table

чување cb3800d5387475aa5b83c98a334b1976a43039c0
родитељ aa4fdf8bd7342ef2f1850121d5a2733406c6f236
Аутор: Страхиња Радић <sr@strahinja.org>
Датум:   Thu, 16 Apr 2026 10:39:44 +0000

Fix edge cases when the input is binary (discovered by afl-fuzz)

Diffstat:
M
defs.h
 | 
1
+
M
table.1.in
 | 
2
++
M
table.c
 | 
31
+++++++++++++++++++++++++++----
M
utf8.c
 | 
2
+-

измењених датотека: 4, додавања: 31(+), брисања: 5(-)
diff --git a/defs.h b/defs.h
@@ -114,6 +114,7 @@ enum {
 	ERR_NOTREG_FILE,
 	ERR_UNMATCHED_QUOTE,
 	ERR_STREAM_ERROR,
+	ERR_BINARY_INPUT,
 };
 
 enum { Q_PROCESS, Q_IGNORE };
diff --git a/table.1.in b/table.1.in
@@ -260,6 +260,8 @@ File is not a regular file.
 A quote is present without its corresponding closing quote.
 .It Li 211
 There has been an input stream error.
+.It Li 212
+Input is not text (it is binary).
 .El
 .
 .Sh EXAMPLES
diff --git a/table.c b/table.c
@@ -101,6 +101,7 @@ const char* errors[] = {
 	[ERR_NOTREG_FILE]	= "'%s' is not a regular file",
 	[ERR_UNMATCHED_QUOTE]	= "Unmatched quote",
 	[ERR_STREAM_ERROR]	= "Input stream error",
+	[ERR_BINARY_INPUT]	= "Input is not text",
 };
 /* clang-format on */
 
@@ -144,10 +145,13 @@ number_of_columns(const u8* input, u32 delimiter)
 	while (*pinput)
 	{
 		if (u8_rune_to_u32(&uch, pinput, &ch_len))
+		{
 			warning("Malformed UTF-8 at position %td: "
 				"uch=%02X, "
 				"pinput=%02X",
 				pinput - input, uch, *pinput);
+			ch_len = 1;
+		}
 		if (uch == delimiter)
 			result++;
 		pinput += ch_len;
@@ -193,7 +197,9 @@ print_aligned(const u8* s, size_t rune_length, size_t to_size, Alignment align,
 			if (!no_ansi && lineno == 0)
 				fputs(ANSI_SGR_BOLD_ON, stdout);
 
-			u8_rune_to_u32(&uch, ps, &ch_len);
+			if (u8_rune_to_u32(&uch, ps, &ch_len))
+				ch_len = 1;
+
 			output_chars = 0;
 			while (output_chars != ch_len)
 			{
@@ -401,7 +407,7 @@ set_format(const char* arg, int** format, size_t* format_size)
 	{
 		*ptoken = 0;
 		errno	= 0;
-		num = strtonumber(token, INT_MIN, INT_MAX, &errornum);
+		num	= strtonumber(token, INT_MIN, INT_MAX, &errornum);
 		if (stn_handle_error(token, INT_MIN, INT_MAX, errornum))
 		{
 			free(token);
@@ -577,12 +583,14 @@ main(int argc, char** argv)
 	char* sep      = NULL;
 	char* progname = NULL;
 	char* eol      = NULL;
+	char* nulbyte  = NULL;
 	ptrdiff_t outbuf_len;
 	size_t outbufrunelen = 0;
 	size_t ch_len	     = 0;
 	size_t output_lines  = 0;
 	size_t column_start  = 0;
 	int in_quote	     = 0;
+	int possible_nulbyte = 0;
 	int u8len	     = 0;
 
 	UNUSED(argc);
@@ -876,7 +884,8 @@ done_arg:
 	CALLOC(line, u8, line_size);
 	outbuf_size = line_max;
 	CALLOC(outbuf, u8, outbuf_size);
-	pline = line;
+	pline		 = line;
+	possible_nulbyte = 0;
 
 do_input:
 	if (feof(input))
@@ -892,6 +901,11 @@ do_input:
 		goto done_input;
 	}
 
+	if (possible_nulbyte)
+	{
+		error_code = ERR_BINARY_INPUT;
+		goto main_done;
+	}
 	eol = strchr((char*)pline, '\n');
 	if (eol)
 	{
@@ -899,7 +913,16 @@ do_input:
 			*(eol - 1) = 0;
 		*eol = 0;
 	}
-	else
+
+	nulbyte = strchr((char*)pline, 0);
+
+	/* We don't know if the NUL byte not being equal to
+	 * (possibly NULL) eol is the result of EOF or not
+	 */
+	if (nulbyte && (nulbyte != eol))
+		possible_nulbyte = 1;
+
+	if (!eol)
 	{
 		line_size += line_max;
 		REALLOC(line, u8, line_size);
diff --git a/utf8.c b/utf8.c
@@ -54,7 +54,7 @@ u8_rune_to_u32(u32* to, const u8* from, size_t* from_delta)
 	*to	    = *from & extract_masks[len - 1];
 	for (int i = 1; i < len; i++)
 	{
-		pfrom = from + i;
+		pfrom = &from[i];
 		if (!*pfrom)
 			return 1;
 		*to <<= 6;