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 8367 2007-08-20 20:26:16Z 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_lang_ref_SoftReference;
  71 extern utf *utf_java_lang_ref_WeakReference;
  72 extern utf *utf_java_lang_ref_PhantomReference;
  73 extern utf *utf_java_io_Serializable;
  74
  75 extern utf *utf_java_lang_Throwable;
  76 extern utf *utf_java_lang_Error;
  77
  78 extern utf *utf_java_lang_AbstractMethodError;
  79 extern utf *utf_java_lang_ClassCircularityError;
  80 extern utf *utf_java_lang_ClassFormatError;
  81 extern utf *utf_java_lang_ExceptionInInitializerError;
  82 extern utf *utf_java_lang_IncompatibleClassChangeError;
  83 extern utf *utf_java_lang_InstantiationError;
  84 extern utf *utf_java_lang_InternalError;
  85 extern utf *utf_java_lang_LinkageError;
  86 extern utf *utf_java_lang_NoClassDefFoundError;
  87 extern utf *utf_java_lang_NoSuchFieldError;
  88 extern utf *utf_java_lang_NoSuchMethodError;
  89 extern utf *utf_java_lang_OutOfMemoryError;
  90 extern utf *utf_java_lang_UnsatisfiedLinkError;
  91 extern utf *utf_java_lang_UnsupportedClassVersionError;
  92 extern utf *utf_java_lang_VerifyError;
  93 extern utf *utf_java_lang_VirtualMachineError;
  94
  95 #if defined(WITH_CLASSPATH_GNU)
  96 extern utf *utf_java_lang_VMThrowable;
  97 #endif
  98
  99 extern utf *utf_java_lang_Exception;
 100
 101 extern utf *utf_java_lang_ArithmeticException;
 102 extern utf *utf_java_lang_ArrayIndexOutOfBoundsException;
 103 extern utf *utf_java_lang_ArrayStoreException;
 104 extern utf *utf_java_lang_ClassCastException;
 105 extern utf *utf_java_lang_ClassNotFoundException;
 106 extern utf *utf_java_lang_CloneNotSupportedException;
 107 extern utf *utf_java_lang_IllegalAccessException;
 108 extern utf *utf_java_lang_IllegalArgumentException;
 109 extern utf *utf_java_lang_IllegalMonitorStateException;
 110 extern utf *utf_java_lang_InstantiationException;
 111 extern utf *utf_java_lang_InterruptedException;
 112 extern utf *utf_java_lang_NegativeArraySizeException;
 113 extern utf *utf_java_lang_NullPointerException;
 114 extern utf *utf_java_lang_StringIndexOutOfBoundsException;
 115
 116 extern utf *utf_java_lang_reflect_InvocationTargetException;
 117
 118 extern utf *utf_java_security_PrivilegedActionException;
 119
 120 #if defined(ENABLE_JAVASE)
 121 extern utf* utf_java_lang_Void;
 122 #endif
 123
 124 extern utf* utf_java_lang_Boolean;
 125 extern utf* utf_java_lang_Byte;
 126 extern utf* utf_java_lang_Character;
 127 extern utf* utf_java_lang_Short;
 128 extern utf* utf_java_lang_Integer;
 129 extern utf* utf_java_lang_Long;
 130 extern utf* utf_java_lang_Float;
 131 extern utf* utf_java_lang_Double;
 132
 133 #if defined(ENABLE_JAVASE)
 134 extern utf *utf_java_lang_StackTraceElement;
 135 extern utf *utf_java_lang_reflect_Constructor;
 136 extern utf *utf_java_lang_reflect_Field;
 137 extern utf *utf_java_lang_reflect_Method;
 138 extern utf *utf_java_util_Vector;
 139 #endif
 140
 141 extern utf *utf_InnerClasses;
 142 extern utf *utf_ConstantValue;
 143 extern utf *utf_Code;
 144 extern utf *utf_Exceptions;
 145 extern utf *utf_LineNumberTable;
 146 extern utf *utf_SourceFile;
 147
 148 #if defined(ENABLE_JAVASE)
 149 extern utf *utf_EnclosingMethod;
 150 extern utf *utf_Signature;
 151 extern utf *utf_StackMapTable;
 152
 153 #if defined(ENABLE_ANNOTATIONS)
 154 /* OpenJDKs sun.reflect.ConstantPool class is for now only
 155  * used by annotations but will probably be used for other
 156  * things in the future. For now I just couple it with
 157  * annotation support.
 158  */
 159 extern utf *utf_sun_reflect_ConstantPool;
 160 #if defined(WITH_CLASSPATH_GNU)
 161 extern utf *utf_sun_reflect_annotation_AnnotationParser;
 162 #endif
 163
 164 extern utf *utf_RuntimeVisibleAnnotations;
 165 extern utf *utf_RuntimeInvisibleAnnotations;
 166 extern utf *utf_RuntimeVisibleParameterAnnotations;
 167 extern utf *utf_RuntimeInvisibleParameterAnnotations;
 168 extern utf *utf_AnnotationDefault;
 169 #endif
 170 #endif
 171
 172 extern utf *utf_init;
 173 extern utf *utf_clinit;
 174 extern utf *utf_clone;
 175 extern utf *utf_finalize;
 176 extern utf *utf_run;
 177
 178 extern utf *utf_add;
 179 extern utf *utf_remove;
 180 extern utf *utf_addThread;
 181 extern utf *utf_removeThread;
 182 extern utf *utf_put;
 183 extern utf *utf_get;
 184 extern utf *utf_uncaughtException;
 185 extern utf *utf_value;
 186
 187 extern utf *utf_fillInStackTrace;
 188 extern utf *utf_findNative;
 189 extern utf *utf_getSystemClassLoader;
 190 extern utf *utf_initCause;
 191 extern utf *utf_loadClass;
 192 extern utf *utf_printStackTrace;
 193
 194 extern utf *utf_division_by_zero;
 195
 196 extern utf *utf_Z;
 197 extern utf *utf_B;
 198 extern utf *utf_C;
 199 extern utf *utf_S;
 200 extern utf *utf_I;
 201 extern utf *utf_J;
 202 extern utf *utf_F;
 203 extern utf *utf_D;
 204
 205 extern utf *utf_void__void;
 206 extern utf *utf_boolean__void;
 207 extern utf *utf_byte__void;
 208 extern utf *utf_char__void;
 209 extern utf *utf_short__void;
 210 extern utf *utf_int__void;
 211 extern utf *utf_long__void;
 212 extern utf *utf_float__void;
 213 extern utf *utf_double__void;
 214
 215 extern utf *utf_void__java_lang_ClassLoader;
 216 extern utf *utf_void__java_lang_Object;
 217 extern utf *utf_void__java_lang_Throwable;
 218 extern utf *utf_java_lang_ClassLoader_java_lang_String__J;
 219 extern utf *utf_java_lang_Exception__V;
 220 extern utf *utf_java_lang_Object__java_lang_Object;
 221 extern utf *utf_java_lang_String__void;
 222 extern utf *utf_java_lang_String__java_lang_Class;
 223 extern utf *utf_java_lang_Thread__V;
 224 extern utf *utf_java_lang_Thread_java_lang_Throwable__V;
 225 extern utf *utf_java_lang_Throwable__void;
 226 extern utf *utf_java_lang_Throwable__java_lang_Throwable;
 227
 228 extern utf *utf_not_named_yet;
 229 extern utf *utf_null;
 230 extern utf *array_packagename;
 231
 232
 233 /* function prototypes ********************************************************/
 234
 235 /* initialize the utf8 subsystem */
 236 bool utf8_init(void);
 237
 238 u4 utf_hashkey(const char *text, u4 length);
 239 u4 utf_full_hashkey(const char *text, u4 length);
 240
 241 /* determine hashkey of a unicode-symbol */
 242 u4 unicode_hashkey(u2 *text, u2 length);
 243
 244 /* create new utf-symbol */
 245 utf *utf_new(const char *text, u2 length);
 246
 247 /* make utf symbol from u2 array */
 248 utf *utf_new_u2(u2 *unicodedata, u4 unicodelength, bool isclassname);
 249
 250 utf *utf_new_char(const char *text);
 251 utf *utf_new_char_classname(const char *text);
 252
 253 /* get number of bytes */
 254 u4 utf_bytes(utf *u);
 255
 256 /* get next unicode character of a utf-string */
 257 u2 utf_nextu2(char **utf);
 258
 259 /* get (number of) unicode characters of a utf string (safe) */
 260 s4 utf8_safe_number_of_u2s(const char *text, s4 nbytes);
 261 void utf8_safe_convert_to_u2s(const char *text, s4 nbytes, u2 *buffer);
 262
 263 /* get (number of) unicode characters of a utf string (UNSAFE!) */
 264 u4 utf_get_number_of_u2s(utf *u);
 265 u4 utf_get_number_of_u2s_for_buffer(const char *buffer, u4 blength);
 266
 267 /* determine utf length in bytes of a u2 array */
 268 u4 u2_utflength(u2 *text, u4 u2_length);
 269
 270 void utf_copy(char *buffer, utf *u);
 271 void utf_cat(char *buffer, utf *u);
 272 void utf_copy_classname(char *buffer, utf *u);
 273 void utf_cat_classname(char *buffer, utf *u);
 274
 275 /* write utf symbol to file/buffer */
 276 void utf_display_printable_ascii(utf *u);
 277 void utf_display_printable_ascii_classname(utf *u);
 278
 279 void utf_sprint_convert_to_latin1(char *buffer, utf *u);
 280 void utf_sprint_convert_to_latin1_classname(char *buffer, utf *u);
 281
 282 void utf_strcat_convert_to_latin1(char *buffer, utf *u);
 283 void utf_strcat_convert_to_latin1_classname(char *buffer, utf *u);
 284
 285 void utf_fprint_printable_ascii(FILE *file, utf *u);
 286 void utf_fprint_printable_ascii_classname(FILE *file, utf *u);
 287
 288 /* check if a UTF-8 string is valid */
 289 bool is_valid_utf(char *utf_ptr, char *end_pos);
 290
 291 /* check if a UTF-8 string may be used as a class/field/method name */
 292 bool is_valid_name(char *utf_ptr, char *end_pos);
 293 bool is_valid_name_utf(utf *u);
 294
 295 /* show utf-table */
 296 void utf_show(void);
 297
 298 #endif /* _UTF_H */
 299
 300
 301 /*
 302  * These are local overrides for various environment variables in Emacs.
 303  * Please do not remove this and leave it at the end of the file, where
 304  * Emacs will automagically detect them.
 305  * ---------------------------------------------------------------------
 306  * Local variables:
 307  * mode: c
 308  * indent-tabs-mode: t
 309  * c-basic-offset: 4
 310  * tab-width: 4
 311  * End:
 312  * vim:noexpandtab:sw=4:ts=4:
 313  */