Add batch processing for pending OCR

scripts/check_ocr.py

@@ -4,7 +4,7 @@
 Run and validate OCR processing for one movie in the SQLite index.
 
 The script accepts a WCX movie ID, retrieves the movie name and thumbnail URL
-from the database, and then coordinates the complete OCR workflow:
+from the selected database, and coordinates the complete OCR workflow:
 
     1. Run ocr.sh for the thumbnail URL.
     2. Pass the raw OCR text to parse_ocr.py.
@@ -30,21 +30,48 @@ updated.
 
 Usage:
     check_ocr.py <movie-id>
+    check_ocr.py <movie-id> --database /path/to/wcx.db
 
 Example:
     check_ocr.py susana-melo_6707
 """
 
-
+import argparse
 import sqlite3
 import subprocess
 import sys
 from pathlib import Path
 
 
-DATABASE_FILE = Path("/storage/disk1/WCX/database/wcx.db")
-OCR_SCRIPT = Path("/storage/disk1/WCX/scripts/ocr.sh")
-PARSER_SCRIPT = Path("/storage/disk1/WCX/scripts/parse_ocr.py")
+DEFAULT_DATABASE_FILE = Path(
+    "/storage/disk1/WCX/database/wcx.db"
+)
+OCR_SCRIPT = Path(
+    "/storage/disk1/WCX/scripts/ocr.sh"
+)
+PARSER_SCRIPT = Path(
+    "/storage/disk1/WCX/scripts/parse_ocr.py"
+)
+
+
+def parse_arguments() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        description="Run OCR processing for one WCX movie."
+    )
+
+    parser.add_argument(
+        "movie_id",
+        help="Movie ID to process.",
+    )
+
+    parser.add_argument(
+        "--database",
+        type=Path,
+        default=DEFAULT_DATABASE_FILE,
+        help=f"SQLite database; default: {DEFAULT_DATABASE_FILE}",
+    )
+
+    return parser.parse_args()
 
 
 def parse_key_value_output(output: str) -> dict[str, str]:
@@ -147,13 +174,26 @@ def set_ocr_completed(
 
 
 def main() -> None:
-    if len(sys.argv) != 2:
-        print(f"Användning: {sys.argv[0]} <movie-id>", file=sys.stderr)
-        sys.exit(1)
+    args = parse_arguments()
+    movie_id = args.movie_id
+    database_file = args.database
 
-    movie_id = sys.argv[1]
+    if not database_file.is_file():
+        raise FileNotFoundError(
+            f"Database file does not exist: {database_file}"
+        )
 
-    with sqlite3.connect(DATABASE_FILE) as connection:
+    if not OCR_SCRIPT.is_file():
+        raise FileNotFoundError(
+            f"OCR script does not exist: {OCR_SCRIPT}"
+        )
+
+    if not PARSER_SCRIPT.is_file():
+        raise FileNotFoundError(
+            f"OCR parser does not exist: {PARSER_SCRIPT}"
+        )
+
+    with sqlite3.connect(database_file) as connection:
         row = connection.execute(
             """
             SELECT name, thumbnail
@@ -165,7 +205,7 @@ def main() -> None:
 
         if row is None:
             print(
-                f"Filmen finns inte i databasen: {movie_id}",
+                f"Movie does not exist in database: {movie_id}",
                 file=sys.stderr,
             )
             sys.exit(1)
@@ -173,7 +213,7 @@ def main() -> None:
         database_name, thumbnail = row
 
         if not thumbnail:
-            error_message = "Filmen saknar thumbnail."
+            error_message = "Movie has no thumbnail."
 
             set_ocr_failed(
                 connection,
@@ -193,7 +233,7 @@ def main() -> None:
         if ocr_result.returncode != 0:
             error_message = (
                 ocr_result.stderr.strip()
-                or "OCR-körningen misslyckades."
+                or "OCR execution failed."
             )
 
             set_ocr_failed(
@@ -203,7 +243,7 @@ def main() -> None:
                 ocr_result.stdout.strip() or None,
             )
 
-            print("OCR-körningen misslyckades:", file=sys.stderr)
+            print("OCR execution failed:", file=sys.stderr)
             print(error_message, file=sys.stderr)
             sys.exit(1)
 
@@ -219,7 +259,7 @@ def main() -> None:
         if parser_result.returncode != 0:
             error_message = (
                 parser_result.stderr.strip()
-                or "Tolkningen av OCR-resultatet misslyckades."
+                or "OCR output parsing failed."
             )
 
             set_manual_review(
@@ -230,7 +270,7 @@ def main() -> None:
             )
 
             print(
-                "Tolkningen av OCR-resultatet misslyckades:",
+                "OCR output parsing failed:",
                 file=sys.stderr,
             )
             print(error_message, file=sys.stderr)
@@ -253,7 +293,7 @@ def main() -> None:
 
         if missing_fields:
             error_message = (
-                "Parsern saknar fält: "
+                "Parser output is missing fields: "
                 + ", ".join(missing_fields)
             )
 
@@ -275,7 +315,7 @@ def main() -> None:
 
         if not names_match:
             error_message = (
-                "OCR-namnet matchar inte databasnamnet: "
+                "OCR name does not match database name: "
                 f"{ocr_name!r} != {database_name!r}"
             )
 
@@ -299,6 +339,7 @@ def main() -> None:
         )
 
     print(f"id={movie_id}")
+    print(f"database={database_file}")
     print(f"database_name={database_name}")
     print(f"ocr_name={ocr_name}")
     print("name_match=yes")
@@ -309,5 +350,11 @@ def main() -> None:
 
 
 if __name__ == "__main__":
-    main()
-
+    try:
+        main()
+    except (
+        FileNotFoundError,
+        sqlite3.Error,
+    ) as error:
+        print(f"Error: {error}", file=sys.stderr)
+        sys.exit(1)