xitrum-framework / scala-xgettext Goto Github PK

"POT-Creation-Date: \n" is all that's there. Why not populate it?

scala-xgettext 1.2 fixes this problem by extracting the key as "don't have to".

The arguments ``singular``, ``plural``, and ``context`` MUST be literal constant string, like in the above examples. Otherwise exception will be thrown during compilation.

Would it be possible to add a configuration option that changes this behavior into ignoring any string that is not a literal constant string instead of throwing an Exception? If you deem this useful, I could look into making a PR that adds this behavior.

I think I need a way to specify the plural forms formula.

https://www.gnu.org/software/gettext/manual/gettext.html#Plural-forms

Plural-Forms: nplurals=2; plural=n != 1; is the default for English, and would make a sane default for this option. Maybe this can be passed in through a compiler property such as '-P:xgettext_plurals:nplurals=2; plural=n != 1'

If lines in .pot and .po files are random, it's very difficult to see diff result when there's modification.

We're observing a behavior when using interpolated strings. If our code is:

println(t(s"Sending node"))

Then the output in i18n.pot is:

msgid new StringContext(scala.this.Predef.wrapRefArray[String](Array[String]{"Sending node"})).s(immutable.this.Nil)

Where interpolation is actually used:

println(t(s"Sending node ${node.name}"))

we see:

msgid new StringContext(scala.this.Predef.wrapRefArray[String](Array[String]{"Sending node ", ""})).s(scala.this.Predef.genericWrapArray[Any](Array[Any]{StreamOfIndexables.this.channel().getClass().toString()}))

Can we not use interpolation with scala-xgettext?

See #3.

Multiple markers for t can be configured like this:

scalacOptions ++= Seq(
  "xitrum.I18n", "t:tr", "t:preparetr", "t:marktr"
).map("-P:xgettext:" + _)

Similar for tn, tc, and tcn.

The current implementation doesn't seem to add the #, java-format flag as documented here:

https://www.gnu.org/software/gettext/manual/html_node/PO-Files.html#The-Format-of-PO-Files

Tools like Poedit understand this flag and displays it to translators:

Translation tools may also use this to validate that all placeholders are used in the translation, and that they are syntactically well formed.

Our legacy project used "t" with constants. We created pot file manualy.
For example:

  val titleKey = "Some title"
  ...
  val messageTitle = t(titleKey)

pot file look like this:

...
msgid titleKey
msgstr ""
...

Can we extract constant value or show warning on such cases?

In Java/Scala-land, I would argue, tr, trn, ... are more common, at least in my projects. I currently still use xgettext which accept the method names with the -koption. Using scala-xgettext is not possible because of incompatible hardcoded method names.

I'd like to "version" the pot files by giving it the project name and the commit hash at which the i18n.pot file was generated. Those aren't really details that xgettext should extract itself, so perhaps let the build system pass that as an option.

I get this:

msgid "Waiting for {0} in-flight document to finish processing..."
msgid_plural "Waiting for {0} in-flight documents to finish processing..."
msgstr[0] ""

when I expect something like

msgid "Waiting for {0} in-flight document to finish processing..."
msgid_plural "Waiting for {0} in-flight documents to finish processing..."
msgstr[0] ""
msgstr[1] ""

https://www.gnu.org/software/gettext/manual/gettext.html#Translating-plural-forms