• Stars
    star
    258
  • Rank 157,189 (Top 4 %)
  • Language
    Rust
  • License
    The Unlicense
  • Created over 9 years ago
  • Updated 11 months ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Fast suffix arrays for Rust (with Unicode support).

suffix

Fast linear time & space suffix arrays for Rust. Supports Unicode!

Build status

Dual-licensed under MIT or the UNLICENSE.

Documentation

https://docs.rs/suffix

If you just want the details on how construction algorithm used, see the documentation for the SuffixTable type. This is where you'll find info on exactly how much overhead is required.

Installation

This crate works with Cargo and is on crates.io. The package is regularly updated. Add it to your Cargo.toml like so:

[dependencies]
suffix = "1.2"

Examples

Usage is simple. Just create a suffix array and search:

use suffix::SuffixTable;

fn main() {
  let st = SuffixTable::new("the quick brown fox was quick.");
  assert_eq!(st.positions("quick"), &[4, 24]);
}

There is also a command line program, stree, that can be used to visualize suffix trees:

git clone git://github.com/BurntSushi/suffix
cd suffix/stree_cmd
cargo build --release
./target/release/stree "banana" | dot -Tpng | xv -

And here's what it looks like:

"banana" suffix tree

Status of implementation

The big thing missing at the moment is a generalized suffix array. I started out with the intention to build them into the construction algorithm, but this has proved more difficult than I thought.

A kind-of-sort-of compromise is to append your distinct texts together, and separate them with a character that doesn't appear in your document. (This is technically incorrect, but maybe your documents don't contain any NUL characters.) During construction of this one giant string, you should record the offsets of where each document starts and stops. Then build a SuffixTable with your giant string. After searching with the SuffixTable, you can find the original document by doing a binary search on your list of documents.

I'm currently experimenting with different techniques to do this.

Benchmarks

Here are some very rough benchmarks that compare suffix table searching with searching in the using standard library functions. Note that these benchmarks explicitly do not include the construction of the suffix table. The premise of a suffix table is that you can afford to do that once---but you hope to gain much faster queries once you do.

test search_scan_exists_many            ... bench:       2,964 ns/iter (+/- 180)
test search_scan_exists_one             ... bench:          19 ns/iter (+/- 1)
test search_scan_not_exists             ... bench:      84,645 ns/iter (+/- 3,558)
test search_suffix_exists_many          ... bench:         228 ns/iter (+/- 65)
test search_suffix_exists_many_contains ... bench:         102 ns/iter (+/- 10)
test search_suffix_exists_one           ... bench:         162 ns/iter (+/- 13)
test search_suffix_exists_one_contains  ... bench:           8 ns/iter (+/- 0)
test search_suffix_not_exists           ... bench:         177 ns/iter (+/- 21)
test search_suffix_not_exists_contains  ... bench:          50 ns/iter (+/- 6)

The "many" benchmarks test repeated queries that match. The "one" benchmarks test a single query that matches. The "not_exists" benchmarks test a single query that does not match. Finally, the "contains" benchmark test existence rather finding all positions.

One thing you might take away from here is that you'll get a very large performance boost if many of your queries don't match. A linear scan takes a long time to fail!

And here are some completely useless benchmarks on suffix array construction. They compare the linear time algorithm with the naive construction algorithm (call sort on all suffixes, which is O(n^2 * logn)).

test naive_dna_medium                   ... bench:  22,307,313 ns/iter (+/- 939,557)
test naive_dna_small                    ... bench:   1,785,734 ns/iter (+/- 43,401)
test naive_small                        ... bench:         228 ns/iter (+/- 10)
test sais_dna_medium                    ... bench:   7,514,327 ns/iter (+/- 280,544)
test sais_dna_small                     ... bench:     712,938 ns/iter (+/- 34,730)
test sais_small                         ... bench:       1,038 ns/iter (+/- 58)

These benchmarks might make you say, "Whoa, the special algorithm isn't that much faster." That's because the data just isn't big enough. And when it is big enough, a micro benchmark is useless. Why? Because using the naive algorithm will just burn your CPUs until the end of the time.

It would be more useful to compare this to other suffix array implementations, but I haven't had time yet. Moreover, most (all?) don't support Unicode and instead operate on bytes, which means they aren't paying the overhead of decoding UTF-8.

More Repositories

1

ripgrep

ripgrep recursively searches directories for a regex pattern while respecting your gitignore
Rust
47,566
star
2

xsv

A fast CSV command line toolkit written in Rust.
Rust
10,306
star
3

toml

TOML parser for Golang with reflection.
Go
4,464
star
4

quickcheck

Automated property based testing for Rust (with shrinking).
Rust
2,323
star
5

erd

Translates a plain text description of a relational database schema to a graphical entity-relationship diagram.
Haskell
1,792
star
6

fst

Represent large sets and maps compactly with finite state transducers.
Rust
1,758
star
7

rust-csv

A CSV parser for Rust, with Serde support.
Rust
1,671
star
8

jiff

A date-time library for Rust that encourages you to jump into the pit of success.
Rust
1,644
star
9

walkdir

Rust library for walking directories recursively.
Rust
1,259
star
10

nflgame

An API to retrieve and read NFL Game Center JSON data. It can work with real-time data, which can be used for fantasy football.
Python
1,257
star
11

nfldb

A library to manage and update NFL data in a relational database.
Python
1,078
star
12

aho-corasick

A fast implementation of Aho-Corasick in Rust.
Rust
1,006
star
13

byteorder

Rust library for reading/writing numbers in big-endian and little-endian.
Rust
966
star
14

wingo

A fully-featured window manager written in Go.
Go
958
star
15

memchr

Optimized string search routines for Rust.
Rust
858
star
16

bstr

A string type for Rust that is not required to be valid UTF-8.
Rust
775
star
17

advent-of-code

Rust solutions to AoC 2018
Rust
479
star
18

xgb

The X Go Binding is a low-level API to communicate with the X server. It is modeled on XCB and supports many X extensions.
Go
472
star
19

termcolor

Cross platform terminal colors for Rust.
Rust
456
star
20

rust-snappy

Snappy compression implemented in Rust (including the Snappy frame format).
Rust
440
star
21

go-sumtype

A simple utility for running exhaustiveness checks on Go "sum types."
Go
417
star
22

chan

Multi-producer, multi-consumer concurrent channel for Rust.
Rust
390
star
23

regex-automata

A low level regular expression library that uses deterministic finite automata.
Rust
351
star
24

cargo-benchcmp

A small utility to compare Rust micro-benchmarks.
Rust
338
star
25

rure-go

Go bindings to Rust's regex engine.
Go
249
star
26

tabwriter

Elastic tabstops for Rust.
Rust
246
star
27

imdb-rename

A command line tool to rename media files based on titles from IMDb.
Rust
224
star
28

rebar

A biased barometer for gauging the relative speed of some regex engines on a curated set of tasks.
Python
223
star
29

critcmp

A command line tool for comparing benchmarks run by Criterion.
Rust
211
star
30

ty

Easy parametric polymorphism at run time using completely unidiomatic Go.
Go
198
star
31

xgbutil

A utility library to make use of the X Go Binding easier. (Implements EWMH and ICCCM specs, key binding support, etc.)
Go
191
star
32

pytyle3

An updated (and much faster) version of pytyle that uses xpybutil and is compatible with Openbox Multihead.
Python
181
star
33

dotfiles

My configuration files and personal collection of scripts.
Vim Script
153
star
34

rsc-regexp

Translations of a simple C program to Rust.
Rust
137
star
35

rust-cbor

CBOR (binary JSON) for Rust with automatic type based decoding and encoding.
Rust
128
star
36

chan-signal

Respond to OS signals with channels.
Rust
125
star
37

goim

Goim is a robust command line utility to maintain and query the Internet Movie Database (IMDb).
Go
117
star
38

same-file

Cross platform Rust library for checking whether two file paths are the same file.
Rust
100
star
39

clibs

A smattering of miscellaneous C libraries. Includes sane argument parsing, a thread-safe multi-producer/multi-consumer queue, and implementation of common data structures (hashmaps, vectors and linked lists).
C
98
star
40

ucd-generate

A command line tool to generate Unicode tables as source code.
Rust
93
star
41

nflvid

An experimental library to map play meta data to footage of that play.
Python
91
star
42

rust-stats

Basic statistical functions on streams for Rust.
Rust
87
star
43

migration

Package migration for Golang automatically handles versioning of a database schema by applying a series of migrations supplied by the client.
Go
80
star
44

winapi-util

Safe wrappers for various Windows specific APIs.
Rust
64
star
45

xpybutil

An incomplete xcb-util port plus some extras
Python
62
star
46

graphics-go

Automatically exported from code.google.com/p/graphics-go
Go
60
star
47

rust-pcre2

High level Rust bindings to PCRE2.
C
54
star
48

blog

My blog.
Rust
52
star
49

rust-sorts

Implementations of common sorting algorithms in Rust with comprehensive tests and benchmarks.
Rust
51
star
50

openbox-multihead

Openbox with patches for enhanced multihead support.
C
46
star
51

nakala

A low level embedded information retrieval system.
Rust
45
star
52

nflfan

View your fantasy teams with nfldb using a web interface.
JavaScript
43
star
53

utf8-ranges

Convert contiguous ranges of Unicode codepoints to UTF-8 byte ranges.
Rust
42
star
54

globset

A globbing library for Rust.
Rust
41
star
55

rtmpdump-ksv

rtmpdump with ksv's patch. Intended to track upstream git://git.ffmpeg.org/rtmpdump as well.
C
40
star
56

regexp

A regular expression library implemented in Rust.
Rust
37
star
57

xdg

A Go package for reading config and data files according to the XDG Base Directory specification.
Go
35
star
58

locker

A simple Golang package for conveniently using named read/write locks. Useful for synchronizing access to session based storage in web applications.
Go
34
star
59

nflcmd

A collection of command line utilities for viewing NFL statistics and rankings with nfldb.
Python
32
star
60

notes

A collection of small notes that aren't appropriate for my blog.
31
star
61

mempool

A fast thread safe memory pool for reusing allocations.
Rust
29
star
62

gribble

A command oriented language whose environment is defined through Go struct types by reflection.
Go
28
star
63

vcr

A simple wrapper tool around ffmpeg to capture video from a VCR.
Rust
27
star
64

encoding_rs_io

Streaming I/O adapters for the encoding_rs crate.
Rust
24
star
65

rust-cmail

A simple command line utility for periodically sending email containing the output of long-running commands.
Rust
21
star
66

cluster

A simple API for managing a network cluster with smart peer discovery.
Go
19
star
67

pager-multihead

A pager that supports per-monitor desktops (compatible with Openbox Multihead and Wingo)
Python
15
star
68

cablastp

Performs BLAST on compressed proteomic data.
Go
15
star
69

rust-error-handling-case-study

Code for the case study in my blog post: http://blog.burntsushi.net/rust-error-handling
Rust
15
star
70

rg-cratesio-typosquat

The source code of the 'rg' crate. It is an intentional typo-squat that redirects folks to 'ripgrep'.
Rust
15
star
71

imgv

An image viewer for Linux written in Go.
Go
14
star
72

cmd

A convenience library for executing commands in Go, including executing commands in parallel with a pool.
Go
14
star
73

cmail

cmail runs a command and sends the output to your email address at certain intervals.
Go
12
star
74

fanfoot

View your fantasy football leagues and get text alerts when one of your players scores.
Python
12
star
75

burntsushi-blog

A small Go application for my old blog.
CSS
12
star
76

gohead

An xrandr wrapper script to manage multi-monitor configurations. With hooks.
Go
12
star
77

intern

A simple package for interning strings, with a focus on efficiently representing dense pairwise data.
Go
11
star
78

crev-proofs

My crev reviews.
10
star
79

pytyle1

A lightweight X11 tool for simulating tiling in a stacking window manager.
Python
9
star
80

cif

A golang package for reading and writing data in the Crystallographic Information File (CIF) format. It mostly conforms to the CIF 1.1 specification.
Go
9
star
81

rucd

WIP
Rust
8
star
82

qcsv

An API to read and analyze CSV files by inferring types for each column of data.
Python
8
star
83

pyndow

A window manager written in Python
Python
8
star
84

csql

Package csql provides convenience functions for use with the types and functions defined in the standard library `database/sql` package.
Go
6
star
85

freetype-go

A fork of freetype-go with bounding box calculations.
Go
6
star
86

sqlsess

Simple database backed session management. Integrates with Gorilla's sessions package.
Go
6
star
87

go-wayland-simple-shm

C
5
star
88

sqlauth

A simple Golang package that provides database backed user authentication with bcrypt.
Vim Script
4
star
89

lcmweb

A Go web application for coding documents with the Linguistic Category Model.
JavaScript
4
star
90

bcbgo

Computational biology tools for the BCB group at Tufts University.
Go
4
star
91

fex

A framework for specifying and executing experiments.
Haskell
3
star
92

present

My presentations.
HTML
3
star
93

memchr-2.6-mov-regression

Rust
3
star
94

genecentric

A tool to generate between-pathway modules and perform GO enrichment on them.
Python
3
star
95

rust-docs

A silly repo for managing my Rust crate documentation.
Python
3
star
96

pcre2-mirror

A git mirror for PCRE2's SVN repository at svn://vcs.exim.org/pcre2/code
2
star
97

xpyb

A clone of xorg-xpyb.
C
2
star
98

burntsushi-homepage

A small PHP web application for my old homepage.
PHP
2
star
99

window-marker

Use vim-like marks on windows.
Python
2
star
100

sudoku

An attempt at a sudoku solver in Haskell.
Haskell
1
star