Section: User Commands (1)
Updated: 20 April 2009 (v6.0)
unzip will list, test, or extract files from a ZIP archive, commonly found on MS-DOS systems. The default behavior (with no options) is to extract into the current directory (and subdirectories below it) all files from the specified ZIP archive. A companion program, zip?(1), creates ZIP archives; both programs are compatible with archives created by PKWARE's PKZIP and PKUNZIP for MS-DOS, but in many cases the program options or default behaviors differ.
Note that, in order to support obsolescent hardware, unzip's usage screen is limited to 22 or 23 lines and should therefore be considered only a reminder of the basic unzip syntax rather than an exhaustive list of all possible flags. The exhaustive list follows:
"*.c" matches "foo.c" but not "mydir/foo.c" "**.c" matches both "foo.c" and "mydir/foo.c" "*/*.c" matches "bar/foo.c" but not "baz/bar/foo.c" "??*/*" matches "ab/foo" and "abc/foo" but not "a/foo" or "a/b/foo"
"a.b.3" -> "a.b;3".:
unzips default behavior may be modified via options placed in an environment variable. This can be done with any option, but it is probably most useful with the -a, -L, -C, -q, -o, or -n' modifiers: make unzip auto-convert text files by default, make it convert filenames from uppercase systems to lowercase, make it match names case-insensitively, make it quieter, or make it always overwrite or never overwrite files as it extracts them. For example, to make unzip act as quietly as possible, only reporting errors, one would use one of the following commands:
Unix Bourne shellUNZIP=-qq; export UNZIP:
Unix C shellsetenv UNZIP -qq:
OS/2 or MS-DOSset UNZIP=-qq:
VMS (quotes for lowercase)define UNZIP_OPTS "-qq":
Environment options are, in effect, considered to be just like any other command-line options, except that they are effectively the first options on the command line. To override an environment option, one may use the ``minus operator'' to remove it. For instance, to override one of the quiet-flags in the example above, use the command
unzip --q[other options] zipfile
The first hyphen is the normal switch character, and the second is a minus sign, acting on the q option. Thus the effect here is to cancel one quantum of quietness. To cancel both quiet flags, two (or more) minuses may be used:
unzip -t--q zipfile unzip ---qt zipfile
As suggested by the examples above, the default variable names are UNZIP_OPTS for VMS (where the symbol used to install unzip as a foreign command would otherwise be confused with the environment variable), and UNZIP for all other operating systems. For compatibility with zip?(1), UNZIPOPT is also accepted (don't ask). If both UNZIP and UNZIPOPT are defined, however, UNZIP takes precedence. unzips diagnostic option (-v' with no zipfile name) can be used to check the values of all four possible unzip and zipinfo environment variables.
The timezone variable (TZ) should be set according to the local timezone in order for the -f and -u to operate correctly. See the description of -f above for details. This variable may also be necessary to get timestamps of extracted files to be set correctly. The WIN32 (Win9x/ME/NT4/2K/XP/2K3) port of unzip gets the timezone configuration from the registry, assuming it is correctly set in the Control Panel. The TZ variable is ignored for this port.
Encrypted archives are fully supported by Info-ZIP software, but due to United States export restrictions, de-/encryption support might be disabled in your compiled binary. However, since spring 2000, US export restrictions have been liberated, and our source archives do now include full crypt code. In case you need binary distributions with crypt support enabled, see the file ``WHERE'' in any Info-ZIP source or binary distribution for locations both inside and outside the US.
Some compiled versions of unzip may not support decryption. To check a version for crypt support, either attempt to test or extract an encrypted archive, or else check unzips diagnostic screen (see the -v' option above) for ``[decryption]'' as one of the special compilation options.
As noted above, the -P option may be used to supply a password on the command line, but at a cost in security. The preferred decryption method is simply to extract normally; if a zipfile member is encrypted, unzip will prompt for the password without echoing what is typed. unzip continues to use the same password as long as it appears to be valid, by testing a 12-byte header on each file. The correct password will always check out against the header, but there is a 1-in-256 chance that an incorrect password will as well. (This is a security feature of the PKWARE zipfile format; it helps prevent brute-force attacks that might otherwise gain a large speed advantage by testing only the header.) In the case that an incorrect password is given but it passes the header test anyway, either an incorrect CRC will be generated for the extracted data or else unzip will fail during the extraction because the ``decrypted'' bytes do not constitute a valid compressed data stream.
If the first password fails the header check on some file, unzip will prompt for another password, and so on until all files are extracted. If a password is not known, entering a null password (that is, just a carriage return or ``Enter) is taken as a signal to skip all further prompting. Only unencrypted files in the archive(s) will thereafter be extracted. (In fact, that's not quite true; older versions of zip?(1) and zipcloak?(1) allowed null passwords, so unzip checks each encrypted file to see if the null password works. This may result in ``false positives and extraction errors, as noted above.)
Archives encrypted with 8-bit passwords (for example, passwords with accented European characters) may not be portable across systems and/or other archivers. This problem stems from the use of multiple encoding methods for such characters, including Latin-1 (ISO 8859-1) and OEM code page 850. DOS PKZIP 2.04g uses the OEM code page; Windows PKZIP 2.50 uses Latin-1 (and is therefore incompatible with DOS PKZIP); Info-ZIP uses the OEM code page on DOS, OS/2 and Win3.x ports but ISO coding (Latin-1 etc.) everywhere else; and Nico Mak's WinZip 6.x does not allow 8-bit passwords at all. UnZip 5.3 (or newer) attempts to use the default character set first (e.g., Latin-1), followed by the alternate one (e.g., OEM code page) to test passwords. On EBCDIC systems, if both of these fail, EBCDIC encoding will be tested as a last resort. (EBCDIC is not tested on non-EBCDIC systems, because there are no known archivers that encrypt using EBCDIC encoding.) ISO character encodings other than Latin-1 are not supported. The new addition of (partially) Unicode (resp. UTF-8) support in UnZip 6.0 has not yet been adapted to the encryption password handling in unzip. On systems that use UTF-8 as native character encoding, unzip simply tries decryption with the native UTF-8 encoded password; the built-in attempts to check the password in translated encoding have not yet been adapted for UTF-8 support and will consequently fail.
To use unzip to extract all members of the archive letters.zip into the current directory and subdirectories below it, creating any subdirectories as necessary:
To extract all members of letters.zip into the current directory only:
unzip -j letters
To test letters.zip, printing only a summary message indicating whether the archive is OK or not:
unzip -tq letters
To test all zipfiles in the current directory, printing only the summaries:
unzip -tq \*.zip
(The backslash before the asterisk is only required if the shell expands wildcards, as in Unix; double quotes could have been used instead, as in the source examples below.) To extract to standard output all members of letters.zip whose names end in .tex, auto-converting to the local end-of-line convention and piping the output into more?(1):
unzip -ca letters \*.tex | more
To extract the binary file paper1.dvi to standard output and pipe it to a printing program:
unzip -p articles paper1.dvi | dvips
To extract all FORTRAN and C source files--*.f, *.c, *.h, and Makefile--into the /tmp directory:
unzip source.zip "*.[fch]" Makefile -d /tmp
(the double quotes are necessary only in Unix and only if globbing is turned on). To extract all FORTRAN and C source files, regardless of case (e.g., both *.c and *.C, and any makefile, Makefile, MAKEFILE or similar):
unzip -C source.zip "*.[fch]" makefile -d /tmp
To extract any such files but convert any uppercase MS-DOS or VMS names to lowercase and convert the line-endings of all of the files to the local standard (without respect to any files that might be marked ``binary''):
unzip -aaCL source.zip "*.[fch]" makefile -d /tmp
To extract only newer versions of the files already in the current directory, without querying (NOTE: be careful of unzipping in one timezone a zipfile created in another--ZIP archives other than those created by Zip 2.1 or later contain no timezone information, and a ``newer'' file from an eastern timezone may, in fact, be older):
unzip -fo sources
To extract newer versions of the files already in the current directory and to create any files not already there (same caveat as previous example):
unzip -uo sources
To display a diagnostic screen showing which unzip and zipinfo options are stored in environment variables, whether decryption support was compiled in, the compiler with which unzip was compiled, etc.:
In the last five examples, assume that UNZIP or UNZIP_OPTS is set to -q. To do a singly quiet listing:
unzip -l file.zip
To do a doubly quiet listing:
unzip -ql file.zip
(Note that the ``.zip'' is generally not necessary.) To do a standard listing:
unzip --ql file.zip
unzip -l-q file.zip
unzip -l--q file.zip
The current maintainer, being a lazy sort, finds it very useful to define a pair of aliases: tt for ``unzip -tq and ii for ``unzip -Z (or ``zipinfo). One may then simply type ``tt zipfile to test an archive, something that is worth making a habit of doing. With luck unzip will report ``No errors detected in compressed data of zipfile.zip,'' after which one may breathe a sigh of relief.
The exit status (or error level) approximates the exit codes defined by PKWARE and takes on the following values, except under VMS:
VMS interprets standard Unix (or PC) return values as other, scarier-looking things, so unzip instead maps them into VMS-style status codes. The current mapping is as follows: 1 (success) for normal exit, 0x7fff0001 for warning errors, and (0x7fff000? + 16*normal_unzip_exit_status) for all other errors, where the `?' is 2 (error) for unzip values 2, 9-11 and 80-82, and 4 (fatal error) for the remaining ones (3-8, 50, 51). In addition, there is a compilation option to expand upon this behavior: defining RETURN_CODES results in a human-readable explanation of what the error status means.
Multi-part archives are not yet supported, except in conjunction with zip. (All parts must be concatenated together in order, and then ``zip -F (for zip 2.x) or ``zip -FF (for zip 3.x) must be performed on the concatenated archive in order to ``fix it. Also, zip 3.0 and later can combine multi-part (split) archives into a combined single-file archive using ``zip -s- inarchive -O outarchive. See the zip 3 manual page for more information.) This will definitely be corrected in the next major release.
Archives read from standard input are not yet supported, except with funzip (and then only the first member of the archive can be extracted).
Archives encrypted with 8-bit passwords (e.g., passwords with accented European characters) may not be portable across systems and/or other archivers. See the discussion in DECRYPTION above.
unzips -M' (``more) option tries to take into account automatic wrapping of long lines. However, the code may fail to detect the correct wrapping locations. First, TAB characters (and similar control sequences) are not taken into account, they are handled as ordinary printable characters. Second, depending on the actual system / OS port, unzip'' may not detect the true screen geometry but rather rely on "commonly used" default dimensions. The correct handling of tabs would require the implementation of a query for the actual tabulator setup on the output console.
Dates, times and permissions of stored directories are not restored except under Unix. (On Windows NT and successors, timestamps are now restored.)
[MS-DOS] When extracting or testing files from an archive on a defective floppy diskette, if the ``Fail option is chosen from DOS's ``Abort, Retry, Fail? message, older versions of unzip may hang the system, requiring a reboot. This problem appears to be fixed, but control-C (or control-Break) can still be used to terminate unzip.
Under DEC Ultrix, unzip would sometimes fail on long zipfiles (bad CRC, not always reproducible). This was apparently due either to a hardware bug (cache memory) or an operating system bug (improper handling of page faults?). Since Ultrix has been abandoned in favor of Digital Unix (OSF/1), this may not be an issue anymore.
[Unix] Unix special files such as FIFO buffers (named pipes), block devices and character devices are not restored even if they are somehow represented in the zipfile, nor are hard-linked files relinked. Basically the only file types restored by unzip are regular files, directories and symbolic (soft) links.
[OS/2] Extended attributes for existing directories are only updated if the -o (``overwrite all) option is given. This is a limitation of the operating system; because directories only have a creation time associated with them, unzip has no way to determine whether the stored attributes are newer or older than those on disk. In practice this may mean a two-pass approach is required: first unpack the archive normally (with or without freshening/updating existing files), then overwrite just the directory entries (e.g., ``unzip -o foo */).
[VMS] When extracting to another directory, only the [.foo] syntax is accepted for the -d option; the simple Unix foo syntax is silently ignored (as is the less common VMS foo.dir syntax).
[VMS] When the file being extracted already exists, unzip's query only allows skipping, overwriting or renaming; there should additionally be a choice for creating a new version of the file. In fact, the ``overwrite'' choice does create a new version; the old version is not overwritten or deleted.
The Info-ZIP home page is currently at
The primary Info-ZIP authors (current semi-active members of the Zip-Bugs workgroup) are: Ed Gordon (Zip, general maintenance, shared code, Zip64, Win32, Unix, Unicode); Christian Spieler (UnZip maintenance coordination, VMS, MS-DOS, Win32, shared code, general Zip and UnZip integration and optimization); Onno van der Linden (Zip); Mike White (Win32, Windows GUI, Windows DLLs); Kai Uwe Rommel (OS/2, Win32); Steven M. Schweda (VMS, Unix, support of new features); Paul Kienitz (Amiga, Win32, Unicode); Chris Herborth (BeOS, QNX, Atari); Jonathan Hudson (SMS/QDOS); Sergio Monesi (Acorn RISC OS); Harald Denker (Atari, MVS); John Bush (Solaris, Amiga); Hunter Goatley (VMS, Info-ZIP Site maintenance); Steve Salisbury (Win32); Steve Miller (Windows CE GUI), Johnny Lee (MS-DOS, Win32, Zip64); and Dave Smith (Tandem NSK).
The following people were former members of the Info-ZIP development group and provided major contributions to key parts of the current code: Greg ``Cave Newt'' Roelofs (UnZip, unshrink decompression); Jean-loup Gailly (deflate compression); Mark Adler (inflate decompression, fUnZip).
The author of the original unzip code upon which Info-ZIP's was based is Samuel H. Smith; Carl Mascott did the first Unix port; and David P. Kirschbaum organized and led Info-ZIP in its early days with Keith Petersen hosting the original mailing list at WSMR-SimTel20. The full list of contributors to UnZip has grown quite large; please refer to the CONTRIBS file in the UnZip source distribution for a relatively complete version.
Tutoriais de Tecnologia Web