danvics/soulsync
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

Improve edition detection and completion logic

Browse source 
Prefer an album's stored track count when owned_tracks meets or exceeds it (avoids marking a complete standard edition as incomplete when an external source reports a larger deluxe count). Add more robust edition-detection regexes and generic catch-alls (parenthesized/bracketed text containing "edition") in music_database.py and web_server.py, and include silver/gold/platinum edition variants. Also tidy related regex handling and comments to improve matching for varied edition naming (e.g. "MMXI Special Edition").
This commit is contained in:
Broque Thomas 2026-02-24 12:05:12 -08:00
parent a8e84fa20f
commit fb7b373d71
2 changed files with 42 additions and 15 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

33
database/music_database.py
View file

@ -2561,8 +2561,18 @@ class MusicDatabase:
result = cursor.fetchone()
stored_track_count = result[0] if result and result[0] else 0
# Use provided expected count if available, otherwise use stored count
expected_tracks = expected_track_count if expected_track_count is not None else stored_track_count
# Use provided expected count if available, otherwise use stored count.
# However, if the album is complete by its own stored metadata, prefer the stored
# count so edition differences don't make a complete album appear incomplete.
# e.g. user has standard edition (12 tracks, all present) but Spotify returns
# deluxe edition count (20) — should show as complete, not 12/20.
if expected_track_count is not None and stored_track_count > 0 and owned_tracks >= stored_track_count:
# Album is complete by its own metadata — don't inflate expected with a different edition's count
expected_tracks = stored_track_count
elif expected_track_count is not None:
expected_tracks = expected_track_count
else:
expected_tracks = stored_track_count
# Determine completeness with refined thresholds
if expected_tracks and expected_tracks > 0:
@ -2736,6 +2746,7 @@ class MusicDatabase:
title_lower = title.lower().strip()
# Define edition patterns and their variations
# Specific patterns first, generic catch-alls last (first match wins due to break)
edition_patterns = {
r'\s*\(deluxe\s*edition?\)': ['deluxe', 'deluxe edition'],
r'\s*\(expanded\s*edition?\)': ['expanded', 'expanded edition'],
@ -2749,6 +2760,9 @@ class MusicDatabase:
r'\s+special\s*edition?$': ['special', 'special edition'],
r'\s*-\s*deluxe': ['deluxe'],
r'\s*-\s*platinum\s*edition?': ['platinum', 'platinum edition'],
# Generic catch-alls for any edition in parens/brackets (e.g. Silver Edition, MMXI Special Edition)
r'\s*\([^)]*\bedition\b[^)]*\)': ['edition'],
r'\s*\[[^\]]*\bedition\b[^\]]*\]': ['edition'],
}
# Check if title contains any edition indicators
@ -3068,11 +3082,11 @@ class MusicDatabase:
def _clean_album_title_for_comparison(self, title: str) -> str:
"""Clean album title by removing edition markers for comparison"""
cleaned = title.lower()
# Remove common edition patterns
# Remove common edition patterns (specific first, then generic catch-alls)
patterns = [
r'\s*\(deluxe\s*edition?\)',
r'\s*\(expanded\s*edition?\)',
r'\s*\(expanded\s*edition?\)',
r'\s*\(platinum\s*edition?\)',
r'\s*\(special\s*edition?\)',
r'\s*\(remastered?\)',
@ -3082,11 +3096,16 @@ class MusicDatabase:
r'\s*-\s*platinum\s*edition?',
r'\s+deluxe\s*edition?$',
r'\s+platinum\s*edition?$',
# Generic catch-alls: any parenthesized/bracketed text containing "edition"
# Handles "Silver Edition", "MMXI Special Edition", "Limited Edition", etc.
r'\s*\([^)]*\bedition\b[^)]*\)',
r'\s*\[[^\]]*\bedition\b[^\]]*\]',
r'\s*-\s+\w+\s+edition\s*$',
]
for pattern in patterns:
cleaned = re.sub(pattern, '', cleaned, flags=re.IGNORECASE)
return cleaned.strip()
def get_album_completion_stats(self, artist_name: str) -> Dict[str, int]:

24
web_server.py
View file

@ -7818,18 +7818,23 @@ def _get_base_album_name(album_name: str) -> str:
"""
Extract the base album name without edition indicators.
E.g., 'good kid, m.A.A.d city (Deluxe Edition)' -> 'good kid, m.A.A.d city'
'Battle Hymns (MMXI Special Edition)' -> 'Battle Hymns'
"""
import re
# Remove common edition suffixes
base_name = album_name
# Remove edition indicators in parentheses or brackets
base_name = re.sub(r'\s*[\[\(](deluxe|special|expanded|extended|bonus|remastered|anniversary|collectors?|limited).*?[\]\)]\s*$', '', base_name, flags=re.IGNORECASE)
# Allow any prefix before the keyword (e.g. "MMXI Special Edition", "20th Anniversary Edition")
base_name = re.sub(r'\s*[\[\(][^)\]]*\b(deluxe|special|expanded|extended|bonus|remaster(?:ed)?|anniversary|collectors?|limited|silver|gold|platinum)\b[^)\]]*[\]\)]\s*$', '', base_name, flags=re.IGNORECASE)
# Generic: any parenthesized/bracketed text ending with "edition"
base_name = re.sub(r'\s*[\[\(][^)\]]*\bedition\b[^)\]]*[\]\)]\s*$', '', base_name, flags=re.IGNORECASE)
# Remove standalone edition words at the end
base_name = re.sub(r'\s+(deluxe|special|expanded|extended|bonus|remastered|anniversary|collectors?|limited)\s*(edition)?\s*$', '', base_name, flags=re.IGNORECASE)
base_name = re.sub(r'\s+(deluxe|special|expanded|extended|bonus|remastered|anniversary|collectors?|limited|silver|gold|platinum)\s*(edition)?\s*$', '', base_name, flags=re.IGNORECASE)
return base_name.strip()
def _detect_deluxe_edition(album_name: str) -> bool:
@ -7845,7 +7850,7 @@ def _detect_deluxe_edition(album_name: str) -> bool:
# Check for deluxe indicators
deluxe_indicators = [
'deluxe',
'deluxe edition',
'deluxe edition',
'special edition',
'expanded edition',
'extended edition',
@ -7853,7 +7858,10 @@ def _detect_deluxe_edition(album_name: str) -> bool:
'remastered',
'anniversary',
'collectors edition',
'limited edition'
'limited edition',
'silver edition',
'gold edition',
'platinum edition',
]
for indicator in deluxe_indicators: