def __parseIFDs(self, base, start, end, IFD=""):
assert IFD != ""
log("Enter", "[%s]" % (IFD), "add")
if not self._file:
assert False
entries = getBytes2(self._file, self._orderAPP1)
log("Number of entries = %d" % (entries))
for idx in xrange(entries):
tag = getBytes2(self._file, self._orderAPP1)
dataFormat = getBytes2(self._file, self._orderAPP1)
numOfComps = getBytes4(self._file, self._orderAPP1)
posBeforeDataOffset = nowAt(self._file)
dataOffset = getBytes4(self._file, self._orderAPP1)
posAfterDataOffset = nowAt(self._file)
if 0 == dataFormat or dataFormat >= len(EXIF_TIFF_DATAFORMAT_LIST):
assert False, "dataformat incorrect = %d" % (dataFormat)
continue
bytesPerComp = EXIF_TIFF_DATAFORMAT_LIST[dataFormat]
dataSize = bytesPerComp * numOfComps
if dataSize > 4:
targetOffset = base + dataOffset
if targetOffset <= start or targetOffset >= end:
continue
else:
seekTo(self._file, targetOffset)
else:
seekTo(self._file, posBeforeDataOffset)
entry = self.__getDataFromFormat(tag, dataFormat, dataSize)
seekTo(self._file, posAfterDataOffset)
log("Leave", "[%s]" % (IFD), "remove")
def __parseIFDs(self, base, start, end, IFD=""):
assert IFD != ""
log("Enter", "[%s]"%(IFD), "add")
if not self._file:
assert False
entries = getBytes2(self._file, self._orderAPP1)
log("Number of entries = %d"%(entries))
for idx in xrange(entries):
tag = getBytes2(self._file, self._orderAPP1)
dataFormat = getBytes2(self._file, self._orderAPP1)
numOfComps = getBytes4(self._file, self._orderAPP1)
posBeforeDataOffset = nowAt(self._file)
dataOffset = getBytes4(self._file, self._orderAPP1)
posAfterDataOffset = nowAt(self._file)
if 0 == dataFormat or dataFormat >= len(EXIF_TIFF_DATAFORMAT_LIST):
assert False, "dataformat incorrect = %d"%(dataFormat)
continue
bytesPerComp = EXIF_TIFF_DATAFORMAT_LIST[dataFormat]
dataSize = bytesPerComp * numOfComps
if dataSize > 4:
targetOffset = base + dataOffset
if targetOffset <= start or targetOffset >= end:
continue
else:
seekTo(self._file, targetOffset)
else:
seekTo(self._file, posBeforeDataOffset)
entry = self.__getDataFromFormat(tag, dataFormat, dataSize)
seekTo(self._file, posAfterDataOffset)
log("Leave", "[%s]"%(IFD), "remove")
def parse(self, filePath):
self._file = open(filePath)
seekTo(self._file, 0)
first = getCharToOrd(self._file)
marker = getCharToOrd(self._file)
if (first != 0xff or marker != JPEG_SOI):
assert False, "Not in JPEG format !!"
while (marker):
first = getCharToOrd(self._file)
if first != 0xff or first < 0:
break
marker = getCharToOrd(self._file)
log("%s-%s" % (hex(first), hex(marker)))
length = getBytes2(self._file)
curPos = nowAt(self._file)
log("length= %d, curPos=%d" % (length, curPos))
if marker in [JPEG_EOI, JPEG_SOS]:
log("EOI or SOS ... exit parsing")
break
elif marker == JPEG_APP0:
log("Enter", "[APP0]", "add")
log("Leave", "[APP0]", "remove")
pass # TBD
elif marker == JPEG_APP1:
log("Enter", "[APP1]", "add")
header = getChar(self._file, 4)
log("header = %s" % (header))
if header.lower() == 'exif':
self.__parseAPP1(curPos + 6, curPos, curPos + length - 2)
elif header.lower() == 'http':
seekTo(self._file, curPos)
xmpBuffer = getChar(self._file, length)
checkURL = "http://ns.adobe.com/xap/1.0/"
if xmpBuffer.startswith(checkURL):
headLen = len(checkURL)
self.__parseXMP(xmpBuffer[headLen:], length - headLen)
pass
log("Leave", "[APP1]", "remove")
elif marker == JPEG_APP2:
log("Enter", "[APP2]", "add")
self.__parseAPP2(length)
log("Leave", "[APP2]", "remove")
pass # TBD
elif marker == JPEG_APP13:
log("Enter", "[APP13]", "add")
log("Leave", "[APP13]", "remove")
pass # TBD
seekTo(self._file, curPos + length - 2)
def parse(self, filePath):
self._file = open(filePath)
seekTo(self._file, 0)
first = getCharToOrd(self._file)
marker = getCharToOrd(self._file)
if (first != 0xff or marker != JPEG_SOI):
assert False, "Not in JPEG format !!"
while (marker):
first = getCharToOrd(self._file)
if first != 0xff or first < 0:
break
marker = getCharToOrd(self._file)
log("%s-%s"%(hex(first), hex(marker)))
length = getBytes2(self._file)
curPos = nowAt(self._file)
log("length= %d, curPos=%d"%(length,curPos))
if marker in [JPEG_EOI, JPEG_SOS]:
log("EOI or SOS ... exit parsing")
break
elif marker == JPEG_APP0:
log("Enter", "[APP0]", "add")
log("Leave", "[APP0]", "remove")
pass # TBD
elif marker == JPEG_APP1:
log("Enter", "[APP1]", "add")
header = getChar(self._file, 4)
log("header = %s"%(header))
if header.lower() == 'exif':
self.__parseAPP1(curPos+6, curPos, curPos+length-2)
elif header.lower() == 'http':
seekTo(self._file, curPos)
xmpBuffer = getChar(self._file, length)
checkURL = "http://ns.adobe.com/xap/1.0/"
if xmpBuffer.startswith(checkURL):
headLen = len(checkURL)
self.__parseXMP(xmpBuffer[headLen:], length-headLen)
pass
log("Leave", "[APP1]", "remove")
elif marker == JPEG_APP2:
log("Enter", "[APP2]", "add")
self.__parseAPP2(length)
log("Leave", "[APP2]", "remove")
pass # TBD
elif marker == JPEG_APP13:
log("Enter", "[APP13]", "add")
log("Leave", "[APP13]", "remove")
pass # TBD
seekTo(self._file, curPos+length-2)
def __parseAPP2(self, length):
curPos = nowAt(self._file)
iptcData = getChar(self._file, length)
iccIdentifier = "ICC_PROFILE"
if (iptcData.startswith(iccIdentifier)):
iccData = iptcData[len(iccIdentifier) + 1:]
iccLen = 0
if ord(iccData[0]) == 0x01 and ord(iccData[1]) == 0x01:
iccLen = length - 14
elif ord(iccData[0]) == 0x01: # multi-page, support header only
iccLen = 128
else:
log("Wrong ICC Profile format !")
return
seekTo(self._file, curPos + 14)
from ICCProfileParser import ICCProfileParser
iccParser = ICCProfileParser(self._file)
iccParser.parse()
else:
log("Wrong ICC Profile format !")
assert False
def __parseAPP2(self, length):
curPos = nowAt(self._file)
iptcData = getChar(self._file, length)
iccIdentifier = "ICC_PROFILE"
if (iptcData.startswith(iccIdentifier)):
iccData = iptcData[len(iccIdentifier)+1:]
iccLen = 0
if ord(iccData[0]) == 0x01 and ord(iccData[1]) == 0x01:
iccLen = length -14
elif ord(iccData[0]) == 0x01: # multi-page, support header only
iccLen = 128
else:
log("Wrong ICC Profile format !")
return
seekTo(self._file, curPos+14)
from ICCProfileParser import ICCProfileParser
iccParser = ICCProfileParser(self._file)
iccParser.parse()
else:
log("Wrong ICC Profile format !")
assert False
def parseTagTable():
log("Enter", "[ICCProfileTagTable]", "add")
tagCount = getBytes4(self._fd)
log("Tag count = %d"%(tagCount))
for idx in xrange(tagCount):
tagStartPos = nowAt(self._fd)
sig = ''.join(getChar(self._fd) for _ in xrange(4))
offset = getBytes4(self._fd)
size = getBytes4(self._fd)
seekTo(self._fd, basePos+offset)
log("Tag sig(%s) / offset(%d) / size(%d) / basePos(%d) / tagSigPos(%d) / tagTypePos(%d) "%(sig, offset, size, basePos, tagStartPos, basePos+offset))
typeDesc = ''.join(getChar(self._fd) for _ in xrange(4))
log("Type Desc(%s)"%(typeDesc))
sigDescObj = GetSigObject(sig, typeDesc, self._fd, size, basePos+offset)
assert sig not in self.__dicSig2TagInfo, "Check this file, two same sig !"
self.__dicSig2TagInfo[sig] = sigDescObj
seekTo(self._fd, tagStartPos+12)
log("Leave", "[ICCProfileTagTable]", "remove")
pprint.pprint(self.__dicSig2TagInfo)
pass
def GetMlucHelper(_fd, sig, tagStartPos):
reserved = getBytes4(_fd)
assert reserved == 0
numOfRecords = getBytes4(_fd)
recordSize = getBytes4(_fd)
log(" numOfRecords = %d / recordSize = %s"%(numOfRecords, recordSize))
sigDescObj = MultiLocalizedUnicode(sig)
for _ in xrange(numOfRecords):
langCode = ''.join(getChar(_fd) for i in xrange(2))
langCountryCode = ''.join(getChar(_fd) for i in xrange(2))
lenRecordString = getBytes4(_fd)
offsetRecordString = getBytes4(_fd)
here = nowAt(_fd)
seekTo(_fd, tagStartPos + offsetRecordString)
uniBytes = getChar(_fd, lenRecordString)
# TODO : Think a better way to store these special unicode glyph
uniChar = unicode(uniBytes, errors='replace')
log(" uniChar = %s"%(uniChar))
sigDescObj.add(langCode, langCountryCode, uniChar)
seekTo(_fd, here)
return sigDescObj
def __parseBasicIFD(self, base, start, end):
log("Enter", "[BasicIFD]", "add")
if not self._file:
assert False
entries = getBytes2(self._file, self._orderAPP1)
log("Number of entries = %d" % (entries))
for idx in xrange(entries):
tag = getBytes2(self._file, self._orderAPP1)
dataFormat = getBytes2(self._file, self._orderAPP1)
numOfComps = getBytes4(self._file, self._orderAPP1)
posBeforeDataOffset = nowAt(self._file)
dataOffset = getBytes4(self._file, self._orderAPP1)
posAfterDataOffset = nowAt(self._file)
if 0 == dataFormat or dataFormat >= len(EXIF_TIFF_DATAFORMAT_LIST):
assert False, "dataformat incorrect = %d" % (dataFormat)
continue
bytesPerComp = EXIF_TIFF_DATAFORMAT_LIST[dataFormat]
dataSize = bytesPerComp * numOfComps
if dataSize > 4:
targetOffset = base + dataOffset
if targetOffset <= start or targetOffset >= end:
continue
else:
seekTo(self._file, targetOffset)
else:
seekTo(self._file, posBeforeDataOffset)
entry = self.__getDataFromFormat(tag, dataFormat, dataSize)
if entry.getTag() == TAGID_ExifIFD:
ifdOffset = entry.getValue()
seekTo(self._file, base + ifdOffset)
self.__parseIFDs(base, start, end, "ExifIFD")
elif entry.getTag() == TAGID_SubIFDs:
log("SubIFDs")
elif entry.getTag() == TAGID_GPSIFD:
ifdOffset = entry.getValue()
seekTo(self._file, base + ifdOffset)
self.__parseIFDs(base, start, end, IFD="GPSIFD")
pass
elif entry.getTag() == TAGID_IPTC:
log("IPTC")
pass
elif entry.getTag() == TAGID_XMP:
log("XMP")
pass
elif entry.getTag() == TAGID_Photoshop:
log("Photoshop")
pass
elif entry.getTag() == TAGID_ICCProfile:
log("ICCProfile")
pass
elif entry.getTag() == TAGID_DNGPrivateData:
log("DNGPrivateData")
pass
seekTo(self._file, posAfterDataOffset)
log("Leave", "[BasicIFD]", "remove")
def __parseBasicIFD(self, base, start, end):
log("Enter", "[BasicIFD]", "add")
if not self._file:
assert False
entries = getBytes2(self._file, self._orderAPP1)
log("Number of entries = %d"%(entries))
for idx in xrange(entries):
tag = getBytes2(self._file, self._orderAPP1)
dataFormat = getBytes2(self._file, self._orderAPP1)
numOfComps = getBytes4(self._file, self._orderAPP1)
posBeforeDataOffset = nowAt(self._file)
dataOffset = getBytes4(self._file, self._orderAPP1)
posAfterDataOffset = nowAt(self._file)
if 0 == dataFormat or dataFormat >= len(EXIF_TIFF_DATAFORMAT_LIST):
assert False, "dataformat incorrect = %d"%(dataFormat)
continue
bytesPerComp = EXIF_TIFF_DATAFORMAT_LIST[dataFormat]
dataSize = bytesPerComp * numOfComps
if dataSize > 4:
targetOffset = base + dataOffset
if targetOffset <= start or targetOffset >= end:
continue
else:
seekTo(self._file, targetOffset)
else:
seekTo(self._file, posBeforeDataOffset)
entry = self.__getDataFromFormat(tag, dataFormat, dataSize)
if entry.getTag() == TAGID_ExifIFD:
ifdOffset = entry.getValue()
seekTo(self._file, base+ifdOffset)
self.__parseIFDs(base, start, end, "ExifIFD")
elif entry.getTag() == TAGID_SubIFDs:
log("SubIFDs")
elif entry.getTag() == TAGID_GPSIFD:
ifdOffset = entry.getValue()
seekTo(self._file, base+ifdOffset)
self.__parseIFDs(base, start, end, IFD="GPSIFD")
pass
elif entry.getTag() == TAGID_IPTC:
log("IPTC")
pass
elif entry.getTag() == TAGID_XMP:
log("XMP")
pass
elif entry.getTag() == TAGID_Photoshop:
log("Photoshop")
pass
elif entry.getTag() == TAGID_ICCProfile:
log("ICCProfile")
pass
elif entry.getTag() == TAGID_DNGPrivateData:
log("DNGPrivateData")
pass
seekTo(self._file, posAfterDataOffset)
log("Leave", "[BasicIFD]", "remove")
def __parseICCProfile(self):
# Refer to http://blog.fpmurphy.com/2012/03/extract-icc-profile-from-images.html
basePos = nowAt(self._fd)
def parseHeader():
log("Enter", "[ICCProfileHeader]", "add")
profileSize = getBytes4(self._fd)
cmmType = ''.join(getChar(self._fd) for _ in xrange(4))
lstVer = [str(getCharToOrd(self._fd)) for _ in xrange(4)]
self.__dicHeaderInfo[H_CMM_TYPE] = cmmType
self.__dicHeaderInfo[H_VERSION] = lstVer[0] + "." + lstVer[1]
deviceClass = ''.join(getChar(self._fd) for _ in xrange(4))
colorSpaceOfData = ''.join(getChar(self._fd) for _ in xrange(4))
pcs = ''.join(getChar(self._fd) for _ in xrange(4))
self.__dicHeaderInfo[H_DEVICE_CLASS] = dicDevCls2Name.get(deviceClass, "Not found")
self.__dicHeaderInfo[H_COLOR_SPACE] = colorSpaceOfData.strip()
self.__dicHeaderInfo[H_PROFILE_CONNECTION_SPACE] = pcs.strip()
lstDatetime = [getBytes2(self._fd) for _ in xrange(6)]
signature = ''.join(getChar(self._fd) for _ in xrange(4))
assert signature == "acsp", "Not a standard ICC Profile !!"
primaryPlatform = ''.join(getChar(self._fd) for _ in xrange(4))
def getBitsList(numBytes):
lstBits = []
for _ in xrange(numBytes):
bits_short = bin(getCharToOrd(self._fd))[2:]
bits_full = '00000000'[len(bits_short):] + bits_short
lstBits.extend([int(b) for b in bits_full])
return lstBits
lstProfileFlags = getBitsList(4)
self.__dicHeaderInfo[H_CREATE_DATETIME] = "Datatime = %d/%d/%d-%d:%d:%d"%tuple(lstDatetime)
self.__dicHeaderInfo[H_SIGNATURE] = signature
self.__dicHeaderInfo[H_PLATFORM] = dicPlatformSig2Desc.get(primaryPlatform, "Not found")
self.__dicHeaderInfo[H_IS_EMBEDED] = True if lstProfileFlags[0] == 1 else False
self.__dicHeaderInfo[H_USED_INDENDENTLY] = False if lstProfileFlags[1] == 1 else True
deviceManufacturer = ''.join(getChar(self._fd) for _ in xrange(4))
deviceModel = ''.join(getChar(self._fd) for _ in xrange(4))
lstDeviceAttributes = getBitsList(8)
renderingIntent, zeroPadding = getBytes2(self._fd), getBytes2(self._fd)
self.__dicHeaderInfo[H_DEVICE_MANUFACTURER] = deviceManufacturer
self.__dicHeaderInfo[H_DEVICE_MODEL] = deviceModel
self.__dicHeaderInfo[H_ATTR_T_R] = "Transparency" if lstDeviceAttributes[0] == 1 else "Reflective"
self.__dicHeaderInfo[H_ATTR_M_G] = "Matte" if lstDeviceAttributes[1] == 1 else "Glossy"
self.__dicHeaderInfo[H_RENDERING_INTENT] = dicRenderingIntent2Desc.get(renderingIntent, "Not found")
intX, intY, intZ = getBytes4(self._fd), getBytes4(self._fd), getBytes4(self._fd)
X = struct.unpack('f', struct.pack('i', intX))
Y = struct.unpack('f', struct.pack('i', intY))
Z = struct.unpack('f', struct.pack('i', intZ))
CIEXYZ_X = X[0] / Y[0]
CIEXYZ_Y = Y[0] / Y[0]
CIEXYZ_Z = Z[0] / Y[0]
profileCreator = ''.join(getChar(self._fd) for _ in xrange(4))
profileID = [hex(getCharToOrd(self._fd)) for _ in xrange(16)]
reserved = [hex(getCharToOrd(self._fd)) for _ in xrange(28)]
self.__dicHeaderInfo[H_PROFILE_CREATOR] = profileCreator
self.__dicHeaderInfo[H_PROFILE_D50_XYZ] = "(%f, %f, %f)"%(CIEXYZ_X, CIEXYZ_Y, CIEXYZ_Z)
log("Header Information : \n%s "%(pprint.pformat(self.__dicHeaderInfo, indent=2)))
log("Leave", "[ICCProfileHeader]", "remove")
def parseTagTable():
log("Enter", "[ICCProfileTagTable]", "add")
tagCount = getBytes4(self._fd)
log("Tag count = %d"%(tagCount))
for idx in xrange(tagCount):
tagStartPos = nowAt(self._fd)
sig = ''.join(getChar(self._fd) for _ in xrange(4))
offset = getBytes4(self._fd)
size = getBytes4(self._fd)
seekTo(self._fd, basePos+offset)
log("Tag sig(%s) / offset(%d) / size(%d) / basePos(%d) / tagSigPos(%d) / tagTypePos(%d) "%(sig, offset, size, basePos, tagStartPos, basePos+offset))
typeDesc = ''.join(getChar(self._fd) for _ in xrange(4))
log("Type Desc(%s)"%(typeDesc))
sigDescObj = GetSigObject(sig, typeDesc, self._fd, size, basePos+offset)
assert sig not in self.__dicSig2TagInfo, "Check this file, two same sig !"
self.__dicSig2TagInfo[sig] = sigDescObj
seekTo(self._fd, tagStartPos+12)
log("Leave", "[ICCProfileTagTable]", "remove")
pprint.pprint(self.__dicSig2TagInfo)
pass
parseHeader()
parseTagTable()
def GetAToBHelper(_fd, sig, tagStartPos, reverse=False):
reserved = getBytes4(_fd)
assert reserved == 0
numOfInputChannel = getCharToOrd(_fd)
numOfOutputChannel = getCharToOrd(_fd)
padding = getBytes2(_fd)
log(" Input(%d) , Output(%d), padding(%d)"%(numOfInputChannel, numOfOutputChannel, padding))
assert padding == 0
sigDescObj = None
lstBCurve = []
mMat = None
lstMCurve = []
clut = []
lstACurve = []
offset2BCurve = getBytes4(_fd)
if offset2BCurve != 0:
here = nowAt(_fd)
seekTo(_fd, tagStartPos + offset2BCurve)
for _ in xrange(numOfOutputChannel):
subType = getChar(_fd, 4)
log(" B Curve subtype = %s"%(subType))
if subType == "para":
sigSubDescObj = GetParaCurveHelper(_fd, sig)
lstBCurve.append(sigSubDescObj)
elif subType == "curv":
sigSubDescObj = GetCurveHelper(_fd, sig)
lstBCurve.append(sigSubDescObj)
seekTo(_fd, here)
assert len(lstBCurve) == numOfOutputChannel
offset2Matrix = getBytes4(_fd)
if offset2Matrix != 0:
here = nowAt(_fd)
seekTo(_fd, tagStartPos + offset2Matrix)
mat = []
for _ in xrange(12):
intUnsigned = getBytes2(_fd)
intSigned = intUnsigned - 65536 if intUnsigned >= 32768 else intUnsigned
fracPart = getBytes2(_fd)
v = intSigned + float(fracPart) / 65536
mat.append(v)
log(" Matrix = %s"%(str(mat)))
mMat = S15Fixed16Array(sig, mat)
seekTo(_fd, here)
offset2MCurve = getBytes4(_fd)
if offset2MCurve != 0:
here = nowAt(_fd)
seekTo(_fd, tagStartPos + offset2MCurve)
for _ in xrange(numOfOutputChannel):
subType = getChar(_fd, 4)
log(" M Curve subtype = %s"%(subType))
if subType == "para":
sigSubDescObj = GetParaCurveHelper(_fd, sig)
lstMCurve.append(sigSubDescObj)
elif subType == "curv":
sigSubDescObj = GetCurveHelper(_fd, sig)
lstMCurve.append(sigSubDescObj)
seekTo(_fd, here)
assert len(lstMCurve) == numOfOutputChannel
offset2CLUT = getBytes4(_fd)
if offset2CLUT != 0:
# TODO : Check the implementation correctness
here = nowAt(_fd)
seekTo(_fd, tagStartPos + offset2CLUT)
lstGridPoints = []
for _ in xrange(16):
gridPts = getCharToOrd(_fd)
if _ >= numOfInputChannel:
assert gridPts == 0
lstGridPoints.append(gridPts)
precision = getCharToOrd(_fd)
padding = getBytes3(_fd)
log(" >>> lstGridPoints : %s"%(str(lstGridPoints)))
log(" >>> precision : %s / padding %s "%(str(precision), str(padding)))
assert padding == 0
getDataPoint = getBytes2 if precision == 2 else getCharToOrd
def fn(x, y):
return x * y if y != 0 else x
totalCLUTPts = reduce(fn, lstGridPoints)
for _ in xrange(totalCLUTPts):
tmp = []
for __ in xrange(numOfOutputChannel):
tmp.append(getDataPoint(_fd))
clut.append(tmp)
seekTo(_fd, here)
offset2ACurve = getBytes4(_fd)
if offset2ACurve != 0:
here = nowAt(_fd)
seekTo(_fd, tagStartPos + offset2ACurve)
for _ in xrange(numOfOutputChannel):
subType = getChar(_fd, 4)
log(" A Curve subtype = %s"%(subType))
if subType == "para":
sigSubDescObj = GetParaCurveHelper(_fd, sig)
lstACurve.append(sigSubDescObj)
elif subType == "curv":
sigSubDescObj = GetCurveHelper(_fd, sig)
lstACurve.append(sigSubDescObj)
seekTo(_fd, here)
assert len(lstACurve) == numOfOutputChannel
log(" O2B(%d) / O2mat(%d) / O2M(%d) / O2CLUT(%d) / O2A(%d)"%(offset2BCurve,\
offset2Matrix, offset2MCurve, offset2CLUT, offset2ACurve))
if reverse:
sigDescObj = LutBToA(sig, lstBCurve, mMat, lstMCurve, clut, lstACurve)
else:
sigDescObj = LutAToB(sig, lstBCurve, mMat, lstMCurve, clut, lstACurve)
return sigDescObj