add scripts from repository python-scripts

2025-04-01 08:12:53 +02:00
parent 38e77b25b1
commit 1b18dae311
8 changed files with 599 additions and 0 deletions
@@ -0,0 +1,73 @@
+"""
+download files with URLs from DB
+"""
+import re
+import subprocess
+from argparse import ArgumentParser, ArgumentDefaultsHelpFormatter
+import mariadb
+from setup import get_database_cursors, create_tables, get_logger
+
+parser = ArgumentParser(formatter_class=ArgumentDefaultsHelpFormatter)
+parser.add_argument('--verbose', '-v', action='count', default=0)
+parser.add_argument('--dir', '-d', default='/data/media')
+parser.add_argument('--dry-run', '-m', action='store_true')
+parser.add_argument('--rename', '-r', action='store_true')
+args = parser.parse_args()
+
+
+def parse_output(lines_list, log):
+    file_name = ""
+    for line in lines_list:
+        if 'has already been downloaded' in line:
+            end_len = len(' has already been downloaded')
+            file_name = line[11:-end_len]
+            log.info('found file: "%s"', file_name)
+        if 'Destination' in line:
+            line_len = len(line)
+            start_len = len('[download] Destination: ')
+            file_len = line_len-start_len
+            file_name = line[-file_len:]
+            log.info('new file: "%s"', file_name)
+    return file_name
+
+
+def download_url(video_url, log):
+    result = subprocess.run(["/home/tpeetz/bin/yt-dlp", video_url], cwd=args.dir, capture_output=True, text=True)
+    if result.returncode == 0:
+        output = result.stdout
+        output = re.sub(' +', ' ', output)
+        lines_list = output.splitlines()
+        return parse_output(lines_list, log)
+    else:
+        return None
+
+
+def download_and_update(link, entry_id, conn):
+    m_cursor = conn.cursor()
+    filename = download_url(link, logger)
+    if filename is None:
+        update_statement = 'UPDATE media_file set review = true WHERE id = ?'
+        logger.debug(f'entry {entry_id} could not downloaded, set to Review')
+        m_cursor.execute(update_statement, (entry_id,))
+    else:
+        update_statement = 'UPDATE media_file set file_name = ?, should_download = false, review = false WHERE id = ?'
+        logger.debug(f'entry {entry_id} successfully downloaded, set review and should_download to false')
+        m_cursor.execute(update_statement, (filename, entry_id))
+    conn.commit()
+
+
+if __name__ == '__main__':
+    logger = get_logger(args.verbose)
+    logger.info('kontor.download started')
+    s_conn, m_conn = get_database_cursors(logger)
+    cursor = m_conn.cursor()
+    cursor.execute('SELECT id, url FROM media_file where should_download is true')
+    for (link_id, url) in cursor.fetchall():
+        if url is None:
+            logger.info('There is no url for id {}'.format(link_id))
+        else:
+            if args.dry_run:
+                logger.info(f'download {url} for {link_id}')
+            else:
+                download_and_update(url, link_id, m_conn)
+    logger.info('kontor.download finished')