Major overhaul of mbstring (part 25)

ext/mbstring/libmbfl/filters/mbfilter_7bit.c

@@ -34,21 +34,18 @@
 static size_t mb_7bit_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state);
 static void mb_wchar_to_7bit(uint32_t *in, size_t len, mb_convert_buf *buf, bool end);
 
-const mbfl_encoding mbfl_encoding_7bit = {
+const struct mbfl_convert_vtbl vtbl_7bit_wchar = {
 	mbfl_no_encoding_7bit,
-	"7bit",
-	"7bit",
-	NULL,
-	NULL,
-	MBFL_ENCTYPE_SBCS,
+	mbfl_no_encoding_wchar,
+	mbfl_filt_conv_common_ctor,
 	NULL,
+	mbfl_filt_conv_7bit_any,
+	mbfl_filt_conv_common_flush,
 	NULL,
-	mb_7bit_to_wchar,
-	mb_wchar_to_7bit
 };
 
-const struct mbfl_convert_vtbl vtbl_8bit_7bit = {
-	mbfl_no_encoding_8bit,
+const struct mbfl_convert_vtbl vtbl_wchar_7bit = {
+	mbfl_no_encoding_wchar,
 	mbfl_no_encoding_7bit,
 	mbfl_filt_conv_common_ctor,
 	NULL,
@@ -57,25 +54,26 @@ const struct mbfl_convert_vtbl vtbl_8bit_7bit = {
 	NULL,
 };
 
-const struct mbfl_convert_vtbl vtbl_7bit_8bit = {
+const mbfl_encoding mbfl_encoding_7bit = {
 	mbfl_no_encoding_7bit,
-	mbfl_no_encoding_8bit,
-	mbfl_filt_conv_common_ctor,
+	"7bit",
+	"7bit",
 	NULL,
-	mbfl_filt_conv_7bit_any,
-	mbfl_filt_conv_common_flush,
 	NULL,
+	MBFL_ENCTYPE_SBCS,
+	&vtbl_7bit_wchar,
+	&vtbl_wchar_7bit,
+	mb_7bit_to_wchar,
+	mb_wchar_to_7bit
 };
 
-
 #define CK(statement)	do { if ((statement) < 0) return (-1); } while (0)
 
 int mbfl_filt_conv_7bit_any(int c, mbfl_convert_filter *filter)
 {
 	return (*filter->output_function)(c < 0x80 ? c : MBFL_BAD_INPUT, filter->data);
 }
 
-
 int mbfl_filt_conv_any_7bit(int c, mbfl_convert_filter *filter)
 {
 	if (c >= 0 && c < 0x80) {

ext/mbstring/libmbfl/filters/mbfilter_base64.c

@@ -337,6 +337,10 @@ static void mb_wchar_to_base64(uint32_t *in, size_t len, mb_convert_buf *buf, bo
 	}
 
 	if (end && bits) {
+		if (chars_output > 72) {
+			out = mb_convert_buf_add2(out, '\r', '\n');
+			chars_output = 0;
+		}
 		if (bits == 8) {
 			out = mb_convert_buf_add4(out, mbfl_base64_table[(cache >> 2) & 0x3F], mbfl_base64_table[(cache & 0x3) << 4], '=', '=');
 		} else {

ext/mbstring/libmbfl/filters/mbfilter_big5.c

@@ -257,6 +257,7 @@ static int mbfl_filt_conv_big5_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status == 1) {
 		/* 2-byte character was truncated */
+		filter->status = 0;
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
 

ext/mbstring/libmbfl/filters/mbfilter_cp5022x.c

@@ -322,6 +322,7 @@ static int mbfl_filt_conv_cp5022x_wchar_flush(mbfl_convert_filter *filter)
 		 * escape sequence was truncated */
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
+	filter->status = 0;
 
 	if (filter->flush_function) {
 		(*filter->flush_function)(filter->data);
@@ -555,7 +556,9 @@ static int mbfl_filt_conv_wchar_cp50220_flush(mbfl_convert_filter *filter)
 
 	if (filter->cache) {
 		int s = mb_convert_kana_codepoint(filter->cache, 0, NULL, NULL, mode);
+		filter->filter_function = mbfl_filt_conv_wchar_cp50221;
 		mbfl_filt_conv_wchar_cp50221(s, filter);
+		filter->filter_function = mbfl_filt_conv_wchar_cp50220;
 		filter->cache = 0;
 	}
 
@@ -822,7 +825,7 @@ static int mbfl_filt_conv_wchar_cp50222_flush(mbfl_convert_filter *filter)
 		CK((*filter->output_function)(0x28, filter->data));		/* '(' */
 		CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
 	}
-	filter->status &= 0xff;
+	filter->status = 0;
 
 	if (filter->flush_function) {
 		(*filter->flush_function)(filter->data);
@@ -1041,7 +1044,6 @@ static void mb_wchar_to_cp50220(uint32_t *in, size_t len, mb_convert_buf *buf, b
 	MB_CONVERT_BUF_LOAD(buf, out, limit);
 	MB_CONVERT_BUF_ENSURE(buf, out, limit, len);
 
-	bool consumed = false;
 	uint32_t w;
 
 	if (buf->state & 0xFFFF00) {
@@ -1060,10 +1062,10 @@ static void mb_wchar_to_cp50220(uint32_t *in, size_t len, mb_convert_buf *buf, b
 			 * but the 'next one' will come in a separate buffer */
 			buf->state |= w << 8;
 			break;
-		} else {
-			w = mb_convert_kana_codepoint(w, len ? *in : 0, &consumed, NULL, MBFL_HAN2ZEN_KATAKANA | MBFL_HAN2ZEN_GLUE);
 		}
 
+		bool consumed = false;
+		w = mb_convert_kana_codepoint(w, len ? *in : 0, &consumed, NULL, MBFL_HAN2ZEN_KATAKANA | MBFL_HAN2ZEN_GLUE);
 		if (consumed) {
 			/* Two successive codepoints were converted into one */
 			in++; len--; consumed = false;

ext/mbstring/libmbfl/filters/mbfilter_cp51932.c

@@ -178,6 +178,7 @@ static int mbfl_filt_conv_cp51932_wchar_flush(mbfl_convert_filter *filter)
 	if (filter->status) {
 		/* Input string was truncated */
 		(*filter->output_function)(MBFL_BAD_INPUT, filter->data);
+		filter->status = 0;
 	}
 
 	if (filter->flush_function) {

ext/mbstring/libmbfl/filters/mbfilter_cp932.c

@@ -217,6 +217,7 @@ static int mbfl_filt_conv_cp932_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status) {
 		(*filter->output_function)(MBFL_BAD_INPUT, filter->data);
+		filter->status = 0;
 	}
 
 	if (filter->flush_function) {

ext/mbstring/libmbfl/filters/mbfilter_cp936.c

@@ -166,6 +166,7 @@ static int mbfl_filt_conv_cp936_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status) {
 		/* 2-byte character was truncated */
+		filter->status = 0;
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
 

ext/mbstring/libmbfl/filters/mbfilter_euc_cn.c

@@ -209,6 +209,7 @@ static int mbfl_filt_conv_euccn_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status == 1) {
 		/* 2-byte character was truncated */
+		filter->status = 0;
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
 

ext/mbstring/libmbfl/filters/mbfilter_euc_jp.c

@@ -180,6 +180,7 @@ static int mbfl_filt_conv_eucjp_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status) {
 		(*filter->output_function)(MBFL_BAD_INPUT, filter->data);
+		filter->status = 0;
 	}
 
 	if (filter->flush_function) {

ext/mbstring/libmbfl/filters/mbfilter_euc_jp_win.c

@@ -226,6 +226,7 @@ static int mbfl_filt_conv_eucjpwin_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status) {
 		(*filter->output_function)(MBFL_BAD_INPUT, filter->data);
+		filter->status = 0;
 	}
 
 	if (filter->flush_function) {

ext/mbstring/libmbfl/filters/mbfilter_euc_kr.c

@@ -193,6 +193,7 @@ static int mbfl_filt_conv_euckr_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status == 1) {
 		/* 2-byte character was truncated */
+		filter->status = 0;
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
 

ext/mbstring/libmbfl/filters/mbfilter_euc_tw.c

@@ -245,6 +245,7 @@ static int mbfl_filt_conv_euctw_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status) {
 		/* 2-byte or 4-byte character was truncated */
+		filter->status = 0;
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
 

ext/mbstring/libmbfl/filters/mbfilter_gb18030.c

@@ -231,6 +231,7 @@ static int mbfl_filt_conv_gb18030_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status) {
 		/* multi-byte character was truncated */
+		filter->status = 0;
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
 

ext/mbstring/libmbfl/filters/mbfilter_htmlent.c

@@ -180,7 +180,7 @@ int mbfl_filt_conv_html_dec(int c, mbfl_convert_filter *filter)
 	int pos;
 	unsigned int ent = 0;
 	mbfl_html_entity_entry *entity;
-	char *buffer = (char*)filter->opaque;
+	unsigned char *buffer = (unsigned char*)filter->opaque;
 
 	if (!filter->status) {
 		if (c == '&' ) {
@@ -196,7 +196,7 @@ int mbfl_filt_conv_html_dec(int c, mbfl_convert_filter *filter)
 					if (filter->status > 3) {
 						/* numeric entity */
 						for (pos=3; pos<filter->status; pos++) {
-							int v =  buffer[pos];
+							int v = buffer[pos];
 							if (v >= '0' && v <= '9') {
 								v = v - '0';
 							} else if (v >= 'A' && v <= 'F') {
@@ -242,13 +242,12 @@ int mbfl_filt_conv_html_dec(int c, mbfl_convert_filter *filter)
 					CK((*filter->output_function)(c, filter->data));
 				}
 				filter->status = 0;
-				/*php_error_docref("ref.mbstring", E_NOTICE, "mbstring decoded '%s'=%d", buffer, ent);*/
 			} else {
 				/* named entity */
 				buffer[filter->status] = 0;
 				entity = (mbfl_html_entity_entry *)mbfl_html_entity_list;
 				while (entity->name) {
-					if (!strcmp(buffer+1, entity->name))	{
+					if (!strcmp((const char*)buffer+1, entity->name)) {
 						ent = entity->code;
 						break;
 					}

ext/mbstring/libmbfl/filters/mbfilter_hz.c

@@ -154,6 +154,8 @@ static int mbfl_filt_conv_hz_wchar_flush(mbfl_convert_filter *filter)
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
 
+	filter->status = 0;
+
 	if (filter->flush_function) {
 		(*filter->flush_function)(filter->data);
 	}

ext/mbstring/libmbfl/filters/mbfilter_iso2022_jp_ms.c

@@ -219,6 +219,7 @@ static int mbfl_filt_conv_2022jpms_wchar_flush(mbfl_convert_filter *filter)
 	if (filter->status & 0xF) {
 		(*filter->output_function)(MBFL_BAD_INPUT, filter->data);
 	}
+	filter->status = 0;
 
 	if (filter->flush_function) {
 		(*filter->flush_function)(filter->data);
@@ -354,6 +355,7 @@ int mbfl_filt_conv_any_2022jpms_flush(mbfl_convert_filter *filter)
 		CK((*filter->output_function)('(', filter->data));
 		CK((*filter->output_function)('B', filter->data));
 	}
+	filter->status = 0;
 
 	if (filter->flush_function) {
 		(*filter->flush_function)(filter->data);

ext/mbstring/libmbfl/filters/mbfilter_iso2022_kr.c

@@ -145,7 +145,7 @@ int mbfl_filt_conv_2022kr_wchar(int c, mbfl_convert_filter *filter)
 		if (c == '$') {
 			filter->status++;
 		} else {
-			filter->status = 0;
+			filter->status &= ~0xF;
 			CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 		}
 		break;
@@ -154,7 +154,7 @@ int mbfl_filt_conv_2022kr_wchar(int c, mbfl_convert_filter *filter)
 		if (c == ')') {
 			filter->status++;
 		} else {
-			filter->status = 0;
+			filter->status &= ~0xF;
 			CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 		}
 		break;
@@ -178,6 +178,7 @@ static int mbfl_filt_conv_2022kr_wchar_flush(mbfl_convert_filter *filter)
 		/* 2-byte character was truncated */
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
+	filter->status = 0;
 
 	if (filter->flush_function) {
 		(*filter->flush_function)(filter->data);
@@ -257,6 +258,10 @@ int mbfl_filt_conv_wchar_2022kr(int c, mbfl_convert_filter *filter)
 
 static int mbfl_filt_conv_any_2022kr_flush(mbfl_convert_filter *filter)
 {
+	if (filter->status & 0xF) {
+		/* Escape sequence or 2-byte character was truncated */
+		(*filter->output_function)(MBFL_BAD_INPUT, filter->data);
+	}
 	/* back to ascii */
 	if (filter->status & 0x10) {
 		CK((*filter->output_function)(0x0f, filter->data)); /* shift in */
@@ -304,7 +309,6 @@ static size_t mb_iso2022kr_to_wchar(unsigned char **in, size_t *in_len, uint32_t
 						p--;
 				}
 				*out++ = MBFL_BAD_INPUT;
-				*state = ASCII;
 			}
 		} else if (c == 0xF) {
 			*state = ASCII;

ext/mbstring/libmbfl/filters/mbfilter_iso2022jp_mobile.c

@@ -313,6 +313,7 @@ static int mbfl_filt_conv_2022jp_mobile_wchar_flush(mbfl_convert_filter *filter)
 	if (filter->status & 0xF) {
 		(*filter->output_function)(MBFL_BAD_INPUT, filter->data);
 	}
+	filter->status = 0;
 
 	if (filter->flush_function) {
 		(*filter->flush_function)(filter->data);
@@ -483,6 +484,7 @@ static int mbfl_filt_conv_wchar_2022jp_mobile_flush(mbfl_convert_filter *filter)
 	if ((filter->status & 0xFF) == 1 && (c1 == '#' || (c1 >= '0' && c1 <= '9'))) {
 		(*filter->output_function)(c1, filter->data);
 	}
+	filter->status = filter->cache = 0;
 
 	if (filter->flush_function) {
 		(*filter->flush_function)(filter->data);

ext/mbstring/libmbfl/filters/mbfilter_jis.c

@@ -271,6 +271,7 @@ static int mbfl_filt_conv_jis_wchar_flush(mbfl_convert_filter *filter)
 		 * or else escape sequence was truncated */
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
+	filter->status = 0;
 
 	if (filter->flush_function) {
 		(*filter->flush_function)(filter->data);
@@ -451,7 +452,7 @@ mbfl_filt_conv_any_jis_flush(mbfl_convert_filter *filter)
 		CK((*filter->output_function)(0x28, filter->data));		/* '(' */
 		CK((*filter->output_function)(0x42, filter->data));		/* 'B' */
 	}
-	filter->status &= 0xff;
+	filter->status = 0;
 
 	if (filter->flush_function != NULL) {
 		return (*filter->flush_function)(filter->data);
@@ -480,6 +481,8 @@ static size_t mb_iso2022jp_to_wchar(unsigned char **in, size_t *in_len, uint32_t
 			/* ESC seen; this is an escape sequence */
 			if ((e - p) < 2) {
 				*out++ = MBFL_BAD_INPUT;
+				if (p != e && (*p == '$' || *p == '('))
+					p++;
 				continue;
 			}
 

ext/mbstring/libmbfl/filters/mbfilter_sjis.c

@@ -183,6 +183,7 @@ static int mbfl_filt_conv_sjis_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status) {
 		(*filter->output_function)(MBFL_BAD_INPUT, filter->data);
+		filter->status = 0;
 	}
 
 	if (filter->flush_function) {

ext/mbstring/libmbfl/filters/mbfilter_sjis_2004.c

@@ -491,6 +491,7 @@ int mbfl_filt_conv_jis2004_wchar_flush(mbfl_convert_filter *filter)
 	if (filter->status & 0xF) {
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}
+	filter->status = 0;
 
 	if (filter->flush_function) {
 		return (*filter->flush_function)(filter->data);
@@ -633,6 +634,8 @@ int mbfl_filt_conv_wchar_jis2004(int c, mbfl_convert_filter *filter)
 			if (filter->to->no_encoding == mbfl_no_encoding_eucjp2004) {
 				CK((*filter->output_function)(0x8e, filter->data));
 				CK((*filter->output_function)(s1, filter->data));
+			} else if (filter->to->no_encoding == mbfl_no_encoding_sjis2004 && (s1 >= 0xA1 && s1 <= 0xDF)) {
+				CK((*filter->output_function)(s1, filter->data));
 			} else {
 				CK(mbfl_filt_conv_illegal_output(c, filter));
 			}

ext/mbstring/libmbfl/filters/mbfilter_sjis_mac.c

@@ -266,6 +266,7 @@ mbfl_filt_conv_sjis_mac_wchar(int c, mbfl_convert_filter *filter)
 static int mbfl_filt_conv_sjis_mac_wchar_flush(mbfl_convert_filter *filter)
 {
 	if (filter->status == 1) {
+		filter->status = 0;
 		CK((*filter->output_function)(MBFL_BAD_INPUT, filter->data));
 	}