A record is read from the input-source-file and the data is checked. If the record can be imported, it is appended to the file XXX_imported.csv, otherwise it is inserted into the file XXX_pending.csv, if no record with the exact same date already exists in the XXX_pending.csv. The check is done by comparing the md5-digests of the new record with the existing records.
When doing the import in pending-only-mode the XXX_pending.csv is taken as input-source-file and if a record can now be imported it is appended to the XXX_imported.csv file and removed from XXX_pending.csv.
Caveat:
if an record is inserted into XXX_pending, it will only be removed by importing in pending-only-mode. If a new input-file contains a record for the same object, with different data, that will not be detected at the moment.
The XXX_imported.csv and XXX_pending file has these columns added:
| imported_by: | the member who did the import |
|---|---|
| import_mode: | edit, create, delete |
| import_date: | the date and time of the import |
| imported_from: | the filename of the input-source-file |
| error: | the error message (empty in XXX_imported) |