从java中的json字符串中提取值

Question

I have a string in json format and I want to extract certain values from that json. For example:

{"foo":"this is foo", "bar":{"nested_bar": "this is nested bar"} }

A user might want to print either foo or bar or both.

Right now, I have a simple class which can read only flat json.

How do I modify the following code to incorporate nested json?

Another question is what is a good way to represent the tags which I want to extract as in flat json? I was passing an array.

public class JsonParser {

public static String[] tagsToExtract;



public JsonParser(String[] tags){
    tagsToExtract = tags;
}


public HashMap<String, Object> extractInformation(Text line) throws JSONException{
    HashMap<String, Object> outputHash = new HashMap<String, Object>();
    JSONObject jsn = new JSONObject(line.toString());
    for (int i =0; i < tagsToExtract.length; i++){
        outputHash.put(tagsToExtract[i],jsn.get(tagsToExtract[i].toString()));
    }
    return outputHash;

}

}

Answer 1

There are quite a few JSON libraries for Java that will do exactly what you want. A couple of the more highly regarded ones are:

• Gson: http://code.google.com/p/google-gson/
• JSONLib: http://json-lib.sourceforge.net/

And you can find a more in-depth discussion of the various options here: https://stackoverflow.com/questions/338586/a-better-java-json-library

If you are really interested in writing your own parser for it, though, the hint I will give is to take advantage of recursion. Suppose you have a JSON object something like this:

{
    prop1: (some value),
    prop2: (some value),
    ...
}

Notice that when you start parsing the top-level object, you're doing exactly the same thing as you will be doing when you parse each value - because the values themselves can be just another JSON object. So a simple way to get started would be to write a parser which just gets the keys and their associated values as strings, without processing the values. Then, call the same algorithm again on each value - and so on, until you get to a value that is just a plain value - a string, a number, etc.