1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
//! Fast, FFI-friendly string interning. A [`Ustr`] (**U**nique **Str**) is a
//! lightweight handle representing a static, immutable entry in a global string
//! cache, allowing for:
//! * Extremely fast string assignment and comparisons -- it's just a pointer
//!   comparison.
//! * Efficient storage -- only one copy of the string is held in memory, and
//!   getting access to it is just a pointer indirection.
//! * Fast hashing -- the precomputed hash is stored with the string
//! * Fast FFI -- the string is stored with a terminating null byte so can be
//!   passed to C directly without doing the CString dance.
//!
//! The downside is no strings are ever freed, so if you're creating lots and
//! lots of strings, you might run out of memory. On the other hand, War and
//! Peace is only 3MB, so it's probably fine.
//!
//! This crate is based on [OpenImageIO's ustring](https://github.com/OpenImageIO/oiio/blob/master/src/include/OpenImageIO/ustring.h) but it is NOT binary-compatible (yet). The underlying hash map implementation is directy ported from OIIO.
//!
//! # Usage
//!
//! ```
//! use ustr_fxhash::{Ustr, ustr, ustr as u};
//!
//! # unsafe { ustr_fxhash::_clear_cache() };
//! // Creation is quick and easy using either `Ustr::from` or the ustr function
//! // and only one copy of any string is stored
//! let u1 = Ustr::from("the quick brown fox");
//! let u2 = ustr("the quick brown fox");
//!
//! // Comparisons and copies are extremely cheap
//! let u3 = u1;
//! assert_eq!(u2, u3);
//!
//! // You can pass straight to FFI
//! let len = unsafe {
//!     libc::strlen(u1.as_char_ptr())
//! };
//! assert_eq!(len, 19);
//!
//! // Use as_str() to get a &str
//! let words: Vec<&str> = u1.as_str().split_whitespace().collect();
//! assert_eq!(words, ["the", "quick", "brown", "fox"]);
//!
//! // For best performance when using Ustr as key for a HashMap or HashSet,
//! // you'll want to use the precomputed hash. To make this easier, just use
//! // the UstrMap and UstrSet exports:
//! use ustr_fxhash::UstrMap;
//!
//! // Key type is always Ustr
//! let mut map: UstrMap<usize> = UstrMap::default();
//! map.insert(u1, 17);
//! assert_eq!(*map.get(&u1).unwrap(), 17);
//! ```
//!
//! By enabling the `"serde"` feature you can serialize individual `Ustr`s
//! or the whole cache with serde.
//!
//! ```
//! # #[cfg(feature = "serde")] {
//! use ustr_fxhash::{Ustr, ustr};
//! let u_ser = ustr("serde");
//! let json = serde_json::to_string(&u_ser).unwrap();
//! let u_de : Ustr = serde_json::from_str(&json).unwrap();
//! assert_eq!(u_ser, u_de);
//! # }
//! ```
//!
//! Since the cache is global, use the `ustr::DeserializedCache` dummy object to
//! drive the deserialization.
//!
//! ```
//! # #[cfg(feature = "serde")] {
//! use ustr_fxhash::{Ustr, ustr};
//! ustr("Send me to JSON and back");
//! let json = serde_json::to_string(ustr::cache()).unwrap();
//!
//! // ... some time later ...
//! let _: ustr::DeserializedCache = serde_json::from_str(&json).unwrap();
//! assert_eq!(ustr_fxhash::num_entries(), 1);
//! assert_eq!(ustr::string_cache_iter().collect::<Vec<_>>(), vec!["Send me to JSON and back"]);
//! # }
//! ```
//!
//! ## Why?
//!
//! It is common in certain types of applications to use strings as identifiers,
//! but not really do any processing with them.
//! To paraphrase from OIIO's `Ustring` documentation -- compared to standard
//! strings, `Ustr`s have several advantages:
//!
//!   - Each individual `Ustr` is very small -- in fact, we guarantee that a
//!     `Ustr` is the same size and memory layout as an ordinary `*u8`.
//!   - Storage is frugal, since there is only one allocated copy of each unique
//!     character sequence, throughout the lifetime of the program.
//!   - Assignment from one `Ustr` to another is just copy of the pointer; no
//!     allocation, no character copying, no reference counting.
//!   - Equality testing (do the strings contain the same characters) is a
//!     single operation, the comparison of the pointer.
//!   - Memory allocation only occurs when a new `Ustr` is constructed from raw
//!     characters the FIRST time -- subsequent constructions of the same string
//!     just finds it in the canonial string set, but doesn't need to allocate
//!     new storage.  Destruction of a `Ustr` is trivial, there is no
//!     de-allocation because the canonical version stays in the set.  Also,
//!     therefore, no user code mistake can lead to memory leaks.
//!
//! But there are some problems, too.  Canonical strings are never freed
//! from the table.  So in some sense all the strings "leak", but they
//! only leak one copy for each unique string that the program ever comes
//! across.
//!
//! On the whole, `Ustr`s are a really great string representation
//!   - if you tend to have (relatively) few unique strings, but many copies of
//!     those strings;
//!   - if the creation of strings from raw characters is relatively rare
//!     compared to copying or comparing to existing strings;
//!   - if you tend to make the same strings over and over again, and if it's
//!     relatively rare that a single unique character sequence is used only
//!     once in the entire lifetime of the program;
//!   - if your most common string operations are assignment and equality
//!     testing and you want them to be as fast as possible;
//!   - if you are doing relatively little character-by-character assembly of
//!     strings, string concatenation, or other "string manipulation" (other
//!     than equality testing).
//!
//! `Ustr`s are not so hot
//!   - if your program tends to have very few copies of each character sequence
//!     over the entire lifetime of the program;
//!   - if your program tends to generate a huge variety of unique strings over
//!     its lifetime, each of which is used only a short time and then
//!     discarded, never to be needed again;
//!   - if you don't need to do a lot of string assignment or equality testing,
//!     but lots of more complex string manipulation.
//!
//! ## Safety and Compatibility
//!
//! This crate contains a significant amount of unsafe but usage has been
//! checked and is well-documented. It is also run through Miri as part of the
//! CI process. I use it regularly on 64-bit systems, and it has passed Miri on
//! a 32-bit system as well, bit 32-bit is not checked regularly. If you want to
//! use it on 32-bit, please make sure to run Miri and open and issue if you
//! find any problems.
use parking_lot::Mutex;
use std::fmt;
use std::mem::size_of;
use std::str::FromStr;

mod stringcache;
pub use stringcache::*;
#[cfg(feature = "serde")]
pub mod serialization;
#[cfg(feature = "serde")]
pub use serialization::DeserializedCache;

mod bumpalloc;

mod hash;
pub use hash::*;
use std::cmp::Ordering;
use std::hash::{Hash, Hasher};
use std::ptr::NonNull;

/// A handle representing a string in the global string cache.
///
/// To use, create one using `Ustr::from` or the `ustr` function. You can freely
/// copy, destroy or send Ustrs to other threads: the underlying string is
/// always valid in memory (and is never destroyed).
#[derive(Copy, Clone, PartialEq)]
#[repr(transparent)]
pub struct Ustr {
    char_ptr: NonNull<u8>,
}

/// Defer to &str for equality - lexicographic ordering will be slower than
/// pointer comparison, but much less surprising if you use Ustrs as keys in
/// e.g. a BTreeMap
impl Ord for Ustr {
    fn cmp(&self, other: &Self) -> Ordering {
        self.as_str().cmp(other.as_str())
    }
}

/// Defer to &str for equality - lexicographic ordering will be slower than
/// pointer comparison, but much less surprising if you use Ustrs as keys in
/// e.g. a BTreeMap
impl PartialOrd for Ustr {
    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
        self.as_str().partial_cmp(other.as_str())
    }
}

impl Ustr {
    /// Create a new Ustr from the given &str.
    ///
    /// You can also use the ustr function
    /// ```
    /// use ustr_fxhash::{Ustr, ustr as u};
    /// # unsafe { ustr_fxhash::_clear_cache() };
    ///
    /// let u1 = Ustr::from("the quick brown fox");
    /// let u2 = u("the quick brown fox");
    /// assert_eq!(u1, u2);
    /// assert_eq!(ustr_fxhash::num_entries(), 1);
    /// ```
    pub fn from(string: &str) -> Ustr {
        let hash = {
            let mut hasher = fxhash::FxHasher::default();
            hasher.write(string.as_bytes());
            hasher.finish()
        };
        let mut sc = STRING_CACHE.0[whichbin(hash)].lock();
        Ustr {
            // SAFETY: sc.insert does not give back a null pointer
            char_ptr: unsafe {
                NonNull::new_unchecked(sc.insert(string, hash) as *mut _)
            },
        }
    }

    pub fn from_existing(string: &str) -> Option<Ustr> {
        let hash = {
            let mut hasher = fxhash::FxHasher::default();
            hasher.write(string.as_bytes());
            hasher.finish()
        };
        let sc = STRING_CACHE.0[whichbin(hash)].lock();
        sc.get_existing(string, hash).map(|ptr| Ustr {
            char_ptr: unsafe { NonNull::new_unchecked(ptr as *mut _) },
        })
    }

    /// Get the cached string as a &str
    /// ```
    /// use ustr_fxhash::ustr as u;
    /// # unsafe { ustr_fxhash::_clear_cache() };
    ///
    /// let u_fox = u("the quick brown fox");
    /// let words: Vec<&str> = u_fox.as_str().split_whitespace().collect();
    /// assert_eq!(words, ["the", "quick", "brown", "fox"]);
    /// ```
    pub fn as_str(&self) -> &'static str {
        // This is safe if:
        // 1) self.char_ptr points to a valid address
        // 2) len is a usize stored usize aligned usize bytes before char_ptr
        // 3) char_ptr points to a valid UTF-8 string of len bytes.
        // All these are guaranteed by StringCache::insert() and by the fact
        // we can only construct a Ustr from a valid &str.
        unsafe {
            std::str::from_utf8_unchecked(std::slice::from_raw_parts(
                self.char_ptr.as_ptr(),
                self.len(),
            ))
        }
    }

    /// Get the cached string as a C char*.
    ///
    /// This includes the null terminator so is safe to pass straight to FFI.
    ///
    /// ```
    /// use ustr_fxhash::ustr as u;
    /// # unsafe { ustr_fxhash::_clear_cache() };
    ///
    /// let u_fox = u("the quick brown fox");
    /// let len = unsafe {
    ///     libc::strlen(u_fox.as_char_ptr())
    /// };
    /// assert_eq!(len, 19);
    /// ```
    ///
    /// # Safety
    /// This is just passing a raw byte array with a null terminator to C.
    /// If your source string contains non-ascii bytes then this will pass them
    /// straight along with no checking.
    /// The string is **immutable**. That means that if you modify it across the
    /// FFI boundary then all sorts of terrible things will happen.
    pub fn as_char_ptr(&self) -> *const std::os::raw::c_char {
        self.char_ptr.as_ptr() as *const std::os::raw::c_char
    }

    /// Get this ustr as a CStr
    ///
    /// This is useful for passing to APIs (like ash) that use CStr
    ///
    /// # Safety
    /// This function by itself is safe as the pointer and length are
    /// guaranteed to be valid. All the same caveats for the use of the CStr
    /// as given in the CSstr docs apply
    pub fn as_cstr(&self) -> &std::ffi::CStr {
        unsafe {
            std::ffi::CStr::from_bytes_with_nul_unchecked(
                std::slice::from_raw_parts(self.as_ptr(), self.len() + 1),
            )
        }
    }

    // Get a raw pointer to the StringCacheEntry
    #[inline]
    fn as_string_cache_entry(&self) -> &StringCacheEntry {
        // The allocator guarantees that the alignment is correct and that
        // this pointer is non-null
        unsafe {
            &*((self.char_ptr.as_ptr() as usize - size_of::<StringCacheEntry>())
                as *const StringCacheEntry)
        }
    }

    /// Get the length (in bytes) of this string.
    #[inline]
    pub fn len(&self) -> usize {
        self.as_string_cache_entry().len
    }

    /// Returns true if the length is zero.
    pub fn is_empty(&self) -> bool {
        self.len() == 0
    }

    /// Get the precomputed hash for this string
    #[inline]
    pub fn precomputed_hash(&self) -> u64 {
        self.as_string_cache_entry().hash
    }

    /// Get an owned String copy of this string.
    pub fn to_owned(&self) -> String {
        self.as_str().to_owned()
    }
}

// We're safe to impl these because the strings they reference are immutable
// and for all intents and purposes 'static since they're never deleted after
// being created
unsafe impl Send for Ustr {}
unsafe impl Sync for Ustr {}

impl PartialEq<&str> for Ustr {
    fn eq(&self, other: &&str) -> bool {
        self.as_str() == *other
    }
}

impl PartialEq<String> for Ustr {
    fn eq(&self, other: &String) -> bool {
        self.as_str() == other
    }
}

impl Eq for Ustr {}

impl AsRef<str> for Ustr {
    fn as_ref(&self) -> &str {
        self.as_str()
    }
}

impl FromStr for Ustr {
    type Err = std::string::ParseError;

    #[inline]
    fn from_str(s: &str) -> Result<Self, Self::Err> {
        Ok(Ustr::from(s))
    }
}

impl From<&str> for Ustr {
    fn from(s: &str) -> Ustr {
        Ustr::from(s)
    }
}

impl From<Ustr> for &'static str {
    fn from(s: Ustr) -> &'static str {
        s.as_str()
    }
}

impl From<String> for Ustr {
    fn from(s: String) -> Ustr {
        Ustr::from(&s)
    }
}

impl Default for Ustr {
    fn default() -> Self {
        Ustr::from("")
    }
}

impl std::ops::Deref for Ustr {
    type Target = str;
    fn deref(&self) -> &Self::Target {
        self.as_str()
    }
}

impl fmt::Display for Ustr {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        write!(f, "{}", self.as_str())
    }
}

impl fmt::Debug for Ustr {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        write!(f, "u!({:?})", self.as_str())
    }
}

// Just feed the precomputed hash into the Hasher. Note that this will of course
// be terrible unless the Hasher in question is expecting a precomputed hash.
#[allow(clippy::derived_hash_with_manual_eq)]
impl Hash for Ustr {
    fn hash<H: Hasher>(&self, state: &mut H) {
        self.precomputed_hash().hash(state);
    }
}

/// DO NOT CALL THIS.
///
/// Clears the cache - used for benchmarking and testing purposes to clear the
/// cache. Calling this will invalidate any previously created `UStr`s and
/// probably cause your house to burn down. DO NOT CALL THIS.
///
/// # Safety
/// DO NOT CALL THIS.
#[doc(hidden)]
pub unsafe fn _clear_cache() {
    for m in STRING_CACHE.0.iter() {
        m.lock().clear();
    }
}

/// Returns the total amount of memory allocated and in use by the cache in
/// bytes
pub fn total_allocated() -> usize {
    STRING_CACHE
        .0
        .iter()
        .map(|sc| {
            let t = sc.lock().total_allocated();

            t
        })
        .sum()
}

/// Returns the total amount of memory reserved by the cache in bytes
pub fn total_capacity() -> usize {
    STRING_CACHE
        .0
        .iter()
        .map(|sc| {
            let t = sc.lock().total_capacity();
            t
        })
        .sum()
}

/// Create a new Ustr from the given &str.
///
/// ```
/// use ustr_fxhash::ustr;
/// # unsafe { ustr_fxhash::_clear_cache() };
///
/// let u1 = ustr("the quick brown fox");
/// let u2 = ustr("the quick brown fox");
/// assert_eq!(u1, u2);
/// assert_eq!(ustr_fxhash::num_entries(), 1);
/// ```
#[inline]
pub fn ustr(s: &str) -> Ustr {
    Ustr::from(s)
}

/// Create a new Ustr from the given &str but only if it already exists in the
/// string cache.
///
/// ```
/// use ustr_fxhash::{ustr, existing_ustr};
/// # unsafe { ustr_fxhash::_clear_cache() };
///
/// let u1 = existing_ustr("the quick brown fox");
/// let u2 = ustr("the quick brown fox");
/// let u3 = existing_ustr("the quick brown fox");
/// assert_eq!(u1, None);
/// assert_eq!(u3, Some(u2));
/// ```
#[inline]
pub fn existing_ustr(s: &str) -> Option<Ustr> {
    Ustr::from_existing(s)
}

/// Utility function to get a reference to the main cache object for use with
/// serialization.
///
/// # Examples
/// ```
/// # use ustr_fxhash::{Ustr, ustr, ustr as u};
/// # #[cfg(feature="serde")]
/// # {
/// # unsafe { ustr_fxhash::_clear_cache() };
/// ustr("Send me to JSON and back");
/// let json = serde_json::to_string(ustr::cache()).unwrap();
/// # }
pub fn cache() -> &'static Bins {
    &STRING_CACHE
}

/// Returns the number of unique strings in the cache
///
/// This may be an underestimate if other threads are writing to the cache
/// concurrently.
///
/// ```
/// use ustr_fxhash::ustr as u;
///
/// let _ = u("Hello");
/// let _ = u(", World!");
/// assert_eq!(ustr_fxhash::num_entries(), 2);
/// ```
pub fn num_entries() -> usize {
    STRING_CACHE
        .0
        .iter()
        .map(|sc| {
            let t = sc.lock().num_entries();
            t
        })
        .sum()
}

#[doc(hidden)]
pub fn num_entries_per_bin() -> Vec<usize> {
    STRING_CACHE
        .0
        .iter()
        .map(|sc| {
            let t = sc.lock().num_entries();
            t
        })
        .collect::<Vec<_>>()
}

/// Return an iterator over the entire string cache.
///
/// If another thread is adding strings concurrently to this call then they
/// might not show up in the view of the cache presented by this iterator.
///
/// # Safety
/// This returns an iterator to the state of the cache at the time when
/// `string_cache_iter()` was called. It is of course possible that another
/// thread will add more strings to the cache after this, but since we never
/// destroy the strings, they remain valid, meaning it's safe to iterate over
/// them, the list just might not be completely up to date.
pub fn string_cache_iter() -> StringCacheIterator {
    let mut allocs = Vec::new();
    for m in STRING_CACHE.0.iter() {
        let sc = m.lock();
        // the start of the allocator's data is actually the ptr, start() just
        // points to the beginning of the allocated region. The first bytes will
        // be uninitialized since we're bumping down
        for a in &sc.old_allocs {
            allocs.push((a.ptr(), a.end()));
        }
        let ptr = sc.alloc.ptr();
        let end = sc.alloc.end();
        if ptr != end {
            allocs.push((sc.alloc.ptr(), sc.alloc.end()));
        }
    }

    let current_ptr = allocs[0].0;
    StringCacheIterator {
        allocs,
        current_alloc: 0,
        current_ptr,
    }
}

#[repr(transparent)]
pub struct Bins(pub(crate) [Mutex<StringCache>; NUM_BINS]);

#[cfg(test)]
mod tests {
    use lazy_static::lazy_static;
    use std::sync::Mutex;

    lazy_static! {
        static ref TEST_LOCK: Mutex<()> = Mutex::new(());
    }

    #[test]
    fn it_works() {
        let _t = TEST_LOCK.lock();
        use super::ustr as u;

        let u_hello = u("hello");
        assert_eq!(u_hello, "hello");
        let u_world = u("world");
        assert_eq!(u_world, String::from("world"));
    }

    #[test]
    fn empty_string() {
        let _t = TEST_LOCK.lock();
        use super::ustr as u;

        unsafe {
            super::_clear_cache();
        }

        let _empty = u("");
        let empty = u("");

        assert!(empty.as_str().is_empty());
        assert_eq!(super::num_entries(), 1);
    }

    #[test]
    fn c_str_works() {
        let _t = TEST_LOCK.lock();
        use super::ustr as u;
        use std::ffi::CStr;

        let s_fox = "The quick brown fox jumps over the lazy dog.";
        let u_fox = u(s_fox);
        let fox = unsafe { CStr::from_ptr(u_fox.as_char_ptr()) }
            .to_string_lossy()
            .into_owned();
        assert_eq!(fox, s_fox);

        let s_odys = "Τη γλώσσα μου έδωσαν ελληνική";
        let u_odys = u(s_odys);
        let odys = unsafe { CStr::from_ptr(u_odys.as_char_ptr()) }
            .to_string_lossy()
            .into_owned();
        assert_eq!(odys, s_odys);
    }

    #[test]
    // We have to disable miri here as it's far too slow unfortunately
    #[cfg_attr(miri, ignore)]
    fn blns() {
        let _t = TEST_LOCK.lock();
        use super::{string_cache_iter, ustr as u};
        use std::collections::HashSet;

        // clear the cache first or our results will be wrong
        unsafe { super::_clear_cache() };

        // let path =
        // std::path::Path::new(&std::env::var("CARGO_MANIFEST_DIR").unwrap())
        //     .join("data")
        //     .join("blns.txt");
        // let blns = std::fs::read_to_string(path).unwrap();
        let blns = include_str!("../data/blns.txt");

        let mut hs = HashSet::new();
        for s in blns.split_whitespace() {
            hs.insert(s);
        }

        let mut us = Vec::new();
        let mut ss = Vec::new();

        for s in blns.split_whitespace().cycle().take(100_000) {
            let u = u(s);
            us.push(u);
            ss.push(s.to_owned());
        }

        let mut hs_u = HashSet::new();
        for s in string_cache_iter() {
            hs_u.insert(s);
        }
        let diff: HashSet<_> = hs.difference(&hs_u).collect();

        // check that the number of entries is the same
        assert_eq!(super::num_entries(), hs.len());

        // check that we have the exact same (unique) strings in the cache as in
        // the source data
        assert_eq!(diff.len(), 0);

        let nbs = super::num_entries_per_bin();
        println!("{:?}", nbs);

        println!("Total allocated: {}", super::total_allocated());
        println!("Total capacity: {}", super::total_capacity());

        println!(
            "size of StringCache: {}",
            std::mem::size_of::<super::StringCache>()
        );
    }

    #[test]
    // We have to disable miri here as it's far too slow unfortunately
    #[cfg_attr(miri, ignore)]
    fn raft() {
        let _t = TEST_LOCK.lock();
        use super::ustr as u;
        use std::sync::Arc;

        // let path =
        // std::path::Path::new(&std::env::var("CARGO_MANIFEST_DIR").unwrap())
        //     .join("data")
        //     .join("raft-large-directories.txt");
        // let raft = std::fs::read_to_string(path).unwrap();
        let raft = include_str!("../data/raft-large-directories.txt");
        let raft = Arc::new(
            raft.split_whitespace()
                .collect::<Vec<_>>()
                .chunks(3)
                .map(|s| {
                    if s.len() == 3 {
                        format!("{}/{}/{}", s[0], s[1], s[2])
                    } else {
                        s[0].to_owned()
                    }
                })
                .collect::<Vec<_>>(),
        );

        let s = raft.clone();
        for _ in 0..600 {
            let mut v = Vec::with_capacity(20_000);
            unsafe { super::_clear_cache() };
            for s in s.iter().cycle().take(20_000) {
                v.push(u(s));
            }
        }
    }

    // This test is to have miri check the allocation code paths, but miri
    // can't open files so it's not usable right now
    // #[test]
    // fn words() {
    //     let _t = TEST_LOCK.lock();
    //     use super::ustr as u;
    //     use std::sync::Arc;

    //     let path = std::path::Path::new("/usr/share/dict/words");
    //     let wordlist = std::fs::read_to_string(path).unwrap();
    //     let wordlist = Arc::new(
    //         wordlist
    //             .split_whitespace()
    //             .collect::<Vec<_>>()
    //             .chunks(7)
    //             .cycle()
    //             .take(4_000_000)
    //             .enumerate()
    //             .map(|(i, s)| u(&format!("{}{}", i, s.join("-"))))
    //             .collect::<Vec<_>>(),
    //     );
    // }

    #[cfg(all(feature = "serde", not(miri)))]
    #[test]
    fn serialization() {
        let _t = TEST_LOCK.lock();
        use super::{string_cache_iter, ustr as u};
        use std::collections::HashSet;

        // clear the cache first or our results will be wrong
        unsafe { super::_clear_cache() };

        let path = std::path::Path::new(
            &std::env::var("CARGO_MANIFEST_DIR")
                .expect("CARGO_MANIFEST_DIR not set"),
        )
        .join("data")
        .join("blns.txt");
        let blns = std::fs::read_to_string(path).unwrap();

        let mut hs = HashSet::new();
        for s in blns.split_whitespace() {
            hs.insert(s);
        }

        let mut us = Vec::new();
        let mut ss = Vec::new();

        for s in blns.split_whitespace().cycle().take(100_000) {
            let u = u(s);
            us.push(u);
            ss.push(s.to_owned());
        }

        let json = serde_json::to_string(super::cache()).unwrap();
        unsafe {
            super::_clear_cache();
        }
        let _: super::DeserializedCache = serde_json::from_str(&json).unwrap();

        // now check that we've got the same data in the cache still
        let mut hs_u = HashSet::new();
        for s in string_cache_iter() {
            hs_u.insert(s);
        }
        let diff: HashSet<_> = hs.difference(&hs_u).collect();

        // check that the number of entries is the same
        assert_eq!(super::num_entries(), hs.len());

        // check that we have the exact same (unique) strings in the cache as in
        // the source data
        assert_eq!(diff.len(), 0);
    }

    #[cfg(all(feature = "serde", not(miri)))]
    #[test]
    fn serialization_ustr() {
        use super::{ustr, Ustr};

        let u_hello = ustr("hello");

        let json = serde_json::to_string(&u_hello).unwrap();
        let me_hello: Ustr = serde_json::from_str(&json).unwrap();

        assert_eq!(u_hello, me_hello);
    }

    #[test]
    fn partial_ord() {
        let _t = TEST_LOCK.lock();
        use super::ustr;
        let str_a = ustr("aaa");
        let str_z = ustr("zzz");
        let str_k = ustr("kkk");
        assert!(str_a < str_k);
        assert!(str_k < str_z);
    }

    #[test]
    fn ord() {
        let _t = TEST_LOCK.lock();
        use super::ustr;
        let u_apple = ustr("apple");
        let u_bravo = ustr("bravo");
        let u_charlie = ustr("charlie");
        let u_delta = ustr("delta");

        let mut v = vec![u_delta, u_bravo, u_charlie, u_apple];
        v.sort();
        assert_eq!(v, vec![u_apple, u_bravo, u_charlie, u_delta]);
    }

    fn takes_into_str<'a, S: Into<&'a str>>(s: S) -> &'a str {
        s.into()
    }

    #[test]
    fn test_into_str() {
        let _t = TEST_LOCK.lock();
        use super::ustr;

        assert_eq!("converted", takes_into_str(ustr("converted")));
    }

    #[test]
    fn test_existing_ustr() {
        let _t = TEST_LOCK.lock();
        use super::{existing_ustr, ustr};
        assert_eq!(existing_ustr("hello world!"), None);
        let s1 = ustr("hello world!");
        let s2 = existing_ustr("hello world!");
        assert_eq!(Some(s1), s2);
    }
}

lazy_static::lazy_static! {
    static ref STRING_CACHE: Bins = {
        use std::mem::{self, MaybeUninit};
        // This deeply unsafe feeling dance allows us to initialize an array of
        // arbitrary size and will have to tide us over until const generics
        // land. See:
        // https://doc.rust-lang.org/beta/std/mem/union.MaybeUninit.html#initializing-an-array-element-by-element

        // Create an uninitialized array of `MaybeUninit`. The `assume_init` is
        // safe because the type we are claiming to have initialized here is a
        // bunch of `MaybeUninit`s, which do not require initialization.
        let mut bins: [MaybeUninit<Mutex<StringCache>>; NUM_BINS] = unsafe {
            MaybeUninit::uninit().assume_init()
        };

        // Dropping a `MaybeUninit` does nothing. Thus using raw pointer
        // assignment instead of `ptr::write` does not cause the old
        // uninitialized value to be dropped. Also if there is a panic during
        // this loop, we have a memory leak, but there is no memory safety
        // issue.
        for bin in &mut bins[..] {
            *bin = MaybeUninit::new(Mutex::new(StringCache::default()));
        }

        // Everything is initialized. Transmute the array to the
        // initialized type.
        unsafe { mem::transmute::<_, Bins>(bins) }
    };
}

// Use the top bits of the hash to choose a bin
#[inline]
fn whichbin(hash: u64) -> usize {
    ((hash >> TOP_SHIFT as u64) % NUM_BINS as u64) as usize
}