column: use \x<hex> for invalid multibyte seq.

Addresses: https://github.com/karelzak/util-linux/issues/542
Signed-off-by: Karel Zak <kzak@redhat.com>

2 files changed, 15 insertions, 3 deletions

diff --git a/text-utils/column.1 b/text-utils/column.1
index 6baa0464a..c6da5615c 100644
--- a/text-utils/column.1
+++ b/text-utils/column.1
@@ -54,7 +54,7 @@ mode is enabled by option \fB-t, \-\-table\fP and columns formatting is
 possible to modify by \fB\-\-table-*\fP options.  Use this mode if not sure.
 .PP
 Input is taken from \fIfile\fR, or otherwise from standard input.  Empty lines
-are ignored.
+are ignored and all invalid multibyte sequences are encoded by \\x<hex> convention.
 .PP
 .SH OPTIONS
 The argument \fIcolumns\fP for \fB\-\-table-*\fP options is comma separated
diff --git a/text-utils/column.c b/text-utils/column.c
index d34cfc3f0..88122a0e7 100644
--- a/text-utils/column.c
+++ b/text-utils/column.c
@@ -52,6 +52,7 @@
 #include "ttyutils.h"
 #include "strv.h"
 #include "optutils.h"
+#include "mbsalign.h"
 
 #include "libsmartcols.h"
 
@@ -458,8 +459,18 @@ static int read_input(struct column_control *ctl, FILE *fp)
 			continue;
 
 		wcs = mbs_to_wcs(str);
-		if (!wcs)
-			err(EXIT_FAILURE, _("read failed"));
+		if (!wcs) {
+			/*
+			 * Convert broken sequences to \x<hex> and continue.
+			 */
+			size_t tmpsz = 0;
+			char *tmp = mbs_invalid_encode(str, &tmpsz);
+
+			if (!tmp)
+				err(EXIT_FAILURE, _("read failed"));
+			wcs = mbs_to_wcs(tmp);
+			free(tmp);
+		}
 
 		switch (ctl->mode) {
 		case COLUMN_MODE_TABLE:
@@ -590,6 +601,7 @@ static void __attribute__((__noreturn__)) usage(void)
 	fputs(_(" -s, --separator <string>         possible table delimiters\n"), out);
 	fputs(_(" -x, --fillrows                   fill rows before columns\n"), out);
 
+
 	fputs(USAGE_SEPARATOR, out);
 	printf(USAGE_HELP_OPTIONS(34));
 	printf(USAGE_MAN_TAIL("column(1)"));