src/vmcore/utf8.h

   1 /* src/vmcore/utf8.h - utf8 string functions
   2
   3    Copyright (C) 1996-2005, 2006, 2007 R. Grafl, A. Krall, C. Kruegel,
   4    C. Oates, R. Obermaisser, M. Platter, M. Probst, S. Ring,
   5    E. Steiner, C. Thalinger, D. Thuernbeck, P. Tomsich, C. Ullrich,
   6    J. Wenninger, Institut f. Computersprachen - TU Wien
   7
   8    This file is part of CACAO.
   9
  10    This program is free software; you can redistribute it and/or
  11    modify it under the terms of the GNU General Public License as
  12    published by the Free Software Foundation; either version 2, or (at
  13    your option) any later version.
  14
  15    This program is distributed in the hope that it will be useful, but
  16    WITHOUT ANY WARRANTY; without even the implied warranty of
  17    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  18    General Public License for more details.
  19
  20    You should have received a copy of the GNU General Public License
  21    along with this program; if not, write to the Free Software
  22    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
  23    02110-1301, USA.
  24
  25    $Id: utf8.h 7257 2007-01-29 23:07:40Z twisti $
  26
  27 */
  28
  29
  30 #ifndef _UTF_H
  31 #define _UTF_H
  32
  33 /* forward typedefs ***********************************************************/
  34
  35 typedef struct utf utf;
  36
  37 #include "config.h"
  38
  39 #include <stdio.h>
  40
  41 #include "vm/types.h"
  42
  43 #include "vm/global.h"
  44
  45
  46 /* data structure for utf8 symbols ********************************************/
  47
  48 struct utf {
  49         utf  *hashlink;                     /* link for external hash chain       */
  50         s4    blength;                      /* text length in bytes               */
  51         char *text;                         /* pointer to text                    */
  52 };
  53
  54 /* to determine the end of utf strings */
  55
  56 #define UTF_END(u)    ((char *) u->text + u->blength)
  57
  58
  59 /* utf-symbols for pointer comparison of frequently used strings **************/
  60
  61 extern utf *utf_java_lang_Object;
  62
  63 extern utf *utf_java_lang_Class;
  64 extern utf *utf_java_lang_ClassLoader;
  65 extern utf *utf_java_lang_Cloneable;
  66 extern utf *utf_java_lang_SecurityManager;
  67 extern utf *utf_java_lang_String;
  68 extern utf *utf_java_lang_System;
  69 extern utf *utf_java_lang_ThreadGroup;
  70 extern utf *utf_java_io_Serializable;
  71
  72 extern utf *utf_java_lang_Throwable;
  73 extern utf *utf_java_lang_Error;
  74
  75 extern utf *utf_java_lang_AbstractMethodError;
  76 extern utf *utf_java_lang_ClassCircularityError;
  77 extern utf *utf_java_lang_ClassFormatError;
  78 extern utf *utf_java_lang_ExceptionInInitializerError;
  79 extern utf *utf_java_lang_IncompatibleClassChangeError;
  80 extern utf *utf_java_lang_InstantiationError;
  81 extern utf *utf_java_lang_InternalError;
  82 extern utf *utf_java_lang_LinkageError;
  83 extern utf *utf_java_lang_NoClassDefFoundError;
  84 extern utf *utf_java_lang_NoSuchFieldError;
  85 extern utf *utf_java_lang_NoSuchMethodError;
  86 extern utf *utf_java_lang_OutOfMemoryError;
  87 extern utf *utf_java_lang_UnsatisfiedLinkError;
  88 extern utf *utf_java_lang_UnsupportedClassVersionError;
  89 extern utf *utf_java_lang_VerifyError;
  90 extern utf *utf_java_lang_VirtualMachineError;
  91
  92 #if defined(WITH_CLASSPATH_GNU)
  93 extern utf *utf_java_lang_VMThrowable;
  94 #endif
  95
  96 extern utf *utf_java_lang_Exception;
  97 extern utf *utf_java_lang_ArithmeticException;
  98 extern utf *utf_java_lang_ArrayIndexOutOfBoundsException;
  99 extern utf *utf_java_lang_ArrayStoreException;
 100 extern utf *utf_java_lang_ClassCastException;
 101 extern utf *utf_java_lang_ClassNotFoundException;
 102 extern utf *utf_java_lang_CloneNotSupportedException;
 103 extern utf *utf_java_lang_IllegalAccessException;
 104 extern utf *utf_java_lang_IllegalArgumentException;
 105 extern utf *utf_java_lang_IllegalMonitorStateException;
 106 extern utf *utf_java_lang_InstantiationException;
 107 extern utf *utf_java_lang_InterruptedException;
 108 extern utf *utf_java_lang_InvocationTargetException;
 109 extern utf *utf_java_lang_NegativeArraySizeException;
 110 extern utf *utf_java_lang_NullPointerException;
 111 extern utf *utf_java_lang_StringIndexOutOfBoundsException;
 112
 113 #if defined(ENABLE_JAVASE)
 114 extern utf* utf_java_lang_Void;
 115 #endif
 116
 117 extern utf* utf_java_lang_Boolean;
 118 extern utf* utf_java_lang_Byte;
 119 extern utf* utf_java_lang_Character;
 120 extern utf* utf_java_lang_Short;
 121 extern utf* utf_java_lang_Integer;
 122 extern utf* utf_java_lang_Long;
 123 extern utf* utf_java_lang_Float;
 124 extern utf* utf_java_lang_Double;
 125
 126 #if defined(ENABLE_JAVASE)
 127 extern utf *utf_java_lang_StackTraceElement;
 128 extern utf *utf_java_lang_reflect_Constructor;
 129 extern utf *utf_java_lang_reflect_Field;
 130 extern utf *utf_java_lang_reflect_Method;
 131 extern utf *utf_java_util_Vector;
 132 #endif
 133
 134 extern utf *utf_InnerClasses;
 135 extern utf *utf_ConstantValue;
 136 extern utf *utf_Code;
 137 extern utf *utf_Exceptions;
 138 extern utf *utf_LineNumberTable;
 139 extern utf *utf_SourceFile;
 140
 141 #if defined(ENABLE_JAVASE)
 142 extern utf *utf_EnclosingMethod;
 143 extern utf *utf_Signature;
 144 extern utf *utf_RuntimeVisibleAnnotations;
 145 extern utf *utf_StackMapTable;
 146 #endif
 147
 148 extern utf *utf_init;
 149 extern utf *utf_clinit;
 150 extern utf *utf_clone;
 151 extern utf *utf_finalize;
 152 extern utf *utf_run;
 153
 154 extern utf *utf_add;
 155 extern utf *utf_remove;
 156 extern utf *utf_addThread;
 157 extern utf *utf_removeThread;
 158 extern utf *utf_put;
 159 extern utf *utf_get;
 160 extern utf *utf_value;
 161
 162 extern utf *utf_fillInStackTrace;
 163 extern utf *utf_getSystemClassLoader;
 164 extern utf *utf_loadClass;
 165 extern utf *utf_printStackTrace;
 166
 167 extern utf *utf_Z;
 168 extern utf *utf_B;
 169 extern utf *utf_C;
 170 extern utf *utf_S;
 171 extern utf *utf_I;
 172 extern utf *utf_J;
 173 extern utf *utf_F;
 174 extern utf *utf_D;
 175
 176 extern utf *utf_void__void;
 177 extern utf *utf_boolean__void;
 178 extern utf *utf_byte__void;
 179 extern utf *utf_char__void;
 180 extern utf *utf_short__void;
 181 extern utf *utf_int__void;
 182 extern utf *utf_long__void;
 183 extern utf *utf_float__void;
 184 extern utf *utf_double__void;
 185
 186 extern utf *utf_void__java_lang_ClassLoader;
 187 extern utf *utf_void__java_lang_Object;
 188 extern utf *utf_void__java_lang_Throwable;
 189 extern utf *utf_java_lang_Object__java_lang_Object;
 190 extern utf *utf_java_lang_String__void;
 191 extern utf *utf_java_lang_String__java_lang_Class;
 192 extern utf *utf_java_lang_Thread__V;
 193 extern utf *utf_java_lang_Throwable__void;
 194
 195 extern utf *utf_not_named_yet;
 196 extern utf *utf_null;
 197 extern utf *array_packagename;
 198
 199
 200 /* function prototypes ********************************************************/
 201
 202 /* initialize the utf8 subsystem */
 203 bool utf8_init(void);
 204
 205 u4 utf_hashkey(const char *text, u4 length);
 206 u4 utf_full_hashkey(const char *text, u4 length);
 207
 208 /* determine hashkey of a unicode-symbol */
 209 u4 unicode_hashkey(u2 *text, u2 length);
 210
 211 /* create new utf-symbol */
 212 utf *utf_new(const char *text, u2 length);
 213
 214 /* make utf symbol from u2 array */
 215 utf *utf_new_u2(u2 *unicodedata, u4 unicodelength, bool isclassname);
 216
 217 utf *utf_new_char(const char *text);
 218 utf *utf_new_char_classname(const char *text);
 219
 220 /* get number of bytes */
 221 u4 utf_bytes(utf *u);
 222
 223 /* get next unicode character of a utf-string */
 224 u2 utf_nextu2(char **utf);
 225
 226 /* get (number of) unicode characters of a utf string (safe) */
 227 s4 utf8_safe_number_of_u2s(const char *text, s4 nbytes);
 228 void utf8_safe_convert_to_u2s(const char *text, s4 nbytes, u2 *buffer);
 229
 230 /* get (number of) unicode characters of a utf string (UNSAFE!) */
 231 u4 utf_get_number_of_u2s(utf *u);
 232 u4 utf_get_number_of_u2s_for_buffer(const char *buffer, u4 blength);
 233
 234 /* determine utf length in bytes of a u2 array */
 235 u4 u2_utflength(u2 *text, u4 u2_length);
 236
 237 void utf_copy(char *buffer, utf *u);
 238 void utf_cat(char *buffer, utf *u);
 239 void utf_copy_classname(char *buffer, utf *u);
 240 void utf_cat_classname(char *buffer, utf *u);
 241
 242 /* write utf symbol to file/buffer */
 243 void utf_display_printable_ascii(utf *u);
 244 void utf_display_printable_ascii_classname(utf *u);
 245
 246 void utf_sprint_convert_to_latin1(char *buffer, utf *u);
 247 void utf_sprint_convert_to_latin1_classname(char *buffer, utf *u);
 248
 249 void utf_strcat_convert_to_latin1(char *buffer, utf *u);
 250 void utf_strcat_convert_to_latin1_classname(char *buffer, utf *u);
 251
 252 void utf_fprint_printable_ascii(FILE *file, utf *u);
 253 void utf_fprint_printable_ascii_classname(FILE *file, utf *u);
 254
 255 /* check if a UTF-8 string is valid */
 256 bool is_valid_utf(char *utf_ptr, char *end_pos);
 257
 258 /* check if a UTF-8 string may be used as a class/field/method name */
 259 bool is_valid_name(char *utf_ptr, char *end_pos);
 260 bool is_valid_name_utf(utf *u);
 261
 262 /* show utf-table */
 263 void utf_show(void);
 264
 265 #endif /* _UTF_H */
 266
 267
 268 /*
 269  * These are local overrides for various environment variables in Emacs.
 270  * Please do not remove this and leave it at the end of the file, where
 271  * Emacs will automagically detect them.
 272  * ---------------------------------------------------------------------
 273  * Local variables:
 274  * mode: c
 275  * indent-tabs-mode: t
 276  * c-basic-offset: 4
 277  * tab-width: 4
 278  * End:
 279  * vim:noexpandtab:sw=4:ts=4:
 280  */