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    /// Iterate over the sequence of `&str`, in chunks of given size.
133    /// Returns an iterator which yields one iterator for each chunk, each of which
134    /// yields `chunk_size` string values.
135    /// The last chunk may be smaller.
136    pub fn chunks(&self, chunk_size: usize) -> impl Iterator<Item = impl Iterator<Item = &str>> {
137        fn helper(b: &[u8]) -> &str {
138            unsafe { str::from_utf8_unchecked(b) }
139        }
140
141        self.inner.chunks(chunk_size).map(|iter| iter.map(helper))
142    }
143
144    /// Truncate to at most the first n `str`
145    pub fn truncate(&mut self, new_size: usize) {
146        self.inner.truncate(new_size)
147    }
148
149    /// Resize to contain only the first n `str`, or pad up to n `str`, with empty `str` added
150    pub fn resize(&mut self, new_size: usize) {
151        self.inner.resize(new_size)
152    }
153
154    /// Retain only those `str` satisfying a predicate.
155    /// The `str` are always visited in order, similar to [std::vec::Vec::retain].
156    pub fn retain(&mut self, mut pred: impl FnMut(&str) -> bool) {
157        self.inner
158            .retain(move |b| pred(unsafe { str::from_utf8_unchecked(b) }));
159    }
160
161    /// Retain only those `str` satisfying a predicate.
162    /// The `str` are always visited in order, similar to [std::vec::Vec::retain_mut].
163    pub fn retain_mut(&mut self, mut pred: impl FnMut(&mut str) -> bool) {
164        self.inner
165            .retain_mut(move |b| pred(unsafe { str::from_utf8_unchecked_mut(b) }));
166    }
167
168    /// Get an `impl std::fmt::Write` which can be used to write the next slice
169    /// directly into the buffer without copying.
170    pub fn in_place_writer(&mut self) -> impl core::fmt::Write {
171        // Wrapper for return
172        struct Sink<T: FnMut(&[u8])>(pub T);
173
174        impl<T: FnMut(&[u8])> core::fmt::Write for Sink<T> {
175            fn write_str(&mut self, s: &str) -> std::fmt::Result {
176                (self.0)(s.as_bytes());
177                Ok(())
178            }
179        }
180
181        Sink(self.inner.in_place_writer_no_std())
182    }
183
184    /// Construct `SeqStr` from any items that implement `Display`,
185    /// formatting them directly into the contiguous buffer.
186    pub fn from_display_iter<T: core::fmt::Display, I: Iterator<Item = T>>(it: I) -> Self {
187        use core::fmt::Write;
188
189        let mut result = Self::default();
190        result.reserve(it.size_hint().0);
191        for item in it {
192            // Note: This error is okay to unwrap, because our destination buffer is unlimited size.
193            // This is the same as what that `ToString::to_string` does with these errors in the stdlib.
194            write!(result.in_place_writer(), "{item}")
195                .expect("a Display implementation returned an error unexpectedly");
196        }
197        result
198    }
199
200    /// Express as a Vec<&str>. The main reason that this may be useful is that there are
201    /// useful methods on slice types `&[&str]`, for example, [core::slice::binary_search],
202    /// but `SeqStr` itself doesn't implement `Deref` the way that `Vec` does and can
203    /// only produce such a slice by allocating.
204    ///
205    /// See [SeqBytes::as_vec] for more discussion of tradeoffs.
206    pub fn as_vec(&self) -> Vec<&str> {
207        self.iter().collect()
208    }
209
210    /// Concatenate the `str` in the sequence into one string
211    pub fn concat(&self) -> &str {
212        unsafe { core::str::from_utf8_unchecked(self.inner.concat()) }
213    }
214
215    /// Join the `str` in the sequence into one string, placing a separator between them
216    pub fn join(&self, separator: &str) -> String {
217        let mut result = String::with_capacity(
218            self.inner.num_bytes() + self.inner.len().saturating_sub(1) * separator.len(),
219        );
220        let mut first = true;
221
222        for s in self.iter() {
223            if first {
224                first = false;
225            } else {
226                result += separator;
227            }
228            result += s;
229        }
230        result
231    }
232}
233
234impl core::ops::Index<usize> for SeqStr {
235    type Output = str;
236
237    fn index(&self, index: usize) -> &str {
238        unsafe { str::from_utf8_unchecked(self.inner.index(index)) }
239    }
240}
241
242impl core::ops::IndexMut<usize> for SeqStr {
243    fn index_mut(&mut self, index: usize) -> &mut str {
244        unsafe { str::from_utf8_unchecked_mut(self.inner.index_mut(index)) }
245    }
246}
247
248impl fmt::Debug for SeqStr {
249    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
250        f.debug_list().entries(self.iter()).finish()
251    }
252}
253
254impl<A: AsRef<str>> Extend<A> for SeqStr {
255    fn extend<T>(&mut self, iter: T)
256    where
257        T: IntoIterator<Item = A>,
258    {
259        let iter = iter.into_iter();
260        self.reserve(iter.size_hint().0);
261        for item in iter {
262            self.push(item);
263        }
264    }
265}
266
267impl<A: AsRef<str>> FromIterator<A> for SeqStr {
268    // Required method
269    fn from_iter<T>(iter: T) -> Self
270    where
271        T: IntoIterator<Item = A>,
272    {
273        let mut result = SeqStr::default();
274        result.extend(iter);
275        result
276    }
277}
278
279// IntoIterator can only be implemented for &'a SeqStr,
280// otherwise the buffer doesn't live long enough.
281impl<'a> IntoIterator for &'a SeqStr {
282    type Item = &'a str;
283    type IntoIter = core::iter::Map<SeqBytesIter<'a>, fn(&[u8]) -> &str>;
284
285    fn into_iter(self) -> Self::IntoIter {
286        fn helper(b: &[u8]) -> &str {
287            unsafe { core::str::from_utf8_unchecked(b) }
288        }
289
290        self.inner.iter().map(helper)
291    }
292}
293
294#[cfg(feature = "serde")]
295mod serde_impls {
296    use super::*;
297    use ::serde::{
298        Deserialize, Serialize,
299        de::{Deserializer, SeqAccess, Visitor},
300        ser::{SerializeSeq, Serializer},
301    };
302
303    impl Serialize for SeqStr {
304        fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
305        where
306            S: Serializer,
307        {
308            let mut seq = serializer.serialize_seq(Some(self.len()))?;
309            for e in self {
310                seq.serialize_element(e)?;
311            }
312            seq.end()
313        }
314    }
315
316    impl<'de> Deserialize<'de> for SeqStr {
317        fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
318        where
319            D: Deserializer<'de>,
320        {
321            deserializer.deserialize_seq(SeqStrVis {})
322        }
323    }
324
325    struct SeqStrVis {}
326
327    impl<'de> Visitor<'de> for SeqStrVis {
328        type Value = SeqStr;
329
330        fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
331            formatter.write_str("a sequence of strings")
332        }
333
334        fn visit_seq<A>(self, mut seq: A) -> Result<Self::Value, A::Error>
335        where
336            A: SeqAccess<'de>,
337        {
338            let mut result = SeqStr::default();
339
340            if let Some(size) = seq.size_hint() {
341                result.reserve(size);
342            }
343
344            loop {
345                let maybe_next_str: Option<&str> = seq.next_element()?;
346                if let Some(next_str) = maybe_next_str {
347                    result.push(next_str);
348                } else {
349                    return Ok(result);
350                }
351            }
352        }
353    }
354}
355
356#[cfg(test)]
357mod tests {
358    use super::*;
359    use alloc::{borrow::ToOwned, vec, vec::Vec};
360
361    #[test]
362    fn vec_str_conversions() {
363        let vec_str = vec!["1", "2", "3"];
364
365        let seq_str: SeqStr = vec_str.into_iter().collect();
366
367        assert_eq!(seq_str.len(), 3);
368        assert_eq!(&seq_str[0], "1");
369        assert_eq!(&seq_str[1], "2");
370        assert_eq!(&seq_str[2], "3");
371
372        let vec_str2 = seq_str.iter().map(ToOwned::to_owned).collect::<Vec<_>>();
373
374        assert_eq!(vec_str2.len(), 3);
375        assert_eq!(vec_str2[0], "1");
376        assert_eq!(vec_str2[1], "2");
377        assert_eq!(vec_str2[2], "3");
378    }
379
380    #[test]
381    fn vec_string_conversions() {
382        let vec_string = vec!["1".to_owned(), "2".to_owned(), "3".to_owned()];
383
384        let seq_str: SeqStr = vec_string.into_iter().collect();
385
386        assert_eq!(seq_str.len(), 3);
387        assert_eq!(&seq_str[0], "1");
388        assert_eq!(&seq_str[1], "2");
389        assert_eq!(&seq_str[2], "3");
390
391        let vec_str2 = seq_str.iter().map(ToOwned::to_owned).collect::<Vec<_>>();
392
393        assert_eq!(vec_str2.len(), 3);
394        assert_eq!(vec_str2[0], "1");
395        assert_eq!(vec_str2[1], "2");
396        assert_eq!(vec_str2[2], "3");
397    }
398
399    #[test]
400    fn from_display_iter() {
401        let v = vec![1, 2, 3, 45, 67];
402
403        let seq_str = SeqStr::from_display_iter(v.into_iter());
404
405        assert_eq!(seq_str.len(), 5);
406        assert_eq!(&seq_str[0], "1");
407        assert_eq!(&seq_str[1], "2");
408        assert_eq!(&seq_str[2], "3");
409        assert_eq!(&seq_str[3], "45");
410        assert_eq!(&seq_str[4], "67");
411    }
412
413    #[test]
414    fn range() {
415        let v = vec![1, 2, 3, 45, 67];
416
417        let seq_str = SeqStr::from_display_iter(v.into_iter());
418
419        let seq_str2: SeqStr = seq_str.range(1..4).collect();
420        assert_eq!(seq_str2.len(), 3);
421        assert_eq!(&seq_str2[0], "2");
422        assert_eq!(&seq_str2[1], "3");
423        assert_eq!(&seq_str2[2], "45");
424
425        let seq_str2: SeqStr = seq_str.range(3..).collect();
426        assert_eq!(seq_str2.len(), 2);
427        assert_eq!(&seq_str2[0], "45");
428        assert_eq!(&seq_str2[1], "67");
429
430        let seq_str2: SeqStr = seq_str.range(..3).collect();
431        assert_eq!(seq_str2.len(), 3);
432        assert_eq!(&seq_str2[0], "1");
433        assert_eq!(&seq_str2[1], "2");
434        assert_eq!(&seq_str2[2], "3");
435
436        let seq_str2: SeqStr = seq_str.range(..).collect();
437        assert_eq!(seq_str2.len(), 5);
438        assert_eq!(&seq_str2[0], "1");
439        assert_eq!(&seq_str2[1], "2");
440        assert_eq!(&seq_str2[2], "3");
441        assert_eq!(&seq_str2[3], "45");
442        assert_eq!(&seq_str2[4], "67");
443    }
444
445    #[test]
446    fn iter_mut() {
447        let mut seq_str = SeqStr::from_iter(["asdf", "jkl;", ""]);
448
449        assert_eq!(seq_str.len(), 3);
450        assert_eq!(&seq_str[0], "asdf");
451        assert_eq!(&seq_str[1], "jkl;");
452        assert_eq!(&seq_str[2], "");
453
454        for s in seq_str.iter_mut() {
455            s.make_ascii_uppercase();
456        }
457
458        assert_eq!(seq_str.len(), 3);
459        assert_eq!(&seq_str[0], "ASDF");
460        assert_eq!(&seq_str[1], "JKL;");
461        assert_eq!(&seq_str[2], "");
462    }
463
464    #[test]
465    fn test_serde() {
466        let seq_str: SeqStr = serde_json::from_str("[\"asdf\", \"jkl;\", \"\"]").unwrap();
467
468        assert_eq!(seq_str.len(), 3);
469        assert_eq!(&seq_str[0], "asdf");
470        assert_eq!(&seq_str[1], "jkl;");
471        assert_eq!(&seq_str[2], "");
472
473        let ser = serde_json::to_string(&seq_str).unwrap();
474
475        let seq_str2: SeqStr = serde_json::from_str(&ser).unwrap();
476
477        assert_eq!(seq_str, seq_str2);
478    }
479
480    #[test]
481    fn test_join() {
482        let seq_str = SeqStr::from_iter(["asdf", "jkl;", "", ":)"]);
483
484        assert_eq!(seq_str.concat(), "asdfjkl;:)");
485        assert_eq!(seq_str.join(", "), "asdf, jkl;, , :)");
486    }
487}