We would like to use an Automatic Action that detects if an asset/task is fully translated (all 100% or ICE matches after segmentation) and if so, we would like to configure the workflow to complete that file automatically and avoid to send it for translation. We would also be able to detect untranslated segments.

In WorldServer, we found 3 Automatic steps that addresses this requirement:

1. Test If All Segments Translated
2. Check Translation Status (LIB)
3. QA Checker

What is the difference between the these automatic steps?

1- Test If All Segments Translated will fail if it finds an untranslated segment that has No Status or No translation. The No Translation status can be achieved by a user who sets a segment to Pending Review or Reviewed status without entering any translation. It corresponds to the Draft status in Online Editor. In this case, the step will return Found 1 or more non-translated segments.



If a target segment contains text and a valid segment status, the result will be No non-translated segments found. This applies also to segments in Draft (for instance as a result of an  Autotranslation) or as Rejected. The segment will still be considered as translated just as if the status was Pending Review or Reviewed and the result will be No non-translated segments found.



2- Check Translation Status (LIB) checks only for segment status and not whether the target segment contains translation text. This Automatic Action can be configured to check for segments in status

All Reviewed
All Pending Review or Reviewed
All Pending Review

If it finds one or more segments that are not in the status configured in the AA, the step will have the result Fail with more details about how many segments do not meet the requirement. For example, if the Automatic Action has been configured to check for All Pending Review or Reviewed the check fails, this message will appear (number of segments will obviously vary):

There are segments with a translation status other than Pending Review or Reviewed (3 out of 11)

If it finds all segments in the expected status, it will have the result Pass and the message:

All segments have a translation status of Pending Review or Reviewed

3- One more way to check for segments is to use the QA Checker Automatic Step. This automatic step includes one setting Check for Untranslated segments.




If this setting is enabled, the QA Checker automatic step will pass if it finds text in the target segment, no matter which segment status it has (included No Status). It will fail if it finds an untranslated target segment that has following status:

1- No Status in Browser Workbench or Draft Status in Online Editor
2- No translation
3- The segment has been manually set to any status like Pending Review or Reviewed in Browser Workbench (or through a Return package import) or Confirmed or Translation Approved in Online Editor, but left empty.

The green color in Browser Workbench always means that a user applied a manual change to a segment. In this scenario that will be detected by the QA Checker, she might have added or removed text, leaving the target segment empty. In all 3 cases above, the result in the Task information will be The QA Checker failed and a QA checker log file associated with the Task as task attribute will be created with detailed information about the type of error and which segment(s) has/have been detected. These screenshots display segments that will be detected by the QA Checker with the setting Check for Untranslated segments enabled:





Example from the QA Checker log. Here you can find why the step failed:



Note: The QA Checker step will *not* fail if it finds a target segment that contains no translation, but comes from the TM as 100%, ICE Match or pre-translated Fuzzy Match. So in this case, the result will be The QA Checker passed. Example from a 100% Match untranslated segment coming from the TM that would pass the QA Checker step:



Here are examples of segments in Online Editor that will be detected by the QA Checker Automatic Step:



This is what the QA Checker log will display. As you can see, segment 4 has not been detected because it comes from the TM so it is assumed that it is left empty on purpose:
 

Segments

Segment No.	Fail Type	Segment	Translation	Comments
1	Untranslated	This is a house
2	Untranslated	Another house and a dog
3	Untranslated	The dog
5	Untranslated	Newspaper and an orange

 

This article might be relevant:

How to bypass the QA Checker results if the issue found should not be corrected from a linguistic point of view

WorldServer - What is the difference between QA Check and QA Checker automatic actions?

WorldServer - what is the purpose of the 'Test for Translatable Segments' Automatic Action?