Passed
Push — master ( 75e659...d8cee1 )
by Jaisen
01:58
created

elodie.external.pyexiftool.ExifTool.__init__()   A

Complexity

Conditions 4

Size

Total Lines 15
Code Lines 10

Duplication

Lines 0
Ratio 0 %

Importance

Changes 0
Metric Value
cc 4
eloc 10
nop 3
dl 0
loc 15
rs 9.9
c 0
b 0
f 0
1
# -*- coding: utf-8 -*-
2
# PyExifTool <http://github.com/smarnach/pyexiftool>
3
# Copyright 2012 Sven Marnach. Enhancements by Leo Broska
4
5
# This file is part of PyExifTool.
6
#
7
# PyExifTool is free software: you can redistribute it and/or modify
8
# it under the terms of the GNU General Public License as published by
9
# the Free Software Foundation, either version 3 of the License, or
10
# (at your option) any later version.
11
#
12
# PyExifTool is distributed in the hope that it will be useful,
13
# but WITHOUT ANY WARRANTY; without even the implied warranty of
14
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15
# GNU General Public License for more details.
16
#
17
# You should have received a copy of the GNU General Public License
18
# along with PyExifTool.  If not, see <http://www.gnu.org/licenses/>.
19
20
"""
21
PyExifTool is a Python library to communicate with an instance of Phil
22
Harvey's excellent ExifTool_ command-line application.  The library
23
provides the class :py:class:`ExifTool` that runs the command-line
24
tool in batch mode and features methods to send commands to that
25
program, including methods to extract meta-information from one or
26
more image files.  Since ``exiftool`` is run in batch mode, only a
27
single instance needs to be launched and can be reused for many
28
queries.  This is much more efficient than launching a separate
29
process for every single query.
30
31
.. _ExifTool: http://www.sno.phy.queensu.ca/~phil/exiftool/
32
33
The source code can be checked out from the github repository with
34
35
::
36
37
    git clone git://github.com/smarnach/pyexiftool.git
38
39
Alternatively, you can download a tarball_.  There haven't been any
40
releases yet.
41
42
.. _tarball: https://github.com/smarnach/pyexiftool/tarball/master
43
44
PyExifTool is licenced under GNU GPL version 3 or later.
45
46
Example usage::
47
48
    import exiftool
49
50
    files = ["a.jpg", "b.png", "c.tif"]
51
    with exiftool.ExifTool() as et:
52
        metadata = et.get_metadata_batch(files)
53
    for d in metadata:
54
        print("{:20.20} {:20.20}".format(d["SourceFile"],
55
                                         d["EXIF:DateTimeOriginal"]))
56
"""
57
58
from __future__ import unicode_literals
59
60
import sys
61
import subprocess
62
import os
63
import json
64
import warnings
65
import logging
66
import codecs
67
68
from future.utils import with_metaclass
69
70
try:        # Py3k compatibility
71
    basestring
0 ignored issues
show
Comprehensibility Best Practice introduced by
The variable basestring does not seem to be defined.
Loading history...
72
except NameError:
73
    basestring = (bytes, str)
74
75
executable = "exiftool"
76
"""The name of the executable to run.
77
78
If the executable is not located in one of the paths listed in the
79
``PATH`` environment variable, the full path should be given here.
80
"""
81
82
# Sentinel indicating the end of the output of a sequence of commands.
83
# The standard value should be fine.
84
sentinel = b"{ready}"
85
86
# The block size when reading from exiftool.  The standard value
87
# should be fine, though other values might give better performance in
88
# some cases.
89
block_size = 4096
90
91
# constants related to keywords manipulations 
92
KW_TAGNAME = "IPTC:Keywords"
93
KW_REPLACE, KW_ADD, KW_REMOVE = range(3)
94
95
96
# This code has been adapted from Lib/os.py in the Python source tree
97
# (sha1 265e36e277f3)
98
def _fscodec():
99
    encoding = sys.getfilesystemencoding()
100
    errors = "strict"
101
    if encoding != "mbcs":
102
        try:
103
            codecs.lookup_error("surrogateescape")
104
        except LookupError:
105
            pass
106
        else:
107
            errors = "surrogateescape"
108
109
    def fsencode(filename):
110
        """
111
        Encode filename to the filesystem encoding with 'surrogateescape' error
112
        handler, return bytes unchanged. On Windows, use 'strict' error handler if
113
        the file system encoding is 'mbcs' (which is the default encoding).
114
        """
115
        if isinstance(filename, bytes):
116
            return filename
117
        else:
118
            return filename.encode(encoding, errors)
119
120
    return fsencode
121
122
fsencode = _fscodec()
123
del _fscodec
124
125
#string helper
126
def strip_nl (s):
127
    return ' '.join(s.splitlines())
128
129
130
# Error checking function
131
# Note: They are quite fragile, beacsue teh just parse the output text from exiftool
132
def check_ok (result):
133
    """Evaluates the output from a exiftool write operation (e.g. `set_tags`)
134
    
135
    The argument is the result from the execute method.
136
    
137
    The result is True or False.
138
    """
139
    return not result is None and (not "due to errors" in result)
140
141
def format_error (result):
142
    """Evaluates the output from a exiftool write operation (e.g. `set_tags`)
143
    
144
    The argument is the result from the execute method.
145
    
146
    The result is a human readable one-line string.
147
    """
148
    if check_ok (result):
149
        return 'exiftool finished probably properly. ("%s")' % strip_nl(result)
150
    else:        
151
        if result is None:
152
            return "exiftool operation can't be evaluated: No result given"
153
        else:
154
            return 'exiftool finished with error: "%s"' % strip_nl(result) 
155
156
class Singleton(type):
157
    """Metaclass to use the singleton [anti-]pattern"""
158
    instance = None
159
160
    def __call__(cls, *args, **kwargs):
161
        if cls.instance is None:
162
            cls.instance = super(Singleton, cls).__call__(*args, **kwargs)
163
        return cls.instance
164
165
class ExifTool(object, with_metaclass(Singleton)):
166
    """Run the `exiftool` command-line tool and communicate to it.
167
168
    You can pass two arguments to the constructor:
169
    - ``addedargs`` (list of strings): contains additional paramaters for
170
      the stay-open instance of exiftool
171
    - ``executable`` (string): file name of the ``exiftool`` executable.
172
      The default value ``exiftool`` will only work if the executable
173
      is in your ``PATH``
174
175
    Most methods of this class are only available after calling
176
    :py:meth:`start()`, which will actually launch the subprocess.  To
177
    avoid leaving the subprocess running, make sure to call
178
    :py:meth:`terminate()` method when finished using the instance.
179
    This method will also be implicitly called when the instance is
180
    garbage collected, but there are circumstance when this won't ever
181
    happen, so you should not rely on the implicit process
182
    termination.  Subprocesses won't be automatically terminated if
183
    the parent process exits, so a leaked subprocess will stay around
184
    until manually killed.
185
186
    A convenient way to make sure that the subprocess is terminated is
187
    to use the :py:class:`ExifTool` instance as a context manager::
188
189
        with ExifTool() as et:
190
            ...
191
192
    .. warning:: Note that there is no error handling.  Nonsensical
193
       options will be silently ignored by exiftool, so there's not
194
       much that can be done in that regard.  You should avoid passing
195
       non-existent files to any of the methods, since this will lead
196
       to undefied behaviour.
197
198
    .. py:attribute:: running
199
200
       A Boolean value indicating whether this instance is currently
201
       associated with a running subprocess.
202
    """
203
204
    def __init__(self, executable_=None, addedargs=None):
205
        
206
        if executable_ is None:
207
            self.executable = executable
208
        else:
209
            self.executable = executable_
210
211
        if addedargs is None:
212
            self.addedargs = []
213
        elif type(addedargs) is list:
214
            self.addedargs = addedargs
215
        else:
216
            raise TypeError("addedargs not a list of strings")
217
        
218
        self.running = False
219
220
    def start(self):
221
        """Start an ``exiftool`` process in batch mode for this instance.
222
223
        This method will issue a ``UserWarning`` if the subprocess is
224
        already running.  The process is started with the ``-G`` and
225
        ``-n`` as common arguments, which are automatically included
226
        in every command you run with :py:meth:`execute()`.
227
        """
228
        if self.running:
229
            warnings.warn("ExifTool already running; doing nothing.")
230
            return
231
        with open(os.devnull, "w") as devnull:
232
            procargs = [self.executable, "-stay_open", "True",  "-@", "-",
233
                 "-common_args", "-G", "-n"];
234
            procargs.extend(self.addedargs)
235
            logging.debug(procargs) 
236
            self._process = subprocess.Popen(
237
                procargs,
238
                stdin=subprocess.PIPE, stdout=subprocess.PIPE,
239
                stderr=devnull)
240
        self.running = True
241
242
    def terminate(self):
243
        """Terminate the ``exiftool`` process of this instance.
244
245
        If the subprocess isn't running, this method will do nothing.
246
        """
247
        if not self.running:
248
            return
249
        self._process.stdin.write(b"-stay_open\nFalse\n")
250
        self._process.stdin.flush()
251
        self._process.communicate()
252
        del self._process
253
        self.running = False
254
255
    def __enter__(self):
256
        self.start()
257
        return self
258
259
    def __exit__(self, exc_type, exc_val, exc_tb):
260
        self.terminate()
261
262
    def __del__(self):
263
        self.terminate()
264
265
    def execute(self, *params):
266
        """Execute the given batch of parameters with ``exiftool``.
267
268
        This method accepts any number of parameters and sends them to
269
        the attached ``exiftool`` process.  The process must be
270
        running, otherwise ``ValueError`` is raised.  The final
271
        ``-execute`` necessary to actually run the batch is appended
272
        automatically; see the documentation of :py:meth:`start()` for
273
        the common options.  The ``exiftool`` output is read up to the
274
        end-of-output sentinel and returned as a raw ``bytes`` object,
275
        excluding the sentinel.
276
277
        The parameters must also be raw ``bytes``, in whatever
278
        encoding exiftool accepts.  For filenames, this should be the
279
        system's filesystem encoding.
280
281
        .. note:: This is considered a low-level method, and should
282
           rarely be needed by application developers.
283
        """
284
        if not self.running:
285
            raise ValueError("ExifTool instance not running.")
286
        self._process.stdin.write(b"\n".join(params + (b"-execute\n",)))
287
        self._process.stdin.flush()
288
        output = b""
289
        fd = self._process.stdout.fileno()
290
        while not output[-32:].strip().endswith(sentinel):
291
            output += os.read(fd, block_size)
292
        return output.strip()[:-len(sentinel)]
293
294
    def execute_json(self, *params):
295
        """Execute the given batch of parameters and parse the JSON output.
296
297
        This method is similar to :py:meth:`execute()`.  It
298
        automatically adds the parameter ``-j`` to request JSON output
299
        from ``exiftool`` and parses the output.  The return value is
300
        a list of dictionaries, mapping tag names to the corresponding
301
        values.  All keys are Unicode strings with the tag names
302
        including the ExifTool group name in the format <group>:<tag>.
303
        The values can have multiple types.  All strings occurring as
304
        values will be Unicode strings.  Each dictionary contains the
305
        name of the file it corresponds to in the key ``"SourceFile"``.
306
307
        The parameters to this function must be either raw strings
308
        (type ``str`` in Python 2.x, type ``bytes`` in Python 3.x) or
309
        Unicode strings (type ``unicode`` in Python 2.x, type ``str``
310
        in Python 3.x).  Unicode strings will be encoded using
311
        system's filesystem encoding.  This behaviour means you can
312
        pass in filenames according to the convention of the
313
        respective Python version – as raw strings in Python 2.x and
314
        as Unicode strings in Python 3.x.
315
        """
316
        params = map(fsencode, params)
317
        # Some latin bytes won't decode to utf-8.
318
        # Try utf-8 and fallback to latin.
319
        # http://stackoverflow.com/a/5552623/1318758
320
        # https://github.com/jmathai/elodie/issues/127
321
        try:
322
            return json.loads(self.execute(b"-j", *params).decode("utf-8"))
323
        except UnicodeDecodeError as e:
324
            return json.loads(self.execute(b"-j", *params).decode("latin-1"))
325
326
    def get_metadata_batch(self, filenames):
327
        """Return all meta-data for the given files.
328
329
        The return value will have the format described in the
330
        documentation of :py:meth:`execute_json()`.
331
        """
332
        return self.execute_json(*filenames)
333
334
    def get_metadata(self, filename):
335
        """Return meta-data for a single file.
336
337
        The returned dictionary has the format described in the
338
        documentation of :py:meth:`execute_json()`.
339
        """
340
        return self.execute_json(filename)[0]
341
342
    def get_tags_batch(self, tags, filenames):
343
        """Return only specified tags for the given files.
344
345
        The first argument is an iterable of tags.  The tag names may
346
        include group names, as usual in the format <group>:<tag>.
347
348
        The second argument is an iterable of file names.
349
350
        The format of the return value is the same as for
351
        :py:meth:`execute_json()`.
352
        """
353
        # Explicitly ruling out strings here because passing in a
354
        # string would lead to strange and hard-to-find errors
355
        if isinstance(tags, basestring):
356
            raise TypeError("The argument 'tags' must be "
357
                            "an iterable of strings")
358
        if isinstance(filenames, basestring):
359
            raise TypeError("The argument 'filenames' must be "
360
                            "an iterable of strings")
361
        params = ["-" + t for t in tags]
362
        params.extend(filenames)
363
        return self.execute_json(*params)
364
365
    def get_tags(self, tags, filename):
366
        """Return only specified tags for a single file.
367
368
        The returned dictionary has the format described in the
369
        documentation of :py:meth:`execute_json()`.
370
        """
371
        return self.get_tags_batch(tags, [filename])[0]
372
373
    def get_tag_batch(self, tag, filenames):
374
        """Extract a single tag from the given files.
375
376
        The first argument is a single tag name, as usual in the
377
        format <group>:<tag>.
378
379
        The second argument is an iterable of file names.
380
381
        The return value is a list of tag values or ``None`` for
382
        non-existent tags, in the same order as ``filenames``.
383
        """
384
        data = self.get_tags_batch([tag], filenames)
385
        result = []
386
        for d in data:
387
            d.pop("SourceFile")
388
            result.append(next(iter(d.values()), None))
389
        return result
390
391
    def get_tag(self, tag, filename):
392
        """Extract a single tag from a single file.
393
394
        The return value is the value of the specified tag, or
395
        ``None`` if this tag was not found in the file.
396
        """
397
        return self.get_tag_batch(tag, [filename])[0]
398
399
    def set_tags_batch(self, tags, filenames):
400
        """Writes the values of the specified tags for the given files.
401
402
        The first argument is a dictionary of tags and values.  The tag names may
403
        include group names, as usual in the format <group>:<tag>.
404
405
        The second argument is an iterable of file names.
406
407
        The format of the return value is the same as for
408
        :py:meth:`execute()`.
409
        
410
        It can be passed into `check_ok()` and `format_error()`.
411
        """
412
        # Explicitly ruling out strings here because passing in a
413
        # string would lead to strange and hard-to-find errors
414
        if isinstance(tags, basestring):
415
            raise TypeError("The argument 'tags' must be dictionary "
416
                            "of strings")
417
        if isinstance(filenames, basestring):
418
            raise TypeError("The argument 'filenames' must be "
419
                            "an iterable of strings")
420
                
421
        params = []
422
        params_utf8 = []
423
        for tag, value in tags.items():
424
            params.append(u'-%s=%s' % (tag, value))
425
            
426
        params.extend(filenames)
427
        params_utf8 = [x.encode('utf-8') for x in params]
428
        return self.execute(*params_utf8)
429
430
    def set_tags(self, tags, filename):
431
        """Writes the values of the specified tags for the given file.
432
433
        This is a convenience function derived from `set_tags_batch()`.
434
        Only difference is that it takes as last arugemnt only one file name
435
        as a string. 
436
        """
437
        return self.set_tags_batch(tags, [filename])
438
    
439
    def set_keywords_batch(self, mode, keywords, filenames):
440
        """Modifies the keywords tag for the given files.
441
442
        The first argument is the operation mode:
443
        KW_REPLACE: Replace (i.e. set) the full keywords tag with `keywords`.
444
        KW_ADD:     Add `keywords` to the keywords tag. 
445
                    If a keyword is present, just keep it.
446
        KW_REMOVE:  Remove `keywords` from the keywords tag. 
447
                    If a keyword wasn't present, just leave it.
448
449
        The second argument is an iterable of key words.    
450
451
        The third argument is an iterable of file names.
452
453
        The format of the return value is the same as for
454
        :py:meth:`execute()`.
455
        
456
        It can be passed into `check_ok()` and `format_error()`.
457
        """
458
        # Explicitly ruling out strings here because passing in a
459
        # string would lead to strange and hard-to-find errors
460
        if isinstance(keywords, basestring):
461
            raise TypeError("The argument 'keywords' must be "
462
                            "an iterable of strings")
463
        if isinstance(filenames, basestring):
464
            raise TypeError("The argument 'filenames' must be "
465
                            "an iterable of strings")
466
                
467
        params = []    
468
            
469
        kw_operation = {KW_REPLACE:"-%s=%s",
470
                        KW_ADD:"-%s+=%s",
471
                        KW_REMOVE:"-%s-=%s"}[mode]
472
473
        kw_params = [ kw_operation % (KW_TAGNAME, w)  for w in keywords ]
474
        
475
        params.extend(kw_params)            
476
        params.extend(filenames)
477
        logging.debug (params)
478
        return self.execute(*params)
479
    
480
    def set_keywords(self, mode, keywords, filename):
481
        """Modifies the keywords tag for the given file.
482
483
        This is a convenience function derived from `set_keywords_batch()`.
484
        Only difference is that it takes as last argument only one file name
485
        as a string. 
486
        """
487
        return self.set_keywords_batch(mode, keywords, [filename])
488