python/mozbuild/mozpack/mozjar.py@925c144e1f1f
python/mozbuild/mozpack/mozjar.py
Fri, 16 Jan 2015 18:13:44 +0100
- author
- Michael Schloh von Bennewitz <michael@schloh.com>
- date
- Fri, 16 Jan 2015 18:13:44 +0100
- branch
- TOR_BUG_9701
- changeset 14
- 925c144e1f1f
- permissions
- -rw-r--r--
Integrate suggestion from review to improve consistency with existing code.
1 # This Source Code Form is subject to the terms of the Mozilla Public
2 # License, v. 2.0. If a copy of the MPL was not distributed with this
3 # file, You can obtain one at http://mozilla.org/MPL/2.0/.
5 from io import BytesIO
6 import struct
7 import zlib
8 import os
9 from zipfile import (
10 ZIP_STORED,
11 ZIP_DEFLATED,
12 )
13 from collections import OrderedDict
14 from urlparse import urlparse, ParseResult
15 import mozpack.path
17 JAR_STORED = ZIP_STORED
18 JAR_DEFLATED = ZIP_DEFLATED
19 MAX_WBITS = 15
22 class JarReaderError(Exception):
23 '''Error type for Jar reader errors.'''
26 class JarWriterError(Exception):
27 '''Error type for Jar writer errors.'''
30 class JarStruct(object):
31 '''
32 Helper used to define ZIP archive raw data structures. Data structures
33 handled by this helper all start with a magic number, defined in
34 subclasses MAGIC field as a 32-bits unsigned integer, followed by data
35 structured as described in subclasses STRUCT field.
37 The STRUCT field contains a list of (name, type) pairs where name is a
38 field name, and the type can be one of 'uint32', 'uint16' or one of the
39 field names. In the latter case, the field is considered to be a string
40 buffer with a length given in that field.
41 For example,
42 STRUCT = [
43 ('version', 'uint32'),
44 ('filename_size', 'uint16'),
45 ('filename', 'filename_size')
46 ]
47 describes a structure with a 'version' 32-bits unsigned integer field,
48 followed by a 'filename_size' 16-bits unsigned integer field, followed by a
49 filename_size-long string buffer 'filename'.
51 Fields that are used as other fields size are not stored in objects. In the
52 above example, an instance of such subclass would only have two attributes:
53 obj['version']
54 obj['filename']
55 filename_size would be obtained with len(obj['filename']).
57 JarStruct subclasses instances can be either initialized from existing data
58 (deserialized), or with empty fields.
59 '''
61 TYPE_MAPPING = {'uint32': ('I', 4), 'uint16': ('H', 2)}
63 def __init__(self, data=None):
64 '''
65 Create an instance from the given data. Data may be omitted to create
66 an instance with empty fields.
67 '''
68 assert self.MAGIC and isinstance(self.STRUCT, OrderedDict)
69 self.size_fields = set(t for t in self.STRUCT.itervalues()
70 if not t in JarStruct.TYPE_MAPPING)
71 self._values = {}
72 if data:
73 self._init_data(data)
74 else:
75 self._init_empty()
77 def _init_data(self, data):
78 '''
79 Initialize an instance from data, following the data structure
80 described in self.STRUCT. The self.MAGIC signature is expected at
81 data[:4].
82 '''
83 assert data is not None
84 self.signature, size = JarStruct.get_data('uint32', data)
85 if self.signature != self.MAGIC:
86 raise JarReaderError('Bad magic')
87 offset = size
88 # For all fields used as other fields sizes, keep track of their value
89 # separately.
90 sizes = dict((t, 0) for t in self.size_fields)
91 for name, t in self.STRUCT.iteritems():
92 if t in JarStruct.TYPE_MAPPING:
93 value, size = JarStruct.get_data(t, data[offset:])
94 else:
95 size = sizes[t]
96 value = data[offset:offset + size]
97 if isinstance(value, memoryview):
98 value = value.tobytes()
99 if not name in sizes:
100 self._values[name] = value
101 else:
102 sizes[name] = value
103 offset += size
105 def _init_empty(self):
106 '''
107 Initialize an instance with empty fields.
108 '''
109 self.signature = self.MAGIC
110 for name, t in self.STRUCT.iteritems():
111 if name in self.size_fields:
112 continue
113 self._values[name] = 0 if t in JarStruct.TYPE_MAPPING else ''
115 @staticmethod
116 def get_data(type, data):
117 '''
118 Deserialize a single field of given type (must be one of
119 JarStruct.TYPE_MAPPING) at the given offset in the given data.
120 '''
121 assert type in JarStruct.TYPE_MAPPING
122 assert data is not None
123 format, size = JarStruct.TYPE_MAPPING[type]
124 data = data[:size]
125 if isinstance(data, memoryview):
126 data = data.tobytes()
127 return struct.unpack('<' + format, data)[0], size
129 def serialize(self):
130 '''
131 Serialize the data structure according to the data structure definition
132 from self.STRUCT.
133 '''
134 serialized = struct.pack('<I', self.signature)
135 sizes = dict((t, name) for name, t in self.STRUCT.iteritems()
136 if not t in JarStruct.TYPE_MAPPING)
137 for name, t in self.STRUCT.iteritems():
138 if t in JarStruct.TYPE_MAPPING:
139 format, size = JarStruct.TYPE_MAPPING[t]
140 if name in sizes:
141 value = len(self[sizes[name]])
142 else:
143 value = self[name]
144 serialized += struct.pack('<' + format, value)
145 else:
146 serialized += self[name]
147 return serialized
149 @property
150 def size(self):
151 '''
152 Return the size of the data structure, given the current values of all
153 variable length fields.
154 '''
155 size = JarStruct.TYPE_MAPPING['uint32'][1]
156 for name, type in self.STRUCT.iteritems():
157 if type in JarStruct.TYPE_MAPPING:
158 size += JarStruct.TYPE_MAPPING[type][1]
159 else:
160 size += len(self[name])
161 return size
163 def __getitem__(self, key):
164 return self._values[key]
166 def __setitem__(self, key, value):
167 if not key in self.STRUCT:
168 raise KeyError(key)
169 if key in self.size_fields:
170 raise AttributeError("can't set attribute")
171 self._values[key] = value
173 def __contains__(self, key):
174 return key in self._values
176 def __iter__(self):
177 return self._values.iteritems()
179 def __repr__(self):
180 return "<%s %s>" % (self.__class__.__name__,
181 ' '.join('%s=%s' % (n, v) for n, v in self))
184 class JarCdirEnd(JarStruct):
185 '''
186 End of central directory record.
187 '''
188 MAGIC = 0x06054b50
189 STRUCT = OrderedDict([
190 ('disk_num', 'uint16'),
191 ('cdir_disk', 'uint16'),
192 ('disk_entries', 'uint16'),
193 ('cdir_entries', 'uint16'),
194 ('cdir_size', 'uint32'),
195 ('cdir_offset', 'uint32'),
196 ('comment_size', 'uint16'),
197 ('comment', 'comment_size'),
198 ])
200 CDIR_END_SIZE = JarCdirEnd().size
203 class JarCdirEntry(JarStruct):
204 '''
205 Central directory file header
206 '''
207 MAGIC = 0x02014b50
208 STRUCT = OrderedDict([
209 ('creator_version', 'uint16'),
210 ('min_version', 'uint16'),
211 ('general_flag', 'uint16'),
212 ('compression', 'uint16'),
213 ('lastmod_time', 'uint16'),
214 ('lastmod_date', 'uint16'),
215 ('crc32', 'uint32'),
216 ('compressed_size', 'uint32'),
217 ('uncompressed_size', 'uint32'),
218 ('filename_size', 'uint16'),
219 ('extrafield_size', 'uint16'),
220 ('filecomment_size', 'uint16'),
221 ('disknum', 'uint16'),
222 ('internal_attr', 'uint16'),
223 ('external_attr', 'uint32'),
224 ('offset', 'uint32'),
225 ('filename', 'filename_size'),
226 ('extrafield', 'extrafield_size'),
227 ('filecomment', 'filecomment_size'),
228 ])
231 class JarLocalFileHeader(JarStruct):
232 '''
233 Local file header
234 '''
235 MAGIC = 0x04034b50
236 STRUCT = OrderedDict([
237 ('min_version', 'uint16'),
238 ('general_flag', 'uint16'),
239 ('compression', 'uint16'),
240 ('lastmod_time', 'uint16'),
241 ('lastmod_date', 'uint16'),
242 ('crc32', 'uint32'),
243 ('compressed_size', 'uint32'),
244 ('uncompressed_size', 'uint32'),
245 ('filename_size', 'uint16'),
246 ('extra_field_size', 'uint16'),
247 ('filename', 'filename_size'),
248 ('extra_field', 'extra_field_size'),
249 ])
252 class JarFileReader(object):
253 '''
254 File-like class for use by JarReader to give access to individual files
255 within a Jar archive.
256 '''
257 def __init__(self, header, data):
258 '''
259 Initialize a JarFileReader. header is the local file header
260 corresponding to the file in the jar archive, data a buffer containing
261 the file data.
262 '''
263 assert header['compression'] in [JAR_DEFLATED, JAR_STORED]
264 self._data = data
265 # Copy some local file header fields.
266 for name in ['filename', 'compressed_size',
267 'uncompressed_size', 'crc32']:
268 setattr(self, name, header[name])
269 self.compressed = header['compression'] == JAR_DEFLATED
271 def read(self, length=-1):
272 '''
273 Read some amount of uncompressed data.
274 '''
275 return self.uncompressed_data.read(length)
277 def readlines(self):
278 '''
279 Return a list containing all the lines of data in the uncompressed
280 data.
281 '''
282 return self.read().splitlines(True)
284 def __iter__(self):
285 '''
286 Iterator, to support the "for line in fileobj" constructs.
287 '''
288 return iter(self.readlines())
290 def seek(self, pos, whence=os.SEEK_SET):
291 '''
292 Change the current position in the uncompressed data. Subsequent reads
293 will start from there.
294 '''
295 return self.uncompressed_data.seek(pos, whence)
297 def close(self):
298 '''
299 Free the uncompressed data buffer.
300 '''
301 self.uncompressed_data.close()
303 @property
304 def compressed_data(self):
305 '''
306 Return the raw compressed data.
307 '''
308 return self._data[:self.compressed_size]
310 @property
311 def uncompressed_data(self):
312 '''
313 Return the uncompressed data.
314 '''
315 if hasattr(self, '_uncompressed_data'):
316 return self._uncompressed_data
317 data = self.compressed_data
318 if self.compressed:
319 data = zlib.decompress(data.tobytes(), -MAX_WBITS)
320 else:
321 data = data.tobytes()
322 if len(data) != self.uncompressed_size:
323 raise JarReaderError('Corrupted file? %s' % self.filename)
324 self._uncompressed_data = BytesIO(data)
325 return self._uncompressed_data
328 class JarReader(object):
329 '''
330 Class with methods to read Jar files. Can open standard jar files as well
331 as Mozilla jar files (see further details in the JarWriter documentation).
332 '''
333 def __init__(self, file=None, fileobj=None):
334 '''
335 Opens the given file as a Jar archive. Use the given file-like object
336 if one is given instead of opening the given file name.
337 '''
338 if fileobj:
339 data = fileobj.read()
340 else:
341 data = open(file, 'rb').read()
342 self._data = memoryview(data)
343 # The End of Central Directory Record has a variable size because of
344 # comments it may contain, so scan for it from the end of the file.
345 offset = -CDIR_END_SIZE
346 while True:
347 signature = JarStruct.get_data('uint32', self._data[offset:])[0]
348 if signature == JarCdirEnd.MAGIC:
349 break
350 if offset == -len(self._data):
351 raise JarReaderError('Not a jar?')
352 offset -= 1
353 self._cdir_end = JarCdirEnd(self._data[offset:])
355 def close(self):
356 '''
357 Free some resources associated with the Jar.
358 '''
359 del self._data
361 @property
362 def entries(self):
363 '''
364 Return an ordered dict of central directory entries, indexed by
365 filename, in the order they appear in the Jar archive central
366 directory. Directory entries are skipped.
367 '''
368 if hasattr(self, '_entries'):
369 return self._entries
370 preload = 0
371 if self.is_optimized:
372 preload = JarStruct.get_data('uint32', self._data)[0]
373 entries = OrderedDict()
374 offset = self._cdir_end['cdir_offset']
375 for e in xrange(self._cdir_end['cdir_entries']):
376 entry = JarCdirEntry(self._data[offset:])
377 offset += entry.size
378 # Creator host system. 0 is MSDOS, 3 is Unix
379 host = entry['creator_version'] >> 8
380 # External attributes values depend on host above. On Unix the
381 # higher bits are the stat.st_mode value. On MSDOS, the lower bits
382 # are the FAT attributes.
383 xattr = entry['external_attr']
384 # Skip directories
385 if (host == 0 and xattr & 0x10) or (host == 3 and
386 xattr & (040000 << 16)):
387 continue
388 entries[entry['filename']] = entry
389 if entry['offset'] < preload:
390 self._last_preloaded = entry['filename']
391 self._entries = entries
392 return entries
394 @property
395 def is_optimized(self):
396 '''
397 Return whether the jar archive is optimized.
398 '''
399 # In optimized jars, the central directory is at the beginning of the
400 # file, after a single 32-bits value, which is the length of data
401 # preloaded.
402 return self._cdir_end['cdir_offset'] == \
403 JarStruct.TYPE_MAPPING['uint32'][1]
405 @property
406 def last_preloaded(self):
407 '''
408 Return the name of the last file that is set to be preloaded.
409 See JarWriter documentation for more details on preloading.
410 '''
411 if hasattr(self, '_last_preloaded'):
412 return self._last_preloaded
413 self._last_preloaded = None
414 self.entries
415 return self._last_preloaded
417 def _getreader(self, entry):
418 '''
419 Helper to create a JarFileReader corresponding to the given central
420 directory entry.
421 '''
422 header = JarLocalFileHeader(self._data[entry['offset']:])
423 for key, value in entry:
424 if key in header and header[key] != value:
425 raise JarReaderError('Central directory and file header ' +
426 'mismatch. Corrupted archive?')
427 return JarFileReader(header,
428 self._data[entry['offset'] + header.size:])
430 def __iter__(self):
431 '''
432 Iterate over all files in the Jar archive, in the form of
433 JarFileReaders.
434 for file in jarReader:
435 ...
436 '''
437 for entry in self.entries.itervalues():
438 yield self._getreader(entry)
440 def __getitem__(self, name):
441 '''
442 Get a JarFileReader for the given file name.
443 '''
444 return self._getreader(self.entries[name])
446 def __contains__(self, name):
447 '''
448 Return whether the given file name appears in the Jar archive.
449 '''
450 return name in self.entries
453 class JarWriter(object):
454 '''
455 Class with methods to write Jar files. Can write more-or-less standard jar
456 archives as well as jar archives optimized for Gecko. See the documentation
457 for the close() member function for a description of both layouts.
458 '''
459 def __init__(self, file=None, fileobj=None, compress=True, optimize=True):
460 '''
461 Initialize a Jar archive in the given file. Use the given file-like
462 object if one is given instead of opening the given file name.
463 The compress option determines the default behavior for storing data
464 in the jar archive. The optimize options determines whether the jar
465 archive should be optimized for Gecko or not.
466 '''
467 if fileobj:
468 self._data = fileobj
469 else:
470 self._data = open(file, 'wb')
471 self._compress = compress
472 self._contents = OrderedDict()
473 self._last_preloaded = None
474 self._optimize = optimize
476 def __enter__(self):
477 '''
478 Context manager __enter__ method for JarWriter.
479 '''
480 return self
482 def __exit__(self, type, value, tb):
483 '''
484 Context manager __exit__ method for JarWriter.
485 '''
486 self.finish()
488 def finish(self):
489 '''
490 Flush and close the Jar archive.
492 Standard jar archives are laid out like the following:
493 - Local file header 1
494 - File data 1
495 - Local file header 2
496 - File data 2
497 - (...)
498 - Central directory entry pointing at Local file header 1
499 - Central directory entry pointing at Local file header 2
500 - (...)
501 - End of central directory, pointing at first central directory
502 entry.
504 Jar archives optimized for Gecko are laid out like the following:
505 - 32-bits unsigned integer giving the amount of data to preload.
506 - Central directory entry pointing at Local file header 1
507 - Central directory entry pointing at Local file header 2
508 - (...)
509 - End of central directory, pointing at first central directory
510 entry.
511 - Local file header 1
512 - File data 1
513 - Local file header 2
514 - File data 2
515 - (...)
516 - End of central directory, pointing at first central directory
517 entry.
518 The duplication of the End of central directory is to accomodate some
519 Zip reading tools that want an end of central directory structure to
520 follow the central directory entries.
521 '''
522 offset = 0
523 headers = {}
524 preload_size = 0
525 # Prepare central directory entries
526 for entry, content in self._contents.itervalues():
527 header = JarLocalFileHeader()
528 for name in entry.STRUCT:
529 if name in header:
530 header[name] = entry[name]
531 entry['offset'] = offset
532 offset += len(content) + header.size
533 if entry['filename'] == self._last_preloaded:
534 preload_size = offset
535 headers[entry] = header
536 # Prepare end of central directory
537 end = JarCdirEnd()
538 end['disk_entries'] = len(self._contents)
539 end['cdir_entries'] = end['disk_entries']
540 end['cdir_size'] = reduce(lambda x, y: x + y[0].size,
541 self._contents.values(), 0)
542 # On optimized archives, store the preloaded size and the central
543 # directory entries, followed by the first end of central directory.
544 if self._optimize:
545 end['cdir_offset'] = 4
546 offset = end['cdir_size'] + end['cdir_offset'] + end.size
547 if preload_size:
548 preload_size += offset
549 self._data.write(struct.pack('<I', preload_size))
550 for entry, _ in self._contents.itervalues():
551 entry['offset'] += offset
552 self._data.write(entry.serialize())
553 self._data.write(end.serialize())
554 # Store local file entries followed by compressed data
555 for entry, content in self._contents.itervalues():
556 self._data.write(headers[entry].serialize())
557 self._data.write(content)
558 # On non optimized archives, store the central directory entries.
559 if not self._optimize:
560 end['cdir_offset'] = offset
561 for entry, _ in self._contents.itervalues():
562 self._data.write(entry.serialize())
563 # Store the end of central directory.
564 self._data.write(end.serialize())
565 self._data.close()
567 def add(self, name, data, compress=None, mode=None):
568 '''
569 Add a new member to the jar archive, with the given name and the given
570 data.
571 The compress option indicates if the given data should be compressed
572 (True), not compressed (False), or compressed according to the default
573 defined when creating the JarWriter (None).
574 When the data should be compressed (True or None with self.compress ==
575 True), it is only really compressed if the compressed size is smaller
576 than the uncompressed size.
577 The mode option gives the unix permissions that should be stored
578 for the jar entry.
579 The given data may be a buffer, a file-like instance, a Deflater or a
580 JarFileReader instance. The latter two allow to avoid uncompressing
581 data to recompress it.
582 '''
583 if name in self._contents:
584 raise JarWriterError("File %s already in JarWriter" % name)
585 if compress is None:
586 compress = self._compress
587 if (isinstance(data, JarFileReader) and data.compressed == compress) \
588 or (isinstance(data, Deflater) and data.compress == compress):
589 deflater = data
590 else:
591 deflater = Deflater(compress)
592 if isinstance(data, basestring):
593 deflater.write(data)
594 elif hasattr(data, 'read'):
595 data.seek(0)
596 deflater.write(data.read())
597 else:
598 raise JarWriterError("Don't know how to handle %s" %
599 type(data))
600 # Fill a central directory entry for this new member.
601 entry = JarCdirEntry()
602 entry['creator_version'] = 20
603 if mode is not None:
604 # Set creator host system (upper byte of creator_version)
605 # to 3 (Unix) so mode is honored when there is one.
606 entry['creator_version'] |= 3 << 8
607 entry['external_attr'] = (mode & 0xFFFF) << 16L
608 if deflater.compressed:
609 entry['min_version'] = 20 # Version 2.0 supports deflated streams
610 entry['general_flag'] = 2 # Max compression
611 entry['compression'] = JAR_DEFLATED
612 else:
613 entry['min_version'] = 10 # Version 1.0 for stored streams
614 entry['general_flag'] = 0
615 entry['compression'] = JAR_STORED
616 # January 1st, 2010. See bug 592369.
617 entry['lastmod_date'] = ((2010 - 1980) << 9) | (1 << 5) | 1
618 entry['lastmod_time'] = 0
619 entry['crc32'] = deflater.crc32
620 entry['compressed_size'] = deflater.compressed_size
621 entry['uncompressed_size'] = deflater.uncompressed_size
622 entry['filename'] = name
623 self._contents[name] = entry, deflater.compressed_data
625 def preload(self, files):
626 '''
627 Set which members of the jar archive should be preloaded when opening
628 the archive in Gecko. This reorders the members according to the order
629 of given list.
630 '''
631 new_contents = OrderedDict()
632 for f in files:
633 if not f in self._contents:
634 continue
635 new_contents[f] = self._contents[f]
636 self._last_preloaded = f
637 for f in self._contents:
638 if not f in new_contents:
639 new_contents[f] = self._contents[f]
640 self._contents = new_contents
643 class Deflater(object):
644 '''
645 File-like interface to zlib compression. The data is actually not
646 compressed unless the compressed form is smaller than the uncompressed
647 data.
648 '''
649 def __init__(self, compress=True):
650 '''
651 Initialize a Deflater. The compress argument determines whether to
652 try to compress at all.
653 '''
654 self._data = BytesIO()
655 self.compress = compress
656 if compress:
657 self._deflater = zlib.compressobj(9, zlib.DEFLATED, -MAX_WBITS)
658 self._deflated = BytesIO()
659 else:
660 self._deflater = None
662 def write(self, data):
663 '''
664 Append a buffer to the Deflater.
665 '''
666 self._data.write(data)
667 if self.compress:
668 if self._deflater:
669 if isinstance(data, memoryview):
670 data = data.tobytes()
671 self._deflated.write(self._deflater.compress(data))
672 else:
673 raise JarWriterError("Can't write after flush")
675 def close(self):
676 '''
677 Close the Deflater.
678 '''
679 self._data.close()
680 if self.compress:
681 self._deflated.close()
683 def _flush(self):
684 '''
685 Flush the underlying zlib compression object.
686 '''
687 if self.compress and self._deflater:
688 self._deflated.write(self._deflater.flush())
689 self._deflater = None
691 @property
692 def compressed(self):
693 '''
694 Return whether the data should be compressed.
695 '''
696 return self._compressed_size < self.uncompressed_size
698 @property
699 def _compressed_size(self):
700 '''
701 Return the real compressed size of the data written to the Deflater. If
702 the Deflater is set not to compress, the uncompressed size is returned.
703 Otherwise, the actual compressed size is returned, whether or not it is
704 a win over the uncompressed size.
705 '''
706 if self.compress:
707 self._flush()
708 return self._deflated.tell()
709 return self.uncompressed_size
711 @property
712 def compressed_size(self):
713 '''
714 Return the compressed size of the data written to the Deflater. If the
715 Deflater is set not to compress, the uncompressed size is returned.
716 Otherwise, if the data should not be compressed (the real compressed
717 size is bigger than the uncompressed size), return the uncompressed
718 size.
719 '''
720 if self.compressed:
721 return self._compressed_size
722 return self.uncompressed_size
724 @property
725 def uncompressed_size(self):
726 '''
727 Return the size of the data written to the Deflater.
728 '''
729 return self._data.tell()
731 @property
732 def crc32(self):
733 '''
734 Return the crc32 of the data written to the Deflater.
735 '''
736 return zlib.crc32(self._data.getvalue()) & 0xffffffff
738 @property
739 def compressed_data(self):
740 '''
741 Return the compressed data, if the data should be compressed (real
742 compressed size smaller than the uncompressed size), or the
743 uncompressed data otherwise.
744 '''
745 if self.compressed:
746 return self._deflated.getvalue()
747 return self._data.getvalue()
750 class JarLog(dict):
751 '''
752 Helper to read the file Gecko generates when setting MOZ_JAR_LOG_FILE.
753 The jar log is then available as a dict with the jar path as key (see
754 canonicalize for more details on the key value), and the corresponding
755 access log as a list value. Only the first access to a given member of
756 a jar is stored.
757 '''
758 def __init__(self, file=None, fileobj=None):
759 if not fileobj:
760 fileobj = open(file, 'r')
761 urlmap = {}
762 for line in fileobj:
763 url, path = line.strip().split(None, 1)
764 if not url or not path:
765 continue
766 if url not in urlmap:
767 urlmap[url] = JarLog.canonicalize(url)
768 jar = urlmap[url]
769 entry = self.setdefault(jar, [])
770 if path not in entry:
771 entry.append(path)
773 @staticmethod
774 def canonicalize(url):
775 '''
776 The jar path is stored in a MOZ_JAR_LOG_FILE log as a url. This method
777 returns a unique value corresponding to such urls.
778 - file:///{path} becomes {path}
779 - jar:file:///{path}!/{subpath} becomes ({path}, {subpath})
780 - jar:jar:file:///{path}!/{subpath}!/{subpath2} becomes
781 ({path}, {subpath}, {subpath2})
782 '''
783 if not isinstance(url, ParseResult):
784 # Assume that if it doesn't start with jar: or file:, it's a path.
785 if not url.startswith(('jar:', 'file:')):
786 url = 'file:///' + os.path.abspath(url)
787 url = urlparse(url)
788 assert url.scheme
789 assert url.scheme in ('jar', 'file')
790 if url.scheme == 'jar':
791 path = JarLog.canonicalize(url.path)
792 if isinstance(path, tuple):
793 return path[:-1] + tuple(path[-1].split('!/', 1))
794 return tuple(path.split('!/', 1))
795 if url.scheme == 'file':
796 assert os.path.isabs(url.path)
797 path = url.path
798 # On Windows, url.path will be /drive:/path ; on Unix systems,
799 # /path. As we want drive:/path instead of /drive:/path on Windows,
800 # remove the leading /.
801 if os.path.isabs(path[1:]):
802 path = path[1:]
803 path = os.path.realpath(path)
804 return mozpack.path.normsep(os.path.normcase(path))
