Laboratoire ICAR » Plateforme TXM

SLH

I don't think that the solution is to move some work at another moment (before the import).

The solution first consists in giving a precise and exhaustive diagnostic messages during the import process:

• if a text file name is not found in 'metadata.csv', the following warning message should be given on one line "Warning: impossible to find a metadata line for the text of name 'X Y' (filename is 'X Y.xml')." [the quotes '...' in the message should help to diagnoze number of spaces problems]
• if the fact that a metadata value is empty is a problem or should be diagnozed, this should be a processing option. For example: EmptyMetadataCellProcessing with values 'Ignore' (take appropriate action so that TXM is not broken but the corpus imported), 'Warning' (print warning message + take appropriate action so that TXM is not broken but the corpus imported - default value) or 'Error' (print error message - metadata name, text name, metadata.csv line number + stop the import).
• if a metadata.csv line is diagnozed as unreadable and this could be considered as an error, this should be a processing option. For example: UnreadableMetadataLineProcessing with values 'Warning' (print warning message + take appropriate action so that TXM is not broken but the corpus imported - default value) or 'Error' (print error message - text name, metadata.csv line number + stop the import).

If we want later to design an independant UI from the import process to help diagnoze metadata.csv files, we should relate that to a metadata editor component (including table import and table edition capabilities) or a properties editor component (with the distinction text properties (metadata) / structures properties / lexical properties).