Major coreutils update.

1 files changed, 75 insertions, 51 deletions

diff --git a/coreutils/uniq.c b/coreutils/uniq.c
index cb63c4277..90686c9af 100644
--- a/coreutils/uniq.c
+++ b/coreutils/uniq.c
@@ -1,10 +1,8 @@
 /* vi: set sw=4 ts=4: */
 /*
- * Mini uniq implementation for busybox
+ * uniq implementation for busybox
  *
- * Copyright (C) 1999 by Lineo, inc. and John Beppu
- * Copyright (C) 1999,2000,2001 by John Beppu <beppu@codepoet.org>
- * Rewritten by Matt Kraai <kraai@alumni.carnegiemellon.edu>
+ * Copyright (C) 2003  Manuel Novoa III  <mjn3@codepoet.org>
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -22,67 +20,93 @@
  *
  */
 
+/* BB_AUDIT SUSv3 compliant */
+/* http://www.opengroup.org/onlinepubs/007904975/utilities/uniq.html */
+
 #include <stdio.h>
-#include <string.h>
-#include <getopt.h>
-#include <errno.h>
 #include <stdlib.h>
+#include <string.h>
+#include <ctype.h>
+#include <unistd.h>
 #include "busybox.h"
+#include "libcoreutils/coreutils.h"
 
-static int print_count;
-static int print_uniq = 1;
-static int print_duplicates = 1;
-
-static void print_line(char *line, int count, FILE *fp)
-{
-	if ((print_duplicates && count > 1) || (print_uniq && count == 1)) {
-		if (print_count)
-			fprintf(fp, "%7d\t%s", count, line);
-		else
-			fputs(line, fp);
-	}
-}
+static const char uniq_opts[] = "f:s:cdu\0\7\3\5\1\2\4";
 
 int uniq_main(int argc, char **argv)
 {
-	FILE *in = stdin, *out = stdout;
-	char *lastline = NULL, *input;
-	int opt, count = 0;
+	FILE *in, *out;
+	/* Note: Ignore the warning about dups and e0 possibly being uninitialized.
+	 * They will be initialized on the fist pass of the loop (since s0 is NULL). */
+	unsigned long dups, skip_fields, skip_chars, i;
+	const char *s0, *e0, *s1, *e1, *input_filename;
+	int opt;
+	int uniq_flags = 6;		/* -u */
 
-	/* parse argv[] */
-	while ((opt = getopt(argc, argv, "cdu")) > 0) {
-		switch (opt) {
-			case 'c':
-				print_count = 1;
-				break;
-			case 'd':
-				print_duplicates = 1;
-				print_uniq = 0;
-				break;
-			case 'u':
-				print_duplicates = 0;
-				print_uniq = 1;
-				break;
+	skip_fields = skip_chars = 0;
+
+	while ((opt = getopt(argc, argv, uniq_opts)) > 0) {
+		if (opt == 'f') {
+			skip_fields = bb_xgetularg10(optarg);
+		} else if (opt == 's') {
+			skip_chars = bb_xgetularg10(optarg);
+		} else if ((s0 = strchr(uniq_opts, opt)) != NULL) {
+			uniq_flags &= s0[4];
+			uniq_flags |= s0[7];
+		} else {
+			bb_show_usage();
 		}
 	}
 
-	if (argv[optind] != NULL) {
-		in = xfopen(argv[optind], "r");
-		if (argv[optind+1] != NULL)
-			out = xfopen(argv[optind+1], "w");
+	input_filename = *(argv += optind);
+
+	in = xgetoptfile_sort_uniq(argv, "r");
+	if (*argv) {
+		++argv;
 	}
+	out = xgetoptfile_sort_uniq(argv, "w");
+	if (*argv && argv[1]) {
+		bb_show_usage();
+	}
+
+	s0 = NULL;
 
-	while ((input = get_line_from_file(in)) != NULL) {
-		if (lastline == NULL || strcmp(input, lastline) != 0) {
-			print_line(lastline, count, out);
-			free(lastline);
-			lastline = input;
-			count = 0;
+	/* gnu uniq ignores newlines */
+	while ((s1 = bb_get_chomped_line_from_file(in)) != NULL) {
+		e1 = s1;
+		for (i=skip_fields ; i ; i--) {
+			e1 = bb_skip_whitespace(e1);
+			while (*e1 && !isspace(*e1)) {
+				++e1;
+			}
+		}
+		for (i = skip_chars ; *e1 && i ; i--) {
+			++e1;
+		}
+		if (s0) {
+			if (strcmp(e0, e1) == 0) {
+				++dups;		/* Note: Testing for overflow seems excessive. */
+				continue;
+			}
+		DO_LAST:
+			if ((dups && (uniq_flags & 2)) || (!dups && (uniq_flags & 4))) {
+				bb_fprintf(out, "\0%7d\t" + (uniq_flags & 1), dups + 1);
+				bb_fprintf(out, "%s\n", s0);
+			}
+			free((void *)s0);
 		}
-		count++;
+
+		s0 = s1;
+		e0 = e1;
+		dups = 0;
+	}
+
+	if (s0) {
+		e1 = NULL;
+		goto DO_LAST;
 	}
-	print_line(lastline, count, out);
-	free(lastline);
 
-	return EXIT_SUCCESS;
+	bb_xferror(in, input_filename);
+
+	bb_fflush_stdout_and_exit(EXIT_SUCCESS);
 }