jenseng / i18nliner-js Goto Github PK

speakingurl is listed under an older version and includes built files that do not cooperate as well with legacy browsers.

if there are check/export errors, we aren't properly reporting the status to grunt... you always get "Done, without errors"

rather than explicitly relying on esprima-fb, we should use recast instead like react-i18nliner does. additionally, we should use config.recastOptions (if provided). that way we'll consistently parse stuff, and it also makes it possible to swap out esprima-fb for both and instead use something like acorn (or even acorn-babel)

needs newlines to make it look prettier. compare:

http://cl.ly/image/3K0q0C3f3N3t/Screen%20Shot%202014-10-08%20at%209.27.48%20AM.png

vs

http://cl.ly/image/1A1O1g0r0d2f/Screen%20Shot%202014-10-08%20at%209.28.07%20AM.png

also should probably say <unsupported expression> instead of <unsupport expression>

with the move of settings from I18nliner over to I18nliner.config, this one got overlooked, so inferred keys currently have no upper bound

Trying to use it with svelte. Got error on
Unexpected character '#' (2:1)

"#" is part of template language

if you have something like:

I18n.t("Session")

and elsewhere you have:

I18n.t("Session", {count: theCount})
// equivalent to
// I18n.t({one: "1 Session", other: "%{count} Sessions"}, {count: theCount})

you get a key as scope error, because the second one infers the same key as the first one (session_3ac1c536). for it to be truly equivalent to the longhand one:/other: object, it should infer the key from the other value (e.g. count_sessions_f79b13d7)

since plugins might define new i18nliner settings, we should initialize all plugins specified in .i18nrc before any other settings in that file ... otherwise the plugin might overwrite those user overrides

like we do in i18nliner.rb ... sometimes you need to break up a string for readability, e.g.

I18n.t("lorem ipsum dolor sit amet bacon narwhals zomg this " +
       "is getting out of control")

there are tests around this, but it could be something isn't wired up correctly...

I18n.t("foo", {count: count})

should be automagically equivalent to

I18n.t({one: "1 foo", other: "%{count} foos"}, {count: count})

but it's behaving like:

I18n.t("foo")

some plugins (e.g. react-i18nliner) may need to add some stuff to the i18n runtime extensions (e.g. stick ComponentInterpolator on the I18n global). so plugins need a way to:

1. signal that they have runtime extensions (perhaps multiple exports? or i18nliner could require("plugin-name/runtime")?)
2. do so in such a way so as not to include its non-runtime extensions in the browserified file ... i.e. it makes no sense to bloat the runtime bundle with extractors/etc

Hi,
I'm trying to export the labels, using:

./node_modules/.bin/i18nliner export

there is a way to configure which file type you want to scan file to extract the translatable labels?
I guess I need to add it in .i18nrc file, but I didn't found any example of it.

Regards.

underscored (and _crc32) keys get truncated so that they don't get too long, but due to an off-by-one oversight, the i18nliner(rb) ones are one char longer. we should make this value configurable both here and there, and make sure they default to the same thing (50 chars, not 51)

need to require it and add it in commands.js

sometimes trailing whitespace is relevant (e.g. when porting something like rails' Array#to_sentence), so we should preserve it by default

that said, we should have a mechanism to override it ... e.g. for i18nliner-handlebars, whitespace stripping is probably desirable

kinda like how eslint does w/ .eslintrc ... basically we should have a config file where you indicate what add-ons you want (i18nliner-handlebars, react-i18nliner, etc), and i18nliner will just figure it out. this would also be a good place for any sort of settings (inferred key format, directories, etc)

that way you won't need to write any glue code ... as can be seen here, it can get a little out of control, and it you really can't just use the i18nliner bin or grunt task as-is.

i18nliner (or rather gglobby) follows symlinks. if a symlink is broken, everything asplodes. it should not.

this might need to be fixed in gglobby rather than i18nliner

CallHelpers.keyifyUnderscored doesn't convert dashes to underscores. this makes it slightly incompatible with i18nliner(.rb)

should audit it for any other obvious discrepancies

Hello, is there a way to tell i18nliner export to only search specific folders and specify an output file?

i.e. i18nliner --search /path/to/folder --output /path/to/json

Thanks :)

running the check command is quite slow, depending on the number of files in your directory... even though we auto-ignore node_modules, we still recursively walk everything (and non-js stuff too!) before selecting/excluding stuff

some ideas:

1. be able to specify a directory/directories where your js files live and only look in there
2. reimplement globby to do lazy enumeration. a big win there is we could break up exclusion patterns and not walk directories we know we don't need to

might be worth looking into the other globby to see if we can build on top of that instead

Is it webpack/rollup compatible?