Merge branch 'master' of https://github.com/noprompt/frak
[frak.git] / README.md
1 # frak
2
3 frak transforms collections of strings into regular expressions for
4 matching those strings. The primary goal of this library is to
5 generate regular expressions from a known set of inputs which avoid
6 backtracking as much as possible.
7
8 ## "Installation"
9
10 Add frak as a dependency to your `project.clj` file.
11
12 ```clojure
13 [frak "0.1.1"]
14 ```
15
16 ## Usage
17
18 ```clojure
19 user> (require 'frak)
20 nil
21 user> (frak/pattern ["foo" "bar" "baz" "quux"])
22 #"(?:ba(?:r|z)|foo|quux)"
23 user> (frak/pattern ["Clojure" "Clojars" "ClojureScript"])
24 #"Cloj(?:ure(?:Script)?|ars)"
25 ```
26
27 ## How?
28
29 A frak pattern is constructed from a trie of characters. As characters
30 are added to it, meta data is stored in it's branches containing
31 information such as which branches are terminal and a record of
32 characters which have "visited" the branch.
33
34 During the rendering process frak will prefer branch characters that
35 have "visited" the most. In the example above, you will notice the
36 `ba(?:r|z)` branch takes precedence over `foo` even though `"foo"` was
37 the first to enter the trie. This is because the character `\b` has
38 frequented the branch more than `\f` and `\q`. The example below
39 illustrates this behavior on the second character of each input.
40
41 ```clojure
42 user> (frak/pattern ["bit" "bat" "ban" "bot" "bar" "box"])
43 #"b(?:a(?:t|n|r)|o(?:t|x)|it)"
44 ```
45
46 ## Why?
47
48 [Here's](https://github.com/guns/vim-clojure-static/blob/249328ee659190babe2b14cd119f972b21b80538/syntax/clojure.vim#L91-L92)
49 why. Also because.
50
51 ## Next
52
53 While the patterns currently generated by frak are correct, there is
54 potential for improvement; the word trie could be converted to a
55 directed acyclic word graph (DAWG).
56
57 By using a DAWG it might be possible to produce more efficient
58 patterns which consider common prefixes and suffixes. This could
59 reduce backtracking and overall pattern size.
60
61 ## And now for something completely different
62
63 Let's build a regular expression for matching any word in
64 `/usr/share/dict/words`.
65
66 ```clojure
67 user> (require '[clojure.java.io :as io])
68 nil
69 user> (def words
70 (-> (io/file "/usr/share/dict/words")
71 io/reader
72 line-seq))
73 #'user/words
74 user> (def word-re (frak/pattern words))
75 #'user/word-re
76 user> (every? #(re-matches word-re %) words)
77 true
78 ```
79
80 You can view the full expression
81 [here](https://gist.github.com/noprompt/6106573/raw/fcb683834bb2e171618ca91bf0b234014b5b957d/word-re.clj)
82 (it's approximately `1.5M`!).
83
84 ## Benchmarks
85
86 ```clojure
87 (use 'criterium.core)
88
89 (def words
90 (-> (io/file "/usr/share/dict/words")
91 io/reader
92 line-seq))
93
94 (defn naive-pattern
95 "Create a naive regular expression pattern for matching every string
96 in strs."
97 [strs]
98 (->> strs
99 (clojure.string/join "|")
100 (format "(?:%s)")
101 re-pattern))
102
103 ;; Shuffle 10000 words and build a naive and frak pattern from them.
104 (def ws (shuffle (take 10000 words)))
105 (def n-pat (naive-pattern ws))
106 (def f-pat (frak/pattern ws))
107
108 ;; Verify the naive pattern matches everything it was constructed from.
109 (every? #(re-matches n-pat %) ws)
110 ;; => true
111
112 ;; Shuffle the words again since the naive pattern is built in the
113 ;; same order as it's inputs.
114 (def ws' (shuffle ws))
115
116 ;;;; Benchmarks
117
118 ;; Naive pattern
119
120 (bench (doseq [w ws'] (re-matches n-pat w)))
121 ;; Execution time mean : 1.499489 sec
122 ;; Execution time std-deviation : 181.365166 ms
123 ;; Execution time lower quantile : 1.337817 sec ( 2.5%)
124 ;; Execution time upper quantile : 1.828733 sec (97.5%)
125
126 ;; frak pattern
127
128 (bench (doseq [w ws'] (re-matches f-pat w)))
129 ;; Execution time mean : 155.515855 ms
130 ;; Execution time std-deviation : 5.663346 ms
131 ;; Execution time lower quantile : 148.168855 ms ( 2.5%)
132 ;; Execution time upper quantile : 164.164294 ms (97.5%)
133 ```