raP File Format - OFP
Gah!
dont' touch this one folks just at moment, it's a cut 'n paste from my website and needs *severe* re-wording.
I've plunked it here in my sandbox to get to work on it as and when....
am including further info about elite and arma
Introduction
raP encoding applies to any humanly readable text file in OFP that contains class statements. Examples of files that are, or should be, raPified, are mission.sqm, config.cpp, description.ext.
In fact, any text file that contains class statements, contains nothing else but class statements. So much so, that entire contents of that file, is considered to be a class !!!
eg
class mission.sqm
{
...
};
The fact of the matter is, if you do not raPify these files, the engine will before using them (and thus causing uneccessary cpu load)
raP encoding simply means that the data inherent in these types of files has been sanitised (stripped of commments and crud) and massaged into a form of indexed lookup table for the engine to use directly. Once done, it is free of the need to check for syntax errors, among other things. Hence, much much faster processing.
These types of files were once known as 'encrypted' or 'binarised' files. They are no such thing. They are simply a cleaner. closer equivalent to what the engine uses internally. For instance, all your savegames are raP encoded (there is no, text equivalent).
A raP encoded file is detected by the magic signature '\0raP' in the first four bytes of the file. Because of the leading 0 byte, no text file can inadvertently have this signature.
Importantly, the filename extension is immaterial.
The engine will work with config.cpp as a raP encoded entity, just as it would work with config.bin.
Tools
Various utilities exist which refer to binary <> cpp compression and extraction (or encoding and decoding). Again, these terms are misleading because the file concerned is not executable binary data, just tokenised strings and values.
Basics
There is no need here to elaborately define what a mission.sqm file is. But, it is worth understanding the basics of these (types of) files to understand the very small requirements needed to raPify them.
class files only contain 3 types of construct
ClassNames, TokenNames, Arrays
class classname [:inherit] {...};
[:inherit] is optional and simply refers to another classname.
(For your interest the [] are known as bacchus naur format and mean optional. Whatever is within the [...] is optional. The [] do not appear in the text file.)
The class body, the {...} contains more, Classnames, TokenNames, Arrays, or nothing at all.
TokenNames come in 3 flavours
aString="A string"; anInteger=77; aFloat 1.855;
For more on this subject, see TokenNameValueTypes
Arrays
anArray[]={......};
an array containing elements including (possibly) more arrays or more TokenNames (but not classes)
thing[]={ 1.0, 7.67, "Elephants", fred[]={......} };
raPifying encodes each of these basic types.
Construct
all raPified data can be expressed as
class filename
{
class FirstEmbeddedClass
{
... tokenames
class FirstEmbeddedEmbeddedClass
{
...
};
...
};
...
class LastEmbeddedClass
{
};
};
number of #defines (optional)
- define table (if any)
the beginning class, the filename, is not recorded in the humanly readable text output. See below for #defines
Header
A raPified file has the first 4 bytes of the file encoded as follows:
"\0raP"
For OFP and RESISTANCE the next three bytes are
"\004\0\0"
see elsewhere for Elite and ArmA.
The rest of the file contain Class Body packets of 3 different construct types with the 1st byte defining what 'type' it is.
Thus
struct ClassBody
{
byte PacketType; // 0 Classname
// 1 TokenName
// 2 Array
....... depends on packet type
};
The very first packet encountered is a classname. It is the enclosing class for *everything* else in the file. The name of his class is the name of the file. It is *not* recorded in humanly readable text output.
Packets
PacketType0: Classname
class Classname: InheritedClassName { Packets... };
struct ClassPacket
{
byte PacketType; // = 0 == class
IndexedString Classname;
Asciiz InheritedClassName; // optional or zero length string
BIS_short nImbeddedPackets; // Iterates thru embedded Packet(s) can be zero
};
Having no embedded packets is quite legal.
The embedded packets, eg, the body of this class, immediately follows this packet.
Bare in mind, that the following data (the body of this class) may indeed have further embedded packets, which may have further embedded packets, which may have.... All of which are contiguous in the datastream (OFP Only).
PacketType1: Variables
The first byte of this packet defines what type of variable. Thus
struct VarPacket
{
byte PacketType; // = 1
byte VarType; // = 0 to 2
IndexedString SomeName;
.... depends on VarType
};
VarType0 String
SomeName="SomeOtherName";
struct VarTypString
{
byte PacketType; // = 1
byte VarType; // = 0
IndexedString SomeName;
IndexedString SomeOtherName;
};
VarType1 Float
SomeName=1.23445;
struct VarTypFloat
{
byte PacketType; // = 1
byte VarType; // = 1
IndexedString SomeName;
float value; // 4 bytes
};
VarType2 Integer
SomeName=123;
struct VarTypLongInteger
{
byte PacketType; // = 1
byte VarType; // = 2
IndexedString SomeName;
int value; // 4 bytes
};
PacketType2: Arrays
Arrays[] contain four possible element types. They are the traditional variables mentioned above with an added tweak of an embedded array type.
thus
SomeName[]={ Element,Element[],"element",....};
struct ArrayPacket
{
byte PacketType; // = 2
IndexedString SomeName;
BIS_short nConstTypes; // iterate thru ConstTypes, can be 0
.... depends on ConstTypes
};
ArrayType0 String
"SomeName",
struct ArrayString
{
byte VarType; // = 0
IndexedString SomeName;
};
ArrayType1 Float
1.234,
struct ArrayFloat
{
byte VarType; // = 1
float value; // 4 bytes
};
ArrayType2 Integer
123,
struct ArrayInteger
{
byte VarType; // = 2
int value; // 4 bytes
};
ArrayType3 Embedded_Array
{array(...},....},
struct EmbeddedArray
{
byte VarType; // = 3
BIS_short nArrayElements; // iterate thru ConstTypes
... depends on elememt types in this embedded array
};
with the above construct (embedded array) each embedded array can contain any ArrayType, including, another embedded array. The difference of course is these embedded arrays have no individual name associated with them (unlike the packet array).
Added Wrinkles
- defines
Optionally, an encrypted file can contain a #define table after the filename class definition.
Long NumberOfDefines
Struct DefTable
{
Asciiz String;
Long value;
}[NumberOfDefines];
type definitions
Bis_Short
the value us either one, or two bytes.
{
int val;
if ((val = GetByte())==EOF) return EOF;
if (val & 0x80)
{
int extra;
if ((extra = GetByte())==EOF) return EOF;
val += (extra - 1) * 0x80;
}
return val;
}
IndexedString struct {
Bis_Short index; Asciiz String;
}; a table of strings is recorded according to it's index number when that specific index number is first encountered. Although the values appear to be ordinal (0,1,2,3,4,5) you should not assume so. 0 ="Peter" 1="Paul" 2="Mary" These are defined index strings and appear, individually, and uniquely, within the mission.sqm as and when they are first encountered. From then on you will only see an index string as 1="" because 1 has been defined earlier on. Note that this is unlike a postscript dictionary in that strings are defined on an add -hoc basis, not at beginning, only when encountered, this 0="peter" 0= 0= 1="mary" 0= 1= 0= 2="fred" 2= 1= etc