add wrapper functions to check for the validity of the database

corrlib/integrity.py

@@ -1,5 +1,32 @@
 import datetime as dt
+from pathlib import Path
+from .tools import get_db_file
+import pandas as pd
+import sqlite3
 
 
-def check_time_validity(created_at: dt.datetime, updated_at: dt.datetime) -> bool:
+def has_valid_times(result: pd.DataFrame) -> bool:
-    return not (created_at > updated_at)
+    # we expect created_at <= updated_at <= now
+    created_at = dt.datetime.fromisoformat(result['created_at'])
+    updated_at = dt.datetime.fromisoformat(result['updated_at'])
+    if created_at > updated_at:
+        return False
+    if updated_at > dt.datetime.now():
+        return False
+    return True
+
+
+def check_db_integrity(path: Path) -> None:
+    db = get_db_file(path)
+    search_expr = "SELECT * FROM 'backlogs'"
+    conn = sqlite3.connect(db)
+    results = pd.read_sql(search_expr, conn)
+
+    for result in results:
+        if not has_valid_times(result):
+            raise ValueError(f"Result with id {result[id]} has wrong time signatures.")
+
+
+def full_integrity_check(path: Path) -> None:
+    check_db_integrity(path)
+