blob: 7c5cf6c7c462e568cff797b519e88e5b85c66df7 [file] [log] [blame]
/* Copyright 2016 The encode_unicode Developers
*
* Licensed under the Apache License, Version 2.0, <LICENSE-APACHE or
* http://apache.org/licenses/LICENSE-2.0> or the MIT license <LICENSE-MIT or
* http://opensource.org/licenses/MIT>, at your option. This file may not be
* copied, modified, or distributed except according to those terms.
*/
use errors::{FromStrError, EmptyStrError, NonAsciiError, InvalidUtf8Slice, InvalidUtf8Array};
use utf8_iterators::Utf8Iterator;
use traits::{CharExt, U8UtfExt};
use utf16_char::Utf16Char;
extern crate core;
use self::core::{hash, fmt, str, ptr};
use self::core::cmp::Ordering;
use self::core::borrow::Borrow;
use self::core::ops::Deref;
use self::core::mem::transmute;
#[cfg(feature="std")]
use self::core::iter::FromIterator;
#[cfg(feature="std")]
#[allow(deprecated)]
use std::ascii::AsciiExt;
#[cfg(feature="ascii")]
extern crate ascii;
#[cfg(feature="ascii")]
use self::ascii::{AsciiChar,ToAsciiChar,ToAsciiCharError};
// I don't think there is any good default value for char, but char does.
#[derive(Default)]
// char doesn't do anything more advanced than u32 for Eq/Ord, so we shouldn't either.
// The default impl of Ord for arrays works out because longer codepoints
// start with more ones, so if they're equal, the length is the same,
// breaks down for values above 0x1f_ff_ff but those can only be created by unsafe code.
#[derive(PartialEq,Eq, PartialOrd,Ord)]
#[derive(Clone,Copy)]
/// An unicode codepoint stored as UTF-8.
///
/// It can be borrowed as a `str`, and has the same size as `char`.
pub struct Utf8Char {
bytes: [u8; 4],
}
/////////////////////
//conversion traits//
/////////////////////
impl str::FromStr for Utf8Char {
type Err = FromStrError;
/// Create an `Utf8Char` from a string slice.
/// The string must contain exactly one codepoint.
///
/// # Examples
///
/// ```
/// use encode_unicode::error::FromStrError::*;
/// use encode_unicode::Utf8Char;
/// use std::str::FromStr;
///
/// assert_eq!(Utf8Char::from_str("a"), Ok(Utf8Char::from('a')));
/// assert_eq!(Utf8Char::from_str("šŸ‚ "), Ok(Utf8Char::from('šŸ‚ ')));
/// assert_eq!(Utf8Char::from_str(""), Err(Empty));
/// assert_eq!(Utf8Char::from_str("ab"), Err(MultipleCodepoints));
/// assert_eq!(Utf8Char::from_str("eĢ"), Err(MultipleCodepoints));// 'e'+u301 combining mark
/// ```
fn from_str(s: &str) -> Result<Self, FromStrError> {
if s.is_empty() {
Err(FromStrError::Empty)
} else if s.len() != 1+s.as_bytes()[0].extra_utf8_bytes_unchecked() {
Err(FromStrError::MultipleCodepoints)
} else {
let mut bytes = [0; 4];
bytes[..s.len()].copy_from_slice(s.as_bytes());
Ok(Utf8Char{bytes: bytes})
}
}
}
impl From<Utf16Char> for Utf8Char {
fn from(utf16: Utf16Char) -> Utf8Char {
match utf16.to_tuple() {
(a @ 0...0x00_7f, _) => {
Utf8Char{ bytes: [a as u8, 0, 0, 0] }
},
(u @ 0...0x07_ff, _) => {
let b = 0x80 | (u & 0x00_3f) as u8;
let a = 0xc0 | ((u & 0x07_c0) >> 6) as u8;
Utf8Char{ bytes: [a, b, 0, 0] }
},
(u, None) => {
let c = 0x80 | (u & 0x00_3f) as u8;
let b = 0x80 | ((u & 0x0f_c0) >> 6) as u8;
let a = 0xe0 | ((u & 0xf0_00) >> 12) as u8;
Utf8Char{ bytes: [a, b, c, 0] }
},
(f, Some(s)) => {
let f = f + (0x01_00_00u32 >> 10) as u16;
let d = 0x80 | (s & 0x00_3f) as u8;
let c = 0x80 | ((s & 0x03_c0) >> 6) as u8
| ((f & 0x00_03) << 4) as u8;
let b = 0x80 | ((f & 0x00_fc) >> 2) as u8;
let a = 0xf0 | ((f & 0x07_00) >> 8) as u8;
Utf8Char{ bytes: [a, b, c, d] }
}
}
}
}
impl From<char> for Utf8Char {
fn from(c: char) -> Self {
Utf8Char{ bytes: c.to_utf8_array().0 }
}
}
impl From<Utf8Char> for char {
fn from(uc: Utf8Char) -> char {
unsafe{ char::from_utf8_exact_slice_unchecked(&uc.bytes[..uc.len()]) }
}
}
impl IntoIterator for Utf8Char {
type Item=u8;
type IntoIter=Utf8Iterator;
/// Iterate over the byte values.
fn into_iter(self) -> Utf8Iterator {
Utf8Iterator::from(self)
}
}
#[cfg(feature="std")]
impl Extend<Utf8Char> for Vec<u8> {
fn extend<I:IntoIterator<Item=Utf8Char>>(&mut self, iter: I) {
let iter = iter.into_iter();
self.reserve(iter.size_hint().0);
for u8c in iter {
// twice as fast as self.extend_from_slice(u8c.as_bytes());
self.push(u8c.bytes[0]);
for &extra in &u8c.bytes[1..] {
if extra != 0 {
self.push(extra);
}
}
}
}
}
#[cfg(feature="std")]
impl<'a> Extend<&'a Utf8Char> for Vec<u8> {
fn extend<I:IntoIterator<Item=&'a Utf8Char>>(&mut self, iter: I) {
self.extend(iter.into_iter().cloned())
}
}
#[cfg(feature="std")]
impl Extend<Utf8Char> for String {
fn extend<I:IntoIterator<Item=Utf8Char>>(&mut self, iter: I) {
unsafe { self.as_mut_vec().extend(iter) }
}
}
#[cfg(feature="std")]
impl<'a> Extend<&'a Utf8Char> for String {
fn extend<I:IntoIterator<Item=&'a Utf8Char>>(&mut self, iter: I) {
self.extend(iter.into_iter().cloned())
}
}
#[cfg(feature="std")]
impl FromIterator<Utf8Char> for String {
fn from_iter<I:IntoIterator<Item=Utf8Char>>(iter: I) -> String {
let mut string = String::new();
string.extend(iter);
return string;
}
}
#[cfg(feature="std")]
impl<'a> FromIterator<&'a Utf8Char> for String {
fn from_iter<I:IntoIterator<Item=&'a Utf8Char>>(iter: I) -> String {
iter.into_iter().cloned().collect()
}
}
#[cfg(feature="std")]
impl FromIterator<Utf8Char> for Vec<u8> {
fn from_iter<I:IntoIterator<Item=Utf8Char>>(iter: I) -> Self {
iter.into_iter().collect::<String>().into_bytes()
}
}
#[cfg(feature="std")]
impl<'a> FromIterator<&'a Utf8Char> for Vec<u8> {
fn from_iter<I:IntoIterator<Item=&'a Utf8Char>>(iter: I) -> Self {
iter.into_iter().cloned().collect::<String>().into_bytes()
}
}
/////////////////
//getter traits//
/////////////////
impl AsRef<[u8]> for Utf8Char {
fn as_ref(&self) -> &[u8] {
&self.bytes[..self.len()]
}
}
impl AsRef<str> for Utf8Char {
fn as_ref(&self) -> &str {
unsafe{ str::from_utf8_unchecked( self.as_ref() ) }
}
}
impl Borrow<[u8]> for Utf8Char {
fn borrow(&self) -> &[u8] {
self.as_ref()
}
}
impl Borrow<str> for Utf8Char {
fn borrow(&self) -> &str {
self.as_ref()
}
}
impl Deref for Utf8Char {
type Target = str;
fn deref(&self) -> &Self::Target {
self.as_ref()
}
}
////////////////
//ascii traits//
////////////////
#[cfg(feature="std")]
#[allow(deprecated)]
impl AsciiExt for Utf8Char {
type Owned = Utf8Char;
fn is_ascii(&self) -> bool {
self.bytes[0].is_ascii()
}
fn eq_ignore_ascii_case(&self, other: &Self) -> bool {
if self.is_ascii() {self.bytes[0].eq_ignore_ascii_case(&other.bytes[0])}
else {self == other}
}
fn to_ascii_uppercase(&self) -> Self::Owned {
let mut uc = *self;
uc.make_ascii_uppercase();
uc
}
fn to_ascii_lowercase(&self) -> Self::Owned {
let mut uc = *self;
uc.make_ascii_lowercase();
uc
}
fn make_ascii_uppercase(&mut self) {
self.bytes[0].make_ascii_uppercase()
}
fn make_ascii_lowercase(&mut self) {
self.bytes[0].make_ascii_lowercase();
}
}
#[cfg(feature="ascii")]
/// Requires the feature "ascii".
impl From<AsciiChar> for Utf8Char {
fn from(ac: AsciiChar) -> Self {
Utf8Char{ bytes: [ac.as_byte(),0,0,0] }
}
}
#[cfg(feature="ascii")]
/// Requires the feature "ascii".
impl ToAsciiChar for Utf8Char {
fn to_ascii_char(self) -> Result<AsciiChar, ToAsciiCharError> {
self.bytes[0].to_ascii_char()
}
unsafe fn to_ascii_char_unchecked(self) -> AsciiChar {
self.bytes[0].to_ascii_char_unchecked()
}
}
/////////////////////////////////////////////////////////
//Genaral traits that cannot be derived to emulate char//
/////////////////////////////////////////////////////////
impl hash::Hash for Utf8Char {
fn hash<H : hash::Hasher>(&self, state: &mut H) {
self.to_char().hash(state);
}
}
impl fmt::Debug for Utf8Char {
fn fmt(&self, fmtr: &mut fmt::Formatter) -> fmt::Result {
fmt::Debug::fmt(&self.to_char(), fmtr)
}
}
impl fmt::Display for Utf8Char {
fn fmt(&self, fmtr: &mut fmt::Formatter) -> fmt::Result {
fmtr.write_str(self.as_str())
}
}
////////////////////////////////
//Comparisons with other types//
////////////////////////////////
impl PartialEq<char> for Utf8Char {
fn eq(&self, u32c: &char) -> bool {
*self == Utf8Char::from(*u32c)
}
}
impl PartialEq<Utf8Char> for char {
fn eq(&self, u8c: &Utf8Char) -> bool {
Utf8Char::from(*self) == *u8c
}
}
impl PartialOrd<char> for Utf8Char {
fn partial_cmp(&self, u32c: &char) -> Option<Ordering> {
self.partial_cmp(&Self::from(*u32c))
}
}
impl PartialOrd<Utf8Char> for char {
fn partial_cmp(&self, u8c: &Utf8Char) -> Option<Ordering> {
Utf8Char::from(*self).partial_cmp(u8c)
}
}
impl PartialEq<Utf16Char> for Utf8Char {
fn eq(&self, u16c: &Utf16Char) -> bool {
*self == Self::from(*u16c)
}
}
impl PartialOrd<Utf16Char> for Utf8Char {
fn partial_cmp(&self, u16c: &Utf16Char) -> Option<Ordering> {
self.partial_cmp(&Self::from(*u16c))
}
}
// The other direction is implemented in utf16_char.rs
/// Only considers the byte equal if both it and the `Utf8Char` represents ASCII characters.
///
/// There is no impl in the opposite direction, as this should only be used to
/// compare `Utf8Char`s against constants.
///
/// # Examples
///
/// ```
/// # use encode_unicode::Utf8Char;
/// assert!(Utf8Char::from('8') == b'8');
/// assert!(Utf8Char::from_array([0xf1,0x80,0x80,0x80]).unwrap() != 0xf1);
/// assert!(Utf8Char::from('\u{ff}') != 0xff);
/// assert!(Utf8Char::from('\u{80}') != 0x80);
/// ```
impl PartialEq<u8> for Utf8Char {
fn eq(&self, byte: &u8) -> bool {
self.bytes[0] == *byte && self.bytes[1] == 0
}
}
#[cfg(feature = "ascii")]
/// `Utf8Char`s that are not ASCII never compare equal.
impl PartialEq<AsciiChar> for Utf8Char {
#[inline]
fn eq(&self, ascii: &AsciiChar) -> bool {
self.bytes[0] == *ascii as u8
}
}
#[cfg(feature = "ascii")]
/// `Utf8Char`s that are not ASCII never compare equal.
impl PartialEq<Utf8Char> for AsciiChar {
#[inline]
fn eq(&self, u8c: &Utf8Char) -> bool {
u8c == self
}
}
#[cfg(feature = "ascii")]
/// `Utf8Char`s that are not ASCII always compare greater.
impl PartialOrd<AsciiChar> for Utf8Char {
#[inline]
fn partial_cmp(&self, ascii: &AsciiChar) -> Option<Ordering> {
self.bytes[0].partial_cmp(ascii)
}
}
#[cfg(feature = "ascii")]
/// `Utf8Char`s that are not ASCII always compare greater.
impl PartialOrd<Utf8Char> for AsciiChar {
#[inline]
fn partial_cmp(&self, u8c: &Utf8Char) -> Option<Ordering> {
self.partial_cmp(&u8c.bytes[0])
}
}
///////////////////////////////////////////////////////
//pub impls that should be together for nicer rustdoc//
///////////////////////////////////////////////////////
impl Utf8Char {
/// Create an `Utf8Char` from the first codepoint in a `str`.
///
/// Returns an error if the `str` is empty.
///
/// # Examples
///
/// ```
/// use encode_unicode::Utf8Char;
///
/// assert_eq!(Utf8Char::from_str_start("a"), Ok((Utf8Char::from('a'),1)));
/// assert_eq!(Utf8Char::from_str_start("ab"), Ok((Utf8Char::from('a'),1)));
/// assert_eq!(Utf8Char::from_str_start("šŸ‚  "), Ok((Utf8Char::from('šŸ‚ '),4)));
/// assert_eq!(Utf8Char::from_str_start("eĢ"), Ok((Utf8Char::from('e'),1)));// 'e'+u301 combining mark
/// assert!(Utf8Char::from_str_start("").is_err());
/// ```
pub fn from_str_start(src: &str) -> Result<(Self,usize),EmptyStrError> {
unsafe {
if src.is_empty() {
Err(EmptyStrError)
} else {
Ok(Utf8Char::from_slice_start_unchecked(src.as_bytes()))
}
}
}
/// Create an `Utf8Char` of the first codepoint in an UTF-8 slice.
/// Also returns the length of the UTF-8 sequence for the codepoint.
///
/// If the slice is from a `str`, use `::from_str_start()` to skip UTF-8 validation.
///
/// # Errors
///
/// Returns an `Err` if the slice is empty, doesn't start with a valid
/// UTF-8 sequence or is too short for the sequence.
///
/// # Examples
///
/// ```
/// use encode_unicode::Utf8Char;
/// use encode_unicode::error::InvalidUtf8Slice::*;
/// use encode_unicode::error::InvalidUtf8::*;
///
/// assert_eq!(Utf8Char::from_slice_start(&[b'A', b'B', b'C']), Ok((Utf8Char::from('A'),1)));
/// assert_eq!(Utf8Char::from_slice_start(&[0xdd, 0xbb]), Ok((Utf8Char::from('\u{77b}'),2)));
///
/// assert_eq!(Utf8Char::from_slice_start(&[]), Err(TooShort(1)));
/// assert_eq!(Utf8Char::from_slice_start(&[0xf0, 0x99]), Err(TooShort(4)));
/// assert_eq!(Utf8Char::from_slice_start(&[0xee, b'F', 0x80]), Err(Utf8(NotAContinuationByte(1))));
/// assert_eq!(Utf8Char::from_slice_start(&[0xee, 0x99, 0x0f]), Err(Utf8(NotAContinuationByte(2))));
/// ```
pub fn from_slice_start(src: &[u8]) -> Result<(Self,usize),InvalidUtf8Slice> {
char::from_utf8_slice_start(src).map(|(_,len)| {
let mut bytes = [0; 4];
bytes[..len].copy_from_slice(&src[..len]);
(Utf8Char{ bytes: bytes }, len)
})
}
/// A `from_slice_start()` that doesn't validate the codepoint.
///
/// # Safety
///
/// The slice must be non-empty and start with a valid UTF-8 codepoint.
/// Invalid or incomplete values might cause reads of uninitalized memory.
pub unsafe fn from_slice_start_unchecked(src: &[u8]) -> (Self,usize) {
let len = 1+src.get_unchecked(0).extra_utf8_bytes_unchecked();
let mut bytes = [0; 4];
ptr::copy_nonoverlapping(src.as_ptr(), &mut bytes[0] as *mut u8, len);
(Utf8Char{ bytes: bytes }, len)
}
/// Create an `Utf8Char` from a byte array after validating it.
///
/// The codepoint must start at the first byte.
/// Unused bytes are set to zero by this function and so can be anything.
///
/// # Errors
///
/// Returns an `Err` if the array doesn't start with a valid UTF-8 sequence.
///
/// # Examples
///
/// ```
/// use encode_unicode::Utf8Char;
/// use encode_unicode::error::InvalidUtf8Array::*;
/// use encode_unicode::error::InvalidUtf8::*;
/// use encode_unicode::error::InvalidCodepoint::*;
///
/// assert_eq!(Utf8Char::from_array([b'A', 0, 0, 0]), Ok(Utf8Char::from('A')));
/// assert_eq!(Utf8Char::from_array([0xf4, 0x8b, 0xbb, 0xbb]), Ok(Utf8Char::from('\u{10befb}')));
/// assert_eq!(Utf8Char::from_array([b'A', b'B', b'C', b'D']), Ok(Utf8Char::from('A')));
/// assert_eq!(Utf8Char::from_array([0, 0, 0xcc, 0xbb]), Ok(Utf8Char::from('\0')));
///
/// assert_eq!(Utf8Char::from_array([0xef, b'F', 0x80, 0x80]), Err(Utf8(NotAContinuationByte(1))));
/// assert_eq!(Utf8Char::from_array([0xc1, 0x80, 0, 0]), Err(Utf8(OverLong)));
/// assert_eq!(Utf8Char::from_array([0xf7, 0xaa, 0x99, 0x88]), Err(Codepoint(TooHigh)));
/// ```
pub fn from_array(utf8: [u8;4]) -> Result<Self,InvalidUtf8Array> {
unsafe {
// perform all validation
try!(char::from_utf8_array(utf8));
let extra = utf8[0].extra_utf8_bytes_unchecked() as u32;
// zero unused bytes in one operation by transmuting the arrary to
// u32, apply an endian-corrected mask and transmute back
let mask = u32::from_le(0xff_ff_ff_ff >> 8*(3-extra));
let unused_zeroed = mask & transmute::<_,u32>(utf8);
Ok(Utf8Char{ bytes: transmute(unused_zeroed) })
}
}
/// Zero-cost constructor.
///
/// # Safety
///
/// Must represent a valid codepoint,
/// starting at the first byte with the unused zeroed.
#[inline]
pub unsafe fn from_array_unchecked(utf8: [u8;4]) -> Self {
Utf8Char{ bytes: utf8 }
}
/// Create an `Utf8Char` from a single byte.
///
/// The byte must be an ASCII character.
///
/// # Errors
///
/// Returns `NonAsciiError` if the byte greater than 127.
///
/// # Examples
///
/// ```
/// # use encode_unicode::Utf8Char;
/// assert_eq!(Utf8Char::from_ascii(b'a').unwrap(), 'a');
/// assert!(Utf8Char::from_ascii(128).is_err());
/// ```
pub fn from_ascii(ascii: u8) -> Result<Self,NonAsciiError> {
if ascii as i8 >= 0 {
Ok(Utf8Char{ bytes: [ascii, 0, 0, 0] })
} else {
Err(NonAsciiError)
}
}
/// Create an `Utf8Char` from a single byte without checking that it's a
/// valid codepoint on its own, which is only true for ASCII characters.
///
/// # Safety
///
/// The byte must be less than 128.
#[inline]
pub unsafe fn from_ascii_unchecked(ascii: u8) -> Self {
Utf8Char{ bytes: [ascii, 0, 0, 0] }
}
/// The number of bytes this character needs.
///
/// Is between 1 and 4 (inclusive) and identical to `.as_ref().len()` or
/// `.as_char().len_utf8()`.
#[inline]
pub fn len(self) -> usize {
// Invariants of the extra bytes enambles algorithms that
// `u8.extra_utf8_bytes_unchecked()` cannot use.
// Some of them turned out to require fewer x86 instructions:
// Exploits that unused bytes are zero and calculates the number of
// trailing zero bytes.
// Setting a bit in the first byte prevents the function from returning
// 0 for '\0' (which has 32 leading zeros).
// trailing and leading is swapped below to optimize for little-endian
// architectures.
(4 - (u32::to_le(unsafe{transmute(self.bytes)})|1).leading_zeros()/8) as usize
// Exploits that the extra bytes have their most significant bit set if
// in use.
// Takes fewer instructions than the one above if popcnt can be used,
// (which it cannot by default,
// set RUSTFLAGS='-C target-cpu=native' to enable)
//let all: u32 = unsafe{transmute(self.bytes)};
//let msb_mask = u32::from_be(0x00808080);
//let add_one = u32::from_be(0x80000000);
//((all & msb_mask) | add_one).count_ones() as usize
}
// There is no .is_emty() because this type is never empty.
/// Checks that the codepoint is an ASCII character.
pub fn is_ascii(&self) -> bool {
self.bytes[0] <= 127
}
/// Checks that two characters are an ASCII case-insensitive match.
///
/// Is equivalent to `a.to_ascii_lowercase() == b.to_ascii_lowercase()`.
#[cfg(feature="std")]
pub fn eq_ignore_ascii_case(&self, other: &Self) -> bool {
if self.is_ascii() {self.bytes[0].eq_ignore_ascii_case(&other.bytes[0])}
else {self == other}
}
/// Converts the character to its ASCII upper case equivalent.
///
/// ASCII letters 'a' to 'z' are mapped to 'A' to 'Z',
/// but non-ASCII letters are unchanged.
#[cfg(feature="std")]
pub fn to_ascii_uppercase(&self) -> Self {
let mut uc = *self;
uc.make_ascii_uppercase();
uc
}
/// Converts the character to its ASCII lower case equivalent.
///
/// ASCII letters 'A' to 'Z' are mapped to 'a' to 'z',
/// but non-ASCII letters are unchanged.
#[cfg(feature="std")]
pub fn to_ascii_lowercase(&self) -> Self {
let mut uc = *self;
uc.make_ascii_lowercase();
uc
}
/// Converts the character to its ASCII upper case equivalent in-place.
///
/// ASCII letters 'a' to 'z' are mapped to 'A' to 'Z',
/// but non-ASCII letters are unchanged.
#[inline]
#[cfg(feature="std")]
pub fn make_ascii_uppercase(&mut self) {
self.bytes[0].make_ascii_uppercase()
}
/// Converts the character to its ASCII lower case equivalent in-place.
///
/// ASCII letters 'A' to 'Z' are mapped to 'a' to 'z',
/// but non-ASCII letters are unchanged.
#[inline]
#[cfg(feature="std")]
pub fn make_ascii_lowercase(&mut self) {
self.bytes[0].make_ascii_lowercase();
}
/// Convert from UTF-8 to UTF-32
pub fn to_char(self) -> char {
self.into()
}
/// Write the internal representation to a slice,
/// and then returns the number of bytes written.
///
/// # Panics
///
/// Will panic the buffer is too small;
/// You can get the required length from `.len()`,
/// but a buffer of length four is always large enough.
pub fn to_slice(self, dst: &mut[u8]) -> usize {
if self.len() > dst.len() {
panic!("The provided buffer is too small.");
}
dst[..self.len()].copy_from_slice(&self.bytes[..self.len()]);
self.len()
}
/// Expose the internal array and the number of used bytes.
pub fn to_array(self) -> ([u8;4],usize) {
(self.bytes, self.len())
}
/// Return a `str` view of the array the codepoint is stored as.
///
/// Is an unambiguous version of `.as_ref()`.
pub fn as_str(&self) -> &str {
self.deref()
}
}