Using regex to 'clean up' the string

Question

I need a clear text with only words, excluding all digits, extra spaces, dashes, commas, dots, brackets, etc. It is used for a word generation algorithm (taken from gamasutra). I suppose that regular expression can help here. How can I do this with help of String.split?

UPD:

Input: I have 1337 such a string with different stuff in it: commas, many spaces, digits - 2 3 4, dashes. How can I remove all stuff?

Output: I have such a string with different stuff in it commas many spaces digits dashes How can I remove all stuff

possible duplicate of [Splitting strings through regular expressions by punctuation and whitespace etc in java](http://stackoverflow.com/questions/7384791/splitting-strings-through-regular-expressions-by-punctuation-and-whitespace-etc) — xlecoustillier, Jun 12 '13 at 08:36
Please add an example with input text and expected output text. — pepuch, Jun 12 '13 at 08:38

assylias · Accepted Answer · 2013-06-12T09:10:06.187

4

In two steps you could do:

String s = "asd asd   asd.asd, asd";
String clean = s.replaceAll("[\\d[^\\w\\s]]+", " ").replaceAll("(\\s{2,})", " ");
System.out.println(clean);

The first step removes all characters that are not a letter or a space and replaces them with a space. The second step removes multiple spaces by only one space.

Output:

asd asd asd asd asd

If all you need is an array containing the words, then this would be enough:

String[] words = s.trim().split("[\\W\\d]+");

edited Jun 12 '13 at 09:10

answered Jun 12 '13 at 08:40

assylias

321,522
82
660
783

1

`\w` is also containing digits, the OP wants to remove them also. – stema Jun 12 '13 at 08:52
@stema I have amended. Your approach is better anyway (+1). – assylias Jun 12 '13 at 09:12

score 3 · Answer 2 · answered Jun 12 '13 at 08:49

3

If you care about Unicode (you should), then use Unicode properties.

String[] result = s.split("\\P{L}+");

\p{L} is the Unicode property for a letter in any language.

\P{L} is the negation of \p{L}, means it will match everything that is not a letter. (I understood that is what you want.)

answered Jun 12 '13 at 08:49

stema

90,351
20
107
135

This perfectly fits in our scenario where we required an accurate length of strings coming from WordPress (via GraphQL). They were showing a different length (`string.length`, usually +1 than the real length) due to the presence of non-Unicode characters and this helped to purge those. – KeshavDulal Dec 15 '21 at 06:31

Evgeniy Dorofeev · Answer 3 · 2013-06-12T09:04:16.240

1

I would do it this way

    str = str.replaceAll("\\s+", " ");
    str = str.replaceAll("\\p{Punct}|\\d", "");
    String[] words = str.split(" ");

edited Jun 12 '13 at 09:04

answered Jun 12 '13 at 08:40

Evgeniy Dorofeev

133,369
30
199
275

This will not remove the digits from the text. – stema Jun 12 '13 at 09:02

Using regex to 'clean up' the string

3 Answers3