Add identify filter for UCS-2, UCS-2BE, and UCS-2LE encodings

Author: alexdowad

ext/mbstring/libmbfl/filters/mbfilter_ucs2.c

@@ -30,6 +30,8 @@
 #include "mbfilter.h"
 #include "mbfilter_ucs2.h"
 
+static int mbfl_filt_ident_ucs2(int c, mbfl_identify_filter *filter);
+
 static const char *mbfl_encoding_ucs2_aliases[] = {"ISO-10646-UCS-2", "UCS2" , "UNICODE", NULL};
 
 const mbfl_encoding mbfl_encoding_ucs2 = {
@@ -65,6 +67,24 @@ const mbfl_encoding mbfl_encoding_ucs2le = {
 	&vtbl_wchar_ucs2le
 };
 
+const struct mbfl_identify_vtbl vtbl_identify_ucs2 = {
+	mbfl_no_encoding_ucs2,
+	mbfl_filt_ident_common_ctor,
+	mbfl_filt_ident_ucs2
+};
+
+const struct mbfl_identify_vtbl vtbl_identify_ucs2be = {
+	mbfl_no_encoding_ucs2be,
+	mbfl_filt_ident_common_ctor,
+	mbfl_filt_ident_ucs2
+};
+
+const struct mbfl_identify_vtbl vtbl_identify_ucs2le = {
+	mbfl_no_encoding_ucs2le,
+	mbfl_filt_ident_common_ctor,
+	mbfl_filt_ident_ucs2
+};
+
 const struct mbfl_convert_vtbl vtbl_ucs2_wchar = {
 	mbfl_no_encoding_ucs2,
 	mbfl_no_encoding_wchar,
@@ -237,3 +257,10 @@ int mbfl_filt_conv_wchar_ucs2le(int c, mbfl_convert_filter *filter)
 
 	return c;
 }
+
+static int mbfl_filt_ident_ucs2(int c, mbfl_identify_filter *filter)
+{
+	/* Input string must be a multiple of 2 bytes */
+	filter->status = (filter->status + 1) % 2;
+	return c;
+}

ext/mbstring/libmbfl/filters/mbfilter_ucs2.h

@@ -35,6 +35,9 @@
 extern const mbfl_encoding mbfl_encoding_ucs2;
 extern const mbfl_encoding mbfl_encoding_ucs2be;
 extern const mbfl_encoding mbfl_encoding_ucs2le;
+extern const struct mbfl_identify_vtbl vtbl_identify_ucs2;
+extern const struct mbfl_identify_vtbl vtbl_identify_ucs2be;
+extern const struct mbfl_identify_vtbl vtbl_identify_ucs2le;
 extern const struct mbfl_convert_vtbl vtbl_ucs2_wchar;
 extern const struct mbfl_convert_vtbl vtbl_wchar_ucs2;
 extern const struct mbfl_convert_vtbl vtbl_ucs2be_wchar;

ext/mbstring/libmbfl/mbfl/mbfl_ident.c

@@ -164,6 +164,9 @@ static const struct mbfl_identify_vtbl *mbfl_identify_filter_list[] = {
 	&vtbl_identify_utf16le,
 	&vtbl_identify_utf16be,
 	&vtbl_identify_8bit,
+	&vtbl_identify_ucs2,
+	&vtbl_identify_ucs2be,
+	&vtbl_identify_ucs2le,
 	&vtbl_identify_false,
 	NULL
 };