contrib/file/names.h

   1 /*
   2  * Names.h - names and types used by ascmagic in file(1).
   3  * These tokens are here because they can appear anywhere in
   4  * the first HOWMANY bytes, while tokens in MAGIC must
   5  * appear at fixed offsets into the file. Don't make HOWMANY
   6  * too high unless you have a very fast CPU.
   7  *
   8  * Copyright (c) Ian F. Darwin, 1987.
   9  * Written by Ian F. Darwin.
  10  *
  11  * See LEGAL.NOTICE
  12  *
  13  * $Id: names.h,v 1.19 2002/05/16 15:01:41 christos Exp $
  14  */
  15
  16 /*
  17         modified by Chris Lowth - 9 April 2000
  18         to add mime type strings to the types table.
  19 */
  20
  21 /* these types are used to index the table 'types': keep em in sync! */
  22 #define L_C     0               /* first and foremost on UNIX */
  23 #define L_CC    1               /* Bjarne's postincrement */
  24 #define L_FORT  2               /* the oldest one */
  25 #define L_MAKE  3               /* Makefiles */
  26 #define L_PLI   4               /* PL/1 */
  27 #define L_MACH  5               /* some kinda assembler */
  28 #define L_ENG   6               /* English */
  29 #define L_PAS   7               /* Pascal */
  30 #define L_MAIL  8               /* Electronic mail */
  31 #define L_NEWS  9               /* Usenet Netnews */
  32 #define L_JAVA  10              /* Java code */
  33 #define L_HTML  11              /* HTML */
  34 #define L_BCPL  12              /* BCPL */
  35 #define L_M4    13              /* M4 */
  36
  37 static const struct {
  38         char *human;
  39         char *mime;
  40 } types[] = {
  41         { "C program",                                  "text/x-c", },
  42         { "C++ program",                                "text/x-c++" },
  43         { "FORTRAN program",                            "text/x-fortran" },
  44         { "make commands",                              "text/x-makefile" },
  45         { "PL/1 program",                               "text/x-pl1" },
  46         { "assembler program",                          "text/x-asm" },
  47         { "English",                                    "text/plain, English" },
  48         { "Pascal program",                             "text/x-pascal" },
  49         { "mail",                                       "text/x-mail" },
  50         { "news",                                       "text/x-news" },
  51         { "Java program",                               "text/x-java" },
  52         { "HTML document",                              "text/html", },
  53         { "BCPL program",                               "text/x-bcpl" },
  54         { "M4 macro language pre-processor",            "text/x-m4" },
  55         { "can't happen error on names.h/types",        "error/x-error" },
  56         { 0, 0}
  57 };
  58
  59 /*
  60  * XXX - how should we distinguish Java from C++?
  61  * The trick used in a Debian snapshot, of having "extends" or "implements"
  62  * as tags for Java, doesn't work very well, given that those keywords
  63  * are often preceded by "class", which flags it as C++.
  64  *
  65  * Perhaps we need to be able to say
  66  *
  67  *      If "class" then
  68  *
  69  *              if "extends" or "implements" then
  70  *                      Java
  71  *              else
  72  *                      C++
  73  *      endif
  74  *
  75  * Or should we use other keywords, such as "package" or "import"?
  76  * Unfortunately, Ada95 uses "package", and Modula-3 uses "import",
  77  * although I infer from the language spec at
  78  *
  79  *      http://www.research.digital.com/SRC/m3defn/html/m3.html
  80  *
  81  * that Modula-3 uses "IMPORT" rather than "import", i.e. it must be
  82  * in all caps.
  83  *
  84  * So, for now, we go with "import".  We must put it before the C++
  85  * stuff, so that we don't misidentify Java as C++.  Not using "package"
  86  * means we won't identify stuff that defines a package but imports
  87  * nothing; hopefully, very little Java code imports nothing (one of the
  88  * reasons for doing OO programming is to import as much as possible
  89  * and write only what you need to, right?).
  90  *
  91  * Unfortunately, "import" may cause us to misidentify English text
  92  * as Java, as it comes after "the" and "The".  Perhaps we need a fancier
  93  * heuristic to identify Java?
  94  */
  95 static struct names {
  96         const char *name;
  97         short type;
  98 } names[] = {
  99         /* These must be sorted by eye for optimal hit rate */
 100         /* Add to this list only after substantial meditation */
 101         {"dnl",         L_M4},
 102         {"import",      L_JAVA},
 103         {"\"libhdr\"",  L_BCPL},
 104         {"\"LIBHDR\"",  L_BCPL},
 105         {"//",          L_CC},
 106         {"template",    L_CC},
 107         {"virtual",     L_CC},
 108         {"class",       L_CC},
 109         {"public:",     L_CC},
 110         {"private:",    L_CC},
 111         {"/*",          L_C},   /* must precede "The", "the", etc. */
 112         {"#include",    L_C},
 113         {"char",        L_C},
 114         {"The",         L_ENG},
 115         {"the",         L_ENG},
 116         {"double",      L_C},
 117         {"extern",      L_C},
 118         {"float",       L_C},
 119         {"struct",      L_C},
 120         {"union",       L_C},
 121         {"CFLAGS",      L_MAKE},
 122         {"LDFLAGS",     L_MAKE},
 123         {"all:",        L_MAKE},
 124         {".PRECIOUS",   L_MAKE},
 125 /* Too many files of text have these words in them.  Find another way
 126  * to recognize Fortrash.
 127  */
 128 #ifdef  NOTDEF
 129         {"subroutine",  L_FORT},
 130         {"function",    L_FORT},
 131         {"block",       L_FORT},
 132         {"common",      L_FORT},
 133         {"dimension",   L_FORT},
 134         {"integer",     L_FORT},
 135         {"data",        L_FORT},
 136 #endif  /*NOTDEF*/
 137         {".ascii",      L_MACH},
 138         {".asciiz",     L_MACH},
 139         {".byte",       L_MACH},
 140         {".even",       L_MACH},
 141         {".globl",      L_MACH},
 142         {".text",       L_MACH},
 143         {"clr",         L_MACH},
 144         {"(input,",     L_PAS},
 145         {"dcl",         L_PLI},
 146         {"Received:",   L_MAIL},
 147         {">From",       L_MAIL},
 148         {"Return-Path:",L_MAIL},
 149         {"Cc:",         L_MAIL},
 150         {"Newsgroups:", L_NEWS},
 151         {"Path:",       L_NEWS},
 152         {"Organization:",L_NEWS},
 153         {"href=",       L_HTML},
 154         {"HREF=",       L_HTML},
 155         {"<body",       L_HTML},
 156         {"<BODY",       L_HTML},
 157         {NULL,          0}
 158 };
 159 #define NNAMES ((sizeof(names)/sizeof(struct names)) - 1)