1ascii..SH NAME - determine file type [ ] [ namefile ] [ magicfile 2] file ... tests each argument in an attempt to classify it. 3There are three sets of tests, performed in this order: 4filesystem tests, magic number tests, and language tests. The 5test that succeeds causes the file type to be printed. The type 6printed will usually contain one of the words (the file contains 7only ASCII characters and is probably safe to read on an ASCII 8terminal), (the file contains the result of compiling a program 9in a form understandable to some UNIX kernel or another), or 10meaning anything else (data is usually `binary' or non- 11printable). Exceptions are well-known file formats (core files, 12tar archives) that are known to contain binary data. When 13modifying the file or the program itself, People depend on 14knowing that all the readable files in a directory have the word 15``text'' printed. Don't do as one computer vendor did - change 16``shell commands text'' to ``shell script''. The filesystem 17tests are based on examining the return from a system call. The 18program checks to see if the file is empty, or if it's some sort 19of special file. Any known file types appropriate to the system 20you are running on (sockets and symbolic links on 4.2BSD, named 21pipes (FIFOs) on System V) are intuited if they are defined in 22the system header file The magic number tests are used to check 23for files with data in particular fixed formats. The canonical 24example of this is a binary executable (compiled program) file, 25whose format is defined in and possibly in the standard include 26directory. These files have a `magic number' stored in a 27particular place near the beginning of the file that tells the 28UNIX operating system that the file is a binary executable, and 29which of several types thereof. The concept of `magic number' 30has been applied by extension to data files. Any file with some 31invariant identifier at a small fixed offset into the file can 32usually be described in this way. The information in these files 33is read from the magic file If an argument appears to be an file, 34attempts to guess its language. The language tests look for 35particular strings (cf _n_a_m_e_s._h) that can appear anywhere in the 36first few blocks of a file. For example, the keyword indicates 37that the file is most likely a troff input file, just as the 38keyword indicates a C program. These tests are less reliable 39than the previous two groups, so they are performed last. The 40language test routines also test for some miscellany (such as 41archives) and determine whether an unknown file should be 42labelled as `ascii text' or `data'. Use to specify an alternate 43file of magic numbers. The option causes a checking printout of 44the parsed form of the magic file. This is usually used in 45conjunction with to debug a new magic file before installing it. 46The option specifies that the names of the files to be examined 47are to be read (one per line) from before the argument list. 48Either or at least one filename argument must be present; to test 49the standard input, use ``-'' as a filename argument. - default 50list of magic numbers - description of magic file format. - 51tools for examining non-textfiles. This program is believed to 52exceed the System V Interface Definition of FILE(CMD), as near as 53one can determine from the vague language contained therein. Its 54behaviour is mostly compatible with the System V program of the 55same name. This version knows more magic, however, so it will 56produce different (albeit more accurate) output in many cases. 57The one significant difference between this version and System V 58is that this version treats any white space as a delimiter, so 59that spaces in pattern strings must be escaped. For example, 60>10 string language impress (imPRESS data) in an 61existing magic file would have to be changed to 62>10 string language\ impress (imPRESS data) The Sun 63Microsystems implementation of System V compatibility includes a 64file(1) command that has some extentions. My version differs 65from Sun's only in minor ways. The significant one is the `&' 66operator, which Sun's program expects as, for example, 67>16 long&0x7fffffff >0 not stripped would be entered 68in my version as >16 long &0x7fffffff not stripped which is a 69little less general; it simply tests (location 16)&0x7ffffff and 70returns its truth value as a C expression. The magic file 71entries have been collected from various sources, mainly USENET, 72and contributed by various authors. Ian Darwin (address below) 73will collect additional or corrected magic file entries. A 74consolidation of magic file entries will be distributed 75periodically. The order of entries in the magic file is 76significant. Depending on what system you are using, the order 77that they are put together may be incorrect. If your old command 78uses a magic file, keep the old magic file around for comparison 79purposes (rename it to There has been a command in every UNIX 80since at least Research Version 6 (man page dated January, 1975). 81The System V version introduced one significant major change: the 82external list of magic number types. This slowed the program 83down slightly but made it a lot more flexible. This program, 84based on the System V version, was written by Ian Darwin without 85looking at anybody else's source code. John Gilmore revised the 86code extensively, making it better than the first version. Geoff 87Collyer found several inadequacies and provided some magic file 88entries. The program has undergone continued evolution since. 89Copyright (c) Ian F. Darwin, 1986 and 1987. Written by Ian F. 90Darwin, UUCP address {utzoo | ihnp4}!darwin!ian, Internet address 91ian@sq.com, postal address: P.O. Box 603, Station F, Toronto, 92Ontario, CANADA M4Y 2L8. and written by and copyright by Henry 93Spencer, utzoo!henry. This software is not subject to any 94license of the American Telephone and Telegraph Company or of the 95Regents of the University of California. Permission is granted 96to anyone to use this software for any purpose on any computer 97system, and to alter it and redistribute it freely, subject to 98the following restrictions: 1. The author is not responsible for 99the consequences of use of this software, no matter how awful, 100even if they arise from flaws in it. 2. The origin of this 101software must not be misrepresented, either by explicit claim or 102by omission. Since few users ever read sources, credits must 103appear in the documentation. 3. Altered versions must be plainly 104marked as such, and must not be misrepresented as being the 105original software. Since few users ever read sources, credits 106must appear in the documentation. 4. This notice may not be 107removed or altered. A few support files (_g_e_t_o_p_t, _s_t_r_t_o_k) 108distributed with this package are by Henry Spencer and are 109subject to the same terms as above. A few simple support files 110(_s_t_r_t_o_l, _s_t_r_c_h_r) distributed with this package are in the public 111domain; they are so marked. The files and were written by John 112Gilmore from his public-domain program, and are not covered by 113the above restrictions. There must be a way to automate the 114construction of the Magic file from all the glop in magdir. What 115is it? uses several algorithms that favor speed over accuracy, 116thus it can be misled about the contents of ASCII files. The 117support for ASCII files (primarily for programming languages) is 118simplistic, inefficient and requires recompilation to update. 119Should there be an ``else'' clause to follow a series of 120continuation lines? Is it worthwhile to implement recursive file 121inspection, so that compressed files, uuencoded, etc., can say 122``compressed ascii text'' or ``compressed executable'' or 123``compressed tar archive" or whatever? The magic file and 124keywords should have regular expression support. It might be 125advisable to allow upper-case letters in keywords for e.g., troff 126commands vs man page macros. Regular expression support would 127make this easy. The program doesn't grok FORTRAN. It should be 128able to figure FORTRAN by seeing some keywords which appear 129indented at the start of line. Regular expression support would 130make this easy. The list of keywords in probably belongs in the 131Magic file. This could be done by using some keyword like `*' 132for the offset value. The program should malloc the magic file 133structures, rather than using a fixed-size array as at present. 134The magic file should be compiled into binary (or better yet, 135fixed-length ASCII strings for use in heterogenous network 136environments) for faster startup. Then the program would run as 137fast as the Version 7 program of the same name, with the 138flexibility of the System V version. But then there would have 139to be yet another magic number for the file. Another 140optimisation would be to sort the magic file so that we can just 141run down all the tests for the first byte, first word, first 142long, etc, once we have fetched it. Complain about conflicts in 143the magic file entries. Make a rule that the magic entries sort 144based on file offset rather than position within the magic file? 145The program should provide a way to give an estimate of ``how 146good'' a guess is. We end up removing guesses (e.g. ``From '' as 147first 5 chars of file) because they are not as good as other 148guesses (e.g. ``Newsgroups:'' versus "Return-Path:"). Still, if 149the others don't pan out, it should be possible to use the first 150guess. Perhaps the program should automatically try all tests 151with byte-swapping done, to avoid having to figure out the byte- 152swapped values when constructing the magic file. Of course this 153will run more slowly, so it should probably be an option (-a?). 154This manual page, and particularly this section, is too long. 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199