num-format

• Add github actions CI
• Fix a few clippy lints

• Bump 3rd party dependencies
• Update cldr-numbers-full git submodule to commit b52a87048985d3052f12d30d05cfe5423ad92709
• Locale::from_str now supports underscore-delimited locales, e.g. "de_DE", in addition to previously supported dash-delimiated locales, e.g "de-DE"

Breaking changes:

• Because SystemLocale requires a lot of dependencies (especially for Windows users), it's now only available behind a feature flag (with-system-locale). The upside is, of course, that those using num-format without this feature will only need to download and compile a small handful of dependencies.

Breaking changes:

• Environment was renamed to SystemLocale.
• Error and ErrorKind were refactored.
• Compiling for Windows depends on bindgen and so requires Clang 3.9 or above.

Bug fixes:

• num-format now works on Windows!

Enhancements:

• Major overhaul of the code that interacts with operating systems (linux, macOS, freebsd, Windows, etc.)
• Methods available_names, from_name, and name were added to both Locale and SystemLocale.
• Several error messages now provide more detail about what exactly went wrong.

Bug fixes:

• WriteFormatted was being unconditionally implemented on std::os::unix::net::UnixStream, which is a unix-only type; so we moved this implementation behind a #[cfg(unix)] flag.

Breaking changes:

• Flatten module structure

  • We decided to flatten the module structure of num-format to both improve code readabiity and to simplify the importing of types by users. Unfortunately this is a breaking change.
  • The num_format::format and num_format::errors modules have been removed and all of their containing types have been moved to the crate root.
  • So whereas one used to do this:

  use num_format::errors::{Error, ErrorKind};
  use num_format::format::{CustomFormat, Environment, Format, Locale};
  use num_format::format::utils::{InfinityStr, MinusSignStr, NanStr};
  

  • one now does this:

  use num_format::{CustomFormat, Environment, Error, ErrorKind, Format, Locale};
  use num_format::utils::{InfinityStr, MinusSignStr, NanStr};
  

  • Thank you BurntSushi for the suggestion

• Refactor Error and ErrorKind

  • Since the future of the failure crate is up in the air, we decided to remove it as a dependency, refactoring num_format::Error and num_format::ErrorKind to work with just plain old std::error::Error
  • Thank you BurntSushi for the suggestion

New features:

• impl From<Environment> and From<Locale> for CustomFormat

  • Now you can easiliy turn an Environment or a Locale into a CustomFormat; so it's easy to start with an already complete "base" format and then just tweak it slightly to the custom format you want.

  use num_format::{CustomFormat, Error, Locale, ToFormattedString};
  
  fn main() -> Result<(), Error> {
      let format = CustomFormat::from(Locale::en_IN)
          .into_builder()
          .separator(Some(' '))
          .build()?;
      let s = 1_000_000.to_formatted_string(&format);
      assert_eq!(&s, "10 00 000");
      Ok(())
  }
  

Initial release