Skip to content

Latest commit

 

History

History
62 lines (46 loc) · 3.5 KB

README.md

File metadata and controls

62 lines (46 loc) · 3.5 KB

deunicode

Documentation

The deunicode library transliterates Unicode strings such as "Æneid" into pure ASCII ones such as "AEneid". It includes support for emoji. It's compatible with no-std Rust environments.

Deunicode is quite fast, supports on-the-fly conversion without allocations. It has a compact representation of Unicode data to minimize memory overhead and executable size (about 75K codepoints mapped to 245K ASCII characters, using 450KB of memory, 160KB gzipped).

Examples

use deunicode::deunicode;

assert_eq!(deunicode("Æneid"), "AEneid");
assert_eq!(deunicode("étude"), "etude");
assert_eq!(deunicode("北亰"), "Bei Jing");
assert_eq!(deunicode("ᔕᓇᓇ"), "shanana");
assert_eq!(deunicode("げんまい茶"), "genmaiCha");
assert_eq!(deunicode("🦄☣"), "unicorn biohazard");

When to use it?

It's a better alternative than just stripping all non-ASCII characters or letting them get mangled by some encoding-ignorant system. It's be okay for one-way conversions for things like search indexes and tokenization, as a stronger version of Unicode NFKD. It may be used for generating nice identifiers for file names and URLs, which aren't too user-facing.

However, like most "universal" libraries of this kind, it has a one-size-fits-all 1:1 mapping of Unicode code points, which can't handle language-specific exceptions nor context-dependent romanization rules. These limitations are only slightly suboptimal for European languages and Korean Hangul, but make a mess of Japanese Kanji.

Guarantees and Warnings

Here are some guarantees you have when calling deunicode():

  • The String returned will be valid ASCII; the decimal representation of every char in the string will be between 0 and 127, inclusive.
  • Every ASCII character (0x00 - 0x7F) is mapped to itself.
  • All Unicode characters will translate to printable ASCII characters (\n or characters in the range 0x20 - 0x7E).

There are, however, some things you should keep in mind:

  • Some transliterations do produce \n characters.
  • Some Unicode characters transliterate to an empty string, either on purpose or because deunicode does not know about the character.
  • Some Unicode characters are unknown and transliterate to "[?]" (or a custom placeholder, or None if you use a chars iterator).
  • Many Unicode characters transliterate to multi-character strings. For example, "北" is transliterated as "Bei".
  • The transliteration is context-free, and not sophisticated enough to produce proper Chinese or Japanese. Han characters used in multiple languages are mapped to a single Mandarin pronounciation, and will be mostly illegible to Japanese readers. Transliteration can't handle cases where a single character has multiple possible pronounciations.

Unicode data

For a detailed explanation on the rationale behind the original dataset, refer to this article written by Burke in 2001.

This is a maintained alternative to the unidecode crate, which started as a Rust port of Text::Unidecode Perl module.