Moodle

At the moment, I've got a script that fetches all get_string() and print_string() calls from .php, .html and .htm files. It recognizes the string identifier and the module. There are some more complicated callers that we will have to deal manually with - eg:

get_string('error'.$this->response['faultCode'], 'mnet')
get_string("format$courseformat","format_$courseformat")
get_string('modulename', $this->cm->modname)

I will try to mark up such places. Hopefully some alpha results soon.

First estimations are here! I need to double check everything before I publish the langchecker.php script but it seems we've got something around 3387 strings that are not called directly. But this number also includes countries.php, currencies.php and timezones.php that we should ignore as they are used to populate <select> form fields.

The problem mentioned above remains. And it influence results in a very bad way. Example: we have the caller
get_string('modulename', $this->cm->modname)
or
get_string('filtername', $myname)

My langchecker.php is unable to determine stringis and module from such caller and therefore it considers
$string['filtername'] = "Filter name";
as an unused string, which is wrong.

The first preview of the results. Please comment the format of the output report.

Interesting document. It means that over 30% of the strings is there for backward compatability. Those should definitly go out the HEAD branch, so new language packs for 2.0 have a smaller workload to start off with.

The second group of strings, the "could not recognise" section": sorry, my head is too small to understand why and what the importance/impact is of those.
I notice "parentlanguage in there - that needs some special attention, since it is a setting, not a language string.Those settings should imho only exist in langconfig.php

The thirth group are calls for a string, that don't have a string. I see some obvious ones, like parentlanguage, descep, etc. Are all the others missing strings in the en_utf8 language pack?

Thanks a lot for your work David!

this is Great David! - I just went looking for a tool like this hoping that someone else had already done the work! thanks for sharing it!

Good work. Is the source of the script available?

David, I think that what you should do with strings like

get_string('error'.$this->response['faultCode'], 'mnet')
get_string("format$courseformat","format_$courseformat")
get_string('modulename', $this->cm->modname)

is, where possible, remember the name of the lang file, so in your report you can list the lang files that we know have extra used strings in them.

Also, make a list of all these tricky get_string calls, and displays the with two lines of surrounding context (a bit like a diff) so we can review them.

Then we can review them, and perhaps make the script a bit smarter, so that, for example, it can correctly handle cases like get_string('modulename', $this->cm->modname).

What's the status of this? It would speed up translation for all translators.

I have partially solved this for now. All the string names that were not found in Moodle 2.0 source code (as a pattern) are now marked as "greylisted" in AMOS. We have ~2000 greylisted strings at the moment, some of them are false positives. We can go through the greylist in AMOS and if we are 100% sure the string is not used any more, it can be removed from CVS HEAD. Although I would suggest to wait after we branch off MOODLE_20_STABLE.

Waiting for forking MOODLE_20_STABLE

Re-triage: Moving into a seprate task in DEV backlog.