#!/usr/bin/env python

import struct
import string

#00 byte values
#08 string
#10 word values
#18 dword
#20 file/data?

THMB_BIG = 0x2007
THMB_SML = 0x2008
FILE_CRW = 0x0816
FILE_THM = 0x0817

#0805 desc str
#0815 fmt:desc str
#080a make NUL model NUL
#080b firmware str
#0810 owner's name
#080d location?
#102a shr[7] == white balance index
#102c g2? white balance  red = shr[51] / shr[50], blue = shr[52] / shr[53]
#0031 eos d30? red = shr[37] / shr[36], blue = shr[38] / shr[39]
#10a9 d60
#1031 wid = shr[1], height = shr[2]
#1835 decoder table?

def readstruct(fh, fmt):
	return struct.unpack(fmt, fh.read(struct.calcsize(fmt)))

#''.join(filter(lambda x: x not in string.printable, map(chr, range(256))))
#lambda x: string.maketrans(x, '.' * len(x))
#def makeprintable(x, a = string.maketrans

def parse_ciff(fh, offset, length, endian):
	fh.seek(offset + length - 4)
	tboff = readstruct(fh, endian + "I")[0] + offset
	fh.seek(tboff)
	nrecs = readstruct(fh, endian + "H")[0]
	for i in range(nrecs):
		type = readstruct(fh, endian + "H")[0]
		if type & 0x4000:
			len = 8
			type &= 0x3fff
			aoff = fh.tell()
			fh.read(8)
		else:
			len, roff = readstruct(fh, endian + "II")
			aoff = offset + roff;
		save = fh.tell()
		olen = len
		if len > 70:
			print 'limiting value, originally', len
			len = 70
		if 1:
			if type >> 8 in [ 0x28, 0x30]:
				print 'recursing in parse_ciff', aoff, olen
				parse_ciff(fh, aoff, olen, endian)
				print 'back'
			elif 0 and type in [ THMB_BIG, THMB_SML ]:
				fh.seek(aoff)
				open('%x.jpg' % type, "a+").write(fh.read(olen))
			else:
				fh.seek(aoff)
				data = fh.read(len)
				print "%04x: %s" % (type, ''.join(map(lambda x: '%02x' % ord(x), data)))
				print "      %s" % repr(data)

		elif type == 0x080a:
			# handle camera name
			pass
		elif type == 0x1835:
			fh.seek(aoff + 2)
			width, height = readstruct(fh, "HH")
		fh.seek(save)

def idcrw(fh):
	fh.seek(0)
	if fh.read(2) == "MM":
		endian = ">"
	else:
		endian = "<"

	hlen = readstruct(fh, endian + "H")[0]

	if hlen == 0x2a:
		#Tiff
		hoff = readstruct(fh, endian + "I")[0]
		raise NotImplementedError, "can't handle TIFF's"
	elif hlen == 0x1a:
		#CRW
		fh.seek(2)
		hlen = readstruct(fh, endian + "I")[0]
		if fh.read(8) != "HEAPCCDR":
			raise ValueError, "not a CRW file"
		fh.seek(0, 2)
		parse_ciff(fh, hlen, fh.tell() - hlen, endian)

if __name__ == '__main__':
	import sys

	map(idcrw, map(open, sys.argv[1:]))