utils-std

commit: 73a04f2dd3faea132f01c810ad5bfdd0b5e08daa
parent 3ee4f0b9884c59fd1d0d33d68d50abd223ce0c06
Author: Haelwenn (lanodan) Monnier <contact@hacktivis.me>
Date:   Thu, 25 Apr 2024 20:47:34 +0200

cmd/wc: Don't output trailing whitespaces

Diffstat:
M
cmd/wc.1
4
++++
M
cmd/wc.c
45
+++++++++++++++++++++++++++------------------
M
test-cmd/wc.t
44
++++++++++++++++++++++++++++++++++++++++++++

3 files changed, 75 insertions(+), 18 deletions(-)
diff --git a/cmd/wc.1 b/cmd/wc.1
@@ -59,6 +59,10 @@ By default the standard output reports each file in the form:
 "%d %d %d %s", <newlines>, <words>, <bytes>, <file>
 .Ed
 .Pp
+Similarly to GNU and BusyBox, this implementation also makes sure
+to not print trailing whitespace, which would have to be trimmed
+in most scripts.
+.Pp
 If more than one
 .Ar file
 is given, a final line is printed with "total" instead of a pathname.
diff --git a/cmd/wc.c b/cmd/wc.c
@@ -24,12 +24,35 @@ static char *argv0 = "wc";
 static enum {
 	WC_OPT_C   = 1 << 0,
 	WC_OPT_L   = 1 << 1,
-	WC_OPT_W   = 1 << 3,
+	WC_OPT_W   = 1 << 2,
 	WC_OPT_ALL = WC_OPT_C | WC_OPT_L | WC_OPT_W,
 } wc_opts = 0;
 
 off_t total_bytes = 0, total_lines = 0, total_words = 0;
 
+static void
+print_counts(off_t lines, off_t words, off_t bytes, char *filename)
+{
+	char *fmt = "%ld";
+	if(FIELD_MATCH(wc_opts, WC_OPT_L))
+	{
+		printf(fmt, lines);
+		fmt = " %ld";
+	}
+	if(FIELD_MATCH(wc_opts, WC_OPT_W))
+	{
+		printf(fmt, words);
+		fmt = " %ld";
+	}
+	if(FIELD_MATCH(wc_opts, WC_OPT_C))
+	{
+		printf(fmt, bytes);
+	}
+
+	if(filename != NULL) printf(" %s", filename);
+	printf("\n");
+}
+
 static int
 wc_file_bytes(FILE *file, char *filename)
 {
@@ -74,11 +97,7 @@ wc_file_bytes(FILE *file, char *filename)
 
 	total_bytes += bytes, total_lines += lines, total_words += words;
 
-	if(FIELD_MATCH(wc_opts, WC_OPT_L)) printf("%ld ", lines);
-	if(FIELD_MATCH(wc_opts, WC_OPT_W)) printf("%ld ", words);
-	if(FIELD_MATCH(wc_opts, WC_OPT_C)) printf("%ld ", bytes);
-	if(filename != NULL) printf("%s", filename);
-	printf("\n");
+	print_counts(lines, words, bytes, filename);
 
 	return 0;
 }
@@ -125,11 +144,7 @@ wc_file_chars(FILE *file, char *filename)
 
 	if(wordlen > 0) words++;
 
-	if(FIELD_MATCH(wc_opts, WC_OPT_L)) printf("%ld ", lines);
-	if(FIELD_MATCH(wc_opts, WC_OPT_W)) printf("%ld ", words);
-	if(FIELD_MATCH(wc_opts, WC_OPT_C)) printf("%ld ", chars);
-	if(filename != NULL) printf("%s", filename);
-	printf("\n");
+	print_counts(lines, words, chars, filename);
 
 	return 0;
 }
@@ -243,13 +258,7 @@ main(int argc, char *argv[])
 		}
 	}
 
-	if(argc > 1)
-	{
-		if(FIELD_MATCH(wc_opts, WC_OPT_L)) printf("%ld ", total_lines);
-		if(FIELD_MATCH(wc_opts, WC_OPT_W)) printf("%ld ", total_words);
-		if(FIELD_MATCH(wc_opts, WC_OPT_C)) printf("%ld ", total_bytes);
-		printf("total\n");
-	}
+	if(argc > 1) print_counts(total_lines, total_words, total_bytes, "total");
 
 	return 0;
 }
diff --git a/test-cmd/wc.t b/test-cmd/wc.t
@@ -41,3 +41,47 @@
   $ wc /var/empty/e/no/ent
   wc: Failed opening file '/var/empty/e/no/ent': No such file or directory
   [1]
+
+Formatting with reduced counts
+  $ echo | wc -l
+  1
+  $ echo a | wc -l
+  1
+
+  $ echo | wc -c
+  1
+  $ echo a | wc -c
+  2
+
+  $ echo | wc -w
+  0
+  $ echo a | wc -w
+  1
+
+  $ echo | wc -cl
+  1 1
+  $ echo a | wc -cl
+  1 2
+
+  $ wc -c empty
+  0 empty
+  $ wc -cl empty
+  0 0 empty
+  $ wc -w empty
+  0 empty
+
+Total when multiple files are specified
+  $ echo foo | wc empty -
+  0 0 0 empty
+  1 1 4
+  1 1 4 total
+  $ printf 'foo bar\n' > foobar
+  $ wc foobar empty
+  1 2 8 foobar
+  0 0 0 empty
+  1 2 8 total
+
+  $ rm foobar
+  $ rm empty
+  $ find .
+  .