X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=src%2Ftext-coding.c;h=b3f80ff4ea181d0cc040f17320f06dd4908621da;hb=3bb53dc907705a4fa7ebc520abf3e4dc93199779;hp=3808ddec5721e69ca30207ea103db3b6613144d9;hpb=a5812bf2ff9a9cf40f4ff78dcb83f5b4c295bd18;p=chise%2Fxemacs-chise.git.1 diff --git a/src/text-coding.c b/src/text-coding.c index 3808dde..b3f80ff 100644 --- a/src/text-coding.c +++ b/src/text-coding.c @@ -1,7 +1,7 @@ /* Code conversion functions. Copyright (C) 1991, 1995 Free Software Foundation, Inc. Copyright (C) 1995 Sun Microsystems, Inc. - Copyright (C) 1999,2000,2001,2002 MORIOKA Tomohiko + Copyright (C) 1999,2000,2001,2002,2003,2004 MORIOKA Tomohiko This file is part of XEmacs. @@ -23,7 +23,7 @@ Boston, MA 02111-1307, USA. */ /* Synched up with: Mule 2.3. Not in FSF. */ /* Rewritten by Ben Wing . */ -/* Rewritten by MORIOKA Tomohiko for XEmacs UTF-2000. */ +/* Rewritten by MORIOKA Tomohiko for XEmacs CHISE. */ #include #include "lisp.h" @@ -2269,6 +2269,8 @@ do { \ #define DECODING_STREAM_DATA(stream) LSTREAM_TYPE_DATA (stream, decoding) +#define ER_BUF_SIZE 24 + struct decoding_stream { /* Coding system that governs the conversion. */ @@ -2310,7 +2312,7 @@ struct decoding_stream #endif #ifdef UTF2000 unsigned char er_counter; - unsigned char er_buf[16]; + unsigned char er_buf[ER_BUF_SIZE]; unsigned combined_char_count; Emchar combined_chars[16]; @@ -3244,6 +3246,8 @@ decode_flush_er_chars (struct decoding_stream *str, unsigned_char_dynarr* dst) } } +EXFUN (Fregexp_quote, 1); + void decode_add_er_char (struct decoding_stream *str, Emchar character, unsigned_char_dynarr* dst); void @@ -3265,7 +3269,7 @@ decode_add_er_char (struct decoding_stream *str, Emchar c, { Lisp_Object string = make_string (str->er_buf, str->er_counter); - Lisp_Object rest = Vcoded_charset_entity_reference_alist; + Lisp_Object rest; Lisp_Object cell; Lisp_Object ret; Lisp_Object pat; @@ -3273,7 +3277,8 @@ decode_add_er_char (struct decoding_stream *str, Emchar c, Lisp_Object char_type; int base; - while (!NILP (rest)) + for ( rest = Vcoded_charset_entity_reference_alist; + !NILP (rest); rest = Fcdr (rest) ) { cell = Fcar (rest); ccs = Fcar (cell); @@ -3293,6 +3298,7 @@ decode_add_er_char (struct decoding_stream *str, Emchar c, pat = ret; else continue; + pat = Fregexp_quote (pat); cell = Fcdr (cell); cell = Fcdr (cell); @@ -3328,23 +3334,22 @@ decode_add_er_char (struct decoding_stream *str, Emchar c, make_int (base))); Emchar chr = NILP (char_type) - ? DECODE_CHAR (ccs, code) + ? DECODE_CHAR (ccs, code, 0) : decode_builtin_char (ccs, code); DECODE_ADD_UCS_CHAR (chr, dst); goto decoded; } - rest = Fcdr (rest); } if (!NILP (Fstring_match (build_string ("^&MCS-\\([0-9A-F]+\\)$"), string, Qnil, Qnil))) { int code - = XINT (Fstring_to_number - (Fsubstring (string, - Fmatch_beginning (make_int (1)), - Fmatch_end (make_int (1))), - make_int (16))); + = XUINT (Fstring_to_number + (Fsubstring (string, + Fmatch_beginning (make_int (1)), + Fmatch_end (make_int (1))), + make_int (16))); DECODE_ADD_UCS_CHAR (code, dst); } @@ -3356,7 +3361,7 @@ decode_add_er_char (struct decoding_stream *str, Emchar c, decoded: str->er_counter = 0; } - else if ( (str->er_counter >= 16) || (c >= 0x7F) ) + else if ( (str->er_counter >= ER_BUF_SIZE) || (c >= 0x7F) ) { Dynarr_add_many (dst, str->er_buf, str->er_counter); str->er_counter = 0; @@ -3375,7 +3380,7 @@ char_encode_as_entity_reference (Emchar ch, char* buf) Lisp_Object ccs; Lisp_Object char_type; int format_columns, idx; - char format[18]; + char format[ER_BUF_SIZE]; while (!NILP (rest)) { @@ -3394,20 +3399,21 @@ char_encode_as_entity_reference (Emchar ch, char* buf) if ( (code_point >= 0) && (NILP (char_type) - || DECODE_CHAR (ccs, code_point) != ch) ) + || DECODE_CHAR (ccs, code_point, 0) != ch) ) { Lisp_Object ret; cell = Fcdr (cell); ret = Fcar (cell); - if (STRINGP (ret) && ((idx = XSTRING_LENGTH (ret)) <= 6)) + if ( STRINGP (ret) && + ( (idx = XSTRING_LENGTH (ret)) <= (ER_BUF_SIZE - 4) ) ) { format[0] = '&'; strncpy (&format[1], XSTRING_DATA (ret), idx); idx++; } else - continue; + goto try_next; cell = Fcdr (cell); ret = Fcar (cell); @@ -3415,12 +3421,15 @@ char_encode_as_entity_reference (Emchar ch, char* buf) { format[idx++] = '%'; format_columns = XINT (ret); - if ( (2 <= format_columns) && (format_columns <= 8) ) + if ( (2 <= format_columns) && (format_columns <= 8) + && (idx + format_columns <= ER_BUF_SIZE - 1) ) { format [idx++] = '0'; format [idx++] = '0' + format_columns; } } + else + goto try_next; cell = Fcdr (cell); ret = Fcar (cell); @@ -3431,7 +3440,7 @@ char_encode_as_entity_reference (Emchar ch, char* buf) else if (EQ (ret, QX)) format [idx++] = 'X'; else - continue; + goto try_next; format [idx++] = ';'; format [idx++] = 0; @@ -3439,6 +3448,7 @@ char_encode_as_entity_reference (Emchar ch, char* buf) return; } } + try_next: rest = Fcdr (rest); } sprintf (buf, "&MCS-%08X;", ch); @@ -3474,7 +3484,8 @@ COMPOSE_ADD_CHAR (struct decoding_stream *str, else if (!CONSP (str->combining_table)) { Lisp_Object ret - = Fget_char_attribute (make_char (character), Qcomposition, Qnil); + = Fchar_feature (make_char (character), Qcomposition, Qnil, + Qnil, Qnil); if (NILP (ret)) decode_add_er_char (str, character, dst); @@ -3493,11 +3504,12 @@ COMPOSE_ADD_CHAR (struct decoding_stream *str, if (CHARP (ret)) { Emchar char2 = XCHARVAL (ret); - ret = Fget_char_attribute (make_char (character), Qcomposition, - Qnil); - if (NILP (ret)) + Lisp_Object ret2 = Fchar_feature (ret, Qcomposition, Qnil, + Qnil, Qnil); + + if (NILP (ret2)) { - decode_add_er_char (str, character, dst); + decode_add_er_char (str, char2, dst); str->combined_char_count = 0; str->combining_table = Qnil; } @@ -3505,13 +3517,23 @@ COMPOSE_ADD_CHAR (struct decoding_stream *str, { str->combined_chars[0] = char2; str->combined_char_count = 1; - str->combining_table = ret; + str->combining_table = ret2; } } else { + ret = Fchar_feature (make_char (character), Qcomposition, Qnil, + Qnil, Qnil); + COMPOSE_FLUSH_CHARS (str, dst); - decode_add_er_char (str, character, dst); + if (NILP (ret)) + decode_add_er_char (str, character, dst); + else + { + str->combined_chars[0] = character; + str->combined_char_count = 1; + str->combining_table = ret; + } } } } @@ -3940,10 +3962,11 @@ decode_coding_big5 (Lstream *decoding, const Extbyte *src, { #ifdef UTF2000 int code_point = (cpos << 8) | c; - Emchar char_id = decode_defined_char (ccs, code_point); + Emchar char_id = decode_defined_char (ccs, code_point, 0); if (char_id < 0) - char_id = DECODE_CHAR (Vcharset_chinese_big5, code_point); + char_id + = DECODE_CHAR (Vcharset_chinese_big5, code_point, 0); DECODE_ADD_UCS_CHAR (char_id, dst); #else unsigned char b1, b2, b3; @@ -4524,13 +4547,13 @@ decode_coding_utf8 (Lstream *decoding, const Extbyte *src, if (!NILP (ccs)) { - char_id = decode_defined_char (ccs, cpos); + char_id = decode_defined_char (ccs, cpos, 0); if (char_id < 0) char_id = cpos; } else - ccs = char_id; + char_id = cpos; COMPOSE_ADD_CHAR (str, char_id, dst); cpos = 0; counter = 0; @@ -4589,21 +4612,23 @@ char_encode_utf8 (struct encoding_stream *str, Emchar ch, = CODING_SYSTEM_ISO2022_INITIAL_CHARSET (str->codesys, 0); int code_point = charset_code_point (ucs_ccs, ch, 0); - if ( (code_point < 0) || (code_point > 0x10FFFF) ) + if ( (code_point < 0) || (code_point > 0xEFFFF) ) { Lisp_Object map = CODING_SYSTEM_ISO2022_INITIAL_CHARSET (str->codesys, 1); Lisp_Object ret; if ( !NILP (map) - && INTP (ret = Fget_char_attribute (make_char (ch), - map, Qnil)) ) + && INTP (ret = Fchar_feature (make_char (ch), + map, Qnil, + Qnil, Qnil)) ) code_point = XINT (ret); else if ( !NILP (map = CODING_SYSTEM_ISO2022_INITIAL_CHARSET (str->codesys, 2)) - && INTP (ret = Fget_char_attribute (make_char (ch), - map, Qnil)) ) + && INTP (ret = Fchar_feature (make_char (ch), + map, Qnil, + Qnil, Qnil)) ) code_point = XINT (ret); else if (CODING_SYSTEM_USE_ENTITY_REFERENCE (str->codesys)) { @@ -5685,7 +5710,7 @@ decode_coding_iso2022 (Lstream *decoding, const Extbyte *src, COMPOSE_ADD_CHAR (str, DECODE_CHAR (charset, ((cpos & 0x7F7F7F) << 8) - | (c & 0x7F)), + | (c & 0x7F), 0), dst); cpos = 0; counter = 0;