Start a new topic

Filter > Segments with numbers

I wonder if we could have this option.

I find it useful when doing manual QA to check correct the configuration of numbers from USA type in source (decimal format and thousands separator) to UE type in target. For the moment, part of this manual check is possible with Filter > Segments with no letters, but it misses segments with numbers with letters.


Hello LAX, here SFO, can you please give some examples?

Hi. Unfortunately, not an airport, it's just my name.

If you notice, in the Filter field we find:

Segments with font color

Segments with highlight color

Segments with letters

Segments with no letters

Segments with no letters gives you numbers, basically.

Segments with letters gives you letters and numbers and letters and no number.

What is missing is the last option, ie. Segments with numbers (both with letters or without, we don't care).

This is the one relevant (at least to my interests). It allows me to quickly check if my thousands separator is correct (e.g. 10,000 > 10 000) and decimal as well (e.g. 0.01 > 0,001).

What I have been doing at the moment is ridiculous, I do several Target searches for:





 and also 





Hopefully it's clear now?


It should be possible to use QA Consistency > Glossary for this.

So I created a sample project like this:


And a QA glossary containing only two expressions:


To notice that this QA fails.

Igor, am I missing something here?

I've also set this preference:


I created a non-translatables glossary:


And ran QA Non-translatables. Not all errors are caught:




First QA failed, because regex is only possible at the source side. Instead of running the second QA, you’ll have to filter on NTFs and check manually.
I have one more idea: QA consistency glossaries with literal number conversions. Will test...

> Filter > Segments with numbers

Update 11 may have it.

1 person likes this

When adding the numbers literally to the QA glossary, the first type is caught, the second isn't:


Why is that, Igor?


Okay, for the second type, I came up with a multiple filtering action (aamof, this one can be used for the first type too, with a adjusted regex):

First filter on all source segments that contain numbers with 3 decimals. Make sure that Multiple is checked:


Keep the dialogue box open, check the target scope box, enter the search string for the target:


My assumption with this is that any deviant number with three decimals, isn't localised (has been forgotten). Hence the search for 


> the second isn't:

Possibly, because of the punctuation characters defined in the "Don not match" list.

If so, can you let CafeTran ignore these characters during qA?
Login to post a comment