-
Type:
Suggestion
-
Resolution: Won't Fix
-
Priority:
Medium
-
None
-
Affects Version/s: 0.8 Sprint 8
-
Component/s: None
-
None
Here's the list of checks I run when receiving files from NZTC:
- The folder structure is the same as untranslated messages export. TAC probably won't be able to do this easily. For reference, I just run the following:
diff -rqs translated/ untranslated/
- All files have been encoded as ISO 8859-1 as required by the .properties spec:
In shell script I use enca in the following fashion:
files=`find * -type f` for i18nfile in $files do echo "Processing $i18nfile" enca -p -L none $i18nfile done
In Java, I believe there is a class in the ICU4J library that allows you to do charset detection.
I also use the rsc bundle check ant task @ http://rscbundlecheck.sourceforge.net/ to perform this series of checks:
- unicode check: All non-ascii characters are replaced by their corresponding unicode escape sequence (e.g. è should be replaced by \u00E8) and unicode escapes in the files are valid.
- line end check: No blanks after the end of the message.
- empty key check: No empty keys ( = some value).
- placeholder check: check for the presence of the placeholders defined in the original message, no additional placeholders allowed and no missing placeholder allowed.
- empty value check: no empty values (some.key = )
- invalid char check: Checks if a value has invalid characters (ASCII value > 127)
- messageformat check: Checks if MessageFormat can be instantiated with each value.
I also check that any choice formats defined in keys are valid.
Can we get a list of all the validation checks performed by TAC during import and consider implementing the checks mentioned above that are not already in place?
- is related to
-
I18N-437 TAC Import check of arguments in the translated message is a bit too strict
-
- Closed
-