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 #define SIZE    8000
  10
  11 int     fields  = 0;
  12 int     letters = 0;
  13 int     linec   = 0;
  14 char    mode;
  15 int     uniq;
  16 char    *b1, *b2;
  17 long    bsize;
  18 Biobuf  fin;
  19 Biobuf  fout;
  20
  21 int     gline(char *buf);
  22 void    pline(char *buf);
  23 int     equal(char *b1, char *b2);
  24 char*   skip(char *s);
  25
  26 void
  27 main(int argc, char *argv[])
  28 {
  29         int f;
  30
  31         argv0 = argv[0];
  32         bsize = SIZE;
  33         b1 = malloc(bsize);
  34         b2 = malloc(bsize);
  35         f = 0;
  36         while(argc > 1) {
  37                 if(*argv[1] == '-') {
  38                         if(isdigit(argv[1][1]))
  39                                 fields = atoi(&argv[1][1]);
  40                         else
  41                                 mode = argv[1][1];
  42                         argc--;
  43                         argv++;
  44                         continue;
  45                 }
  46                 if(*argv[1] == '+') {
  47                         letters = atoi(&argv[1][1]);
  48                         argc--;
  49                         argv++;
  50                         continue;
  51                 }
  52                 f = open(argv[1], 0);
  53                 if(f < 0)
  54                         sysfatal("cannot open %s", argv[1]);
  55                 break;
  56         }
  57         if(argc > 2)
  58                 sysfatal("unexpected argument %s", argv[2]);
  59         Binit(&fin, f, OREAD);
  60         Binit(&fout, 1, OWRITE);
  61
  62         if(gline(b1))
  63                 exits(0);
  64         for(;;) {
  65                 linec++;
  66                 if(gline(b2)) {
  67                         pline(b1);
  68                         exits(0);
  69                 }
  70                 if(!equal(b1, b2)) {
  71                         pline(b1);
  72                         linec = 0;
  73                         do {
  74                                 linec++;
  75                                 if(gline(b1)) {
  76                                         pline(b2);
  77                                         exits(0);
  78                                 }
  79                         } while(equal(b2, b1));
  80                         pline(b2);
  81                         linec = 0;
  82                 }
  83         }
  84 }
  85
  86 int
  87 gline(char *buf)
  88 {
  89         int len;
  90         char *p;
  91
  92         p = Brdline(&fin, '\n');
  93         if(p == 0)
  94                 return 1;
  95         len = Blinelen(&fin);
  96         if(len >= bsize-1)
  97                 sysfatal("line too long");
  98         memmove(buf, p, len);
  99         buf[len-1] = 0;
 100         return 0;
 101 }
 102
 103 void
 104 pline(char *buf)
 105 {
 106         switch(mode) {
 107
 108         case 'u':
 109                 if(uniq) {
 110                         uniq = 0;
 111                         return;
 112                 }
 113                 break;
 114
 115         case 'd':
 116                 if(uniq)
 117                         break;
 118                 return;
 119
 120         case 'c':
 121                 Bprint(&fout, "%4d ", linec);
 122         }
 123         uniq = 0;
 124         Bprint(&fout, "%s\n", buf);
 125 }
 126
 127 int
 128 equal(char *b1, char *b2)
 129 {
 130         char c;
 131
 132         if(fields || letters) {
 133                 b1 = skip(b1);
 134                 b2 = skip(b2);
 135         }
 136         for(;;) {
 137                 c = *b1++;
 138                 if(c != *b2++) {
 139                         if(c == 0 && mode == 's')
 140                                 return 1;
 141                         return 0;
 142                 }
 143                 if(c == 0) {
 144                         uniq++;
 145                         return 1;
 146                 }
 147         }
 148 }
 149
 150 char*
 151 skip(char *s)
 152 {
 153         int nf, nl;
 154
 155         nf = nl = 0;
 156         while(nf++ < fields) {
 157                 while(*s == ' ' || *s == '\t')
 158                         s++;
 159                 while(!(*s == ' ' || *s == '\t' || *s == 0) )
 160                         s++;
 161         }
 162         while(nl++ < letters && *s != 0)
 163                         s++;
 164         return s;
 165 }