用String.split在两个字符序列之间不匹配正则表达式

Question

I'm using Scala to work with some very messy data that it is not practical to clean. It comes in the form of delimited key-value pairs, something like this: "a=1, b=2, c=3" . I am using String.split to break up the String into key-value pairs. Most of the string value parts of these pairs are quoted if the need to be, so this works to not match , inside of quotes: <string-instance>.split(", (?=(?:[^\\"]*\\"[^\\"]*\\")*[^\\"]*$)")

However, I have come across a url field that is neither quoted nor in all cases URL-encoded, so I have to deal with something like this:

"foo=bar, url=http://city.com/Boston, MA US, is_test=false"

In this case, I'm trying to match the comma-space after bar and the one after US and ignore the one after Boston . Fortunately, I can rely on these bad cases falling in between url= and , is_test= everywhere they occur (and that's about it). I've been banging my head on the Java regex tester here: https://www.freeformatter.com/java-regex-tester.html and failing. The closest I could get with the above input was this: (?<!url=[.]{0,300}^, is_test), (?!.*, is_test) , which only matched the comma-space after US , not the one after bar . The {0,300} is there to alleviate the problem of Java Regex not being able to handle potentially infinite look-behind expressions: java.util.regex.PatternSyntaxException: Look-behind group does not have an obvious maximum length

How can I solve this? Ideally, I could or the expression with the quoted comma-space ignoring one. One possibility too would be to match in between url= and , is_test and replace them with %20 . Unfortunately on that Regex expression, the closest I got was (?<=url=.{0,300})\\s(?!^\\w*, is_test) which matched the white-space right before is_test which I don't want to touch.

==edit==

My first example did not include a query string with a = which is a major part of my problem. Here is a more complete example of what I am dealing with:

foo="bar, harbor", url=http://city.com/start_city=Boston, MA US&end_city=New York, NY US, is_test=false

Answer 1

As your key value pair is separated by = and each of your pair is separated by a comma and some space, you can split on every comma which is just immediately before a = character using this regex,

,\s*(?=\w+=)

Online Demo

Check these Java codes which split your string at desired positions,

String[] data = "foo=\"bar, harbor\", url=http://city.com/start_city=Boston, MAUS&end_city=New York, NY US, is_test=false".split(",\\s*(?=\\w+=)");
Arrays.stream(data).forEach(System.out::println);

Prints,

foo="bar, harbor"
url=http://city.com/start_city=Boston, MAUS&end_city=New York, NY US
is_test=false

Let me know if this works for your cases and if not, please add the cases for which it doesn't work.