Serialize Class to JSON with correct sObject API field names

Question

I want to use the standard Force.com REST API to create a case remotely from another Salesforce instance in my instance. I want to populate two fields on creation: "Subject" and "txt_UserId__c". Because the source org doesn't have the case object, I wrote a custom class that is populated from inputfield values on a visualforce page:

public class CaseMap {
public String Subject {get; set;}
public Id txt_UserId__c {get; set;}
}

Unforunately the property "txt_UserId__c" is not allowed. This means I have to rename it to something else:

public class CaseMap {
public String Subject {get; set;}
public Id UserId {get; set;}
}

But this means, when serializing the class into JSON, one of the field names (i.e., 'UserId'), does not match the API field name that is expected on the target org.

What I did so far is a simple JSON string find and replace operation. But I am sure there must be a more elegant and less fault-prone way to change field names, for example include a mapping table in the serialization or so.

Any ideas?

Thanks!

Answer 1

You could using the stream JSON api to build your json, or you could build a Map that represents your record, and use regular JSON.serialize to build the json, that's probably easier than the streaming approach, and will be considerably more robust than trying to regex the generated json. e.g.

Map<String, String> r = new Map<String, String>();
r.put('name', 'bob');
r.put('account__c', '0010100000000123');
System.debug(JSON.serialize(r));

generates

09:30:33:071 USER_DEBUG [5]|DEBUG|{"name":"bob","account__c":"0010100000000123"}

Answer 2

I ran into a similar issue not too long ago. There is no annotation you can use to serialize/deserialize reserved keywords in apex. Your options are to either do the string replace or build everything manually using the JSON writer.

I would definitely recommend using a string replace. Something like the following should work for all custom fields:

public class CaseMap {
public String Subject {get; set;}
public Id txt_UserId_c {get; set;}
}

//pass in json body
private String filterRequest(String value) {
    //regex not thoroughly tested
    return value.replaceall('_c"\\s*:', '__c" :');
}

Answer 3

In case of suspicious JSON structure, I usually prefer using "JSON.deserializeUntyped". It will safeguard from such issues, and bit safer then string token replacements in JSON.

Pros

Easy and less risky conversion, end result is a simple Map instance, which could be parsed and consumed with ease.

Cons

• Missing type conversion to CaseMap etc
• Risk of hitting Heap and Script statement issues: If your JSON is huge, with too many iterations and conversions needed in Apex to create CaseMap type structure (assuming more fields in real code).

More details:

• JSON Class Apex Docs