Java library for generating regular expression from input string

Question

There are many websites which take a string as user input and allow you to create a regular expression (regex) from pieces of the string.

But I could not find any java library which does the same. Is there any Java library available which generates a regular expression that exactly matches a string?

String inputString = "ABC345";
String regularExpression = Something.generateRegEx(inputString);

or something like that.

Note: I have a condition wherein I want to take some string from user, generate regular expression and then match for that pattern on some data-sets to extract similar patterns. I have created a small utility, but it is not that reliable yet. Moreover, I am looking for some well-tested library.

EDIT :

Please visit txt2re.com. I want a java library which performs the same function.

Do you just want to escape the input? See this question: http://stackoverflow.com/questions/60160/how-to-escape-text-for-regular-expression-in-java and the answer regarding `quote` method. — Dave L., Jul 30 '12 at 14:21
I used to write a library that do this. It always return `.*` — Kien Truong, Jul 30 '12 at 14:24
@ all - sorry for the inadequate description. I have just updated the question. — Saurabh, Jul 30 '12 at 14:31

score 3 · Answer 1 · answered Jul 30 '12 at 14:24

3

Pattern.quote(String) returns a (string) regex that matches the specified string exactly.

answered Jul 30 '12 at 14:24

Louis Wasserman

191,574
25
345
413

I have just used Pattern.quote(String) and the output was \QABC123\E – Saurabh Jul 30 '12 at 14:29
1

...Yes? What's the problem with that? – Louis Wasserman Jul 30 '12 at 14:31
@Saurabh: Is this a problem? What was your input? – Chris Dargis Jul 30 '12 at 14:32
@DougRamsey : sorry for inadequate information. I have just updated my question. – Saurabh Jul 30 '12 at 14:34
I can't figure out what that site is doing. Can you clarify the question yourself, rather than pointing to another source? – Louis Wasserman Jul 30 '12 at 14:36

score 2 · Accepted Answer · answered Jul 30 '12 at 15:06

I think, the txt2re.com has a database from known regular expressions, since the tool extends its answers with semantics like "date" or "email" for date and email formats. Otherwise, it gives an expression, which validates only a string but not a "regular language". Regular languages are expressed by regular expressions and they can be calculated by finite-state machines, but they are sets of limited words (all finite languages are regular). For example a simple language like:

L = { (a^n)(b^n) | n >= 0 } is not regular. (proof with pumping lemma)

L = {ab, aabb, aaabbb,...} (not- regular)

if you consider, that the input is a set of infinite words (inclusive natural languages), however, the regular expressions can not describe all of them. In order to generate regular expressions for a language, you had to first describe it with a (TYPE-3) grammar.

if your language has only a word like this:

L = { your.name@example.com }

then you can write a basic compiler iterating over the chars while checking their types, pseudo:

s = size(input) 
result = ""
for (i = 0; i < s; i++) {
   if input[i] is numeric
      result += "d"
   else if input[i] is word
      result += "w" 
   ...
}
return result

Thanks for your detailed answer. So, there's no such library already available. And to make such library, one need to have database of known regular expressions included in that library. rite? Thanks for your pseudo code, in fact my current running code (work-around) uses same logic for generating regular expression. — Saurabh, Jul 31 '12 at 04:42

score 0 · Answer 3 · answered May 10 '19 at 03:14

0

A genetic algorithm based java library like regex++ url: https://github.com/MaLeLabTs/RegexGenerator can be used for the same purpose.

answered May 10 '19 at 03:14

Aayush

220
1
11

kgautron · Answer 4 · 2012-07-30T14:41:01.307

-1

If what you want is to find a regex matching a given String, this does not make sense because there exists an infinite number of it.

On a contrary if you want to build a Pattern object from a regex that is input from the user, use the standard java API (java.util.regex.*) this way :

Pattern p = Pattern.compile(inputString);

edited Jul 30 '12 at 14:41

answered Jul 30 '12 at 14:20

kgautron

7,915
9
39
60

he wants to get regular expressions from a string given. – Erhan Bagdemir Jul 30 '12 at 14:21
@ErhanBagdemir yes, so what is the problem? – kgautron Jul 30 '12 at 14:25
1

Pattern.compile takes the regular expression as parameter, but it doesn't give the expression self for a given string. – Erhan Bagdemir Jul 30 '12 at 14:27
How could that be possible? For any String there is an infinite number of regex matching it. – kgautron Jul 30 '12 at 14:30
an answer like your latest comment could get +1. but the current answer -1. – Erhan Bagdemir Jul 30 '12 at 14:31
Why does the fact that there are an infinite number of regular expressions for a given String make it impossible for one to be generated? – Chris Dargis Jul 30 '12 at 14:48
I never said it was impossible, I said it made no sense. Because how do you know the generated regex will do what you want? For example ".*" would meet the criteria while not being particulary useful. – kgautron Jul 30 '12 at 14:54

Java library for generating regular expression from input string

4 Answers4