Fix offset value when escape characters exist

[SebastienGllmt]

Currently, the code assumes that 1 character in JSON corresponds to 1 character in the input. However, this isn't true because JSON supports escape characters (so multiple characters can map to a single JSON character)

This PR fixes it, and adds tests for this case

[juanjoDiaz]

Hi @SebastienGllmt ,

This PR seems a duplicate of #24. Or at least related.

Currently, the offset property counts the bytes in the stream. Not the characters in the string.
That's because the parser can take in a string but also an array of bytes, a Uint8Array, etc...

It's impossible for the library to know if you passed the string"д", or the Uint8Array [0xD0, 0x94] or the Uint16Array [0x0414]. So, currently, it will offset 2 because that's the amount of bytes.
Escaped sequences are typically 1 bye and unicode characters can be 1, 2, 3 or 4 bytes...

What you changed here simply counts all escaped chars as 2 bytes and all the unicode sequences as 6.
So it doesn't seem correct.

What do you think?

[SebastienGllmt]

Escaped sequences are typically 1 bye and unicode characters can be 1, 2, 3 or 4 bytes...

While this is true in general, you can see from the image I posted above that this doesn't matter for JSON specification purposes. \u is always followed by exactly 4 hexadecimal digits (so 6 total since it's prefixed with \u), and other escape codes are always exactly 2

[SebastienGllmt]

It's impossible for the library to know if you passed the string"д", or the Uint8Array [0xD0, 0x94] or the Uint16Array [0x0414]. So, currently, it will offset 2 because that's the amount of bytes.

I don't think this matters. The two below are two totally separate and valid JSON strings. The escape is not done by Javascript - the escape is done by the JSON parser itself.

console.log(JSON.parse('{"char": "\\u0434"}'));
console.log(JSON.parse('{"char": "д"}'));

That is to say, the input to the parser are not both d0b4. One is d0b4 and the other is 5c7530343334

[juanjoDiaz]

I think that you are right, but I have a few coding styles comments.

Once they are fixed, I'm happy to merge this.