uniq: support -w. closes bug 3094.

function                                             old     new   delta
packed_usage                                       24136   24132      -4
uniq_main                                            399     384     -15

3 files changed, 33 insertions, 17 deletions

diff --git a/coreutils/uniq.c b/coreutils/uniq.c
index 32327c6ce..41f1fed7b 100644
--- a/coreutils/uniq.c
+++ b/coreutils/uniq.c
@@ -12,8 +12,6 @@
 
 #include "libbb.h"
 
-static const char uniq_opts[] ALIGN1 = "cdu" "f:s:" "cdu\0\1\2\4";
-
 static FILE *xgetoptfile_uniq_s(char **argv, int read0write2)
 {
         const char *n;
@@ -31,9 +29,11 @@ int uniq_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 int uniq_main(int argc ATTRIBUTE_UNUSED, char **argv)
 {
         FILE *in, *out;
-        unsigned long dups, skip_fields, skip_chars, i;
         const char *s0, *e0, *s1, *e1, *input_filename;
+        unsigned long dups;
+        unsigned skip_fields, skip_chars, max_chars;
         unsigned opt;
+        unsigned i;
 
         enum {
                 OPT_c = 0x1,
@@ -41,15 +41,14 @@ int uniq_main(int argc ATTRIBUTE_UNUSED, char **argv)
                 OPT_u = 0x4,
                 OPT_f = 0x8,
                 OPT_s = 0x10,
+                OPT_w = 0x20,
         };
 
         skip_fields = skip_chars = 0;
+        max_chars = -1;
 
-        opt = getopt32(argv, "cduf:s:", &s0, &s1);
-        if (opt & OPT_f)
-                skip_fields = xatoul(s0);
-        if (opt & OPT_s)
-                skip_chars = xatoul(s1);
+        opt_complementary = "f+:s+:w+";
+        opt = getopt32(argv, "cduf:s:w:", &skip_fields, &skip_chars, &max_chars);
         argv += optind;
 
         input_filename = *argv;
@@ -63,7 +62,7 @@ int uniq_main(int argc ATTRIBUTE_UNUSED, char **argv)
                 bb_show_usage();
         }
 
-        s1 = e1 = NULL;                         /* prime the pump */
+        s1 = e1 = NULL; /* prime the pump */
 
         do {
                 s0 = s1;
@@ -81,16 +80,16 @@ int uniq_main(int argc ATTRIBUTE_UNUSED, char **argv)
                                 ++e1;
                         }
 
-                        if (!s0 || strcmp(e0, e1)) {
+                        if (!s0 || strncmp(e0, e1, max_chars)) {
                                 break;
                         }
 
-                        ++dups;          /* Note: Testing for overflow seems excessive. */
+                        ++dups;  /* note: testing for overflow seems excessive. */
                 }
 
                 if (s0) {
                         if (!(opt & (OPT_d << !!dups))) { /* (if dups, opt & OPT_e) */
-                                fprintf(out, "\0%d " + (opt & 1), dups + 1);
+                                fprintf(out, "\0%ld " + (opt & 1), dups + 1); /* 1 == OPT_c */
                                 fprintf(out, "%s\n", s0);
                         }
                         free((void *)s0);
diff --git a/include/usage.h b/include/usage.h
index cbc5cb0a2..e791ba6bf 100644
--- a/include/usage.h
+++ b/include/usage.h
@@ -4303,16 +4303,16 @@
         )
 
 #define uniq_trivial_usage \
-       "[-fscdu]... [INPUT [OUTPUT]]"
+       "[-fscduw]... [INPUT [OUTPUT]]"
 #define uniq_full_usage "\n\n" \
-       "Discard all but one of successive identical lines from INPUT\n" \
-       "(or standard input), writing to OUTPUT (or standard output)\n" \
+       "Discard duplicate lines\n" \
      "\nOptions:" \
      "\n        -c      Prefix lines by the number of occurrences" \
      "\n        -d      Only print duplicate lines" \
      "\n        -u      Only print unique lines" \
-     "\n        -f N    Skip the first N fields" \
-     "\n        -s N    Skip the first N chars (after any skipped fields)" \
+     "\n        -f N    Skip first N fields" \
+     "\n        -s N    Skip first N chars (after any skipped fields)" \
+     "\n        -w N    Compare N characters in line" \
 
 #define uniq_example_usage \
        "$ echo -e \"a\\na\\nb\\nc\\nc\\na\" | sort | uniq\n" \
diff --git a/testsuite/uniq.tests b/testsuite/uniq.tests
index 49d4bed9c..8961d669c 100755
--- a/testsuite/uniq.tests
+++ b/testsuite/uniq.tests
@@ -41,6 +41,7 @@ testing "uniq input - (specify stdout)" "uniq input -" \
 #-c occurrences
 #-d dups only
 #-u
+#-w max chars
 
 # Test various command line options
 
@@ -60,6 +61,22 @@ aa	bb	cc9
 bb      cc      dd8
 aa      bb      cc9
 "
+testing "uniq -w (compare max characters)" "uniq -w 2" \
+"cc1
+" "" \
+"cc1
+cc2
+cc3
+"
+
+testing "uniq -s -w (skip fields and compare max chars)" \
+"uniq -s 2 -w 2" \
+"aaccaa
+" "" \
+"aaccaa
+aaccbb
+bbccaa
+"
 
 # -d is "Suppress the writing fo lines that are not repeated in the input."
 # -u is "Suppress the writing of lines that are repeated in the input."