Parser
Jodd JSON parser reads JSON string and converts it into objects (i.e. object graph). Mapping JSON data to Java objects may be tricky. Jodd JSON uses class and property types to map string JSON values into specific Java type. And as a bonus, it is super fast.
The parser class is JsonParser. When no target type is provided, parser will convert JSON string to Java Map and List. Like this:
JsonParser jsonParser = new JsonParser();Map map = jsonParser.parse("{ \"one\" : { \"two\" : 285 }, \"three\" : true}");
or as a one-liner:
Map map = JsonParser.create().parse("{ ... }");
JSON string here is converted into a Map that contains 2 keys. One of the keys (one) holds another Map instance. As you can see from the example, simple JSON data types are converted into their counter-part Java types. Boolean value is converted to Boolean, number is converted to a Number or BigInteger etc. JsonParser finds the best possible type; for example, if number can be stored into an int then Integer is returned, but if it is a decimal value then Double is returned and so on.
Ok, let's now see how to map JSON to types.
Let's start with the JSON data:
{"name" : "Mak","bars" : {"123": {"amount" : 12300},"456": {"amount" : 45600}},"inters" : {"letterJ" : {"sign" : "J"},"letterO" : {"sign" : "O"},"letterD" : {"sign" : "D"}}}
This data can be mapped to Java type like:
public class User {private String name;private Map<String, Bar> bars;private Map<String, Inter> inters;// ...}
Property name is a simple property. But bars is a Map of Bars:
public class Bar {private Integer amount;// ...}
This is no problem for JsonParser. It will parse inner JSON maps to Bar types, since the generic type of the property specifies the map's content type. In the same way JsonParser can handle the inters map.
JsonParser jsonParser = new JsonParser();User user = jsonParser.parse(json, User.class);
JsonParser now converts JSON directly to a target type. Let's remember:
JsonParser lookups for type information from the target's property: the property type and generics informations. {: .attn}
Using this approach, JsonParser can parse complex JSON strings into Java object graph, as long it can resolve the type information.
To introduce some more complexity, let's say that Inter is an interface:
public interface Inter {public char getSign();}
and one of it's implementation is:
public class InterImpl implements Inter {protected char sign;​public char getSign() {return sign;}public void setSign(char sign) {this.sign = sign;}}
Now, this is something JsonParser can't figure out by looking at the inters property types and it's generic information. We need to explicitly specify the target type for maps values. As you could guess, we can use path to specify the mapping. But in this case, the path needs to address the values of the map! No problem - by using special path name value we can address all the values of a map:
User user = new JsonParser().map("inters.values", InterImpl.class).parse(json, User.class);
Nice! Similar with the value, we could specify the key's type, using the path reference keys. Look at the following example:
String json = "{\"eee\" : {\"123\" : \"name\"}}";​Map<String, Map<Long, String>> map =new JsonParser().map("values.keys", Long.class).parse(json);
We changed the key type of the inner map. This is one more thing to remember:
Use map() method to map target type in the result object graph, specified by it's path. {: .attn}
Now we have a powerful tool that can parse about any JSON to any Java type. Here is another example:
{"1": {"first": {"areaCode": "404"},"second": {"name": "Jodd"}}}
May be parsed like this:
Map<String, Pair<Phone, Network>> complex = new JsonParser().map("values", Pair.class).map("values.first", Phone.class).map("values.second", Network.class).parse(json);
Each value of returned map is going to be a Pair of two different types.
As seen, path can contain special names like values or keys to reference all values of a map or all keys of a map (or of an array). But you can not change the type of particular map value, since these special paths address all items.
But there is a solution for this. By enabling the alternative paths with .useAltPaths() we are telling JsonParser to match paths to current map values! By default this option is disabled, for performance reasons (there is some penalty because more paths are matched).
With alt paths enabled, you can reference any value in the map, too.
Sometimes JSON string does contain an information about the target type, stored in 'special' key like class or __class. If you have such JSON or if you have used this option with JsonSerializer, you can enable this feature with JsonParser as well:
Target target =new JsonParser().setClassMetadataName("class").parse(json);
Now every JSON map will be scanned for this special key class that holds the full class name of the target. But be careful:
Using class metadata name with JsonParser has some performance penalty and may introduce potential security risk. {: .attn}
Every JSON map first must be converted to a Map so we can fetch the class name and then converted to a target class. Because of this double conversion expect performance penalties if using class metadata name.
There are security risks using class meta-data. You may expose a security hole in case untrusted source manages to specify a class that is accessible through class loader and exposes set of methods and/or fields, access of which opens an actual security hole. Such classes are known as deserialization gadgets.
Because of this, use of "default typing" is not encouraged in general, and in particular is recommended against if the source of content is not trusted. Conversely, default typing may be used for processing content in cases where both ends (sender and receiver) are controlled by same entity.
For additional security control there is method to whitelist allowed class names wildcards: allowClass(). If class name does not match one of set wildcard patterns, JsonParser will throw an exception. To reset the whitelist, use allowAllClasses().
As for everything in Jodd, JoddParser uses powerful TypeConverter to convert between strings to real types.
Sometimes data comes in different flavors. For example, Date may be specified as a string in yyyy/MM/dd format and not as a number of milliseconds. So we need to explicitly convert the string into Date. For that we can use ValueConverter:
final SimpleDateFormat dateFormat = new SimpleDateFormat("yyyy/MM/dd");​Person person = new JsonParser().withValueConverter("birthdate", new ValueConverter<String, Date>() {public Date convert(String data) {try {return dateFormat.parse(data);} catch (ParseException pe) {throw new JsonException(pe);}}}).parse(json, Person.class);
Json parser consumes only valid JSON strings. If JSON string is not valid, an exception is thrown with detailed message of why parsing failed.
But real-world often does not play by the rules ;) Therefore, JsonParser may run in so-called loose mode, when it can process more:
JsonParser jsonParser = new JsonParser().looseMode(true);
Here is what loose mode may handle:
both single and double quotes are accepted.
invalid escape characters are just added to the ouput.
strings may be unquoted, too.
For example, JsonParser in loose mode may parse the following input:
{key : value,'my' : 'hol\\x'}
This JSON is not valid, but it can be parsed, too.
Finally, the performance gem. Common scenario is parsing a (big) JSON document only to access few keys. For this situations we actually don't need to parse the complete JSON, only the elements that we need.
Jodd JSON has the lazy mode that provides exactly that feature. It returns the Map or the List i.e. in the lazy mode there is no sense to parse to concrete types. Returned object is lazy, since the parsing happens only on key retrieval.
This may boost performance a lot (in explained scenario)! The downside of this approach is that JSON string is kept in memory while the returned object exist. Please use lazy mode with care.
JsonParser jsonParser = new JsonParser().lazy(true);
or as a one-liner:
JsonParser jsonParser = JsonParser.createLazyOne();
