Safe handling of invalid timestamps

[Anjum48]

In many Microsoft apps, the equivalent of a null timestamp is 1601/01/01 00:00:00. These can pop up sometimes if a file has been recovered or there was some other issue during saving.

This causes an OSError when trying to convert to a Unix timestamp: datetime(1601, 1, 1, 0, 0, 0).timestamp(),

Suggested fix:

In https://github.com/Unstructured-IO/unstructured-ingest/blob/main/unstructured_ingest/utils/string_and_date_utils.py add:

def safe_timestamp(dt: datetime) -> Optional[str]:
    """
    Converts a datetime object to a string representation of its timestamp.
    Handles potential exceptions that may arise during conversion.
    """
    try:
        return str(dt.timestamp())
    except (OSError, ValueError, OverflowError):
        return None

In the OneDrive ingester:

unstructured-ingest/unstructured_ingest/processes/connectors/onedrive.py

Lines 208 to 225 in 1234157

	return FileData(
	identifier=drive_item.id,
	connector_type=self.connector_type,
	source_identifiers=SourceIdentifiers(
	fullpath=server_path, filename=drive_item.name, rel_path=rel_path
	),
	metadata=FileDataSourceMetadata(
	url=drive_item.parent_reference.path + "/" + drive_item.name,
	version=drive_item.etag,
	date_modified=str(date_modified_dt.timestamp()) if date_modified_dt else None,
	date_created=str(date_created_at.timestamp()) if date_created_at else None,
	date_processed=str(time()),
	record_locator={
	"user_pname": self.connection_config.user_pname,
	"server_relative_path": server_path,
	},
	),
	additional_metadata=self.get_properties_sync(drive_item=drive_item),

use safe_timestamp()

        return FileData(
            identifier=drive_item.id,
            connector_type=self.connector_type,
            source_identifiers=SourceIdentifiers(
                fullpath=server_path, filename=drive_item.name, rel_path=rel_path
            ),
            metadata=FileDataSourceMetadata(
                url=drive_item.parent_reference.path + "/" + drive_item.name,
                version=drive_item.etag,
                date_modified=safe_timestamp(date_modified_dt),
                date_created=safe_timestamp(date_created_at),
                date_processed=str(time()),
                record_locator={
                    "user_pname": self.connection_config.user_pname,
                    "server_relative_path": server_path,
                },
            ),
            additional_metadata=self.get_properties_sync(drive_item=drive_item),
        )

Happy to open a PR if needed