extend check_kontor.py for comparing json file with db

scripts/check_kontor.py

@@ -2,6 +2,7 @@
 Checks the database kontor
 """
 from enum import Enum, auto
+import json
 import mariadb
 import requests
 from pathlib import Path
@@ -12,7 +13,8 @@ from config import get_logger, get_database_cursors
 parser = ArgumentParser(formatter_class=ArgumentDefaultsHelpFormatter)
 parser.add_argument('--verbose', '-v', action='count', default=0)
 parser.add_argument('--config', '-c', default='kontor')
-parser.add_argument('--dir', '-d', default='/data/media')
+parser.add_argument('--file', '-f')
+parser.add_argument('--dir', '-d')
 parser.add_argument('--dry-run', '-m', action='store_true')
 parser.add_argument('--reset-cloud-link', '-r', action='store_true')
 args = parser.parse_args()
@@ -128,13 +130,48 @@ def reset_cloud_link(conn, dry_run, log):
         cursor.execute('UPDATE media_file SET cloud_link="" WHERE id is NOT NULL')
         conn.commit()
 
+def check_file_with_db(data_file: Path, m_conn, log):
+    log.info(f"read json file: {data_file}")
+    cursor = m_conn.cursor()
+    with open(data_file, 'r') as json_file:
+        json_load = json.load(json_file)
+        for table in json_load:
+            log.info(f"{table}: {len(json_load[table])}")
+            items = json_load[table]
+            for item in items:
+                item_id = item['id']
+                select_statement = f"SELECT * FROM {table} WHERE id='{item_id}'"
+                cursor.execute(select_statement)
+                rows = cursor.fetchall()
+                count = len(rows)
+                log.info(f"{count} entries found for {item_id}")
+                if count == 0:
+                    log.info(f"entry for {item_id} not found")
+                if count == 1:
+                    log.info(f"check entry {item_id}")
+                    #log.info(f"entry {rows[0]}")
+                    columns = []
+                    values = []
+                    for (key, value) in item.items():
+                        columns.append(key)
+                        values.append(value)
+                    for index, _ in enumerate(columns):
+                        log.info(f"compare {values[index]} with {rows[0][index]}")
+
+
 
 if __name__ == '__main__':
     log = get_logger(args.verbose, args.config)
     log.info("kontor.check_kontor started")
     _, m_conn = get_database_cursors(log, args.config)
+    if args.dir:
         log.info("kontor.check_kontor.rename_files_to_id")
         rename_files_to_id(args.dir, args.dry_run, m_conn, log)
+    if args.file:
+        data_file = Path(args.file)
+        if data_file.exists():
+            log.info("kontor.check_kontor.check_file_with_db")
+            check_file_with_db(data_file, m_conn, log)
     #logger.info("kontor.check_kontor.update_cloud_link_with_found_files")
     #update_cloud_link_with_found_files(data_dir, mariadb_conn, args.dry_run)
     #logger.info("kontor.check_kontor.get_ids_from_column_cloud_link")

scripts/json_to_mariadb.py

@@ -4,7 +4,7 @@ copy data from SQLite to MariaDB
 from argparse import ArgumentParser, ArgumentDefaultsHelpFormatter
 from pathlib import Path
 from typing import Dict
-from config import get_logger, get_database_cursors, get_meta_data, get_scripts
+from config import get_logger, get_database_cursors
 import mariadb
 import json
 
@@ -40,7 +40,7 @@ def copy_data(mariadb_conn, data_file: Path, log):
                     columns.append(key)
                     values.append(value)
                 row = tuple(values)
-                #log.info(f"values: {row}")
+                log.info(f"values: {row}")
                 insert_statement = 'INSERT INTO {}({}) VALUES({})'.format(table, ', '.join(columns), ', '.join(['?']*len(columns)))
                 #log.info(f"statement: {insert_statement}")
                 mariadb_cursor.execute(insert_statement, row)