src/vm/utf8.h

   1 /* src/vm/utf8.h - utf8 string functions
   2
   3    Copyright (C) 1996-2005, 2006, 2007, 2008
   4    CACAOVM - Verein zur Foerderung der freien virtuellen Maschine CACAO
   5
   6    This file is part of CACAO.
   7
   8    This program is free software; you can redistribute it and/or
   9    modify it under the terms of the GNU General Public License as
  10    published by the Free Software Foundation; either version 2, or (at
  11    your option) any later version.
  12
  13    This program is distributed in the hope that it will be useful, but
  14    WITHOUT ANY WARRANTY; without even the implied warranty of
  15    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16    General Public License for more details.
  17
  18    You should have received a copy of the GNU General Public License
  19    along with this program; if not, write to the Free Software
  20    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
  21    02110-1301, USA.
  22
  23 */
  24
  25
  26 #ifndef _UTF_H
  27 #define _UTF_H
  28
  29 /* forward typedefs ***********************************************************/
  30
  31 typedef struct utf utf;
  32
  33 #include "config.h"
  34
  35 #include <stdio.h>
  36
  37 #include "vm/types.h"
  38
  39 #include "vm/global.h"
  40
  41
  42 /* data structure for utf8 symbols ********************************************/
  43
  44 struct utf {
  45         utf  *hashlink;                     /* link for external hash chain       */
  46         s4    blength;                      /* text length in bytes               */
  47         char *text;                         /* pointer to text                    */
  48 };
  49
  50 /* to determine the end of utf strings */
  51
  52 #define UTF_END(u)    ((char *) u->text + u->blength)
  53
  54
  55 /* utf-symbols for pointer comparison of frequently used strings **************/
  56
  57 extern utf *utf_java_lang_Object;
  58
  59 extern utf *utf_java_lang_Class;
  60 extern utf *utf_java_lang_ClassLoader;
  61 extern utf *utf_java_lang_ClassLoader_NativeLibrary;
  62 extern utf *utf_java_lang_Cloneable;
  63 extern utf *utf_java_lang_SecurityManager;
  64 extern utf *utf_java_lang_String;
  65 extern utf *utf_java_lang_ThreadGroup;
  66 extern utf *utf_java_lang_ref_SoftReference;
  67 extern utf *utf_java_lang_ref_WeakReference;
  68 extern utf *utf_java_lang_ref_PhantomReference;
  69 extern utf *utf_java_io_Serializable;
  70
  71 extern utf *utf_java_lang_Throwable;
  72 extern utf *utf_java_lang_Error;
  73
  74 extern utf *utf_java_lang_AbstractMethodError;
  75 extern utf *utf_java_lang_ClassCircularityError;
  76 extern utf *utf_java_lang_ClassFormatError;
  77 extern utf *utf_java_lang_ExceptionInInitializerError;
  78 extern utf *utf_java_lang_IncompatibleClassChangeError;
  79 extern utf *utf_java_lang_InstantiationError;
  80 extern utf *utf_java_lang_InternalError;
  81 extern utf *utf_java_lang_LinkageError;
  82 extern utf *utf_java_lang_NoClassDefFoundError;
  83 extern utf *utf_java_lang_NoSuchFieldError;
  84 extern utf *utf_java_lang_NoSuchMethodError;
  85 extern utf *utf_java_lang_OutOfMemoryError;
  86 extern utf *utf_java_lang_UnsatisfiedLinkError;
  87 extern utf *utf_java_lang_UnsupportedClassVersionError;
  88 extern utf *utf_java_lang_VerifyError;
  89 extern utf *utf_java_lang_VirtualMachineError;
  90
  91 extern utf *utf_java_lang_Exception;
  92
  93 extern utf *utf_java_lang_ArithmeticException;
  94 extern utf *utf_java_lang_ArrayIndexOutOfBoundsException;
  95 extern utf *utf_java_lang_ArrayStoreException;
  96 extern utf *utf_java_lang_ClassCastException;
  97 extern utf *utf_java_lang_ClassNotFoundException;
  98 extern utf *utf_java_lang_CloneNotSupportedException;
  99 extern utf *utf_java_lang_IllegalAccessException;
 100 extern utf *utf_java_lang_IllegalArgumentException;
 101 extern utf *utf_java_lang_IllegalMonitorStateException;
 102 extern utf *utf_java_lang_InstantiationException;
 103 extern utf *utf_java_lang_InterruptedException;
 104 extern utf *utf_java_lang_NegativeArraySizeException;
 105 extern utf *utf_java_lang_NullPointerException;
 106 extern utf *utf_java_lang_RuntimeException;
 107 extern utf *utf_java_lang_StringIndexOutOfBoundsException;
 108
 109 extern utf *utf_java_lang_reflect_InvocationTargetException;
 110
 111 extern utf *utf_java_security_PrivilegedActionException;
 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
 132 # if defined(WITH_JAVA_RUNTIME_LIBRARY_GNU_CLASSPATH)
 133 extern utf *utf_java_lang_reflect_VMConstructor;
 134 extern utf *utf_java_lang_reflect_VMField;
 135 extern utf *utf_java_lang_reflect_VMMethod;
 136 # endif
 137
 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_JVMTI)
 154 extern utf *utf_LocalVariableTable;
 155 # endif
 156
 157 # if defined(ENABLE_ANNOTATIONS)
 158 extern utf *utf_RuntimeVisibleAnnotations;
 159 extern utf *utf_RuntimeInvisibleAnnotations;
 160 extern utf *utf_RuntimeVisibleParameterAnnotations;
 161 extern utf *utf_RuntimeInvisibleParameterAnnotations;
 162 extern utf *utf_AnnotationDefault;
 163 # endif
 164 #endif
 165
 166 extern utf *utf_init;
 167 extern utf *utf_clinit;
 168 extern utf *utf_clone;
 169 extern utf *utf_finalize;
 170 extern utf *utf_invoke;
 171 extern utf *utf_main;
 172 extern utf *utf_run;
 173
 174 extern utf *utf_add;
 175 extern utf *utf_dispatch;
 176 extern utf *utf_remove;
 177 extern utf *utf_addThread;
 178 extern utf *utf_removeThread;
 179 extern utf *utf_put;
 180 extern utf *utf_get;
 181 extern utf *utf_uncaughtException;
 182 extern utf *utf_value;
 183
 184 extern utf *utf_fillInStackTrace;
 185 extern utf *utf_findNative;
 186 extern utf *utf_getSystemClassLoader;
 187 extern utf *utf_initCause;
 188 extern utf *utf_loadClass;
 189 extern utf *utf_loadClassInternal;
 190 extern utf *utf_printStackTrace;
 191
 192 extern utf *utf_division_by_zero;
 193
 194 extern utf *utf_Z;
 195 extern utf *utf_B;
 196 extern utf *utf_C;
 197 extern utf *utf_S;
 198 extern utf *utf_I;
 199 extern utf *utf_J;
 200 extern utf *utf_F;
 201 extern utf *utf_D;
 202
 203 extern utf *utf_void__void;
 204 extern utf *utf_boolean__void;
 205 extern utf *utf_byte__void;
 206 extern utf *utf_char__void;
 207 extern utf *utf_short__void;
 208 extern utf *utf_int__void;
 209 extern utf *utf_long__void;
 210 extern utf *utf_float__void;
 211 extern utf *utf_double__void;
 212
 213 extern utf *utf_void__java_lang_ClassLoader;
 214 extern utf *utf_void__java_lang_Object;
 215 extern utf *utf_void__java_lang_Throwable;
 216 extern utf *utf_java_lang_ClassLoader_java_lang_String__J;
 217 extern utf *utf_java_lang_Exception__V;
 218 extern utf *utf_java_lang_Object__java_lang_Object;
 219 extern utf *utf_java_lang_String__void;
 220 extern utf *utf_java_lang_String__java_lang_Class;
 221 extern utf *utf_java_lang_Thread__V;
 222 extern utf *utf_java_lang_Thread_java_lang_Throwable__V;
 223 extern utf *utf_Ljava_lang_ThreadGroup_Ljava_lang_String__V;
 224 extern utf *utf_java_lang_Throwable__void;
 225 extern utf *utf_java_lang_Throwable__java_lang_Throwable;
 226
 227 extern utf *utf_not_named_yet;
 228 extern utf *utf_null;
 229 extern utf *array_packagename;
 230
 231
 232 /* function prototypes ********************************************************/
 233
 234 #ifdef __cplusplus
 235 extern "C" {
 236 #endif
 237
 238 /* initialize the utf8 subsystem */
 239 void utf8_init(void);
 240
 241 u4 utf_hashkey(const char *text, u4 length);
 242 u4 utf_full_hashkey(const char *text, u4 length);
 243
 244 /* determine hashkey of a unicode-symbol */
 245 u4 unicode_hashkey(u2 *text, u2 length);
 246
 247 /* create new utf-symbol */
 248 utf *utf_new(const char *text, u2 length);
 249
 250 /* make utf symbol from u2 array */
 251 utf *utf_new_u2(u2 *unicodedata, u4 unicodelength, bool isclassname);
 252
 253 utf *utf_new_char(const char *text);
 254 utf *utf_new_char_classname(const char *text);
 255
 256 /* get number of bytes */
 257 u4 utf_bytes(utf *u);
 258
 259 /* get next unicode character of a utf-string */
 260 u2 utf_nextu2(char **utf);
 261
 262 /* get (number of) unicode characters of a utf string (safe) */
 263 s4 utf8_safe_number_of_u2s(const char *text, s4 nbytes);
 264 void utf8_safe_convert_to_u2s(const char *text, s4 nbytes, u2 *buffer);
 265
 266 /* get (number of) unicode characters of a utf string (UNSAFE!) */
 267 u4 utf_get_number_of_u2s(utf *u);
 268 u4 utf_get_number_of_u2s_for_buffer(const char *buffer, u4 blength);
 269
 270 /* determine utf length in bytes of a u2 array */
 271 u4 u2_utflength(u2 *text, u4 u2_length);
 272
 273 void utf_copy(char *buffer, utf *u);
 274 void utf_cat(char *buffer, utf *u);
 275 void utf_copy_classname(char *buffer, utf *u);
 276 void utf_cat_classname(char *buffer, utf *u);
 277
 278 /* write utf symbol to file/buffer */
 279 void utf_display_printable_ascii(utf *u);
 280 void utf_display_printable_ascii_classname(utf *u);
 281
 282 void utf_sprint_convert_to_latin1(char *buffer, utf *u);
 283 void utf_sprint_convert_to_latin1_classname(char *buffer, utf *u);
 284
 285 void utf_strcat_convert_to_latin1(char *buffer, utf *u);
 286 void utf_strcat_convert_to_latin1_classname(char *buffer, utf *u);
 287
 288 void utf_fprint_printable_ascii(FILE *file, utf *u);
 289 void utf_fprint_printable_ascii_classname(FILE *file, utf *u);
 290
 291 /* check if a UTF-8 string is valid */
 292 bool is_valid_utf(char *utf_ptr, char *end_pos);
 293
 294 /* check if a UTF-8 string may be used as a class/field/method name */
 295 bool is_valid_name(char *utf_ptr, char *end_pos);
 296 bool is_valid_name_utf(utf *u);
 297
 298 /* show utf-table */
 299 void utf_show(void);
 300
 301 #ifdef __cplusplus
 302 }
 303 #endif
 304
 305 #endif /* _UTF_H */
 306
 307
 308 /*
 309  * These are local overrides for various environment variables in Emacs.
 310  * Please do not remove this and leave it at the end of the file, where
 311  * Emacs will automagically detect them.
 312  * ---------------------------------------------------------------------
 313  * Local variables:
 314  * mode: c
 315  * indent-tabs-mode: t
 316  * c-basic-offset: 4
 317  * tab-width: 4
 318  * End:
 319  * vim:noexpandtab:sw=4:ts=4:
 320  */