unicode.rs

/* soon to be auto-generated */

use std::cmp::{Equal, Less, Greater};
use std::vec::ImmutableVector;
use std::option::None;

static LETTER_UPPERCASE : &'static [(char, char)] = &[
  ('\x41', '\x5a')
];

static LETTER_LOWERCASE : &'static [(char, char)] = &[
  ('\x61', '\x7a')
];

// table for common and simple case folding
// from data/6.3.0/CaseFolding.txt
static CASE_FOLDING_SIMPLE : &'static [(char, char)] = &[
  ('\U00000041', '\U00000061'), // 0041; C; 0061; # LATIN CAPITAL LETTER A
  ('\U00000042', '\U00000062'), // 0042; C; 0062; # LATIN CAPITAL LETTER B

  ('\U0001D6E2', '\U00000391') // '\U0001D6E2'//;MATHEMATICAL ITALIC CAPITAL ALPHA;Lu;0;L;<font> 0391;;;;N;;;;;'
];

fn bsearch_range_table(c: char, r: &'static [(char,char)]) -> bool {
  r.bsearch(|&(lo,hi)| {
      if lo <= c && c <= hi { Equal }
      else if hi < c { Less }
      else { Greater }
  }) != None
}

fn bsearch_lo(c: char, r: &'static [(char,char)]) -> Option<uint> {
  r.bsearch(|&(x, _)| {
      if x == c { Equal }
      else if x < c { Less }
      else { Greater }
  })
}

fn bsearch_hi(c: char, r: &'static [(char,char)]) -> Option<uint> {
  r.bsearch(|&(_, x)| {
      if x == c { Equal }
      else if x < c { Less }
      else { Greater }
  })
}

fn to_lower(c: char) -> char {
  match bsearch_lo(c, CASE_FOLDING_SIMPLE) {
    None        => c,
    Some(index) => CASE_FOLDING_SIMPLE[index].val1()
  }
}

fn to_upper(c: char) -> char {
  match bsearch_hi(c, CASE_FOLDING_SIMPLE) {
    None        => c,
    Some(index) => CASE_FOLDING_SIMPLE[index].val0()
  }
}

fn is_upper(c: char) -> bool {
  bsearch_range_table(c, LETTER_UPPERCASE)
}

fn is_lower(c: char) -> bool {
  bsearch_range_table(c, LETTER_LOWERCASE)
}

#[test]
fn test_is_upper(){
  assert!(is_upper('\U00000041'));
  assert!(is_upper('A'));
  assert!(!is_upper('a'));
}

#[test]
fn test_is_lower(){
  assert!(is_lower('\U00000061'));
  assert!(is_lower('a'));
  assert!(!is_lower('A'));
}

#[test]
fn test_to_lower(){
  assert_eq!(to_lower('A'), 'a');
  assert_eq!(to_lower('💩'), '💩');
}

#[test]
fn test_to_upper(){
  assert_eq!(to_upper('a'), 'A');
  assert_eq!(to_lower('💩'), '💩');
}