Extraction Verification Step

From Okapi Framework
Jump to navigation Jump to search


This step verifies that the input document can be extracted, merged, then extracted again and produces the same set of events during both extractions.

Takes: Raw document. Sends: Raw document (unmodified).

This verification allows a high level of confidence that the document can be merged safely after translation or other modifications. this step can be used as the first one of a pipeline that make use of newly developed and less tested filters: any error or warning would indicate that there is will be some problem with merging the files back after at the end of the process.


Perform extraction verification — Set this option to use this step. If this option is not set the input document is simply passed to the next step without any verification.

Verify all events (otherwise only text units are verified) — Set this option to verify every events generated by the document. If this option is not set only text unit events are verified. Note that often verifying only text units is a good indicator if the merge was done without error as if it was not the text units are usually different between the two extractions. Disable this option to save processing time.

Compare skeleton — Set this option to ompare the skeleton parts of the verified events. The skeleton is the non-extractable part of the input document.

Maximum number of warnings per document — Enter the maximum number of warnings to allow for a document before moving to the next one.

Interrupt after reaching the maximum number of warnings — Set this option to stop the process when the maxium number of warning has been reached.


  • Support for multilingual documents such as XLIFF, TMX, PO, TTX, etc. is BETA.