table

чување d1c6ab6711445cba5ffb974c8d65b4b3c0a855b5
родитељ cb3800d5387475aa5b83c98a334b1976a43039c0
Аутор: Страхиња Радић <sr@strahinja.org>
Датум:   Sat, 18 Apr 2026 07:50:33 +0000

Handle malformed UTF-8 sequences better (replace with FFFD)

Diffstat:
M
table.c
 | 
75
+++++++++++++++++++++++++++++++++++++++++++++++++++++++--------------------
M
utf8.c
 | 
4
++++
M
utf8.h
 | 
2
++

измењених датотека: 3, додавања: 61(+), брисања: 20(-)
diff --git a/table.c b/table.c
@@ -147,10 +147,8 @@ number_of_columns(const u8* input, u32 delimiter)
 		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;
+				"from=%02X",
+				pinput - input, *pinput);
 		}
 		if (uch == delimiter)
 			result++;
@@ -198,7 +196,11 @@ print_aligned(const u8* s, size_t rune_length, size_t to_size, Alignment align,
 				fputs(ANSI_SGR_BOLD_ON, stdout);
 
 			if (u8_rune_to_u32(&uch, ps, &ch_len))
-				ch_len = 1;
+			{
+				warning("Malformed UTF-8 at position %td: "
+					"from=%02X",
+					ps - s, *ps);
+			}
 
 			output_chars = 0;
 			while (output_chars != ch_len)
@@ -218,7 +220,12 @@ print_aligned(const u8* s, size_t rune_length, size_t to_size, Alignment align,
 			skip_while_sgr(ps, &skipped_chars, 1, 0);
 			ps += skipped_chars;
 
-			u8_rune_to_u32(&uch, ps, &ch_len);
+			if (u8_rune_to_u32(&uch, ps, &ch_len))
+			{
+				warning("Malformed UTF-8 at position %td: "
+					"from=%02X",
+					ps - s, *ps);
+			}
 			ps += ch_len;
 			skipped_runes++;
 		}
@@ -232,7 +239,12 @@ 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))
+			{
+				warning("Malformed UTF-8 at position %td: "
+					"from=%02X",
+					ps - s, *ps);
+			}
 			output_chars = 0;
 			while (output_chars != ch_len)
 			{
@@ -251,7 +263,12 @@ print_aligned(const u8* s, size_t rune_length, size_t to_size, Alignment align,
 			skip_while_sgr(ps, &skipped_chars, 1, 0);
 			ps += skipped_chars;
 
-			u8_rune_to_u32(&uch, ps, &ch_len);
+			if (u8_rune_to_u32(&uch, ps, &ch_len))
+			{
+				warning("Malformed UTF-8 at position %td: "
+					"from=%02X",
+					ps - s, *ps);
+			}
 			ps += ch_len;
 			skipped_runes++;
 		}
@@ -264,7 +281,12 @@ 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))
+			{
+				warning("Malformed UTF-8 at position %td: "
+					"from=%02X",
+					ps - s, *ps);
+			}
 			output_chars = 0;
 			while (output_chars != ch_len)
 			{
@@ -343,8 +365,8 @@ set_delimiter(const char* arg, u32* delimiter)
 	size_t delimiter_len = 0;
 	u32 uch;
 	assert((arg != NULL) && (delimiter != NULL));
-	u8_rune_to_u32(&uch, (u8*)arg, &delimiter_len);
-	*delimiter = uch;
+	if (!u8_rune_to_u32(&uch, (u8*)arg, &delimiter_len))
+		*delimiter = uch;
 	return 0;
 }
 
@@ -978,6 +1000,20 @@ do_input:
 		format_value
 			= (rune_columns - table_columns - 2) / table_columns;
 
+	if (format_value < 1)
+	{
+		warning("Impossible to fit %ld table columns into %ld rune "
+			"columns (%ld-%ld-2)/%ld=%d), adjusting rune columns "
+			"to %ld",
+			table_columns, rune_columns, rune_columns,
+			table_columns, table_columns, format_value,
+			2 * table_columns + 2);
+		rune_columns = 2 * table_columns + 2;
+		format_value
+			= (rune_columns - table_columns - 2) / table_columns;
+	}
+	assert(format_value > 0);
+
 	current_table_column = 0;
 	column_start	     = 0;
 
@@ -1117,7 +1153,12 @@ continue_line:
 		(ssize_t)(poutbuf - outbuf + sizeof((u8*)ANSI_SGR_BOLD_OFF)),
 		(ssize_t)(outbuf_size + sizeof((u8*)ANSI_SGR_BOLD_OFF)), u8);
 	poutbuf = outbuf + outbuf_len;
-	u8_rune_to_u32(&uch, pline, &ch_len);
+	if (u8_rune_to_u32(&uch, pline, &ch_len))
+	{
+		warning("Malformed UTF-8 at position %td: "
+			"from=%02X",
+			pline - line, *pline);
+	}
 	if (uch == U32('\033') && *(pline + 1) == '[')
 	{
 		in_sgr = 1;
@@ -1206,14 +1247,6 @@ continue_line:
 		pline += ch_len;
 		colno += ch_len;
 	}
-	else if (!ch_len) /* UTF-8 error */
-	{
-		*poutbuf++ = *pline;
-		*poutbuf   = 0;
-		outbufrunelen++;
-		pline++;
-		colno++;
-	}
 	else if (uch == TAB && expand_tabs)
 	{
 		while ((current_rune_column + outbufrunelen) % tab_length != 0)
@@ -1247,6 +1280,8 @@ done_line:
 		size_t colwidth = format ? *(format + current_table_column)
 					 : format_value;
 
+		assert(colwidth != 0);
+
 		if (!no_ansi && lineno == 0)
 			snprintf((char*)poutbuf, sizeof ANSI_SGR_BOLD_OFF, "%s",
 				ANSI_SGR_BOLD_OFF);
diff --git a/utf8.c b/utf8.c
@@ -56,7 +56,11 @@ u8_rune_to_u32(u32* to, const u8* from, size_t* from_delta)
 	{
 		pfrom = &from[i];
 		if (!*pfrom)
+		{
+			*from_delta = 1;
+			*to	    = U32_REPL_CHAR;
 			return 1;
+		}
 		*to <<= 6;
 		*to |= *pfrom & XMASKR;
 	}
diff --git a/utf8.h b/utf8.h
@@ -37,6 +37,8 @@
 typedef uint8_t u8;
 typedef uint32_t u32;
 
+#define U32_REPL_CHAR ((u32)0xFFFD)
+
 int u8_rune_to_u32(u32* to, const u8* from, size_t* from_delta);
 int u32_rune_to_u8(u8* to, const u32 from); /* max == 1 */
 int u8_to_u32(u32* to, const u8* from, const ssize_t max, size_t* from_delta);