|
102
|
1 |
"""
|
|
|
2 |
A custom EXIF parsing module, aimed at high performance.
|
|
|
3 |
"""
|
|
|
4 |
|
|
|
5 |
import struct, mmap, os
|
|
|
6 |
|
|
|
7 |
def read_struct (file, fmt) :
|
|
|
8 |
"""
|
|
|
9 |
Utility function to read data from the a file using struct
|
|
|
10 |
"""
|
|
|
11 |
|
|
|
12 |
# length of data
|
|
|
13 |
fmt_size = struct.calcsize(fmt)
|
|
|
14 |
|
|
|
15 |
# get data
|
|
|
16 |
file_data = file.read(fmt_size)
|
|
|
17 |
|
|
|
18 |
# unpack single item, this should raise an error if file_data is too short
|
|
|
19 |
return struct.unpack(fmt, file_data)
|
|
|
20 |
|
|
|
21 |
class Buffer (object) :
|
|
|
22 |
"""
|
|
|
23 |
Wraps a buffer object (anything that supports the python buffer protocol) for read-only access.
|
|
|
24 |
|
|
|
25 |
Includes an offset for relative values, and an endianess for reading binary data.
|
|
|
26 |
"""
|
|
|
27 |
|
|
|
28 |
def __init__ (self, obj, offset=None, size=None, struct_prefix='=') :
|
|
|
29 |
"""
|
|
|
30 |
Create a new Buffer object with a new underlying buffer, created from the given object, offset and size.
|
|
|
31 |
|
|
|
32 |
The endiannes is given in the form of a struct-module prefix, which should be one of '<' or '>'.
|
|
|
33 |
Standard size/alignment are assumed.
|
|
|
34 |
"""
|
|
|
35 |
|
|
|
36 |
# store
|
|
|
37 |
self.buf = buffer(obj, offset, size)
|
|
|
38 |
self.offset = offset
|
|
|
39 |
self.size = size
|
|
|
40 |
self.prefix = struct_prefix
|
|
|
41 |
|
|
|
42 |
def subregion (self, offset, length=None) :
|
|
|
43 |
"""
|
|
|
44 |
Create a new sub-Buffer referencing a view of this buffer, at the given offset, and with the given
|
|
|
45 |
length, if any, and the same struct_prefix.
|
|
|
46 |
"""
|
|
|
47 |
|
|
|
48 |
return Buffer(self.buf, offset, length, struct_prefix=self.prefix)
|
|
|
49 |
|
|
|
50 |
def pread (self, offset, length) :
|
|
|
51 |
"""
|
|
|
52 |
Read a random-access region of raw data
|
|
|
53 |
"""
|
|
|
54 |
|
|
|
55 |
return self.buf[offset:offset + length]
|
|
|
56 |
|
|
|
57 |
def pread_struct (self, offset, fmt) :
|
|
|
58 |
"""
|
|
|
59 |
Read structured data using the given struct format from the given offset.
|
|
|
60 |
"""
|
|
|
61 |
|
|
|
62 |
return struct.unpack_from(self.prefix + fmt, self.buf, offset=offset)
|
|
|
63 |
|
|
|
64 |
def pread_item (self, offset, fmt) :
|
|
|
65 |
"""
|
|
|
66 |
Read a single item of structured data from the given offset.
|
|
|
67 |
"""
|
|
|
68 |
|
|
|
69 |
value, = self.pread_struct(offset, fmt)
|
|
|
70 |
|
|
|
71 |
return value
|
|
|
72 |
|
|
|
73 |
def iter_offsets (self, count, size, offset=0) :
|
|
|
74 |
"""
|
|
|
75 |
Yield a series of offsets for `count` items of `size` bytes, beginning at `offset`.
|
|
|
76 |
"""
|
|
|
77 |
|
|
|
78 |
return xrange(offset, offset + count * size, size)
|
|
|
79 |
|
|
|
80 |
def item_size (self, fmt) :
|
|
|
81 |
"""
|
|
|
82 |
Returns the size in bytes of the given item format
|
|
|
83 |
"""
|
|
|
84 |
|
|
|
85 |
return struct.calcsize(self.prefix + fmt)
|
|
|
86 |
|
|
|
87 |
def unpack_item (self, fmt, data) :
|
|
|
88 |
"""
|
|
|
89 |
Unpacks a single item from the given data
|
|
|
90 |
"""
|
|
|
91 |
|
|
|
92 |
value, = struct.unpack(self.prefix + fmt, data)
|
|
|
93 |
|
|
|
94 |
return value
|
|
|
95 |
|
|
|
96 |
def mmap_buffer (file, size) :
|
|
|
97 |
"""
|
|
|
98 |
Create and return a new read-only mmap'd region
|
|
|
99 |
"""
|
|
|
100 |
|
|
|
101 |
return mmap.mmap(file.fileno(), size, access=mmap.ACCESS_READ)
|
|
|
102 |
|
|
|
103 |
import exif_data
|
|
|
104 |
|
|
|
105 |
class Tag (object) :
|
|
|
106 |
"""
|
|
|
107 |
Represents a single Tag in an IFD
|
|
|
108 |
"""
|
|
|
109 |
|
|
|
110 |
def __init__ (self, offset, tag, type, count, value_ref) :
|
|
|
111 |
"""
|
|
|
112 |
Build a Tag with the given binary items from the IFD entry
|
|
|
113 |
"""
|
|
|
114 |
|
|
|
115 |
self.offset = offset
|
|
|
116 |
self.tag = tag
|
|
|
117 |
self.type = type
|
|
|
118 |
self.count = count
|
|
|
119 |
self.value_ref = value_ref
|
|
|
120 |
|
|
|
121 |
# lookup the type for this tag
|
|
|
122 |
self.type_data = exif_data.FIELD_TYPES.get(type)
|
|
|
123 |
|
|
|
124 |
# unpack it
|
|
|
125 |
if self.type_data :
|
|
|
126 |
self.type_format, self.type_name = self.type_data
|
|
|
127 |
|
|
|
128 |
# lookup the tag data for this tag
|
|
|
129 |
self.tag_data = exif_data.EXIF_TAGS.get(tag)
|
|
|
130 |
|
|
|
131 |
# unpack it
|
|
|
132 |
if self.tag_data :
|
|
|
133 |
# the EXIF tag name
|
|
|
134 |
self.tag_name = tag_data[0]
|
|
|
135 |
|
|
|
136 |
# the optional value formatting specification
|
|
|
137 |
if len(self.tag_data) > 1 :
|
|
|
138 |
self.tag_value_spec = self.tag_data[1]
|
|
|
139 |
|
|
|
140 |
else :
|
|
|
141 |
self.tag_value_spec = None
|
|
|
142 |
|
|
|
143 |
@property
|
|
|
144 |
def name (self) :
|
|
|
145 |
"""
|
|
|
146 |
Lookup the name of this tag via its code, returns None if unknown.
|
|
|
147 |
"""
|
|
|
148 |
|
|
|
149 |
if self.tag_data :
|
|
|
150 |
return self.tag_name
|
|
|
151 |
|
|
|
152 |
else :
|
|
|
153 |
return None
|
|
|
154 |
|
|
|
155 |
def readable_value (self, value) :
|
|
|
156 |
"""
|
|
|
157 |
Convert the given value for this tag into a human-readable string.
|
|
|
158 |
|
|
|
159 |
Returns the value itself by default.
|
|
|
160 |
"""
|
|
|
161 |
|
|
|
162 |
if self.tag_data and self.tag_value_spec :
|
|
|
163 |
# map it
|
|
|
164 |
return exif_data.tag_value(self.tag_value_spec, value)
|
|
|
165 |
|
|
|
166 |
else :
|
|
|
167 |
# nope...
|
|
|
168 |
return value
|
|
|
169 |
|
|
|
170 |
# size of an IFD entry in bytes
|
|
|
171 |
IFD_ENTRY_SIZE = 12
|
|
|
172 |
|
|
|
173 |
class IFD (Buffer) :
|
|
|
174 |
"""
|
|
|
175 |
Represents an IFD (Image file directory) region in EXIF data.
|
|
|
176 |
"""
|
|
|
177 |
|
|
|
178 |
def __init__ (self, buffer, **buffer_opts) :
|
|
|
179 |
"""
|
|
|
180 |
Access the IFD data from the given bufferable object with given buffer opts.
|
|
|
181 |
|
|
|
182 |
This will read the `count` and `next_offset` values.
|
|
|
183 |
"""
|
|
|
184 |
|
|
|
185 |
# init
|
|
|
186 |
super(IFD, self).__init__(buffer, **buffer_opts)
|
|
|
187 |
|
|
|
188 |
# read header
|
|
|
189 |
self.count = self.pread_item(0, 'H')
|
|
|
190 |
|
|
|
191 |
# read next-offset
|
|
|
192 |
self.next_offset = self.pread_item(0x02 + self.count * IFD_ENTRY_SIZE, 'I')
|
|
|
193 |
|
|
|
194 |
def iter_tags (self) :
|
|
|
195 |
"""
|
|
|
196 |
Iterate over all the Tag objects in this IFD
|
|
|
197 |
"""
|
|
|
198 |
|
|
|
199 |
# read each tag
|
|
|
200 |
for offset in self.iter_offsets(self.count, IFD_ENTRY_SIZE, 0x02) :
|
|
|
201 |
# read the tag data
|
|
|
202 |
tag, type, count, value_ref = self.pread_struct(offset, 'HHII')
|
|
|
203 |
|
|
|
204 |
# yield the new Tag
|
|
|
205 |
yield Tag(offset, tag, type, count, value_ref)
|
|
|
206 |
|
|
|
207 |
class EXIF (Buffer) :
|
|
|
208 |
"""
|
|
|
209 |
Represents the EXIF data embedded in some image file in the form of a Region.
|
|
|
210 |
"""
|
|
|
211 |
|
|
|
212 |
def __init__ (self, buffer, tags=None, **buffer_opts) :
|
|
|
213 |
"""
|
|
|
214 |
Access the EXIF data from the given bufferable object with the given buffer options.
|
|
|
215 |
|
|
|
216 |
`tags`, if given, specifies that only the given named tags should be loaded.
|
|
|
217 |
"""
|
|
|
218 |
|
|
|
219 |
# init Buffer
|
|
|
220 |
super(EXIF, self).__init__(buffer, **buffer_opts)
|
|
|
221 |
|
|
|
222 |
# store
|
|
|
223 |
self.buffer = buffer
|
|
|
224 |
|
|
|
225 |
def iter_ifds (self) :
|
|
|
226 |
"""
|
|
|
227 |
Iterate over all of the IFD objects in this EXIF.
|
|
|
228 |
"""
|
|
|
229 |
|
|
|
230 |
# starting offset
|
|
|
231 |
offset = self.pread_item(0x04, 'I')
|
|
|
232 |
|
|
|
233 |
while offset :
|
|
|
234 |
# create and read the IFD
|
|
|
235 |
ifd = IFD(self, offset=offset)
|
|
|
236 |
|
|
|
237 |
# yield it
|
|
|
238 |
yield ifd
|
|
|
239 |
|
|
|
240 |
# skip to next offset
|
|
|
241 |
offset = ifd.next_offset
|
|
|
242 |
|
|
|
243 |
__iter__ = iter_ifds
|
|
|
244 |
|
|
|
245 |
def tag_values (self, tag) :
|
|
|
246 |
"""
|
|
|
247 |
Get the raw values for the given tag as a tuple.
|
|
|
248 |
|
|
|
249 |
Returns None if the tag could not be recognized.
|
|
|
250 |
"""
|
|
|
251 |
|
|
|
252 |
# unknown tag?
|
|
|
253 |
if not tag.type_data :
|
|
|
254 |
return None
|
|
|
255 |
|
|
|
256 |
# size of the data
|
|
|
257 |
data_size = tag.count * self.item_size(tag.type_format)
|
|
|
258 |
|
|
|
259 |
# inline or external?
|
|
|
260 |
if data_size > 0x04 :
|
|
|
261 |
# point at the external data
|
|
|
262 |
offset = self.unpack_item('I', tag.value_ref)
|
|
|
263 |
|
|
|
264 |
else :
|
|
|
265 |
# point at the inline data
|
|
|
266 |
offset = tag.offset + 0x08
|
|
|
267 |
|
|
|
268 |
# read values
|
|
|
269 |
return self.pread_struct(offset, "%d%s" % (tag.count, tag.type_format))
|
|
|
270 |
|
|
|
271 |
def tag_value (self, tag) :
|
|
|
272 |
"""
|
|
|
273 |
Return the human-readable string value for the given tag.
|
|
|
274 |
"""
|
|
|
275 |
|
|
|
276 |
# load the raw values
|
|
|
277 |
values = self.tag_values(tag)
|
|
|
278 |
|
|
|
279 |
# unknown?
|
|
|
280 |
if not values :
|
|
|
281 |
return ""
|
|
|
282 |
|
|
|
283 |
# return as comma-separated formatted string, yes
|
|
|
284 |
return ", ".join(tag.readable_value(value) for value in values)
|
|
|
285 |
|
|
|
286 |
# mapping from two-byte TIFF byte order marker to struct prefix
|
|
|
287 |
TIFF_BYTE_ORDER = {
|
|
|
288 |
'II': '<',
|
|
|
289 |
'MM': '>',
|
|
|
290 |
}
|
|
|
291 |
|
|
|
292 |
# "An arbitrary but carefully chosen number (42) that further identifies the file as a TIFF file"
|
|
|
293 |
TIFF_BYTEORDER_MAGIC = 42
|
|
|
294 |
|
|
|
295 |
def tiff_load (file, length=0, **opts) :
|
|
|
296 |
"""
|
|
|
297 |
Load the Exif/TIFF data from the given file at its current position with optional length, using exif_load.
|
|
|
298 |
"""
|
|
|
299 |
|
|
|
300 |
# all Exif data offsets are relative to the beginning of this TIFF header
|
|
|
301 |
offset = file.tell()
|
|
|
302 |
|
|
|
303 |
# mmap the region for the EXIF data
|
|
|
304 |
buffer = mmap_region(file, length)
|
|
|
305 |
|
|
|
306 |
# read byte-order header
|
|
|
307 |
byte_order = file.read(2)
|
|
|
308 |
|
|
|
309 |
# map to struct prefix
|
|
|
310 |
struct_prefix = TIFF_BYTE_ORDER[byte_order]
|
|
|
311 |
|
|
|
312 |
# validate
|
|
|
313 |
check_value, = read_struct(file, struct_prefix + 'H')
|
|
|
314 |
|
|
|
315 |
if check_value != TIFF_BYTEORDER_MAGIC :
|
|
|
316 |
raise Exception("Invalid byte-order for TIFF: %2c -> %d" % (byte_order, check_value))
|
|
|
317 |
|
|
|
318 |
# build and return the EXIF object with the correct offset/size from the mmap region
|
|
|
319 |
return EXIF(buffer, offset=offset, size=length, **opts)
|
|
|
320 |
|
|
|
321 |
# the JPEG markers that don't have any data
|
|
|
322 |
JPEG_NOSIZE_MARKERS = (0xD8, 0xD9)
|
|
|
323 |
|
|
|
324 |
# the first marker in a JPEG File
|
|
|
325 |
JPEG_START_MARKER = 0xD8
|
|
|
326 |
|
|
|
327 |
# the JPEG APP1 marker used for EXIF
|
|
|
328 |
JPEG_EXIF_MARKER = 0xE1
|
|
|
329 |
|
|
|
330 |
# the JPEG APP1 Exif header
|
|
|
331 |
JPEG_EXIF_HEADER = "Exif\x00\x00"
|
|
|
332 |
|
|
|
333 |
def jpeg_markers (file) :
|
|
|
334 |
"""
|
|
|
335 |
Iterate over the JPEG markers in the given file, yielding (type_byte, size) tuples.
|
|
|
336 |
|
|
|
337 |
The size fields will be 0 for markers with no data. The file will be positioned at the beginning of the data
|
|
|
338 |
region, and may be seek'd around if needed.
|
|
|
339 |
|
|
|
340 |
XXX: find a real implementation of this somewhere?
|
|
|
341 |
"""
|
|
|
342 |
|
|
|
343 |
while True :
|
|
|
344 |
# read type
|
|
|
345 |
marker_byte, marker_type = read_struct(file, '!BB')
|
|
|
346 |
|
|
|
347 |
# validate
|
|
|
348 |
if marker_byte != 0xff :
|
|
|
349 |
raise Exception("Not a JPEG marker: %x%x" % (marker_byte, marker_type))
|
|
|
350 |
|
|
|
351 |
# special cases for no data
|
|
|
352 |
if marker_byte in JPEG_NOSIZE_MARKERS :
|
|
|
353 |
size = 0
|
|
|
354 |
|
|
|
355 |
else :
|
|
|
356 |
# read size field
|
|
|
357 |
size, = read_struct(file, '!H')
|
|
|
358 |
|
|
|
359 |
# validate
|
|
|
360 |
if size < 0x02 :
|
|
|
361 |
raise Exception("Invalid size for marker %x%x: %x" % (marker_byte, marker_type, size))
|
|
|
362 |
|
|
|
363 |
else :
|
|
|
364 |
# do not count the size field itself
|
|
|
365 |
size = size - 2
|
|
|
366 |
|
|
|
367 |
# ok, data is at current position
|
|
|
368 |
offset = file.tell()
|
|
|
369 |
|
|
|
370 |
# yield
|
|
|
371 |
yield marker_type, size
|
|
|
372 |
|
|
|
373 |
# absolute seek to next marker
|
|
|
374 |
file.seek(offset + size)
|
|
|
375 |
|
|
|
376 |
def jpeg_find_exif (file) :
|
|
|
377 |
"""
|
|
|
378 |
Find the Exif/TIFF section in the given JPEG file.
|
|
|
379 |
|
|
|
380 |
If found, the file will be seek'd to the start of the Exif/TIFF header, and the size of the Exif/TIFF data will
|
|
|
381 |
be returned.
|
|
|
382 |
|
|
|
383 |
Returns None if no EXIF section was found.
|
|
|
384 |
"""
|
|
|
385 |
|
|
|
386 |
for count, (marker, size) in enumerate(jpeg_markers(file)) :
|
|
|
387 |
# verify that it's a JPEG file
|
|
|
388 |
if count == 0 :
|
|
|
389 |
# must start with the right marker
|
|
|
390 |
if marker != JPEG_START_MARKER :
|
|
|
391 |
raise Exception("JPEG file must start with 0xFF%02x marker" % (marker, ))
|
|
|
392 |
|
|
|
393 |
# look for APP1 marker (0xE1) with EXIF signature
|
|
|
394 |
elif marker == JPEG_EXIF_MARKER and file.read(len(JPEG_EXIF_HEADER)) == JPEG_EXIF_HEADER:
|
|
|
395 |
# skipped the initial Exif marker signature
|
|
|
396 |
return size - JPEG_EXIF_HEADER
|
|
|
397 |
|
|
|
398 |
# nothing
|
|
|
399 |
return None
|
|
|
400 |
|
|
|
401 |
def jpeg_load (file, **opts) :
|
|
|
402 |
"""
|
|
|
403 |
Loads the embedded Exif TIFF data from the given JPEG file using tiff_load.
|
|
|
404 |
|
|
|
405 |
Returns None if no EXIF data could be found.
|
|
|
406 |
"""
|
|
|
407 |
|
|
|
408 |
# look for the right section
|
|
|
409 |
size = jpeg_find_exif(file)
|
|
|
410 |
|
|
|
411 |
# not found?
|
|
|
412 |
if not res :
|
|
|
413 |
# nothing
|
|
|
414 |
return
|
|
|
415 |
|
|
|
416 |
else :
|
|
|
417 |
# load it as TIFF data
|
|
|
418 |
return tiff_load(file, size, **opts)
|
|
|
419 |
|
|
|
420 |
def load_path (path, **opts) :
|
|
|
421 |
"""
|
|
|
422 |
Loads an EXIF object from the given filesystem path.
|
|
|
423 |
|
|
|
424 |
Returns None if it could not be parsed.
|
|
|
425 |
"""
|
|
|
426 |
|
|
|
427 |
# file extension
|
|
|
428 |
root, fext = os.path.splitext(path)
|
|
|
429 |
|
|
|
430 |
# map
|
|
|
431 |
func = {
|
|
|
432 |
'.jpeg': jpeg_load,
|
|
|
433 |
'.jpg': jpeg_load,
|
|
|
434 |
'.tiff': tiff_load, # XXX: untested
|
|
|
435 |
}.get(fext.lower())
|
|
|
436 |
|
|
|
437 |
# not recognized?
|
|
|
438 |
if not func :
|
|
|
439 |
# XXX: sniff the file
|
|
|
440 |
return None
|
|
|
441 |
|
|
|
442 |
# open it
|
|
|
443 |
file = open(path, 'rb')
|
|
|
444 |
|
|
|
445 |
# try and load it
|
|
|
446 |
return func(file, **opts)
|
|
|
447 |
|
|
|
448 |
def dump_exif (exif) :
|
|
|
449 |
"""
|
|
|
450 |
Dump all tags from the given EXIF object to stdout
|
|
|
451 |
"""
|
|
|
452 |
|
|
|
453 |
print "EXIF offset=%d, size=%d:" % (exif.offset, exif.size)
|
|
|
454 |
|
|
|
455 |
for i, ifd in enumerate(exif.iter_ifds()) :
|
|
|
456 |
print "\tIFD %d, offset=%d, size=%d, count=%d, next=%d:" % (i, ifd.offset, ifd.size, ifd.count, ifd.next_offset)
|
|
|
457 |
|
|
|
458 |
for i, tag in enumerate(exif.iter_tags()) :
|
|
|
459 |
print "\t\tTag %d, offset=%d, tag=%d/%s, type=%d/%s, count=%d:" % (
|
|
|
460 |
i,
|
|
|
461 |
tag.offset,
|
|
|
462 |
tag.code, tag.name or '???',
|
|
|
463 |
tag.type, tag.type_name if tag.type_data else '???',
|
|
|
464 |
tag.count,
|
|
|
465 |
)
|
|
|
466 |
|
|
|
467 |
for i, value in enumerate(exif.tag_values(tag)) :
|
|
|
468 |
print "\t\t\t%02d: %s" % (i, tag.readable_value(value))
|
|
|
469 |
|
|
|
470 |
def main (path) :
|
|
|
471 |
"""
|
|
|
472 |
Load and dump EXIF data from the given path
|
|
|
473 |
"""
|
|
|
474 |
|
|
|
475 |
# try and load it
|
|
|
476 |
exif = load_path(path)
|
|
|
477 |
|
|
|
478 |
if not exif :
|
|
|
479 |
raise Exception("No EXIF data found")
|
|
|
480 |
|
|
|
481 |
# dump it
|
|
|
482 |
print "%s: " % path
|
|
|
483 |
print
|
|
|
484 |
|
|
|
485 |
dump_exif(exif)
|
|
|
486 |
|
|
|
487 |
if __name__ == '__main__' :
|
|
|
488 |
from sys import argv
|
|
|
489 |
|
|
|
490 |
main(argv[1])
|
|
|
491 |
|