commands/comm/comm.c

   1 /* comm - select lines from two sorted files    Author: Martin C. Atkins */
   2
   3 /*
   4  *      This program was written by:
   5  *              Martin C. Atkins,
   6  *              University of York,
   7  *              Heslington,
   8  *              York. Y01 5DD
   9  *              England
  10  *      and is released into the public domain, on the condition
  11  *      that this comment is always included without alteration.
  12  */
  13
  14 #include <sys/types.h>
  15 #include <fcntl.h>
  16 #include <stdlib.h>
  17 #include <string.h>
  18 #include <unistd.h>
  19 #include <minix/minlib.h>
  20 #include <stdio.h>
  21
  22 #define BUFFER_SIZE (512)
  23 #define LINMAX (600)
  24
  25 struct file {
  26   char *name;                   /* the file's name */
  27   int fd;                       /* the file descripter */
  28   char buf[BUFFER_SIZE];                /* buffer storage */
  29   char *next;                   /* the next character to read */
  30   char *endp;                   /* the first invalid character */
  31   int seeneof;                  /* an end of file has been seen */
  32 } files[2];
  33
  34 char lines[2][LINMAX];
  35
  36 int colflgs[3] = {1, 2, 3};     /* number of tabs + 1: 0 => no column */
  37
  38 static char *umsg = "Usage: comm [-[123]] file1 file2\n";
  39
  40 int main(int argc, char **argv);
  41 void usage(void);
  42 void error(char *s, char *f);
  43 void eopen(char *fn, struct file *file);
  44 int getbuf(struct file *file);
  45 int readline(int fno);
  46 void comm(void);
  47 void putcol(int col, char *buf);
  48 void cpycol(int col);
  49
  50 int main(argc, argv)
  51 int argc;
  52 char *argv[];
  53 {
  54   int cnt;
  55   if (argc > 1 && argv[1][0] == '-' && argv[1][1] != '\0') {
  56         char *ap;
  57         for (ap = &argv[1][1]; *ap; ap++) switch (*ap) {
  58                     case '1':
  59                     case '2':
  60                     case '3':
  61                         cnt = *ap - '1';
  62                         if (colflgs[cnt] == 0) break;
  63                         colflgs[cnt] = 0;
  64                         for (cnt++; cnt < 3; cnt++) colflgs[cnt]--;
  65                         break;
  66                     default:    usage();
  67                 }
  68         argc--;
  69         argv++;
  70   }
  71   if (argc != 3) usage();
  72   eopen(argv[1], &files[0]);
  73   eopen(argv[2], &files[1]);
  74   comm();
  75   return(0);
  76 }
  77
  78 void usage()
  79 {
  80
  81   std_err(umsg);
  82   exit(1);
  83 }
  84
  85 void error(s, f)
  86 char *s, *f;
  87 {
  88   std_err("comm: ");
  89   std_err(s);
  90   if (f) std_err(f);
  91   std_err("\n");
  92   exit(1);
  93 }
  94
  95 void eopen(fn, file)
  96 char *fn;
  97 struct file *file;
  98 {
  99   file->name = fn;
 100   file->next = file->endp = &file->buf[0];
 101   file->seeneof = 0;
 102   if (fn[0] == '-' && fn[1] == '\0')
 103         file->fd = 0;
 104   else if ((file->fd = open(fn, O_RDONLY)) < 0)
 105         error("can't open ", fn);
 106 }
 107
 108
 109 int getbuf(file)
 110 struct file *file;
 111 {
 112 /* Get a buffer-full from the file.  Return true if no characters
 113  * were obtained because we are at end of file.
 114  */
 115   int n;
 116
 117   if (file->seeneof) return(1);
 118   if ((n = read(file->fd, &file->buf[0], BUFFER_SIZE)) < 0)
 119         error("read error on ", file->name);
 120   if (n == 0) {
 121         file->seeneof++;
 122         return 1;
 123   }
 124   file->next = &file->buf[0];
 125   file->endp = &file->buf[n];
 126   return(0);
 127 }
 128
 129
 130 int readline(fno)
 131 int fno;
 132 {
 133 /* Read up to the next '\n' character to buf.
 134  * Return a complete line, even if end of file occurs within a line.
 135  * Return false at end of file/
 136  */
 137   register struct file *file = &files[fno];
 138   char *buf = lines[fno];
 139
 140   if (file->next == file->endp && getbuf(file)) return(0);
 141   while ((*buf++ = *file->next++) != '\n')
 142         if (file->next == file->endp && getbuf(file)) {
 143                 *buf++ = '\n';
 144                 *buf = '\0';
 145                 return(1);
 146         }
 147   *buf = '\0';
 148   return(1);
 149 }
 150
 151 void comm()
 152 {
 153   register int res;
 154
 155   if (!readline(0)) {
 156         cpycol(1);
 157         return;
 158   }
 159   if (!readline(1)) {
 160         putcol(0, lines[0]);
 161         cpycol(0);
 162         return;
 163   }
 164   for (;;) {
 165         if ((res = strcmp(lines[0], lines[1])) != 0) {
 166                 res = res > 0;
 167                 putcol(res, lines[res]);
 168                 if (!readline(res)) {
 169                         putcol(!res, lines[!res]);
 170                         cpycol(!res);
 171                         return;
 172                 }
 173         } else {
 174                 putcol(2, lines[0]);    /* files[1]lin == f2lin */
 175                 if (!readline(0)) {
 176                         cpycol(1);
 177                         return;
 178                 }
 179                 if (!readline(1)) {
 180                         putcol(0, lines[0]);
 181                         cpycol(0);
 182                         return;
 183                 }
 184         }
 185   }
 186
 187   /* NOTREACHED */
 188 }
 189
 190 void putcol(col, buf)
 191 int col;
 192 char *buf;
 193 {
 194   int cnt;
 195
 196   if (colflgs[col] == 0) return;
 197   for (cnt = 0; cnt < colflgs[col] - 1; cnt++) printf("\t");
 198   printf("%s", buf);
 199 }
 200
 201 void cpycol(col)
 202 int col;
 203 {
 204   if (colflgs[col]) while (readline(col))
 205                 putcol(col, lines[col]);
 206 }