modules/unsupported.c

   1 // This file is part of Deark.
   2 // Copyright (C) 2016 Jason Summers
   3 // See the file COPYING for terms of use.
   4
   5 // File types we recognize but don't support.
   6 // The point is to print a better error message than "unknown format".
   7 // This is most useful for file types that might be mistaken for one we
   8 // do support.
   9
  10 #include <deark-config.h>
  11 #include <deark-private.h>
  12 DE_DECLARE_MODULE(de_module_unsupported);
  13
  14 struct fmtinfo_struct {
  15         int confidence;
  16         int special_message;
  17         const char *descr;
  18 };
  19
  20 // fmti is allocated by the caller.
  21 // get_fmt initializes it. If a format is unidentified,
  22 // it sets ->confidence to 0.
  23 static void get_fmt(deark *c, struct fmtinfo_struct *fmti)
  24 {
  25         u8 b[32];
  26
  27         de_zeromem(fmti, sizeof(struct fmtinfo_struct));
  28
  29         de_read(b, 0, sizeof(b));
  30
  31         if(!de_memcmp(b, "\x42\x5a\x68", 3) &&
  32                 !de_memcmp(&b[4], "\x31\x41\x59\x26\x53\x59", 6) )
  33         {
  34                 fmti->confidence = 90;
  35                 fmti->descr = "a bzip2-compressed file";
  36                 return;
  37         }
  38
  39         if(!de_memcmp(b, "7z\xbc\xaf\x27\x1c", 6)) {
  40                 fmti->confidence = 90;
  41                 fmti->descr = "a 7z file";
  42                 return;
  43         }
  44
  45         // Note - Make sure BSAVE has lower confidence.
  46         if(!de_memcmp(b, "\xfd\x37\x7a\x58\x5a\x00", 6)) {
  47                 fmti->confidence = 90;
  48                 fmti->descr = "an xz-compressed file";
  49                 return;
  50         }
  51
  52         if(!de_memcmp(b, "LZIP", 4)) {
  53                 fmti->confidence = 50;
  54                 fmti->descr = "an lzip-compressed file";
  55                 return;
  56         }
  57
  58         if(!de_memcmp(b, "<?xpacket", 9)) {
  59                 fmti->confidence = 20;
  60                 fmti->descr = "an XMP file";
  61                 return;
  62         }
  63
  64         if(!de_memcmp(b, "ISc(", 4)) {
  65                 fmti->confidence = 40;
  66                 fmti->descr = "an InstallShield CAB file";
  67                 return;
  68         }
  69
  70         if(b[0]=='H' && b[1]=='P' && b[2]=='H' && b[3]=='P' &&
  71                 b[4]=='4' && (b[5]=='8' || b[5]=='9'))
  72         {
  73                 fmti->confidence = 90;
  74                 fmti->descr = "a non-GROB HP-48/49 file";
  75                 return;
  76         }
  77
  78         if(!de_memcmp(b, "%PDF-", 5)) {
  79                 fmti->confidence = 90;
  80                 fmti->descr = "a PDF file";
  81                 return;
  82         }
  83
  84         if(!de_memcmp(b, "\x7f" "ELF", 4)) {
  85                 fmti->confidence = 40;
  86                 fmti->descr = "an ELF binary";
  87                 return;
  88         }
  89
  90         if(!de_memcmp(b, "\xff" "WPC", 4)) {
  91                 fmti->confidence = 40;
  92                 fmti->descr = "a WordPerfect document";
  93                 return;
  94         }
  95
  96         if(!de_memcmp(b, "Rar!\x1a\x07", 6)) {
  97                 fmti->confidence = 90;
  98                 fmti->descr = "a RAR archive";
  99                 return;
 100         }
 101
 102         if((!de_memcmp(b, "StuffIt", 7)) && (b[7]=='!' || b[7]=='?')) {
 103                 fmti->confidence = 90;
 104                 fmti->descr = "a StuffIt X archive";
 105                 return;
 106         }
 107
 108         if(!de_memcmp(b, "ICE!", 4) ||
 109                 !de_memcmp(b, "Ice!", 4))
 110         {
 111                 fmti->confidence = 75;
 112                 fmti->descr = "a Pack-Ice compressed file";
 113                 return;
 114         }
 115
 116         // Note - Make sure Atari CAS has lower confidence.
 117         if(!de_memcmp(b, "FUJIFILMCCD-RAW", 15)) {
 118                 fmti->confidence = 100;
 119                 fmti->descr = "a Fujifilm RAF file";
 120                 return;
 121         }
 122
 123         if(!de_memcmp(b, "AutoCAD Slide\r\n\x1a", 16)) {
 124                 fmti->confidence = 100;
 125                 fmti->descr = "an AutoCAD Slide file";
 126                 return;
 127         }
 128
 129         if(!de_memcmp(b, "Top!", 4)) {
 130                 // A format often found alongside RISC OS Draw files
 131                 fmti->confidence = 9;
 132                 fmti->descr = "an ArtWorks image";
 133                 return;
 134         }
 135
 136         if(!de_memcmp(b, "CPT", 3) &&
 137                 (b[3]>='7' && b[3]<='9') &&
 138                 !de_memcmp(&b[4], "FILE", 4))
 139         {
 140                 fmti->confidence = 91;
 141                 fmti->descr = "a Corel Photo-Paint image";
 142                 return;
 143         }
 144
 145         // We're not trying to detect every HTML file, but we want to make sure
 146         // we can detect the ones we generate.
 147         if(!de_memcmp(b, "<!DOCTYPE html", 14) ||
 148                 !de_memcmp(b, "\xef\xbb\xbf<!DOCTYPE html", 17) ||
 149                 !de_memcmp(b, "<html", 5))
 150         {
 151                 fmti->confidence = 20;
 152                 fmti->descr = "an HTML file";
 153                 return;
 154         }
 155 }
 156
 157 static void de_run_unsupported(deark *c, de_module_params *mparams)
 158 {
 159         struct fmtinfo_struct fmti;
 160         get_fmt(c, &fmti);
 161         if(fmti.confidence>0 && fmti.descr) {
 162                 if(fmti.special_message) {
 163                         de_err(c, "This looks like %s", fmti.descr);
 164                 }
 165                 else {
 166                         de_err(c, "This looks like %s, which is not a supported format.", fmti.descr);
 167                 }
 168         }
 169 }
 170
 171 static int de_identify_unsupported(deark *c)
 172 {
 173         struct fmtinfo_struct fmti;
 174         get_fmt(c, &fmti);
 175         return fmti.confidence;
 176 }
 177
 178 void de_module_unsupported(deark *c, struct deark_module_info *mi)
 179 {
 180         mi->id = "unsupported";
 181         mi->desc = "Identify some unsupported formats";
 182         mi->run_fn = de_run_unsupported;
 183         mi->identify_fn = de_identify_unsupported;
 184         mi->flags |= DE_MODFLAG_HIDDEN | DE_MODFLAG_NOEXTRACT;
 185         mi->unique_id = 1;
 186 }