使用正则表达式将文本解析为对象

Question

I'm using an API which returns text in the following format:

#start
#p 12345 foo
#p 12346 bar
#end
#start
#p 12345 foo2
#p 12346 bar2
#end

My parsing function:

function parseApiResponse(data) {

    var results = [], match, obj;

    while (match = CST.REGEX.POST.exec(/(#start)|(#end)|#p\s+(\S+)\s+(\S+)/ig)) {

        if (match[1]) {           // #start
            obj = {};

        } else if (match[2]) {    // #end
            results.push(obj);
            obj = null;           // prevent accidental reuse 
                                  // if input is malformed

        } else {                  // #p something something
            obj[match[3]] = match[4];
        }
    }

    return results;
}

This will give me a list of objects which looks something like this:

[{ '12345': 'foo', '12346': 'bar'}, /* etc... */]

However, if a line is formatted like this

#start
#p 12345
#p 12346 bar
#end

The line would actually be #p 12345\\n and my match[4] would contain the next row's #p .

How do I adjust the pattern to adapt to this?

Answer 1

Assuming you have one #start , #end , or #p element per line, you can make your regex aware of this and add an additional non-capturing group to indicate that the last \\s+(\\S+) in a line is optional:

/(#start)|(#end)|#p\\s+(\\S+)(?:\\s+(\\S+))?$/igm

(?: ) is saying "treat this as a group, but don't capture the pattern it matches" (so it won't create an element in match ). The ? that follows that group means "this group is optional and may or may not match anything in the pattern". The $ right after that, in conjunction with the m flag, matches the end of the line.

You can also avoid the (?: ) trickery by using * instead of + quantifiers, meaning "match zero or more times": change \\s+(\\S+) to \\s*(\\S*) . This has the side effect that the space between the number and the data that follows it is now optional.

I would rewrite the regex and refactor the code a bit as follows:

while (match = CST.REGEX.POST.exec(/^#(start|end|p)(?:\s+(\d+)(?:[^\S\r\n]+([^\r\n]+))?)?$/igm)) {
  switch (match[1]) {
    case 'start':
      obj = {};
      break;
    case 'end':
      results.push(obj);
      obj = null;
      break;
    case 'p':
      obj[match[2]] = match[3];
      break;
  }
}

I like capturing start , end , or p in the one capture group so I can use it in a switch statement. The version of the regex I use here is a little more discriminating (expects the token that follows #p to be numeric) and a little more forgiving (allows the last token on a #p line to contain any non-linebreak whitespace, eg #p 1138 this is only a test ).