X-Git-Url: http://git.chise.org/gitweb/?a=blobdiff_plain;f=src%2Fmule-charset.c;h=2db3cbcebfc59250dcea85a408b636e6f39276be;hb=975655e6b5b1526ee82b159b3eadf69888c42090;hp=cf3b4aca30d646412d3d989017037c39a24c959e;hpb=5fb1ce6331fce54c98b69b0092456a1a4b927227;p=chise%2Fxemacs-chise.git- diff --git a/src/mule-charset.c b/src/mule-charset.c index cf3b4ac..2db3cbc 100644 --- a/src/mule-charset.c +++ b/src/mule-charset.c @@ -63,9 +63,11 @@ Lisp_Object Vcharset_chinese_cns11643_2; Lisp_Object Vcharset_ucs; Lisp_Object Vcharset_ucs_bmp; Lisp_Object Vcharset_latin_viscii; +Lisp_Object Vcharset_latin_tcvn5712; Lisp_Object Vcharset_latin_viscii_lower; Lisp_Object Vcharset_latin_viscii_upper; Lisp_Object Vcharset_ideograph_daikanwa; +Lisp_Object Vcharset_mojikyo; Lisp_Object Vcharset_mojikyo_pj_1; Lisp_Object Vcharset_mojikyo_pj_2; Lisp_Object Vcharset_mojikyo_pj_3; @@ -109,7 +111,7 @@ static int composite_char_col_next; struct charset_lookup *chlook; static const struct lrecord_description charset_lookup_description_1[] = { - { XD_LISP_OBJECT, offsetof(struct charset_lookup, charset_by_leading_byte), + { XD_LISP_OBJECT_ARRAY, offsetof (struct charset_lookup, charset_by_leading_byte), #ifdef UTF2000 128+4*128 #else @@ -119,7 +121,7 @@ static const struct lrecord_description charset_lookup_description_1[] = { }; static const struct struct_description charset_lookup_description = { - sizeof(struct charset_lookup), + sizeof (struct charset_lookup), charset_lookup_description_1 }; @@ -158,9 +160,9 @@ const Bytecount rep_bytes_by_first_byte[0xA0] = #ifdef UTF2000 static Lisp_Object -mark_char_byte_table (Lisp_Object obj) +mark_byte_table (Lisp_Object obj) { - struct Lisp_Char_Byte_Table *cte = XCHAR_BYTE_TABLE (obj); + Lisp_Byte_Table *cte = XBYTE_TABLE (obj); int i; for (i = 0; i < 256; i++) @@ -171,19 +173,19 @@ mark_char_byte_table (Lisp_Object obj) } static int -char_byte_table_equal (Lisp_Object obj1, Lisp_Object obj2, int depth) +byte_table_equal (Lisp_Object obj1, Lisp_Object obj2, int depth) { - struct Lisp_Char_Byte_Table *cte1 = XCHAR_BYTE_TABLE (obj1); - struct Lisp_Char_Byte_Table *cte2 = XCHAR_BYTE_TABLE (obj2); + Lisp_Byte_Table *cte1 = XBYTE_TABLE (obj1); + Lisp_Byte_Table *cte2 = XBYTE_TABLE (obj2); int i; for (i = 0; i < 256; i++) - if (CHAR_BYTE_TABLE_P (cte1->property[i])) + if (BYTE_TABLE_P (cte1->property[i])) { - if (CHAR_BYTE_TABLE_P (cte2->property[i])) + if (BYTE_TABLE_P (cte2->property[i])) { - if (!char_byte_table_equal (cte1->property[i], - cte2->property[i], depth + 1)) + if (!byte_table_equal (cte1->property[i], + cte2->property[i], depth + 1)) return 0; } else @@ -196,224 +198,236 @@ char_byte_table_equal (Lisp_Object obj1, Lisp_Object obj2, int depth) } static unsigned long -char_byte_table_hash (Lisp_Object obj, int depth) +byte_table_hash (Lisp_Object obj, int depth) { - struct Lisp_Char_Byte_Table *cte = XCHAR_BYTE_TABLE (obj); + Lisp_Byte_Table *cte = XBYTE_TABLE (obj); return internal_array_hash (cte->property, 256, depth); } -static const struct lrecord_description char_byte_table_description[] = { - { XD_LISP_OBJECT, offsetof(struct Lisp_Char_Byte_Table, property), 256 }, +static const struct lrecord_description byte_table_description[] = { + { XD_LISP_OBJECT_ARRAY, offsetof(Lisp_Byte_Table, property), 256 }, { XD_END } }; -DEFINE_LRECORD_IMPLEMENTATION ("char-byte-table", char_byte_table, - mark_char_byte_table, +DEFINE_LRECORD_IMPLEMENTATION ("byte-table", byte_table, + mark_byte_table, internal_object_printer, - 0, char_byte_table_equal, - char_byte_table_hash, - char_byte_table_description, - struct Lisp_Char_Byte_Table); + 0, byte_table_equal, + byte_table_hash, + byte_table_description, + Lisp_Byte_Table); static Lisp_Object -make_char_byte_table (Lisp_Object initval) +make_byte_table (Lisp_Object initval, int older) { Lisp_Object obj; int i; - struct Lisp_Char_Byte_Table *cte = - alloc_lcrecord_type (struct Lisp_Char_Byte_Table, - &lrecord_char_byte_table); + Lisp_Byte_Table *cte; + + if (older) + cte = alloc_older_lcrecord_type (Lisp_Byte_Table, &lrecord_byte_table); + else + cte = alloc_lcrecord_type (Lisp_Byte_Table, &lrecord_byte_table); for (i = 0; i < 256; i++) cte->property[i] = initval; - XSETCHAR_BYTE_TABLE (obj, cte); + XSETBYTE_TABLE (obj, cte); return obj; } static Lisp_Object -copy_char_byte_table (Lisp_Object entry) +copy_byte_table (Lisp_Object entry) { - struct Lisp_Char_Byte_Table *cte = XCHAR_BYTE_TABLE (entry); + Lisp_Byte_Table *cte = XBYTE_TABLE (entry); Lisp_Object obj; int i; - struct Lisp_Char_Byte_Table *ctenew = - alloc_lcrecord_type (struct Lisp_Char_Byte_Table, - &lrecord_char_byte_table); + Lisp_Byte_Table *ctenew + = alloc_lcrecord_type (Lisp_Byte_Table, &lrecord_byte_table); for (i = 0; i < 256; i++) { Lisp_Object new = cte->property[i]; - if (CHAR_BYTE_TABLE_P (new)) - ctenew->property[i] = copy_char_byte_table (new); + if (BYTE_TABLE_P (new)) + ctenew->property[i] = copy_byte_table (new); else ctenew->property[i] = new; } - XSETCHAR_BYTE_TABLE (obj, ctenew); + XSETBYTE_TABLE (obj, ctenew); return obj; } static Lisp_Object -mark_char_code_table (Lisp_Object obj) +mark_char_id_table (Lisp_Object obj) { - struct Lisp_Char_Code_Table *cte = XCHAR_CODE_TABLE (obj); + Lisp_Char_ID_Table *cte = XCHAR_ID_TABLE (obj); return cte->table; } static int -char_code_table_equal (Lisp_Object obj1, Lisp_Object obj2, int depth) +char_id_table_equal (Lisp_Object obj1, Lisp_Object obj2, int depth) { - struct Lisp_Char_Code_Table *cte1 = XCHAR_CODE_TABLE (obj1); - struct Lisp_Char_Code_Table *cte2 = XCHAR_CODE_TABLE (obj2); + Lisp_Char_ID_Table *cte1 = XCHAR_ID_TABLE (obj1); + Lisp_Char_ID_Table *cte2 = XCHAR_ID_TABLE (obj2); - return char_byte_table_equal (cte1->table, cte2->table, depth + 1); + return byte_table_equal (cte1->table, cte2->table, depth + 1); } static unsigned long -char_code_table_hash (Lisp_Object obj, int depth) +char_id_table_hash (Lisp_Object obj, int depth) { - struct Lisp_Char_Code_Table *cte = XCHAR_CODE_TABLE (obj); + Lisp_Char_ID_Table *cte = XCHAR_ID_TABLE (obj); - return char_code_table_hash (cte->table, depth + 1); + return char_id_table_hash (cte->table, depth + 1); } -static const struct lrecord_description char_code_table_description[] = { - { XD_LISP_OBJECT, offsetof(struct Lisp_Char_Code_Table, table), 1 }, +static const struct lrecord_description char_id_table_description[] = { + { XD_LISP_OBJECT, offsetof(Lisp_Char_ID_Table, table) }, { XD_END } }; -DEFINE_LRECORD_IMPLEMENTATION ("char-code-table", char_code_table, - mark_char_code_table, +DEFINE_LRECORD_IMPLEMENTATION ("char-id-table", char_id_table, + mark_char_id_table, internal_object_printer, - 0, char_code_table_equal, - char_code_table_hash, - char_code_table_description, - struct Lisp_Char_Code_Table); + 0, char_id_table_equal, + char_id_table_hash, + char_id_table_description, + Lisp_Char_ID_Table); static Lisp_Object -make_char_code_table (Lisp_Object initval) +make_char_id_table (Lisp_Object initval, int older) { Lisp_Object obj; - struct Lisp_Char_Code_Table *cte = - alloc_lcrecord_type (struct Lisp_Char_Code_Table, - &lrecord_char_code_table); + Lisp_Char_ID_Table *cte; - cte->table = make_char_byte_table (initval); + if (older) + cte = alloc_older_lcrecord_type (Lisp_Char_ID_Table, + &lrecord_char_id_table); + else + cte = alloc_lcrecord_type (Lisp_Char_ID_Table, &lrecord_char_id_table); - XSETCHAR_CODE_TABLE (obj, cte); + cte->table = make_byte_table (initval, older); + + XSETCHAR_ID_TABLE (obj, cte); return obj; } +/* not used */ +#if 0 static Lisp_Object -copy_char_code_table (Lisp_Object entry) +copy_char_id_table (Lisp_Object entry) { - struct Lisp_Char_Code_Table *cte = XCHAR_CODE_TABLE (entry); + Lisp_Char_ID_Table *cte = XCHAR_ID_TABLE (entry); Lisp_Object obj; - struct Lisp_Char_Code_Table *ctenew = - alloc_lcrecord_type (struct Lisp_Char_Code_Table, - &lrecord_char_code_table); + Lisp_Char_ID_Table *ctenew + = alloc_lcrecord_type (Lisp_Char_ID_Table, &lrecord_char_id_table); - ctenew->table = copy_char_byte_table (cte->table); - XSETCHAR_CODE_TABLE (obj, ctenew); + ctenew->table = copy_byte_table (cte->table); + XSETCHAR_ID_TABLE (obj, ctenew); return obj; } +#endif Lisp_Object -get_char_code_table (Emchar ch, Lisp_Object table) +get_char_id_table (Emchar ch, Lisp_Object table) { unsigned int code = ch; - struct Lisp_Char_Byte_Table* cpt - = XCHAR_BYTE_TABLE (XCHAR_CODE_TABLE (table)->table); + Lisp_Byte_Table* cpt + = XBYTE_TABLE (XCHAR_ID_TABLE (table)->table); Lisp_Object ret = cpt->property [(unsigned char)(code >> 24)]; - if (CHAR_BYTE_TABLE_P (ret)) - cpt = XCHAR_BYTE_TABLE (ret); + if (BYTE_TABLE_P (ret)) + cpt = XBYTE_TABLE (ret); else return ret; ret = cpt->property [(unsigned char) (code >> 16)]; - if (CHAR_BYTE_TABLE_P (ret)) - cpt = XCHAR_BYTE_TABLE (ret); + if (BYTE_TABLE_P (ret)) + cpt = XBYTE_TABLE (ret); else return ret; ret = cpt->property [(unsigned char) (code >> 8)]; - if (CHAR_BYTE_TABLE_P (ret)) - cpt = XCHAR_BYTE_TABLE (ret); + if (BYTE_TABLE_P (ret)) + cpt = XBYTE_TABLE (ret); else return ret; return cpt->property [(unsigned char) code]; } +void put_char_id_table (Emchar ch, Lisp_Object value, Lisp_Object table); void -put_char_code_table (Emchar ch, Lisp_Object value, Lisp_Object table) +put_char_id_table (Emchar ch, Lisp_Object value, Lisp_Object table) { unsigned int code = ch; - struct Lisp_Char_Byte_Table* cpt1 - = XCHAR_BYTE_TABLE (XCHAR_CODE_TABLE (table)->table); + Lisp_Byte_Table* cpt1 = XBYTE_TABLE (XCHAR_ID_TABLE (table)->table); Lisp_Object ret = cpt1->property[(unsigned char)(code >> 24)]; - if (CHAR_BYTE_TABLE_P (ret)) + if (BYTE_TABLE_P (ret)) { - struct Lisp_Char_Byte_Table* cpt2 = XCHAR_BYTE_TABLE (ret); + Lisp_Byte_Table* cpt2 = XBYTE_TABLE (ret); ret = cpt2->property[(unsigned char)(code >> 16)]; - if (CHAR_BYTE_TABLE_P (ret)) + if (BYTE_TABLE_P (ret)) { - struct Lisp_Char_Byte_Table* cpt3 = XCHAR_BYTE_TABLE (ret); + Lisp_Byte_Table* cpt3 = XBYTE_TABLE (ret); ret = cpt3->property[(unsigned char)(code >> 8)]; - if (CHAR_BYTE_TABLE_P (ret)) + if (BYTE_TABLE_P (ret)) { - struct Lisp_Char_Byte_Table* cpt4 - = XCHAR_BYTE_TABLE (ret); + Lisp_Byte_Table* cpt4 = XBYTE_TABLE (ret); cpt4->property[(unsigned char)code] = value; } else if (!EQ (ret, value)) { - Lisp_Object cpt4 = make_char_byte_table (ret); - - XCHAR_BYTE_TABLE(cpt4)->property[(unsigned char)code] = value; + Lisp_Object cpt4 + = make_byte_table (ret, OLDER_RECORD_P (table)); + + XBYTE_TABLE(cpt4)->property[(unsigned char)code] = value; cpt3->property[(unsigned char)(code >> 8)] = cpt4; } } else if (!EQ (ret, value)) { - Lisp_Object cpt3 = make_char_byte_table (ret); - Lisp_Object cpt4 = make_char_byte_table (ret); - - XCHAR_BYTE_TABLE(cpt4)->property[(unsigned char)code] = value; - XCHAR_BYTE_TABLE(cpt3)->property[(unsigned char)(code >> 8)] + int older = OLDER_RECORD_P (table); + Lisp_Object cpt3 = make_byte_table (ret, older); + Lisp_Object cpt4 = make_byte_table (ret, older); + + XBYTE_TABLE(cpt4)->property[(unsigned char)code] = value; + XBYTE_TABLE(cpt3)->property[(unsigned char)(code >> 8)] = cpt4; cpt2->property[(unsigned char)(code >> 16)] = cpt3; } } else if (!EQ (ret, value)) { - Lisp_Object cpt2 = make_char_byte_table (ret); - Lisp_Object cpt3 = make_char_byte_table (ret); - Lisp_Object cpt4 = make_char_byte_table (ret); - - XCHAR_BYTE_TABLE(cpt4)->property[(unsigned char)code] = value; - XCHAR_BYTE_TABLE(cpt3)->property[(unsigned char)(code >> 8)] = cpt4; - XCHAR_BYTE_TABLE(cpt2)->property[(unsigned char)(code >> 16)] = cpt3; + int older = OLDER_RECORD_P (table); + Lisp_Object cpt2 = make_byte_table (ret, older); + Lisp_Object cpt3 = make_byte_table (ret, older); + Lisp_Object cpt4 = make_byte_table (ret, older); + + XBYTE_TABLE(cpt4)->property[(unsigned char)code] = value; + XBYTE_TABLE(cpt3)->property[(unsigned char)(code >> 8)] = cpt4; + XBYTE_TABLE(cpt2)->property[(unsigned char)(code >> 16)] = cpt3; cpt1->property[(unsigned char)(code >> 24)] = cpt2; } } -Lisp_Object Vcharacter_attribute_table; +Lisp_Object Vchar_attribute_hash_table; Lisp_Object Vcharacter_composition_table; Lisp_Object Vcharacter_variant_table; +Lisp_Object Qideograph_daikanwa; Lisp_Object Q_decomposition; +Lisp_Object Qucs; Lisp_Object Q_ucs; Lisp_Object Qcompat; Lisp_Object Qisolated; @@ -432,8 +446,14 @@ Lisp_Object Qnarrow; Lisp_Object Qsmall; Lisp_Object Qfont; +Emchar to_char_id (Lisp_Object v, char* err_msg, Lisp_Object err_arg); + +Lisp_Object put_char_ccs_code_point (Lisp_Object character, + Lisp_Object ccs, Lisp_Object value); +Lisp_Object remove_char_ccs (Lisp_Object character, Lisp_Object ccs); + Emchar -to_char_code (Lisp_Object v, char* err_msg, Lisp_Object err_arg) +to_char_id (Lisp_Object v, char* err_msg, Lisp_Object err_arg) { if (INTP (v)) return XINT (v); @@ -487,21 +507,21 @@ Return character corresponding with list. { Lisp_Object v = Fcar (rest); Lisp_Object ret; - Emchar c = to_char_code (v, "Invalid value for composition", list); + Emchar c = to_char_id (v, "Invalid value for composition", list); - ret = get_char_code_table (c, table); + ret = get_char_id_table (c, table); rest = Fcdr (rest); if (NILP (rest)) { - if (!CHAR_CODE_TABLE_P (ret)) + if (!CHAR_ID_TABLE_P (ret)) return ret; else return Qt; } else if (!CONSP (rest)) break; - else if (CHAR_CODE_TABLE_P (ret)) + else if (CHAR_ID_TABLE_P (ret)) table = ret; else signal_simple_error ("Invalid table is found with", list); @@ -515,8 +535,79 @@ Return variants of CHARACTER. (character)) { CHECK_CHAR (character); - return Fcopy_list (get_char_code_table (XCHAR (character), - Vcharacter_variant_table)); + return Fcopy_list (get_char_id_table (XCHAR (character), + Vcharacter_variant_table)); +} + + +/* We store the char-attributes in hash tables with the names as the + key and the actual char-id-table object as the value. Occasionally + we need to use them in a list format. These routines provide us + with that. */ +struct char_attribute_list_closure +{ + Lisp_Object *char_attribute_list; +}; + +static int +add_char_attribute_to_list_mapper (Lisp_Object key, Lisp_Object value, + void *char_attribute_list_closure) +{ + /* This function can GC */ + struct char_attribute_list_closure *calcl + = (struct char_attribute_list_closure*) char_attribute_list_closure; + Lisp_Object *char_attribute_list = calcl->char_attribute_list; + + *char_attribute_list = Fcons (key, *char_attribute_list); + return 0; +} + +DEFUN ("char-attribute-list", Fchar_attribute_list, 0, 0, 0, /* +Return the list of all existing character attributes except coded-charsets. +*/ + ()) +{ + Lisp_Object char_attribute_list = Qnil; + struct gcpro gcpro1; + struct char_attribute_list_closure char_attribute_list_closure; + + GCPRO1 (char_attribute_list); + char_attribute_list_closure.char_attribute_list = &char_attribute_list; + elisp_maphash (add_char_attribute_to_list_mapper, + Vchar_attribute_hash_table, + &char_attribute_list_closure); + UNGCPRO; + return char_attribute_list; +} + + +/* We store the char-id-tables in hash tables with the attributes as + the key and the actual char-id-table object as the value. Each + char-id-table stores values of an attribute corresponding with + characters. Occasionally we need to get attributes of a character + in a association-list format. These routines provide us with + that. */ +struct char_attribute_alist_closure +{ + Emchar char_id; + Lisp_Object *char_attribute_alist; +}; + +static int +add_char_attribute_alist_mapper (Lisp_Object key, Lisp_Object value, + void *char_attribute_alist_closure) +{ + /* This function can GC */ + struct char_attribute_alist_closure *caacl = + (struct char_attribute_alist_closure*) char_attribute_alist_closure; + Lisp_Object ret = get_char_id_table (caacl->char_id, value); + if (!UNBOUNDP (ret)) + { + Lisp_Object *char_attribute_alist = caacl->char_attribute_alist; + *char_attribute_alist + = Fcons (Fcons (key, ret), *char_attribute_alist); + } + return 0; } DEFUN ("char-attribute-alist", Fchar_attribute_alist, 1, 1, 0, /* @@ -524,9 +615,41 @@ Return the alist of attributes of CHARACTER. */ (character)) { + Lisp_Object alist = Qnil; + int i; + CHECK_CHAR (character); - return Fcopy_alist (get_char_code_table (XCHAR (character), - Vcharacter_attribute_table)); + { + struct gcpro gcpro1; + struct char_attribute_alist_closure char_attribute_alist_closure; + + GCPRO1 (alist); + char_attribute_alist_closure.char_id = XCHAR (character); + char_attribute_alist_closure.char_attribute_alist = &alist; + elisp_maphash (add_char_attribute_alist_mapper, + Vchar_attribute_hash_table, + &char_attribute_alist_closure); + UNGCPRO; + } + + for (i = 0; i < countof (chlook->charset_by_leading_byte); i++) + { + Lisp_Object ccs = chlook->charset_by_leading_byte[i]; + + if (!NILP (ccs)) + { + Lisp_Object encoding_table = XCHARSET_ENCODING_TABLE (ccs); + Lisp_Object cpos; + + if ( CHAR_ID_TABLE_P (encoding_table) + && INTP (cpos = get_char_id_table (XCHAR (character), + encoding_table)) ) + { + alist = Fcons (Fcons (ccs, cpos), alist); + } + } + } + return alist; } DEFUN ("get-char-attribute", Fget_char_attribute, 2, 2, 0, /* @@ -534,48 +657,150 @@ Return the value of CHARACTER's ATTRIBUTE. */ (character, attribute)) { - Lisp_Object ret; Lisp_Object ccs; CHECK_CHAR (character); - ret = get_char_code_table (XCHAR (character), - Vcharacter_attribute_table); - if (EQ (ret, Qnil)) - return Qnil; - if (!NILP (ccs = Ffind_charset (attribute))) - attribute = ccs; + { + Lisp_Object encoding_table = XCHARSET_ENCODING_TABLE (ccs); - return Fcdr (Fassq (attribute, ret)); + if (CHAR_ID_TABLE_P (encoding_table)) + return get_char_id_table (XCHAR (character), encoding_table); + else + return Qnil; + } + else + { + Lisp_Object table = Fgethash (attribute, + Vchar_attribute_hash_table, + Qunbound); + if (!UNBOUNDP (table)) + { + Lisp_Object ret = get_char_id_table (XCHAR (character), table); + if (!UNBOUNDP (ret)) + return ret; + } + } + return Qnil; } -Lisp_Object -put_char_attribute (Lisp_Object character, Lisp_Object attribute, - Lisp_Object value) +DEFUN ("put-char-attribute", Fput_char_attribute, 3, 3, 0, /* +Store CHARACTER's ATTRIBUTE with VALUE. +*/ + (character, attribute, 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); + Lisp_Object ccs; - if (NILP (cell)) + CHECK_CHAR (character); + ccs = Ffind_charset (attribute); + if (!NILP (ccs)) { - ret = Fcons (Fcons (attribute, value), ret); + return put_char_ccs_code_point (character, ccs, value); } - else if (!EQ (Fcdr (cell), value)) + else if (EQ (attribute, Q_decomposition)) { - Fsetcdr (cell, value); + Lisp_Object seq; + + if (!CONSP (value)) + signal_simple_error ("Invalid value for ->decomposition", + value); + + if (CONSP (Fcdr (value))) + { + Lisp_Object rest = value; + Lisp_Object table = Vcharacter_composition_table; + size_t len; + int i = 0; + + GET_EXTERNAL_LIST_LENGTH (rest, len); + seq = make_vector (len, Qnil); + + while (CONSP (rest)) + { + Lisp_Object v = Fcar (rest); + Lisp_Object ntable; + Emchar c + = to_char_id (v, "Invalid value for ->decomposition", value); + + if (c < 0) + XVECTOR_DATA(seq)[i++] = v; + else + XVECTOR_DATA(seq)[i++] = make_char (c); + rest = Fcdr (rest); + if (!CONSP (rest)) + { + put_char_id_table (c, character, table); + break; + } + else + { + ntable = get_char_id_table (c, table); + if (!CHAR_ID_TABLE_P (ntable)) + { + ntable + = make_char_id_table (Qnil, OLDER_RECORD_P (table)); + put_char_id_table (c, ntable, table); + } + table = ntable; + } + } + } + else + { + Lisp_Object v = Fcar (value); + + if (INTP (v)) + { + Emchar c = XINT (v); + Lisp_Object ret + = get_char_id_table (c, Vcharacter_variant_table); + + if (NILP (Fmemq (v, ret))) + { + put_char_id_table (c, Fcons (character, ret), + Vcharacter_variant_table); + } + } + seq = make_vector (1, v); + } + value = seq; + } + else if (EQ (attribute, Q_ucs)) + { + Lisp_Object ret; + Emchar c; + + if (!INTP (value)) + signal_simple_error ("Invalid value for ->ucs", value); + + c = XINT (value); + + ret = get_char_id_table (c, Vcharacter_variant_table); + if (NILP (Fmemq (character, ret))) + { + put_char_id_table (c, Fcons (character, ret), + Vcharacter_variant_table); + } } - put_char_code_table (char_code, ret, Vcharacter_attribute_table); - return ret; + { + Lisp_Object table = Fgethash (attribute, + Vchar_attribute_hash_table, + Qnil); + + if (NILP (table)) + { + table = make_char_id_table (Qunbound, 0); + Fputhash (attribute, table, Vchar_attribute_hash_table); + } + put_char_id_table (XCHAR (character), value, table); + return value; + } } -DEFUN ("put-char-attribute", Fput_char_attribute, 3, 3, 0, /* -Store CHARACTER's ATTRIBUTE with VALUE. +DEFUN ("remove-char-attribute", Fremove_char_attribute, 2, 2, 0, /* +Remove CHARACTER's ATTRIBUTE. */ - (character, attribute, value)) + (character, attribute)) { Lisp_Object ccs; @@ -583,24 +808,135 @@ Store CHARACTER's ATTRIBUTE with VALUE. ccs = Ffind_charset (attribute); if (!NILP (ccs)) { - Lisp_Object cpos, rest; - Lisp_Object v = XCHARSET_DECODING_TABLE (ccs); + return remove_char_ccs (character, ccs); + } + else + { + Lisp_Object table = Fgethash (attribute, + Vchar_attribute_hash_table, + Qunbound); + if (!UNBOUNDP (table)) + { + put_char_id_table (XCHAR (character), Qunbound, table); + return Qt; + } + } + return Qnil; +} + +INLINE_HEADER int CHARSET_BYTE_SIZE (Lisp_Charset* cs); +INLINE_HEADER int +CHARSET_BYTE_SIZE (Lisp_Charset* cs) +{ + /* ad-hoc method for `ascii' */ + if ((CHARSET_CHARS (cs) == 94) && + (CHARSET_BYTE_OFFSET (cs) != 33)) + return 128 - CHARSET_BYTE_OFFSET (cs); + else + return CHARSET_CHARS (cs); +} + +#define XCHARSET_BYTE_SIZE(ccs) CHARSET_BYTE_SIZE (XCHARSET (ccs)) + +int decoding_table_check_elements (Lisp_Object v, int dim, int ccs_len); +int +decoding_table_check_elements (Lisp_Object v, int dim, int ccs_len) +{ + int i; + + if (XVECTOR_LENGTH (v) > ccs_len) + return -1; + + for (i = 0; i < XVECTOR_LENGTH (v); i++) + { + Lisp_Object c = XVECTOR_DATA(v)[i]; + + if (!NILP (c) && !CHARP (c)) + { + if (VECTORP (c)) + { + int ret = decoding_table_check_elements (c, dim - 1, ccs_len); + if (ret) + return ret; + } + else + return -2; + } + } + return 0; +} + +INLINE_HEADER void +decoding_table_remove_char (Lisp_Object v, int dim, int byte_offset, + int code_point); +INLINE_HEADER void +decoding_table_remove_char (Lisp_Object v, int dim, int byte_offset, + int code_point) +{ + int i = -1; + + while (dim > 0) + { Lisp_Object nv; - int i = -1; - int ccs_len; - int dim; - int code_point; - - /* ad-hoc method for `ascii' */ - if ((XCHARSET_CHARS (ccs) == 94) && - (XCHARSET_BYTE_OFFSET (ccs) != 33)) - ccs_len = 128 - XCHARSET_BYTE_OFFSET (ccs); + + dim--; + i = ((code_point >> (8 * dim)) & 255) - byte_offset; + nv = XVECTOR_DATA(v)[i]; + if (!VECTORP (nv)) + break; + v = nv; + } + if (i >= 0) + XVECTOR_DATA(v)[i] = Qnil; +} + +INLINE_HEADER void +decoding_table_put_char (Lisp_Object v, int dim, int byte_offset, + int code_point, Lisp_Object character); +INLINE_HEADER void +decoding_table_put_char (Lisp_Object v, int dim, int byte_offset, + int code_point, Lisp_Object character) +{ + int i = -1; + Lisp_Object nv; + int ccs_len = XVECTOR_LENGTH (v); + + while (dim > 0) + { + dim--; + i = ((code_point >> (8 * dim)) & 255) - byte_offset; + nv = XVECTOR_DATA(v)[i]; + if (dim > 0) + { + if (!VECTORP (nv)) + nv = (XVECTOR_DATA(v)[i] = make_older_vector (ccs_len, Qnil)); + v = nv; + } else - ccs_len = XCHARSET_CHARS (ccs); - + break; + } + XVECTOR_DATA(v)[i] = character; +} + +Lisp_Object +put_char_ccs_code_point (Lisp_Object character, + Lisp_Object ccs, Lisp_Object value) +{ + Lisp_Object encoding_table; + + if (!EQ (XCHARSET_NAME (ccs), Qucs) + || (XCHAR (character) != XINT (value))) + { + Lisp_Object v = XCHARSET_DECODING_TABLE (ccs); + int dim = XCHARSET_DIMENSION (ccs); + int ccs_len = XCHARSET_BYTE_SIZE (ccs); + int byte_offset = XCHARSET_BYTE_OFFSET (ccs); + int code_point; + if (CONSP (value)) - { + { /* obsolete representation: value must be a list of bytes */ Lisp_Object ret = Fcar (value); + Lisp_Object rest; if (!INTP (ret)) signal_simple_error ("Invalid value for coded-charset", value); @@ -610,132 +946,85 @@ Store CHARACTER's ATTRIBUTE with VALUE. rest = Fcdr (value); while (!NILP (rest)) { - int i; + int j; if (!CONSP (rest)) - signal_simple_error ("Invalid value for coded-charset", value); + signal_simple_error ("Invalid value for coded-charset", + value); ret = Fcar (rest); if (!INTP (ret)) - signal_simple_error ("Invalid value for coded-charset", value); - i = XINT (ret); + signal_simple_error ("Invalid value for coded-charset", + value); + j = XINT (ret); if (XCHARSET_GRAPHIC (ccs) == 1) - i &= 0x7F; - code_point = (code_point << 8) | i; + j &= 0x7F; + code_point = (code_point << 8) | j; rest = Fcdr (rest); } value = make_int (code_point); } else if (INTP (value)) { + code_point = XINT (value); if (XCHARSET_GRAPHIC (ccs) == 1) - value = make_int (XINT (value) & 0x7F7F7F7F); + { + code_point &= 0x7F7F7F7F; + value = make_int (code_point); + } } else signal_simple_error ("Invalid value for coded-charset", value); - attribute = ccs; - cpos = Fget_char_attribute (character, attribute); if (VECTORP (v)) { + Lisp_Object cpos = Fget_char_attribute (character, ccs); if (!NILP (cpos)) { - dim = XCHARSET_DIMENSION (ccs); - code_point = XINT (cpos); - while (dim > 0) - { - dim--; - i = ((code_point >> (8 * dim)) & 255) - - XCHARSET_BYTE_OFFSET (ccs); - nv = XVECTOR_DATA(v)[i]; - if (!VECTORP (nv)) - break; - v = nv; - } - if (i >= 0) - XVECTOR_DATA(v)[i] = Qnil; - v = XCHARSET_DECODING_TABLE (ccs); + decoding_table_remove_char (v, dim, byte_offset, XINT (cpos)); } } else { - XCHARSET_DECODING_TABLE (ccs) = v = make_vector (ccs_len, Qnil); + XCHARSET_DECODING_TABLE (ccs) + = v = make_older_vector (ccs_len, Qnil); } - dim = XCHARSET_DIMENSION (ccs); - code_point = XINT (value); - i = -1; - while (dim > 0) - { - dim--; - i = ((code_point >> (8 * dim)) & 255) - XCHARSET_BYTE_OFFSET (ccs); - nv = XVECTOR_DATA(v)[i]; - if (dim > 0) - { - if (!VECTORP (nv)) - nv = (XVECTOR_DATA(v)[i] = make_vector (ccs_len, Qnil)); - v = nv; - } - else - break; - } - XVECTOR_DATA(v)[i] = character; + decoding_table_put_char (v, dim, byte_offset, code_point, character); } - else if (EQ (attribute, Q_decomposition)) + if (NILP (encoding_table = XCHARSET_ENCODING_TABLE (ccs))) { - Lisp_Object rest = value; - Lisp_Object table = Vcharacter_composition_table; - - if (!CONSP (value)) - signal_simple_error ("Invalid value for ->decomposition", - value); - - while (CONSP (rest)) - { - Lisp_Object v = Fcar (rest); - Lisp_Object ntable; - Emchar c - = to_char_code (v, "Invalid value for ->decomposition", value); - - rest = Fcdr (rest); - if (!CONSP (rest)) - { - put_char_code_table (c, character, table); - break; - } - else - { - ntable = get_char_code_table (c, table); - if (!CHAR_CODE_TABLE_P (ntable)) - { - ntable = make_char_code_table (Qnil); - put_char_code_table (c, ntable, table); - } - table = ntable; - } - } + XCHARSET_ENCODING_TABLE (ccs) + = encoding_table = make_char_id_table (Qnil, -1); } - else if (EQ (attribute, Q_ucs)) - { - Lisp_Object ret; - Emchar c; + put_char_id_table (XCHAR (character), value, encoding_table); + return Qt; +} - if (!INTP (value)) - signal_simple_error ("Invalid value for ->ucs", value); +Lisp_Object +remove_char_ccs (Lisp_Object character, Lisp_Object ccs) +{ + Lisp_Object decoding_table = XCHARSET_DECODING_TABLE (ccs); + Lisp_Object encoding_table = XCHARSET_ENCODING_TABLE (ccs); - c = XINT (value); + if (VECTORP (decoding_table)) + { + Lisp_Object cpos = Fget_char_attribute (character, ccs); - ret = get_char_code_table (c, Vcharacter_variant_table); - if (NILP (Fmemq (character, ret))) + if (!NILP (cpos)) { - put_char_code_table (c, Fcons (character, ret), - Vcharacter_variant_table); + decoding_table_remove_char (decoding_table, + XCHARSET_DIMENSION (ccs), + XCHARSET_BYTE_OFFSET (ccs), + XINT (cpos)); } } - return put_char_attribute (character, attribute, value); + if (CHAR_ID_TABLE_P (encoding_table)) + { + put_char_id_table (XCHAR (character), Qnil, encoding_table); + } + return Qt; } -Lisp_Object Qucs; - EXFUN (Fmake_char, 3); EXFUN (Fdecode_char, 2); @@ -747,6 +1036,9 @@ Store character's ATTRIBUTES. Lisp_Object rest = attributes; Lisp_Object code = Fcdr (Fassq (Qucs, attributes)); Lisp_Object character; +#if 0 + Lisp_Object daikanwa = Qnil; +#endif if (NILP (code)) { @@ -757,7 +1049,9 @@ Store character's ATTRIBUTES. if (!LISTP (cell)) signal_simple_error ("Invalid argument", attributes); - if (!NILP (ccs = Ffind_charset (Fcar (cell)))) + if (!NILP (ccs = Ffind_charset (Fcar (cell))) + && ((XCHARSET_FINAL (ccs) != 0) || + (XCHARSET_UCS_MAX (ccs) > 0)) ) { cell = Fcdr (cell); if (CONSP (cell)) @@ -788,14 +1082,39 @@ Store character's ATTRIBUTES. while (CONSP (rest)) { Lisp_Object cell = Fcar (rest); +#if 0 + Lisp_Object key = Fcar (cell); + Lisp_Object value = Fcdr (cell); +#endif if (!LISTP (cell)) signal_simple_error ("Invalid argument", attributes); + +#if 0 + if (EQ (key, Qmorohashi_daikanwa)) + { + size_t len; + GET_EXTERNAL_LIST_LENGTH (value, len); + + if (len == 1) + { + if (NILP (daikanwa)) + daikanwa = Fcdr (Fassq (Qideograph_daikanwa, rest)); + if (EQ (Fcar (value), daikanwa)) + goto ignored; + } + } + else if (EQ (key, Qideograph_daikanwa)) + daikanwa = value; +#endif + Fput_char_attribute (character, Fcar (cell), Fcdr (cell)); +#if 0 + ignored: +#endif rest = Fcdr (rest); } - return - get_char_code_table (XCHAR (character), Vcharacter_attribute_table); + return character; } Lisp_Object Vutf_2000_version; @@ -839,11 +1158,12 @@ Lisp_Object Qascii, #ifdef UTF2000 Qucs_bmp, Qlatin_viscii, + Qlatin_tcvn5712, Qlatin_viscii_lower, Qlatin_viscii_upper, Qvietnamese_viscii_lower, Qvietnamese_viscii_upper, - Qideograph_daikanwa, + Qmojikyo, Qmojikyo_pj_1, Qmojikyo_pj_2, Qmojikyo_pj_3, @@ -875,13 +1195,6 @@ 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. @@ -992,7 +1305,7 @@ non_ascii_set_charptr_emchar (Bufbyte *str, Emchar c) Use the macro charptr_emchar() instead. */ Emchar -non_ascii_charptr_emchar (CONST Bufbyte *str) +non_ascii_charptr_emchar (const Bufbyte *str) { #ifdef UTF2000 Bufbyte b; @@ -1148,7 +1461,7 @@ non_ascii_valid_char_p (Emchar ch) charptr_copy_char() instead. */ Bytecount -non_ascii_charptr_copy_char (CONST Bufbyte *ptr, Bufbyte *str) +non_ascii_charptr_copy_char (const Bufbyte *ptr, Bufbyte *str) { Bufbyte *strptr = str; *strptr = *ptr++; @@ -1241,7 +1554,7 @@ Lstream_funget_emchar (Lstream *stream, Emchar ch) static Lisp_Object mark_charset (Lisp_Object obj) { - struct Lisp_Charset *cs = XCHARSET (obj); + Lisp_Charset *cs = XCHARSET (obj); mark_object (cs->short_name); mark_object (cs->long_name); @@ -1249,7 +1562,8 @@ mark_charset (Lisp_Object obj) mark_object (cs->registry); mark_object (cs->ccl_program); #ifdef UTF2000 - mark_object (cs->decoding_table); + /* mark_object (cs->encoding_table); */ + /* mark_object (cs->decoding_table); */ #endif return cs->name; } @@ -1257,7 +1571,7 @@ mark_charset (Lisp_Object obj) static void print_charset (Lisp_Object obj, Lisp_Object printcharfun, int escapeflag) { - struct Lisp_Charset *cs = XCHARSET (obj); + Lisp_Charset *cs = XCHARSET (obj); char buf[200]; if (print_readably) @@ -1287,9 +1601,16 @@ print_charset (Lisp_Object obj, Lisp_Object printcharfun, int escapeflag) } static const struct lrecord_description charset_description[] = { - { XD_LISP_OBJECT, offsetof(struct Lisp_Charset, name), 7 }, + { XD_LISP_OBJECT, offsetof (Lisp_Charset, name) }, + { XD_LISP_OBJECT, offsetof (Lisp_Charset, doc_string) }, + { XD_LISP_OBJECT, offsetof (Lisp_Charset, registry) }, + { XD_LISP_OBJECT, offsetof (Lisp_Charset, short_name) }, + { XD_LISP_OBJECT, offsetof (Lisp_Charset, long_name) }, + { XD_LISP_OBJECT, offsetof (Lisp_Charset, reverse_direction_charset) }, + { XD_LISP_OBJECT, offsetof (Lisp_Charset, ccl_program) }, #ifdef UTF2000 - { XD_LISP_OBJECT, offsetof(struct Lisp_Charset, decoding_table), 2 }, + { XD_LISP_OBJECT, offsetof (Lisp_Charset, decoding_table) }, + { XD_LISP_OBJECT, offsetof (Lisp_Charset, encoding_table) }, #endif { XD_END } }; @@ -1297,7 +1618,7 @@ static const struct lrecord_description charset_description[] = { DEFINE_LRECORD_IMPLEMENTATION ("charset", charset, mark_charset, print_charset, 0, 0, 0, charset_description, - struct Lisp_Charset); + Lisp_Charset); /* Make a new charset. */ static Lisp_Object @@ -1313,8 +1634,10 @@ make_charset (Charset_ID id, Lisp_Object name, { unsigned char type = 0; Lisp_Object obj; - struct Lisp_Charset *cs = - alloc_lcrecord_type (struct Lisp_Charset, &lrecord_charset); + Lisp_Charset *cs = alloc_lcrecord_type (Lisp_Charset, &lrecord_charset); + + zero_lcrecord (cs); + XSETCHARSET (obj, cs); CHARSET_ID (cs) = id; @@ -1333,6 +1656,7 @@ 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; @@ -1434,24 +1758,24 @@ get_unallocated_leading_byte (int dimension) Charset_ID lb; #ifdef UTF2000 - if (next_allocated_leading_byte > MAX_LEADING_BYTE_PRIVATE) + if (chlook->next_allocated_leading_byte > MAX_LEADING_BYTE_PRIVATE) lb = 0; else - lb = next_allocated_leading_byte++; + lb = chlook->next_allocated_leading_byte++; #else if (dimension == 1) { - if (next_allocated_1_byte_leading_byte > MAX_LEADING_BYTE_PRIVATE_1) + if (chlook->next_allocated_1_byte_leading_byte > MAX_LEADING_BYTE_PRIVATE_1) lb = 0; else - lb = next_allocated_1_byte_leading_byte++; + lb = chlook->next_allocated_1_byte_leading_byte++; } else { - if (next_allocated_2_byte_leading_byte > MAX_LEADING_BYTE_PRIVATE_2) + if (chlook->next_allocated_2_byte_leading_byte > MAX_LEADING_BYTE_PRIVATE_2) lb = 0; else - lb = next_allocated_2_byte_leading_byte++; + lb = chlook->next_allocated_2_byte_leading_byte++; } #endif @@ -1618,12 +1942,12 @@ encode_builtin_char_1 (Emchar c, Lisp_Object* charset) *charset = Vcharset_latin_iso8859_1; return c & 0x7F; } + /* else if ((MIN_CHAR_GREEK <= c) && (c <= MAX_CHAR_GREEK)) { *charset = Vcharset_greek_iso8859_7; return c - MIN_CHAR_GREEK + 0x20; } - /* else if ((MIN_CHAR_CYRILLIC <= c) && (c <= MAX_CHAR_CYRILLIC)) { *charset = Vcharset_cyrillic_iso8859_5; @@ -1658,11 +1982,18 @@ encode_builtin_char_1 (Emchar c, Lisp_Object* charset) *charset = Vcharset_ucs; return c; } + /* else if (c <= MAX_CHAR_DAIKANWA) { *charset = Vcharset_ideograph_daikanwa; return c - MIN_CHAR_DAIKANWA; } + */ + else if (c <= MAX_CHAR_MOJIKYO) + { + *charset = Vcharset_mojikyo; + return c - MIN_CHAR_MOJIKYO; + } else if (c < MIN_CHAR_94) { *charset = Vcharset_ucs; @@ -1793,7 +2124,7 @@ add_charset_to_list_mapper (Lisp_Object key, Lisp_Object value, (struct charset_list_closure*) charset_list_closure; Lisp_Object *charset_list = chcl->charset_list; - *charset_list = Fcons (XCHARSET_NAME (value), *charset_list); + *charset_list = Fcons (key /* XCHARSET_NAME (value) */, *charset_list); return 0; } @@ -2038,7 +2369,7 @@ NEW-NAME is the name of the new charset. Return the new charset. int id, chars, dimension, columns, graphic, final; int direction; Lisp_Object registry, doc_string, short_name, long_name; - struct Lisp_Charset *cs; + Lisp_Charset *cs; charset = Fget_charset (charset); if (!NILP (XCHARSET_REVERSE_DIRECTION_CHARSET (charset))) @@ -2207,7 +2538,7 @@ Recognized properties are those listed in `make-charset', as well as */ (charset, prop)) { - struct Lisp_Charset *cs; + Lisp_Charset *cs; charset = Fget_charset (charset); cs = XCHARSET (charset); @@ -2302,39 +2633,42 @@ Set mapping-table of CHARSET to TABLE. (charset, table)) { struct Lisp_Charset *cs; - Lisp_Object old_table; size_t i; + int byte_offset; charset = Fget_charset (charset); cs = XCHARSET (charset); - if (EQ (table, Qnil)) + if (NILP (table)) { - CHARSET_DECODING_TABLE(cs) = table; + if (VECTORP (CHARSET_DECODING_TABLE(cs))) + make_vector_newer (CHARSET_DECODING_TABLE(cs)); + CHARSET_DECODING_TABLE(cs) = Qnil; return table; } else if (VECTORP (table)) { - int ccs_len; - - /* ad-hoc method for `ascii' */ - if ((CHARSET_CHARS (cs) == 94) && - (CHARSET_BYTE_OFFSET (cs) != 33)) - ccs_len = 128 - CHARSET_BYTE_OFFSET (cs); - else - ccs_len = CHARSET_CHARS (cs); - - if (XVECTOR_LENGTH (table) > ccs_len) - args_out_of_range (table, make_int (CHARSET_CHARS (cs))); - old_table = CHARSET_DECODING_TABLE(cs); - CHARSET_DECODING_TABLE(cs) = table; + int ccs_len = CHARSET_BYTE_SIZE (cs); + int ret = decoding_table_check_elements (table, + CHARSET_DIMENSION (cs), + ccs_len); + if (ret) + { + if (ret == -1) + signal_simple_error ("Too big table", table); + else if (ret == -2) + signal_simple_error ("Invalid element is found", table); + else + signal_simple_error ("Something wrong", table); + } + CHARSET_DECODING_TABLE(cs) = Qnil; } else signal_error (Qwrong_type_argument, list2 (build_translated_string ("vector-or-nil-p"), table)); - /* signal_simple_error ("Wrong type argument: vector-or-nil-p", table); */ + byte_offset = CHARSET_BYTE_OFFSET (cs); switch (CHARSET_DIMENSION (cs)) { case 1: @@ -2343,9 +2677,8 @@ Set mapping-table of CHARSET to TABLE. Lisp_Object c = XVECTOR_DATA(table)[i]; if (CHARP (c)) - put_char_attribute - (c, charset, - make_int (i + CHARSET_BYTE_OFFSET (cs))); + put_char_ccs_code_point (c, charset, + make_int (i + byte_offset)); } break; case 2: @@ -2357,25 +2690,21 @@ Set mapping-table of CHARSET to TABLE. { size_t j; - if (XVECTOR_LENGTH (v) > CHARSET_CHARS (cs)) - { - CHARSET_DECODING_TABLE(cs) = old_table; - args_out_of_range (v, make_int (CHARSET_CHARS (cs))); - } for (j = 0; j < XVECTOR_LENGTH (v); j++) { Lisp_Object c = XVECTOR_DATA(v)[j]; if (CHARP (c)) - put_char_attribute + put_char_ccs_code_point (c, charset, - make_int ( ((i + CHARSET_BYTE_OFFSET (cs)) << 8) - | (j + CHARSET_BYTE_OFFSET (cs)) )); + make_int ( ( (i + byte_offset) << 8 ) + | (j + byte_offset) + ) ); } } else if (CHARP (v)) - put_char_attribute (v, charset, - make_int (i + CHARSET_BYTE_OFFSET (cs))); + put_char_ccs_code_point (v, charset, + make_int (i + byte_offset)); } break; } @@ -2397,11 +2726,83 @@ Make a character from CHARSET and code-point CODE. int c; charset = Fget_charset (charset); + CHECK_INT (code); c = XINT (code); if (XCHARSET_GRAPHIC (charset) == 1) c &= 0x7F7F7F7F; return make_char (DECODE_CHAR (charset, c)); } + +DEFUN ("decode-builtin-char", Fdecode_builtin_char, 2, 2, 0, /* +Make a builtin character from CHARSET and code-point CODE. +*/ + (charset, code)) +{ + int c; + int final; + + charset = Fget_charset (charset); + CHECK_INT (code); + c = XINT (code); + + if ((final = XCHARSET_FINAL (charset)) >= '0') + { + if (XCHARSET_DIMENSION (charset) == 1) + { + switch (XCHARSET_CHARS (charset)) + { + case 94: + return + make_char (MIN_CHAR_94 + (final - '0') * 94 + + ((c & 0x7F) - 33)); + case 96: + return + make_char (MIN_CHAR_96 + (final - '0') * 96 + + ((c & 0x7F) - 32)); + default: + return Fdecode_char (charset, code); + } + } + else + { + switch (XCHARSET_CHARS (charset)) + { + case 94: + return + make_char (MIN_CHAR_94x94 + + (final - '0') * 94 * 94 + + (((c >> 8) & 0x7F) - 33) * 94 + + ((c & 0x7F) - 33)); + case 96: + return + make_char (MIN_CHAR_96x96 + + (final - '0') * 96 * 96 + + (((c >> 8) & 0x7F) - 32) * 96 + + ((c & 0x7F) - 32)); + default: + return Fdecode_char (charset, code); + } + } + } + else if (XCHARSET_UCS_MAX (charset)) + { + Emchar cid + = (XCHARSET_DIMENSION (charset) == 1 + ? + c - XCHARSET_BYTE_OFFSET (charset) + : + ((c >> 8) - XCHARSET_BYTE_OFFSET (charset)) + * XCHARSET_CHARS (charset) + + (c & 0xFF) - XCHARSET_BYTE_OFFSET (charset)) + - XCHARSET_CODE_OFFSET (charset) + XCHARSET_UCS_MIN (charset); + if ((cid < XCHARSET_UCS_MIN (charset)) + || (XCHARSET_UCS_MAX (charset) < cid)) + return Fdecode_char (charset, code); + return make_char (cid); + } + else + return Fdecode_char (charset, code); +} #endif DEFUN ("make-char", Fmake_char, 2, 3, 0, /* @@ -2412,7 +2813,7 @@ character s with caron. */ (charset, arg1, arg2)) { - struct Lisp_Charset *cs; + Lisp_Charset *cs; int a1, a2; int lowlim, highlim; @@ -2623,6 +3024,12 @@ Return a string of the characters comprising a composite character. void syms_of_mule_charset (void) { +#ifdef UTF2000 + INIT_LRECORD_IMPLEMENTATION (byte_table); + INIT_LRECORD_IMPLEMENTATION (char_id_table); +#endif + INIT_LRECORD_IMPLEMENTATION (charset); + DEFSUBR (Fcharsetp); DEFSUBR (Ffind_charset); DEFSUBR (Fget_charset); @@ -2642,9 +3049,11 @@ syms_of_mule_charset (void) DEFSUBR (Fset_charset_ccl_program); DEFSUBR (Fset_charset_registry); #ifdef UTF2000 + DEFSUBR (Fchar_attribute_list); DEFSUBR (Fchar_attribute_alist); DEFSUBR (Fget_char_attribute); DEFSUBR (Fput_char_attribute); + DEFSUBR (Fremove_char_attribute); DEFSUBR (Fdefine_char); DEFSUBR (Fchar_variants); DEFSUBR (Fget_composite_char); @@ -2654,6 +3063,7 @@ syms_of_mule_charset (void) #ifdef UTF2000 DEFSUBR (Fdecode_char); + DEFSUBR (Fdecode_builtin_char); #endif DEFSUBR (Fmake_char); DEFSUBR (Fchar_charset); @@ -2723,11 +3133,13 @@ syms_of_mule_charset (void) defsymbol (&Qucs, "ucs"); defsymbol (&Qucs_bmp, "ucs-bmp"); defsymbol (&Qlatin_viscii, "latin-viscii"); + defsymbol (&Qlatin_tcvn5712, "latin-tcvn5712"); defsymbol (&Qlatin_viscii_lower, "latin-viscii-lower"); defsymbol (&Qlatin_viscii_upper, "latin-viscii-upper"); defsymbol (&Qvietnamese_viscii_lower, "vietnamese-viscii-lower"); defsymbol (&Qvietnamese_viscii_upper, "vietnamese-viscii-upper"); defsymbol (&Qideograph_daikanwa, "ideograph-daikanwa"); + defsymbol (&Qmojikyo, "mojikyo"); defsymbol (&Qmojikyo_pj_1, "mojikyo-pj-1"); defsymbol (&Qmojikyo_pj_2, "mojikyo-pj-2"); defsymbol (&Qmojikyo_pj_3, "mojikyo-pj-3"); @@ -2786,10 +3198,10 @@ vars_of_mule_charset (void) #endif #ifdef UTF2000 - next_allocated_leading_byte = MIN_LEADING_BYTE_PRIVATE; + chlook->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; + chlook->next_allocated_1_byte_leading_byte = MIN_LEADING_BYTE_PRIVATE_1; + chlook->next_allocated_2_byte_leading_byte = MIN_LEADING_BYTE_PRIVATE_2; #endif #ifndef UTF2000 @@ -2801,19 +3213,16 @@ Leading-code of private TYPE9N charset of column-width 1. #endif #ifdef UTF2000 - Vutf_2000_version = build_string("0.14 (Kawachi-Katakami)"); + Vutf_2000_version = build_string("0.16 (Ōji)"); 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); - staticpro (&Vcharacter_composition_table); - Vcharacter_composition_table = make_char_code_table (Qnil); + Vcharacter_composition_table = make_char_id_table (Qnil, -1); staticpro (&Vcharacter_variant_table); - Vcharacter_variant_table = make_char_code_table (Qnil); + Vcharacter_variant_table = make_char_id_table (Qnil, 0); Vdefault_coded_charset_priority_list = Qnil; DEFVAR_LISP ("default-coded-charset-priority-list", @@ -2834,6 +3243,10 @@ complex_vars_of_mule_charset (void) ease of access. */ #ifdef UTF2000 + staticpro (&Vchar_attribute_hash_table); + Vchar_attribute_hash_table + = make_lisp_hash_table (16, HASH_TABLE_NON_WEAK, HASH_TABLE_EQ); + staticpro (&Vcharset_ucs); Vcharset_ucs = make_charset (LEADING_BYTE_UCS, Qucs, 256, 4, @@ -2855,14 +3268,10 @@ complex_vars_of_mule_charset (void) #else # define MIN_CHAR_THAI 0 # define MAX_CHAR_THAI 0 -# define MIN_CHAR_GREEK 0 -# define MAX_CHAR_GREEK 0 # define MIN_CHAR_HEBREW 0 # define MAX_CHAR_HEBREW 0 # define MIN_CHAR_HALFWIDTH_KATAKANA 0 # define MAX_CHAR_HALFWIDTH_KATAKANA 0 -# define MIN_CHAR_CYRILLIC 0 -# define MAX_CHAR_CYRILLIC 0 #endif staticpro (&Vcharset_ascii); Vcharset_ascii = @@ -2935,7 +3344,9 @@ complex_vars_of_mule_charset (void) build_string ("ISO8859-7 (Greek)"), build_string ("ISO8859-7 (Greek)"), build_string ("iso8859-7"), - Qnil, MIN_CHAR_GREEK, MAX_CHAR_GREEK, 0, 32); + Qnil, + 0 /* MIN_CHAR_GREEK */, + 0 /* MAX_CHAR_GREEK */, 0, 32); staticpro (&Vcharset_arabic_iso8859_6); Vcharset_arabic_iso8859_6 = make_charset (LEADING_BYTE_ARABIC_ISO8859_6, Qarabic_iso8859_6, 96, 1, @@ -3076,6 +3487,15 @@ complex_vars_of_mule_charset (void) build_string (CHINESE_CNS_PLANE_RE("2")), Qnil, 0, 0, 0, 33); #ifdef UTF2000 + staticpro (&Vcharset_latin_tcvn5712); + Vcharset_latin_tcvn5712 = + make_charset (LEADING_BYTE_LATIN_TCVN5712, Qlatin_tcvn5712, 96, 1, + 1, 1, 'Z', CHARSET_LEFT_TO_RIGHT, + build_string ("TCVN 5712"), + build_string ("TCVN 5712 (VSCII-2)"), + build_string ("Vietnamese TCVN 5712:1983 (VSCII-2)"), + build_string ("tcvn5712-1"), + Qnil, 0, 0, 0, 32); staticpro (&Vcharset_latin_viscii_lower); Vcharset_latin_viscii_lower = make_charset (LEADING_BYTE_LATIN_VISCII_LOWER, Qlatin_viscii_lower, 96, 1, @@ -3112,6 +3532,15 @@ complex_vars_of_mule_charset (void) build_string ("Daikanwa dictionary by MOROHASHI Tetsuji"), build_string ("Daikanwa"), Qnil, MIN_CHAR_DAIKANWA, MAX_CHAR_DAIKANWA, 0, 0); + staticpro (&Vcharset_mojikyo); + Vcharset_mojikyo = + make_charset (LEADING_BYTE_MOJIKYO, Qmojikyo, 256, 3, + 2, 2, 0, CHARSET_LEFT_TO_RIGHT, + build_string ("Mojikyo"), + build_string ("Mojikyo"), + build_string ("Konjaku-Mojikyo"), + build_string (""), + Qnil, MIN_CHAR_MOJIKYO, MAX_CHAR_MOJIKYO, 0, 0); staticpro (&Vcharset_mojikyo_pj_1); Vcharset_mojikyo_pj_1 = make_charset (LEADING_BYTE_MOJIKYO_PJ_1, Qmojikyo_pj_1, 94, 2,