2013-10-11 17:28:32 +00:00
|
|
|
#
|
|
|
|
# The Python Imaging Library.
|
|
|
|
# $Id$
|
|
|
|
#
|
|
|
|
# IPTC/NAA file handling
|
|
|
|
#
|
|
|
|
# history:
|
|
|
|
# 1995-10-01 fl Created
|
|
|
|
# 1998-03-09 fl Cleaned up and added to PIL
|
|
|
|
# 2002-06-18 fl Added getiptcinfo helper
|
|
|
|
#
|
|
|
|
# Copyright (c) Secret Labs AB 1997-2002.
|
|
|
|
# Copyright (c) Fredrik Lundh 1995.
|
|
|
|
#
|
|
|
|
# See the README file for information on usage and redistribution.
|
|
|
|
#
|
|
|
|
|
2014-09-30 16:15:32 +00:00
|
|
|
from __future__ import print_function
|
2013-10-11 17:28:32 +00:00
|
|
|
|
|
|
|
__version__ = "0.3"
|
|
|
|
|
|
|
|
|
2014-09-30 16:15:32 +00:00
|
|
|
from PIL import Image, ImageFile, _binary
|
2013-10-11 17:28:32 +00:00
|
|
|
import os, tempfile
|
|
|
|
|
2014-09-30 16:15:32 +00:00
|
|
|
i8 = _binary.i8
|
|
|
|
i16 = _binary.i16be
|
|
|
|
i32 = _binary.i32be
|
|
|
|
o8 = _binary.o8
|
2013-10-11 17:28:32 +00:00
|
|
|
|
|
|
|
COMPRESSION = {
|
|
|
|
1: "raw",
|
|
|
|
5: "jpeg"
|
|
|
|
}
|
|
|
|
|
2014-09-30 16:15:32 +00:00
|
|
|
PAD = o8(0) * 4
|
2013-10-11 17:28:32 +00:00
|
|
|
|
|
|
|
#
|
|
|
|
# Helpers
|
|
|
|
|
|
|
|
def i(c):
|
|
|
|
return i32((PAD + c)[-4:])
|
|
|
|
|
|
|
|
def dump(c):
|
|
|
|
for i in c:
|
2014-09-30 16:15:32 +00:00
|
|
|
print("%02x" % i8(i), end=' ')
|
|
|
|
print()
|
2013-10-11 17:28:32 +00:00
|
|
|
|
|
|
|
##
|
|
|
|
# Image plugin for IPTC/NAA datastreams. To read IPTC/NAA fields
|
|
|
|
# from TIFF and JPEG files, use the <b>getiptcinfo</b> function.
|
|
|
|
|
|
|
|
class IptcImageFile(ImageFile.ImageFile):
|
|
|
|
|
|
|
|
format = "IPTC"
|
|
|
|
format_description = "IPTC/NAA"
|
|
|
|
|
|
|
|
def getint(self, key):
|
|
|
|
return i(self.info[key])
|
|
|
|
|
|
|
|
def field(self):
|
|
|
|
#
|
|
|
|
# get a IPTC field header
|
|
|
|
s = self.fp.read(5)
|
|
|
|
if not len(s):
|
|
|
|
return None, 0
|
|
|
|
|
2014-09-30 16:15:32 +00:00
|
|
|
tag = i8(s[1]), i8(s[2])
|
2013-10-11 17:28:32 +00:00
|
|
|
|
|
|
|
# syntax
|
2014-09-30 16:15:32 +00:00
|
|
|
if i8(s[0]) != 0x1C or tag[0] < 1 or tag[0] > 9:
|
|
|
|
raise SyntaxError("invalid IPTC/NAA file")
|
2013-10-11 17:28:32 +00:00
|
|
|
|
|
|
|
# field size
|
2014-09-30 16:15:32 +00:00
|
|
|
size = i8(s[3])
|
2013-10-11 17:28:32 +00:00
|
|
|
if size > 132:
|
2014-09-30 16:15:32 +00:00
|
|
|
raise IOError("illegal field length in IPTC/NAA file")
|
2013-10-11 17:28:32 +00:00
|
|
|
elif size == 128:
|
|
|
|
size = 0
|
|
|
|
elif size > 128:
|
|
|
|
size = i(self.fp.read(size-128))
|
|
|
|
else:
|
|
|
|
size = i16(s[3:])
|
|
|
|
|
|
|
|
return tag, size
|
|
|
|
|
|
|
|
def _is_raw(self, offset, size):
|
|
|
|
#
|
|
|
|
# check if the file can be mapped
|
|
|
|
|
|
|
|
# DISABLED: the following only slows things down...
|
|
|
|
return 0
|
|
|
|
|
|
|
|
self.fp.seek(offset)
|
|
|
|
t, sz = self.field()
|
|
|
|
if sz != size[0]:
|
|
|
|
return 0
|
|
|
|
y = 1
|
2014-09-30 16:15:32 +00:00
|
|
|
while True:
|
2013-10-11 17:28:32 +00:00
|
|
|
self.fp.seek(sz, 1)
|
|
|
|
t, s = self.field()
|
|
|
|
if t != (8, 10):
|
|
|
|
break
|
|
|
|
if s != sz:
|
|
|
|
return 0
|
2014-09-30 16:15:32 +00:00
|
|
|
y += 1
|
2013-10-11 17:28:32 +00:00
|
|
|
return y == size[1]
|
|
|
|
|
|
|
|
def _open(self):
|
|
|
|
|
|
|
|
# load descriptive fields
|
2014-09-30 16:15:32 +00:00
|
|
|
while True:
|
2013-10-11 17:28:32 +00:00
|
|
|
offset = self.fp.tell()
|
|
|
|
tag, size = self.field()
|
|
|
|
if not tag or tag == (8,10):
|
|
|
|
break
|
|
|
|
if size:
|
|
|
|
tagdata = self.fp.read(size)
|
|
|
|
else:
|
|
|
|
tagdata = None
|
2014-09-30 16:15:32 +00:00
|
|
|
if tag in list(self.info.keys()):
|
2013-10-11 17:28:32 +00:00
|
|
|
if isinstance(self.info[tag], list):
|
|
|
|
self.info[tag].append(tagdata)
|
|
|
|
else:
|
|
|
|
self.info[tag] = [self.info[tag], tagdata]
|
|
|
|
else:
|
|
|
|
self.info[tag] = tagdata
|
|
|
|
|
|
|
|
# print tag, self.info[tag]
|
|
|
|
|
|
|
|
# mode
|
2014-09-30 16:15:32 +00:00
|
|
|
layers = i8(self.info[(3,60)][0])
|
|
|
|
component = i8(self.info[(3,60)][1])
|
|
|
|
if (3,65) in self.info:
|
|
|
|
id = i8(self.info[(3,65)][0])-1
|
2013-10-11 17:28:32 +00:00
|
|
|
else:
|
|
|
|
id = 0
|
|
|
|
if layers == 1 and not component:
|
|
|
|
self.mode = "L"
|
|
|
|
elif layers == 3 and component:
|
|
|
|
self.mode = "RGB"[id]
|
|
|
|
elif layers == 4 and component:
|
|
|
|
self.mode = "CMYK"[id]
|
|
|
|
|
|
|
|
# size
|
|
|
|
self.size = self.getint((3,20)), self.getint((3,30))
|
|
|
|
|
|
|
|
# compression
|
|
|
|
try:
|
|
|
|
compression = COMPRESSION[self.getint((3,120))]
|
|
|
|
except KeyError:
|
2014-09-30 16:15:32 +00:00
|
|
|
raise IOError("Unknown IPTC image compression")
|
2013-10-11 17:28:32 +00:00
|
|
|
|
|
|
|
# tile
|
|
|
|
if tag == (8,10):
|
|
|
|
if compression == "raw" and self._is_raw(offset, self.size):
|
|
|
|
self.tile = [(compression, (offset, size + 5, -1),
|
|
|
|
(0, 0, self.size[0], self.size[1]))]
|
|
|
|
else:
|
|
|
|
self.tile = [("iptc", (compression, offset),
|
|
|
|
(0, 0, self.size[0], self.size[1]))]
|
|
|
|
|
|
|
|
def load(self):
|
|
|
|
|
|
|
|
if len(self.tile) != 1 or self.tile[0][0] != "iptc":
|
|
|
|
return ImageFile.ImageFile.load(self)
|
|
|
|
|
|
|
|
type, tile, box = self.tile[0]
|
|
|
|
|
|
|
|
encoding, offset = tile
|
|
|
|
|
|
|
|
self.fp.seek(offset)
|
|
|
|
|
|
|
|
# Copy image data to temporary file
|
2014-09-30 16:15:32 +00:00
|
|
|
o_fd, outfile = tempfile.mkstemp(text=False)
|
|
|
|
o = os.fdopen(o_fd)
|
2013-10-11 17:28:32 +00:00
|
|
|
if encoding == "raw":
|
|
|
|
# To simplify access to the extracted file,
|
|
|
|
# prepend a PPM header
|
|
|
|
o.write("P5\n%d %d\n255\n" % self.size)
|
2014-09-30 16:15:32 +00:00
|
|
|
while True:
|
2013-10-11 17:28:32 +00:00
|
|
|
type, size = self.field()
|
|
|
|
if type != (8, 10):
|
|
|
|
break
|
|
|
|
while size > 0:
|
|
|
|
s = self.fp.read(min(size, 8192))
|
|
|
|
if not s:
|
|
|
|
break
|
|
|
|
o.write(s)
|
2014-09-30 16:15:32 +00:00
|
|
|
size -= len(s)
|
2013-10-11 17:28:32 +00:00
|
|
|
o.close()
|
|
|
|
|
|
|
|
try:
|
|
|
|
try:
|
|
|
|
# fast
|
|
|
|
self.im = Image.core.open_ppm(outfile)
|
|
|
|
except:
|
|
|
|
# slightly slower
|
|
|
|
im = Image.open(outfile)
|
|
|
|
im.load()
|
|
|
|
self.im = im.im
|
|
|
|
finally:
|
|
|
|
try: os.unlink(outfile)
|
|
|
|
except: pass
|
|
|
|
|
|
|
|
|
|
|
|
Image.register_open("IPTC", IptcImageFile)
|
|
|
|
|
|
|
|
Image.register_extension("IPTC", ".iim")
|
|
|
|
|
|
|
|
##
|
|
|
|
# Get IPTC information from TIFF, JPEG, or IPTC file.
|
|
|
|
#
|
|
|
|
# @param im An image containing IPTC data.
|
|
|
|
# @return A dictionary containing IPTC information, or None if
|
|
|
|
# no IPTC information block was found.
|
|
|
|
|
|
|
|
def getiptcinfo(im):
|
|
|
|
|
2014-09-30 16:15:32 +00:00
|
|
|
from PIL import TiffImagePlugin, JpegImagePlugin
|
|
|
|
import io
|
2013-10-11 17:28:32 +00:00
|
|
|
|
|
|
|
data = None
|
|
|
|
|
|
|
|
if isinstance(im, IptcImageFile):
|
|
|
|
# return info dictionary right away
|
|
|
|
return im.info
|
|
|
|
|
|
|
|
elif isinstance(im, JpegImagePlugin.JpegImageFile):
|
|
|
|
# extract the IPTC/NAA resource
|
|
|
|
try:
|
|
|
|
app = im.app["APP13"]
|
|
|
|
if app[:14] == "Photoshop 3.0\x00":
|
|
|
|
app = app[14:]
|
|
|
|
# parse the image resource block
|
|
|
|
offset = 0
|
|
|
|
while app[offset:offset+4] == "8BIM":
|
2014-09-30 16:15:32 +00:00
|
|
|
offset += 4
|
2013-10-11 17:28:32 +00:00
|
|
|
# resource code
|
|
|
|
code = JpegImagePlugin.i16(app, offset)
|
2014-09-30 16:15:32 +00:00
|
|
|
offset += 2
|
2013-10-11 17:28:32 +00:00
|
|
|
# resource name (usually empty)
|
2014-09-30 16:15:32 +00:00
|
|
|
name_len = i8(app[offset])
|
2013-10-11 17:28:32 +00:00
|
|
|
name = app[offset+1:offset+1+name_len]
|
|
|
|
offset = 1 + offset + name_len
|
|
|
|
if offset & 1:
|
2014-09-30 16:15:32 +00:00
|
|
|
offset += 1
|
2013-10-11 17:28:32 +00:00
|
|
|
# resource data block
|
|
|
|
size = JpegImagePlugin.i32(app, offset)
|
2014-09-30 16:15:32 +00:00
|
|
|
offset += 4
|
2013-10-11 17:28:32 +00:00
|
|
|
if code == 0x0404:
|
|
|
|
# 0x0404 contains IPTC/NAA data
|
|
|
|
data = app[offset:offset+size]
|
|
|
|
break
|
|
|
|
offset = offset + size
|
|
|
|
if offset & 1:
|
2014-09-30 16:15:32 +00:00
|
|
|
offset += 1
|
2013-10-11 17:28:32 +00:00
|
|
|
except (AttributeError, KeyError):
|
|
|
|
pass
|
|
|
|
|
|
|
|
elif isinstance(im, TiffImagePlugin.TiffImageFile):
|
|
|
|
# get raw data from the IPTC/NAA tag (PhotoShop tags the data
|
|
|
|
# as 4-byte integers, so we cannot use the get method...)
|
|
|
|
try:
|
2014-09-30 16:15:32 +00:00
|
|
|
data = im.tag.tagdata[TiffImagePlugin.IPTC_NAA_CHUNK]
|
2013-10-11 17:28:32 +00:00
|
|
|
except (AttributeError, KeyError):
|
|
|
|
pass
|
|
|
|
|
|
|
|
if data is None:
|
|
|
|
return None # no properties
|
|
|
|
|
|
|
|
# create an IptcImagePlugin object without initializing it
|
|
|
|
class FakeImage:
|
|
|
|
pass
|
|
|
|
im = FakeImage()
|
|
|
|
im.__class__ = IptcImageFile
|
|
|
|
|
|
|
|
# parse the IPTC information chunk
|
|
|
|
im.info = {}
|
2014-09-30 16:15:32 +00:00
|
|
|
im.fp = io.BytesIO(data)
|
2013-10-11 17:28:32 +00:00
|
|
|
|
|
|
|
try:
|
|
|
|
im._open()
|
|
|
|
except (IndexError, KeyError):
|
|
|
|
pass # expected failure
|
|
|
|
|
|
|
|
return im.info
|