HGxExtract

Extract HGx files from the cs2 VN engine

I've had great trouble with the HGx file format, but I guess I've got it working perfectly now!

Here's what I've understood about the HGx (here I did HG3 but HG2 is almost the same): (It may contain minor errors as I wrote this when I didn't fully understand how this worked)

HG3 File specifications:
Header: 20 Bytes
	-Magic "HG-3" or "HG-2" 	(4B)
	-Unknown values         	(12B)
	-Entry count            	(4B) //Not always relevant
	
Tag   : 16 Bytes
	-Signature              	(8B)
	-OffsetNextTag          	(4B) //Not always relevant
	-Length                 	(4B)
	
Tag signature:"stdinfo"
	StdInfo: 40 Bytes
		-Width                  (4B)
		-Height                 (4B)
		-bpp                    (4B)
		-OffsetX                (4B)
		-OffsetY                (4B)
		-TotalWidth             (4B)
		-TotalHeight            (4B)
		-Unknown values         (12B)

Tag signature:"img0000"
	ImgHdr:  24 Bytes
		-Unknown value          (4B)
		-Height                 (4B)
		-Compressed length      (4B)
		-Original length        (4B)
		-Compressed CMD length  (4B)
		-Original CMD length    (4B)

		Read image:
			1. Read compressed buffer and uncompress it -> buff
			2. Read compressed CMD and uncompress it    -> cmdBuff
			3. Un-RLE (buff, cmdBuff) -> (outBuff, outLen)
			4. RGBABuff[outLen]
			5. Un-delta (outBuff, Width, Height, Bpp{=bpp/8}) -> RGBABuff
			
		Un-RLE:(buff, cmdBuff) -> (outBuff, outLen)
			1. cmdBuff is a bit buffer, 1st bit is copyFlag (boolean)
			2. cmdBuff EliasGamma is outLen;
			3. outBuff[outLen]
			4. n = 0; for(i = 0; i < outLen; i += n)
				a. n = cmdBuff EliasGamma
				b. if(copyFlag)
					Copy n bytes from buff to (outBuff + i)
					buff += n
				c. else
					Set n bytes of (outBuff to 0
				d. Invert copyFlag
			5. Return (outBuff, outLen)
			
		Un-delta:(buff, Width, Height, Bpp) -> (RGBABuff)
			1. table1[256] //Optimisation: table1 to table4 could be hard-coded
			2. table2[256] //That's what I did ^^
			3. table3[256]
			4. table4[256]
			5. for(i = 0; i < 256; i++)
				a. Value = (((((i & 0xC0) << 6 | (i & 0x30)) << 6) | (i & 0x0C)) << 6) | (i & 0x03) //Didn't find how to put this simply...
				b. table4[i] = val << 0
				c. table3[i] = val << 2
				d. table2[i] = val << 4
				e. table1[i] = val << 6
			06. sectLen = buffLen / 4
			07. sect1   = buff
			08. sect2   = sect1 + sectLen
			09. sect3   = sect2 + sectLen
			10. sect4   = sect3 + sectLen
			11. outPtr  = RGBABuff
			12. outEnd  = RGBABuff + buffLen
			13. while(outPtr < outEnd)
				a. Value = table1[*(sect1++)] | table2[*(sect2++)] | table3[*(sect3++)] | table4[*(sect4++)] //Same ^^
				b. *(outPtr++) = ((uchar)(val >>  0) >> 1) ^ ((uchar)(val >>  0) & 1 ? 0xFF : 0x00) //Same ^^
				c. *(outPtr++) = ((uchar)(val >>  8) >> 1) ^ ((uchar)(val >>  8) & 1 ? 0xFF : 0x00)
				d. *(outPtr++) = ((uchar)(val >> 16) >> 1) ^ ((uchar)(val >> 16) & 1 ? 0xFF : 0x00)
				e. *(outPtr++) = ((uchar)(val >> 32) >> 1) ^ ((uchar)(val >> 32) & 1 ? 0xFF : 0x00)
			14. lineLen = Width * Bpp
			15. for(x = Bpp; x < lineLen; x++)
				RGBABuff[x] += RGBABuff[x - Bpp]
			16. for(y = 1; y < height; y++)
				a. line = RGBABuff + y * lineLen
				b. prev = RGBABuff + (y - 1) * lineLen
				c. for(x = 0; x < lineLen; x++)
					line[x] += prev[x]