1 /*!
2 This library provides heavily optimized routines for string search primitives.
3 
4 # Overview
5 
6 This section gives a brief high level overview of what this crate offers.
7 
8 * The top-level module provides routines for searching for 1, 2 or 3 bytes
9   in the forward or reverse direction. When searching for more than one byte,
10   positions are considered a match if the byte at that position matches any
11   of the bytes.
12 * The [`memmem`] sub-module provides forward and reverse substring search
13   routines.
14 
15 In all such cases, routines operate on `&[u8]` without regard to encoding. This
16 is exactly what you want when searching either UTF-8 or arbitrary bytes.
17 
18 # Example: using `memchr`
19 
20 This example shows how to use `memchr` to find the first occurrence of `z` in
21 a haystack:
22 
23 ```
24 use memchr::memchr;
25 
26 let haystack = b"foo bar baz quuz";
27 assert_eq!(Some(10), memchr(b'z', haystack));
28 ```
29 
30 # Example: matching one of three possible bytes
31 
32 This examples shows how to use `memrchr3` to find occurrences of `a`, `b` or
33 `c`, starting at the end of the haystack.
34 
35 ```
36 use memchr::memchr3_iter;
37 
38 let haystack = b"xyzaxyzbxyzc";
39 
40 let mut it = memchr3_iter(b'a', b'b', b'c', haystack).rev();
41 assert_eq!(Some(11), it.next());
42 assert_eq!(Some(7), it.next());
43 assert_eq!(Some(3), it.next());
44 assert_eq!(None, it.next());
45 ```
46 
47 # Example: iterating over substring matches
48 
49 This example shows how to use the [`memmem`] sub-module to find occurrences of
50 a substring in a haystack.
51 
52 ```
53 use memchr::memmem;
54 
55 let haystack = b"foo bar foo baz foo";
56 
57 let mut it = memmem::find_iter(haystack, "foo");
58 assert_eq!(Some(0), it.next());
59 assert_eq!(Some(8), it.next());
60 assert_eq!(Some(16), it.next());
61 assert_eq!(None, it.next());
62 ```
63 
64 # Example: repeating a search for the same needle
65 
66 It may be possible for the overhead of constructing a substring searcher to be
67 measurable in some workloads. In cases where the same needle is used to search
68 many haystacks, it is possible to do construction once and thus to avoid it for
69 subsequent searches. This can be done with a [`memmem::Finder`]:
70 
71 ```
72 use memchr::memmem;
73 
74 let finder = memmem::Finder::new("foo");
75 
76 assert_eq!(Some(4), finder.find(b"baz foo quux"));
77 assert_eq!(None, finder.find(b"quux baz bar"));
78 ```
79 
80 # Why use this crate?
81 
82 At first glance, the APIs provided by this crate might seem weird. Why provide
83 a dedicated routine like `memchr` for something that could be implemented
84 clearly and trivially in one line:
85 
86 ```
87 fn memchr(needle: u8, haystack: &[u8]) -> Option<usize> {
88     haystack.iter().position(|&b| b == needle)
89 }
90 ```
91 
92 Or similarly, why does this crate provide substring search routines when Rust's
93 core library already provides them?
94 
95 ```
96 fn search(haystack: &str, needle: &str) -> Option<usize> {
97     haystack.find(needle)
98 }
99 ```
100 
101 The primary reason for both of them to exist is performance. When it comes to
102 performance, at a high level at least, there are two primary ways to look at
103 it:
104 
105 * **Throughput**: For this, think about it as, "given some very large haystack
106   and a byte that never occurs in that haystack, how long does it take to
107   search through it and determine that it, in fact, does not occur?"
108 * **Latency**: For this, think about it as, "given a tiny haystack---just a
109   few bytes---how long does it take to determine if a byte is in it?"
110 
111 The `memchr` routine in this crate has _slightly_ worse latency than the
112 solution presented above, however, its throughput can easily be over an
113 order of magnitude faster. This is a good general purpose trade off to make.
114 You rarely lose, but often gain big.
115 
116 **NOTE:** The name `memchr` comes from the corresponding routine in libc. A key
117 advantage of using this library is that its performance is not tied to its
118 quality of implementation in the libc you happen to be using, which can vary
119 greatly from platform to platform.
120 
121 But what about substring search? This one is a bit more complicated. The
122 primary reason for its existence is still indeed performance, but it's also
123 useful because Rust's core library doesn't actually expose any substring
124 search routine on arbitrary bytes. The only substring search routine that
125 exists works exclusively on valid UTF-8.
126 
127 So if you have valid UTF-8, is there a reason to use this over the standard
128 library substring search routine? Yes. This routine is faster on almost every
129 metric, including latency. The natural question then, is why isn't this
130 implementation in the standard library, even if only for searching on UTF-8?
131 The reason is that the implementation details for using SIMD in the standard
132 library haven't quite been worked out yet.
133 
134 **NOTE:** Currently, only `x86_64` targets have highly accelerated
135 implementations of substring search. For `memchr`, all targets have
136 somewhat-accelerated implementations, while only `x86_64` targets have highly
137 accelerated implementations. This limitation is expected to be lifted once the
138 standard library exposes a platform independent SIMD API.
139 
140 # Crate features
141 
142 * **std** - When enabled (the default), this will permit this crate to use
143   features specific to the standard library. Currently, the only thing used
144   from the standard library is runtime SIMD CPU feature detection. This means
145   that this feature must be enabled to get AVX accelerated routines. When
146   `std` is not enabled, this crate will still attempt to use SSE2 accelerated
147   routines on `x86_64`.
148 * **libc** - When enabled (**not** the default), this library will use your
149   platform's libc implementation of `memchr` (and `memrchr` on Linux). This
150   can be useful on non-`x86_64` targets where the fallback implementation in
151   this crate is not as good as the one found in your libc. All other routines
152   (e.g., `memchr[23]` and substring search) unconditionally use the
153   implementation in this crate.
154 */
155 
156 #![deny(missing_docs)]
157 #![cfg_attr(not(feature = "std"), no_std)]
158 // It's not worth trying to gate all code on just miri, so turn off relevant
159 // dead code warnings.
160 #![cfg_attr(miri, allow(dead_code, unused_macros))]
161 
162 // Supporting 8-bit (or others) would be fine. If you need it, please submit a
163 // bug report at https://github.com/BurntSushi/rust-memchr
164 #[cfg(not(any(
165     target_pointer_width = "16",
166     target_pointer_width = "32",
167     target_pointer_width = "64"
168 )))]
169 compile_error!("memchr currently not supported on non-{16,32,64}");
170 
171 pub use crate::memchr::{
172     memchr, memchr2, memchr2_iter, memchr3, memchr3_iter, memchr_iter,
173     memrchr, memrchr2, memrchr2_iter, memrchr3, memrchr3_iter, memrchr_iter,
174     Memchr, Memchr2, Memchr3,
175 };
176 
177 mod cow;
178 mod memchr;
179 pub mod memmem;
180 #[cfg(test)]
181 mod tests;
182