1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
|
<section> <date> 1. June 2000 </date>
<h2> ZIP File Access </h2> Using Zipped Files Transparently
<!--border-->
<section>
<h3>The Typedef</h3>
<P>
The typedef <code>ZZIP_FILE</code> can serve as a replacement
for a normal file descriptor. As long as it is only used
for reading a file, the zzlib-user can actually replace
the posix functions <code>open/read/close</code>
by their counterparts from the
<a href="zziplib.html">zziplib library</a>:
<code>zzip_open/zzip_read/zzip_close</code>.
</P>
<P>
As long as the filename path given to <code>zzip_open</code>
refers to a real file in the filesystem, it will almost
directly forward the call to the respective posix <code>open</code>
call. The returned file descriptor is then stored in
a member-variable of the <code>ZZIP_FILE</code> structure.
</P>
<P>
Any subsequent calls to <code>zzip_read</code> will then
be forwarded to the posix <code>read</code> call on the
memorized file descriptor. The same about <code>zzip_close</code>
which will call the posix <code>close</code> function and then
<code>free</code> the <code>ZZIP_FILE</code> structure.
</P>
<P>
The real benefit of the
<a href="zziplib.html">zziplib library</a>
comes about when the filename argument does actually refer
to a file that is zipped in a zip-archive. It happens that
even both a real file and a zipped file can live under the
same pathname given to the <code>zzip_open</code> call,
whereas the real file is used in preference.
</P>
</section><section>
<h3>Zipped File</h3>
<P>
Suppose you have subdirectory called '<tt>test/</tt>'. In
this directory is just one file, called '<tt>README</tt>'.
Calling the <code>zzip_open</code> function with an
argument of '<i>optional-path/</i> <tt>test/README</tt>',
then it will open that file for subsequent reading with
<code>zzip_read</code>. In this case the real (<i>stat'able</i>)
file is opened.
</P>
<P>
Now you can go to the '<tt>test/</tt>' directory and zip up
the files in there by calling
<nobr><tt>`zip ../test.zip *`</tt></nobr>.
After this, you can delete the '<tt>test/</tt>' directory and
the call to <code>zzip_open</code> will still succeed.
The reason is that the part of the path saying
'<tt>test/README</tt>' will be replaced by sth. like
'<tt>test.zip:README</tt>' - that is the real file '<tt>test.zip</tt>'
is opened and searched for a contained file '<tt>README</tt>'.
</P>
<P>
Calling <code>zzip_read</code> on the zipped '<tt>README</tt>' file
will return the very same data as if it is a real file in a
real directory. If the zipped file is compressed it will be
decompressed on the fly.
</P>
</section><section>
<h3>Zip Directory</h3>
<P>
The same applies to the use of <code>opendir/readdir/closedir</code>
which can safely be replaced with their counterparts from the
<a href="zziplib.html">zziplib library</a> - again their prototype
follows the scheme of the original calls, just prepend <tt>zzip_</tt>
to the function calls and <tt>ZZIP_</tt> to the struct-typedefs.
</P>
<P>
To call <code>zzip_opendir</code> on a real directory will then
return a <code>ZZIP_DIR</code> whose member-variable
<code>realdir</code> points to the actual <code>DIR</code>-structure
returned by the underlying posix <code>opendir</code>-call.
</P>
<P>
If a real directory '<tt>test</tt>' does not exist, then the
<code>zzip_opendir</code> will try to open a file '<tt>test.zip</tt>'
with a call to <code>zzip_dir_open</code>.
Subsequent calls to <code>zzip_readdir</code> will then return
information as being obtained from the central archive directory
of the zip-file.
</P>
</section><section>
<h3>Differences</h3>
<P>
There are no differences between the posix calls and their counterparts
from the <a href="zziplib.html">zziplib library</a> - well, just
as long as the zip-file contains just the plain files from a directory.
</P>
<P>
If the zip-file contains directory entries you may be prompted with
some awkward behaviour, since in zip-file a directory happens to be
just an empty file. Note that the posix function <code>open</code>
may also open a directory for reading - it will only return
<code>EISDIR</code> if the <code>open</code> mode-argument included
write-access.
</P>
<P>
What the current of version of the
<a href="zziplib.html">zziplib library</a>
can definitly not do: calling zzip_opendir on a directory zippend
<em>inside</em> a zip-file.
</P>
<P>
To prevent the enrollment of directories into the zip-archive, you
can use the <tt>-D</tt> option of the <tt>zip</tt> program. That
is in any <tt>Makefile</tt> you may want to use
<nobr><tt>`cd $(dir) && zip -D ../$(dir).zip *`</tt></nobr>.
</P>
</section><section>
<h3>Advantages</h3>
<P>
Distribution of a set of files is much easier if it just means
to wrap up a group of files into a zip-archive - and copy that
zip-archive to the respective destination directory.
Even more the files can be compressed and unlike a <tt>tar.gz</tt>
archive there is no need to decompress the archive in temporary
location before accessing a member-file.
</P>
<P>
On the other hand, there is no chance to scatter files around
on the disk like it could easily happen with a set of gzip'ed
man-pages in a single `<tt>man</tt>`-directory. The reader
application does not specifically need to know that the file
is compressed, so that reading a script like
`<tt>share/guile/x.x.x/ice-9/popen.scm</tt>` is done by simple
calls to <code>zzip_read</code> which works on zip-file named
`<tt>share/guile/x.x.x/ice-9.zip</tt>`.
</P>
<P>
A version mismatch between different files in a group is now
obvious: either the opened file belongs to the distribution
archive, or otherwise in resides in a real directory <em>just
next to the zip-archive that contains the original</em>.
</P>
</section><section>
<h3>Issues</h3>
<P>
The <a href="zziplib.html">zziplib library</a> does not
use any code piece from the <code>zip</code> programs, neither
<em>pkzip</em> nor <em>infozip</em>, so there is no license
issue here. The decompression is done by using the free
<a href="http://www.gzip.org/zlib">zlib library</a> which has no special
issues with respect to licensing.
The rights to the <a href="zziplib.html">zziplib library</a>
are reserved to the copyright holders, there is a public
license that puts most the sources themselves under
<a href="COPYING.LIB">the GNU Lesser General Public License</a>,
so that the use of a shared library instance of the
<a href="zziplib.html">zziplib library</a>
has no restrictions of interest to application programmers.
For more details and hints about static linking, check
the <a href="copying.html">COPYING</a> information.
</P>
<P>
The only issue you have with the
<a href="zziplib.html">zziplib library</a>
is the fact that you can only <em>read</em> the contained files.
Writing/Compression is not implemented. Even more, a compressed
file is not seekable at the moment although I hope that someone
will stand up to implement that functionality someday.
</P>
</section></section>
|