Using Regular Expressions to Extract a Value in Java

Question

I have several strings in the rough form:

[some text] [some number] [some more text]

I want to extract the text in [some number] using the Java Regex classes.

I know roughly what regular expression I want to use (though all suggestions are welcome). What I'm really interested in are the Java calls to take the regex string and use it on the source data to produce the value of [some number].

EDIT: I should add that I'm only interested in a single [some number] (basically, the first instance). The source strings are short and I'm not going to be looking for multiple occurrences of [some number].

Answer 1

From memory, the following should work:

Pattern p = Pattern.compile("^[a-zA-Z]+([0-9]+).*");
Matcher m = p.matcher("Testing123Testing");

if (m.find()) {
    System.out.println(m.group(1));
}

Answer 2

In Java 1.4 and up:

String input = "...";
Matcher matcher = Pattern.compile("[^0-9]+([0-9]+)[^0-9]+").matcher(input);
if (matcher.find()) {
    String someNumberStr = matcher.group(1);
    // if you need this to be an int:
    int someNumberInt = Integer.parseInt(someNumberStr);
}

Answer 3

Allain basically has the java code, so you can use that. However, his expression only matches if you numbers are only preceded by a stream of word characters.

"(\\d+)"

should be able to find the first string of digits. You don't need to specify what's before it, if you're sure that it's going to be the first string of digits. Likewise, there is no use to specify what's after it, unless you want that. If you just want the number, and are sure that it will be the first string of one or more digits then that's all you need.

If you expect it to be offset by spaces, it will make it even more distinct to specify `

"\\s+(\\d+)\\s+"

might be better.

If you need all three parts, this will do:

"([^\\d]+)(\\d+)(.*)"

EDIT The Expressions given by Allain and Jack suggest that you need to specify some subset of non-digits in order to capture digits. If you tell the regex engine you're looking for \d then it's going to ignore everything before the digits. If J or A's expression fits your pattern, then the whole match equals the input string. And there's no reason to specify it. It probably slows a clean match down, it it isn't totally ignored.

Answer 4

How about [^\\d]*([0-9]+[\\s]*[.,]{0,1}[\\s]*[0-9]*).* I think it would take care of numbers with fractional part. I included white spaces and included , as possible separator. I'm trying to get the numbers out of a string including floats and taking into account that the user might make a mistake and include white spaces while typing the number.