sbase

suckless unix tools
git clone git://git.2f30.org/sbase.git
Log | Files | Refs | README | LICENSE

uniq.c (2526B)


      1 /* See LICENSE file for copyright and license details. */
      2 #include <ctype.h>
      3 #include <stdio.h>
      4 #include <stdlib.h>
      5 #include <string.h>
      6 
      7 #include "text.h"
      8 #include "util.h"
      9 
     10 static const char *countfmt = "";
     11 static int dflag = 0;
     12 static int uflag = 0;
     13 static int fskip = 0;
     14 static int sskip = 0;
     15 
     16 static struct line prevl;
     17 static ssize_t prevoff    = -1;
     18 static long prevlinecount = 0;
     19 
     20 static size_t
     21 uniqskip(struct line *l)
     22 {
     23 	size_t i;
     24 	int f = fskip, s = sskip;
     25 
     26 	for (i = 0; i < l->len && f; --f) {
     27 		while (isblank(l->data[i]))
     28 			i++;
     29 		while (i < l->len && !isblank(l->data[i]))
     30 			i++;
     31 	}
     32 	for (; s && i < l->len && l->data[i] != '\n'; --s, i++)
     33 		;
     34 
     35 	return i;
     36 }
     37 
     38 static void
     39 uniqline(FILE *ofp, struct line *l)
     40 {
     41 	size_t loff;
     42 
     43 	if (l) {
     44 		loff = uniqskip(l);
     45 
     46 		if (prevoff >= 0 && (l->len - loff) == (prevl.len - prevoff) &&
     47 		    !memcmp(l->data + loff, prevl.data + prevoff, l->len - loff)) {
     48 			++prevlinecount;
     49 			return;
     50 		}
     51 	}
     52 
     53 	if (prevoff >= 0) {
     54 		if ((prevlinecount == 1 && !dflag) ||
     55 		    (prevlinecount != 1 && !uflag)) {
     56 			if (*countfmt)
     57 				fprintf(ofp, countfmt, prevlinecount);
     58 			fwrite(prevl.data, 1, prevl.len, ofp);
     59 		}
     60 		prevoff = -1;
     61 	}
     62 
     63 	if (l) {
     64 		if (!prevl.data || l->len >= prevl.len) {
     65 			prevl.data = erealloc(prevl.data, l->len);
     66 		}
     67 		prevl.len = l->len;
     68 		memcpy(prevl.data, l->data, prevl.len);
     69 		prevoff = loff;
     70 	}
     71 	prevlinecount = 1;
     72 }
     73 
     74 static void
     75 uniq(FILE *fp, FILE *ofp)
     76 {
     77 	static struct line line;
     78 	static size_t size;
     79 	ssize_t len;
     80 
     81 	while ((len = getline(&line.data, &size, fp)) > 0) {
     82 		line.len = len;
     83 		uniqline(ofp, &line);
     84 	}
     85 }
     86 
     87 static void
     88 uniqfinish(FILE *ofp)
     89 {
     90 	uniqline(ofp, NULL);
     91 }
     92 
     93 static void
     94 usage(void)
     95 {
     96 	eprintf("usage: %s [-c] [-d | -u] [-f fields] [-s chars]"
     97 	        " [input [output]]\n", argv0);
     98 }
     99 
    100 int
    101 main(int argc, char *argv[])
    102 {
    103 	FILE *fp[2] = { stdin, stdout };
    104 	int ret = 0, i;
    105 	char *fname[2] = { "<stdin>", "<stdout>" };
    106 
    107 	ARGBEGIN {
    108 	case 'c':
    109 		countfmt = "%7ld ";
    110 		break;
    111 	case 'd':
    112 		dflag = 1;
    113 		break;
    114 	case 'u':
    115 		uflag = 1;
    116 		break;
    117 	case 'f':
    118 		fskip = estrtonum(EARGF(usage()), 0, INT_MAX);
    119 		break;
    120 	case 's':
    121 		sskip = estrtonum(EARGF(usage()), 0, INT_MAX);
    122 		break;
    123 	default:
    124 		usage();
    125 	} ARGEND
    126 
    127 	if (argc > 2)
    128 		usage();
    129 
    130 	for (i = 0; i < argc; i++) {
    131 		if (strcmp(argv[i], "-")) {
    132 			fname[i] = argv[i];
    133 			if (!(fp[i] = fopen(argv[i], (i == 0) ? "r" : "w")))
    134 				eprintf("fopen %s:", argv[i]);
    135 		}
    136 	}
    137 
    138 	uniq(fp[0], fp[1]);
    139 	uniqfinish(fp[1]);
    140 
    141 	ret |= fshut(fp[0], fname[0]) | fshut(fp[1], fname[1]);
    142 
    143 	return ret;
    144 }