audiobook edit sqlite database to remove audiobook genre

audiobook_metadata/metadata_edit_database.py

@@ -0,0 +1,90 @@
+import json
+import re
+import sqlite3
+import argparse
+from pathlib import Path
+
+def load_genre(conn, unique=False):
+    """
+    Return all genres from the books table.
+
+    Args:
+        conn: sqlite3 connection
+        unique: if True, return unique genres only
+
+    Returns:
+        list[str]
+    """
+    genres = []
+
+    for (genres_json,) in conn.execute("SELECT genres FROM books"):
+        if not genres_json:
+            continue
+
+        try:
+            genres.extend(json.loads(genres_json))
+        except Exception:
+            pass
+
+    if unique:
+        return sorted(set(genres))
+
+    return genres
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "database",
+        type=Path,
+        help="SQL Database Path",
+    )
+
+    args = parser.parse_args()
+
+    database = args.database.resolve()
+
+    if not database.is_file():
+        parser.error(f"Database is not a file: {database}")
+
+    db = sqlite3.connect(database)
+    cur = db.cursor()
+
+    list_genres = load_genre(cur,True)
+
+    for genre in list_genres:
+        print(genre)
+
+    for book_id, genres_json in cur.execute("SELECT id, genres FROM books"):
+        if not genres_json:
+            continue
+
+        genres = json.loads(genres_json)
+        new_genres = []
+
+        for genre in genres:
+            genre = re.sub(r'(?i)^audio\s*books?$', '', genre)
+            genre = re.sub(r'(?i)^audiobooks?$', '', genre)
+            genre = re.sub(r'(?i)^hörb(uch|ücher)$', '', genre)
+            genre = re.sub(r'(?i)^hörspiele?$', '', genre)
+            genre = re.sub(r'(?i)^spoken\s*word$', '', genre)
+            genre = re.sub(r'(?i)^speech$', '', genre)
+
+            genre = re.sub(r'(?i)^(sci[- ]?fi|sf)$', 'Science Fiction', genre)
+            genre = re.sub(r'(?i)^nonfiction$', 'NonFiction', genre)
+
+            genre = genre.strip()
+
+            if genre:
+                new_genres.append(genre)
+
+        new_genres = sorted(set(new_genres))
+
+        cur.execute(
+            "UPDATE books SET genres=? WHERE id=?",
+            (json.dumps(new_genres), book_id),
+        )
+
+    db.commit()
+
+if __name__ == "__main__":
+    main()