Add date validation

I'm only concerned with validating issue dates here. In DSpace they
are generally always YYYY, YYY-MM, or YYYY-MM-DD (though in theory
they could be any valid ISO8601 format).

This also checks for cases where the date is missing and where the
metadata has specified multiple dates like "1990||1991", as this is
valid, but there is no practical value for it in our system.

csv_metadata_quality/check.py

@@ -73,3 +73,54 @@ def separators(field):
 
 
     return field
+
+
+def date(field):
+    """Check if a date is valid.
+
+    In DSpace the issue date is usually 1990, 1990-01, or 1990-01-01, but it
+    could technically even include time as long as it is ISO8601.
+
+    Also checks for other invalid cases like missing and multiple dates.
+
+    Prints the date if invalid.
+    """
+    from datetime import datetime
+
+    if pd.isna(field):
+        print(f'Missing date.')
+
+        return
+
+    # Try to split multi-value field on "||" separator
+    multiple_dates = field.split('||')
+
+    # We don't allow multi-value date fields
+    if len(multiple_dates) > 1:
+        print(f'Multiple dates not allowed: {field}')
+
+        return field
+
+    try:
+        # Check if date is valid YYYY format
+        datetime.strptime(field, '%Y')
+
+        return field
+    except ValueError:
+        pass
+
+    try:
+        # Check if date is valid YYYY-MM format
+        datetime.strptime(field, '%Y-%m')
+
+        return field
+    except ValueError:
+        pass
+
+    try:
+        # Check if date is valid YYYY-MM-DD format
+        datetime.strptime(field, '%Y-%m-%d')
+
+        return field
+    except ValueError:
+        print(f'Invalid date: {field}')