JIS7/JIS8 encoding: treat truncated multibyte characters as error

Author: alexdowad

ext/mbstring/libmbfl/filters/mbfilter_jis.c

@@ -33,6 +33,8 @@
 #include "unicode_table_cp932_ext.h"
 #include "unicode_table_jis.h"
 
+static int mbfl_filt_conv_jis_wchar_flush(mbfl_convert_filter *filter);
+
 const mbfl_encoding mbfl_encoding_jis = {
 	mbfl_no_encoding_jis,
 	"JIS",
@@ -61,7 +63,7 @@ const struct mbfl_convert_vtbl vtbl_jis_wchar = {
 	mbfl_filt_conv_common_ctor,
 	NULL,
 	mbfl_filt_conv_jis_wchar,
-	mbfl_filt_conv_common_flush,
+	mbfl_filt_conv_jis_wchar_flush,
 	NULL,
 };
 
@@ -81,7 +83,7 @@ const struct mbfl_convert_vtbl vtbl_2022jp_wchar = {
 	mbfl_filt_conv_common_ctor,
 	NULL,
 	mbfl_filt_conv_jis_wchar,
-	mbfl_filt_conv_common_flush,
+	mbfl_filt_conv_jis_wchar_flush,
 	NULL,
 };
 
@@ -264,6 +266,15 @@ mbfl_filt_conv_jis_wchar(int c, mbfl_convert_filter *filter)
 	return c;
 }
 
+static int mbfl_filt_conv_jis_wchar_flush(mbfl_convert_filter *filter)
+{
+	if ((filter->status & 0xF) == 1) {
+		/* 2-byte (JIS X 0208 or 0212) character was truncated */
+		CK((*filter->output_function)(filter->cache | MBFL_WCSGROUP_THROUGH, filter->data));
+	}
+	return 0;
+}
+
 /*
  * wchar => JIS
  */

ext/mbstring/tests/iso2022jp_encoding.phpt

@@ -125,6 +125,12 @@ for ($i = 0x21; $i <= 0x7E; $i++) {
 	}
 }
 
+/* Try truncated JISX0208 characters */
+for ($i = 0x21; $i <= 0x7E; $i++) {
+	testInvalid("\x1B\$B" . chr($i), "\x00%", 'JIS');
+	testInvalid("\x1B\$B" . chr($i), "\x00%", 'ISO-2022-JP');
+}
+
 echo "JIS X 0208 support OK\n";
 
 /* JIS7 supports escape to switch to JIS X 0212 charset, but ISO-2022-JP does not */
@@ -144,6 +150,11 @@ for ($i = 0x21; $i <= 0x7E; $i++) {
 	}
 }
 
+/* Try truncated JISX0212 characters */
+for ($i = 0x21; $i <= 0x7E; $i++) {
+	testInvalid("\x1B\$(D" . chr($i), "\x00%", 'JIS');
+}
+
 echo "JIS X 0212 support OK\n";
 
 /* All possible escape sequences */