|
10 | 10 |
|
11 | 11 |
|
12 | 12 | class ZipFile(zipfile.ZipFile):
|
13 |
| - """ Class with methods to open, read, write, close, list zip files. |
14 |
| -
|
15 |
| - Subclassing zipfile.ZipFile to allow for overwriting of existing |
16 |
| - entries, though only for writestr, not for write. |
17 |
| - """ |
18 |
| - def __init__(self, file, mode="r", compression=zipfile.ZIP_STORED, |
19 |
| - lock = False): |
20 |
| - if lock: |
21 |
| - assert isinstance(file, basestring) |
22 |
| - self.lockfile = lock_file(file + '.lck') |
23 |
| - else: |
24 |
| - self.lockfile = None |
25 |
| - |
26 |
| - if mode == 'a' and lock: |
27 |
| - # appending to a file which doesn't exist fails, but we can't check |
28 |
| - # existence util we hold the lock |
29 |
| - if (not os.path.isfile(file)) or os.path.getsize(file) == 0: |
30 |
| - mode = 'w' |
31 |
| - |
32 |
| - zipfile.ZipFile.__init__(self, file, mode, compression) |
33 |
| - self._remove = [] |
34 |
| - self.end = self.fp.tell() |
35 |
| - self.debug = 0 |
36 |
| - |
37 |
| - def writestr(self, zinfo_or_arcname, bytes): |
38 |
| - """Write contents into the archive. |
39 |
| -
|
40 |
| - The contents is the argument 'bytes', 'zinfo_or_arcname' is either |
41 |
| - a ZipInfo instance or the name of the file in the archive. |
42 |
| - This method is overloaded to allow overwriting existing entries. |
| 13 | + """ Class with methods to open, read, write, close, list zip files. |
| 14 | +
|
| 15 | + Subclassing zipfile.ZipFile to allow for overwriting of existing |
| 16 | + entries, though only for writestr, not for write. |
43 | 17 | """
|
44 |
| - if not isinstance(zinfo_or_arcname, zipfile.ZipInfo): |
45 |
| - zinfo = zipfile.ZipInfo(filename=zinfo_or_arcname, |
46 |
| - date_time=time.localtime(time.time())) |
47 |
| - zinfo.compress_type = self.compression |
48 |
| - # Add some standard UNIX file access permissions (-rw-r--r--). |
49 |
| - zinfo.external_attr = (0x81a4 & 0xFFFF) << 16L |
50 |
| - else: |
51 |
| - zinfo = zinfo_or_arcname |
52 |
| - |
53 |
| - # Now to the point why we overwrote this in the first place, |
54 |
| - # remember the entry numbers if we already had this entry. |
55 |
| - # Optimizations: |
56 |
| - # If the entry to overwrite is the last one, just reuse that. |
57 |
| - # If we store uncompressed and the new content has the same size |
58 |
| - # as the old, reuse the existing entry. |
59 |
| - |
60 |
| - doSeek = False # store if we need to seek to the eof after overwriting |
61 |
| - if self.NameToInfo.has_key(zinfo.filename): |
62 |
| - # Find the last ZipInfo with our name. |
63 |
| - # Last, because that's catching multiple overwrites |
64 |
| - i = len(self.filelist) |
65 |
| - while i > 0: |
66 |
| - i -= 1 |
67 |
| - if self.filelist[i].filename == zinfo.filename: |
68 |
| - break |
69 |
| - zi = self.filelist[i] |
70 |
| - if ((zinfo.compress_type == zipfile.ZIP_STORED |
71 |
| - and zi.compress_size == len(bytes)) |
72 |
| - or (i + 1) == len(self.filelist)): |
73 |
| - # make sure we're allowed to write, otherwise done by writestr below |
74 |
| - self._writecheck(zi) |
75 |
| - # overwrite existing entry |
76 |
| - self.fp.seek(zi.header_offset) |
77 |
| - if (i + 1) == len(self.filelist): |
78 |
| - # this is the last item in the file, just truncate |
79 |
| - self.fp.truncate() |
| 18 | + |
| 19 | + def __init__(self, file, mode="r", compression=zipfile.ZIP_STORED, |
| 20 | + lock=False): |
| 21 | + if lock: |
| 22 | + assert isinstance(file, basestring) |
| 23 | + self.lockfile = lock_file(file + '.lck') |
80 | 24 | else:
|
81 |
| - # we need to move to the end of the file afterwards again |
82 |
| - doSeek = True |
83 |
| - # unhook the current zipinfo, the writestr of our superclass |
84 |
| - # will add a new one |
85 |
| - self.filelist.pop(i) |
86 |
| - self.NameToInfo.pop(zinfo.filename) |
87 |
| - else: |
88 |
| - # Couldn't optimize, sadly, just remember the old entry for removal |
89 |
| - self._remove.append(self.filelist.pop(i)) |
90 |
| - zipfile.ZipFile.writestr(self, zinfo, bytes) |
91 |
| - self.filelist.sort(lambda l, r: cmp(l.header_offset, r.header_offset)) |
92 |
| - if doSeek: |
93 |
| - self.fp.seek(self.end) |
94 |
| - self.end = self.fp.tell() |
95 |
| - |
96 |
| - def close(self): |
97 |
| - """Close the file, and for mode "w" and "a" write the ending |
98 |
| - records. |
99 |
| -
|
100 |
| - Overwritten to compact overwritten entries. |
101 |
| - """ |
102 |
| - if not self._remove: |
103 |
| - # we don't have anything special to do, let's just call base |
104 |
| - r = zipfile.ZipFile.close(self) |
105 |
| - self.lockfile = None |
106 |
| - return r |
107 |
| - |
108 |
| - if self.fp.mode != 'r+b': |
109 |
| - # adjust file mode if we originally just wrote, now we rewrite |
110 |
| - self.fp.close() |
111 |
| - self.fp = open(self.filename, 'r+b') |
112 |
| - all = map(lambda zi: (zi, True), self.filelist) + \ |
113 |
| - map(lambda zi: (zi, False), self._remove) |
114 |
| - all.sort(lambda l, r: cmp(l[0].header_offset, r[0].header_offset)) |
115 |
| - # empty _remove for multiple closes |
116 |
| - self._remove = [] |
117 |
| - |
118 |
| - lengths = [all[i+1][0].header_offset - all[i][0].header_offset |
119 |
| - for i in xrange(len(all)-1)] |
120 |
| - lengths.append(self.end - all[-1][0].header_offset) |
121 |
| - to_pos = 0 |
122 |
| - for (zi, keep), length in zip(all, lengths): |
123 |
| - if not keep: |
124 |
| - continue |
125 |
| - oldoff = zi.header_offset |
126 |
| - # python <= 2.4 has file_offset |
127 |
| - if hasattr(zi, 'file_offset'): |
128 |
| - zi.file_offset = zi.file_offset + to_pos - oldoff |
129 |
| - zi.header_offset = to_pos |
130 |
| - self.fp.seek(oldoff) |
131 |
| - content = self.fp.read(length) |
132 |
| - self.fp.seek(to_pos) |
133 |
| - self.fp.write(content) |
134 |
| - to_pos += length |
135 |
| - self.fp.truncate() |
136 |
| - zipfile.ZipFile.close(self) |
137 |
| - self.lockfile = None |
| 25 | + self.lockfile = None |
| 26 | + |
| 27 | + if mode == 'a' and lock: |
| 28 | + # appending to a file which doesn't exist fails, but we can't check |
| 29 | + # existence util we hold the lock |
| 30 | + if (not os.path.isfile(file)) or os.path.getsize(file) == 0: |
| 31 | + mode = 'w' |
| 32 | + |
| 33 | + zipfile.ZipFile.__init__(self, file, mode, compression) |
| 34 | + self._remove = [] |
| 35 | + self.end = self.fp.tell() |
| 36 | + self.debug = 0 |
| 37 | + |
| 38 | + def writestr(self, zinfo_or_arcname, bytes): |
| 39 | + """Write contents into the archive. |
| 40 | +
|
| 41 | + The contents is the argument 'bytes', 'zinfo_or_arcname' is either |
| 42 | + a ZipInfo instance or the name of the file in the archive. |
| 43 | + This method is overloaded to allow overwriting existing entries. |
| 44 | + """ |
| 45 | + if not isinstance(zinfo_or_arcname, zipfile.ZipInfo): |
| 46 | + zinfo = zipfile.ZipInfo(filename=zinfo_or_arcname, |
| 47 | + date_time=time.localtime(time.time())) |
| 48 | + zinfo.compress_type = self.compression |
| 49 | + # Add some standard UNIX file access permissions (-rw-r--r--). |
| 50 | + zinfo.external_attr = (0x81a4 & 0xFFFF) << 16L |
| 51 | + else: |
| 52 | + zinfo = zinfo_or_arcname |
| 53 | + |
| 54 | + # Now to the point why we overwrote this in the first place, |
| 55 | + # remember the entry numbers if we already had this entry. |
| 56 | + # Optimizations: |
| 57 | + # If the entry to overwrite is the last one, just reuse that. |
| 58 | + # If we store uncompressed and the new content has the same size |
| 59 | + # as the old, reuse the existing entry. |
| 60 | + |
| 61 | + doSeek = False # store if we need to seek to the eof after overwriting |
| 62 | + if self.NameToInfo.has_key(zinfo.filename): |
| 63 | + # Find the last ZipInfo with our name. |
| 64 | + # Last, because that's catching multiple overwrites |
| 65 | + i = len(self.filelist) |
| 66 | + while i > 0: |
| 67 | + i -= 1 |
| 68 | + if self.filelist[i].filename == zinfo.filename: |
| 69 | + break |
| 70 | + zi = self.filelist[i] |
| 71 | + if ((zinfo.compress_type == zipfile.ZIP_STORED |
| 72 | + and zi.compress_size == len(bytes)) |
| 73 | + or (i + 1) == len(self.filelist)): |
| 74 | + # make sure we're allowed to write, otherwise done by writestr below |
| 75 | + self._writecheck(zi) |
| 76 | + # overwrite existing entry |
| 77 | + self.fp.seek(zi.header_offset) |
| 78 | + if (i + 1) == len(self.filelist): |
| 79 | + # this is the last item in the file, just truncate |
| 80 | + self.fp.truncate() |
| 81 | + else: |
| 82 | + # we need to move to the end of the file afterwards again |
| 83 | + doSeek = True |
| 84 | + # unhook the current zipinfo, the writestr of our superclass |
| 85 | + # will add a new one |
| 86 | + self.filelist.pop(i) |
| 87 | + self.NameToInfo.pop(zinfo.filename) |
| 88 | + else: |
| 89 | + # Couldn't optimize, sadly, just remember the old entry for removal |
| 90 | + self._remove.append(self.filelist.pop(i)) |
| 91 | + zipfile.ZipFile.writestr(self, zinfo, bytes) |
| 92 | + self.filelist.sort(lambda l, r: cmp(l.header_offset, r.header_offset)) |
| 93 | + if doSeek: |
| 94 | + self.fp.seek(self.end) |
| 95 | + self.end = self.fp.tell() |
| 96 | + |
| 97 | + def close(self): |
| 98 | + """Close the file, and for mode "w" and "a" write the ending |
| 99 | + records. |
| 100 | +
|
| 101 | + Overwritten to compact overwritten entries. |
| 102 | + """ |
| 103 | + if not self._remove: |
| 104 | + # we don't have anything special to do, let's just call base |
| 105 | + r = zipfile.ZipFile.close(self) |
| 106 | + self.lockfile = None |
| 107 | + return r |
| 108 | + |
| 109 | + if self.fp.mode != 'r+b': |
| 110 | + # adjust file mode if we originally just wrote, now we rewrite |
| 111 | + self.fp.close() |
| 112 | + self.fp = open(self.filename, 'r+b') |
| 113 | + all = map(lambda zi: (zi, True), self.filelist) + \ |
| 114 | + map(lambda zi: (zi, False), self._remove) |
| 115 | + all.sort(lambda l, r: cmp(l[0].header_offset, r[0].header_offset)) |
| 116 | + # empty _remove for multiple closes |
| 117 | + self._remove = [] |
| 118 | + |
| 119 | + lengths = [all[i+1][0].header_offset - all[i][0].header_offset |
| 120 | + for i in xrange(len(all)-1)] |
| 121 | + lengths.append(self.end - all[-1][0].header_offset) |
| 122 | + to_pos = 0 |
| 123 | + for (zi, keep), length in zip(all, lengths): |
| 124 | + if not keep: |
| 125 | + continue |
| 126 | + oldoff = zi.header_offset |
| 127 | + # python <= 2.4 has file_offset |
| 128 | + if hasattr(zi, 'file_offset'): |
| 129 | + zi.file_offset = zi.file_offset + to_pos - oldoff |
| 130 | + zi.header_offset = to_pos |
| 131 | + self.fp.seek(oldoff) |
| 132 | + content = self.fp.read(length) |
| 133 | + self.fp.seek(to_pos) |
| 134 | + self.fp.write(content) |
| 135 | + to_pos += length |
| 136 | + self.fp.truncate() |
| 137 | + zipfile.ZipFile.close(self) |
| 138 | + self.lockfile = None |
0 commit comments