Unpacks(/packs) QtBinaryJson to json
This defines a binary data structure for Json data. The data structure is optimised for fast reading and minimum allocations. The whole data structure can be mmap'ed and used directly.
In most cases the binary structure is not as space efficient as a utf8 encoded text representation, but much faster to access.
The size requirements are:
String:
Latin1 data: 2 bytes header + string.length()
Full Unicode: 4 bytes header + 2*(string.length())
Values: 4 bytes + size of data (size can be 0 for some data)
bool: 0 bytes
double: 8 bytes (0 if integer with less than 27bits)
string: see above
array: size of array
object: size of object
Array: 12 bytes + 4*length + size of Value data
Object: 12 bytes + 8*length + size of Key Strings + size of Value data
For an example such as
{ // object: 12 + 5*8 = 52
"firstName": "John", // key 12, value 8 = 20
"lastName" : "Smith", // key 12, value 8 = 20
"age" : 25, // key 8, value 0 = 8
"address" : // key 12, object below = 140
{ // object: 12 + 4*8
"streetAddress": "21 2nd Street", // key 16, value 16
"city" : "New York", // key 8, value 12
"state" : "NY", // key 8, value 4
"postalCode" : "10021" // key 12, value 8
}, // object total: 128
"phoneNumber": // key: 16, value array below = 172
[ // array: 12 + 2*4 + values below: 156
{ // object 12 + 2*8
"type" : "home", // key 8, value 8
"number": "212 555-1234" // key 8, value 16
}, // object total: 68
{ // object 12 + 2*8
"type" : "fax", // key 8, value 8
"number": "646 555-4567" // key 8, value 16
} // object total: 68
] // array total: 156
} // great total: 412 bytes
The uncompressed text file used roughly 500 bytes, so in this case we end up using about the same space as the text representation.
Other measurements have shown a slightly bigger binary size than a compact text representation where all possible whitespace was stripped out.