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;
33 use hash::{Hash, Hasher};
34 use iter::FromIterator;
41 use sys_common::AsInner;
43 const UTF8_REPLACEMENT_CHARACTER: &'static str = "\u{FFFD}";
45 /// A Unicode code point: from U+0000 to U+10FFFF.
47 /// Compare with the `char` type,
48 /// which represents a Unicode scalar value:
49 /// a code point that is not a surrogate (U+D800 to U+DFFF).
50 #[derive(Eq, PartialEq, Ord, PartialOrd, Clone, Copy)]
51 pub struct CodePoint {
55 /// Format the code point as `U+` followed by four to six hexadecimal digits.
56 /// Example: `U+1F4A9`
57 impl fmt::Debug for CodePoint {
59 fn fmt(&self, formatter: &mut fmt::Formatter) -> Result<(), fmt::Error> {
60 write!(formatter, "U+{:04X}", self.value)
65 /// Unsafely creates a new `CodePoint` without checking the value.
67 /// Only use when `value` is known to be less than or equal to 0x10FFFF.
69 pub unsafe fn from_u32_unchecked(value: u32) -> CodePoint {
70 CodePoint { value: value }
73 /// Creates a new `CodePoint` if the value is a valid code point.
75 /// Returns `None` if `value` is above 0x10FFFF.
77 pub fn from_u32(value: u32) -> Option<CodePoint> {
79 0 ... 0x10FFFF => Some(CodePoint { value: value }),
84 /// Creates a new `CodePoint` from a `char`.
86 /// Since all Unicode scalar values are code points, this always succeeds.
88 pub fn from_char(value: char) -> CodePoint {
89 CodePoint { value: value as u32 }
92 /// Returns the numeric value of the code point.
94 pub fn to_u32(&self) -> u32 {
98 /// Optionally returns a Unicode scalar value for the code point.
100 /// Returns `None` if the code point is a surrogate (from U+D800 to U+DFFF).
102 pub fn to_char(&self) -> Option<char> {
104 0xD800 ... 0xDFFF => None,
105 _ => Some(unsafe { char::from_u32_unchecked(self.value) })
109 /// Returns a Unicode scalar value for the code point.
111 /// Returns `'\u{FFFD}'` (the replacement character “�”)
112 /// if the code point is a surrogate (from U+D800 to U+DFFF).
114 pub fn to_char_lossy(&self) -> char {
115 self.to_char().unwrap_or('\u{FFFD}')
119 /// An owned, growable string of well-formed WTF-8 data.
121 /// Similar to `String`, but can additionally contain surrogate code points
122 /// if they’re not in a surrogate pair.
123 #[derive(Eq, PartialEq, Ord, PartialOrd, Clone)]
128 impl ops::Deref for Wtf8Buf {
131 fn deref(&self) -> &Wtf8 {
136 impl ops::DerefMut for Wtf8Buf {
137 fn deref_mut(&mut self) -> &mut Wtf8 {
142 /// Format the string with double quotes,
143 /// and surrogates as `\u` followed by four hexadecimal digits.
144 /// Example: `"a\u{D800}"` for a string with code points [U+0061, U+D800]
145 impl fmt::Debug for Wtf8Buf {
147 fn fmt(&self, formatter: &mut fmt::Formatter) -> Result<(), fmt::Error> {
148 fmt::Debug::fmt(&**self, formatter)
153 /// Creates a new, empty WTF-8 string.
155 pub fn new() -> Wtf8Buf {
156 Wtf8Buf { bytes: Vec::new() }
159 /// Creates a new, empty WTF-8 string with pre-allocated capacity for `n` bytes.
161 pub fn with_capacity(n: usize) -> Wtf8Buf {
162 Wtf8Buf { bytes: Vec::with_capacity(n) }
165 /// Creates a WTF-8 string from a UTF-8 `String`.
167 /// This takes ownership of the `String` and does not copy.
169 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
171 pub fn from_string(string: String) -> Wtf8Buf {
172 Wtf8Buf { bytes: string.into_bytes() }
175 /// Creates a WTF-8 string from a UTF-8 `&str` slice.
177 /// This copies the content of the slice.
179 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
181 pub fn from_str(str: &str) -> Wtf8Buf {
182 Wtf8Buf { bytes: <[_]>::to_vec(str.as_bytes()) }
185 pub fn clear(&mut self) {
189 /// Creates a WTF-8 string from a potentially ill-formed UTF-16 slice of 16-bit code units.
191 /// This is lossless: calling `.encode_wide()` on the resulting string
192 /// will always return the original code units.
193 pub fn from_wide(v: &[u16]) -> Wtf8Buf {
194 let mut string = Wtf8Buf::with_capacity(v.len());
195 for item in char::decode_utf16(v.iter().cloned()) {
197 Ok(ch) => string.push_char(ch),
199 let surrogate = surrogate.unpaired_surrogate();
200 // Surrogates are known to be in the code point range.
201 let code_point = unsafe {
202 CodePoint::from_u32_unchecked(surrogate as u32)
204 // Skip the WTF-8 concatenation check,
205 // surrogate pairs are already decoded by decode_utf16
206 string.push_code_point_unchecked(code_point)
213 /// Copied from String::push
214 /// This does **not** include the WTF-8 concatenation check.
215 fn push_code_point_unchecked(&mut self, code_point: CodePoint) {
217 char::from_u32_unchecked(code_point.value)
219 let mut bytes = [0; 4];
220 let bytes = c.encode_utf8(&mut bytes).as_bytes();
221 self.bytes.extend_from_slice(bytes)
225 pub fn as_slice(&self) -> &Wtf8 {
226 unsafe { Wtf8::from_bytes_unchecked(&self.bytes) }
230 pub fn as_mut_slice(&mut self) -> &mut Wtf8 {
231 unsafe { Wtf8::from_mut_bytes_unchecked(&mut self.bytes) }
234 /// Reserves capacity for at least `additional` more bytes to be inserted
235 /// in the given `Wtf8Buf`.
236 /// The collection may reserve more space to avoid frequent reallocations.
240 /// Panics if the new capacity overflows `usize`.
242 pub fn reserve(&mut self, additional: usize) {
243 self.bytes.reserve(additional)
247 pub fn reserve_exact(&mut self, additional: usize) {
248 self.bytes.reserve_exact(additional)
252 pub fn shrink_to_fit(&mut self) {
253 self.bytes.shrink_to_fit()
256 /// Returns the number of bytes that this string buffer can hold without reallocating.
258 pub fn capacity(&self) -> usize {
259 self.bytes.capacity()
262 /// Append a UTF-8 slice at the end of the string.
264 pub fn push_str(&mut self, other: &str) {
265 self.bytes.extend_from_slice(other.as_bytes())
268 /// Append a WTF-8 slice at the end of the string.
270 /// This replaces newly paired surrogates at the boundary
271 /// with a supplementary code point,
272 /// like concatenating ill-formed UTF-16 strings effectively would.
274 pub fn push_wtf8(&mut self, other: &Wtf8) {
275 match ((&*self).final_lead_surrogate(), other.initial_trail_surrogate()) {
276 // Replace newly paired surrogates by a supplementary code point.
277 (Some(lead), Some(trail)) => {
278 let len_without_lead_surrogate = self.len() - 3;
279 self.bytes.truncate(len_without_lead_surrogate);
280 let other_without_trail_surrogate = &other.bytes[3..];
281 // 4 bytes for the supplementary code point
282 self.bytes.reserve(4 + other_without_trail_surrogate.len());
283 self.push_char(decode_surrogate_pair(lead, trail));
284 self.bytes.extend_from_slice(other_without_trail_surrogate);
286 _ => self.bytes.extend_from_slice(&other.bytes)
290 /// Append a Unicode scalar value at the end of the string.
292 pub fn push_char(&mut self, c: char) {
293 self.push_code_point_unchecked(CodePoint::from_char(c))
296 /// Append a code point at the end of the string.
298 /// This replaces newly paired surrogates at the boundary
299 /// with a supplementary code point,
300 /// like concatenating ill-formed UTF-16 strings effectively would.
302 pub fn push(&mut self, code_point: CodePoint) {
303 if let trail @ 0xDC00...0xDFFF = code_point.to_u32() {
304 if let Some(lead) = (&*self).final_lead_surrogate() {
305 let len_without_lead_surrogate = self.len() - 3;
306 self.bytes.truncate(len_without_lead_surrogate);
307 self.push_char(decode_surrogate_pair(lead, trail as u16));
312 // No newly paired surrogates at the boundary.
313 self.push_code_point_unchecked(code_point)
316 /// Shortens a string to the specified length.
320 /// Panics if `new_len` > current length,
321 /// or if `new_len` is not a code point boundary.
323 pub fn truncate(&mut self, new_len: usize) {
324 assert!(is_code_point_boundary(self, new_len));
325 self.bytes.truncate(new_len)
328 /// Consumes the WTF-8 string and tries to convert it to UTF-8.
330 /// This does not copy the data.
332 /// If the contents are not well-formed UTF-8
333 /// (that is, if the string contains surrogates),
334 /// the original WTF-8 string is returned instead.
335 pub fn into_string(self) -> Result<String, Wtf8Buf> {
336 match self.next_surrogate(0) {
337 None => Ok(unsafe { String::from_utf8_unchecked(self.bytes) }),
338 Some(_) => Err(self),
342 /// Consumes the WTF-8 string and converts it lossily to UTF-8.
344 /// This does not copy the data (but may overwrite parts of it in place).
346 /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”)
347 pub fn into_string_lossy(mut self) -> String {
350 match self.next_surrogate(pos) {
351 Some((surrogate_pos, _)) => {
352 pos = surrogate_pos + 3;
353 self.bytes[surrogate_pos..pos]
354 .copy_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes());
356 None => return unsafe { String::from_utf8_unchecked(self.bytes) }
361 /// Converts this `Wtf8Buf` into a boxed `Wtf8`.
363 pub fn into_box(self) -> Box<Wtf8> {
364 unsafe { mem::transmute(self.bytes.into_boxed_slice()) }
367 /// Converts a `Box<Wtf8>` into a `Wtf8Buf`.
368 pub fn from_box(boxed: Box<Wtf8>) -> Wtf8Buf {
369 let bytes: Box<[u8]> = unsafe { mem::transmute(boxed) };
370 Wtf8Buf { bytes: bytes.into_vec() }
374 /// Create a new WTF-8 string from an iterator of code points.
376 /// This replaces surrogate code point pairs with supplementary code points,
377 /// like concatenating ill-formed UTF-16 strings effectively would.
378 impl FromIterator<CodePoint> for Wtf8Buf {
379 fn from_iter<T: IntoIterator<Item=CodePoint>>(iter: T) -> Wtf8Buf {
380 let mut string = Wtf8Buf::new();
386 /// Append code points from an iterator to the string.
388 /// This replaces surrogate code point pairs with supplementary code points,
389 /// like concatenating ill-formed UTF-16 strings effectively would.
390 impl Extend<CodePoint> for Wtf8Buf {
391 fn extend<T: IntoIterator<Item=CodePoint>>(&mut self, iter: T) {
392 let iterator = iter.into_iter();
393 let (low, _high) = iterator.size_hint();
394 // Lower bound of one byte per code point (ASCII only)
395 self.bytes.reserve(low);
396 for code_point in iterator {
397 self.push(code_point);
402 /// A borrowed slice of well-formed WTF-8 data.
404 /// Similar to `&str`, but can additionally contain surrogate code points
405 /// if they’re not in a surrogate pair.
406 #[derive(Eq, Ord, PartialEq, PartialOrd)]
411 impl AsInner<[u8]> for Wtf8 {
412 fn as_inner(&self) -> &[u8] { &self.bytes }
415 /// Format the slice with double quotes,
416 /// and surrogates as `\u` followed by four hexadecimal digits.
417 /// Example: `"a\u{D800}"` for a slice with code points [U+0061, U+D800]
418 impl fmt::Debug for Wtf8 {
419 fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
420 fn write_str_escaped(f: &mut fmt::Formatter, s: &str) -> fmt::Result {
422 for c in s.chars().flat_map(|c| c.escape_debug()) {
428 formatter.write_str("\"")?;
430 while let Some((surrogate_pos, surrogate)) = self.next_surrogate(pos) {
433 unsafe { str::from_utf8_unchecked(
434 &self.bytes[pos .. surrogate_pos]
437 write!(formatter, "\\u{{{:x}}}", surrogate)?;
438 pos = surrogate_pos + 3;
442 unsafe { str::from_utf8_unchecked(&self.bytes[pos..]) },
444 formatter.write_str("\"")
448 impl fmt::Display for Wtf8 {
449 fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
450 let wtf8_bytes = &self.bytes;
453 match self.next_surrogate(pos) {
454 Some((surrogate_pos, _)) => {
455 formatter.write_str(unsafe {
456 str::from_utf8_unchecked(&wtf8_bytes[pos .. surrogate_pos])
458 formatter.write_str(UTF8_REPLACEMENT_CHARACTER)?;
459 pos = surrogate_pos + 3;
463 str::from_utf8_unchecked(&wtf8_bytes[pos..])
466 return s.fmt(formatter)
468 return formatter.write_str(s)
477 /// Creates a WTF-8 slice from a UTF-8 `&str` slice.
479 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
481 pub fn from_str(value: &str) -> &Wtf8 {
482 unsafe { Wtf8::from_bytes_unchecked(value.as_bytes()) }
485 /// Creates a WTF-8 slice from a WTF-8 byte slice.
487 /// Since the byte slice is not checked for valid WTF-8, this functions is
490 unsafe fn from_bytes_unchecked(value: &[u8]) -> &Wtf8 {
491 mem::transmute(value)
494 /// Creates a mutable WTF-8 slice from a mutable WTF-8 byte slice.
496 /// Since the byte slice is not checked for valid WTF-8, this functions is
499 unsafe fn from_mut_bytes_unchecked(value: &mut [u8]) -> &mut Wtf8 {
500 mem::transmute(value)
503 /// Returns the length, in WTF-8 bytes.
505 pub fn len(&self) -> usize {
510 pub fn is_empty(&self) -> bool {
511 self.bytes.is_empty()
514 /// Returns the code point at `position` if it is in the ASCII range,
515 /// or `b'\xFF' otherwise.
519 /// Panics if `position` is beyond the end of the string.
521 pub fn ascii_byte_at(&self, position: usize) -> u8 {
522 match self.bytes[position] {
523 ascii_byte @ 0x00 ... 0x7F => ascii_byte,
528 /// Returns an iterator for the string’s code points.
530 pub fn code_points(&self) -> Wtf8CodePoints {
531 Wtf8CodePoints { bytes: self.bytes.iter() }
534 /// Tries to convert the string to UTF-8 and return a `&str` slice.
536 /// Returns `None` if the string contains surrogates.
538 /// This does not copy the data.
540 pub fn as_str(&self) -> Option<&str> {
541 // Well-formed WTF-8 is also well-formed UTF-8
542 // if and only if it contains no surrogate.
543 match self.next_surrogate(0) {
544 None => Some(unsafe { str::from_utf8_unchecked(&self.bytes) }),
549 /// Lossily converts the string to UTF-8.
550 /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8.
552 /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”).
554 /// This only copies the data if necessary (if it contains any surrogate).
555 pub fn to_string_lossy(&self) -> Cow<str> {
556 let surrogate_pos = match self.next_surrogate(0) {
557 None => return Cow::Borrowed(unsafe { str::from_utf8_unchecked(&self.bytes) }),
558 Some((pos, _)) => pos,
560 let wtf8_bytes = &self.bytes;
561 let mut utf8_bytes = Vec::with_capacity(self.len());
562 utf8_bytes.extend_from_slice(&wtf8_bytes[..surrogate_pos]);
563 utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes());
564 let mut pos = surrogate_pos + 3;
566 match self.next_surrogate(pos) {
567 Some((surrogate_pos, _)) => {
568 utf8_bytes.extend_from_slice(&wtf8_bytes[pos .. surrogate_pos]);
569 utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes());
570 pos = surrogate_pos + 3;
573 utf8_bytes.extend_from_slice(&wtf8_bytes[pos..]);
574 return Cow::Owned(unsafe { String::from_utf8_unchecked(utf8_bytes) })
580 /// Converts the WTF-8 string to potentially ill-formed UTF-16
581 /// and return an iterator of 16-bit code units.
583 /// This is lossless:
584 /// calling `Wtf8Buf::from_ill_formed_utf16` on the resulting code units
585 /// would always return the original WTF-8 string.
587 pub fn encode_wide(&self) -> EncodeWide {
588 EncodeWide { code_points: self.code_points(), extra: 0 }
592 fn next_surrogate(&self, mut pos: usize) -> Option<(usize, u16)> {
593 let mut iter = self.bytes[pos..].iter();
595 let b = *iter.next()?;
601 } else if b == 0xED {
602 match (iter.next(), iter.next()) {
603 (Some(&b2), Some(&b3)) if b2 >= 0xA0 => {
604 return Some((pos, decode_surrogate(b2, b3)))
622 fn final_lead_surrogate(&self) -> Option<u16> {
623 let len = self.len();
627 match &self.bytes[(len - 3)..] {
628 &[0xED, b2 @ 0xA0...0xAF, b3] => Some(decode_surrogate(b2, b3)),
634 fn initial_trail_surrogate(&self) -> Option<u16> {
635 let len = self.len();
639 match &self.bytes[..3] {
640 &[0xED, b2 @ 0xB0...0xBF, b3] => Some(decode_surrogate(b2, b3)),
645 /// Boxes this `Wtf8`.
647 pub fn into_box(&self) -> Box<Wtf8> {
648 let boxed: Box<[u8]> = self.bytes.into();
649 unsafe { mem::transmute(boxed) }
652 /// Creates a boxed, empty `Wtf8`.
653 pub fn empty_box() -> Box<Wtf8> {
654 let boxed: Box<[u8]> = Default::default();
655 unsafe { mem::transmute(boxed) }
659 pub fn into_arc(&self) -> Arc<Wtf8> {
660 let arc: Arc<[u8]> = Arc::from(&self.bytes);
661 unsafe { Arc::from_raw(Arc::into_raw(arc) as *const Wtf8) }
665 pub fn into_rc(&self) -> Rc<Wtf8> {
666 let rc: Rc<[u8]> = Rc::from(&self.bytes);
667 unsafe { Rc::from_raw(Rc::into_raw(rc) as *const Wtf8) }
672 /// Return a slice of the given string for the byte range [`begin`..`end`).
676 /// Panics when `begin` and `end` do not point to code point boundaries,
677 /// or point beyond the end of the string.
678 impl ops::Index<ops::Range<usize>> for Wtf8 {
682 fn index(&self, range: ops::Range<usize>) -> &Wtf8 {
683 // is_code_point_boundary checks that the index is in [0, .len()]
684 if range.start <= range.end &&
685 is_code_point_boundary(self, range.start) &&
686 is_code_point_boundary(self, range.end) {
687 unsafe { slice_unchecked(self, range.start, range.end) }
689 slice_error_fail(self, range.start, range.end)
694 /// Return a slice of the given string from byte `begin` to its end.
698 /// Panics when `begin` is not at a code point boundary,
699 /// or is beyond the end of the string.
700 impl ops::Index<ops::RangeFrom<usize>> for Wtf8 {
704 fn index(&self, range: ops::RangeFrom<usize>) -> &Wtf8 {
705 // is_code_point_boundary checks that the index is in [0, .len()]
706 if is_code_point_boundary(self, range.start) {
707 unsafe { slice_unchecked(self, range.start, self.len()) }
709 slice_error_fail(self, range.start, self.len())
714 /// Return a slice of the given string from its beginning to byte `end`.
718 /// Panics when `end` is not at a code point boundary,
719 /// or is beyond the end of the string.
720 impl ops::Index<ops::RangeTo<usize>> for Wtf8 {
724 fn index(&self, range: ops::RangeTo<usize>) -> &Wtf8 {
725 // is_code_point_boundary checks that the index is in [0, .len()]
726 if is_code_point_boundary(self, range.end) {
727 unsafe { slice_unchecked(self, 0, range.end) }
729 slice_error_fail(self, 0, range.end)
734 impl ops::Index<ops::RangeFull> for Wtf8 {
738 fn index(&self, _range: ops::RangeFull) -> &Wtf8 {
744 fn decode_surrogate(second_byte: u8, third_byte: u8) -> u16 {
745 // The first byte is assumed to be 0xED
746 0xD800 | (second_byte as u16 & 0x3F) << 6 | third_byte as u16 & 0x3F
750 fn decode_surrogate_pair(lead: u16, trail: u16) -> char {
751 let code_point = 0x10000 + ((((lead - 0xD800) as u32) << 10) | (trail - 0xDC00) as u32);
752 unsafe { char::from_u32_unchecked(code_point) }
755 /// Copied from core::str::StrPrelude::is_char_boundary
757 pub fn is_code_point_boundary(slice: &Wtf8, index: usize) -> bool {
758 if index == slice.len() { return true; }
759 match slice.bytes.get(index) {
761 Some(&b) => b < 128 || b >= 192,
765 /// Copied from core::str::raw::slice_unchecked
767 pub unsafe fn slice_unchecked(s: &Wtf8, begin: usize, end: usize) -> &Wtf8 {
768 // memory layout of an &[u8] and &Wtf8 are the same
769 Wtf8::from_bytes_unchecked(slice::from_raw_parts(
770 s.bytes.as_ptr().offset(begin as isize),
775 /// Copied from core::str::raw::slice_error_fail
777 pub fn slice_error_fail(s: &Wtf8, begin: usize, end: usize) -> ! {
778 assert!(begin <= end);
779 panic!("index {} and/or {} in `{:?}` do not lie on character boundary",
783 /// Iterator for the code points of a WTF-8 string.
785 /// Created with the method `.code_points()`.
787 pub struct Wtf8CodePoints<'a> {
788 bytes: slice::Iter<'a, u8>
791 impl<'a> Iterator for Wtf8CodePoints<'a> {
792 type Item = CodePoint;
795 fn next(&mut self) -> Option<CodePoint> {
796 next_code_point(&mut self.bytes).map(|c| CodePoint { value: c })
800 fn size_hint(&self) -> (usize, Option<usize>) {
801 let len = self.bytes.len();
802 (len.saturating_add(3) / 4, Some(len))
806 /// Generates a wide character sequence for potentially ill-formed UTF-16.
807 #[stable(feature = "rust1", since = "1.0.0")]
809 pub struct EncodeWide<'a> {
810 code_points: Wtf8CodePoints<'a>,
814 // Copied from libunicode/u_str.rs
815 #[stable(feature = "rust1", since = "1.0.0")]
816 impl<'a> Iterator for EncodeWide<'a> {
820 fn next(&mut self) -> Option<u16> {
822 let tmp = self.extra;
827 let mut buf = [0; 2];
828 self.code_points.next().map(|code_point| {
830 char::from_u32_unchecked(code_point.value)
832 let n = c.encode_utf16(&mut buf).len();
841 fn size_hint(&self) -> (usize, Option<usize>) {
842 let (low, high) = self.code_points.size_hint();
843 // every code point gets either one u16 or two u16,
844 // so this iterator is between 1 or 2 times as
845 // long as the underlying iterator.
846 (low, high.and_then(|n| n.checked_mul(2)))
850 impl Hash for CodePoint {
852 fn hash<H: Hasher>(&self, state: &mut H) {
853 self.value.hash(state)
857 impl Hash for Wtf8Buf {
859 fn hash<H: Hasher>(&self, state: &mut H) {
860 state.write(&self.bytes);
867 fn hash<H: Hasher>(&self, state: &mut H) {
868 state.write(&self.bytes);
874 pub fn is_ascii(&self) -> bool {
875 self.bytes.is_ascii()
877 pub fn to_ascii_uppercase(&self) -> Wtf8Buf {
878 Wtf8Buf { bytes: self.bytes.to_ascii_uppercase() }
880 pub fn to_ascii_lowercase(&self) -> Wtf8Buf {
881 Wtf8Buf { bytes: self.bytes.to_ascii_lowercase() }
883 pub fn eq_ignore_ascii_case(&self, other: &Wtf8) -> bool {
884 self.bytes.eq_ignore_ascii_case(&other.bytes)
887 pub fn make_ascii_uppercase(&mut self) { self.bytes.make_ascii_uppercase() }
888 pub fn make_ascii_lowercase(&mut self) { self.bytes.make_ascii_lowercase() }
897 fn code_point_from_u32() {
898 assert!(CodePoint::from_u32(0).is_some());
899 assert!(CodePoint::from_u32(0xD800).is_some());
900 assert!(CodePoint::from_u32(0x10FFFF).is_some());
901 assert!(CodePoint::from_u32(0x110000).is_none());
905 fn code_point_to_u32() {
906 fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
907 assert_eq!(c(0).to_u32(), 0);
908 assert_eq!(c(0xD800).to_u32(), 0xD800);
909 assert_eq!(c(0x10FFFF).to_u32(), 0x10FFFF);
913 fn code_point_from_char() {
914 assert_eq!(CodePoint::from_char('a').to_u32(), 0x61);
915 assert_eq!(CodePoint::from_char('💩').to_u32(), 0x1F4A9);
919 fn code_point_to_string() {
920 assert_eq!(format!("{:?}", CodePoint::from_char('a')), "U+0061");
921 assert_eq!(format!("{:?}", CodePoint::from_char('💩')), "U+1F4A9");
925 fn code_point_to_char() {
926 fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
927 assert_eq!(c(0x61).to_char(), Some('a'));
928 assert_eq!(c(0x1F4A9).to_char(), Some('💩'));
929 assert_eq!(c(0xD800).to_char(), None);
933 fn code_point_to_char_lossy() {
934 fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
935 assert_eq!(c(0x61).to_char_lossy(), 'a');
936 assert_eq!(c(0x1F4A9).to_char_lossy(), '💩');
937 assert_eq!(c(0xD800).to_char_lossy(), '\u{FFFD}');
942 assert_eq!(Wtf8Buf::new().bytes, b"");
946 fn wtf8buf_from_str() {
947 assert_eq!(Wtf8Buf::from_str("").bytes, b"");
948 assert_eq!(Wtf8Buf::from_str("aé 💩").bytes,
949 b"a\xC3\xA9 \xF0\x9F\x92\xA9");
953 fn wtf8buf_from_string() {
954 assert_eq!(Wtf8Buf::from_string(String::from("")).bytes, b"");
955 assert_eq!(Wtf8Buf::from_string(String::from("aé 💩")).bytes,
956 b"a\xC3\xA9 \xF0\x9F\x92\xA9");
960 fn wtf8buf_from_wide() {
961 assert_eq!(Wtf8Buf::from_wide(&[]).bytes, b"");
962 assert_eq!(Wtf8Buf::from_wide(
963 &[0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]).bytes,
964 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 { CodePoint::from_u32(value).unwrap() }
992 let mut string = Wtf8Buf::new();
993 string.push(c(0xD83D)); // lead
994 string.push(c(0xDCA9)); // trail
995 assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic!
997 let mut string = Wtf8Buf::new();
998 string.push(c(0xD83D)); // lead
999 string.push(c(0x20)); // not surrogate
1000 string.push(c(0xDCA9)); // trail
1001 assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9");
1003 let mut string = Wtf8Buf::new();
1004 string.push(c(0xD800)); // lead
1005 string.push(c(0xDBFF)); // lead
1006 assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF");
1008 let mut string = Wtf8Buf::new();
1009 string.push(c(0xD800)); // lead
1010 string.push(c(0xE000)); // not surrogate
1011 assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80");
1013 let mut string = Wtf8Buf::new();
1014 string.push(c(0xD7FF)); // not surrogate
1015 string.push(c(0xDC00)); // trail
1016 assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80");
1018 let mut string = Wtf8Buf::new();
1019 string.push(c(0x61)); // not surrogate, < 3 bytes
1020 string.push(c(0xDC00)); // trail
1021 assert_eq!(string.bytes, b"\x61\xED\xB0\x80");
1023 let mut string = Wtf8Buf::new();
1024 string.push(c(0xDC00)); // trail
1025 assert_eq!(string.bytes, b"\xED\xB0\x80");
1029 fn wtf8buf_push_wtf8() {
1030 let mut string = Wtf8Buf::from_str("aé");
1031 assert_eq!(string.bytes, b"a\xC3\xA9");
1032 string.push_wtf8(Wtf8::from_str(" 💩"));
1033 assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9");
1035 fn w(v: &[u8]) -> &Wtf8 { unsafe { Wtf8::from_bytes_unchecked(v) } }
1037 let mut string = Wtf8Buf::new();
1038 string.push_wtf8(w(b"\xED\xA0\xBD")); // lead
1039 string.push_wtf8(w(b"\xED\xB2\xA9")); // trail
1040 assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic!
1042 let mut string = Wtf8Buf::new();
1043 string.push_wtf8(w(b"\xED\xA0\xBD")); // lead
1044 string.push_wtf8(w(b" ")); // not surrogate
1045 string.push_wtf8(w(b"\xED\xB2\xA9")); // trail
1046 assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9");
1048 let mut string = Wtf8Buf::new();
1049 string.push_wtf8(w(b"\xED\xA0\x80")); // lead
1050 string.push_wtf8(w(b"\xED\xAF\xBF")); // lead
1051 assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF");
1053 let mut string = Wtf8Buf::new();
1054 string.push_wtf8(w(b"\xED\xA0\x80")); // lead
1055 string.push_wtf8(w(b"\xEE\x80\x80")); // not surrogate
1056 assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80");
1058 let mut string = Wtf8Buf::new();
1059 string.push_wtf8(w(b"\xED\x9F\xBF")); // not surrogate
1060 string.push_wtf8(w(b"\xED\xB0\x80")); // trail
1061 assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80");
1063 let mut string = Wtf8Buf::new();
1064 string.push_wtf8(w(b"a")); // not surrogate, < 3 bytes
1065 string.push_wtf8(w(b"\xED\xB0\x80")); // trail
1066 assert_eq!(string.bytes, b"\x61\xED\xB0\x80");
1068 let mut string = Wtf8Buf::new();
1069 string.push_wtf8(w(b"\xED\xB0\x80")); // trail
1070 assert_eq!(string.bytes, b"\xED\xB0\x80");
1074 fn wtf8buf_truncate() {
1075 let mut string = Wtf8Buf::from_str("aé");
1077 assert_eq!(string.bytes, b"a");
1082 fn wtf8buf_truncate_fail_code_point_boundary() {
1083 let mut string = Wtf8Buf::from_str("aé");
1089 fn wtf8buf_truncate_fail_longer() {
1090 let mut string = Wtf8Buf::from_str("aé");
1095 fn wtf8buf_into_string() {
1096 let mut string = Wtf8Buf::from_str("aé 💩");
1097 assert_eq!(string.clone().into_string(), Ok(String::from("aé 💩")));
1098 string.push(CodePoint::from_u32(0xD800).unwrap());
1099 assert_eq!(string.clone().into_string(), Err(string));
1103 fn wtf8buf_into_string_lossy() {
1104 let mut string = Wtf8Buf::from_str("aé 💩");
1105 assert_eq!(string.clone().into_string_lossy(), String::from("aé 💩"));
1106 string.push(CodePoint::from_u32(0xD800).unwrap());
1107 assert_eq!(string.clone().into_string_lossy(), String::from("aé 💩�"));
1111 fn wtf8buf_from_iterator() {
1112 fn f(values: &[u32]) -> Wtf8Buf {
1113 values.iter().map(|&c| CodePoint::from_u32(c).unwrap()).collect::<Wtf8Buf>()
1115 assert_eq!(f(&[0x61, 0xE9, 0x20, 0x1F4A9]).bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9");
1117 assert_eq!(f(&[0xD83D, 0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic!
1118 assert_eq!(f(&[0xD83D, 0x20, 0xDCA9]).bytes, b"\xED\xA0\xBD \xED\xB2\xA9");
1119 assert_eq!(f(&[0xD800, 0xDBFF]).bytes, b"\xED\xA0\x80\xED\xAF\xBF");
1120 assert_eq!(f(&[0xD800, 0xE000]).bytes, b"\xED\xA0\x80\xEE\x80\x80");
1121 assert_eq!(f(&[0xD7FF, 0xDC00]).bytes, b"\xED\x9F\xBF\xED\xB0\x80");
1122 assert_eq!(f(&[0x61, 0xDC00]).bytes, b"\x61\xED\xB0\x80");
1123 assert_eq!(f(&[0xDC00]).bytes, b"\xED\xB0\x80");
1127 fn wtf8buf_extend() {
1128 fn e(initial: &[u32], extended: &[u32]) -> Wtf8Buf {
1129 fn c(value: &u32) -> CodePoint { CodePoint::from_u32(*value).unwrap() }
1130 let mut string = initial.iter().map(c).collect::<Wtf8Buf>();
1131 string.extend(extended.iter().map(c));
1135 assert_eq!(e(&[0x61, 0xE9], &[0x20, 0x1F4A9]).bytes,
1136 b"a\xC3\xA9 \xF0\x9F\x92\xA9");
1138 assert_eq!(e(&[0xD83D], &[0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic!
1139 assert_eq!(e(&[0xD83D, 0x20], &[0xDCA9]).bytes, b"\xED\xA0\xBD \xED\xB2\xA9");
1140 assert_eq!(e(&[0xD800], &[0xDBFF]).bytes, b"\xED\xA0\x80\xED\xAF\xBF");
1141 assert_eq!(e(&[0xD800], &[0xE000]).bytes, b"\xED\xA0\x80\xEE\x80\x80");
1142 assert_eq!(e(&[0xD7FF], &[0xDC00]).bytes, b"\xED\x9F\xBF\xED\xB0\x80");
1143 assert_eq!(e(&[0x61], &[0xDC00]).bytes, b"\x61\xED\xB0\x80");
1144 assert_eq!(e(&[], &[0xDC00]).bytes, b"\xED\xB0\x80");
1149 let mut string = Wtf8Buf::from_str("a\té \u{7f}💩\r");
1150 string.push(CodePoint::from_u32(0xD800).unwrap());
1151 assert_eq!(format!("{:?}", string), "\"a\\té \\u{7f}\u{1f4a9}\\r\\u{d800}\"");
1155 fn wtf8buf_as_slice() {
1156 assert_eq!(Wtf8Buf::from_str("aé").as_slice(), Wtf8::from_str("aé"));
1160 fn wtf8buf_show_str() {
1161 let text = "a\té 💩\r";
1162 let string = Wtf8Buf::from_str(text);
1163 assert_eq!(format!("{:?}", text), format!("{:?}", string));
1167 fn wtf8_from_str() {
1168 assert_eq!(&Wtf8::from_str("").bytes, b"");
1169 assert_eq!(&Wtf8::from_str("aé 💩").bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9");
1174 assert_eq!(Wtf8::from_str("").len(), 0);
1175 assert_eq!(Wtf8::from_str("aé 💩").len(), 8);
1180 assert_eq!(&Wtf8::from_str("aé 💩")[1.. 4].bytes, b"\xC3\xA9 ");
1185 fn wtf8_slice_not_code_point_boundary() {
1186 &Wtf8::from_str("aé 💩")[2.. 4];
1190 fn wtf8_slice_from() {
1191 assert_eq!(&Wtf8::from_str("aé 💩")[1..].bytes, b"\xC3\xA9 \xF0\x9F\x92\xA9");
1196 fn wtf8_slice_from_not_code_point_boundary() {
1197 &Wtf8::from_str("aé 💩")[2..];
1201 fn wtf8_slice_to() {
1202 assert_eq!(&Wtf8::from_str("aé 💩")[..4].bytes, b"a\xC3\xA9 ");
1207 fn wtf8_slice_to_not_code_point_boundary() {
1208 &Wtf8::from_str("aé 💩")[5..];
1212 fn wtf8_ascii_byte_at() {
1213 let slice = Wtf8::from_str("aé 💩");
1214 assert_eq!(slice.ascii_byte_at(0), b'a');
1215 assert_eq!(slice.ascii_byte_at(1), b'\xFF');
1216 assert_eq!(slice.ascii_byte_at(2), b'\xFF');
1217 assert_eq!(slice.ascii_byte_at(3), b' ');
1218 assert_eq!(slice.ascii_byte_at(4), b'\xFF');
1222 fn wtf8_code_points() {
1223 fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
1224 fn cp(string: &Wtf8Buf) -> Vec<Option<char>> {
1225 string.code_points().map(|c| c.to_char()).collect::<Vec<_>>()
1227 let mut string = Wtf8Buf::from_str("é ");
1228 assert_eq!(cp(&string), [Some('é'), Some(' ')]);
1229 string.push(c(0xD83D));
1230 assert_eq!(cp(&string), [Some('é'), Some(' '), None]);
1231 string.push(c(0xDCA9));
1232 assert_eq!(cp(&string), [Some('é'), Some(' '), Some('💩')]);
1237 assert_eq!(Wtf8::from_str("").as_str(), Some(""));
1238 assert_eq!(Wtf8::from_str("aé 💩").as_str(), Some("aé 💩"));
1239 let mut string = Wtf8Buf::new();
1240 string.push(CodePoint::from_u32(0xD800).unwrap());
1241 assert_eq!(string.as_str(), None);
1245 fn wtf8_to_string_lossy() {
1246 assert_eq!(Wtf8::from_str("").to_string_lossy(), Cow::Borrowed(""));
1247 assert_eq!(Wtf8::from_str("aé 💩").to_string_lossy(), Cow::Borrowed("aé 💩"));
1248 let mut string = Wtf8Buf::from_str("aé 💩");
1249 string.push(CodePoint::from_u32(0xD800).unwrap());
1250 let expected: Cow<str> = Cow::Owned(String::from("aé 💩�"));
1251 assert_eq!(string.to_string_lossy(), expected);
1256 fn d(b: &[u8]) -> String {
1257 format!("{}", &unsafe { Wtf8::from_bytes_unchecked(b) })
1260 assert_eq!("", d("".as_bytes()));
1261 assert_eq!("aé 💩", d("aé 💩".as_bytes()));
1263 let mut string = Wtf8Buf::from_str("aé 💩");
1264 string.push(CodePoint::from_u32(0xD800).unwrap());
1265 assert_eq!("aé 💩�", d(string.as_inner()));
1269 fn wtf8_encode_wide() {
1270 let mut string = Wtf8Buf::from_str("aé ");
1271 string.push(CodePoint::from_u32(0xD83D).unwrap());
1272 string.push_char('💩');
1273 assert_eq!(string.encode_wide().collect::<Vec<_>>(),
1274 vec![0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]);