The last three fields are the modification date and time of the file, and its name. The case of the filename is respected; thus files that come from MS-DOS PKZIP are always capitalized. If the file was zipped with a stored directory name, that is also displayed as part of the filename.
The second and third fields indicate that the file was zipped under Unix with version 1.9 of zip. Since it comes from Unix, the file permissions at the beginning of the line are printed in Unix format. The uncompressed file-size (2802 in this example) is the fourth field.
The fifth field consists of two characters, either of which may take on several values. The first character may be either `t' or `b', indicating that zip believes the file to be text or binary, respectively; but if the file is encrypted, zipinfo notes this fact by capitalizing the character (`T' or `B'). The second character may also take on four values, depending on whether there is an extended local header and/or an ``extra field'' associated with the file (fully explained in PKWare's APPNOTE.TXT, but basically analogous to pragmas in ANSI C--i.e., they provide a standard way to include non-standard information in the archive). If neither exists, the character will be a hyphen (`-'); if there is an extended local header but no extra field, `l'; if the reverse, `x'; and if both exist, `X'. Thus the file in this example is (probably) a text file, is not encrypted, and has neither an extra field nor an extended local header associated with it. The example below, on the other hand, is an encrypted binary file with an extra field:
Extra fields are used for various purposes (see discussion of the -v option below) including the storage of VMS file attributes, which is presumably the case here. Note that the file attributes are listed in VMS format. Some other possibilities for the host operating system (which is actually a misnomer--host file system is more correct) include OS/2 or NT with High Performance File System (HPFS), MS-DOS, OS/2 or NT with File Allocation Table (FAT) file system, and Macintosh. These are denoted as follows:
File attributes in the first two cases are indicated in a Unix-like format, where the seven subfields indicate whether the file: (1) is a directory, (2) is readable (always true), (3) is writable, (4) is executable (guessed on the basis of the extension--.exe, .com, .bat, .cmd and .btm files are assumed to be so), (5) has its archive bit set, (6) is hidden, and (7) is a system file. Interpretation of Macintosh file attributes is unreliable because some Macintosh archivers don't store any attributes in the archive.
Finally, the sixth field indicates the compression method and possible sub-method used. There are six methods known at present: storing (no compression), reducing, shrinking, imploding, tokenizing (never publicly released), and deflating. In addition, there are four levels of reducing (1 through 4); four types of imploding (4K or 8K sliding dictionary, and 2 or 3 Shannon-Fano trees); and four levels of deflating (superfast, fast, normal, maximum compression). zipinfo represents these methods and their sub-methods as follows: stor; re:1, re:2, etc.; shrk; i4:2, i8:3, etc.; tokn; and defS, defF, defN, and defX.
The medium and long listings are almost identical to the short format except that they add information on the file's compression. The medium format lists the file's compression factor as a percentage indicating the amount of space that has been ``removed'':
In this example, the file has been compressed by more than a factor of five; the compressed data are only 19% of the original size. The long format gives the compressed file's size in bytes, instead:
In contrast to the unzip listings, the compressed size figures in this listing format denote the complete size of compressed data, including the 12 extra header bytes in case of encrypted entries.
Adding the -T option changes the file date and time to decimal format:
Note that because of limitations in the MS-DOS format used to store file times, the seconds field is always rounded to the nearest even second. For Unix files this is expected to change in the next major releases of zip(1) and unzip.
In addition to individual file information, a default zipfile listing also includes header and trailer lines:
The header line gives the name of the archive, its total size, and the total number of files; the trailer gives the number of files listed, their total uncompressed size, and their total compressed size (not including any of zip's internal overhead). If, however, one or more file(s) are provided, the header and trailer lines are not listed. This behavior is also similar to that of Unix's ``ls -l''; it may be overridden by specifying the -h and -t options explicitly. In such a case the listing format must also be specified explicitly, since -h or -t (or both) in the absence of other options implies that ONLY the header or trailer line (or both) is listed. See the EXAMPLES section below for a semi-intelligible translation of this nonsense.
The verbose listing is mostly self-explanatory. It also lists file comments and the zipfile comment, if any, and the type and number of bytes in any stored extra fields. Currently known types of extra fields include PKWARE's authentication (``AV'') info; OS/2 extended attributes; VMS filesystem info, both PKWARE and Info-ZIP versions; Macintosh resource forks; Acorn/Archimedes SparkFS info; and so on. (Note that in the case of OS/2 extended attributes--perhaps the most common use of zipfile extra fields--the size of the stored EAs as reported by zipinfo may not match the number given by OS/2's dir command: OS/2 always reports the number of bytes required in 16-bit format, whereas zipinfo always reports the 32-bit storage.)
Again, the compressed size figures of the individual entries include the 12 extra header bytes for encrypted entries. In contrast, the archive total compressed size and the average compression ratio shown in the summary bottom line are calculated without the extra 12 header bytes of encrypted entries.
The default listing format, as noted above, corresponds roughly to the "zipinfo -hst" command (except when individual zipfile members are specified). A user who prefers the long-listing format (-l) can make use of the zipinfo's environment variable to change this default:
If, in addition, the user dislikes the trailer line, zipinfo's concept of ``negative options'' may be used to override the default inclusion of the line. This is accomplished by preceding the undesired option with one or more minuses: e.g., ``-l-t'' or ``--tl'', in this example. The first hyphen is the regular switch character, but the one before the `t' is a minus sign. The dual use of hyphens may seem a little awkward, but it's reasonably intuitive nonetheless: simply ignore the first hyphen and go from there. It is also consistent with the behavior of the Unix command nice(1).
As suggested above, the default variable names are ZIPINFO_OPTS for VMS (where the symbol used to install zipinfo as a foreign command would otherwise be confused with the environment variable), and ZIPINFO for all other operating systems. For compatibility with zip(1), ZIPINFOOPT is also accepted (don't ask). If both ZIPINFO and ZIPINFOOPT are defined, however, ZIPINFO takes precedence. unzip's diagnostic option (-v with no zipfile name) can be used to check the values of all four possible unzip and zipinfo environment variables.
To produce a basic, long-format listing (not verbose), including header and totals lines, use -l:
zipinfo -l storage
To list the complete contents of the archive without header and totals lines, either negate the -h and -t options or else specify the contents explicitly:
zipinfo --h-t storage zipinfo storage \*
(where the backslash is required only if the shell would otherwise expand the `*' wildcard, as in Unix when globbing is turned on--double quotes around the asterisk would have worked as well). To turn off the totals line by default, use the environment variable (C shell is assumed here):
setenv ZIPINFO --t zipinfo storage
To get the full, short-format listing of the first example again, given that the environment variable is set as in the previous example, it is necessary to specify the -s option explicitly, since the -t option by itself implies that ONLY the footer line is to be printed:
setenv ZIPINFO --t zipinfo -t storage [only totals line] zipinfo -st storage [full listing]
The -s option, like -m and -l, includes headers and footers by default, unless otherwise specified. Since the environment variable specified no footers and that has a higher precedence than the default behavior of -s, an explicit -t option was necessary to produce the full listing. Nothing was indicated about the header, however, so the -s option was sufficient. Note that both the -h and -t options, when used by themselves or with each other, override any default listing of member files; only the header and/or footer are printed. This behavior is useful when zipinfo is used with a wildcard zipfile specification; the contents of all zipfiles are then summarized with a single command.
To list information on a single file within the archive, in medium format, specify the filename explicitly:
zipinfo -m storage unshrink.c
The specification of any member file, as in this example, will override the default header and totals lines; only the single line of information about the requested file will be printed. This is intuitively what one would expect when requesting information about a single file. For multiple files, it is often useful to know the total compressed and uncompressed size; in such cases -t may be specified explicitly:
zipinfo -mt storage "*.[ch]" Mak\*
To get maximal information about the ZIP archive, use the verbose option. It is usually wise to pipe the output into a filter such as Unix more(1) if the operating system allows it:
zipinfo -v storage | more
zipinfo -T storage | sort -nr -k 7 | sed 15q
The -nr option to sort(1) tells it to sort numerically in reverse order rather than in textual order, and the -k 7 option tells it to sort on the seventh field. This assumes the default short-listing format; if -m or -l is used, the proper sort(1) option would be -k 8. Older versions of sort(1) do not support the -k option, but you can use the traditional + option instead, e.g., +6 instead of -k 7. The sed(1) command filters out all but the first 15 lines of the listing. Future releases of zipinfo may incorporate date/time and filename sorting as built-in options.
zipinfo's listing-format behavior is unnecessarily complex and should be simplified. (This is not to say that it will be.)