sys/src/cmd/uniq.c

   1 /*
   2  * Deal with duplicated lines in a file
   3  */
   4 #include <u.h>
   5 #include <libc.h>
   6 #include <bio.h>
   7 #include <ctype.h>
   8
   9 int     fields  = 0;
  10 int     letters = 0;
  11 int     linec   = 0;
  12 char    mode;
  13 int     uniq;
  14 char    *b1, *b2;
  15 Biobuf  fin;
  16 Biobuf  fout;
  17
  18 int     gline(char *buf);
  19 void    pline(char *buf);
  20 int     equal(char *b1, char *b2);
  21 char*   skip(char *s);
  22
  23 void
  24 main(int argc, char *argv[])
  25 {
  26         int f;
  27
  28         argv0 = argv[0];
  29         b1 = malloc(Bsize);
  30         b2 = malloc(Bsize);
  31         f = 0;
  32         while(argc > 1) {
  33                 if(*argv[1] == '-') {
  34                         if(isdigit(argv[1][1]))
  35                                 fields = atoi(&argv[1][1]);
  36                         else
  37                                 mode = argv[1][1];
  38                         argc--;
  39                         argv++;
  40                         continue;
  41                 }
  42                 if(*argv[1] == '+') {
  43                         letters = atoi(&argv[1][1]);
  44                         argc--;
  45                         argv++;
  46                         continue;
  47                 }
  48                 f = open(argv[1], 0);
  49                 if(f < 0)
  50                         sysfatal("cannot open %s", argv[1]);
  51                 break;
  52         }
  53         if(argc > 2)
  54                 sysfatal("unexpected argument %s", argv[2]);
  55         Binit(&fin, f, OREAD);
  56         Binit(&fout, 1, OWRITE);
  57
  58         if(gline(b1))
  59                 exits(0);
  60         for(;;) {
  61                 linec++;
  62                 if(gline(b2)) {
  63                         pline(b1);
  64                         exits(0);
  65                 }
  66                 if(!equal(b1, b2)) {
  67                         pline(b1);
  68                         linec = 0;
  69                         do {
  70                                 linec++;
  71                                 if(gline(b1)) {
  72                                         pline(b2);
  73                                         exits(0);
  74                                 }
  75                         } while(equal(b2, b1));
  76                         pline(b2);
  77                         linec = 0;
  78                 }
  79         }
  80 }
  81
  82 int
  83 gline(char *buf)
  84 {
  85         int len;
  86         char *p;
  87
  88         p = Brdline(&fin, '\n');
  89         if(p == 0)
  90                 return 1;
  91         len = Blinelen(&fin);
  92         if(len > Bsize)
  93                 sysfatal("line too long");
  94         memmove(buf, p, len);
  95         buf[len-1] = 0;
  96         return 0;
  97 }
  98
  99 void
 100 pline(char *buf)
 101 {
 102         switch(mode) {
 103
 104         case 'u':
 105                 if(uniq) {
 106                         uniq = 0;
 107                         return;
 108                 }
 109                 break;
 110
 111         case 'd':
 112                 if(uniq)
 113                         break;
 114                 return;
 115
 116         case 'c':
 117                 Bprint(&fout, "%4d ", linec);
 118         }
 119         uniq = 0;
 120         Bprint(&fout, "%s\n", buf);
 121 }
 122
 123 int
 124 equal(char *b1, char *b2)
 125 {
 126         char c;
 127
 128         if(fields || letters) {
 129                 b1 = skip(b1);
 130                 b2 = skip(b2);
 131         }
 132         for(;;) {
 133                 c = *b1++;
 134                 if(c != *b2++) {
 135                         if(c == 0 && mode == 's')
 136                                 return 1;
 137                         return 0;
 138                 }
 139                 if(c == 0) {
 140                         uniq++;
 141                         return 1;
 142                 }
 143         }
 144 }
 145
 146 char*
 147 skip(char *s)
 148 {
 149         int nf, nl;
 150
 151         nf = nl = 0;
 152         while(nf++ < fields) {
 153                 while(*s == ' ' || *s == '\t')
 154                         s++;
 155                 while(!(*s == ' ' || *s == '\t' || *s == 0) )
 156                         s++;
 157         }
 158         while(nl++ < letters && *s != 0)
 159                         s++;
 160         return s;
 161 }