-
Notifications
You must be signed in to change notification settings - Fork 2
/
make_regex2.pi
391 lines (325 loc) · 13.9 KB
/
make_regex2.pi
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
/*
A port of my Perl/Java/Python/Julia (Prefix) MakeRegex in Picat.
make_regex(Words)
generates a (prefix) regex for words. It use a simple approach which
combines common prefixes in generating the regexp.
Some examples:
* words = ["a", "al", "all", "alla", "an", "ann", "anna", "annas", "ananas"]
regex: a(l(la?)?|n(anas|n(as?)?)?)?
* words: ["and", "at", "do", "end", "for", "in", "is", "not", "of", "or", "use"]
regex: (a(nd|t)|do|end|for|i[ns]|not|o[fr]|use)
There is a simple way of handling character classes
* words: ["price1", "price2", "price3", "price4"]
regex: price[1234]
If there is no common prefix then it just put '|' between the words
* words: ["this", "is", "a", "very", "boring", "example", "with", "no", "common", "prefix"]
regex: (a|boring|common|example|is|no|prefix|this|very|with)
Also, see the (very old) page for my Perl package MakeRegex:
http://hakank.org/makeregex/index.html
The REAME file in that package states:
"""
The Perl package MakeRegex composes a regex-expression from a list of
words. It had been inspired by the emacs elisp module make-regex.el,
by Simon Marshall.
"""
See some other implementations:
- Perl: http://hakank.org/makeregex/index.html
- Julia: http://www.hakank.org/julia/make_regex.jl
And I should say that this is NOT a typical/paradigmatic Picat program,
it's a rather faithful port of the Julia program.
It should be completely re-written using some of the Picat specific
features...
Some experiments
----------------
Note that using just a few words might make the regexp larger than the concatenated words.
But, as we see in go3/0, when using many words, the compression is quite good.
* go2/0: Some random words
- words = [fee,lorinda,second,splenetic,callaghan,airline,scene,cater,resident,cambric,cowry,apothegm,sinew,osgood,nine,ott,ejector,aurochs,diluent,morristown,sulfur,turin,inseminate,cog,interpol,1st,move,gerundial,endoderm,pauline,construct,origin,cornstarch,abyss,java,styli,lithography,seminar,liniment,eidetic,impelled,revery,fickle,downright,treat,interference,hurdle,rampart,audible,crystallography,fib,tyburn,chemisorb,fateful,ogress,checkout,father,revelatory,augite,optometry]
regexp: (1st|a(byss|irline|pothegm|u(dible|gite|rochs))|c(a(llaghan|mbric|ter)|he(ckout|misorb)|o(g|nstruct|rnstarch|wry)|rystallography)|d(iluent|ownright)|e(idetic|jector|ndoderm)|f(at(eful|her)|ee|i(b|ckle))|gerundial|hurdle|i(mpelled|n(seminate|ter(ference|pol)))|java|l(i(niment|thography)|orinda)|mo(rristown|ve)|nine|o(gress|ptometry|rigin|sgood|tt)|pauline|r(ampart|e(sident|ve(latory|ry)))|s(cene|e(cond|minar)|inew|plenetic|tyli|ulfur)|t(reat|urin|yburn))
wordsLen = 414
regexLen = 457
- words = [piscataway,total,kruger,enzyme,lafayette,touchstone,bromide,naked,transfer,reddish]
regex (bromide|enzyme|kruger|lafayette|naked|piscataway|reddish|t(o(tal|uchstone)|ransfer))
wordsLen = 73
regexLen = 85
- words = [desuetude,dialectic,solitary,safety,array,angeles,plutonium,somebody'll,north,racetrack,abort,disjunct,collateral,foamy,alleviate,sadden,nashua,yokuts,canoe,cancerous]
- regex = (a(bort|lleviate|ngeles|rray)|c(an(cerous|oe)|ollateral)|d(esuetude|i(alectic|sjunct))|foamy|n(ashua|orth)|plutonium|racetrack|s(a(dden|fety)|o(litary|mebody'll))|yokuts)
wordsLen = 147
regexLen = 170
* go3/0: Generating all the words from some wordlists:
- /usr/share/dict/words (102305 words) takes 6.505s.
Compression: 868999 chars -> 439066 chars.
- Using a smaller wordlist (25104 words) takes 0.465s.
Comparession: 181299 chars -> 123515 chars
This Picat program was created by Hakan Kjellerstrand, hakank@gmail.com
See also my Picat page: http://www.hakank.org/picat/
*/
import util.
import regex.
main => go.
go ?=>
%
% Test cases of the form:
% [Words, Regexp]
%
Tests = [
[["all","alla"], "alla?"],
% A lot of Swedish words
[[ "all", "alla", "alle", "alls", "palle", "palla", "pelle", "perkele",
"ann", "anna", "annas", "anders", "håkan", "ångest", "ärlig",
"solsken", "sture", "stina", "hörapparat", "hörsel", "hårig"],
"(a(ll[aes]?|n(ders|n(as?)?))|h(å(kan|rig)|ör(apparat|sel))|p(all[ae]|e(lle|rkele))|s(olsken|t(ina|ure))|ärlig|ångest)"],
[["alla", "palla", "balla", "kalla", "all", "pall", "ball", "kall"],
"(alla?|balla?|kalla?|palla?)"
],
% "ananas" is the Swedish word for pineapple
[["a", "al", "all", "alla", "an", "ann", "anna", "annas", "ananas"],
"a(l(la?)?|n(anas|n(as?)?)?)?"],
[["a", "an", "ann", "anna", "annan", "annas", "annans", "ananas", "ananasens"],
"a(n(anas(ens)?|n(a(ns?|s)?)?)?)?"],
[["a", "ab", "abc", "abcd", "abcde", "abcdef", "b", "bc", "bcd", "bcde", "bcdef",
"bcdefg", "abb", "abbc", "abbcc", "abbccdd"],
"(a(b(b(c(c(dd)?)?)?|c(d(ef?)?)?)?)?|b(c(d(e(fg?)?)?)?)?)"
],
[["this", "is", "a", "very", "boring", "example", "with", "no", "common", "prefix"],
"(a|boring|common|example|is|no|prefix|this|very|with)"
],
[["price1","price2","price3","price4"],
"price[1234]"],
% This is from Marshall's make-regex.el
[["and", "at", "do", "end", "for", "in", "is", "not", "of", "or", "use"],
"(a(nd|t)|do|end|for|i[ns]|not|o[fr]|use)"
],
% This is from Marshall's make-regex.el
[["cond", "if", "while", "let*?", "prog1", "prog2", "progn",
"catch", "throw", "save-restriction", "save-excursion",
"save-window-excursion", "save-match-data", "unwind-protect",
"condition-case", "track-mouse"],
"(c(atch|ond(ition-case)?)|if|let*?|prog[12n]|save-(excursion|match-data|restriction|window-excursion)|t(hrow|rack-mouse)|unwind-protect|while)"
],
% This is from Marshall's make-regex.el
[["abort", "abs", "accept", "access", "array",
"begin", "body", "case", "constant", "declare",
"delay", "delta", "digits", "else", "elsif", "entry",
"exception", "exit", "function", "generic", "goto",
"if", "others", "limited", "loop", "mod", "new",
"null", "out", "subtype", "package", "pragma",
"private", "procedure", "raise", "range", "record",
"rem", "renames", "return", "reverse", "select",
"separate", "task", "terminate", "then", "type",
"when", "while", "with", "xor"],
"(a(b(ort|s)|cce(pt|ss)|rray)|b(egin|ody)|c(ase|onstant)|d(e(clare|l(ay|ta))|igits)|e(ls(e|if)|ntry|x(ception|it))|function|g(eneric|oto)|if|l(imited|oop)|mod|n(ew|ull)|o(thers|ut)|p(ackage|r(agma|ivate|ocedure))|r(a(ise|nge)|e(cord|m|names|turn|verse))|s(e(lect|parate)|ubtype)|t(ask|erminate|hen|ype)|w(h(en|ile)|ith)|xor)"
],
% This is from https://stackoverflow.com/questions/5365283/regular-expression-to-search-for-gadaffi
% Different spelling of Kadaffi
[["Gadaffi","Gadafi","Gadafy","Gaddafi","Gaddafy","Gaddhafi","Gadhafi","Gathafi","Ghadaffi","Ghadafi","Ghaddafi",
"Ghaddafy","Gheddafi","Kadaffi","Kadafi","Kaddafi","Kadhafi","Kazzafi","Khadaffy","Khadafy","Khaddafi","Qadafi",
"Qaddafi","Qadhafi","Qadhdhafi","Qadthafi","Qathafi","Quathafi","Qudhafi","Kadafi]"],"(G(a(d(af(fi|i|y)|d(af[iy]|hafi)|hafi)|thafi)|h(ad(af(fi|i)|daf[iy])|eddafi))|K(a(d(af(fi|i]?)|dafi|hafi)|zzafi)|had(af(fy|y)|dafi))|Q(a(d(afi|dafi|h(afi|dhafi)|thafi)|thafi)|u(athafi|dhafi)))"],
% See mankell_v3.pi, regex_generating_strings_v3.pi.
% The "best" regex is
% k(je|ä)ll(er|ar)?(st|b)r?an?d"
% but it's not found by make_regex/1.
[["källbad","kjellbad","källstad","källband","källbrad","kjellstad","kjellband","kjellbrad","källstand","källstrad","källbrand","källerbad","källarbad","kjellstand","kjellstrad","kjellbrand","kjellerbad","kjellarbad","källstrand","källerstad","källerband","källerbrad","källarstad","källarband","källarbrad","kjellstrand","kjellerstad","kjellerband","kjellerbrad","kjellarstad","kjellarband","kjellarbrad","källerstand","källerstrad","källerbrand","källarstand","källarstrad","källarbrand","kjellerstand","kjellerstrad","kjellerbrand","kjellarstand","kjellarstrad","kjellarbrand","källerstrand","källarstrand","kjellerstrand","kjellarstrand"],xxx]
],
% Just testing first case
% Tests := [
% ["all","alla"]
% % alla?
% % ["alla", "palla", "balla", "kalla", "all", "pall", "ball", "kall"]
% % ["a", "al", "all", "alla", "an", "ann", "anna", "annas", "ananas"]
% ],
FailedTests = [],
foreach([T,Check] in Tests)
println(words=T),
Regexp = make_regex(T),
println(regexp=Regexp),
if Regexp == Check then
println(ok)
else
println(not_ok),
FailedTests := FailedTests ++ [T],
end,
RegexFailed = [],
regex_compile(Regex),
foreach(Word in T)
if regex_match(Word) then
true % println(Word=ok)
else
println(Word=not_ok),
RegexFailed := RegexFailed ++ [Word]
end
end,
println(regexFailed=RegexFailed),
nl
end,
println(failed_tests=FailedTests),
nl.
go => true.
%
% Test some random words
%
go2 ?=>
garbage_collect(300_000_000),
_ = random2(),
NumWords = 130,
AllWords = read_file_lines("unixdict.txt"),
% AllWords = read_file_lines("/usr/share/dict/words"),
AllWordsLen = AllWords.len,
println(allWordsLen=AllWordsLen),
% Pick about NumWords words
Words = [AllWords[R] : _ in 1..NumWords, R = random(1,AllWordsLen)].remove_dups,
println(words=Words),
WordsString = join(Words,''),
Regex = make_regex(Words),
println(regex=Regex),
println(wordsLen=WordsString.len),
println(regexLen=Regex.len),
regex_compile(Regex),
Failed = [],
foreach(Word in Words)
if regex_match(Word) then
true % println(Word=ok)
else
Failed := Failed ++ [Word],
println(Word=not_ok)
end
end,
nl,
println(failed=Failed),
nl.
go2 => true.
%
% Testing all the words in the wordlists.
%
go3 ?=>
garbage_collect(300_000_000),
% Words = read_file_lines("unixdict.txt"),
% Words = read_file_lines("/usr/share/dict/words"),
% Words = read_file_lines("words_lower.txt"),
Words = read_file_lines("wordle_small.txt"),
% Pick words of a certain length
% Words := [W.to_lowercase : W in Words, W.len == 3,not membchk('\'',W),not membchk(' ',W)],
% print(words=Words),
WordsString = join(Words,''),
WordsLen = Words.len,
println(wordsLen=WordsLen),
Regex = make_regex(Words),
println(Regex),
println(wordsLen=WordsString.len),
println(regexLen=Regex.len),
% Check against pcre2 regex
regex_compile(Regex),
Failed = [],
foreach(Word in Words)
if regex_match(Word) then
println(Word=ok)
else
Failed := Failed ++ [Word],
println(Word=not_ok)
end
end,
nl,
println(failed=Failed),
nl.
go3 => true.
make_regex(Words) = common_prefix("", sort(Words)).
% println($make_regex(Words)).
% replace.(words,r"([*?+])"=>s"\\\1") % replace meta characters
% Words := replace_string(Words,"(","\\("),
% Words := replace_string(Words,"[","\\["),
% Words := replace_string(Words,"*","\\*"),
% Words := replace_string(Words,"+","\\+"),
% Words := replace_string(Words,"]","\\]"),
% Words := replace_string(Words,")","\\)"),
% We sort the words to induce more common prefixes
% println(words=Words).
%
% common_prefix(p, list)
%
% Here is where the main work is done. It's somewhat magically ported from my
% Perl/Java/Python versions...
%
common_prefix(P, List) = Ret =>
ListLen = List.len,
Ret1 = _,
if ListLen == 0 then
Ret1 := ""
end,
if var(Ret1), ListLen == 1 then
Ret1 := ([P] ++ join(List,'')).flatten
end,
%
% fix for some - as of now - unknown bug. To fix in later version!
%
if var(Ret1), P == "", List[1] == "", List[2] == "" then
Ret1 := ""
end,
if var(Ret1) then
% Collect all the strings with the same prefix-char
Hash = new_map(),
foreach(Word in List.sort())
Prefix = "",
SuffixedWord = "",
if Word.len > 0 then
Prefix := Word[1],
SuffixedWord := [Word[I] : I in 2..Word.len] % Word[2..Word.len],
end,
% Put the suffix in the list of other suffixes for
% this prefix
Hash.put(Prefix, Hash.get(Prefix,"") ++ [SuffixedWord])
end,
% And recurse this list
All = [],
foreach(Key in Hash.keys().sort())
Comm = "",
Values = Hash.get(Key),
if Key.len > 0, Values != [] then
Values := sort(Values),
Comm := common_prefix(Key, Values)
end,
% hack to be able to use the '?' char . Should be re-written!
if Comm == "" then
Comm := " "
end,
All := All ++ [Comm]
end,
All := sort(All),
% paren: what to put in parenthesis ('()' or '[]') if anything
Paren = "",
AllLen = All.len,
if AllLen == 1 then
Paren := join(All,'')
else
Len := max([A.len : A in All]),
JoinChar = cond(Len != 1,"|",""),
% joins all entries except for " "
JoinStr = "",
Mark = "",
Count = 0,
foreach(W in All, W != ' ')
GotHackMark = cond(W == " ",true, false), % This is a hack for handling '?'
if W.len > 0, W != " ", W != ' ' then
JoinStr := JoinStr ++ W,
if Count < AllLen-1 then
JoinStr := JoinStr ++ JoinChar
end
end,
if GotHackMark then
Mark := "?"
end,
Count := Count + 1
end,
Paren = "",
if JoinStr.len == 1 then
Paren := JoinStr ++ [Mark]
else
if Len == 1 then
Paren := "[" ++ JoinStr ++ "]" ++ Mark
else
Paren := "(" ++ JoinStr ++ ")" ++ Mark
end
end
end,
Ret1 := ([P] ++ Paren).flatten
end,
Ret = Ret1.flatten.