1 // Copyright 2015 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution and at
3 // http://rust-lang.org/COPYRIGHT.
5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
8 // option. This file may not be copied, modified, or distributed
9 // except according to those terms.
11 //! Implementation of [the WTF-8 encoding](https://simonsapin.github.io/wtf-8/).
13 //! This library uses Rust’s type system to maintain
14 //! [well-formedness](https://simonsapin.github.io/wtf-8/#well-formed),
15 //! like the `String` and `&str` types do for UTF-8.
17 //! Since [WTF-8 must not be used
18 //! for interchange](https://simonsapin.github.io/wtf-8/#intended-audience),
19 //! this library deliberately does not provide access to the underlying bytes
21 //! nor can it decode WTF-8 from arbitrary bytes.
22 //! WTF-8 strings can be obtained from UTF-8, UTF-16, or code points.
24 // this module is imported from @SimonSapin's repo and has tons of dead code on
25 // unix (it's mostly used on windows), so don't worry about dead code here.
28 use core
::str::next_code_point
;
34 use hash
::{Hash, Hasher}
;
35 use iter
::FromIterator
;
40 use sys_common
::AsInner
;
42 const UTF8_REPLACEMENT_CHARACTER
: &'
static [u8] = b
"\xEF\xBF\xBD";
44 /// A Unicode code point: from U+0000 to U+10FFFF.
46 /// Compare with the `char` type,
47 /// which represents a Unicode scalar value:
48 /// a code point that is not a surrogate (U+D800 to U+DFFF).
49 #[derive(Eq, PartialEq, Ord, PartialOrd, Clone, Copy)]
50 pub struct CodePoint
{
54 /// Format the code point as `U+` followed by four to six hexadecimal digits.
55 /// Example: `U+1F4A9`
56 impl fmt
::Debug
for CodePoint
{
58 fn fmt(&self, formatter
: &mut fmt
::Formatter
) -> Result
<(), fmt
::Error
> {
59 write
!(formatter
, "U+{:04X}", self.value
)
64 /// Unsafely creates a new `CodePoint` without checking the value.
66 /// Only use when `value` is known to be less than or equal to 0x10FFFF.
68 pub unsafe fn from_u32_unchecked(value
: u32) -> CodePoint
{
69 CodePoint { value: value }
72 /// Creates a new `CodePoint` if the value is a valid code point.
74 /// Returns `None` if `value` is above 0x10FFFF.
76 pub fn from_u32(value
: u32) -> Option
<CodePoint
> {
78 0 ... 0x10FFFF => Some(CodePoint { value: value }
),
83 /// Creates a new `CodePoint` from a `char`.
85 /// Since all Unicode scalar values are code points, this always succeeds.
87 pub fn from_char(value
: char) -> CodePoint
{
88 CodePoint { value: value as u32 }
91 /// Returns the numeric value of the code point.
93 pub fn to_u32(&self) -> u32 {
97 /// Optionally returns a Unicode scalar value for the code point.
99 /// Returns `None` if the code point is a surrogate (from U+D800 to U+DFFF).
101 pub fn to_char(&self) -> Option
<char> {
103 0xD800 ... 0xDFFF => None
,
104 _
=> Some(unsafe { char::from_u32_unchecked(self.value) }
)
108 /// Returns a Unicode scalar value for the code point.
110 /// Returns `'\u{FFFD}'` (the replacement character “�”)
111 /// if the code point is a surrogate (from U+D800 to U+DFFF).
113 pub fn to_char_lossy(&self) -> char {
114 self.to_char().unwrap_or('
\u{FFFD}'
)
118 /// An owned, growable string of well-formed WTF-8 data.
120 /// Similar to `String`, but can additionally contain surrogate code points
121 /// if they’re not in a surrogate pair.
122 #[derive(Eq, PartialEq, Ord, PartialOrd, Clone)]
127 impl ops
::Deref
for Wtf8Buf
{
130 fn deref(&self) -> &Wtf8
{
135 /// Format the string with double quotes,
136 /// and surrogates as `\u` followed by four hexadecimal digits.
137 /// Example: `"a\u{D800}"` for a string with code points [U+0061, U+D800]
138 impl fmt
::Debug
for Wtf8Buf
{
140 fn fmt(&self, formatter
: &mut fmt
::Formatter
) -> Result
<(), fmt
::Error
> {
141 fmt
::Debug
::fmt(&**self, formatter
)
146 /// Creates a new, empty WTF-8 string.
148 pub fn new() -> Wtf8Buf
{
149 Wtf8Buf { bytes: Vec::new() }
152 /// Creates a new, empty WTF-8 string with pre-allocated capacity for `n` bytes.
154 pub fn with_capacity(n
: usize) -> Wtf8Buf
{
155 Wtf8Buf { bytes: Vec::with_capacity(n) }
158 /// Creates a WTF-8 string from a UTF-8 `String`.
160 /// This takes ownership of the `String` and does not copy.
162 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
164 pub fn from_string(string
: String
) -> Wtf8Buf
{
165 Wtf8Buf { bytes: string.into_bytes() }
168 /// Creates a WTF-8 string from a UTF-8 `&str` slice.
170 /// This copies the content of the slice.
172 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
174 pub fn from_str(str: &str) -> Wtf8Buf
{
175 Wtf8Buf { bytes: <[_]>::to_vec(str.as_bytes()) }
178 pub fn clear(&mut self) {
182 /// Creates a WTF-8 string from a potentially ill-formed UTF-16 slice of 16-bit code units.
184 /// This is lossless: calling `.encode_wide()` on the resulting string
185 /// will always return the original code units.
186 pub fn from_wide(v
: &[u16]) -> Wtf8Buf
{
187 let mut string
= Wtf8Buf
::with_capacity(v
.len());
188 for item
in char::decode_utf16(v
.iter().cloned()) {
190 Ok(ch
) => string
.push_char(ch
),
192 let surrogate
= surrogate
.unpaired_surrogate();
193 // Surrogates are known to be in the code point range.
194 let code_point
= unsafe {
195 CodePoint
::from_u32_unchecked(surrogate
as u32)
197 // Skip the WTF-8 concatenation check,
198 // surrogate pairs are already decoded by decode_utf16
199 string
.push_code_point_unchecked(code_point
)
206 /// Copied from String::push
207 /// This does **not** include the WTF-8 concatenation check.
208 fn push_code_point_unchecked(&mut self, code_point
: CodePoint
) {
210 char::from_u32_unchecked(code_point
.value
)
212 let mut bytes
= [0; 4];
213 let bytes
= c
.encode_utf8(&mut bytes
).as_bytes();
214 self.bytes
.extend_from_slice(bytes
)
218 pub fn as_slice(&self) -> &Wtf8
{
219 unsafe { Wtf8::from_bytes_unchecked(&self.bytes) }
222 /// Reserves capacity for at least `additional` more bytes to be inserted
223 /// in the given `Wtf8Buf`.
224 /// The collection may reserve more space to avoid frequent reallocations.
228 /// Panics if the new capacity overflows `usize`.
230 pub fn reserve(&mut self, additional
: usize) {
231 self.bytes
.reserve(additional
)
235 pub fn reserve_exact(&mut self, additional
: usize) {
236 self.bytes
.reserve_exact(additional
)
240 pub fn shrink_to_fit(&mut self) {
241 self.bytes
.shrink_to_fit()
244 /// Returns the number of bytes that this string buffer can hold without reallocating.
246 pub fn capacity(&self) -> usize {
247 self.bytes
.capacity()
250 /// Append a UTF-8 slice at the end of the string.
252 pub fn push_str(&mut self, other
: &str) {
253 self.bytes
.extend_from_slice(other
.as_bytes())
256 /// Append a WTF-8 slice at the end of the string.
258 /// This replaces newly paired surrogates at the boundary
259 /// with a supplementary code point,
260 /// like concatenating ill-formed UTF-16 strings effectively would.
262 pub fn push_wtf8(&mut self, other
: &Wtf8
) {
263 match ((&*self).final_lead_surrogate(), other
.initial_trail_surrogate()) {
264 // Replace newly paired surrogates by a supplementary code point.
265 (Some(lead
), Some(trail
)) => {
266 let len_without_lead_surrogate
= self.len() - 3;
267 self.bytes
.truncate(len_without_lead_surrogate
);
268 let other_without_trail_surrogate
= &other
.bytes
[3..];
269 // 4 bytes for the supplementary code point
270 self.bytes
.reserve(4 + other_without_trail_surrogate
.len());
271 self.push_char(decode_surrogate_pair(lead
, trail
));
272 self.bytes
.extend_from_slice(other_without_trail_surrogate
);
274 _
=> self.bytes
.extend_from_slice(&other
.bytes
)
278 /// Append a Unicode scalar value at the end of the string.
280 pub fn push_char(&mut self, c
: char) {
281 self.push_code_point_unchecked(CodePoint
::from_char(c
))
284 /// Append a code point at the end of the string.
286 /// This replaces newly paired surrogates at the boundary
287 /// with a supplementary code point,
288 /// like concatenating ill-formed UTF-16 strings effectively would.
290 pub fn push(&mut self, code_point
: CodePoint
) {
291 if let trail @
0xDC00...0xDFFF = code_point
.to_u32() {
292 if let Some(lead
) = (&*self).final_lead_surrogate() {
293 let len_without_lead_surrogate
= self.len() - 3;
294 self.bytes
.truncate(len_without_lead_surrogate
);
295 self.push_char(decode_surrogate_pair(lead
, trail
as u16));
300 // No newly paired surrogates at the boundary.
301 self.push_code_point_unchecked(code_point
)
304 /// Shortens a string to the specified length.
308 /// Panics if `new_len` > current length,
309 /// or if `new_len` is not a code point boundary.
311 pub fn truncate(&mut self, new_len
: usize) {
312 assert
!(is_code_point_boundary(self, new_len
));
313 self.bytes
.truncate(new_len
)
316 /// Consumes the WTF-8 string and tries to convert it to UTF-8.
318 /// This does not copy the data.
320 /// If the contents are not well-formed UTF-8
321 /// (that is, if the string contains surrogates),
322 /// the original WTF-8 string is returned instead.
323 pub fn into_string(self) -> Result
<String
, Wtf8Buf
> {
324 match self.next_surrogate(0) {
325 None
=> Ok(unsafe { String::from_utf8_unchecked(self.bytes) }
),
326 Some(_
) => Err(self),
330 /// Consumes the WTF-8 string and converts it lossily to UTF-8.
332 /// This does not copy the data (but may overwrite parts of it in place).
334 /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”)
335 pub fn into_string_lossy(mut self) -> String
{
338 match self.next_surrogate(pos
) {
339 Some((surrogate_pos
, _
)) => {
340 pos
= surrogate_pos
+ 3;
341 self.bytes
[surrogate_pos
..pos
]
342 .copy_from_slice(UTF8_REPLACEMENT_CHARACTER
);
344 None
=> return unsafe { String::from_utf8_unchecked(self.bytes) }
349 /// Converts this `Wtf8Buf` into a boxed `Wtf8`.
351 pub fn into_box(self) -> Box
<Wtf8
> {
352 unsafe { mem::transmute(self.bytes.into_boxed_slice()) }
355 /// Converts a `Box<Wtf8>` into a `Wtf8Buf`.
356 pub fn from_box(boxed
: Box
<Wtf8
>) -> Wtf8Buf
{
357 let bytes
: Box
<[u8]> = unsafe { mem::transmute(boxed) }
;
358 Wtf8Buf { bytes: bytes.into_vec() }
362 /// Create a new WTF-8 string from an iterator of code points.
364 /// This replaces surrogate code point pairs with supplementary code points,
365 /// like concatenating ill-formed UTF-16 strings effectively would.
366 impl FromIterator
<CodePoint
> for Wtf8Buf
{
367 fn from_iter
<T
: IntoIterator
<Item
=CodePoint
>>(iter
: T
) -> Wtf8Buf
{
368 let mut string
= Wtf8Buf
::new();
374 /// Append code points from an iterator to the string.
376 /// This replaces surrogate code point pairs with supplementary code points,
377 /// like concatenating ill-formed UTF-16 strings effectively would.
378 impl Extend
<CodePoint
> for Wtf8Buf
{
379 fn extend
<T
: IntoIterator
<Item
=CodePoint
>>(&mut self, iter
: T
) {
380 let iterator
= iter
.into_iter();
381 let (low
, _high
) = iterator
.size_hint();
382 // Lower bound of one byte per code point (ASCII only)
383 self.bytes
.reserve(low
);
384 for code_point
in iterator
{
385 self.push(code_point
);
390 /// A borrowed slice of well-formed WTF-8 data.
392 /// Similar to `&str`, but can additionally contain surrogate code points
393 /// if they’re not in a surrogate pair.
394 #[derive(Eq, Ord, PartialEq, PartialOrd)]
399 impl AsInner
<[u8]> for Wtf8
{
400 fn as_inner(&self) -> &[u8] { &self.bytes }
403 /// Format the slice with double quotes,
404 /// and surrogates as `\u` followed by four hexadecimal digits.
405 /// Example: `"a\u{D800}"` for a slice with code points [U+0061, U+D800]
406 impl fmt
::Debug
for Wtf8
{
407 fn fmt(&self, formatter
: &mut fmt
::Formatter
) -> fmt
::Result
{
408 fn write_str_escaped(f
: &mut fmt
::Formatter
, s
: &str) -> fmt
::Result
{
410 for c
in s
.chars().flat_map(|c
| c
.escape_debug()) {
416 formatter
.write_str("\"")?
;
419 match self.next_surrogate(pos
) {
421 Some((surrogate_pos
, surrogate
)) => {
424 unsafe { str::from_utf8_unchecked(
425 &self.bytes
[pos
.. surrogate_pos
]
428 write
!(formatter
, "\\u{{{:x}}}", surrogate
)?
;
429 pos
= surrogate_pos
+ 3;
435 unsafe { str::from_utf8_unchecked(&self.bytes[pos..]) }
,
437 formatter
.write_str("\"")
442 /// Creates a WTF-8 slice from a UTF-8 `&str` slice.
444 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
446 pub fn from_str(value
: &str) -> &Wtf8
{
447 unsafe { Wtf8::from_bytes_unchecked(value.as_bytes()) }
450 /// Creates a WTF-8 slice from a WTF-8 byte slice.
452 /// Since the byte slice is not checked for valid WTF-8, this functions is
455 unsafe fn from_bytes_unchecked(value
: &[u8]) -> &Wtf8
{
456 mem
::transmute(value
)
459 /// Returns the length, in WTF-8 bytes.
461 pub fn len(&self) -> usize {
466 pub fn is_empty(&self) -> bool
{
467 self.bytes
.is_empty()
470 /// Returns the code point at `position` if it is in the ASCII range,
471 /// or `b'\xFF' otherwise.
475 /// Panics if `position` is beyond the end of the string.
477 pub fn ascii_byte_at(&self, position
: usize) -> u8 {
478 match self.bytes
[position
] {
479 ascii_byte @
0x00 ... 0x7F => ascii_byte
,
484 /// Returns an iterator for the string’s code points.
486 pub fn code_points(&self) -> Wtf8CodePoints
{
487 Wtf8CodePoints { bytes: self.bytes.iter() }
490 /// Tries to convert the string to UTF-8 and return a `&str` slice.
492 /// Returns `None` if the string contains surrogates.
494 /// This does not copy the data.
496 pub fn as_str(&self) -> Option
<&str> {
497 // Well-formed WTF-8 is also well-formed UTF-8
498 // if and only if it contains no surrogate.
499 match self.next_surrogate(0) {
500 None
=> Some(unsafe { str::from_utf8_unchecked(&self.bytes) }
),
505 /// Lossily converts the string to UTF-8.
506 /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8.
508 /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”).
510 /// This only copies the data if necessary (if it contains any surrogate).
511 pub fn to_string_lossy(&self) -> Cow
<str> {
512 let surrogate_pos
= match self.next_surrogate(0) {
513 None
=> return Cow
::Borrowed(unsafe { str::from_utf8_unchecked(&self.bytes) }
),
514 Some((pos
, _
)) => pos
,
516 let wtf8_bytes
= &self.bytes
;
517 let mut utf8_bytes
= Vec
::with_capacity(self.len());
518 utf8_bytes
.extend_from_slice(&wtf8_bytes
[..surrogate_pos
]);
519 utf8_bytes
.extend_from_slice(UTF8_REPLACEMENT_CHARACTER
);
520 let mut pos
= surrogate_pos
+ 3;
522 match self.next_surrogate(pos
) {
523 Some((surrogate_pos
, _
)) => {
524 utf8_bytes
.extend_from_slice(&wtf8_bytes
[pos
.. surrogate_pos
]);
525 utf8_bytes
.extend_from_slice(UTF8_REPLACEMENT_CHARACTER
);
526 pos
= surrogate_pos
+ 3;
529 utf8_bytes
.extend_from_slice(&wtf8_bytes
[pos
..]);
530 return Cow
::Owned(unsafe { String::from_utf8_unchecked(utf8_bytes) }
)
536 /// Converts the WTF-8 string to potentially ill-formed UTF-16
537 /// and return an iterator of 16-bit code units.
539 /// This is lossless:
540 /// calling `Wtf8Buf::from_ill_formed_utf16` on the resulting code units
541 /// would always return the original WTF-8 string.
543 pub fn encode_wide(&self) -> EncodeWide
{
544 EncodeWide { code_points: self.code_points(), extra: 0 }
548 fn next_surrogate(&self, mut pos
: usize) -> Option
<(usize, u16)> {
549 let mut iter
= self.bytes
[pos
..].iter();
551 let b
= match iter
.next() {
560 } else if b
== 0xED {
561 match (iter
.next(), iter
.next()) {
562 (Some(&b2
), Some(&b3
)) if b2
>= 0xA0 => {
563 return Some((pos
, decode_surrogate(b2
, b3
)))
581 fn final_lead_surrogate(&self) -> Option
<u16> {
582 let len
= self.len();
586 match &self.bytes
[(len
- 3)..] {
587 &[0xED, b2 @
0xA0...0xAF, b3
] => Some(decode_surrogate(b2
, b3
)),
593 fn initial_trail_surrogate(&self) -> Option
<u16> {
594 let len
= self.len();
598 match &self.bytes
[..3] {
599 &[0xED, b2 @
0xB0...0xBF, b3
] => Some(decode_surrogate(b2
, b3
)),
604 /// Boxes this `Wtf8`.
606 pub fn into_box(&self) -> Box
<Wtf8
> {
607 let boxed
: Box
<[u8]> = self.bytes
.into();
608 unsafe { mem::transmute(boxed) }
611 /// Creates a boxed, empty `Wtf8`.
612 pub fn empty_box() -> Box
<Wtf8
> {
613 let boxed
: Box
<[u8]> = Default
::default();
614 unsafe { mem::transmute(boxed) }
619 /// Return a slice of the given string for the byte range [`begin`..`end`).
623 /// Panics when `begin` and `end` do not point to code point boundaries,
624 /// or point beyond the end of the string.
625 impl ops
::Index
<ops
::Range
<usize>> for Wtf8
{
629 fn index(&self, range
: ops
::Range
<usize>) -> &Wtf8
{
630 // is_code_point_boundary checks that the index is in [0, .len()]
631 if range
.start
<= range
.end
&&
632 is_code_point_boundary(self, range
.start
) &&
633 is_code_point_boundary(self, range
.end
) {
634 unsafe { slice_unchecked(self, range.start, range.end) }
636 slice_error_fail(self, range
.start
, range
.end
)
641 /// Return a slice of the given string from byte `begin` to its end.
645 /// Panics when `begin` is not at a code point boundary,
646 /// or is beyond the end of the string.
647 impl ops
::Index
<ops
::RangeFrom
<usize>> for Wtf8
{
651 fn index(&self, range
: ops
::RangeFrom
<usize>) -> &Wtf8
{
652 // is_code_point_boundary checks that the index is in [0, .len()]
653 if is_code_point_boundary(self, range
.start
) {
654 unsafe { slice_unchecked(self, range.start, self.len()) }
656 slice_error_fail(self, range
.start
, self.len())
661 /// Return a slice of the given string from its beginning to byte `end`.
665 /// Panics when `end` is not at a code point boundary,
666 /// or is beyond the end of the string.
667 impl ops
::Index
<ops
::RangeTo
<usize>> for Wtf8
{
671 fn index(&self, range
: ops
::RangeTo
<usize>) -> &Wtf8
{
672 // is_code_point_boundary checks that the index is in [0, .len()]
673 if is_code_point_boundary(self, range
.end
) {
674 unsafe { slice_unchecked(self, 0, range.end) }
676 slice_error_fail(self, 0, range
.end
)
681 impl ops
::Index
<ops
::RangeFull
> for Wtf8
{
685 fn index(&self, _range
: ops
::RangeFull
) -> &Wtf8
{
691 fn decode_surrogate(second_byte
: u8, third_byte
: u8) -> u16 {
692 // The first byte is assumed to be 0xED
693 0xD800 | (second_byte
as u16 & 0x3F) << 6 | third_byte
as u16 & 0x3F
697 fn decode_surrogate_pair(lead
: u16, trail
: u16) -> char {
698 let code_point
= 0x10000 + ((((lead
- 0xD800) as u32) << 10) | (trail
- 0xDC00) as u32);
699 unsafe { char::from_u32_unchecked(code_point) }
702 /// Copied from core::str::StrPrelude::is_char_boundary
704 pub fn is_code_point_boundary(slice
: &Wtf8
, index
: usize) -> bool
{
705 if index
== slice
.len() { return true; }
706 match slice
.bytes
.get(index
) {
708 Some(&b
) => b
< 128 || b
>= 192,
712 /// Copied from core::str::raw::slice_unchecked
714 pub unsafe fn slice_unchecked(s
: &Wtf8
, begin
: usize, end
: usize) -> &Wtf8
{
715 // memory layout of an &[u8] and &Wtf8 are the same
716 Wtf8
::from_bytes_unchecked(slice
::from_raw_parts(
717 s
.bytes
.as_ptr().offset(begin
as isize),
722 /// Copied from core::str::raw::slice_error_fail
724 pub fn slice_error_fail(s
: &Wtf8
, begin
: usize, end
: usize) -> ! {
725 assert
!(begin
<= end
);
726 panic
!("index {} and/or {} in `{:?}` do not lie on character boundary",
730 /// Iterator for the code points of a WTF-8 string.
732 /// Created with the method `.code_points()`.
734 pub struct Wtf8CodePoints
<'a
> {
735 bytes
: slice
::Iter
<'a
, u8>
738 impl<'a
> Iterator
for Wtf8CodePoints
<'a
> {
739 type Item
= CodePoint
;
742 fn next(&mut self) -> Option
<CodePoint
> {
743 next_code_point(&mut self.bytes
).map(|c
| CodePoint { value: c }
)
747 fn size_hint(&self) -> (usize, Option
<usize>) {
748 let len
= self.bytes
.len();
749 (len
.saturating_add(3) / 4, Some(len
))
753 #[stable(feature = "rust1", since = "1.0.0")]
755 pub struct EncodeWide
<'a
> {
756 code_points
: Wtf8CodePoints
<'a
>,
760 // Copied from libunicode/u_str.rs
761 #[stable(feature = "rust1", since = "1.0.0")]
762 impl<'a
> Iterator
for EncodeWide
<'a
> {
766 fn next(&mut self) -> Option
<u16> {
768 let tmp
= self.extra
;
773 let mut buf
= [0; 2];
774 self.code_points
.next().map(|code_point
| {
776 char::from_u32_unchecked(code_point
.value
)
778 let n
= c
.encode_utf16(&mut buf
).len();
787 fn size_hint(&self) -> (usize, Option
<usize>) {
788 let (low
, high
) = self.code_points
.size_hint();
789 // every code point gets either one u16 or two u16,
790 // so this iterator is between 1 or 2 times as
791 // long as the underlying iterator.
792 (low
, high
.and_then(|n
| n
.checked_mul(2)))
796 impl Hash
for CodePoint
{
798 fn hash
<H
: Hasher
>(&self, state
: &mut H
) {
799 self.value
.hash(state
)
803 impl Hash
for Wtf8Buf
{
805 fn hash
<H
: Hasher
>(&self, state
: &mut H
) {
806 state
.write(&self.bytes
);
813 fn hash
<H
: Hasher
>(&self, state
: &mut H
) {
814 state
.write(&self.bytes
);
819 impl AsciiExt
for Wtf8
{
820 type Owned
= Wtf8Buf
;
822 fn is_ascii(&self) -> bool
{
823 self.bytes
.is_ascii()
825 fn to_ascii_uppercase(&self) -> Wtf8Buf
{
826 Wtf8Buf { bytes: self.bytes.to_ascii_uppercase() }
828 fn to_ascii_lowercase(&self) -> Wtf8Buf
{
829 Wtf8Buf { bytes: self.bytes.to_ascii_lowercase() }
831 fn eq_ignore_ascii_case(&self, other
: &Wtf8
) -> bool
{
832 self.bytes
.eq_ignore_ascii_case(&other
.bytes
)
835 fn make_ascii_uppercase(&mut self) { self.bytes.make_ascii_uppercase() }
836 fn make_ascii_lowercase(&mut self) { self.bytes.make_ascii_lowercase() }
845 fn code_point_from_u32() {
846 assert
!(CodePoint
::from_u32(0).is_some());
847 assert
!(CodePoint
::from_u32(0xD800).is_some());
848 assert
!(CodePoint
::from_u32(0x10FFFF).is_some());
849 assert
!(CodePoint
::from_u32(0x110000).is_none());
853 fn code_point_to_u32() {
854 fn c(value
: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
855 assert_eq
!(c(0).to_u32(), 0);
856 assert_eq
!(c(0xD800).to_u32(), 0xD800);
857 assert_eq
!(c(0x10FFFF).to_u32(), 0x10FFFF);
861 fn code_point_from_char() {
862 assert_eq
!(CodePoint
::from_char('a'
).to_u32(), 0x61);
863 assert_eq
!(CodePoint
::from_char('💩'
).to_u32(), 0x1F4A9);
867 fn code_point_to_string() {
868 assert_eq
!(format
!("{:?}", CodePoint
::from_char('a'
)), "U+0061");
869 assert_eq
!(format
!("{:?}", CodePoint
::from_char('💩'
)), "U+1F4A9");
873 fn code_point_to_char() {
874 fn c(value
: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
875 assert_eq
!(c(0x61).to_char(), Some('a'
));
876 assert_eq
!(c(0x1F4A9).to_char(), Some('💩'
));
877 assert_eq
!(c(0xD800).to_char(), None
);
881 fn code_point_to_char_lossy() {
882 fn c(value
: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
883 assert_eq
!(c(0x61).to_char_lossy(), 'a'
);
884 assert_eq
!(c(0x1F4A9).to_char_lossy(), '💩'
);
885 assert_eq
!(c(0xD800).to_char_lossy(), '
\u{FFFD}'
);
890 assert_eq
!(Wtf8Buf
::new().bytes
, b
"");
894 fn wtf8buf_from_str() {
895 assert_eq
!(Wtf8Buf
::from_str("").bytes
, b
"");
896 assert_eq
!(Wtf8Buf
::from_str("aé 💩").bytes
,
897 b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
901 fn wtf8buf_from_string() {
902 assert_eq
!(Wtf8Buf
::from_string(String
::from("")).bytes
, b
"");
903 assert_eq
!(Wtf8Buf
::from_string(String
::from("aé 💩")).bytes
,
904 b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
908 fn wtf8buf_from_wide() {
909 assert_eq
!(Wtf8Buf
::from_wide(&[]).bytes
, b
"");
910 assert_eq
!(Wtf8Buf
::from_wide(
911 &[0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]).bytes
,
912 b
"a\xC3\xA9 \xED\xA0\xBD\xF0\x9F\x92\xA9");
916 fn wtf8buf_push_str() {
917 let mut string
= Wtf8Buf
::new();
918 assert_eq
!(string
.bytes
, b
"");
919 string
.push_str("aé 💩");
920 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
924 fn wtf8buf_push_char() {
925 let mut string
= Wtf8Buf
::from_str("aé ");
926 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 ");
927 string
.push_char('💩'
);
928 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
933 let mut string
= Wtf8Buf
::from_str("aé ");
934 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 ");
935 string
.push(CodePoint
::from_char('💩'
));
936 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
938 fn c(value
: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
940 let mut string
= Wtf8Buf
::new();
941 string
.push(c(0xD83D)); // lead
942 string
.push(c(0xDCA9)); // trail
943 assert_eq
!(string
.bytes
, b
"\xF0\x9F\x92\xA9"); // Magic!
945 let mut string
= Wtf8Buf
::new();
946 string
.push(c(0xD83D)); // lead
947 string
.push(c(0x20)); // not surrogate
948 string
.push(c(0xDCA9)); // trail
949 assert_eq
!(string
.bytes
, b
"\xED\xA0\xBD \xED\xB2\xA9");
951 let mut string
= Wtf8Buf
::new();
952 string
.push(c(0xD800)); // lead
953 string
.push(c(0xDBFF)); // lead
954 assert_eq
!(string
.bytes
, b
"\xED\xA0\x80\xED\xAF\xBF");
956 let mut string
= Wtf8Buf
::new();
957 string
.push(c(0xD800)); // lead
958 string
.push(c(0xE000)); // not surrogate
959 assert_eq
!(string
.bytes
, b
"\xED\xA0\x80\xEE\x80\x80");
961 let mut string
= Wtf8Buf
::new();
962 string
.push(c(0xD7FF)); // not surrogate
963 string
.push(c(0xDC00)); // trail
964 assert_eq
!(string
.bytes
, b
"\xED\x9F\xBF\xED\xB0\x80");
966 let mut string
= Wtf8Buf
::new();
967 string
.push(c(0x61)); // not surrogate, < 3 bytes
968 string
.push(c(0xDC00)); // trail
969 assert_eq
!(string
.bytes
, b
"\x61\xED\xB0\x80");
971 let mut string
= Wtf8Buf
::new();
972 string
.push(c(0xDC00)); // trail
973 assert_eq
!(string
.bytes
, b
"\xED\xB0\x80");
977 fn wtf8buf_push_wtf8() {
978 let mut string
= Wtf8Buf
::from_str("aé");
979 assert_eq
!(string
.bytes
, b
"a\xC3\xA9");
980 string
.push_wtf8(Wtf8
::from_str(" 💩"));
981 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
983 fn w(v
: &[u8]) -> &Wtf8 { unsafe { Wtf8::from_bytes_unchecked(v) }
}
985 let mut string
= Wtf8Buf
::new();
986 string
.push_wtf8(w(b
"\xED\xA0\xBD")); // lead
987 string
.push_wtf8(w(b
"\xED\xB2\xA9")); // trail
988 assert_eq
!(string
.bytes
, b
"\xF0\x9F\x92\xA9"); // Magic!
990 let mut string
= Wtf8Buf
::new();
991 string
.push_wtf8(w(b
"\xED\xA0\xBD")); // lead
992 string
.push_wtf8(w(b
" ")); // not surrogate
993 string
.push_wtf8(w(b
"\xED\xB2\xA9")); // trail
994 assert_eq
!(string
.bytes
, b
"\xED\xA0\xBD \xED\xB2\xA9");
996 let mut string
= Wtf8Buf
::new();
997 string
.push_wtf8(w(b
"\xED\xA0\x80")); // lead
998 string
.push_wtf8(w(b
"\xED\xAF\xBF")); // lead
999 assert_eq
!(string
.bytes
, b
"\xED\xA0\x80\xED\xAF\xBF");
1001 let mut string
= Wtf8Buf
::new();
1002 string
.push_wtf8(w(b
"\xED\xA0\x80")); // lead
1003 string
.push_wtf8(w(b
"\xEE\x80\x80")); // not surrogate
1004 assert_eq
!(string
.bytes
, b
"\xED\xA0\x80\xEE\x80\x80");
1006 let mut string
= Wtf8Buf
::new();
1007 string
.push_wtf8(w(b
"\xED\x9F\xBF")); // not surrogate
1008 string
.push_wtf8(w(b
"\xED\xB0\x80")); // trail
1009 assert_eq
!(string
.bytes
, b
"\xED\x9F\xBF\xED\xB0\x80");
1011 let mut string
= Wtf8Buf
::new();
1012 string
.push_wtf8(w(b
"a")); // not surrogate, < 3 bytes
1013 string
.push_wtf8(w(b
"\xED\xB0\x80")); // trail
1014 assert_eq
!(string
.bytes
, b
"\x61\xED\xB0\x80");
1016 let mut string
= Wtf8Buf
::new();
1017 string
.push_wtf8(w(b
"\xED\xB0\x80")); // trail
1018 assert_eq
!(string
.bytes
, b
"\xED\xB0\x80");
1022 fn wtf8buf_truncate() {
1023 let mut string
= Wtf8Buf
::from_str("aé");
1025 assert_eq
!(string
.bytes
, b
"a");
1030 fn wtf8buf_truncate_fail_code_point_boundary() {
1031 let mut string
= Wtf8Buf
::from_str("aé");
1037 fn wtf8buf_truncate_fail_longer() {
1038 let mut string
= Wtf8Buf
::from_str("aé");
1043 fn wtf8buf_into_string() {
1044 let mut string
= Wtf8Buf
::from_str("aé 💩");
1045 assert_eq
!(string
.clone().into_string(), Ok(String
::from("aé 💩")));
1046 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1047 assert_eq
!(string
.clone().into_string(), Err(string
));
1051 fn wtf8buf_into_string_lossy() {
1052 let mut string
= Wtf8Buf
::from_str("aé 💩");
1053 assert_eq
!(string
.clone().into_string_lossy(), String
::from("aé 💩"));
1054 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1055 assert_eq
!(string
.clone().into_string_lossy(), String
::from("aé 💩�"));
1059 fn wtf8buf_from_iterator() {
1060 fn f(values
: &[u32]) -> Wtf8Buf
{
1061 values
.iter().map(|&c
| CodePoint
::from_u32(c
).unwrap()).collect
::<Wtf8Buf
>()
1063 assert_eq
!(f(&[0x61, 0xE9, 0x20, 0x1F4A9]).bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
1065 assert_eq
!(f(&[0xD83D, 0xDCA9]).bytes
, b
"\xF0\x9F\x92\xA9"); // Magic!
1066 assert_eq
!(f(&[0xD83D, 0x20, 0xDCA9]).bytes
, b
"\xED\xA0\xBD \xED\xB2\xA9");
1067 assert_eq
!(f(&[0xD800, 0xDBFF]).bytes
, b
"\xED\xA0\x80\xED\xAF\xBF");
1068 assert_eq
!(f(&[0xD800, 0xE000]).bytes
, b
"\xED\xA0\x80\xEE\x80\x80");
1069 assert_eq
!(f(&[0xD7FF, 0xDC00]).bytes
, b
"\xED\x9F\xBF\xED\xB0\x80");
1070 assert_eq
!(f(&[0x61, 0xDC00]).bytes
, b
"\x61\xED\xB0\x80");
1071 assert_eq
!(f(&[0xDC00]).bytes
, b
"\xED\xB0\x80");
1075 fn wtf8buf_extend() {
1076 fn e(initial
: &[u32], extended
: &[u32]) -> Wtf8Buf
{
1077 fn c(value
: &u32) -> CodePoint { CodePoint::from_u32(*value).unwrap() }
1078 let mut string
= initial
.iter().map(c
).collect
::<Wtf8Buf
>();
1079 string
.extend(extended
.iter().map(c
));
1083 assert_eq
!(e(&[0x61, 0xE9], &[0x20, 0x1F4A9]).bytes
,
1084 b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
1086 assert_eq
!(e(&[0xD83D], &[0xDCA9]).bytes
, b
"\xF0\x9F\x92\xA9"); // Magic!
1087 assert_eq
!(e(&[0xD83D, 0x20], &[0xDCA9]).bytes
, b
"\xED\xA0\xBD \xED\xB2\xA9");
1088 assert_eq
!(e(&[0xD800], &[0xDBFF]).bytes
, b
"\xED\xA0\x80\xED\xAF\xBF");
1089 assert_eq
!(e(&[0xD800], &[0xE000]).bytes
, b
"\xED\xA0\x80\xEE\x80\x80");
1090 assert_eq
!(e(&[0xD7FF], &[0xDC00]).bytes
, b
"\xED\x9F\xBF\xED\xB0\x80");
1091 assert_eq
!(e(&[0x61], &[0xDC00]).bytes
, b
"\x61\xED\xB0\x80");
1092 assert_eq
!(e(&[], &[0xDC00]).bytes
, b
"\xED\xB0\x80");
1097 let mut string
= Wtf8Buf
::from_str("a\té \u{7f}💩\r");
1098 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1099 assert_eq
!(format
!("{:?}", string
), "\"a\\té \\u{7f}\u{1f4a9}\\r\\u{d800}\"");
1103 fn wtf8buf_as_slice() {
1104 assert_eq
!(Wtf8Buf
::from_str("aé").as_slice(), Wtf8
::from_str("aé"));
1108 fn wtf8buf_show_str() {
1109 let text
= "a\té 💩\r";
1110 let string
= Wtf8Buf
::from_str(text
);
1111 assert_eq
!(format
!("{:?}", text
), format
!("{:?}", string
));
1115 fn wtf8_from_str() {
1116 assert_eq
!(&Wtf8
::from_str("").bytes
, b
"");
1117 assert_eq
!(&Wtf8
::from_str("aé 💩").bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
1122 assert_eq
!(Wtf8
::from_str("").len(), 0);
1123 assert_eq
!(Wtf8
::from_str("aé 💩").len(), 8);
1128 assert_eq
!(&Wtf8
::from_str("aé 💩")[1.. 4].bytes
, b
"\xC3\xA9 ");
1133 fn wtf8_slice_not_code_point_boundary() {
1134 &Wtf8
::from_str("aé 💩")[2.. 4];
1138 fn wtf8_slice_from() {
1139 assert_eq
!(&Wtf8
::from_str("aé 💩")[1..].bytes
, b
"\xC3\xA9 \xF0\x9F\x92\xA9");
1144 fn wtf8_slice_from_not_code_point_boundary() {
1145 &Wtf8
::from_str("aé 💩")[2..];
1149 fn wtf8_slice_to() {
1150 assert_eq
!(&Wtf8
::from_str("aé 💩")[..4].bytes
, b
"a\xC3\xA9 ");
1155 fn wtf8_slice_to_not_code_point_boundary() {
1156 &Wtf8
::from_str("aé 💩")[5..];
1160 fn wtf8_ascii_byte_at() {
1161 let slice
= Wtf8
::from_str("aé 💩");
1162 assert_eq
!(slice
.ascii_byte_at(0), b'a'
);
1163 assert_eq
!(slice
.ascii_byte_at(1), b'
\xFF'
);
1164 assert_eq
!(slice
.ascii_byte_at(2), b'
\xFF'
);
1165 assert_eq
!(slice
.ascii_byte_at(3), b' '
);
1166 assert_eq
!(slice
.ascii_byte_at(4), b'
\xFF'
);
1170 fn wtf8_code_points() {
1171 fn c(value
: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
1172 fn cp(string
: &Wtf8Buf
) -> Vec
<Option
<char>> {
1173 string
.code_points().map(|c
| c
.to_char()).collect
::<Vec
<_
>>()
1175 let mut string
= Wtf8Buf
::from_str("é ");
1176 assert_eq
!(cp(&string
), [Some('é'
), Some(' '
)]);
1177 string
.push(c(0xD83D));
1178 assert_eq
!(cp(&string
), [Some('é'
), Some(' '
), None
]);
1179 string
.push(c(0xDCA9));
1180 assert_eq
!(cp(&string
), [Some('é'
), Some(' '
), Some('💩'
)]);
1185 assert_eq
!(Wtf8
::from_str("").as_str(), Some(""));
1186 assert_eq
!(Wtf8
::from_str("aé 💩").as_str(), Some("aé 💩"));
1187 let mut string
= Wtf8Buf
::new();
1188 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1189 assert_eq
!(string
.as_str(), None
);
1193 fn wtf8_to_string_lossy() {
1194 assert_eq
!(Wtf8
::from_str("").to_string_lossy(), Cow
::Borrowed(""));
1195 assert_eq
!(Wtf8
::from_str("aé 💩").to_string_lossy(), Cow
::Borrowed("aé 💩"));
1196 let mut string
= Wtf8Buf
::from_str("aé 💩");
1197 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1198 let expected
: Cow
<str> = Cow
::Owned(String
::from("aé 💩�"));
1199 assert_eq
!(string
.to_string_lossy(), expected
);
1203 fn wtf8_encode_wide() {
1204 let mut string
= Wtf8Buf
::from_str("aé ");
1205 string
.push(CodePoint
::from_u32(0xD83D).unwrap());
1206 string
.push_char('💩'
);
1207 assert_eq
!(string
.encode_wide().collect
::<Vec
<_
>>(),
1208 vec
![0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]);