Refactor JSON Pointer code to be more self-contained

[tatu-at-salesforce]

It would be nice to make existing JSON pointer more self-contained, to make it easier to reuse -- I am hoping to try to build JSON Pointer support for Jackson, either in core or as add-on module.

[fge]

Agreed, the current code has many adherences which make it uneasy to reuse:

• a heavy dependency on Guava;
• all token decoding/encoding into the class itself;
• "tree resolving" into the class itself as well.

And tree resolving is heavily JsonNode-dependent.

And as a more general rule, the code is quite complicated. As a first step, the plan is to split this class into two disctinct classes:

• ReferenceToken, which would represent one token of a JSON Pointer;
• JsonPointer which would just be a Collection<ReferenceToken>.

This would simplify matters a lot:

• JSON Pointers are always split according to '/', so a simple .split('/') is enough to grab tokens, no need for Guava's Splitter;
• it would be the role of ReferenceToken to err out in the event of a malformed token, by throwing an exception -- which can be changed if the code is salvaged;
• ReferenceToken's .toString() would then output the cooked form of a token, so that JsonPointer's .toString() would just look like:

@Override
public String toString()
{
    final String ret = "";
    for (final ReferenceToken token: tokens)
        ret += '/' + token;
    return ret;
}

That is not all: the current class' .isParentOf() (to be renamed) would still work fine if ReferenceToken's .equals() and .hashCode() just use this token's cooked representation -- while its .resolve() method would use the raw one.

So: code simplification, more portable and adaptable. A win win. Will do it.

[fge]

This leaves one question open, however: even with this refactoring, the code will be dependent on JsonNode. Especially given the special rules of a JSON Pointer with regards to an array:

• any numeric value starting with 0 and which representation is not exactly 0 is not a legal array index (ie, 0 is the first element of an array, 00 in the event of an array is ENOENT);
• what of other tree formats which are not JSON?

Meh. I'll do a first implementation and then we'll see what's what.

[fge]

See here:

java-json-tools/json-schema-core#1