1
0
mirror of https://github.com/vlang/v.git synced 2023-08-10 21:13:21 +03:00
v/vlib/encoding/utf8/utf8_util_test.v

94 lines
2.5 KiB
V
Raw Normal View History

import encoding.utf8
2019-11-29 16:25:25 +03:00
fn test_utf8_util() {
2019-12-01 20:41:29 +03:00
// string test
src := 'ăĂ ôÔ testo æ' //_\u1E5A\u1E5B<=>\u1F49\u1F41<=>\u0128\u012a\u012c" // len 29 runes, raw 49 bytes
src_upper := 'ĂĂ ÔÔ TESTO Æ' //_\u1E5A\u1E5A<=>\u1F49\u1F49<=>\u0128\u012A\u012C"
src_lower := 'ăă ôô testo æ' //_\u1E5B\u1E5B<=>\u1F41\u1F41<=>\u0129\u012B\u012D"
upper := utf8.to_upper(src)
lower := utf8.to_lower(src)
assert upper == src_upper
assert lower == src_lower
assert utf8.to_upper('абвёabc12') == 'АБВЁABC12'
assert utf8.to_lower('АБВЁABC12') == 'абвёabc12'
2019-12-01 20:41:29 +03:00
// test len function
assert utf8.len('') == 0
assert utf8.len('pippo') == 5
assert utf8.len(src) == 15 // 29
assert src.len == 24 // 49
2019-12-12 21:08:44 +03:00
// western punctuation
a := '.abc?abcòàè.'
assert utf8.is_punct(a, 0) == true
assert utf8.is_punct('b', 0) == false
assert utf8.is_uchar_punct(0x002E) == true
assert utf8.is_punct(a, 4) == true // ?
assert utf8.is_punct(a, 14) == true // last .
assert utf8.is_punct(a, 12) == false // è
println('OK western')
2019-12-12 21:08:44 +03:00
// global punctuation
b := '.ĂĂa. ÔÔ TESTO Æ'
assert utf8.is_global_punct(b, 0) == true
assert utf8.is_global_punct('.', 0) == true
assert utf8.is_uchar_punct(0x002E) == true
assert utf8.is_global_punct(b, 6) == true // .
assert utf8.is_global_punct(b, 1) == false // a
// test utility functions
assert utf8.get_uchar(b, 0) == 0x002E
2019-11-29 16:25:25 +03:00
}
fn test_raw_indexing() {
a := 'V Lang!'
// test non ascii characters
assert utf8.raw_index(a, 0) == ''
assert utf8.raw_index(a, 1) == ''
// test ascii characters
assert utf8.raw_index(a, 2) == 'V'
assert utf8.raw_index(a, 3) == ' '
assert utf8.raw_index(a, 4) == 'L'
assert utf8.raw_index(a, 5) == 'a'
assert utf8.raw_index(a, 6) == 'n'
assert utf8.raw_index(a, 7) == 'g'
assert utf8.raw_index(a, 8) == '!'
}
fn test_reversed() {
a := 'V Lang!'
b := 'hello world'
assert utf8.reverse(a) == '!gnaL V'
assert utf8.reverse(b) == 'dlrow olleh'
}
fn test_is_control() {
for ra in `a` .. `z` {
assert utf8.is_control(ra) == false
}
for ra in `A` .. `Z` {
assert utf8.is_control(ra) == false
}
assert utf8.is_control('\x01'.runes()[0]) == true
assert utf8.is_control('\u0100'.runes()[0]) == false
}
fn test_is_letter() {
for ra in `a` .. `z` {
assert utf8.is_letter(ra) == true
}
for ra in `A` .. `Z` {
assert utf8.is_letter(ra) == true
}
assert utf8.is_letter(`ɀ`) == true
assert utf8.is_letter(`ȶ`) == true
assert utf8.is_letter(`ȹ`) == true
}