How to split a line with priority regex/pattern matching, Java

Question

So I know this question has probably been asked a bunch of times before, but I'm essentially trying to do the same thing as the JVM when it looks at run arguments on the command line, eg:

java MyProgram arg1 arg2 "argument three" arg4

The priority match is if the argument is in quotes, treat it as one argument; otherwise, separate them by spaces.

I'm reading through a CSV file, but sometimes one section is contained in quotes, so it might look something like this:

value, value, value, value, "value, value", value

Thus it adds one more element to the returned array from String.split().

The regex I'm trying to use:

String[] data = line.split("(\".*\")|,", -1);

So essentially I'm trying to say, if there's a double quote followed by anything, followed by another quote, treat that as priority (left - right); otherwise, split it based on the comma.

That regex doesn't seem to be working though, because I still get one more value on that line than there are fields (headers) in the file.

Any help would be appreciated, I'm not the best with regex. Thanks.

Answer 1

You can do the following (matches strings using delimiters as space and comma and ignores delimiters inside quotes..same problem different approach):

List<String> matchList = new ArrayList<String>();
Pattern regex = Pattern.compile("[^\\s,\\\"']+|\\\"([^\\\"]*)\\\"|'([^']*)'");
Matcher regexMatcher = regex.matcher(line);
while (regexMatcher.find()) {
    matchList.add(regexMatcher.group());
}

Edit: You can use [^\\\\s,\\\\\\"]+|\\\\\\"([^\\\\\\"]*)\\\\\\" for allowing only double quotes (as suggested by uraimo).

Output:

[value, value, value, value, "value, value", value]

Answer 2

You are looking for either:

• the start of the string or a comma (?:^|,) followed by zero-or-more whitespaces \\s* followed by a quote " then any number of non-quote characters ([^"]*) then another quote " then zero-or-more whitespace \\s* and either a trailing comma or the end-of-line (?=,|$) - which when combined gives (?:^|,)\\s*"([^"]*)"\\s*(?=,|$) or
• the start of the string or a comma (?:^|,) followed by zero-or-more non-comma characters ([^,]*) and either a trailing comma or the end-of-line (?=,|$) which when combined gives (?:^|,)([^,]*)(?=,|$)

Putting the two together you get the regular expression:

(?:^|,)(?:\s*"([^"]*)"\s*|([^,]*))(?=,|$)

And you can implement it like this:

String test = "value, value, value, value, \"value, value\", value";

Pattern pattern = Pattern.compile( "(?:^|,)(?:\\s*\"([^\"]*)\"\\s*|([^,]*))(?=,|$)" );
Matcher matcher = pattern.matcher( test );
while( matcher.find() ){
    String value = matcher.group(1);
    if ( value == null )
        value = matcher.group(2).trim();
    System.out.println( value );
}

If you want to expand it to allowing escaped quotes in the quoted string then you want:

(?:^|,)(?:\s*"((?:[^"]|\\")*)"\s*|([^,]*))(?=,|$)

Which can be written, in Java, as:

Pattern pattern = Pattern.compile( "(?:^|,)(?:\\s*\"((?:[^\"]|\\\\\")*)\"\\s*|([^,]*))(?=,|$)" );