commands/simple/uniq.c

   1 /* uniq - compact repeated lines                Author: John Woods */
   2 /* Uniq [-udc] [-n] [+n] [infile [outfile]]
   3  *
   4  *      Written 02/08/86 by John Woods, placed into public domain.  Enjoy.
   5  *
   6  */
   7
   8 /* If the symbol WRITE_ERROR is defined, uniq will exit(1) if it gets a
   9  * write error on the output.  This is not (of course) how V7 uniq does it,
  10  * so undefine the symbol if you want to lose your output to a full disk
  11  */
  12
  13 #define WRITE_ERROR 1
  14 #include <ctype.h>
  15 #include <errno.h>
  16 #include <string.h>
  17 #include <stdlib.h>
  18 #include <stdio.h>
  19
  20 char buffer[BUFSIZ];
  21 int uflag = 1;                  /* default is union of -d and -u outputs */
  22 int dflag = 1;                  /* flags are mutually exclusive */
  23 int cflag = 0;
  24 int fields = 0;
  25 int chars = 0;
  26
  27 _PROTOTYPE(int main, (int argc, char **argv));
  28 _PROTOTYPE(FILE *xfopen, (char *fn, char *mode));
  29 _PROTOTYPE(char *skip, (char *s));
  30 _PROTOTYPE(int equal, (char *s1, char *s2));
  31 _PROTOTYPE(void show, (char *line, int count));
  32 _PROTOTYPE(int uniq, (void));
  33 _PROTOTYPE(void usage, (void));
  34 _PROTOTYPE(int getline, (char *buf, int count));
  35
  36 FILE *xfopen(fn, mode)
  37 char *fn, *mode;
  38 {
  39   FILE *p;
  40
  41   if ((p = fopen(fn, mode)) == NULL) {
  42         perror("uniq");
  43         fflush(stdout);
  44         exit(1);
  45   }
  46   return(p);
  47 }
  48
  49 int main(argc, argv)
  50 int argc;
  51 char *argv[];
  52 {
  53   char *p;
  54   int inf = -1, outf;
  55
  56   setbuf(stdout, buffer);
  57   for (--argc, ++argv; argc > 0 && (**argv == '-' || **argv == '+');
  58        --argc, ++argv) {
  59         if (**argv == '+')
  60                 chars = atoi(*argv + 1);
  61         else if (isdigit(argv[0][1]))
  62                 fields = atoi(*argv + 1);
  63         else if (argv[0][1] == '\0')
  64                 inf = 0;        /* - is stdin */
  65         else
  66                 for (p = *argv + 1; *p; p++) {
  67                         switch (*p) {
  68                             case 'd':
  69                                 dflag = 1;
  70                                 uflag = 0;
  71                                 break;
  72                             case 'u':
  73                                 uflag = 1;
  74                                 dflag = 0;
  75                                 break;
  76                             case 'c':   cflag = 1;      break;
  77                             default:    usage();
  78                         }
  79                 }
  80   }
  81
  82   /* Input file */
  83   if (argc == 0)
  84         inf = 0;
  85   else if (inf == -1) {         /* if - was not given */
  86         fclose(stdin);
  87         xfopen(*argv++, "r");
  88         argc--;
  89   }
  90   if (argc == 0)
  91         outf = 1;
  92   else {
  93         fclose(stdout);
  94         xfopen(*argv++, "w");
  95         argc--;
  96   }
  97
  98   uniq();
  99   fflush(stdout);
 100   return(0);
 101 }
 102
 103 char *skip(s)
 104 char *s;
 105 {
 106   int n;
 107
 108   /* Skip fields */
 109   for (n = fields; n > 0; --n) {
 110         /* Skip blanks */
 111         while (*s && (*s == ' ' || *s == '\t')) s++;
 112         if (!*s) return s;
 113         while (*s && (*s != ' ' && *s != '\t')) s++;
 114         if (!*s) return s;
 115   }
 116
 117   /* Skip characters */
 118   for (n = chars; n > 0; --n) {
 119         if (!*s) return s;
 120         s++;
 121   }
 122   return s;
 123 }
 124
 125 int equal(s1, s2)
 126 char *s1, *s2;
 127 {
 128   return !strcmp(skip(s1), skip(s2));
 129 }
 130
 131 void show(line, count)
 132 char *line;
 133 int count;
 134 {
 135   if (cflag)
 136         printf("%4d %s", count, line);
 137   else {
 138         if ((uflag && count == 1) || (dflag && count != 1))
 139                 printf("%s", line);
 140   }
 141 }
 142
 143 /* The meat of the whole affair */
 144 char *nowline, *prevline, buf1[1024], buf2[1024];
 145
 146 int uniq()
 147 {
 148   char *p;
 149   int seen;
 150
 151   /* Setup */
 152   prevline = buf1;
 153   if (getline(prevline, 1024) < 0) return(0);
 154   seen = 1;
 155   nowline = buf2;
 156
 157   /* Get nowline and compare if not equal, dump prevline and swap
 158    * pointers else continue, bumping seen count */
 159   while (getline(nowline, 1024) > 0) {
 160         if (!equal(prevline, nowline)) {
 161                 show(prevline, seen);
 162                 seen = 1;
 163                 p = nowline;
 164                 nowline = prevline;
 165                 prevline = p;
 166         } else
 167                 seen += 1;
 168   }
 169   show(prevline, seen);
 170   return 0;
 171 }
 172
 173 void usage()
 174 {
 175   fprintf(stderr, "Usage: uniq [-udc] [+n] [-n] [input [output]]\n");
 176 }
 177
 178 int getline(buf, count)
 179 char *buf;
 180 int count;
 181 {
 182   int c;
 183   int ct = 0;
 184
 185   while (ct++ < count) {
 186         c = getc(stdin);
 187         if (c < 0) return(-1);
 188         *buf++ = c;
 189         if (c == '\n') {
 190                 *buf++ = 0;
 191                 return(ct);
 192         }
 193   }
 194   return(ct);
 195 }