Regular Expression for reading CSV in Java

Question

Possible Duplicate:
Parsing CSV input with a RegEx in java

I have an input file in which each line has input values of string of the following form:

   "  ab  cd  " ,    "  efgh,ijk.",  4,"lmno"

i.e.,

The words are either in quotes or they have no quotes.
The space before and after the start and end word respectively is not allowed.

EDIT: 3. It can have inputs just separated by commas.(abc,"Hi Mary,Joe",5)

Using .Split() in java, I need a regular expression to output this:

ab  cd
efgh,ijk.
4
lmno

I tried this:

[^",]*[\",]

But this doesn't work on "efgh,ijk."

Here is a link for regex testing: http://regexpal.com/ I need some assistance on this. Please help. Thank you

That's the [CSV format](http://tools.ietf.org/html/rfc4180). There are [plenty of](https://www.google.com/search?q=java+csv+parser) parsers for this format. You could even easily [write one](http://stackoverflow.com/a/2241950) yourself. Don't try to use regex for non-regular patterns, it'll only hurt you. — BalusC, Nov 06 '12 at 21:17
How is `5, ebo"eu ooeu" euoe, oeuou` supposed to be handled? — durron597, Nov 06 '12 at 21:17
@durron597 - That violates OP's condition for what the input looks like. Values should either have no quotes or be _surrounded_ by quotes. — Ted Hopp, Nov 06 '12 at 21:23
@TedHopp his program (if it's going to be in production) should handle illegal input. The answer to my question could be "throw an exception" — durron597, Nov 06 '12 at 21:24
@BalusC - looks like a regular expression to me. How not so? — Ed Staub, Nov 06 '12 at 21:49

Javier Diaz · Accepted Answer · 2012-11-07T10:38:56.310

2

DEMO

Regex pattern: (?:\s*(?:\"([^\"]*)\"|([^,]+))\s*,?)+?

Update for null values: (?:\s*(?:\"([^\"]*)\"|([^,]+))\s*,?|(?<=,)(),?)+? DEMO

An example of it working, I know it's kinda CSV Format but as long as you dont write really really weird things it'll match all of them.

Matcher ma = Pattern.compile("(?:\\s*(?:\\\"([^\\\"]*)\\\"|([^,]+))\\s*,?)+?").matcher("   \"  ab  cd  \" ,    \"  efgh,ijk.\",  4,\"lmno\"");
while (ma.find()) {
    if (ma.group(1) == null) {
        System.out.println(ma.group(2));
    } else {
        System.out.println(ma.group(1));
    }
}

Edit, btw if you wanted us to give the code for you, don't tell us about a regex online tester, if you do so it's because you know how to handle regex, if you have no idea of how to do that, ask it too.

edited Nov 07 '12 at 10:38

answered Nov 06 '12 at 21:25

Javier Diaz

1,791
1
17
25

This takes the quotes as well as the comma after it. – Crocode Nov 06 '12 at 22:06
But it is printing only spaces and comma when i execute the program – Crocode Nov 06 '12 at 22:13
Because you are replacing it with null or \n, what I've given you is a backreference(s) to what you need, you just need to loop through it and get them. – Javier Diaz Nov 06 '12 at 22:20
Can you please tell me how to do that? – Crocode Nov 06 '12 at 23:04
1

There you got your code. – Javier Diaz Nov 06 '12 at 23:11
Btw (I can't edit) if you downvote so fast, you won't be able to upvote again, if the answer is not useful for you, I'll myself (or whoever) delete the post, but remember that. – Javier Diaz Nov 06 '12 at 23:31
I did not do anything on purpose. I tried to select your answer as final one, but couldn't do it. – Crocode Nov 07 '12 at 02:06
Could you please edit your expression so that it identifies even this input: "abc",,"lmno",, (4 inputs here) so it takes it as 4 values. – Crocode Nov 07 '12 at 02:08

Ωmega · Answer 2 · 2012-11-06T22:57:07.247

2

I suggest to find matches and then trim them to get final results.

Matcher m = Pattern.compile("\\s*(?:\"[^\"]*\"|(?:^|(?<=,))[^,]*)").matcher(s);
while (m.find()) {
  System.out.println(m.group().replaceAll("^\\s*\"?\\s*(.*?)\\s*\"?\\s*$", "$1"));
}

See this demo.

edited Nov 06 '12 at 22:57

answered Nov 06 '12 at 22:28

Ωmega

42,614
34
134
203

barfuin · Answer 3 · 2012-11-06T21:54:23.460

Try calling split() with (?:^\s*"\s*|\s*"\s*$|\s*"?\s*,\s*"?\s*) (demo).

This will match the comma contained in the one String as well, which is wrong in your case. But it's the only way if you are going to use split(). You could introduce some way of escaping the contained comma (like \,), which could easily be added to the regex.

Else you will have to use some other means of disscting the String, and split() won't help you.

score 0 · Answer 4 · answered Nov 06 '12 at 21:38

In case you don't want to use a regex, as a regex implies it is a 'regular' expression. "i think there is a pattern here" does not a regular expression make. They're good, they're fast, and I only use them when I completely control the input being fed into the regex.

//no development environment up, there may be compilation errors.
private static String[] csv(final String input){
  String[] inputArray = input.split(",");
  for(int i =0;i < inputArray.length;i++){
//org.apache.commons.lang.StringUtils
    String value = StringUtils.removeEnd(inputArray[i],"\"");
    value = StringUtils.removeStart(value,"\"");
    value = StringUtils.trim(value);

    inputArray[i] = value;
  }
  return inputArray;
}

You should trim the String twice: before and after removing the *"* at end and beginning of the String. — Luiggi Mendoza, Nov 06 '12 at 21:42
If the string is "a,b,c","def" it would split into a b c def. But I want it to be a,b,c and def — Crocode, Nov 06 '12 at 21:59

Regular Expression for reading CSV in Java

4 Answers4