mikros/rust
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

char_range_at perf work

Browse Source 
Moves multibyte code to it's own function to make char_range_at
easier to inline, and faster for single and multibyte chars.

Benchmarked reading example.json 100 times, 1.18s before, 1.08s
after.
This commit is contained in:
Gary Linscott 2013-07-10 14:32:59 -04:00
parent e95fcfafc7
commit 179637304a
1 changed files with 60 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

88
src/libstd/str.rs
View File

@ -722,17 +722,29 @@ pub fn count_bytes<'b>(s: &'b str, start: uint, n: uint) -> uint {
end - start
}
// https://tools.ietf.org/html/rfc3629
static UTF8_CHAR_WIDTH: [u8, ..256] = [
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x1F
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x3F
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x5F
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x7F
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0x9F
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0xBF
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, // 0xDF
3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, // 0xEF
4,4,4,4,4,0,0,0,0,0,0,0,0,0,0,0, // 0xFF
];
/// Given a first byte, determine how many bytes are in this UTF-8 character
pub fn utf8_char_width(b: u8) -> uint {
let byte: uint = b as uint;
if byte < 128u { return 1u; }
// Not a valid start byte
if byte < 192u { return 0u; }
if byte < 224u { return 2u; }
if byte < 240u { return 3u; }
if byte < 248u { return 4u; }
if byte < 252u { return 5u; }
return 6u;
return UTF8_CHAR_WIDTH[b] as uint;
}
#[allow(missing_doc)]
@ -1714,26 +1726,29 @@ fn is_char_boundary(&self, index: uint) -> bool {
* If `i` is greater than or equal to the length of the string.
* If `i` is not the index of the beginning of a valid UTF-8 character.
*/
#[inline]
fn char_range_at(&self, i: uint) -> CharRange {
let b0 = self[i];
let w = utf8_char_width(b0);
assert!((w != 0u));
if w == 1u { return CharRange {ch: b0 as char, next: i + 1u}; }
let mut val = 0u;
let end = i + w;
let mut i = i + 1u;
while i < end {
let byte = self[i];
assert_eq!(byte & 192u8, TAG_CONT_U8);
val <<= 6u;
val += (byte & 63u8) as uint;
i += 1u;
if (self[i] < 128u8) {
return CharRange {ch: self[i] as char, next: i + 1 };
}
// Clunky way to get the right bits from the first byte. Uses two shifts,
// the first to clip off the marker bits at the left of the byte, and then
// a second (as uint) to get it to the right position.
val += ((b0 << ((w + 1u) as u8)) as uint) << ((w - 1u) * 6u - w - 1u);
return CharRange {ch: val as char, next: i};
// Multibyte case is a fn to allow char_range_at to inline cleanly
fn multibyte_char_range_at(s: &str, i: uint) -> CharRange {
let mut val = s[i] as uint;
let w = UTF8_CHAR_WIDTH[val] as uint;
assert!((w != 0));
// First byte is special, only want bottom 5 bits for width 2, 4 bits
// for width 3, and 3 bits for width 4
val &= 0x7Fu >> w;
val = (val << 6) | (s[i + 1] & 63u8) as uint;
if w > 2 { val = (val << 6) | (s[i + 2] & 63u8) as uint; }
if w > 3 { val = (val << 6) | (s[i + 3] & 63u8) as uint; }
return CharRange {ch: val as char, next: i + w};
}
return multibyte_char_range_at(*self, i);
}
/// Plucks the character starting at the `i`th byte of a string
@ -2430,7 +2445,11 @@ fn test_pop_char_fail() {
fn test_push_char() {
let mut data = ~"ประเทศไทย中";
data.push_char('华');
assert_eq!(~"ประเทศไทย中华", data);
data.push_char('b'); // 1 byte
data.push_char('¢'); // 2 byte
data.push_char('€'); // 3 byte
data.push_char('𤭢'); // 4 byte
assert_eq!(~"ประเทศไทย中华b¢€𤭢", data);
}
#[test]
@ -3240,6 +3259,19 @@ fn test_total_ord() {
"22".cmp(& &"1234") == Greater;
}
#[test]
fn test_char_range_at() {
let data = ~"b¢€𤭢𤭢€¢b";
assert_eq!('b', data.char_range_at(0).ch);
assert_eq!('¢', data.char_range_at(1).ch);
assert_eq!('€', data.char_range_at(3).ch);
assert_eq!('𤭢', data.char_range_at(6).ch);
assert_eq!('𤭢', data.char_range_at(10).ch);
assert_eq!('€', data.char_range_at(14).ch);
assert_eq!('¢', data.char_range_at(17).ch);
assert_eq!('b', data.char_range_at(19).ch);
}
#[test]
fn test_char_range_at_reverse_underflow() {
assert_eq!("abc".char_range_at_reverse(0).next, 0);