1 | use std::borrow::{Borrow, Cow, ToOwned}; |
2 | use std::convert::{AsRef, TryFrom}; |
3 | use std::fmt; |
4 | use std::hash::{Hash, Hasher}; |
5 | use std::ops::Deref; |
6 | use std::str::from_utf8; |
7 | |
8 | const MAX_INLINE_STR_LEN: usize = 3 * std::mem::size_of::<isize>() - 2; |
9 | |
10 | /// Returned when trying to convert a `&str` into a `InlineStr` |
11 | /// but it fails because it doesn't fit. |
12 | #[derive (Debug)] |
13 | pub struct StringTooLongError; |
14 | |
15 | /// An inline string that can contain almost three words |
16 | /// of utf-8 text. |
17 | #[derive (Debug, Clone, Copy, Eq)] |
18 | pub struct InlineStr { |
19 | inner: [u8; MAX_INLINE_STR_LEN], |
20 | len: u8, |
21 | } |
22 | |
23 | impl<'a> AsRef<str> for InlineStr { |
24 | fn as_ref(&self) -> &str { |
25 | self.deref() |
26 | } |
27 | } |
28 | |
29 | impl Hash for InlineStr { |
30 | fn hash<H: Hasher>(&self, state: &mut H) { |
31 | self.deref().hash(state); |
32 | } |
33 | } |
34 | |
35 | impl From<char> for InlineStr { |
36 | fn from(c: char) -> Self { |
37 | let mut inner: [u8; 22] = [0u8; MAX_INLINE_STR_LEN]; |
38 | c.encode_utf8(&mut inner); |
39 | let len: u8 = c.len_utf8() as u8; |
40 | Self { inner, len } |
41 | } |
42 | } |
43 | |
44 | impl<'a> std::cmp::PartialEq<InlineStr> for InlineStr { |
45 | fn eq(&self, other: &InlineStr) -> bool { |
46 | self.deref() == other.deref() |
47 | } |
48 | } |
49 | |
50 | impl TryFrom<&str> for InlineStr { |
51 | type Error = StringTooLongError; |
52 | |
53 | fn try_from(s: &str) -> Result<InlineStr, StringTooLongError> { |
54 | let len: usize = s.len(); |
55 | if len <= MAX_INLINE_STR_LEN { |
56 | let mut inner: [u8; 22] = [0u8; MAX_INLINE_STR_LEN]; |
57 | inner[..len].copy_from_slice(src:s.as_bytes()); |
58 | let len: u8 = len as u8; |
59 | Ok(Self { inner, len }) |
60 | } else { |
61 | Err(StringTooLongError) |
62 | } |
63 | } |
64 | } |
65 | |
66 | impl Deref for InlineStr { |
67 | type Target = str; |
68 | |
69 | fn deref(&self) -> &str { |
70 | let len: usize = self.len as usize; |
71 | from_utf8(&self.inner[..len]).unwrap() |
72 | } |
73 | } |
74 | |
75 | impl fmt::Display for InlineStr { |
76 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
77 | write!(f, " {}" , self.as_ref()) |
78 | } |
79 | } |
80 | |
81 | /// A copy-on-write string that can be owned, borrowed |
82 | /// or inlined. |
83 | /// |
84 | /// It is three words long. |
85 | #[derive (Debug, Eq)] |
86 | pub enum CowStr<'a> { |
87 | /// An owned, immutable string. |
88 | Boxed(Box<str>), |
89 | /// A borrowed string. |
90 | Borrowed(&'a str), |
91 | /// A short inline string. |
92 | Inlined(InlineStr), |
93 | } |
94 | |
95 | #[cfg (feature = "serde" )] |
96 | mod serde_impl { |
97 | use super::CowStr; |
98 | use serde::{de, Deserialize, Deserializer, Serialize, Serializer}; |
99 | use std::fmt; |
100 | |
101 | impl<'a> Serialize for CowStr<'a> { |
102 | fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error> |
103 | where |
104 | S: Serializer, |
105 | { |
106 | serializer.serialize_str(self.as_ref()) |
107 | } |
108 | } |
109 | |
110 | struct CowStrVisitor; |
111 | |
112 | impl<'de> de::Visitor<'de> for CowStrVisitor { |
113 | type Value = CowStr<'de>; |
114 | |
115 | fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { |
116 | formatter.write_str("a string" ) |
117 | } |
118 | |
119 | fn visit_borrowed_str<E>(self, v: &'de str) -> Result<Self::Value, E> |
120 | where |
121 | E: de::Error, |
122 | { |
123 | Ok(CowStr::Borrowed(v)) |
124 | } |
125 | } |
126 | |
127 | impl<'a, 'de: 'a> Deserialize<'de> for CowStr<'a> { |
128 | fn deserialize<D>(deserializer: D) -> Result<Self, D::Error> |
129 | where |
130 | D: Deserializer<'de>, |
131 | { |
132 | deserializer.deserialize_str(CowStrVisitor) |
133 | } |
134 | } |
135 | } |
136 | |
137 | impl<'a> AsRef<str> for CowStr<'a> { |
138 | fn as_ref(&self) -> &str { |
139 | self.deref() |
140 | } |
141 | } |
142 | |
143 | impl<'a> Hash for CowStr<'a> { |
144 | fn hash<H: Hasher>(&self, state: &mut H) { |
145 | self.deref().hash(state); |
146 | } |
147 | } |
148 | |
149 | impl<'a> std::clone::Clone for CowStr<'a> { |
150 | fn clone(&self) -> Self { |
151 | match self { |
152 | CowStr::Boxed(s: &Box) => match InlineStr::try_from(&**s) { |
153 | Ok(inline: InlineStr) => CowStr::Inlined(inline), |
154 | Err(..) => CowStr::Boxed(s.clone()), |
155 | }, |
156 | CowStr::Borrowed(s: &&str) => CowStr::Borrowed(s), |
157 | CowStr::Inlined(s: &InlineStr) => CowStr::Inlined(*s), |
158 | } |
159 | } |
160 | } |
161 | |
162 | impl<'a> std::cmp::PartialEq<CowStr<'a>> for CowStr<'a> { |
163 | fn eq(&self, other: &CowStr) -> bool { |
164 | self.deref() == other.deref() |
165 | } |
166 | } |
167 | |
168 | impl<'a> From<&'a str> for CowStr<'a> { |
169 | fn from(s: &'a str) -> Self { |
170 | CowStr::Borrowed(s) |
171 | } |
172 | } |
173 | |
174 | impl<'a> From<String> for CowStr<'a> { |
175 | fn from(s: String) -> Self { |
176 | CowStr::Boxed(s.into_boxed_str()) |
177 | } |
178 | } |
179 | |
180 | impl<'a> From<char> for CowStr<'a> { |
181 | fn from(c: char) -> Self { |
182 | CowStr::Inlined(c.into()) |
183 | } |
184 | } |
185 | |
186 | impl<'a> From<Cow<'a, str>> for CowStr<'a> { |
187 | fn from(s: Cow<'a, str>) -> Self { |
188 | match s { |
189 | Cow::Borrowed(s: &str) => CowStr::Borrowed(s), |
190 | Cow::Owned(s: String) => CowStr::Boxed(s.into_boxed_str()), |
191 | } |
192 | } |
193 | } |
194 | |
195 | impl<'a> From<CowStr<'a>> for Cow<'a, str> { |
196 | fn from(s: CowStr<'a>) -> Self { |
197 | match s { |
198 | CowStr::Boxed(s: Box) => Cow::Owned(s.to_string()), |
199 | CowStr::Inlined(s: InlineStr) => Cow::Owned(s.to_string()), |
200 | CowStr::Borrowed(s: &str) => Cow::Borrowed(s), |
201 | } |
202 | } |
203 | } |
204 | |
205 | impl<'a> From<Cow<'a, char>> for CowStr<'a> { |
206 | fn from(s: Cow<'a, char>) -> Self { |
207 | CowStr::Inlined(InlineStr::from(*s)) |
208 | } |
209 | } |
210 | |
211 | impl<'a> Deref for CowStr<'a> { |
212 | type Target = str; |
213 | |
214 | fn deref(&self) -> &str { |
215 | match self { |
216 | CowStr::Boxed(ref b: &Box) => &*b, |
217 | CowStr::Borrowed(b: &&str) => b, |
218 | CowStr::Inlined(ref s: &InlineStr) => s.deref(), |
219 | } |
220 | } |
221 | } |
222 | |
223 | impl<'a> Borrow<str> for CowStr<'a> { |
224 | fn borrow(&self) -> &str { |
225 | self.deref() |
226 | } |
227 | } |
228 | |
229 | impl<'a> CowStr<'a> { |
230 | pub fn into_string(self) -> String { |
231 | match self { |
232 | CowStr::Boxed(b: Box) => b.into(), |
233 | CowStr::Borrowed(b: &str) => b.to_owned(), |
234 | CowStr::Inlined(s: InlineStr) => s.deref().to_owned(), |
235 | } |
236 | } |
237 | } |
238 | |
239 | impl<'a> fmt::Display for CowStr<'a> { |
240 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
241 | write!(f, " {}" , self.as_ref()) |
242 | } |
243 | } |
244 | |
245 | #[cfg (test)] |
246 | mod test_special_string { |
247 | use super::*; |
248 | |
249 | #[test ] |
250 | fn inlinestr_ascii() { |
251 | let s: InlineStr = 'a' .into(); |
252 | assert_eq!("a" , s.deref()); |
253 | } |
254 | |
255 | #[test ] |
256 | fn inlinestr_unicode() { |
257 | let s: InlineStr = '🍔' .into(); |
258 | assert_eq!("🍔" , s.deref()); |
259 | } |
260 | |
261 | #[test ] |
262 | fn cowstr_size() { |
263 | let size = std::mem::size_of::<CowStr>(); |
264 | let word_size = std::mem::size_of::<isize>(); |
265 | assert_eq!(3 * word_size, size); |
266 | } |
267 | |
268 | #[test ] |
269 | fn cowstr_char_to_string() { |
270 | let c = '藏' ; |
271 | let smort: CowStr = c.into(); |
272 | let owned: String = smort.to_string(); |
273 | let expected = "藏" .to_owned(); |
274 | assert_eq!(expected, owned); |
275 | } |
276 | |
277 | #[test ] |
278 | fn max_inline_str_len_atleast_four() { |
279 | // we need 4 bytes to store a char |
280 | assert!(MAX_INLINE_STR_LEN >= 4); |
281 | } |
282 | |
283 | #[test ] |
284 | #[cfg (target_pointer_width = "64" )] |
285 | fn inlinestr_fits_twentytwo() { |
286 | let s = "0123456789abcdefghijkl" ; |
287 | let stack_str = InlineStr::try_from(s).unwrap(); |
288 | assert_eq!(stack_str.deref(), s); |
289 | } |
290 | |
291 | #[test ] |
292 | #[cfg (target_pointer_width = "64" )] |
293 | fn inlinestr_not_fits_twentythree() { |
294 | let s = "0123456789abcdefghijklm" ; |
295 | let _stack_str = InlineStr::try_from(s).unwrap_err(); |
296 | } |
297 | |
298 | #[test ] |
299 | #[cfg (target_pointer_width = "64" )] |
300 | fn small_boxed_str_clones_to_stack() { |
301 | let s = "0123456789abcde" .to_owned(); |
302 | let smort: CowStr = s.into(); |
303 | let smort_clone = smort.clone(); |
304 | |
305 | if let CowStr::Inlined(..) = smort_clone { |
306 | } else { |
307 | panic!("Expected a Inlined variant!" ); |
308 | } |
309 | } |
310 | |
311 | #[test ] |
312 | fn cow_to_cow_str() { |
313 | let s = "some text" ; |
314 | let cow = Cow::Borrowed(s); |
315 | let actual = CowStr::from(cow); |
316 | let expected = CowStr::Borrowed(s); |
317 | assert_eq!(actual, expected); |
318 | assert!(variant_eq(&actual, &expected)); |
319 | |
320 | let s = "some text" .to_string(); |
321 | let cow: Cow<str> = Cow::Owned(s.clone()); |
322 | let actual = CowStr::from(cow); |
323 | let expected = CowStr::Boxed(s.into_boxed_str()); |
324 | assert_eq!(actual, expected); |
325 | assert!(variant_eq(&actual, &expected)); |
326 | } |
327 | |
328 | #[test ] |
329 | fn cow_str_to_cow() { |
330 | let s = "some text" ; |
331 | let cow_str = CowStr::Borrowed(s); |
332 | let actual = Cow::from(cow_str); |
333 | let expected = Cow::Borrowed(s); |
334 | assert_eq!(actual, expected); |
335 | assert!(variant_eq(&actual, &expected)); |
336 | |
337 | let s = "s" ; |
338 | let inline_str: InlineStr = InlineStr::try_from(s).unwrap(); |
339 | let cow_str = CowStr::Inlined(inline_str); |
340 | let actual = Cow::from(cow_str); |
341 | let expected: Cow<str> = Cow::Owned(s.to_string()); |
342 | assert_eq!(actual, expected); |
343 | assert!(variant_eq(&actual, &expected)); |
344 | |
345 | let s = "s" ; |
346 | let cow_str = CowStr::Boxed(s.to_string().into_boxed_str()); |
347 | let actual = Cow::from(cow_str); |
348 | let expected: Cow<str> = Cow::Owned(s.to_string()); |
349 | assert_eq!(actual, expected); |
350 | assert!(variant_eq(&actual, &expected)); |
351 | } |
352 | |
353 | #[test ] |
354 | fn cow_char_to_cow_str() { |
355 | let c = 'c' ; |
356 | let cow: Cow<char> = Cow::Owned(c); |
357 | let actual = CowStr::from(cow); |
358 | let expected = CowStr::Inlined(InlineStr::from(c)); |
359 | assert_eq!(actual, expected); |
360 | assert!(variant_eq(&actual, &expected)); |
361 | |
362 | let c = 'c' ; |
363 | let cow: Cow<char> = Cow::Borrowed(&c); |
364 | let actual = CowStr::from(cow); |
365 | let expected = CowStr::Inlined(InlineStr::from(c)); |
366 | assert_eq!(actual, expected); |
367 | assert!(variant_eq(&actual, &expected)); |
368 | } |
369 | |
370 | fn variant_eq<T>(a: &T, b: &T) -> bool { |
371 | std::mem::discriminant(a) == std::mem::discriminant(b) |
372 | } |
373 | } |
374 | |