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;
23 use hash::{Hash, Hasher};
24 use iter::FromIterator;
31 use 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 `n` bytes.
151 pub fn with_capacity(n: usize) -> Wtf8Buf {
152 Wtf8Buf { bytes: Vec::with_capacity(n) }
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 {
192 CodePoint::from_u32_unchecked(surrogate as u32)
194 // Skip the WTF-8 concatenation check,
195 // surrogate pairs are already decoded by decode_utf16
196 string.push_code_point_unchecked(code_point)
203 /// Copied from String::push
204 /// This does **not** include the WTF-8 concatenation check.
205 fn push_code_point_unchecked(&mut self, code_point: CodePoint) {
207 char::from_u32_unchecked(code_point.value)
209 let mut bytes = [0; 4];
210 let bytes = c.encode_utf8(&mut bytes).as_bytes();
211 self.bytes.extend_from_slice(bytes)
215 pub fn as_slice(&self) -> &Wtf8 {
216 unsafe { Wtf8::from_bytes_unchecked(&self.bytes) }
220 pub fn as_mut_slice(&mut self) -> &mut Wtf8 {
221 unsafe { Wtf8::from_mut_bytes_unchecked(&mut self.bytes) }
224 /// Reserves capacity for at least `additional` more bytes to be inserted
225 /// in the given `Wtf8Buf`.
226 /// The collection may reserve more space to avoid frequent reallocations.
230 /// Panics if the new capacity overflows `usize`.
232 pub fn reserve(&mut self, additional: usize) {
233 self.bytes.reserve(additional)
237 pub fn reserve_exact(&mut self, additional: usize) {
238 self.bytes.reserve_exact(additional)
242 pub fn shrink_to_fit(&mut self) {
243 self.bytes.shrink_to_fit()
247 pub fn shrink_to(&mut self, min_capacity: usize) {
248 self.bytes.shrink_to(min_capacity)
251 /// Returns the number of bytes that this string buffer can hold without reallocating.
253 pub fn capacity(&self) -> usize {
254 self.bytes.capacity()
257 /// Append a UTF-8 slice at the end of the string.
259 pub fn push_str(&mut self, other: &str) {
260 self.bytes.extend_from_slice(other.as_bytes())
263 /// Append a WTF-8 slice at the end of the string.
265 /// This replaces newly paired surrogates at the boundary
266 /// with a supplementary code point,
267 /// like concatenating ill-formed UTF-16 strings effectively would.
269 pub fn push_wtf8(&mut self, other: &Wtf8) {
270 match ((&*self).final_lead_surrogate(), other.initial_trail_surrogate()) {
271 // Replace newly paired surrogates by a supplementary code point.
272 (Some(lead), Some(trail)) => {
273 let len_without_lead_surrogate = self.len() - 3;
274 self.bytes.truncate(len_without_lead_surrogate);
275 let other_without_trail_surrogate = &other.bytes[3..];
276 // 4 bytes for the supplementary code point
277 self.bytes.reserve(4 + other_without_trail_surrogate.len());
278 self.push_char(decode_surrogate_pair(lead, trail));
279 self.bytes.extend_from_slice(other_without_trail_surrogate);
281 _ => self.bytes.extend_from_slice(&other.bytes)
285 /// Append a Unicode scalar value at the end of the string.
287 pub fn push_char(&mut self, c: char) {
288 self.push_code_point_unchecked(CodePoint::from_char(c))
291 /// Append a code point at the end of the string.
293 /// This replaces newly paired surrogates at the boundary
294 /// with a supplementary code point,
295 /// like concatenating ill-formed UTF-16 strings effectively would.
297 pub fn push(&mut self, code_point: CodePoint) {
298 if let trail @ 0xDC00..=0xDFFF = code_point.to_u32() {
299 if let Some(lead) = (&*self).final_lead_surrogate() {
300 let len_without_lead_surrogate = self.len() - 3;
301 self.bytes.truncate(len_without_lead_surrogate);
302 self.push_char(decode_surrogate_pair(lead, trail as u16));
307 // No newly paired surrogates at the boundary.
308 self.push_code_point_unchecked(code_point)
311 /// Shortens a string to the specified length.
315 /// Panics if `new_len` > current length,
316 /// or if `new_len` is not a code point boundary.
318 pub fn truncate(&mut self, new_len: usize) {
319 assert!(is_code_point_boundary(self, new_len));
320 self.bytes.truncate(new_len)
323 /// Consumes the WTF-8 string and tries to convert it to UTF-8.
325 /// This does not copy the data.
327 /// If the contents are not well-formed UTF-8
328 /// (that is, if the string contains surrogates),
329 /// the original WTF-8 string is returned instead.
330 pub fn into_string(self) -> Result<String, Wtf8Buf> {
331 match self.next_surrogate(0) {
332 None => Ok(unsafe { String::from_utf8_unchecked(self.bytes) }),
333 Some(_) => Err(self),
337 /// Consumes the WTF-8 string and converts it lossily to UTF-8.
339 /// This does not copy the data (but may overwrite parts of it in place).
341 /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”)
342 pub fn into_string_lossy(mut self) -> String {
345 match self.next_surrogate(pos) {
346 Some((surrogate_pos, _)) => {
347 pos = surrogate_pos + 3;
348 self.bytes[surrogate_pos..pos]
349 .copy_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes());
351 None => return unsafe { String::from_utf8_unchecked(self.bytes) }
356 /// Converts this `Wtf8Buf` into a boxed `Wtf8`.
358 pub fn into_box(self) -> Box<Wtf8> {
359 unsafe { mem::transmute(self.bytes.into_boxed_slice()) }
362 /// Converts a `Box<Wtf8>` into a `Wtf8Buf`.
363 pub fn from_box(boxed: Box<Wtf8>) -> Wtf8Buf {
364 let bytes: Box<[u8]> = unsafe { mem::transmute(boxed) };
365 Wtf8Buf { bytes: bytes.into_vec() }
369 /// Creates a new WTF-8 string from an iterator of code points.
371 /// This replaces surrogate code point pairs with supplementary code points,
372 /// like concatenating ill-formed UTF-16 strings effectively would.
373 impl FromIterator<CodePoint> for Wtf8Buf {
374 fn from_iter<T: IntoIterator<Item=CodePoint>>(iter: T) -> Wtf8Buf {
375 let mut string = Wtf8Buf::new();
381 /// Append code points from an iterator to the string.
383 /// This replaces surrogate code point pairs with supplementary code points,
384 /// like concatenating ill-formed UTF-16 strings effectively would.
385 impl Extend<CodePoint> for Wtf8Buf {
386 fn extend<T: IntoIterator<Item=CodePoint>>(&mut self, iter: T) {
387 let iterator = iter.into_iter();
388 let (low, _high) = iterator.size_hint();
389 // Lower bound of one byte per code point (ASCII only)
390 self.bytes.reserve(low);
391 for code_point in iterator {
392 self.push(code_point);
397 /// A borrowed slice of well-formed WTF-8 data.
399 /// Similar to `&str`, but can additionally contain surrogate code points
400 /// if they’re not in a surrogate pair.
401 #[derive(Eq, Ord, PartialEq, PartialOrd)]
406 impl AsInner<[u8]> for Wtf8 {
407 fn as_inner(&self) -> &[u8] { &self.bytes }
410 /// Format the slice with double quotes,
411 /// and surrogates as `\u` followed by four hexadecimal digits.
412 /// Example: `"a\u{D800}"` for a slice with code points [U+0061, U+D800]
413 impl fmt::Debug for Wtf8 {
414 fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
415 fn write_str_escaped(f: &mut fmt::Formatter, s: &str) -> fmt::Result {
417 for c in s.chars().flat_map(|c| c.escape_debug()) {
423 formatter.write_str("\"")?;
425 while let Some((surrogate_pos, surrogate)) = self.next_surrogate(pos) {
428 unsafe { str::from_utf8_unchecked(
429 &self.bytes[pos .. surrogate_pos]
432 write!(formatter, "\\u{{{:x}}}", surrogate)?;
433 pos = surrogate_pos + 3;
437 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;
458 str::from_utf8_unchecked(&wtf8_bytes[pos..])
461 return s.fmt(formatter)
463 return formatter.write_str(s)
472 /// Creates a WTF-8 slice from a UTF-8 `&str` slice.
474 /// Since WTF-8 is a superset of UTF-8, this always succeeds.
476 pub fn from_str(value: &str) -> &Wtf8 {
477 unsafe { Wtf8::from_bytes_unchecked(value.as_bytes()) }
480 /// Creates a WTF-8 slice from a WTF-8 byte slice.
482 /// Since the byte slice is not checked for valid WTF-8, this functions is
485 unsafe fn from_bytes_unchecked(value: &[u8]) -> &Wtf8 {
486 mem::transmute(value)
489 /// Creates a mutable WTF-8 slice from a mutable WTF-8 byte slice.
491 /// Since the byte slice is not checked for valid WTF-8, this functions is
494 unsafe fn from_mut_bytes_unchecked(value: &mut [u8]) -> &mut Wtf8 {
495 mem::transmute(value)
498 /// Returns the length, in WTF-8 bytes.
500 pub fn len(&self) -> usize {
505 pub fn is_empty(&self) -> bool {
506 self.bytes.is_empty()
509 /// Returns the code point at `position` if it is in the ASCII range,
510 /// or `b'\xFF' otherwise.
514 /// Panics if `position` is beyond the end of the string.
516 pub fn ascii_byte_at(&self, position: usize) -> u8 {
517 match self.bytes[position] {
518 ascii_byte @ 0x00 ..= 0x7F => ascii_byte,
523 /// Returns an iterator for the string’s code points.
525 pub fn code_points(&self) -> Wtf8CodePoints {
526 Wtf8CodePoints { bytes: self.bytes.iter() }
529 /// Tries to convert the string to UTF-8 and return a `&str` slice.
531 /// Returns `None` if the string contains surrogates.
533 /// This does not copy the data.
535 pub fn as_str(&self) -> Option<&str> {
536 // Well-formed WTF-8 is also well-formed UTF-8
537 // if and only if it contains no surrogate.
538 match self.next_surrogate(0) {
539 None => Some(unsafe { str::from_utf8_unchecked(&self.bytes) }),
544 /// Lossily converts the string to UTF-8.
545 /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8.
547 /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character “�”).
549 /// This only copies the data if necessary (if it contains any surrogate).
550 pub fn to_string_lossy(&self) -> Cow<str> {
551 let surrogate_pos = match self.next_surrogate(0) {
552 None => return Cow::Borrowed(unsafe { str::from_utf8_unchecked(&self.bytes) }),
553 Some((pos, _)) => pos,
555 let wtf8_bytes = &self.bytes;
556 let mut utf8_bytes = Vec::with_capacity(self.len());
557 utf8_bytes.extend_from_slice(&wtf8_bytes[..surrogate_pos]);
558 utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes());
559 let mut pos = surrogate_pos + 3;
561 match self.next_surrogate(pos) {
562 Some((surrogate_pos, _)) => {
563 utf8_bytes.extend_from_slice(&wtf8_bytes[pos .. surrogate_pos]);
564 utf8_bytes.extend_from_slice(UTF8_REPLACEMENT_CHARACTER.as_bytes());
565 pos = surrogate_pos + 3;
568 utf8_bytes.extend_from_slice(&wtf8_bytes[pos..]);
569 return Cow::Owned(unsafe { String::from_utf8_unchecked(utf8_bytes) })
575 /// Converts the WTF-8 string to potentially ill-formed UTF-16
576 /// and return an iterator of 16-bit code units.
578 /// This is lossless:
579 /// calling `Wtf8Buf::from_ill_formed_utf16` on the resulting code units
580 /// would always return the original WTF-8 string.
582 pub fn encode_wide(&self) -> EncodeWide {
583 EncodeWide { code_points: self.code_points(), extra: 0 }
587 fn next_surrogate(&self, mut pos: usize) -> Option<(usize, u16)> {
588 let mut iter = self.bytes[pos..].iter();
590 let b = *iter.next()?;
596 } else if b == 0xED {
597 match (iter.next(), iter.next()) {
598 (Some(&b2), Some(&b3)) if b2 >= 0xA0 => {
599 return Some((pos, decode_surrogate(b2, b3)))
617 fn final_lead_surrogate(&self) -> Option<u16> {
618 let len = self.len();
622 match &self.bytes[(len - 3)..] {
623 &[0xED, b2 @ 0xA0..=0xAF, b3] => Some(decode_surrogate(b2, b3)),
629 fn initial_trail_surrogate(&self) -> Option<u16> {
630 let len = self.len();
634 match &self.bytes[..3] {
635 &[0xED, b2 @ 0xB0..=0xBF, b3] => Some(decode_surrogate(b2, b3)),
640 /// Boxes this `Wtf8`.
642 pub fn into_box(&self) -> Box<Wtf8> {
643 let boxed: Box<[u8]> = self.bytes.into();
644 unsafe { mem::transmute(boxed) }
647 /// Creates a boxed, empty `Wtf8`.
648 pub fn empty_box() -> Box<Wtf8> {
649 let boxed: Box<[u8]> = Default::default();
650 unsafe { mem::transmute(boxed) }
654 pub fn into_arc(&self) -> Arc<Wtf8> {
655 let arc: Arc<[u8]> = Arc::from(&self.bytes);
656 unsafe { Arc::from_raw(Arc::into_raw(arc) as *const Wtf8) }
660 pub fn into_rc(&self) -> Rc<Wtf8> {
661 let rc: Rc<[u8]> = Rc::from(&self.bytes);
662 unsafe { Rc::from_raw(Rc::into_raw(rc) as *const Wtf8) }
667 /// Returns a slice of the given string for the byte range [`begin`..`end`).
671 /// Panics when `begin` and `end` do not point to code point boundaries,
672 /// or point beyond the end of the string.
673 impl ops::Index<ops::Range<usize>> for Wtf8 {
677 fn index(&self, range: ops::Range<usize>) -> &Wtf8 {
678 // is_code_point_boundary checks that the index is in [0, .len()]
679 if range.start <= range.end &&
680 is_code_point_boundary(self, range.start) &&
681 is_code_point_boundary(self, range.end) {
682 unsafe { slice_unchecked(self, range.start, range.end) }
684 slice_error_fail(self, range.start, range.end)
689 /// Returns a slice of the given string from byte `begin` to its end.
693 /// Panics when `begin` is not at a code point boundary,
694 /// or is beyond the end of the string.
695 impl ops::Index<ops::RangeFrom<usize>> for Wtf8 {
699 fn index(&self, range: ops::RangeFrom<usize>) -> &Wtf8 {
700 // is_code_point_boundary checks that the index is in [0, .len()]
701 if is_code_point_boundary(self, range.start) {
702 unsafe { slice_unchecked(self, range.start, self.len()) }
704 slice_error_fail(self, range.start, self.len())
709 /// Returns a slice of the given string from its beginning to byte `end`.
713 /// Panics when `end` is not at a code point boundary,
714 /// or is beyond the end of the string.
715 impl ops::Index<ops::RangeTo<usize>> for Wtf8 {
719 fn index(&self, range: ops::RangeTo<usize>) -> &Wtf8 {
720 // is_code_point_boundary checks that the index is in [0, .len()]
721 if is_code_point_boundary(self, range.end) {
722 unsafe { slice_unchecked(self, 0, range.end) }
724 slice_error_fail(self, 0, range.end)
729 impl ops::Index<ops::RangeFull> for Wtf8 {
733 fn index(&self, _range: ops::RangeFull) -> &Wtf8 {
739 fn decode_surrogate(second_byte: u8, third_byte: u8) -> u16 {
740 // The first byte is assumed to be 0xED
741 0xD800 | (second_byte as u16 & 0x3F) << 6 | third_byte as u16 & 0x3F
745 fn decode_surrogate_pair(lead: u16, trail: u16) -> char {
746 let code_point = 0x10000 + ((((lead - 0xD800) as u32) << 10) | (trail - 0xDC00) as u32);
747 unsafe { char::from_u32_unchecked(code_point) }
750 /// Copied from core::str::StrPrelude::is_char_boundary
752 pub fn is_code_point_boundary(slice: &Wtf8, index: usize) -> bool {
753 if index == slice.len() { return true; }
754 match slice.bytes.get(index) {
756 Some(&b) => b < 128 || b >= 192,
760 /// Copied from core::str::raw::slice_unchecked
762 pub unsafe fn slice_unchecked(s: &Wtf8, begin: usize, end: usize) -> &Wtf8 {
763 // memory layout of an &[u8] and &Wtf8 are the same
764 Wtf8::from_bytes_unchecked(slice::from_raw_parts(
765 s.bytes.as_ptr().add(begin),
770 /// Copied from core::str::raw::slice_error_fail
772 pub fn slice_error_fail(s: &Wtf8, begin: usize, end: usize) -> ! {
773 assert!(begin <= end);
774 panic!("index {} and/or {} in `{:?}` do not lie on character boundary",
778 /// Iterator for the code points of a WTF-8 string.
780 /// Created with the method `.code_points()`.
782 pub struct Wtf8CodePoints<'a> {
783 bytes: slice::Iter<'a, u8>
786 impl<'a> Iterator for Wtf8CodePoints<'a> {
787 type Item = CodePoint;
790 fn next(&mut self) -> Option<CodePoint> {
791 next_code_point(&mut self.bytes).map(|c| CodePoint { value: c })
795 fn size_hint(&self) -> (usize, Option<usize>) {
796 let len = self.bytes.len();
797 (len.saturating_add(3) / 4, Some(len))
801 /// Generates a wide character sequence for potentially ill-formed UTF-16.
802 #[stable(feature = "rust1", since = "1.0.0")]
804 pub struct EncodeWide<'a> {
805 code_points: Wtf8CodePoints<'a>,
809 // Copied from libunicode/u_str.rs
810 #[stable(feature = "rust1", since = "1.0.0")]
811 impl<'a> Iterator for EncodeWide<'a> {
815 fn next(&mut self) -> Option<u16> {
817 let tmp = self.extra;
822 let mut buf = [0; 2];
823 self.code_points.next().map(|code_point| {
825 char::from_u32_unchecked(code_point.value)
827 let n = c.encode_utf16(&mut buf).len();
836 fn size_hint(&self) -> (usize, Option<usize>) {
837 let (low, high) = self.code_points.size_hint();
838 // every code point gets either one u16 or two u16,
839 // so this iterator is between 1 or 2 times as
840 // long as the underlying iterator.
841 (low, high.and_then(|n| n.checked_mul(2)))
845 impl Hash for CodePoint {
847 fn hash<H: Hasher>(&self, state: &mut H) {
848 self.value.hash(state)
852 impl Hash for Wtf8Buf {
854 fn hash<H: Hasher>(&self, state: &mut H) {
855 state.write(&self.bytes);
862 fn hash<H: Hasher>(&self, state: &mut H) {
863 state.write(&self.bytes);
869 pub fn make_ascii_uppercase(&mut self) { self.bytes.make_ascii_uppercase() }
878 fn code_point_from_u32() {
879 assert!(CodePoint::from_u32(0).is_some());
880 assert!(CodePoint::from_u32(0xD800).is_some());
881 assert!(CodePoint::from_u32(0x10FFFF).is_some());
882 assert!(CodePoint::from_u32(0x110000).is_none());
886 fn code_point_to_u32() {
887 fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
888 assert_eq!(c(0).to_u32(), 0);
889 assert_eq!(c(0xD800).to_u32(), 0xD800);
890 assert_eq!(c(0x10FFFF).to_u32(), 0x10FFFF);
894 fn code_point_from_char() {
895 assert_eq!(CodePoint::from_char('a').to_u32(), 0x61);
896 assert_eq!(CodePoint::from_char('💩').to_u32(), 0x1F4A9);
900 fn code_point_to_string() {
901 assert_eq!(format!("{:?}", CodePoint::from_char('a')), "U+0061");
902 assert_eq!(format!("{:?}", CodePoint::from_char('💩')), "U+1F4A9");
906 fn code_point_to_char() {
907 fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
908 assert_eq!(c(0x61).to_char(), Some('a'));
909 assert_eq!(c(0x1F4A9).to_char(), Some('💩'));
910 assert_eq!(c(0xD800).to_char(), None);
914 fn code_point_to_char_lossy() {
915 fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
916 assert_eq!(c(0x61).to_char_lossy(), 'a');
917 assert_eq!(c(0x1F4A9).to_char_lossy(), '💩');
918 assert_eq!(c(0xD800).to_char_lossy(), '\u{FFFD}');
923 assert_eq!(Wtf8Buf::new().bytes, b"");
927 fn wtf8buf_from_str() {
928 assert_eq!(Wtf8Buf::from_str("").bytes, b"");
929 assert_eq!(Wtf8Buf::from_str("aé 💩").bytes,
930 b"a\xC3\xA9 \xF0\x9F\x92\xA9");
934 fn wtf8buf_from_string() {
935 assert_eq!(Wtf8Buf::from_string(String::from("")).bytes, b"");
936 assert_eq!(Wtf8Buf::from_string(String::from("aé 💩")).bytes,
937 b"a\xC3\xA9 \xF0\x9F\x92\xA9");
941 fn wtf8buf_from_wide() {
942 assert_eq!(Wtf8Buf::from_wide(&[]).bytes, b"");
943 assert_eq!(Wtf8Buf::from_wide(
944 &[0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]).bytes,
945 b"a\xC3\xA9 \xED\xA0\xBD\xF0\x9F\x92\xA9");
949 fn wtf8buf_push_str() {
950 let mut string = Wtf8Buf::new();
951 assert_eq!(string.bytes, b"");
952 string.push_str("aé 💩");
953 assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9");
957 fn wtf8buf_push_char() {
958 let mut string = Wtf8Buf::from_str("aé ");
959 assert_eq!(string.bytes, b"a\xC3\xA9 ");
960 string.push_char('💩');
961 assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9");
966 let mut string = Wtf8Buf::from_str("aé ");
967 assert_eq!(string.bytes, b"a\xC3\xA9 ");
968 string.push(CodePoint::from_char('💩'));
969 assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9");
971 fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
973 let mut string = Wtf8Buf::new();
974 string.push(c(0xD83D)); // lead
975 string.push(c(0xDCA9)); // trail
976 assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic!
978 let mut string = Wtf8Buf::new();
979 string.push(c(0xD83D)); // lead
980 string.push(c(0x20)); // not surrogate
981 string.push(c(0xDCA9)); // trail
982 assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9");
984 let mut string = Wtf8Buf::new();
985 string.push(c(0xD800)); // lead
986 string.push(c(0xDBFF)); // lead
987 assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF");
989 let mut string = Wtf8Buf::new();
990 string.push(c(0xD800)); // lead
991 string.push(c(0xE000)); // not surrogate
992 assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80");
994 let mut string = Wtf8Buf::new();
995 string.push(c(0xD7FF)); // not surrogate
996 string.push(c(0xDC00)); // trail
997 assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80");
999 let mut string = Wtf8Buf::new();
1000 string.push(c(0x61)); // not surrogate, < 3 bytes
1001 string.push(c(0xDC00)); // trail
1002 assert_eq!(string.bytes, b"\x61\xED\xB0\x80");
1004 let mut string = Wtf8Buf::new();
1005 string.push(c(0xDC00)); // trail
1006 assert_eq!(string.bytes, b"\xED\xB0\x80");
1010 fn wtf8buf_push_wtf8() {
1011 let mut string = Wtf8Buf::from_str("aé");
1012 assert_eq!(string.bytes, b"a\xC3\xA9");
1013 string.push_wtf8(Wtf8::from_str(" 💩"));
1014 assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9");
1016 fn w(v: &[u8]) -> &Wtf8 { unsafe { Wtf8::from_bytes_unchecked(v) } }
1018 let mut string = Wtf8Buf::new();
1019 string.push_wtf8(w(b"\xED\xA0\xBD")); // lead
1020 string.push_wtf8(w(b"\xED\xB2\xA9")); // trail
1021 assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic!
1023 let mut string = Wtf8Buf::new();
1024 string.push_wtf8(w(b"\xED\xA0\xBD")); // lead
1025 string.push_wtf8(w(b" ")); // not surrogate
1026 string.push_wtf8(w(b"\xED\xB2\xA9")); // trail
1027 assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9");
1029 let mut string = Wtf8Buf::new();
1030 string.push_wtf8(w(b"\xED\xA0\x80")); // lead
1031 string.push_wtf8(w(b"\xED\xAF\xBF")); // lead
1032 assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF");
1034 let mut string = Wtf8Buf::new();
1035 string.push_wtf8(w(b"\xED\xA0\x80")); // lead
1036 string.push_wtf8(w(b"\xEE\x80\x80")); // not surrogate
1037 assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80");
1039 let mut string = Wtf8Buf::new();
1040 string.push_wtf8(w(b"\xED\x9F\xBF")); // not surrogate
1041 string.push_wtf8(w(b"\xED\xB0\x80")); // trail
1042 assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80");
1044 let mut string = Wtf8Buf::new();
1045 string.push_wtf8(w(b"a")); // not surrogate, < 3 bytes
1046 string.push_wtf8(w(b"\xED\xB0\x80")); // trail
1047 assert_eq!(string.bytes, b"\x61\xED\xB0\x80");
1049 let mut string = Wtf8Buf::new();
1050 string.push_wtf8(w(b"\xED\xB0\x80")); // trail
1051 assert_eq!(string.bytes, b"\xED\xB0\x80");
1055 fn wtf8buf_truncate() {
1056 let mut string = Wtf8Buf::from_str("aé");
1058 assert_eq!(string.bytes, b"a");
1063 fn wtf8buf_truncate_fail_code_point_boundary() {
1064 let mut string = Wtf8Buf::from_str("aé");
1070 fn wtf8buf_truncate_fail_longer() {
1071 let mut string = Wtf8Buf::from_str("aé");
1076 fn wtf8buf_into_string() {
1077 let mut string = Wtf8Buf::from_str("aé 💩");
1078 assert_eq!(string.clone().into_string(), Ok(String::from("aé 💩")));
1079 string.push(CodePoint::from_u32(0xD800).unwrap());
1080 assert_eq!(string.clone().into_string(), Err(string));
1084 fn wtf8buf_into_string_lossy() {
1085 let mut string = Wtf8Buf::from_str("aé 💩");
1086 assert_eq!(string.clone().into_string_lossy(), String::from("aé 💩"));
1087 string.push(CodePoint::from_u32(0xD800).unwrap());
1088 assert_eq!(string.clone().into_string_lossy(), String::from("aé 💩�"));
1092 fn wtf8buf_from_iterator() {
1093 fn f(values: &[u32]) -> Wtf8Buf {
1094 values.iter().map(|&c| CodePoint::from_u32(c).unwrap()).collect::<Wtf8Buf>()
1096 assert_eq!(f(&[0x61, 0xE9, 0x20, 0x1F4A9]).bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9");
1098 assert_eq!(f(&[0xD83D, 0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic!
1099 assert_eq!(f(&[0xD83D, 0x20, 0xDCA9]).bytes, b"\xED\xA0\xBD \xED\xB2\xA9");
1100 assert_eq!(f(&[0xD800, 0xDBFF]).bytes, b"\xED\xA0\x80\xED\xAF\xBF");
1101 assert_eq!(f(&[0xD800, 0xE000]).bytes, b"\xED\xA0\x80\xEE\x80\x80");
1102 assert_eq!(f(&[0xD7FF, 0xDC00]).bytes, b"\xED\x9F\xBF\xED\xB0\x80");
1103 assert_eq!(f(&[0x61, 0xDC00]).bytes, b"\x61\xED\xB0\x80");
1104 assert_eq!(f(&[0xDC00]).bytes, b"\xED\xB0\x80");
1108 fn wtf8buf_extend() {
1109 fn e(initial: &[u32], extended: &[u32]) -> Wtf8Buf {
1110 fn c(value: &u32) -> CodePoint { CodePoint::from_u32(*value).unwrap() }
1111 let mut string = initial.iter().map(c).collect::<Wtf8Buf>();
1112 string.extend(extended.iter().map(c));
1116 assert_eq!(e(&[0x61, 0xE9], &[0x20, 0x1F4A9]).bytes,
1117 b"a\xC3\xA9 \xF0\x9F\x92\xA9");
1119 assert_eq!(e(&[0xD83D], &[0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic!
1120 assert_eq!(e(&[0xD83D, 0x20], &[0xDCA9]).bytes, b"\xED\xA0\xBD \xED\xB2\xA9");
1121 assert_eq!(e(&[0xD800], &[0xDBFF]).bytes, b"\xED\xA0\x80\xED\xAF\xBF");
1122 assert_eq!(e(&[0xD800], &[0xE000]).bytes, b"\xED\xA0\x80\xEE\x80\x80");
1123 assert_eq!(e(&[0xD7FF], &[0xDC00]).bytes, b"\xED\x9F\xBF\xED\xB0\x80");
1124 assert_eq!(e(&[0x61], &[0xDC00]).bytes, b"\x61\xED\xB0\x80");
1125 assert_eq!(e(&[], &[0xDC00]).bytes, b"\xED\xB0\x80");
1130 let mut string = Wtf8Buf::from_str("a\té \u{7f}💩\r");
1131 string.push(CodePoint::from_u32(0xD800).unwrap());
1132 assert_eq!(format!("{:?}", string), "\"a\\té \\u{7f}\u{1f4a9}\\r\\u{d800}\"");
1136 fn wtf8buf_as_slice() {
1137 assert_eq!(Wtf8Buf::from_str("aé").as_slice(), Wtf8::from_str("aé"));
1141 fn wtf8buf_show_str() {
1142 let text = "a\té 💩\r";
1143 let string = Wtf8Buf::from_str(text);
1144 assert_eq!(format!("{:?}", text), format!("{:?}", string));
1148 fn wtf8_from_str() {
1149 assert_eq!(&Wtf8::from_str("").bytes, b"");
1150 assert_eq!(&Wtf8::from_str("aé 💩").bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9");
1155 assert_eq!(Wtf8::from_str("").len(), 0);
1156 assert_eq!(Wtf8::from_str("aé 💩").len(), 8);
1161 assert_eq!(&Wtf8::from_str("aé 💩")[1.. 4].bytes, b"\xC3\xA9 ");
1166 fn wtf8_slice_not_code_point_boundary() {
1167 &Wtf8::from_str("aé 💩")[2.. 4];
1171 fn wtf8_slice_from() {
1172 assert_eq!(&Wtf8::from_str("aé 💩")[1..].bytes, b"\xC3\xA9 \xF0\x9F\x92\xA9");
1177 fn wtf8_slice_from_not_code_point_boundary() {
1178 &Wtf8::from_str("aé 💩")[2..];
1182 fn wtf8_slice_to() {
1183 assert_eq!(&Wtf8::from_str("aé 💩")[..4].bytes, b"a\xC3\xA9 ");
1188 fn wtf8_slice_to_not_code_point_boundary() {
1189 &Wtf8::from_str("aé 💩")[5..];
1193 fn wtf8_ascii_byte_at() {
1194 let slice = Wtf8::from_str("aé 💩");
1195 assert_eq!(slice.ascii_byte_at(0), b'a');
1196 assert_eq!(slice.ascii_byte_at(1), b'\xFF');
1197 assert_eq!(slice.ascii_byte_at(2), b'\xFF');
1198 assert_eq!(slice.ascii_byte_at(3), b' ');
1199 assert_eq!(slice.ascii_byte_at(4), b'\xFF');
1203 fn wtf8_code_points() {
1204 fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() }
1205 fn cp(string: &Wtf8Buf) -> Vec<Option<char>> {
1206 string.code_points().map(|c| c.to_char()).collect::<Vec<_>>()
1208 let mut string = Wtf8Buf::from_str("é ");
1209 assert_eq!(cp(&string), [Some('é'), Some(' ')]);
1210 string.push(c(0xD83D));
1211 assert_eq!(cp(&string), [Some('é'), Some(' '), None]);
1212 string.push(c(0xDCA9));
1213 assert_eq!(cp(&string), [Some('é'), Some(' '), Some('💩')]);
1218 assert_eq!(Wtf8::from_str("").as_str(), Some(""));
1219 assert_eq!(Wtf8::from_str("aé 💩").as_str(), Some("aé 💩"));
1220 let mut string = Wtf8Buf::new();
1221 string.push(CodePoint::from_u32(0xD800).unwrap());
1222 assert_eq!(string.as_str(), None);
1226 fn wtf8_to_string_lossy() {
1227 assert_eq!(Wtf8::from_str("").to_string_lossy(), Cow::Borrowed(""));
1228 assert_eq!(Wtf8::from_str("aé 💩").to_string_lossy(), Cow::Borrowed("aé 💩"));
1229 let mut string = Wtf8Buf::from_str("aé 💩");
1230 string.push(CodePoint::from_u32(0xD800).unwrap());
1231 let expected: Cow<str> = Cow::Owned(String::from("aé 💩�"));
1232 assert_eq!(string.to_string_lossy(), expected);
1237 fn d(b: &[u8]) -> String {
1238 (&unsafe { Wtf8::from_bytes_unchecked(b) }).to_string()
1241 assert_eq!("", d("".as_bytes()));
1242 assert_eq!("aé 💩", d("aé 💩".as_bytes()));
1244 let mut string = Wtf8Buf::from_str("aé 💩");
1245 string.push(CodePoint::from_u32(0xD800).unwrap());
1246 assert_eq!("aé 💩�", d(string.as_inner()));
1250 fn wtf8_encode_wide() {
1251 let mut string = Wtf8Buf::from_str("aé ");
1252 string.push(CodePoint::from_u32(0xD83D).unwrap());
1253 string.push_char('💩');
1254 assert_eq!(string.encode_wide().collect::<Vec<_>>(),
1255 vec![0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]);