utils-std

commit: 8aec98a517fdbbb0fc303e64ad5c7f61babdfd1b
parent 7c50cbab04faffa818b127dd3b55047dc01cfc4f
Author: Haelwenn (lanodan) Monnier <contact@hacktivis.me>
Date:   Fri, 16 Jan 2026 23:50:47 +0100

cmd/cut: implement -n option

Fixes: https://todo.sr.ht/~lanodan/utils-std/4

Diffstat:
M
cmd/cut.1
5
++---
M
cmd/cut.c
60
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
M
test-cmd/cut.sh
11
++++++++++-

3 files changed, 71 insertions(+), 5 deletions(-)
diff --git a/cmd/cut.1 b/cmd/cut.1
@@ -74,7 +74,8 @@ if an empty string is passed, then NULL is used as separator.
 Cut fields based on
 .Ar list .
 .It Fl n
-Do not split codepoints. (Currently unsupported in this implementation)
+Prevent splitting codepoints.
+If the last byte of a codepoint is part of the selection, it gets printed, otherwise it gets dropped.
 .It Fl s
 Suppress lines with no delimiter characters,
 otherwise whole delimiter-less lines are printed as-is.
@@ -84,8 +85,6 @@ Use NULL as line separator instead of newline.
 .Sh EXIT STATUS
 .Ex -std
 .Sh STANDARDS
-Except for the lack of support for
-.Fl n ,
 .Nm
 should be compliant with the
 IEEE Std 1003.1-2024 (“POSIX.1”)
diff --git a/cmd/cut.c b/cmd/cut.c
@@ -193,6 +193,64 @@ cut_b(FILE *in, const char *filename)
 }
 
 static int
+cut_b_n(FILE *in, const char *filename)
+{
+	char *line = NULL;
+	size_t line_len = 0;
+	int err = 0;
+
+	while(err == 0)
+	{
+		errno = 0;
+		ssize_t sread = getdelim(&line, &line_len, line_delim, in);
+		if(sread < 0)
+		{
+			if(errno != 0)
+			{
+				fprintf(
+				    stderr, "%s: error: Failed reading file '%s': %s\n", argv0, filename, strerror(errno));
+				err = 1;
+			}
+			break;
+		}
+		size_t nread = (size_t)sread;
+
+		if(nread == 0)
+		{
+			fputc(line_delim, stdout);
+			continue;
+		}
+
+		if(line[nread - 1] == '\n') line[nread--] = '\0';
+
+		for(size_t i = 0; i < MIN(list_len, nread); i++)
+		{
+			size_t isz = mbrlen(line + i, nread, NULL);
+
+			if(isz == 0 || isz == (size_t)-2 || isz == (size_t)-1) continue;
+
+			/*
+			 * Check that last byte is part of the low-high selection.
+			 * Per POSIX.1-2024 high and low only decrements, and otherwise drops the character.
+			 */
+			if(list[i + isz - 1])
+			{
+				fwrite(line + i, isz, 1, stdout);
+				if(isz > 1) i += isz - 1;
+			}
+		}
+
+		if(nostop && nread > list_len) fwrite(line + list_len, nread - list_len, 1, stdout);
+
+		fputc(line_delim, stdout);
+	}
+
+	free(line);
+
+	return err;
+}
+
+static int
 cut_c(FILE *in, const char *filename)
 {
 	char *line = NULL;
@@ -360,7 +418,7 @@ cut(FILE *in, const char *filename)
 		fprintf(stderr, "%s: error: No action (-b, -c, -f) specified\n", argv0);
 		return 1;
 	case CUT_MODE_B:
-		return cut_b(in, filename);
+		return opt_n ? cut_b_n(in, filename) : cut_b(in, filename);
 	case CUT_MODE_C:
 		return cut_c(in, filename);
 	case CUT_MODE_F:
diff --git a/test-cmd/cut.sh b/test-cmd/cut.sh
@@ -2,9 +2,9 @@
 # SPDX-FileCopyrightText: 2017 Haelwenn (lanodan) Monnier <contact+utils@hacktivis.me>
 # SPDX-License-Identifier: MPL-2.0
 
+plans=18
 WD="$(dirname "$0")/../"
 target="${WD}/cmd/cut"
-plans=15
 . "${WD}/test-cmd/tap.sh"
 . "${WD}/test-cmd/init_env.sh"
 
@@ -33,6 +33,15 @@ t 'bytes:-10' "-b -10 ${WD}/test-cmd/inputs/alnum" '0123456789
 '
 t --exit=1 'bytes:,' "-b , ${WD}/test-cmd/inputs/alnum" 'cut: error: Empty list element
 '
+# The cafés tests assumes a UTF-8 environment
+t --input='cafés' 'bytes:cafés[4]' '-b 4' '�
+'
+# per POSIX high only decrements so here it gets dropped
+t --input='cafés' 'bytes:cafés[4]' '-b 4 -n' '
+'
+t --input='cafés' 'bytes:cafés[5]' '-b 5 -n' 'é
+'
+
 # Example taken from POSIX cut(1)
 t --input='abcdefghi' 'chars:6,2,4-7,1' '-c 6,2,4-7,1' 'abdefg
 '