episodes")
|
||||
|
||||
missing = 0
|
||||
for i, episode in enumerate(episodes):
|
||||
if progress_callback and total > 0:
|
||||
progress = int((i / total) * 100)
|
||||
progress_callback(progress, f"Verifying episode {i+1}/{total}")
|
||||
for i, episode in enumerate(downloaded_episodes):
|
||||
if progress_callback and total_downloaded > 0:
|
||||
progress = int((i / total_downloaded) * 50) # Use first half of progress for verification
|
||||
progress_callback(progress, f"Verifying episode {i+1}/{total_downloaded}")
|
||||
|
||||
if not episode.file_path or not os.path.exists(episode.file_path):
|
||||
episode.downloaded = False
|
||||
|
|
@ -312,15 +329,133 @@ def verify_downloaded_episodes(podcast_id=None, progress_callback=None):
|
|||
|
||||
db.session.commit()
|
||||
|
||||
if progress_callback:
|
||||
progress_callback(100, f"Verification complete. {missing} episodes marked as not downloaded.")
|
||||
# Now check for existing files for episodes that aren't marked as downloaded
|
||||
query = Episode.query.filter(Episode.downloaded == False)
|
||||
if podcast_id:
|
||||
query = query.filter(Episode.podcast_id == podcast_id)
|
||||
|
||||
logger.info(f"Verified {total} episodes. {missing} were missing.")
|
||||
undownloaded_episodes = query.all()
|
||||
total_undownloaded = len(undownloaded_episodes)
|
||||
|
||||
if progress_callback:
|
||||
progress_callback(50, f"Checking for existing files for {total_undownloaded} undownloaded episodes")
|
||||
|
||||
found = 0
|
||||
if total_undownloaded > 0 and podcast_id:
|
||||
# Get the podcast
|
||||
podcast = Podcast.query.get(podcast_id)
|
||||
if not podcast:
|
||||
logger.error(f"Podcast with ID {podcast_id} not found")
|
||||
return {
|
||||
'total_checked': total_downloaded,
|
||||
'missing': missing,
|
||||
'found': 0
|
||||
}
|
||||
|
||||
# Get settings
|
||||
settings = Settings.query.first()
|
||||
if not settings:
|
||||
settings = Settings(
|
||||
download_path=current_app.config['DOWNLOAD_PATH'],
|
||||
naming_format="{podcast_title}/{episode_title}"
|
||||
)
|
||||
db.session.add(settings)
|
||||
db.session.commit()
|
||||
|
||||
# Use podcast's naming format if available, otherwise use global settings
|
||||
naming_format = podcast.naming_format or settings.naming_format
|
||||
download_path = settings.download_path
|
||||
|
||||
# Check each undownloaded episode for existing files
|
||||
for i, episode in enumerate(undownloaded_episodes):
|
||||
if progress_callback:
|
||||
progress = 50 + int((i / total_undownloaded) * 50) # Use second half of progress for file matching
|
||||
progress_callback(progress, f"Checking for file for episode {i+1}/{total_undownloaded}")
|
||||
|
||||
try:
|
||||
# Format filename using the naming format
|
||||
filename = format_filename(naming_format, podcast, episode)
|
||||
|
||||
# Check for common audio file extensions
|
||||
extensions = ['.mp3', '.m4a', '.ogg', '.wav']
|
||||
for ext in extensions:
|
||||
file_path = os.path.normpath(os.path.join(download_path, filename + ext))
|
||||
if os.path.exists(file_path):
|
||||
logger.info(f"Found existing file for episode: {file_path}")
|
||||
episode.downloaded = True
|
||||
episode.file_path = file_path
|
||||
found += 1
|
||||
break
|
||||
except Exception as e:
|
||||
logger.error(f"Error checking for existing file for episode {episode.title}: {str(e)}")
|
||||
|
||||
db.session.commit()
|
||||
|
||||
if progress_callback:
|
||||
progress_callback(100, f"Verification complete. {missing} episodes marked as not downloaded, {found} files matched.")
|
||||
|
||||
logger.info(f"Verified {total_downloaded} episodes. {missing} were missing. Found files for {found} undownloaded episodes.")
|
||||
return {
|
||||
'total_checked': total,
|
||||
'missing': missing
|
||||
'total_checked': total_downloaded,
|
||||
'missing': missing,
|
||||
'found': found
|
||||
}
|
||||
|
||||
def download_all_episodes(podcast_id, progress_callback=None):
|
||||
"""
|
||||
Download all episodes of a podcast that haven't been downloaded yet.
|
||||
|
||||
Args:
|
||||
podcast_id: ID of the Podcast to download all episodes for.
|
||||
progress_callback (callable, optional): Callback function for progress updates.
|
||||
|
||||
Returns:
|
||||
dict: Statistics about the download process.
|
||||
"""
|
||||
from app.models.podcast import Podcast, Episode
|
||||
|
||||
if progress_callback:
|
||||
progress_callback(2, "Loading podcast data")
|
||||
|
||||
# Load the podcast
|
||||
podcast = Podcast.query.get(podcast_id)
|
||||
if not podcast:
|
||||
raise ValueError(f"Podcast with ID {podcast_id} not found")
|
||||
|
||||
# Get all episodes that haven't been downloaded yet
|
||||
episodes = Episode.query.filter_by(podcast_id=podcast_id, downloaded=False).all()
|
||||
total_episodes = len(episodes)
|
||||
|
||||
if progress_callback:
|
||||
progress_callback(5, f"Found {total_episodes} episodes to download")
|
||||
|
||||
if total_episodes == 0:
|
||||
if progress_callback:
|
||||
progress_callback(100, "No episodes to download")
|
||||
return {"total": 0, "downloaded": 0, "failed": 0}
|
||||
|
||||
stats = {"total": total_episodes, "downloaded": 0, "failed": 0}
|
||||
|
||||
# Download each episode
|
||||
for i, episode in enumerate(episodes):
|
||||
if progress_callback:
|
||||
progress = 5 + int((i / total_episodes) * 90) # Scale from 5% to 95%
|
||||
progress_callback(progress, f"Downloading episode {i+1}/{total_episodes}: {episode.title}")
|
||||
|
||||
try:
|
||||
download_episode(episode.id)
|
||||
stats["downloaded"] += 1
|
||||
logger.info(f"Downloaded episode {i+1}/{total_episodes}: {episode.title}")
|
||||
except Exception as e:
|
||||
stats["failed"] += 1
|
||||
logger.error(f"Error downloading episode {episode.title}: {str(e)}")
|
||||
|
||||
if progress_callback:
|
||||
progress_callback(100, f"Download complete. Downloaded {stats['downloaded']} episodes, {stats['failed']} failed.")
|
||||
|
||||
logger.info(f"Podcast archive download completed: {stats}")
|
||||
return stats
|
||||
|
||||
def rename_episode(episode_id, new_format=None, progress_callback=None):
|
||||
"""
|
||||
Rename a downloaded episode file using a new format.
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue