Hey Jed folks! A couple of us were having a discussion about the new Jed format downstream at mikeedwards/po2json#32 . We're trying to figure out what the best output for something like this:

msgid "one product"
msgid_plural "%d products"
msgstr[0] "jeden produkt"
msgstr[1] "%d produkty"
msgstr[2] "%d produktów"

We were outputting (correctly or not):

         "one product": [
            "%d products",
            "jeden produkt",
            "%d produkty",
            "%d produktów"
         ],

Would it be like this now that the zeroth element is being lopped off?

         "one product": [
            "jeden produkt"
         ],
         "%d products": [
            "jeden produkt",
            "%d produkty",
            "%d produktów"
         ]

po2json outputs the plural forms with a key of Plural-Forms (https://metacpan.org/source/GETTY/Locale-Simple-0.011/bin/po2json#L110) and it looks like Jed wants it under a key of plural_forms

I'm guessing there is no such thing as a spec for PO-as-json so perhaps Jed just needs to look at both keys?

Thoughts? If you are happy with this approach than I'll throw together a quick PR.

Hi, this has been bugging me for a while

// Note: by convention, the 0-index location of the translations // is never accessed. It's just a thing, I guess.

what does this mean, convention by whom ? I look at my po files and all of them have all useful data.

They don't have any useless translation that is never accessed, so where is this convention coming from. For kicks I've been going through the source code and deleting all ' + 1' there and there, and with an ugly

val_list = _.compact(val_list);

It just works, but there's no waste of an empty index in an array. Why would I parse an mo file to generate empty data ?

Hi,

I'm currently using xgettext in Python mode, but there are quite a few edge cases where it doesn't work correctly. How do you extract gettext keys from your JS files?

thanks, Florian

In commit 8cc37367d55eda9f800f3a4fa5fab6b5756e9ad7, it looks like the expected format of locale_data changed so that the first element of the translated data array is no longer supposed to be a null. That's at odds with what's on the doc site http://slexaxton.github.io/Jed/ and a lot of the examples on the web. po2json, which I was using to parse my po files, doesn't output the right format anymore. Please update the doc and examples.

Hey,

With this fix make test is error free! My Mac needs 6 seconds with the current HEAD, Travis CI needs 3 seconds. This patched version is a bit less than 1 second on my Mac, so it should be half a second on Travis CI which is way less than mocha's default timeout of 2 seconds...

expect.js's be and to functions aren't really complex but heavy enough to be more than 6x slower than a simple !== + throw.

Cheers, Philip

Hi,

I'm trying to get my messages.po from Javascript files using Jed. I'm using xgettext and this command:

  xgettext --keyword=translate --from-code utf-8 -L Perl `find . -iname "*.js"`

It seems to work fine, but the problem comes when xgettext finds plurals like the next ones:

i18n.translate("Pass through %1$s %2$s").fetch(var1, var2);

Because it doesn't recognizes $ as a valid character. Is there any work around using xgettext, or otherwise, any other way to use a different numbering method xgettext-compatible?

Thanks.

This requires the sprintf.js npm module instead of inlining it and updates the library from 0.7-beta1 to version 1.0.3. A few changes to the tests were required to accommodate for the change.

See #41.

At the top of https://github.com/messageformat/Jed and in https://www.npmjs.com/package/jed, the URL http://slexaxton.github.com/Jed is written instead of http://messageformat.github.io/Jed

All green.

If using semver, should be a major version because not backwards compatible ... my bad.

Should go hand in hand with WrinklyNinja/jed-gettext-parser#6

I would say don't merge this until its analog for parsing is ready.

Thanks for Jed!

I'm looking into generating the .po files from my source. I wondered how you do/did that?

I started working on contributions to an existing package (handelbars-xgettext), but there must be something already out there?

The repository url in the package.json contains a username, which causes problems in some browsers. https://github.com/messageformat/Jed/blob/351c47d5c57c5c81e418414c53ca84075c518edb/package.json#L8

By opening this link in firefox the browser attempts to login to github with the SlexAxton username and shows the following error:

You are about to log in to the site “github.com” with the username “SlexAxton”, but the website does not require authentication. This may be an attempt to trick you.

I wrote a small implementation of the gettext concept recently and wanted to provide two new methods to the community. I was writing these in React Native, as shown below, and would be happy to incorporate the pure JS equivalents into this project for Hacktoberfest! 🎃🎃🎃

In short, the first method pboolgettext returns the plural form if a boolean is true or the singular form if the boolean is false. This is a case when no variables need to be inserted.

The second, pvboolgettext, does the same, but formats the plural message with variables.

/*
    * Returns the singular form of the message if a given boolean condition is true, otherwise the plural form.
    *
    * usage : pgettext(singularMessageId: string, pluralMessageId: string, b: boolean)
    *
    * @param singularMessageId <string> : the message ID to be used if the number of elements specified is zero
    * @param pluralMessageId <string> : the message ID to be sued if the number of elements specified is greater than zero
    * @param b <boolean> : a condition indicating if the singular message form should be displayed or not
     */
    pboolgettext(singularMessageId: string, pluralMessageId: string, b: boolean) {
        if (typeof singularMessageId === "undefined" || typeof singularMessageId === "undefined" || typeof b === "undefined") {
            throw "pboolgettext: Invalid arguments";
        }
        return b ? this.translations[singularMessageId] : this.translations[pluralMessageId];
    }

    /*
    * Returns the singular form of the message if a given boolean condition is true, otherwise the plural form formatted with variables.
    *
    * usage : pgettext(singularMessageId: string, pluralMessageId: string, b: boolean, value: any)
    * usage : pgettext(singularMessageId: string, pluralMessageId: string, b: boolean, values: [any, ... ,any])
    *
    * Note : variables can be mixed string and number.
    *
    * @param singularMessageId <string> : the message ID to be used if the number of elements specified is zero
    * @param pluralMessageId <string> : the message ID to be sued if the number of elements specified is greater than zero
    * @param b <boolean> : a condition indicating if the singular message form should be displayed or not
    * @param v <any> : the variable or variables to insert into the plural string
     */
    pvboolgettext(singularMessageId: string, pluralMessageId: string, b: boolean, ...values) {
        if (typeof singularMessageId === "undefined" || typeof singularMessageId === "undefined" || typeof b === "undefined" || typeof values === "undefined") {
            throw "pvboolgettext: Invalid arguments";
        }
        if (b) {
            return this.translations[singularMessageId];
        }
        return this.vngettext(pluralMessageId, ...values);
    }

Hi,

How does one generate .pot from the source code using Jed? I assume that gettext won't be able to parse the calls to Jed since the syntax is too complex. If that's so, what are the alternatives?

Cheers, Andrej

There are two examples of plural forms on this page

1. "%d key" : [ "%d key", "%d keys" ],
2. "test singular": ["test plural", "test_1 singular", "test_1 plural"],

These appear to be different structures. Using Gettext terminology they look like:

1. msgid: [ msgstr1, msgstr2 ]
2. msgid: [ msgid_plural, msgstr1, msgstr2 ]

Can you please clarify how this structure should be used and whether it has changed at some point?

Currently, Jed uses sprintf 0.7 which is more than four years old and contains several bugs which are meanwhile fixed (for example huh?).

@SlexAxton do you think, this could be updated? maybe even made a dependency in package.json?