Implement fast text conversion interface for EUC-CN

Author: alexdowad

ext/mbstring/libmbfl/filters/mbfilter_euc_cn.c

@@ -33,6 +33,8 @@
 #include "unicode_table_cp936.h"
 
 static int mbfl_filt_conv_euccn_wchar_flush(mbfl_convert_filter *filter);
+static size_t mb_euccn_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state);
+static void mb_wchar_to_euccn(uint32_t *in, size_t len, mb_convert_buf *buf, bool end);
 
 static const unsigned char mblen_table_euccn[] = { /* 0xA1-0xFE */
   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
@@ -64,8 +66,8 @@ const mbfl_encoding mbfl_encoding_euc_cn = {
 	0,
 	&vtbl_euccn_wchar,
 	&vtbl_wchar_euccn,
-	NULL,
-	NULL
+	mb_euccn_to_wchar,
+	mb_wchar_to_euccn
 };
 
 const struct mbfl_convert_vtbl vtbl_euccn_wchar = {
@@ -216,3 +218,107 @@ static int mbfl_filt_conv_euccn_wchar_flush(mbfl_convert_filter *filter)
 
 	return 0;
 }
+
+static size_t mb_euccn_to_wchar(unsigned char **in, size_t *in_len, uint32_t *buf, size_t bufsize, unsigned int *state)
+{
+	unsigned char *p = *in, *e = p + *in_len;
+	uint32_t *out = buf, *limit = buf + bufsize;
+
+	while (p < e && out < limit) {
+		unsigned char c = *p++;
+
+		if (c < 0x80) {
+			*out++ = c;
+		} else if (((c >= 0xA1 && c <= 0xA9) || (c >= 0xB0 && c <= 0xF7)) && p < e) {
+			unsigned char c2 = *p++;
+
+			if (c2 >= 0xA1 && c2 <= 0xFE) {
+				unsigned int w = (c - 0x81)*192 + c2 - 0x40;
+				ZEND_ASSERT(w < cp936_ucs_table_size);
+				if (w == 0x1864) {
+					w = 0x30FB;
+				} else if (w == 0x186A) {
+					w = 0x2015;
+				} else if ((w >= 0x1921 && w <= 0x192A) || w == 0x1963 || (w >= 0x1C59 && w <= 0x1C7E) || (w >= 0x1DBB && w <= 0x1DC4)) {
+					w = 0;
+				} else {
+					w = cp936_ucs_table[w];
+				}
+
+				if (!w)
+					w = MBFL_BAD_INPUT;
+				*out++ = w;
+			} else {
+				*out++ = MBFL_BAD_INPUT;
+			}
+		} else {
+			*out++ = MBFL_BAD_INPUT;
+		}
+	}
+
+	*in_len = e - p;
+	*in = p;
+	return out - buf;
+}
+
+static void mb_wchar_to_euccn(uint32_t *in, size_t len, mb_convert_buf *buf, bool end)
+{
+	unsigned char *out, *limit;
+	MB_CONVERT_BUF_LOAD(buf, out, limit);
+	MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 2);
+
+	while (len--) {
+		uint32_t w = *in++;
+		unsigned int s = 0;
+
+		if (w >= ucs_a1_cp936_table_min && w < ucs_a1_cp936_table_max) {
+			if (w != 0xB7 && w != 0x144 && w != 0x148 && w != 0x251 && w != 0x261) {
+				s = ucs_a1_cp936_table[w - ucs_a1_cp936_table_min];
+			}
+		} else if (w >= ucs_a2_cp936_table_min && w < ucs_a2_cp936_table_max) {
+			if (w == 0x2015) {
+				s = 0xA1AA;
+			} else if (w != 0x2014 && (w < 0x2170 || w > 0x2179)) {
+				s = ucs_a2_cp936_table[w - ucs_a2_cp936_table_min];
+			}
+		} else if (w >= ucs_a3_cp936_table_min && w < ucs_a3_cp936_table_max) {
+			if (w == 0x30FB) {
+				s = 0xA1A4;
+			} else {
+				s = ucs_a3_cp936_table[w - ucs_a3_cp936_table_min];
+			}
+		} else if (w >= ucs_i_cp936_table_min && w < ucs_i_cp936_table_max) {
+			s = ucs_i_cp936_table[w - ucs_i_cp936_table_min];
+		} else if (w >= ucs_hff_cp936_table_min && w < ucs_hff_cp936_table_max) {
+			if (w == 0xFF04) {
+				s = 0xA1E7;
+			} else if (w == 0xFF5E) {
+				s = 0xA1AB;
+			} else if (w >= 0xFF01 && w <= 0xFF5D) {
+				s = w - 0xFF01 + 0xA3A1;
+			} else if (w >= 0xFFE0 && w <= 0xFFE5) {
+				s = ucs_hff_s_cp936_table[w - 0xFFE0];
+			}
+		}
+
+		/* Exclude CP936 extensions */
+		if (((s >> 8) & 0xFF) < 0xA1 || (s & 0xFF) < 0xA1) {
+			s = 0;
+		}
+
+		if (!s) {
+			if (w < 0x80) {
+				out = mb_convert_buf_add(out, w);
+			} else {
+				MB_CONVERT_ERROR(buf, out, limit, w, mb_wchar_to_euccn);
+				MB_CONVERT_BUF_ENSURE(buf, out, limit, len * 2);
+			}
+		} else if (s < 0x80) {
+			out = mb_convert_buf_add(out, s);
+		} else {
+			out = mb_convert_buf_add2(out, (s >> 8) & 0xFF, s & 0xFF);
+		}
+	}
+
+	MB_CONVERT_BUF_STORE(buf, out, limit);
+}