1 //! Implementation of [the WTF-8 encoding](https://simonsapin.github.io/wtf-8/).
3 //! This library uses Rust’s type system to maintain
4 //! [well-formedness](https://simonsapin.github.io/wtf-8/#well-formed),
5 //! like the `String` and `&str` types do for UTF-8.
7 //! Since [WTF-8 must not be used
8 //! for interchange](https://simonsapin.github.io/wtf-8/#intended-audience),
9 //! this library deliberately does not provide access to the underlying bytes
11 //! nor can it decode WTF-8 from arbitrary bytes.
12 //! WTF-8 strings can be obtained from UTF-8, UTF-16, or code points.
14 // this module is imported from @SimonSapin's repo and has tons of dead code on
15 // unix (it's mostly used on windows), so don't worry about dead code here.
18 use core
::str::next_code_point
;
20 use crate::borrow
::Cow
;
23 use crate::hash
::{Hash, Hasher}
;
24 use crate::iter
::FromIterator
;
31 use crate::sys_common
::AsInner
;
33 const UTF8_REPLACEMENT_CHARACTER
: &str = "\u{FFFD}";
35 /// A Unicode code point: from U+0000 to U+10FFFF.
37 /// Compares with the `char` type,
38 /// which represents a Unicode scalar value:
39 /// a code point that is not a surrogate (U+D800 to U+DFFF).
40 #[derive(Eq, PartialEq, Ord, PartialOrd, Clone, Copy)]
41 pub struct CodePoint
{
45 /// Format the code point as `U+` followed by four to six hexadecimal digits.
46 /// Example: `U+1F4A9`
47 impl fmt
::Debug
for CodePoint
{
49 fn fmt(&self, formatter
: &mut fmt
::Formatter
<'_
>) -> fmt
::Result
{
50 write
!(formatter
, "U+{:04X}", self.value
)
55 /// Unsafely creates a new `CodePoint` without checking the value.
57 /// Only use when `value` is known to be less than or equal to 0x10FFFF.
59 pub unsafe fn from_u32_unchecked(value
: u32) -> CodePoint
{
63 /// Creates a new `CodePoint` if the value is a valid code point.
65 /// Returns `None` if `value` is above 0x10FFFF.
67 pub fn from_u32(value
: u32) -> Option
<CodePoint
> {
69 0..=0x10FFFF => Some(CodePoint { value }
),
74 /// Creates a new `CodePoint` from a `char`.
76 /// Since all Unicode scalar values are code points, this always succeeds.
78 pub fn from_char(value
: char) -> CodePoint
{
79 CodePoint { value: value as u32 }
82 /// Returns the numeric value of the code point.
84 pub fn to_u32(&self) -> u32 {
88 /// Optionally returns a Unicode scalar value for the code point.
90 /// Returns `None` if the code point is a surrogate (from U+D800 to U+DFFF).
92 pub fn to_char(&self) -> Option
<char> {
94 0xD800..=0xDFFF => None
,
95 _
=> Some(unsafe { char::from_u32_unchecked(self.value) }
),
99 /// Returns a Unicode scalar value for the code point.
101 /// Returns `'\u{FFFD}'` (the replacement character “�”)
102 /// if the code point is a surrogate (from U+D800 to U+DFFF).
104 pub fn to_char_lossy(&self) -> char {
105 self.to_char().unwrap_or('
\u{FFFD}'
)
109 /// An owned, growable string of well-formed WTF-8 data.
111 /// Similar to `String`, but can additionally contain surrogate code points
112 /// if they’re not in a surrogate pair.
113 #[derive(Eq, PartialEq, Ord, PartialOrd, Clone)]
118 impl ops
::Deref
for Wtf8Buf
{
121 fn deref(&self) -> &Wtf8
{
126 impl ops
::DerefMut
for Wtf8Buf
{
127 fn deref_mut(&mut self) -> &mut Wtf8
{
132 /// Format the string with double quotes,
133 /// and surrogates as `\u` followed by four hexadecimal digits.
134 /// Example: `"a\u{D800}"` for a string with code points [U+0061, U+D800]
135 impl fmt
::Debug
for Wtf8Buf
{
137 fn fmt(&self, formatter
: &mut fmt
::Formatter
<'_
>) -> fmt
::Result
{
138 fmt
::Debug
::fmt(&**self, formatter
)
143 /// Creates a new, empty WTF-8 string.
145 pub fn new() -> Wtf8Buf
{
146 Wtf8Buf { bytes: Vec::new() }
149 /// Creates a new, empty WTF-8 string with pre-allocated capacity for `capacity` bytes.
151 pub fn with_capacity(capacity
: usize) -> Wtf8Buf
{
152 Wtf8Buf { bytes: Vec::with_capacity(capacity) }
155 /// Creates a WTF-8 string from a UTF-8 `String`.
157 /// This takes ownership of the `String` and does not copy.
159 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
161 pub fn from_string(string
: String
) -> Wtf8Buf
{
162 Wtf8Buf { bytes: string.into_bytes() }
165 /// Creates a WTF-8 string from a UTF-8 `&str` slice.
167 /// This copies the content of the slice.
169 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
171 pub fn from_str(str: &str) -> Wtf8Buf
{
172 Wtf8Buf { bytes: <[_]>::to_vec(str.as_bytes()) }
175 pub fn clear(&mut self) {
179 /// Creates a WTF-8 string from a potentially ill-formed UTF-16 slice of 16-bit code units.
181 /// This is lossless: calling `.encode_wide()` on the resulting string
182 /// will always return the original code units.
183 pub fn from_wide(v
: &[u16]) -> Wtf8Buf
{
184 let mut string
= Wtf8Buf
::with_capacity(v
.len());
185 for item
in char::decode_utf16(v
.iter().cloned()) {
187 Ok(ch
) => string
.push_char(ch
),
189 let surrogate
= surrogate
.unpaired_surrogate();
190 // Surrogates are known to be in the code point range.
191 let code_point
= unsafe { CodePoint::from_u32_unchecked(surrogate as u32) }
;
192 // Skip the WTF-8 concatenation check,
193 // surrogate pairs are already decoded by decode_utf16
194 string
.push_code_point_unchecked(code_point
)
201 /// Copied from String::push
202 /// This does **not** include the WTF-8 concatenation check.
203 fn push_code_point_unchecked(&mut self, code_point
: CodePoint
) {
204 let mut bytes
= [0; 4];
205 let bytes
= char::encode_utf8_raw(code_point
.value
, &mut bytes
);
206 self.bytes
.extend_from_slice(bytes
)
210 pub fn as_slice(&self) -> &Wtf8
{
211 unsafe { Wtf8::from_bytes_unchecked(&self.bytes) }
215 pub fn as_mut_slice(&mut self) -> &mut Wtf8
{
216 unsafe { Wtf8::from_mut_bytes_unchecked(&mut self.bytes) }
219 /// Reserves capacity for at least `additional` more bytes to be inserted
220 /// in the given `Wtf8Buf`.
221 /// The collection may reserve more space to avoid frequent reallocations.
225 /// Panics if the new capacity overflows `usize`.
227 pub fn reserve(&mut self, additional
: usize) {
228 self.bytes
.reserve(additional
)
232 pub fn reserve_exact(&mut self, additional
: usize) {
233 self.bytes
.reserve_exact(additional
)
237 pub fn shrink_to_fit(&mut self) {
238 self.bytes
.shrink_to_fit()
242 pub fn shrink_to(&mut self, min_capacity
: usize) {
243 self.bytes
.shrink_to(min_capacity
)
246 /// Returns the number of bytes that this string buffer can hold without reallocating.
248 pub fn capacity(&self) -> usize {
249 self.bytes
.capacity()
252 /// Append a UTF-8 slice at the end of the string.
254 pub fn push_str(&mut self, other
: &str) {
255 self.bytes
.extend_from_slice(other
.as_bytes())
258 /// Append a WTF-8 slice at the end of the string.
260 /// This replaces newly paired surrogates at the boundary
261 /// with a supplementary code point,
262 /// like concatenating ill-formed UTF-16 strings effectively would.
264 pub fn push_wtf8(&mut self, other
: &Wtf8
) {
265 match ((&*self).final_lead_surrogate(), other
.initial_trail_surrogate()) {
266 // Replace newly paired surrogates by a supplementary code point.
267 (Some(lead
), Some(trail
)) => {
268 let len_without_lead_surrogate
= self.len() - 3;
269 self.bytes
.truncate(len_without_lead_surrogate
);
270 let other_without_trail_surrogate
= &other
.bytes
[3..];
271 // 4 bytes for the supplementary code point
272 self.bytes
.reserve(4 + other_without_trail_surrogate
.len());
273 self.push_char(decode_surrogate_pair(lead
, trail
));
274 self.bytes
.extend_from_slice(other_without_trail_surrogate
);
276 _
=> self.bytes
.extend_from_slice(&other
.bytes
),
280 /// Append a Unicode scalar value at the end of the string.
282 pub fn push_char(&mut self, c
: char) {
283 self.push_code_point_unchecked(CodePoint
::from_char(c
))
286 /// Append a code point at the end of the string.
288 /// This replaces newly paired surrogates at the boundary
289 /// with a supplementary code point,
290 /// like concatenating ill-formed UTF-16 strings effectively would.
292 pub fn push(&mut self, code_point
: CodePoint
) {
293 if let trail @
0xDC00..=0xDFFF = code_point
.to_u32() {
294 if let Some(lead
) = (&*self).final_lead_surrogate() {
295 let len_without_lead_surrogate
= self.len() - 3;
296 self.bytes
.truncate(len_without_lead_surrogate
);
297 self.push_char(decode_surrogate_pair(lead
, trail
as u16));
302 // No newly paired surrogates at the boundary.
303 self.push_code_point_unchecked(code_point
)
306 /// Shortens a string to the specified length.
310 /// Panics if `new_len` > current length,
311 /// or if `new_len` is not a code point boundary.
313 pub fn truncate(&mut self, new_len
: usize) {
314 assert
!(is_code_point_boundary(self, new_len
));
315 self.bytes
.truncate(new_len
)
318 /// Consumes the WTF-8 string and tries to convert it to UTF-8.
320 /// This does not copy the data.
322 /// If the contents are not well-formed UTF-8
323 /// (that is, if the string contains surrogates),
324 /// the original WTF-8 string is returned instead.
325 pub fn into_string(self) -> Result
<String
, Wtf8Buf
> {
326 match self.next_surrogate(0) {
327 None
=> Ok(unsafe { String::from_utf8_unchecked(self.bytes) }
),
328 Some(_
) => Err(self),
332 /// Consumes the WTF-8 string and converts it lossily to UTF-8.
334 /// This does not copy the data (but may overwrite parts of it in place).
336 /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”)
337 pub fn into_string_lossy(mut self) -> String
{
340 match self.next_surrogate(pos
) {
341 Some((surrogate_pos
, _
)) => {
342 pos
= surrogate_pos
+ 3;
343 self.bytes
[surrogate_pos
..pos
]
344 .copy_from_slice(UTF8_REPLACEMENT_CHARACTER
.as_bytes());
346 None
=> return unsafe { String::from_utf8_unchecked(self.bytes) }
,
351 /// Converts this `Wtf8Buf` into a boxed `Wtf8`.
353 pub fn into_box(self) -> Box
<Wtf8
> {
354 unsafe { mem::transmute(self.bytes.into_boxed_slice()) }
357 /// Converts a `Box<Wtf8>` into a `Wtf8Buf`.
358 pub fn from_box(boxed
: Box
<Wtf8
>) -> Wtf8Buf
{
359 let bytes
: Box
<[u8]> = unsafe { mem::transmute(boxed) }
;
360 Wtf8Buf { bytes: bytes.into_vec() }
364 /// Creates a new WTF-8 string from an iterator of code points.
366 /// This replaces surrogate code point pairs with supplementary code points,
367 /// like concatenating ill-formed UTF-16 strings effectively would.
368 impl FromIterator
<CodePoint
> for Wtf8Buf
{
369 fn from_iter
<T
: IntoIterator
<Item
= CodePoint
>>(iter
: T
) -> Wtf8Buf
{
370 let mut string
= Wtf8Buf
::new();
376 /// Append code points from an iterator to the string.
378 /// This replaces surrogate code point pairs with supplementary code points,
379 /// like concatenating ill-formed UTF-16 strings effectively would.
380 impl Extend
<CodePoint
> for Wtf8Buf
{
381 fn extend
<T
: IntoIterator
<Item
= CodePoint
>>(&mut self, iter
: T
) {
382 let iterator
= iter
.into_iter();
383 let (low
, _high
) = iterator
.size_hint();
384 // Lower bound of one byte per code point (ASCII only)
385 self.bytes
.reserve(low
);
386 iterator
.for_each(move |code_point
| self.push(code_point
));
390 fn extend_one(&mut self, code_point
: CodePoint
) {
391 self.push(code_point
);
395 fn extend_reserve(&mut self, additional
: usize) {
396 // Lower bound of one byte per code point (ASCII only)
397 self.bytes
.reserve(additional
);
401 /// A borrowed slice of well-formed WTF-8 data.
403 /// Similar to `&str`, but can additionally contain surrogate code points
404 /// if they’re not in a surrogate pair.
405 #[derive(Eq, Ord, PartialEq, PartialOrd)]
410 impl AsInner
<[u8]> for Wtf8
{
411 fn as_inner(&self) -> &[u8] {
416 /// Format the slice with double quotes,
417 /// and surrogates as `\u` followed by four hexadecimal digits.
418 /// Example: `"a\u{D800}"` for a slice with code points [U+0061, U+D800]
419 impl fmt
::Debug
for Wtf8
{
420 fn fmt(&self, formatter
: &mut fmt
::Formatter
<'_
>) -> fmt
::Result
{
421 fn write_str_escaped(f
: &mut fmt
::Formatter
<'_
>, s
: &str) -> fmt
::Result
{
422 use crate::fmt
::Write
;
423 for c
in s
.chars().flat_map(|c
| c
.escape_debug()) {
429 formatter
.write_str("\"")?
;
431 while let Some((surrogate_pos
, surrogate
)) = self.next_surrogate(pos
) {
432 write_str_escaped(formatter
, unsafe {
433 str::from_utf8_unchecked(&self.bytes
[pos
..surrogate_pos
])
435 write
!(formatter
, "\\u{{{:x}}}", surrogate
)?
;
436 pos
= surrogate_pos
+ 3;
438 write_str_escaped(formatter
, unsafe { str::from_utf8_unchecked(&self.bytes[pos..]) }
)?
;
439 formatter
.write_str("\"")
443 impl fmt
::Display
for Wtf8
{
444 fn fmt(&self, formatter
: &mut fmt
::Formatter
<'_
>) -> fmt
::Result
{
445 let wtf8_bytes
= &self.bytes
;
448 match self.next_surrogate(pos
) {
449 Some((surrogate_pos
, _
)) => {
450 formatter
.write_str(unsafe {
451 str::from_utf8_unchecked(&wtf8_bytes
[pos
..surrogate_pos
])
453 formatter
.write_str(UTF8_REPLACEMENT_CHARACTER
)?
;
454 pos
= surrogate_pos
+ 3;
457 let s
= unsafe { str::from_utf8_unchecked(&wtf8_bytes[pos..]) }
;
458 if pos
== 0 { return s.fmt(formatter) }
else { return formatter.write_str(s) }
466 /// Creates a WTF-8 slice from a UTF-8 `&str` slice.
468 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
470 pub fn from_str(value
: &str) -> &Wtf8
{
471 unsafe { Wtf8::from_bytes_unchecked(value.as_bytes()) }
474 /// Creates a WTF-8 slice from a WTF-8 byte slice.
476 /// Since the byte slice is not checked for valid WTF-8, this functions is
479 unsafe fn from_bytes_unchecked(value
: &[u8]) -> &Wtf8
{
480 mem
::transmute(value
)
483 /// Creates a mutable WTF-8 slice from a mutable WTF-8 byte slice.
485 /// Since the byte slice is not checked for valid WTF-8, this functions is
488 unsafe fn from_mut_bytes_unchecked(value
: &mut [u8]) -> &mut Wtf8
{
489 mem
::transmute(value
)
492 /// Returns the length, in WTF-8 bytes.
494 pub fn len(&self) -> usize {
499 pub fn is_empty(&self) -> bool
{
500 self.bytes
.is_empty()
503 /// Returns the code point at `position` if it is in the ASCII range,
504 /// or `b'\xFF' otherwise.
508 /// Panics if `position` is beyond the end of the string.
510 pub fn ascii_byte_at(&self, position
: usize) -> u8 {
511 match self.bytes
[position
] {
512 ascii_byte @
0x00..=0x7F => ascii_byte
,
517 /// Returns an iterator for the string’s code points.
519 pub fn code_points(&self) -> Wtf8CodePoints
<'_
> {
520 Wtf8CodePoints { bytes: self.bytes.iter() }
523 /// Tries to convert the string to UTF-8 and return a `&str` slice.
525 /// Returns `None` if the string contains surrogates.
527 /// This does not copy the data.
529 pub fn as_str(&self) -> Option
<&str> {
530 // Well-formed WTF-8 is also well-formed UTF-8
531 // if and only if it contains no surrogate.
532 match self.next_surrogate(0) {
533 None
=> Some(unsafe { str::from_utf8_unchecked(&self.bytes) }
),
538 /// Lossily converts the string to UTF-8.
539 /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8.
541 /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”).
543 /// This only copies the data if necessary (if it contains any surrogate).
544 pub fn to_string_lossy(&self) -> Cow
<'_
, str> {
545 let surrogate_pos
= match self.next_surrogate(0) {
546 None
=> return Cow
::Borrowed(unsafe { str::from_utf8_unchecked(&self.bytes) }
),
547 Some((pos
, _
)) => pos
,
549 let wtf8_bytes
= &self.bytes
;
550 let mut utf8_bytes
= Vec
::with_capacity(self.len());
551 utf8_bytes
.extend_from_slice(&wtf8_bytes
[..surrogate_pos
]);
552 utf8_bytes
.extend_from_slice(UTF8_REPLACEMENT_CHARACTER
.as_bytes());
553 let mut pos
= surrogate_pos
+ 3;
555 match self.next_surrogate(pos
) {
556 Some((surrogate_pos
, _
)) => {
557 utf8_bytes
.extend_from_slice(&wtf8_bytes
[pos
..surrogate_pos
]);
558 utf8_bytes
.extend_from_slice(UTF8_REPLACEMENT_CHARACTER
.as_bytes());
559 pos
= surrogate_pos
+ 3;
562 utf8_bytes
.extend_from_slice(&wtf8_bytes
[pos
..]);
563 return Cow
::Owned(unsafe { String::from_utf8_unchecked(utf8_bytes) }
);
569 /// Converts the WTF-8 string to potentially ill-formed UTF-16
570 /// and return an iterator of 16-bit code units.
572 /// This is lossless:
573 /// calling `Wtf8Buf::from_ill_formed_utf16` on the resulting code units
574 /// would always return the original WTF-8 string.
576 pub fn encode_wide(&self) -> EncodeWide
<'_
> {
577 EncodeWide { code_points: self.code_points(), extra: 0 }
581 fn next_surrogate(&self, mut pos
: usize) -> Option
<(usize, u16)> {
582 let mut iter
= self.bytes
[pos
..].iter();
584 let b
= *iter
.next()?
;
590 } else if b
== 0xED {
591 match (iter
.next(), iter
.next()) {
592 (Some(&b2
), Some(&b3
)) if b2
>= 0xA0 => {
593 return Some((pos
, decode_surrogate(b2
, b3
)));
611 fn final_lead_surrogate(&self) -> Option
<u16> {
613 [.., 0xED, b2 @
0xA0..=0xAF, b3
] => Some(decode_surrogate(b2
, b3
)),
619 fn initial_trail_surrogate(&self) -> Option
<u16> {
621 [0xED, b2 @
0xB0..=0xBF, b3
, ..] => Some(decode_surrogate(b2
, b3
)),
626 pub fn clone_into(&self, buf
: &mut Wtf8Buf
) {
627 self.bytes
.clone_into(&mut buf
.bytes
)
630 /// Boxes this `Wtf8`.
632 pub fn into_box(&self) -> Box
<Wtf8
> {
633 let boxed
: Box
<[u8]> = self.bytes
.into();
634 unsafe { mem::transmute(boxed) }
637 /// Creates a boxed, empty `Wtf8`.
638 pub fn empty_box() -> Box
<Wtf8
> {
639 let boxed
: Box
<[u8]> = Default
::default();
640 unsafe { mem::transmute(boxed) }
644 pub fn into_arc(&self) -> Arc
<Wtf8
> {
645 let arc
: Arc
<[u8]> = Arc
::from(&self.bytes
);
646 unsafe { Arc::from_raw(Arc::into_raw(arc) as *const Wtf8) }
650 pub fn into_rc(&self) -> Rc
<Wtf8
> {
651 let rc
: Rc
<[u8]> = Rc
::from(&self.bytes
);
652 unsafe { Rc::from_raw(Rc::into_raw(rc) as *const Wtf8) }
656 pub fn make_ascii_lowercase(&mut self) {
657 self.bytes
.make_ascii_lowercase()
661 pub fn make_ascii_uppercase(&mut self) {
662 self.bytes
.make_ascii_uppercase()
666 pub fn to_ascii_lowercase(&self) -> Wtf8Buf
{
667 Wtf8Buf { bytes: self.bytes.to_ascii_lowercase() }
671 pub fn to_ascii_uppercase(&self) -> Wtf8Buf
{
672 Wtf8Buf { bytes: self.bytes.to_ascii_uppercase() }
676 pub fn is_ascii(&self) -> bool
{
677 self.bytes
.is_ascii()
681 pub fn eq_ignore_ascii_case(&self, other
: &Self) -> bool
{
682 self.bytes
.eq_ignore_ascii_case(&other
.bytes
)
686 /// Returns a slice of the given string for the byte range [`begin`..`end`).
690 /// Panics when `begin` and `end` do not point to code point boundaries,
691 /// or point beyond the end of the string.
692 impl ops
::Index
<ops
::Range
<usize>> for Wtf8
{
696 fn index(&self, range
: ops
::Range
<usize>) -> &Wtf8
{
697 // is_code_point_boundary checks that the index is in [0, .len()]
698 if range
.start
<= range
.end
699 && is_code_point_boundary(self, range
.start
)
700 && is_code_point_boundary(self, range
.end
)
702 unsafe { slice_unchecked(self, range.start, range.end) }
704 slice_error_fail(self, range
.start
, range
.end
)
709 /// Returns a slice of the given string from byte `begin` to its end.
713 /// Panics when `begin` is not at a code point boundary,
714 /// or is beyond the end of the string.
715 impl ops
::Index
<ops
::RangeFrom
<usize>> for Wtf8
{
719 fn index(&self, range
: ops
::RangeFrom
<usize>) -> &Wtf8
{
720 // is_code_point_boundary checks that the index is in [0, .len()]
721 if is_code_point_boundary(self, range
.start
) {
722 unsafe { slice_unchecked(self, range.start, self.len()) }
724 slice_error_fail(self, range
.start
, self.len())
729 /// Returns a slice of the given string from its beginning to byte `end`.
733 /// Panics when `end` is not at a code point boundary,
734 /// or is beyond the end of the string.
735 impl ops
::Index
<ops
::RangeTo
<usize>> for Wtf8
{
739 fn index(&self, range
: ops
::RangeTo
<usize>) -> &Wtf8
{
740 // is_code_point_boundary checks that the index is in [0, .len()]
741 if is_code_point_boundary(self, range
.end
) {
742 unsafe { slice_unchecked(self, 0, range.end) }
744 slice_error_fail(self, 0, range
.end
)
749 impl ops
::Index
<ops
::RangeFull
> for Wtf8
{
753 fn index(&self, _range
: ops
::RangeFull
) -> &Wtf8
{
759 fn decode_surrogate(second_byte
: u8, third_byte
: u8) -> u16 {
760 // The first byte is assumed to be 0xED
761 0xD800 | (second_byte
as u16 & 0x3F) << 6 | third_byte
as u16 & 0x3F
765 fn decode_surrogate_pair(lead
: u16, trail
: u16) -> char {
766 let code_point
= 0x10000 + ((((lead
- 0xD800) as u32) << 10) | (trail
- 0xDC00) as u32);
767 unsafe { char::from_u32_unchecked(code_point) }
770 /// Copied from core::str::StrPrelude::is_char_boundary
772 pub fn is_code_point_boundary(slice
: &Wtf8
, index
: usize) -> bool
{
773 if index
== slice
.len() {
776 match slice
.bytes
.get(index
) {
778 Some(&b
) => b
< 128 || b
>= 192,
782 /// Copied from core::str::raw::slice_unchecked
784 pub unsafe fn slice_unchecked(s
: &Wtf8
, begin
: usize, end
: usize) -> &Wtf8
{
785 // memory layout of an &[u8] and &Wtf8 are the same
786 Wtf8
::from_bytes_unchecked(slice
::from_raw_parts(s
.bytes
.as_ptr().add(begin
), end
- begin
))
789 /// Copied from core::str::raw::slice_error_fail
791 pub fn slice_error_fail(s
: &Wtf8
, begin
: usize, end
: usize) -> ! {
792 assert
!(begin
<= end
);
793 panic
!("index {} and/or {} in `{:?}` do not lie on character boundary", begin
, end
, s
);
796 /// Iterator for the code points of a WTF-8 string.
798 /// Created with the method `.code_points()`.
800 pub struct Wtf8CodePoints
<'a
> {
801 bytes
: slice
::Iter
<'a
, u8>,
804 impl<'a
> Iterator
for Wtf8CodePoints
<'a
> {
805 type Item
= CodePoint
;
808 fn next(&mut self) -> Option
<CodePoint
> {
809 next_code_point(&mut self.bytes
).map(|c
| CodePoint { value: c }
)
813 fn size_hint(&self) -> (usize, Option
<usize>) {
814 let len
= self.bytes
.len();
815 (len
.saturating_add(3) / 4, Some(len
))
819 /// Generates a wide character sequence for potentially ill-formed UTF-16.
820 #[stable(feature = "rust1", since = "1.0.0")]
822 pub struct EncodeWide
<'a
> {
823 code_points
: Wtf8CodePoints
<'a
>,
827 // Copied from libunicode/u_str.rs
828 #[stable(feature = "rust1", since = "1.0.0")]
829 impl<'a
> Iterator
for EncodeWide
<'a
> {
833 fn next(&mut self) -> Option
<u16> {
835 let tmp
= self.extra
;
840 let mut buf
= [0; 2];
841 self.code_points
.next().map(|code_point
| {
842 let n
= char::encode_utf16_raw(code_point
.value
, &mut buf
).len();
851 fn size_hint(&self) -> (usize, Option
<usize>) {
852 let (low
, high
) = self.code_points
.size_hint();
853 // every code point gets either one u16 or two u16,
854 // so this iterator is between 1 or 2 times as
855 // long as the underlying iterator.
856 (low
, high
.and_then(|n
| n
.checked_mul(2)))
860 impl Hash
for CodePoint
{
862 fn hash
<H
: Hasher
>(&self, state
: &mut H
) {
863 self.value
.hash(state
)
867 impl Hash
for Wtf8Buf
{
869 fn hash
<H
: Hasher
>(&self, state
: &mut H
) {
870 state
.write(&self.bytes
);
877 fn hash
<H
: Hasher
>(&self, state
: &mut H
) {
878 state
.write(&self.bytes
);
886 use crate::borrow
::Cow
;
889 fn code_point_from_u32() {
890 assert
!(CodePoint
::from_u32(0).is_some());
891 assert
!(CodePoint
::from_u32(0xD800).is_some());
892 assert
!(CodePoint
::from_u32(0x10FFFF).is_some());
893 assert
!(CodePoint
::from_u32(0x110000).is_none());
897 fn code_point_to_u32() {
898 fn c(value
: u32) -> CodePoint
{
899 CodePoint
::from_u32(value
).unwrap()
901 assert_eq
!(c(0).to_u32(), 0);
902 assert_eq
!(c(0xD800).to_u32(), 0xD800);
903 assert_eq
!(c(0x10FFFF).to_u32(), 0x10FFFF);
907 fn code_point_from_char() {
908 assert_eq
!(CodePoint
::from_char('a'
).to_u32(), 0x61);
909 assert_eq
!(CodePoint
::from_char('💩'
).to_u32(), 0x1F4A9);
913 fn code_point_to_string() {
914 assert_eq
!(format
!("{:?}", CodePoint
::from_char('a'
)), "U+0061");
915 assert_eq
!(format
!("{:?}", CodePoint
::from_char('💩'
)), "U+1F4A9");
919 fn code_point_to_char() {
920 fn c(value
: u32) -> CodePoint
{
921 CodePoint
::from_u32(value
).unwrap()
923 assert_eq
!(c(0x61).to_char(), Some('a'
));
924 assert_eq
!(c(0x1F4A9).to_char(), Some('💩'
));
925 assert_eq
!(c(0xD800).to_char(), None
);
929 fn code_point_to_char_lossy() {
930 fn c(value
: u32) -> CodePoint
{
931 CodePoint
::from_u32(value
).unwrap()
933 assert_eq
!(c(0x61).to_char_lossy(), 'a'
);
934 assert_eq
!(c(0x1F4A9).to_char_lossy(), '💩'
);
935 assert_eq
!(c(0xD800).to_char_lossy(), '
\u{FFFD}'
);
940 assert_eq
!(Wtf8Buf
::new().bytes
, b
"");
944 fn wtf8buf_from_str() {
945 assert_eq
!(Wtf8Buf
::from_str("").bytes
, b
"");
946 assert_eq
!(Wtf8Buf
::from_str("aé 💩").bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
950 fn wtf8buf_from_string() {
951 assert_eq
!(Wtf8Buf
::from_string(String
::from("")).bytes
, b
"");
953 Wtf8Buf
::from_string(String
::from("aé 💩")).bytes
,
954 b
"a\xC3\xA9 \xF0\x9F\x92\xA9"
959 fn wtf8buf_from_wide() {
960 assert_eq
!(Wtf8Buf
::from_wide(&[]).bytes
, b
"");
962 Wtf8Buf
::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]).bytes
,
963 b
"a\xC3\xA9 \xED\xA0\xBD\xF0\x9F\x92\xA9"
968 fn wtf8buf_push_str() {
969 let mut string
= Wtf8Buf
::new();
970 assert_eq
!(string
.bytes
, b
"");
971 string
.push_str("aé 💩");
972 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
976 fn wtf8buf_push_char() {
977 let mut string
= Wtf8Buf
::from_str("aé ");
978 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 ");
979 string
.push_char('💩'
);
980 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
985 let mut string
= Wtf8Buf
::from_str("aé ");
986 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 ");
987 string
.push(CodePoint
::from_char('💩'
));
988 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
990 fn c(value
: u32) -> CodePoint
{
991 CodePoint
::from_u32(value
).unwrap()
994 let mut string
= Wtf8Buf
::new();
995 string
.push(c(0xD83D)); // lead
996 string
.push(c(0xDCA9)); // trail
997 assert_eq
!(string
.bytes
, b
"\xF0\x9F\x92\xA9"); // Magic!
999 let mut string
= Wtf8Buf
::new();
1000 string
.push(c(0xD83D)); // lead
1001 string
.push(c(0x20)); // not surrogate
1002 string
.push(c(0xDCA9)); // trail
1003 assert_eq
!(string
.bytes
, b
"\xED\xA0\xBD \xED\xB2\xA9");
1005 let mut string
= Wtf8Buf
::new();
1006 string
.push(c(0xD800)); // lead
1007 string
.push(c(0xDBFF)); // lead
1008 assert_eq
!(string
.bytes
, b
"\xED\xA0\x80\xED\xAF\xBF");
1010 let mut string
= Wtf8Buf
::new();
1011 string
.push(c(0xD800)); // lead
1012 string
.push(c(0xE000)); // not surrogate
1013 assert_eq
!(string
.bytes
, b
"\xED\xA0\x80\xEE\x80\x80");
1015 let mut string
= Wtf8Buf
::new();
1016 string
.push(c(0xD7FF)); // not surrogate
1017 string
.push(c(0xDC00)); // trail
1018 assert_eq
!(string
.bytes
, b
"\xED\x9F\xBF\xED\xB0\x80");
1020 let mut string
= Wtf8Buf
::new();
1021 string
.push(c(0x61)); // not surrogate, < 3 bytes
1022 string
.push(c(0xDC00)); // trail
1023 assert_eq
!(string
.bytes
, b
"\x61\xED\xB0\x80");
1025 let mut string
= Wtf8Buf
::new();
1026 string
.push(c(0xDC00)); // trail
1027 assert_eq
!(string
.bytes
, b
"\xED\xB0\x80");
1031 fn wtf8buf_push_wtf8() {
1032 let mut string
= Wtf8Buf
::from_str("aé");
1033 assert_eq
!(string
.bytes
, b
"a\xC3\xA9");
1034 string
.push_wtf8(Wtf8
::from_str(" 💩"));
1035 assert_eq
!(string
.bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
1037 fn w(v
: &[u8]) -> &Wtf8
{
1038 unsafe { Wtf8::from_bytes_unchecked(v) }
1041 let mut string
= Wtf8Buf
::new();
1042 string
.push_wtf8(w(b
"\xED\xA0\xBD")); // lead
1043 string
.push_wtf8(w(b
"\xED\xB2\xA9")); // trail
1044 assert_eq
!(string
.bytes
, b
"\xF0\x9F\x92\xA9"); // Magic!
1046 let mut string
= Wtf8Buf
::new();
1047 string
.push_wtf8(w(b
"\xED\xA0\xBD")); // lead
1048 string
.push_wtf8(w(b
" ")); // not surrogate
1049 string
.push_wtf8(w(b
"\xED\xB2\xA9")); // trail
1050 assert_eq
!(string
.bytes
, b
"\xED\xA0\xBD \xED\xB2\xA9");
1052 let mut string
= Wtf8Buf
::new();
1053 string
.push_wtf8(w(b
"\xED\xA0\x80")); // lead
1054 string
.push_wtf8(w(b
"\xED\xAF\xBF")); // lead
1055 assert_eq
!(string
.bytes
, b
"\xED\xA0\x80\xED\xAF\xBF");
1057 let mut string
= Wtf8Buf
::new();
1058 string
.push_wtf8(w(b
"\xED\xA0\x80")); // lead
1059 string
.push_wtf8(w(b
"\xEE\x80\x80")); // not surrogate
1060 assert_eq
!(string
.bytes
, b
"\xED\xA0\x80\xEE\x80\x80");
1062 let mut string
= Wtf8Buf
::new();
1063 string
.push_wtf8(w(b
"\xED\x9F\xBF")); // not surrogate
1064 string
.push_wtf8(w(b
"\xED\xB0\x80")); // trail
1065 assert_eq
!(string
.bytes
, b
"\xED\x9F\xBF\xED\xB0\x80");
1067 let mut string
= Wtf8Buf
::new();
1068 string
.push_wtf8(w(b
"a")); // not surrogate, < 3 bytes
1069 string
.push_wtf8(w(b
"\xED\xB0\x80")); // trail
1070 assert_eq
!(string
.bytes
, b
"\x61\xED\xB0\x80");
1072 let mut string
= Wtf8Buf
::new();
1073 string
.push_wtf8(w(b
"\xED\xB0\x80")); // trail
1074 assert_eq
!(string
.bytes
, b
"\xED\xB0\x80");
1078 fn wtf8buf_truncate() {
1079 let mut string
= Wtf8Buf
::from_str("aé");
1081 assert_eq
!(string
.bytes
, b
"a");
1086 fn wtf8buf_truncate_fail_code_point_boundary() {
1087 let mut string
= Wtf8Buf
::from_str("aé");
1093 fn wtf8buf_truncate_fail_longer() {
1094 let mut string
= Wtf8Buf
::from_str("aé");
1099 fn wtf8buf_into_string() {
1100 let mut string
= Wtf8Buf
::from_str("aé 💩");
1101 assert_eq
!(string
.clone().into_string(), Ok(String
::from("aé 💩")));
1102 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1103 assert_eq
!(string
.clone().into_string(), Err(string
));
1107 fn wtf8buf_into_string_lossy() {
1108 let mut string
= Wtf8Buf
::from_str("aé 💩");
1109 assert_eq
!(string
.clone().into_string_lossy(), String
::from("aé 💩"));
1110 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1111 assert_eq
!(string
.clone().into_string_lossy(), String
::from("aé 💩�"));
1115 fn wtf8buf_from_iterator() {
1116 fn f(values
: &[u32]) -> Wtf8Buf
{
1117 values
.iter().map(|&c
| CodePoint
::from_u32(c
).unwrap()).collect
::<Wtf8Buf
>()
1119 assert_eq
!(f(&[0x61, 0xE9, 0x20, 0x1F4A9]).bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
1121 assert_eq
!(f(&[0xD83D, 0xDCA9]).bytes
, b
"\xF0\x9F\x92\xA9"); // Magic!
1122 assert_eq
!(f(&[0xD83D, 0x20, 0xDCA9]).bytes
, b
"\xED\xA0\xBD \xED\xB2\xA9");
1123 assert_eq
!(f(&[0xD800, 0xDBFF]).bytes
, b
"\xED\xA0\x80\xED\xAF\xBF");
1124 assert_eq
!(f(&[0xD800, 0xE000]).bytes
, b
"\xED\xA0\x80\xEE\x80\x80");
1125 assert_eq
!(f(&[0xD7FF, 0xDC00]).bytes
, b
"\xED\x9F\xBF\xED\xB0\x80");
1126 assert_eq
!(f(&[0x61, 0xDC00]).bytes
, b
"\x61\xED\xB0\x80");
1127 assert_eq
!(f(&[0xDC00]).bytes
, b
"\xED\xB0\x80");
1131 fn wtf8buf_extend() {
1132 fn e(initial
: &[u32], extended
: &[u32]) -> Wtf8Buf
{
1133 fn c(value
: &u32) -> CodePoint
{
1134 CodePoint
::from_u32(*value
).unwrap()
1136 let mut string
= initial
.iter().map(c
).collect
::<Wtf8Buf
>();
1137 string
.extend(extended
.iter().map(c
));
1141 assert_eq
!(e(&[0x61, 0xE9], &[0x20, 0x1F4A9]).bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
1143 assert_eq
!(e(&[0xD83D], &[0xDCA9]).bytes
, b
"\xF0\x9F\x92\xA9"); // Magic!
1144 assert_eq
!(e(&[0xD83D, 0x20], &[0xDCA9]).bytes
, b
"\xED\xA0\xBD \xED\xB2\xA9");
1145 assert_eq
!(e(&[0xD800], &[0xDBFF]).bytes
, b
"\xED\xA0\x80\xED\xAF\xBF");
1146 assert_eq
!(e(&[0xD800], &[0xE000]).bytes
, b
"\xED\xA0\x80\xEE\x80\x80");
1147 assert_eq
!(e(&[0xD7FF], &[0xDC00]).bytes
, b
"\xED\x9F\xBF\xED\xB0\x80");
1148 assert_eq
!(e(&[0x61], &[0xDC00]).bytes
, b
"\x61\xED\xB0\x80");
1149 assert_eq
!(e(&[], &[0xDC00]).bytes
, b
"\xED\xB0\x80");
1154 let mut string
= Wtf8Buf
::from_str("a\té \u{7f}💩\r");
1155 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1156 assert_eq
!(format
!("{:?}", string
), "\"a\\té \\u{7f}\u{1f4a9}\\r\\u{d800}\"");
1160 fn wtf8buf_as_slice() {
1161 assert_eq
!(Wtf8Buf
::from_str("aé").as_slice(), Wtf8
::from_str("aé"));
1165 fn wtf8buf_show_str() {
1166 let text
= "a\té 💩\r";
1167 let string
= Wtf8Buf
::from_str(text
);
1168 assert_eq
!(format
!("{:?}", text
), format
!("{:?}", string
));
1172 fn wtf8_from_str() {
1173 assert_eq
!(&Wtf8
::from_str("").bytes
, b
"");
1174 assert_eq
!(&Wtf8
::from_str("aé 💩").bytes
, b
"a\xC3\xA9 \xF0\x9F\x92\xA9");
1179 assert_eq
!(Wtf8
::from_str("").len(), 0);
1180 assert_eq
!(Wtf8
::from_str("aé 💩").len(), 8);
1185 assert_eq
!(&Wtf8
::from_str("aé 💩")[1..4].bytes
, b
"\xC3\xA9 ");
1190 fn wtf8_slice_not_code_point_boundary() {
1191 &Wtf8
::from_str("aé 💩")[2..4];
1195 fn wtf8_slice_from() {
1196 assert_eq
!(&Wtf8
::from_str("aé 💩")[1..].bytes
, b
"\xC3\xA9 \xF0\x9F\x92\xA9");
1201 fn wtf8_slice_from_not_code_point_boundary() {
1202 &Wtf8
::from_str("aé 💩")[2..];
1206 fn wtf8_slice_to() {
1207 assert_eq
!(&Wtf8
::from_str("aé 💩")[..4].bytes
, b
"a\xC3\xA9 ");
1212 fn wtf8_slice_to_not_code_point_boundary() {
1213 &Wtf8
::from_str("aé 💩")[5..];
1217 fn wtf8_ascii_byte_at() {
1218 let slice
= Wtf8
::from_str("aé 💩");
1219 assert_eq
!(slice
.ascii_byte_at(0), b'a'
);
1220 assert_eq
!(slice
.ascii_byte_at(1), b'
\xFF'
);
1221 assert_eq
!(slice
.ascii_byte_at(2), b'
\xFF'
);
1222 assert_eq
!(slice
.ascii_byte_at(3), b' '
);
1223 assert_eq
!(slice
.ascii_byte_at(4), b'
\xFF'
);
1227 fn wtf8_code_points() {
1228 fn c(value
: u32) -> CodePoint
{
1229 CodePoint
::from_u32(value
).unwrap()
1231 fn cp(string
: &Wtf8Buf
) -> Vec
<Option
<char>> {
1232 string
.code_points().map(|c
| c
.to_char()).collect
::<Vec
<_
>>()
1234 let mut string
= Wtf8Buf
::from_str("é ");
1235 assert_eq
!(cp(&string
), [Some('é'
), Some(' '
)]);
1236 string
.push(c(0xD83D));
1237 assert_eq
!(cp(&string
), [Some('é'
), Some(' '
), None
]);
1238 string
.push(c(0xDCA9));
1239 assert_eq
!(cp(&string
), [Some('é'
), Some(' '
), Some('💩'
)]);
1244 assert_eq
!(Wtf8
::from_str("").as_str(), Some(""));
1245 assert_eq
!(Wtf8
::from_str("aé 💩").as_str(), Some("aé 💩"));
1246 let mut string
= Wtf8Buf
::new();
1247 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1248 assert_eq
!(string
.as_str(), None
);
1252 fn wtf8_to_string_lossy() {
1253 assert_eq
!(Wtf8
::from_str("").to_string_lossy(), Cow
::Borrowed(""));
1254 assert_eq
!(Wtf8
::from_str("aé 💩").to_string_lossy(), Cow
::Borrowed("aé 💩"));
1255 let mut string
= Wtf8Buf
::from_str("aé 💩");
1256 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1257 let expected
: Cow
<'_
, str> = Cow
::Owned(String
::from("aé 💩�"));
1258 assert_eq
!(string
.to_string_lossy(), expected
);
1263 fn d(b
: &[u8]) -> String
{
1264 (&unsafe { Wtf8::from_bytes_unchecked(b) }
).to_string()
1267 assert_eq
!("", d("".as_bytes()));
1268 assert_eq
!("aé 💩", d("aé 💩".as_bytes()));
1270 let mut string
= Wtf8Buf
::from_str("aé 💩");
1271 string
.push(CodePoint
::from_u32(0xD800).unwrap());
1272 assert_eq
!("aé 💩�", d(string
.as_inner()));
1276 fn wtf8_encode_wide() {
1277 let mut string
= Wtf8Buf
::from_str("aé ");
1278 string
.push(CodePoint
::from_u32(0xD83D).unwrap());
1279 string
.push_char('💩'
);
1281 string
.encode_wide().collect
::<Vec
<_
>>(),
1282 vec
![0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]