Extracting messages

[christopanayotovvia]

Iw would be helpful if there's a CLI to extract all messages and build JSON :

ex. talkt extract /**/*.ts|x/ should find all keys, sort them alphabetically, nest them if needed ( nesting would be one option, the other is to be dot-notated ) and return as json. Example results

{
 "name": {
  "first": 'First name'
  "last": 'Last name'
 }
}

{
"name.first": "First name",
"name.last": "Last name"
}

[DoneDeal0]

Hi Christoph, thanks for your message and your interest in Talkr! However, I don't really see the added value of your proposal. What would be the point to find all the keys used in your app and then output a json file? You are supposed to have a json in the first place to use the package anyways. Also, some keys may be dynamic, so we couldn't infer the full path in some cases.

For exemple, the following script could return day.monday or day.sunday.

function DisplayDay({day}){
return (
<h1>{ T(`date.${date}`) }</h1>
)
}

But maybe I didn't understand well your proposal. If so, feel free to send a pr or to correct me.

[ablankenship10]

Talkr doesn't really work this way at least the way I use it, maybe there are other ways. 100% of all my strings are in my default language file (en.json). There are no default strings or anything within the codebase. There is nothing to extract from the codebase other than keys to the json locations where strings would be. If, for example, you wanted to get translations done you would just upload the JSON file and select the languages you want and I would presume translators send you back your translation files as basically a mirrored copy of the original JSON object with all the same keys but with other language text filled in.

Only tricky aspect maybe in the case where keys don't exist for other tenses or number of values (zero, one, few, many, etc) that would require the translator to add additional keys and probably wouldn't be aware of the Talkr required format for that. Not sure of the standards around this stuff.

[ablankenship10]

I think, however, a middle ground of what @christopanayotovvia is asking @DoneDeal0 would be you know how it shows a warning in console if a key doesn't exist in your JSON that you tried to use in your code? Maybe a command you could run that could extract all those and put them in your json file for you to fill in instead of creating manually. But the thing with that is those warnings only show on render of a component containing a call to T(), not the codebase as a whole. Also, anyone could rename the T() function whatever they want in each file so detecting it in the code files themselves rather than at runtime is still not trivial (or is trivial? I think either wat is correct.. English is weird).

IMO its just easier to fill in your JSON file first and then use the key in your code after it already exists. That said, is there a way to detect discrepancies between JSON files, like if I update en.ts but not fr.ts, would there be an easy way to diff them and find the updates I need to make to the other file?

[DoneDeal0]

A command line to check differences between json files should be doable. But it's not a big feature for Talkr, and there are probably existing tools to do it anyways. Actual improvements could be:

• making Talkr work with React Native.
• adding tests
• find a way to drive adoption for this lib lol (it's the lightest 18n package out there, we should aim for more downloads)

[samuelstroschein]

Hi everyone,

I am building an ecosystem of localization solutions that seem to partially serve @christopanayotovvia needs and @ablankenship10 idea regarding a CLI that can "lint" translations (see https://github.com/inlang/inlang/discussions/108. )

Talkr could work with the inlang project through a converter. Is there interest from your side?

@DoneDeal0
PS I went for the "push fluent as standard format" approach (we chatted on reddit).

The VS Code extension is one of the tools of the inlang project and makes extraction of text easy in addition to showing the messages inline.

[DoneDeal0]

Hi Samuel,

Yes I remember our chat on Reddit, I'm glad to see that Inlang has made such progresses! A CLI that checks all missing translations in json files would be very nice indeed.

The thing is that despite all its recent improvements (autocomplete, code refactor for faster performance and conditional rendering, official website, etc.), Talkr doesn't have a big user base yet. So I don't know whether it's worth it to spend a lot of time in coding such a feature with converters & whatnot.

Except if it looks easy/quick to implement, or if there is a crazy demand for that, I'll put this task on the back burner for now. Of course, any PR is welcome.

[christopanayotovvia]

Hi All, Thanks for sharing this. We’re currently using formatJS with formatted messages extractor and automatically upload/download from localise. Do you want to do a meeting on what are some issues that we have been facing ( and did workarounds ) and how your project approaches this? Best, Chris

…

On 20 Feb 2022, 19:08 +0200, Samuel Stroschein ***@***.***>, wrote: Hi everyone, I am building an ecosystem of localization solutions that seem to partially serve @christopanayotovvia needs and @ablankenship10 idea regarding a CLI that can "lint" translations (see inlang/inlang#108. ) Talkr could work with the inlang project through a converter. Is there interest from your side? @DoneDeal0 PS I went for the "push fluent as standard format" approach (we chatted on reddit). The VS Code extension is one of the tools of the inlang project and makes extraction of text easy in addition to showing the messages inline. — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you were mentioned.Message ID: ***@***.***>

-- This communication and any attachments may contain confidential information and are intended to be viewed only by the intended recipients. If you have received this message in error, please notify the sender immediately by replying to the original message and then delete all copies of the email from your systems.

[samuelstroschein]

@christopanayotovvia

Bumping in case you missed it.

I am super curious about your workflows and workarounds. Can you write me a mail to proxy at stroschein.me ?