1 #include <unistd.h>
2 #include <stdlib.h>
3 #include <fcntl.h>
4 #include <libgen.h>
5 #include <stdio.h>
6 #include <assert.h>
7 #include <math.h>
8 #include <string.h>
9 #include <sys/types.h>
10 #include <sys/stat.h>
11 #include <sys/mman.h>
13 #include "rabinpoly.h"
14 #include "gsimm.h"
16 #define MIN(x,y) ((y)<(x) ? (y) : (x))
17 #define MAX(x,y) ((y)>(x) ? (y) : (x))
19 /* The RABIN_WINDOW_SIZE is the size of fingerprint window used by
20 Rabin algorithm. This is not a modifiable parameter.
22 The first RABIN_WINDOW_SIZE - 1 bytes are skipped, in order to ensure
23 fingerprints are good hashes. This does somewhat reduce the
24 influence of the first few bytes in the file (they're part of
25 fewer windows, like the last few bytes), but that actually isn't
26 so bad as files often start with fixed content that may bias comparisons.
27 */
29 typedef struct fileinfo
30 { char *name;
31 size_t length;
32 u_char md[MD_LENGTH];
33 int match;
34 } File;
36 int flag_verbose = 0;
37 int flag_debug = 0;
38 char *flag_relative = 0;
40 char cmd[12] = " ...";
41 char md_strbuf[MD_LENGTH * 2 + 1];
42 u_char relative_md [MD_LENGTH];
44 File *file;
45 int file_count;
46 size_t file_bytes;
48 char hex[17] = "0123456789abcdef";
50 void usage()
51 { fprintf (stderr, "usage: %s [-dhvw] [-r fingerprint] file ...\n", cmd);
52 fprintf (stderr, " -d\tdebug output, repeate for more verbosity\n");
53 fprintf (stderr, " -h\tshow this usage information\n");
54 fprintf (stderr, " -r\tshow distance relative to fingerprint "
55 "(%u hex digits)\n", MD_LENGTH * 2);
56 fprintf (stderr, " -v\tverbose output, repeat for even more verbosity\n");
57 fprintf (stderr, " -w\tenable warnings for suspect statistics\n");
58 exit (1);
59 }
61 char *md_to_str(u_char *md)
62 { int j;
64 for (j = 0; j < MD_LENGTH; j++)
65 { u_char ch = md[j];
67 md_strbuf[j*2] = hex[ch >> 4];
68 md_strbuf[j*2+1] = hex[ch & 0xF];
69 }
71 md_strbuf[j*2] = 0;
72 return md_strbuf;
73 }
75 void process_file (char *name)
76 { int fd;
77 struct stat fs;
78 u_char *data;
79 File *fi = file+file_count;;
81 fd = open (name, O_RDONLY, 0);
82 if (fd < 0)
83 { perror (name);
84 exit (2);
85 }
87 if (fstat (fd, &fs))
88 { perror (name);
89 exit (2);
90 }
92 if (fs.st_size >= GB_SIMM_MIN_FILE_SIZE
93 && fs.st_size <= GB_SIMM_MAX_FILE_SIZE)
94 { fi->length = fs.st_size;
95 fi->name = name;
97 data = (u_char *) mmap (0, fs.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
99 if (data == (u_char *) -1)
100 { perror (name);
101 exit (2);
102 }
104 gb_simm_process (data, fs.st_size, fi->md);
105 if (flag_relative)
106 fprintf (stdout, "%s %llu %u %s %u %3.1f\n",
107 md_to_str (fi->md), (long long unsigned) 0,
108 (unsigned) fs.st_size, name,
109 (unsigned) 0,
110 100.0 * gb_simm_score(fi->md, relative_md));
111 else
112 {
113 fprintf (stdout, "%s %llu %u %s\n",
114 md_to_str (fi->md), (long long unsigned) 0,
115 (unsigned) fs.st_size, name);
116 }
117 munmap (data, fs.st_size);
118 file_bytes += fs.st_size;
119 file_count++;
120 } else if (flag_verbose)
121 { fprintf (stdout, "skipping %s (size %llu)\n", name, (long long unsigned) fs.st_size); }
123 close (fd);
124 }
126 u_char *str_to_md(char *str, u_char *md)
127 { int j;
129 if (!md || !str) return 0;
131 bzero (md, MD_LENGTH);
133 for (j = 0; j < MD_LENGTH * 2; j++)
134 { char ch = str[j];
136 if (ch >= '0' && ch <= '9')
137 { md [j/2] = (md [j/2] << 4) + (ch - '0');
138 }
139 else
140 { ch |= 32;
142 if (ch < 'a' || ch > 'f') break;
143 md [j/2] = (md[j/2] << 4) + (ch - 'a' + 10);
144 } }
146 return (j != MD_LENGTH * 2 || str[j] != 0) ? 0 : md;
147 }
149 int main (int argc, char *argv[])
150 { int ch, j;
152 strncpy (cmd, basename (argv[0]), 8);
154 while ((ch = getopt(argc, argv, "dhr:vw")) != -1)
155 { switch (ch)
156 { case 'd': flag_debug++;
157 break;
158 case 'r': if (!optarg)
159 { fprintf (stderr, "%s: missing argument for -r\n", cmd);
160 return 1;
161 }
162 if (str_to_md (optarg, relative_md)) flag_relative = optarg;
163 else
164 { fprintf (stderr, "%s: not a valid fingerprint\n", optarg);
165 return 1;
166 }
167 break;
168 case 'v': flag_verbose++;
169 break;
170 case 'w': break;
171 default : usage();
172 return (ch != 'h');
173 } }
175 argc -= optind;
176 argv += optind;
178 if (argc == 0) usage();
180 rabin_reset ();
181 if (flag_verbose && flag_relative)
182 { fprintf (stdout, "distances are relative to %s\n", flag_relative);
183 }
185 file = (File *) calloc (argc, sizeof (File));
187 for (j = 0; j < argc; j++) process_file (argv[j]);
189 if (flag_verbose)
190 { fprintf (stdout, "%li bytes in %i files\n", (long) file_bytes, file_count);
191 }
193 return 0;
194 }