seq_str/
seq_str.rs

1use crate::{SeqBytes, SeqBytesIter, SeqBytesIterMut};
2use alloc::{string::String, vec::Vec};
3use core::fmt;
4
5/// A sequence of &str, stored contiguously
6///
7/// This can be used as a drop-in replacement for `Vec<String>` in some cases,
8/// with better memory locality and fewer memory allocations.
9///
10/// When using `SeqBytes` instead of `Vec<String>`, the individual strings
11/// cannot be resized, but when this isn't needed there isn't much downside otherwise.
12///
13/// The container also supports "emplace"-style APIs like `in_place_writer`, which allow you to
14/// write the next element directly into the contiguous buffer with minimal overhead.
15///
16/// `SeqStr::from_display_iter` allows you to collect an iterator of `impl Display` items,
17/// formatting them directly into the contiguous buffer.
18#[derive(Clone, Default, Eq, PartialEq, Hash)]
19pub struct SeqStr {
20    inner: SeqBytes,
21}
22
23impl SeqStr {
24    /// Create a new SeqStr
25    pub fn new() -> Self {
26        Self::default()
27    }
28
29    /// Check if the sequence is empty
30    pub fn is_empty(&self) -> bool {
31        self.inner.is_empty()
32    }
33
34    /// Get the number of str in the sequence
35    pub fn len(&self) -> usize {
36        self.inner.len()
37    }
38
39    /// Reserve capacity for more str's
40    pub fn reserve(&mut self, extra: usize) {
41        self.inner.reserve(extra);
42    }
43
44    /// Shrink container to fit the current data
45    pub fn shrink_to_fit(&mut self) {
46        self.inner.shrink_to_fit();
47    }
48
49    /// Get the i'th element of the sequence in a checked manner
50    pub fn get(&self, idx: usize) -> Option<&str> {
51        self.inner
52            .get(idx)
53            .map(|b| unsafe { str::from_utf8_unchecked(b) })
54    }
55
56    /// Get the i'th element of the sequence in a checked manner
57    pub fn get_mut(&mut self, idx: usize) -> Option<&mut str> {
58        self.inner
59            .get_mut(idx)
60            .map(|b| unsafe { str::from_utf8_unchecked_mut(b) })
61    }
62
63    /// Check if the sequence contains a particular string
64    pub fn contains(&self, s: impl AsRef<str>) -> bool {
65        self.inner.contains(s.as_ref().as_bytes())
66    }
67
68    /// Push a `&str` onto the sequence
69    pub fn push(&mut self, s: impl AsRef<str>) {
70        self.inner.push(s.as_ref().as_bytes());
71    }
72
73    /// Get the last `str` of the sequence
74    pub fn last(&self) -> Option<&str> {
75        self.inner
76            .last()
77            .map(|b| unsafe { str::from_utf8_unchecked(b) })
78    }
79
80    /// Pop the last element of the container
81    /// Note that we can't return it because of lifetimes, so call [last] before popping.
82    pub fn pop(&mut self) {
83        self.inner.pop()
84    }
85
86    /// Iterate over a range of the sequence of `&str`
87    ///
88    /// This resembles [std::collections::BTreeMap::range], and is needed becuase like `BTreeMap`,
89    /// we can't implement `Deref` or `SliceIndex<Range>` and produce a slice of our contents.
90    /// See also [as_vec].
91    pub fn range(
92        &self,
93        range_bounds: impl core::ops::RangeBounds<usize>,
94    ) -> core::iter::Map<SeqBytesIter<'_>, fn(&[u8]) -> &str> {
95        fn helper(b: &[u8]) -> &str {
96            unsafe { str::from_utf8_unchecked(b) }
97        }
98
99        self.inner.range(range_bounds).map(helper)
100    }
101
102    /// Iterate over a range of the sequence of &mut [u8]
103    pub fn range_mut(
104        &mut self,
105        range_bounds: impl core::ops::RangeBounds<usize>,
106    ) -> core::iter::Map<SeqBytesIterMut<'_>, fn(&mut [u8]) -> &mut str> {
107        fn helper(b: &mut [u8]) -> &mut str {
108            unsafe { str::from_utf8_unchecked_mut(b) }
109        }
110
111        self.inner.range_mut(range_bounds).map(helper)
112    }
113
114    /// Iterate over the sequence of `&str`
115    pub fn iter(&self) -> core::iter::Map<SeqBytesIter<'_>, fn(&[u8]) -> &str> {
116        fn helper(b: &[u8]) -> &str {
117            unsafe { str::from_utf8_unchecked(b) }
118        }
119
120        self.inner.iter().map(helper)
121    }
122
123    /// Iterate over the sequence of `&mut str`
124    pub fn iter_mut(&mut self) -> core::iter::Map<SeqBytesIterMut<'_>, fn(&mut [u8]) -> &mut str> {
125        fn helper(b: &mut [u8]) -> &mut str {
126            unsafe { str::from_utf8_unchecked_mut(b) }
127        }
128
129        self.inner.iter_mut().map(helper)
130    }
131
132    /// Truncate to at most the first n `str`
133    pub fn truncate(&mut self, new_size: usize) {
134        self.inner.truncate(new_size)
135    }
136
137    /// Resize to contain only the first n `str`, or pad up to n `str`, with empty `str` added
138    pub fn resize(&mut self, new_size: usize) {
139        self.inner.resize(new_size)
140    }
141
142    /// Retain only those `str` satisfying a predicate.
143    /// The `str` are always visited in order, similar to [std::vec::Vec::retain].
144    pub fn retain(&mut self, mut pred: impl FnMut(&str) -> bool) {
145        self.inner
146            .retain(move |b| pred(unsafe { str::from_utf8_unchecked(b) }));
147    }
148
149    /// Retain only those `str` satisfying a predicate.
150    /// The `str` are always visited in order, similar to [std::vec::Vec::retain_mut].
151    pub fn retain_mut(&mut self, mut pred: impl FnMut(&mut str) -> bool) {
152        self.inner
153            .retain_mut(move |b| pred(unsafe { str::from_utf8_unchecked_mut(b) }));
154    }
155
156    /// Get an `impl std::fmt::Write` which can be used to write the next slice
157    /// directly into the buffer without copying.
158    pub fn in_place_writer(&mut self) -> impl core::fmt::Write {
159        // Wrapper for return
160        struct Sink<T: FnMut(&[u8])>(pub T);
161
162        impl<T: FnMut(&[u8])> core::fmt::Write for Sink<T> {
163            fn write_str(&mut self, s: &str) -> std::fmt::Result {
164                (self.0)(s.as_bytes());
165                Ok(())
166            }
167        }
168
169        Sink(self.inner.in_place_writer_no_std())
170    }
171
172    /// Construct `SeqStr` from any items that implement `Display`,
173    /// formatting them directly into the contiguous buffer.
174    pub fn from_display_iter<T: core::fmt::Display, I: Iterator<Item = T>>(it: I) -> Self {
175        use core::fmt::Write;
176
177        let mut result = Self::default();
178        result.reserve(it.size_hint().0);
179        for item in it {
180            // Note: This error is okay to unwrap, because our destination buffer is unlimited size.
181            // This is the same as what that `ToString::to_string` does with these errors in the stdlib.
182            write!(result.in_place_writer(), "{item}")
183                .expect("a Display implementation returned an error unexpectedly");
184        }
185        result
186    }
187
188    /// Express as a Vec<&str>. The main reason that this may be useful is that there are
189    /// useful methods on slice types `&[&str]`, for example, [core::slice::binary_search],
190    /// but `SeqStr` itself doesn't implement `Deref` the way that `Vec` does and can
191    /// only produce such a slice by allocating.
192    ///
193    /// See [SeqBytes::as_vec] for more discussion of tradeoffs.
194    pub fn as_vec(&self) -> Vec<&str> {
195        self.iter().collect()
196    }
197
198    /// Concatenate the `str` in the sequence into one string
199    pub fn concat(&self) -> &str {
200        unsafe { core::str::from_utf8_unchecked(self.inner.concat()) }
201    }
202
203    /// Join the `str` in the sequence into one string, placing a separator between them
204    pub fn join(&self, separator: &str) -> String {
205        let mut result = String::with_capacity(self.inner.num_bytes() + self.inner.len().saturating_sub(1) * separator.len());
206        let mut first = true;
207
208        for s in self.iter() {
209            if first {
210                first = false;
211            } else {
212                result += separator;
213            }
214            result += s;
215        }
216        result
217    }
218}
219
220impl core::ops::Index<usize> for SeqStr {
221    type Output = str;
222
223    fn index(&self, index: usize) -> &str {
224        unsafe { str::from_utf8_unchecked(self.inner.index(index)) }
225    }
226}
227
228impl core::ops::IndexMut<usize> for SeqStr {
229    fn index_mut(&mut self, index: usize) -> &mut str {
230        unsafe { str::from_utf8_unchecked_mut(self.inner.index_mut(index)) }
231    }
232}
233
234impl fmt::Debug for SeqStr {
235    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
236        f.debug_list().entries(self.iter()).finish()
237    }
238}
239
240impl<A: AsRef<str>> Extend<A> for SeqStr {
241    fn extend<T>(&mut self, iter: T)
242    where
243        T: IntoIterator<Item = A>,
244    {
245        let iter = iter.into_iter();
246        self.reserve(iter.size_hint().0);
247        for item in iter {
248            self.push(item);
249        }
250    }
251}
252
253impl<A: AsRef<str>> FromIterator<A> for SeqStr {
254    // Required method
255    fn from_iter<T>(iter: T) -> Self
256    where
257        T: IntoIterator<Item = A>,
258    {
259        let mut result = SeqStr::default();
260        result.extend(iter);
261        result
262    }
263}
264
265// IntoIterator can only be implemented for &'a SeqStr,
266// otherwise the buffer doesn't live long enough.
267impl<'a> IntoIterator for &'a SeqStr {
268    type Item = &'a str;
269    type IntoIter = core::iter::Map<SeqBytesIter<'a>, fn(&[u8]) -> &str>;
270
271    fn into_iter(self) -> Self::IntoIter {
272        fn helper(b: &[u8]) -> &str {
273            unsafe { core::str::from_utf8_unchecked(b) }
274        }
275
276        self.inner.iter().map(helper)
277    }
278}
279
280#[cfg(feature = "serde")]
281mod serde_impls {
282    use super::*;
283    use ::serde::{
284        Deserialize, Serialize,
285        de::{Deserializer, SeqAccess, Visitor},
286        ser::{SerializeSeq, Serializer},
287    };
288
289    impl Serialize for SeqStr {
290        fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
291        where
292            S: Serializer,
293        {
294            let mut seq = serializer.serialize_seq(Some(self.len()))?;
295            for e in self {
296                seq.serialize_element(e)?;
297            }
298            seq.end()
299        }
300    }
301
302    impl<'de> Deserialize<'de> for SeqStr {
303        fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
304        where
305            D: Deserializer<'de>,
306        {
307            deserializer.deserialize_seq(SeqStrVis {})
308        }
309    }
310
311    struct SeqStrVis {}
312
313    impl<'de> Visitor<'de> for SeqStrVis {
314        type Value = SeqStr;
315
316        fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
317            formatter.write_str("a sequence of strings")
318        }
319
320        fn visit_seq<A>(self, mut seq: A) -> Result<Self::Value, A::Error>
321        where
322            A: SeqAccess<'de>,
323        {
324            let mut result = SeqStr::default();
325
326            if let Some(size) = seq.size_hint() {
327                result.reserve(size);
328            }
329
330            loop {
331                let maybe_next_str: Option<&str> = seq.next_element()?;
332                if let Some(next_str) = maybe_next_str {
333                    result.push(next_str);
334                } else {
335                    return Ok(result);
336                }
337            }
338        }
339    }
340}
341
342#[cfg(test)]
343mod tests {
344    use super::*;
345    use alloc::{borrow::ToOwned, vec, vec::Vec};
346
347    #[test]
348    fn vec_str_conversions() {
349        let vec_str = vec!["1", "2", "3"];
350
351        let seq_str: SeqStr = vec_str.into_iter().collect();
352
353        assert_eq!(seq_str.len(), 3);
354        assert_eq!(&seq_str[0], "1");
355        assert_eq!(&seq_str[1], "2");
356        assert_eq!(&seq_str[2], "3");
357
358        let vec_str2 = seq_str.iter().map(ToOwned::to_owned).collect::<Vec<_>>();
359
360        assert_eq!(vec_str2.len(), 3);
361        assert_eq!(vec_str2[0], "1");
362        assert_eq!(vec_str2[1], "2");
363        assert_eq!(vec_str2[2], "3");
364    }
365
366    #[test]
367    fn vec_string_conversions() {
368        let vec_string = vec!["1".to_owned(), "2".to_owned(), "3".to_owned()];
369
370        let seq_str: SeqStr = vec_string.into_iter().collect();
371
372        assert_eq!(seq_str.len(), 3);
373        assert_eq!(&seq_str[0], "1");
374        assert_eq!(&seq_str[1], "2");
375        assert_eq!(&seq_str[2], "3");
376
377        let vec_str2 = seq_str.iter().map(ToOwned::to_owned).collect::<Vec<_>>();
378
379        assert_eq!(vec_str2.len(), 3);
380        assert_eq!(vec_str2[0], "1");
381        assert_eq!(vec_str2[1], "2");
382        assert_eq!(vec_str2[2], "3");
383    }
384
385    #[test]
386    fn from_display_iter() {
387        let v = vec![1, 2, 3, 45, 67];
388
389        let seq_str = SeqStr::from_display_iter(v.into_iter());
390
391        assert_eq!(seq_str.len(), 5);
392        assert_eq!(&seq_str[0], "1");
393        assert_eq!(&seq_str[1], "2");
394        assert_eq!(&seq_str[2], "3");
395        assert_eq!(&seq_str[3], "45");
396        assert_eq!(&seq_str[4], "67");
397    }
398
399    #[test]
400    fn range() {
401        let v = vec![1, 2, 3, 45, 67];
402
403        let seq_str = SeqStr::from_display_iter(v.into_iter());
404
405        let seq_str2: SeqStr = seq_str.range(1..4).collect();
406        assert_eq!(seq_str2.len(), 3);
407        assert_eq!(&seq_str2[0], "2");
408        assert_eq!(&seq_str2[1], "3");
409        assert_eq!(&seq_str2[2], "45");
410
411        let seq_str2: SeqStr = seq_str.range(3..).collect();
412        assert_eq!(seq_str2.len(), 2);
413        assert_eq!(&seq_str2[0], "45");
414        assert_eq!(&seq_str2[1], "67");
415
416        let seq_str2: SeqStr = seq_str.range(..3).collect();
417        assert_eq!(seq_str2.len(), 3);
418        assert_eq!(&seq_str2[0], "1");
419        assert_eq!(&seq_str2[1], "2");
420        assert_eq!(&seq_str2[2], "3");
421
422        let seq_str2: SeqStr = seq_str.range(..).collect();
423        assert_eq!(seq_str2.len(), 5);
424        assert_eq!(&seq_str2[0], "1");
425        assert_eq!(&seq_str2[1], "2");
426        assert_eq!(&seq_str2[2], "3");
427        assert_eq!(&seq_str2[3], "45");
428        assert_eq!(&seq_str2[4], "67");
429    }
430
431    #[test]
432    fn iter_mut() {
433        let mut seq_str = SeqStr::from_iter(["asdf", "jkl;", ""]);
434
435        assert_eq!(seq_str.len(), 3);
436        assert_eq!(&seq_str[0], "asdf");
437        assert_eq!(&seq_str[1], "jkl;");
438        assert_eq!(&seq_str[2], "");
439
440        for s in seq_str.iter_mut() {
441            s.make_ascii_uppercase();
442        }
443
444        assert_eq!(seq_str.len(), 3);
445        assert_eq!(&seq_str[0], "ASDF");
446        assert_eq!(&seq_str[1], "JKL;");
447        assert_eq!(&seq_str[2], "");
448    }
449
450    #[test]
451    fn test_serde() {
452        let seq_str: SeqStr = serde_json::from_str("[\"asdf\", \"jkl;\", \"\"]").unwrap();
453
454        assert_eq!(seq_str.len(), 3);
455        assert_eq!(&seq_str[0], "asdf");
456        assert_eq!(&seq_str[1], "jkl;");
457        assert_eq!(&seq_str[2], "");
458
459        let ser = serde_json::to_string(&seq_str).unwrap();
460
461        let seq_str2: SeqStr = serde_json::from_str(&ser).unwrap();
462
463        assert_eq!(seq_str, seq_str2);
464    }
465
466    #[test]
467    fn test_join() {
468        let seq_str = SeqStr::from_iter(["asdf", "jkl;", "", ":)"]);
469
470        assert_eq!(seq_str.concat(), "asdfjkl;:)");
471        assert_eq!(seq_str.join(", "), "asdf, jkl;, , :)");    
472    }
473}