X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=src%2Fmule-charset.c;h=324a6100c11fd6df1bc446118f4a6b9c7c2e971e;hb=e66d0a27c2fd4cdf439328ae014419a38fb99152;hp=ee0ce80731771a5d27f76ff42eb4ab4b2108b963;hpb=d4fd14df47400ce4a63d921ec1a4c9de649622dc;p=chise%2Fxemacs-chise.git- diff --git a/src/mule-charset.c b/src/mule-charset.c index ee0ce80..324a610 100644 --- a/src/mule-charset.c +++ b/src/mule-charset.c @@ -308,6 +308,246 @@ put_char_code_table (Emchar ch, Lisp_Object value, Lisp_Object table) } +Lisp_Object Vcharacter_attribute_table; + +DEFUN ("char-attribute-alist", Fchar_attribute_alist, 1, 1, 0, /* +Return the alist of attributes of CHARACTER. +*/ + (character)) +{ + return get_char_code_table (XCHAR (character), Vcharacter_attribute_table); +} + +DEFUN ("get-char-attribute", Fget_char_attribute, 2, 2, 0, /* +Return the value of CHARACTER's ATTRIBUTE. +*/ + (character, attribute)) +{ + Lisp_Object ret + = get_char_code_table (XCHAR (character), Vcharacter_attribute_table); + Lisp_Object ccs; + + if (EQ (ret, Qnil)) + return Qnil; + + if (!NILP (ccs = Ffind_charset (attribute))) + attribute = ccs; + + return Fcdr (Fassq (attribute, ret)); +} + +Lisp_Object +put_char_attribute (Lisp_Object character, Lisp_Object attribute, + Lisp_Object value) +{ + Emchar char_code = XCHAR (character); + Lisp_Object ret + = get_char_code_table (char_code, Vcharacter_attribute_table); + Lisp_Object cell; + + cell = Fassq (attribute, ret); + + if (NILP (cell)) + { + ret = Fcons (Fcons (attribute, value), ret); + } + else if (!EQ (Fcdr (cell), value)) + { + Fsetcdr (cell, value); + } + put_char_code_table (char_code, ret, Vcharacter_attribute_table); + return ret; +} + +DEFUN ("put-char-attribute", Fput_char_attribute, 3, 3, 0, /* +Store CHARACTER's ATTRIBUTE with VALUE. +*/ + (character, attribute, value)) +{ + Lisp_Object ccs; + + ccs = Ffind_charset (attribute); + if (!NILP (ccs)) + { + Lisp_Object rest; + Lisp_Object v = XCHARSET_DECODING_TABLE (ccs); + Lisp_Object nv; + int i = -1; + int ccs_len; + + /* ad-hoc method for `ascii' */ + if ((XCHARSET_CHARS (ccs) == 94) && + (XCHARSET_BYTE_OFFSET (ccs) != 33)) + ccs_len = 128 - XCHARSET_BYTE_OFFSET (ccs); + else + ccs_len = XCHARSET_CHARS (ccs); + + if (!CONSP (value)) + signal_simple_error ("Invalid value for coded-charset", + value); + + attribute = ccs; + rest = Fget_char_attribute (character, attribute); + if (VECTORP (v)) + { + if (!NILP (rest)) + { + while (!NILP (rest)) + { + Lisp_Object ei = Fcar (rest); + + i = XINT (ei) - XCHARSET_BYTE_OFFSET (ccs); + nv = XVECTOR_DATA(v)[i]; + if (!VECTORP (nv)) + break; + v = nv; + rest = Fcdr (rest); + } + if (i >= 0) + XVECTOR_DATA(v)[i] = Qnil; + v = XCHARSET_DECODING_TABLE (ccs); + } + } + else + { + XCHARSET_DECODING_TABLE (ccs) = v = make_vector (ccs_len, Qnil); + } + + rest = value; + i = -1; + while (CONSP (rest)) + { + Lisp_Object ei = Fcar (rest); + + if (!INTP (ei)) + signal_simple_error ("Invalid value for coded-charset", + value); + i = XINT (ei) - XCHARSET_BYTE_OFFSET (ccs); + nv = XVECTOR_DATA(v)[i]; + rest = Fcdr (rest); + if (CONSP (rest)) + { + if (!VECTORP (nv)) + { + nv = (XVECTOR_DATA(v)[i] = make_vector (ccs_len, Qnil)); + } + v = nv; + } + else + break; + } + XVECTOR_DATA(v)[i] = character; + } + return put_char_attribute (character, attribute, value); +} + +Lisp_Object Qucs; + +DEFUN ("define-char", Fdefine_char, 1, 1, 0, /* +Store character's ATTRIBUTES. +*/ + (attributes)) +{ + Lisp_Object rest = attributes; + Lisp_Object code = Fcdr (Fassq (Qucs, attributes)); + Lisp_Object character; + + if (NILP (code)) + { + while (CONSP (rest)) + { + Lisp_Object cell = Fcar (rest); + Lisp_Object ccs; + + if (!LISTP (cell)) + signal_simple_error ("Invalid argument", attributes); + if (!NILP (ccs = Ffind_charset (Fcar (cell))) + && XCHARSET_FINAL (ccs)) + { + Emchar code; + + if (XCHARSET_DIMENSION (ccs) == 1) + { + Lisp_Object eb1 = Fcar (Fcdr (cell)); + int b1; + + if (!INTP (eb1)) + signal_simple_error ("Invalid argument", attributes); + b1 = XINT (eb1); + switch (XCHARSET_CHARS (ccs)) + { + case 94: + code = MIN_CHAR_94 + + (XCHARSET_FINAL (ccs) - '0') * 94 + (b1 - 33); + break; + case 96: + code = MIN_CHAR_96 + + (XCHARSET_FINAL (ccs) - '0') * 96 + (b1 - 32); + break; + default: + abort (); + } + } + else if (XCHARSET_DIMENSION (ccs) == 2) + { + Lisp_Object eb1 = Fcar (Fcdr (cell)); + Lisp_Object eb2 = Fcar (Fcdr (Fcdr (cell))); + int b1, b2; + + if (!INTP (eb1)) + signal_simple_error ("Invalid argument", attributes); + b1 = XINT (eb1); + if (!INTP (eb2)) + signal_simple_error ("Invalid argument", attributes); + b2 = XINT (eb2); + switch (XCHARSET_CHARS (ccs)) + { + case 94: + code = MIN_CHAR_94x94 + + (XCHARSET_FINAL (ccs) - '0') * 94 * 94 + + (b1 - 33) * 94 + (b2 - 33); + break; + case 96: + code = MIN_CHAR_96x96 + + (XCHARSET_FINAL (ccs) - '0') * 96 * 96 + + (b1 - 32) * 96 + (b2 - 32); + break; + default: + abort (); + } + } + else + { + rest = Fcdr (rest); + continue; + } + character = make_char (code); + goto setup_attributes; + } + rest = Fcdr (rest); + } + return Qnil; + } + else if (!INTP (code)) + signal_simple_error ("Invalid argument", attributes); + else + character = make_char (XINT (code)); + + setup_attributes: + rest = attributes; + while (CONSP (rest)) + { + Lisp_Object cell = Fcar (rest); + + if (!LISTP (cell)) + signal_simple_error ("Invalid argument", attributes); + Fput_char_attribute (character, Fcar (cell), Fcdr (cell)); + rest = Fcdr (rest); + } + return + get_char_code_table (XCHAR (character), Vcharacter_attribute_table); +} + Lisp_Object Vutf_2000_version; #endif @@ -363,8 +603,12 @@ Lisp_Object Ql2r, Qr2l; Lisp_Object Vcharset_hash_table; +#ifdef UTF2000 +static Charset_ID next_allocated_leading_byte; +#else static Charset_ID next_allocated_1_byte_leading_byte; static Charset_ID next_allocated_2_byte_leading_byte; +#endif /* Composite characters are characters constructed by overstriking two or more regular characters. @@ -728,7 +972,6 @@ mark_charset (Lisp_Object obj, void (*markobj) (Lisp_Object)) markobj (cs->ccl_program); #ifdef UTF2000 markobj (cs->decoding_table); - markobj (cs->encoding_table); #endif return cs->name; } @@ -812,7 +1055,6 @@ make_charset (Charset_ID id, Lisp_Object name, CHARSET_REVERSE_DIRECTION_CHARSET (cs) = Qnil; #ifdef UTF2000 CHARSET_DECODING_TABLE(cs) = Qnil; - CHARSET_ENCODING_TABLE(cs) = Qnil; CHARSET_UCS_MIN(cs) = ucs_min; CHARSET_UCS_MAX(cs) = ucs_max; CHARSET_CODE_OFFSET(cs) = code_offset; @@ -903,6 +1145,12 @@ get_unallocated_leading_byte (int dimension) { Charset_ID lb; +#ifdef UTF2000 + if (next_allocated_leading_byte > MAX_LEADING_BYTE_PRIVATE) + lb = 0; + else + lb = next_allocated_leading_byte++; +#else if (dimension == 1) { if (next_allocated_1_byte_leading_byte > MAX_LEADING_BYTE_PRIVATE_1) @@ -917,6 +1165,7 @@ get_unallocated_leading_byte (int dimension) else lb = next_allocated_2_byte_leading_byte++; } +#endif if (!lb) signal_simple_error @@ -927,46 +1176,47 @@ get_unallocated_leading_byte (int dimension) } #ifdef UTF2000 -unsigned char -charset_get_byte1 (Lisp_Object charset, Emchar ch) +Lisp_Object +range_charset_code_point (Lisp_Object charset, Emchar ch) { - Lisp_Object table; int d; - if (!EQ (table = XCHARSET_ENCODING_TABLE (charset), Qnil)) - { - Lisp_Object value = get_char_code_table (ch, table); - - if (INTP (value)) - { - Emchar code = XINT (value); - - if (code < (1 << 8)) - return code; - else if (code < (1 << 16)) - return code >> 8; - else if (code < (1 << 24)) - return code >> 16; - else - return code >> 24; - } - } if ((XCHARSET_UCS_MIN (charset) <= ch) && (ch <= XCHARSET_UCS_MAX (charset))) - return (ch - XCHARSET_UCS_MIN (charset) - + XCHARSET_CODE_OFFSET (charset)) - / (XCHARSET_DIMENSION (charset) == 1 ? - 1 - : - XCHARSET_DIMENSION (charset) == 2 ? - XCHARSET_CHARS (charset) - : - XCHARSET_DIMENSION (charset) == 3 ? - XCHARSET_CHARS (charset) * XCHARSET_CHARS (charset) - : - XCHARSET_CHARS (charset) - * XCHARSET_CHARS (charset) * XCHARSET_CHARS (charset)) - + XCHARSET_BYTE_OFFSET (charset); + { + d = ch - XCHARSET_UCS_MIN (charset) + XCHARSET_CODE_OFFSET (charset); + + if (XCHARSET_DIMENSION (charset) == 1) + return list1 (make_int (d + XCHARSET_BYTE_OFFSET (charset))); + else if (XCHARSET_DIMENSION (charset) == 2) + return list2 (make_int (d / XCHARSET_CHARS (charset) + + XCHARSET_BYTE_OFFSET (charset)), + make_int (d % XCHARSET_CHARS (charset) + + XCHARSET_BYTE_OFFSET (charset))); + else if (XCHARSET_DIMENSION (charset) == 3) + return list3 (make_int (d / (XCHARSET_CHARS (charset) + * XCHARSET_CHARS (charset)) + + XCHARSET_BYTE_OFFSET (charset)), + make_int (d / XCHARSET_CHARS (charset) + % XCHARSET_CHARS (charset) + + XCHARSET_BYTE_OFFSET (charset)), + make_int (d % XCHARSET_CHARS (charset) + + XCHARSET_BYTE_OFFSET (charset))); + else /* if (XCHARSET_DIMENSION (charset) == 4) */ + return list4 (make_int (d / (XCHARSET_CHARS (charset) + * XCHARSET_CHARS (charset) + * XCHARSET_CHARS (charset)) + + XCHARSET_BYTE_OFFSET (charset)), + make_int (d / (XCHARSET_CHARS (charset) + * XCHARSET_CHARS (charset)) + % XCHARSET_CHARS (charset) + + XCHARSET_BYTE_OFFSET (charset)), + make_int (d / XCHARSET_CHARS (charset) + % XCHARSET_CHARS (charset) + + XCHARSET_BYTE_OFFSET (charset)), + make_int (d % XCHARSET_CHARS (charset) + + XCHARSET_BYTE_OFFSET (charset))); + } else if (XCHARSET_CODE_OFFSET (charset) == 0) { if (XCHARSET_DIMENSION (charset) == 1) @@ -976,17 +1226,17 @@ charset_get_byte1 (Lisp_Object charset, Emchar ch) if (((d = ch - (MIN_CHAR_94 + (XCHARSET_FINAL (charset) - '0') * 94)) >= 0) && (d < 94)) - return d + 33; + return list1 (make_int (d + 33)); } else if (XCHARSET_CHARS (charset) == 96) { if (((d = ch - (MIN_CHAR_96 + (XCHARSET_FINAL (charset) - '0') * 96)) >= 0) && (d < 96)) - return d + 32; + return list1 (make_int (d + 32)); } else - return 0; + return Qnil; } else if (XCHARSET_DIMENSION (charset) == 2) { @@ -996,7 +1246,7 @@ charset_get_byte1 (Lisp_Object charset, Emchar ch) + (XCHARSET_FINAL (charset) - '0') * 94 * 94)) >= 0) && (d < 94 * 94)) - return (d / 94) + 33; + return list2 ((d / 94) + 33, d % 94 + 33); } else if (XCHARSET_CHARS (charset) == 96) { @@ -1004,64 +1254,26 @@ charset_get_byte1 (Lisp_Object charset, Emchar ch) + (XCHARSET_FINAL (charset) - '0') * 96 * 96)) >= 0) && (d < 96 * 96)) - return (d / 96) + 32; + return list2 ((d / 96) + 32, d % 96 + 32); } } } - return 0; + return Qnil; } -unsigned char -charset_get_byte2 (Lisp_Object charset, Emchar ch) +Lisp_Object +charset_code_point (Lisp_Object charset, Emchar ch) { - if (XCHARSET_DIMENSION (charset) == 1) - return 0; - else - { - Lisp_Object table; + Lisp_Object cdef = get_char_code_table (ch, Vcharacter_attribute_table); - if (!EQ (table = XCHARSET_ENCODING_TABLE (charset), Qnil)) - { - Lisp_Object value = get_char_code_table (ch, table); - - if (INTP (value)) - { - Emchar code = XINT (value); + if (!EQ (cdef, Qnil)) + { + Lisp_Object field = Fassq (charset, cdef); - if (code < (1 << 16)) - return (unsigned char)code; - else if (code < (1 << 24)) - return (unsigned char)(code >> 16); - else - return (unsigned char)(code >> 24); - } - } - if ((XCHARSET_UCS_MIN (charset) <= ch) - && (ch <= XCHARSET_UCS_MAX (charset))) - return ((ch - XCHARSET_UCS_MIN (charset) - + XCHARSET_CODE_OFFSET (charset)) - / (XCHARSET_DIMENSION (charset) == 2 ? - 1 - : - XCHARSET_DIMENSION (charset) == 3 ? - XCHARSET_CHARS (charset) - : - XCHARSET_CHARS (charset) * XCHARSET_CHARS (charset))) - % XCHARSET_CHARS (charset) - + XCHARSET_BYTE_OFFSET (charset); - else if (XCHARSET_CHARS (charset) == 94) - return (MIN_CHAR_94x94 - + (XCHARSET_FINAL (charset) - '0') * 94 * 94 <= ch) - && (ch < MIN_CHAR_94x94 - + (XCHARSET_FINAL (charset) - '0' + 1) * 94 * 94) ? - ((ch - MIN_CHAR_94x94) % 94) + 33 : 0; - else /* if (XCHARSET_CHARS (charset) == 96) */ - return (MIN_CHAR_96x96 - + (XCHARSET_FINAL (charset) - '0') * 96 * 96 <= ch) - && (ch < MIN_CHAR_96x96 - + (XCHARSET_FINAL (charset) - '0' + 1) * 96 * 96) ? - ((ch - MIN_CHAR_96x96) % 96) + 32 : 0; + if (!EQ (field, Qnil)) + return Fcdr (field); } + return range_charset_code_point (charset, ch); } Lisp_Object Vdefault_coded_charset_priority_list; @@ -1213,9 +1425,7 @@ character set. Recognized properties are: Lisp_Object rest, keyword, value; Lisp_Object ccl_program = Qnil; Lisp_Object short_name = Qnil, long_name = Qnil; -#ifdef UTF2000 unsigned char byte_offset = 0; -#endif CHECK_SYMBOL (name); if (!NILP (doc_string)) @@ -1326,52 +1536,7 @@ character set. Recognized properties are: error ("Character set already defined for this DIMENSION/CHARS/FINAL combo"); -#ifdef UTF2000 - if (dimension == 1) - { - if (chars == 94) - { - /* id = CHARSET_ID_OFFSET_94 + final; */ - id = get_unallocated_leading_byte (dimension); - } - else if (chars == 96) - { - id = get_unallocated_leading_byte (dimension); - } - else - { - abort (); - } - } - else if (dimension == 2) - { - if (chars == 94) - { - id = get_unallocated_leading_byte (dimension); - } - else if (chars == 96) - { - id = get_unallocated_leading_byte (dimension); - } - else - { - abort (); - } - } - else - { - abort (); - } - if (final) - { - if (chars == 94) - byte_offset = 33; - else if (chars == 96) - byte_offset = 32; - } -#else id = get_unallocated_leading_byte (dimension); -#endif if (NILP (doc_string)) doc_string = build_string (""); @@ -1680,14 +1845,13 @@ Set mapping-table of CHARSET to TABLE. if (EQ (table, Qnil)) { CHARSET_DECODING_TABLE(cs) = table; - CHARSET_ENCODING_TABLE(cs) = Qnil; return table; } else if (VECTORP (table)) { if (XVECTOR_LENGTH (table) > CHARSET_CHARS (cs)) args_out_of_range (table, make_int (CHARSET_CHARS (cs))); - old_table = CHARSET_ENCODING_TABLE(cs); + old_table = CHARSET_DECODING_TABLE(cs); CHARSET_DECODING_TABLE(cs) = table; } else @@ -1699,19 +1863,17 @@ Set mapping-table of CHARSET to TABLE. switch (CHARSET_DIMENSION (cs)) { case 1: - CHARSET_ENCODING_TABLE(cs) = make_char_code_table (Qnil); for (i = 0; i < XVECTOR_LENGTH (table); i++) { Lisp_Object c = XVECTOR_DATA(table)[i]; if (CHARP (c)) - put_char_code_table (XCHAR (c), - make_int (i + CHARSET_BYTE_OFFSET (cs)), - CHARSET_ENCODING_TABLE(cs)); + put_char_attribute + (c, charset, + list1 (make_int (i + CHARSET_BYTE_OFFSET (cs)))); } break; case 2: - CHARSET_ENCODING_TABLE(cs) = make_char_code_table (Qnil); for (i = 0; i < XVECTOR_LENGTH (table); i++) { Lisp_Object v = XVECTOR_DATA(table)[i]; @@ -1730,17 +1892,18 @@ Set mapping-table of CHARSET to TABLE. Lisp_Object c = XVECTOR_DATA(v)[j]; if (CHARP (c)) - put_char_code_table - (XCHAR (c), - make_int (( (i + CHARSET_BYTE_OFFSET (cs)) << 8) - | (j + CHARSET_BYTE_OFFSET (cs))), - CHARSET_ENCODING_TABLE(cs)); + put_char_attribute (c, charset, + list2 + (make_int + (i + CHARSET_BYTE_OFFSET (cs)), + make_int + (j + CHARSET_BYTE_OFFSET (cs)))); } } else if (CHARP (v)) - put_char_code_table (XCHAR (v), - make_int (i + CHARSET_BYTE_OFFSET (cs)), - CHARSET_ENCODING_TABLE(cs)); + put_char_attribute (v, charset, + list1 + (make_int (i + CHARSET_BYTE_OFFSET (cs)))); } break; } @@ -1953,6 +2116,10 @@ syms_of_mule_charset (void) DEFSUBR (Fset_charset_ccl_program); DEFSUBR (Fset_charset_registry); #ifdef UTF2000 + DEFSUBR (Fchar_attribute_alist); + DEFSUBR (Fget_char_attribute); + DEFSUBR (Fput_char_attribute); + DEFSUBR (Fdefine_char); DEFSUBR (Fcharset_mapping_table); DEFSUBR (Fset_charset_mapping_table); #endif @@ -2002,6 +2169,7 @@ syms_of_mule_charset (void) defsymbol (&Qchinese_cns11643_1, "chinese-cns11643-1"); defsymbol (&Qchinese_cns11643_2, "chinese-cns11643-2"); #ifdef UTF2000 + defsymbol (&Qucs, "ucs"); defsymbol (&Qucs_bmp, "ucs-bmp"); defsymbol (&Qlatin_viscii, "latin-viscii"); defsymbol (&Qlatin_viscii_lower, "latin-viscii-lower"); @@ -2042,10 +2210,10 @@ vars_of_mule_charset (void) charset_by_attributes[i][j][k] = Qnil; #endif - next_allocated_1_byte_leading_byte = MIN_LEADING_BYTE_PRIVATE_1; #ifdef UTF2000 - next_allocated_2_byte_leading_byte = LEADING_BYTE_CHINESE_BIG5_2 + 1; + next_allocated_leading_byte = MIN_LEADING_BYTE_PRIVATE; #else + next_allocated_1_byte_leading_byte = MIN_LEADING_BYTE_PRIVATE_1; next_allocated_2_byte_leading_byte = MIN_LEADING_BYTE_PRIVATE_2; #endif @@ -2058,15 +2226,18 @@ Leading-code of private TYPE9N charset of column-width 1. #endif #ifdef UTF2000 - Vutf_2000_version = build_string("0.8 (Kami)"); + Vutf_2000_version = build_string("0.11 (Shiki)"); DEFVAR_LISP ("utf-2000-version", &Vutf_2000_version /* Version number of UTF-2000. */ ); + staticpro (&Vcharacter_attribute_table); + Vcharacter_attribute_table = make_char_code_table (Qnil); + Vdefault_coded_charset_priority_list = Qnil; DEFVAR_LISP ("default-coded-charset-priority-list", &Vdefault_coded_charset_priority_list /* -Default order of preferred coded-character-set. +Default order of preferred coded-character-sets. */ ); #endif }