ntex_bytes/
string.rs

1//! A UTF-8 encoded read-only string using Bytes as storage.
2use std::{borrow, fmt, hash, ops, slice, str};
3
4use crate::{Bytes, BytesMut, BytesVec};
5
6/// An immutable UTF-8 encoded string with [`Bytes`] as a storage.
7#[derive(Clone, Default, Eq, PartialOrd, Ord)]
8pub struct ByteString(Bytes);
9
10impl ByteString {
11    /// Creates a new empty `ByteString`.
12    #[inline]
13    pub const fn new() -> Self {
14        ByteString(Bytes::new())
15    }
16
17    /// Get a str slice.
18    #[inline]
19    pub fn as_str(&self) -> &str {
20        self
21    }
22
23    /// Get a reference to the underlying bytes.
24    #[inline]
25    pub fn as_slice(&self) -> &[u8] {
26        self.0.as_ref()
27    }
28
29    /// Get a reference to the underlying `Bytes` object.
30    #[inline]
31    pub fn as_bytes(&self) -> &Bytes {
32        &self.0
33    }
34
35    /// Unwraps this `ByteString` into the underlying `Bytes` object.
36    #[inline]
37    pub fn into_bytes(self) -> Bytes {
38        self.0
39    }
40
41    /// Creates a new `ByteString` from a `&'static str`.
42    #[inline]
43    pub const fn from_static(src: &'static str) -> ByteString {
44        Self(Bytes::from_static(src.as_bytes()))
45    }
46
47    /// Returns a slice of self for the provided range.
48    ///
49    /// This will increment the reference count for the underlying memory and
50    /// return a new `ByteString` handle set to the slice.
51    ///
52    /// This operation is `O(1)`.
53    ///
54    /// # Examples
55    ///
56    /// ```
57    /// use ntex_bytes::ByteString;
58    ///
59    /// let a = ByteString::from("hello world");
60    /// let b = a.slice(2..5);
61    ///
62    /// assert_eq!(b, "llo");
63    /// ```
64    ///
65    /// # Panics
66    ///
67    /// Requires that `begin <= end` and `end <= self.len()`, otherwise slicing
68    /// will panic.
69    pub fn slice(
70        &self,
71        range: impl ops::RangeBounds<usize> + slice::SliceIndex<str> + Clone,
72    ) -> ByteString {
73        ops::Index::index(self.as_ref(), range.clone());
74        ByteString(self.0.slice(range))
75    }
76
77    /// Splits the bytestring into two at the given index.
78    ///
79    /// Afterwards `self` contains elements `[0, at)`, and the returned `ByteString`
80    /// contains elements `[at, len)`.
81    ///
82    /// This is an `O(1)` operation that just increases the reference count and
83    /// sets a few indices.
84    ///
85    /// # Examples
86    ///
87    /// ```
88    /// use ntex_bytes::ByteString;
89    ///
90    /// let mut a = ByteString::from("hello world");
91    /// let b = a.split_off(5);
92    ///
93    /// assert_eq!(a, "hello");
94    /// assert_eq!(b, " world");
95    /// ```
96    ///
97    /// # Panics
98    ///
99    /// Panics if `at > len`.
100    pub fn split_off(&mut self, at: usize) -> ByteString {
101        // check str
102        let _ = self.split_at(at);
103
104        ByteString(self.0.split_off(at))
105    }
106
107    /// Splits the bytestring into two at the given index.
108    ///
109    /// Afterwards `self` contains elements `[at, len)`, and the returned
110    /// `Bytes` contains elements `[0, at)`.
111    ///
112    /// This is an `O(1)` operation that just increases the reference count and
113    /// sets a few indices.
114    ///
115    /// # Examples
116    ///
117    /// ```
118    /// use ntex_bytes::ByteString;
119    ///
120    /// let mut a = ByteString::from("hello world");
121    /// let b = a.split_to(5);
122    ///
123    /// assert_eq!(a, " world");
124    /// assert_eq!(b, "hello");
125    /// ```
126    ///
127    /// # Panics
128    ///
129    /// Panics if `at > len`.
130    pub fn split_to(&mut self, at: usize) -> ByteString {
131        // check str
132        let _ = self.split_at(at);
133
134        ByteString(self.0.split_to(at))
135    }
136
137    /// Shortens the buffer to `len` bytes and dropping the rest.
138    #[inline]
139    pub fn trimdown(&mut self) {
140        self.0.trimdown()
141    }
142
143    /// Clears the buffer, removing all data.
144    ///
145    /// # Examples
146    ///
147    /// ```
148    /// use ntex_bytes::ByteString;
149    ///
150    /// let mut a = ByteString::from("hello world");
151    /// a.clear();
152    ///
153    /// assert!(a.is_empty());
154    /// ```
155    #[inline]
156    pub fn clear(&mut self) {
157        self.0.clear()
158    }
159
160    /// Creates a new `ByteString` from a Bytes.
161    ///
162    /// # Safety
163    /// This function is unsafe because it does not check the bytes passed to it are valid UTF-8.
164    /// If this constraint is violated, it may cause memory unsafety issues with future users of
165    /// the `ByteString`, as we assume that `ByteString`s are valid UTF-8. However, the most likely
166    /// issue is that the data gets corrupted.
167    #[inline]
168    pub const unsafe fn from_bytes_unchecked(src: Bytes) -> ByteString {
169        Self(src)
170    }
171}
172
173impl PartialEq<str> for ByteString {
174    fn eq(&self, other: &str) -> bool {
175        &self[..] == other
176    }
177}
178
179impl<T: AsRef<str>> PartialEq<T> for ByteString {
180    fn eq(&self, other: &T) -> bool {
181        &self[..] == other.as_ref()
182    }
183}
184
185impl AsRef<str> for ByteString {
186    #[inline]
187    fn as_ref(&self) -> &str {
188        self
189    }
190}
191
192impl hash::Hash for ByteString {
193    fn hash<H: hash::Hasher>(&self, state: &mut H) {
194        (**self).hash(state);
195    }
196}
197
198impl ops::Deref for ByteString {
199    type Target = str;
200
201    #[inline]
202    fn deref(&self) -> &str {
203        let bytes = self.0.as_ref();
204        // SAFETY:
205        // UTF-8 validity is guaranteed during construction.
206        unsafe { str::from_utf8_unchecked(bytes) }
207    }
208}
209
210impl borrow::Borrow<str> for ByteString {
211    #[inline]
212    fn borrow(&self) -> &str {
213        self
214    }
215}
216
217impl From<String> for ByteString {
218    #[inline]
219    fn from(value: String) -> Self {
220        Self(Bytes::from(value))
221    }
222}
223
224impl From<&str> for ByteString {
225    #[inline]
226    fn from(value: &str) -> Self {
227        Self(Bytes::copy_from_slice(value.as_ref()))
228    }
229}
230
231impl<'a> From<borrow::Cow<'a, str>> for ByteString {
232    #[inline]
233    fn from(value: borrow::Cow<'a, str>) -> Self {
234        match value {
235            borrow::Cow::Owned(s) => Self::from(s),
236            borrow::Cow::Borrowed(s) => Self::from(s),
237        }
238    }
239}
240
241impl TryFrom<&[u8]> for ByteString {
242    type Error = ();
243
244    #[inline]
245    fn try_from(value: &[u8]) -> Result<Self, Self::Error> {
246        if utf8::is_valid(value) {
247            Ok(ByteString(Bytes::copy_from_slice(value)))
248        } else {
249            Err(())
250        }
251    }
252}
253
254impl TryFrom<Vec<u8>> for ByteString {
255    type Error = ();
256
257    #[inline]
258    fn try_from(value: Vec<u8>) -> Result<Self, Self::Error> {
259        if utf8::is_valid(&value) {
260            Ok(ByteString(Bytes::from(value)))
261        } else {
262            Err(())
263        }
264    }
265}
266
267impl TryFrom<Bytes> for ByteString {
268    type Error = ();
269
270    #[inline]
271    fn try_from(value: Bytes) -> Result<Self, Self::Error> {
272        if utf8::is_valid(&value) {
273            Ok(ByteString(value))
274        } else {
275            Err(())
276        }
277    }
278}
279
280impl TryFrom<&Bytes> for ByteString {
281    type Error = ();
282
283    #[inline]
284    fn try_from(value: &Bytes) -> Result<Self, Self::Error> {
285        if utf8::is_valid(value) {
286            Ok(ByteString(value.clone()))
287        } else {
288            Err(())
289        }
290    }
291}
292
293impl TryFrom<BytesMut> for ByteString {
294    type Error = ();
295
296    #[inline]
297    fn try_from(value: BytesMut) -> Result<Self, Self::Error> {
298        if utf8::is_valid(&value) {
299            Ok(ByteString(value.freeze()))
300        } else {
301            Err(())
302        }
303    }
304}
305
306impl TryFrom<BytesVec> for ByteString {
307    type Error = ();
308
309    #[inline]
310    fn try_from(value: BytesVec) -> Result<Self, Self::Error> {
311        if utf8::is_valid(&value) {
312            Ok(ByteString(value.freeze()))
313        } else {
314            Err(())
315        }
316    }
317}
318
319impl fmt::Debug for ByteString {
320    fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result {
321        (**self).fmt(fmt)
322    }
323}
324
325impl fmt::Display for ByteString {
326    fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result {
327        (**self).fmt(fmt)
328    }
329}
330
331mod serde {
332    use serde::de::{Deserialize, Deserializer};
333    use serde::ser::{Serialize, Serializer};
334
335    use super::ByteString;
336
337    impl Serialize for ByteString {
338        #[inline]
339        fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
340        where
341            S: Serializer,
342        {
343            serializer.serialize_str(self.as_ref())
344        }
345    }
346
347    impl<'de> Deserialize<'de> for ByteString {
348        #[inline]
349        fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
350        where
351            D: Deserializer<'de>,
352        {
353            String::deserialize(deserializer).map(ByteString::from)
354        }
355    }
356}
357
358#[cfg(feature = "simd")]
359mod utf8 {
360    pub(super) fn is_valid(input: &[u8]) -> bool {
361        simdutf8::basic::from_utf8(input).is_ok()
362    }
363}
364
365#[cfg(not(feature = "simd"))]
366mod utf8 {
367    pub(super) fn is_valid(input: &[u8]) -> bool {
368        std::str::from_utf8(input).is_ok()
369    }
370}
371
372#[cfg(test)]
373mod test {
374    use std::borrow::{Borrow, Cow};
375    use std::collections::hash_map::DefaultHasher;
376    use std::hash::{Hash, Hasher};
377
378    use super::*;
379
380    #[test]
381    fn test_basics() {
382        let mut s = ByteString::from_static("test");
383        s.trimdown();
384        assert_eq!(s, "test");
385        assert_eq!(s, *"test");
386        assert_eq!(s, "test".to_owned());
387        assert_eq!(s.as_str(), "test");
388        assert_eq!(s.as_slice(), b"test");
389        assert_eq!(s.as_bytes(), &Bytes::copy_from_slice(b"test"));
390        assert_eq!(Borrow::<str>::borrow(&s), "test");
391
392        assert_eq!(format!("{}", s), "test");
393        assert_eq!(format!("{:?}", s), "\"test\"");
394
395        let b = s.into_bytes();
396        assert_eq!(b, Bytes::copy_from_slice(b"test"));
397
398        let s = unsafe { ByteString::from_bytes_unchecked(b) };
399        assert_eq!(s, "test");
400        assert_eq!(s.slice(0..2), "te");
401
402        let s = ByteString::from(Cow::Borrowed("test"));
403        assert_eq!(s, "test");
404        let mut s = ByteString::from(Cow::Owned("test".to_string()));
405        assert_eq!(s, "test");
406
407        s.clear();
408        assert_eq!(s, "");
409    }
410
411    #[test]
412    fn test_split() {
413        let mut s = ByteString::from_static("helloworld");
414        let s1 = s.split_off(5);
415        assert_eq!(s, "hello");
416        assert_eq!(s1, "world");
417
418        let mut s = ByteString::from_static("helloworld");
419        let s1 = s.split_to(5);
420        assert_eq!(s, "world");
421        assert_eq!(s1, "hello");
422    }
423
424    #[test]
425    fn test_new() {
426        let _: ByteString = ByteString::new();
427    }
428
429    #[test]
430    fn test_hash() {
431        let mut hasher1 = DefaultHasher::default();
432        "str".hash(&mut hasher1);
433
434        let mut hasher2 = DefaultHasher::default();
435        let s = ByteString::from_static("str");
436        s.hash(&mut hasher2);
437        assert_eq!(hasher1.finish(), hasher2.finish());
438    }
439
440    #[test]
441    fn test_from_string() {
442        let s: ByteString = "hello".to_owned().into();
443        assert_eq!(&s, "hello");
444        let t: &str = s.as_ref();
445        assert_eq!(t, "hello");
446    }
447
448    #[test]
449    fn test_from_str() {
450        let _: ByteString = "str".into();
451    }
452
453    #[test]
454    fn test_from_static_str() {
455        static _S: ByteString = ByteString::from_static("hello");
456        let _ = ByteString::from_static("str");
457    }
458
459    #[test]
460    fn test_try_from() {
461        let _ = ByteString::try_from(&b"nice bytes"[..]).unwrap();
462        assert!(ByteString::try_from(b"\xc3\x28".as_ref()).is_err());
463
464        let _ = ByteString::try_from(b"nice bytes".to_vec()).unwrap();
465        assert!(ByteString::try_from(vec![b'\xc3']).is_err());
466
467        let _ = ByteString::try_from(Bytes::from_static(b"nice bytes")).unwrap();
468        assert!(ByteString::try_from(Bytes::from_static(b"\xc3\x28")).is_err());
469
470        let _ = ByteString::try_from(&Bytes::from_static(b"nice bytes")).unwrap();
471        assert!(ByteString::try_from(&Bytes::from_static(b"\xc3\x28")).is_err());
472
473        let _ = ByteString::try_from(BytesMut::from(&b"nice bytes"[..])).unwrap();
474        assert!(ByteString::try_from(BytesMut::copy_from_slice(b"\xc3\x28")).is_err());
475
476        let _ =
477            ByteString::try_from(BytesVec::copy_from_slice(&b"nice bytes"[..])).unwrap();
478        assert!(ByteString::try_from(BytesVec::copy_from_slice(b"\xc3\x28")).is_err());
479    }
480
481    #[test]
482    fn test_serialize() {
483        let s: ByteString = serde_json::from_str(r#""nice bytes""#).unwrap();
484        assert_eq!(s, "nice bytes");
485    }
486
487    #[test]
488    fn test_deserialize() {
489        let s = serde_json::to_string(&ByteString::from_static("nice bytes")).unwrap();
490        assert_eq!(s, r#""nice bytes""#);
491    }
492}
ntex_bytes/string.rs

ntex_bytes/
string.rs
