Basic segmentation and pseudo-translation

[Yves Savourel <ysavourel@enlaso.com>]

Hi everyone,

 

In case it can help you test your implementation. I've added two function to Lynx.

 

·       Segmentation: if you rewrite a document (command -rw) with the option -seg, the tool will re-segment each existing segment (when allowed) applying a very basic rule: break after a period followed by a whitespace. Entries with existing target will not be re-segmented.

 

·       Pseudo-translation: if you rewrite a document (command -rw) with the option -pseudo, the tool will add some simple pseudo-translation based on the source (digits are replaced by 'N' and letters by 'Z'). Existing translations will be overwritten.

 

Both option can be use together. The segmentation is always done before the pseudo-translation. For example:

 

lynx -rw -seg -pseudo example.xlf

 

will give the example.out.xlf attached.

 

One note: Currently the pseudo-translation is not compliant because it implements inheritance of the translate state on a segment from the siblings then the parent, rather than the parent alone. You can see the effect of that in the output file attached. See also my email on that issue.

 

The new options of Lynx are in the latest snapshot:

http://okapi.opentag.com/snapshots/okapi-xliffLib_all-platforms_0.21-SNAPSHOT.zip

 

Cheers,

-yves

 

Attachment: example.out.xlf
Description: Binary data

Attachment: example.xlf
Description: Binary data