Andreas Krall
Christian Thalinger
- $Id: utf8.c 2097 2005-03-27 18:59:15Z edwin $
+ $Id: utf8.c 3687 2005-11-16 19:13:37Z edwin $
*/
#include <string.h>
+#include <assert.h>
#include "mm/memory.h"
#include "vm/exceptions.h"
#include "vm/options.h"
#include "vm/statistics.h"
+#include "vm/stringlocal.h"
#include "vm/tables.h"
#include "vm/utf8.h"
+/* global variables ***********************************************************/
+
+#if defined(USE_THREADS)
+static java_objectheader *lock_utf_hashtable;
+#endif
hashtable utf_hash; /* hashtable for utf8-symbols */
utf *utf_java_lang_SecurityManager;
utf *utf_java_lang_String;
utf *utf_java_lang_System;
+utf *utf_java_lang_ThreadGroup;
utf *utf_java_io_Serializable;
utf *utf_java_lang_Throwable;
utf *utf_java_lang_VMThrowable;
-utf *utf_java_lang_Exception;
utf *utf_java_lang_Error;
+utf *utf_java_lang_NoClassDefFoundError;
+utf *utf_java_lang_NoSuchMethodError;
utf *utf_java_lang_OutOfMemoryError;
+utf *utf_java_lang_Exception;
+utf *utf_java_lang_ClassNotFoundException;
+utf *utf_java_lang_IllegalArgumentException;
+utf *utf_java_lang_IllegalMonitorStateException;
+
+utf *utf_java_lang_NullPointerException;
+
utf* utf_java_lang_Void;
utf* utf_java_lang_Boolean;
utf* utf_java_lang_Byte;
utf* utf_java_lang_Float;
utf* utf_java_lang_Double;
+utf *utf_java_lang_StackTraceElement;
+utf *utf_java_lang_reflect_Constructor;
+utf *utf_java_lang_reflect_Field;
+utf *utf_java_lang_reflect_Method;
utf *utf_java_util_Vector;
utf *utf_InnerClasses; /* InnerClasses */
utf *utf_init; /* <init> */
utf *utf_clinit; /* <clinit> */
+utf *utf_clone; /* clone */
utf *utf_finalize; /* finalize */
+utf *utf_run; /* run */
+
+utf *utf_add; /* add */
+utf *utf_remove; /* remove */
+utf *utf_put; /* put */
+utf *utf_get; /* get */
+utf *utf_value; /* value */
-utf *utf_printStackTrace;
utf *utf_fillInStackTrace;
+utf *utf_getSystemClassLoader;
utf *utf_loadClass;
+utf *utf_printStackTrace;
+
+utf *utf_Z; /* Z */
+utf *utf_B; /* B */
+utf *utf_C; /* C */
+utf *utf_S; /* S */
+utf *utf_I; /* I */
+utf *utf_J; /* J */
+utf *utf_F; /* F */
+utf *utf_D; /* D */
utf *utf_void__void; /* ()V */
utf *utf_boolean__void; /* (Z)V */
utf *utf_long__void; /* (J)V */
utf *utf_float__void; /* (F)V */
utf *utf_double__void; /* (D)V */
+
+utf *utf_void__java_lang_ClassLoader; /* ()Ljava/lang/ClassLoader; */
+utf *utf_void__java_lang_Object; /* ()Ljava/lang/Object; */
utf *utf_void__java_lang_Throwable; /* ()Ljava/lang/Throwable; */
+utf *utf_java_lang_Object__java_lang_Object;
utf *utf_java_lang_String__void; /* (Ljava/lang/String;)V */
utf *utf_java_lang_String__java_lang_Class;
utf *utf_java_lang_Throwable__void; /* (Ljava/lang/Throwable;)V */
+utf *utf_not_named_yet; /* special name for unnamed classes */
+
utf *array_packagename;
*******************************************************************************/
-void utf8_init(void)
+bool utf8_init(void)
{
+#if defined(USE_THREADS)
+ /* create utf hashtable lock object */
+
+ lock_utf_hashtable = NEW(java_objectheader);
+
+# if defined(NATIVE_THREADS)
+ initObjectLock(lock_utf_hashtable);
+# endif
+#endif
+
/* create utf-symbols for pointer comparison of frequently used strings */
utf_java_lang_Object = utf_new_char("java/lang/Object");
utf_java_lang_SecurityManager = utf_new_char("java/lang/SecurityManager");
utf_java_lang_String = utf_new_char("java/lang/String");
utf_java_lang_System = utf_new_char("java/lang/System");
+ utf_java_lang_ThreadGroup = utf_new_char("java/lang/ThreadGroup");
utf_java_io_Serializable = utf_new_char("java/io/Serializable");
- utf_java_lang_Throwable = utf_new_char("java/lang/Throwable");
- utf_java_lang_VMThrowable = utf_new_char("java/lang/VMThrowable");
- utf_java_lang_Exception = utf_new_char("java/lang/Exception");
- utf_java_lang_Error = utf_new_char("java/lang/Error");
- utf_java_lang_OutOfMemoryError = utf_new_char("java/lang/OutOfMemoryError");
+ utf_java_lang_Throwable = utf_new_char(string_java_lang_Throwable);
+ utf_java_lang_VMThrowable = utf_new_char(string_java_lang_VMThrowable);
+ utf_java_lang_Error = utf_new_char(string_java_lang_Error);
+
+ utf_java_lang_NoClassDefFoundError =
+ utf_new_char(string_java_lang_NoClassDefFoundError);
+
+ utf_java_lang_NoSuchMethodError =
+ utf_new_char(string_java_lang_NoSuchMethodError);
+
+ utf_java_lang_OutOfMemoryError =
+ utf_new_char(string_java_lang_OutOfMemoryError);
+
+ utf_java_lang_Exception = utf_new_char(string_java_lang_Exception);
+
+ utf_java_lang_ClassNotFoundException =
+ utf_new_char(string_java_lang_ClassNotFoundException);
+
+ utf_java_lang_IllegalArgumentException =
+ utf_new_char(string_java_lang_IllegalArgumentException);
+
+ utf_java_lang_IllegalMonitorStateException =
+ utf_new_char(string_java_lang_IllegalMonitorStateException);
+
+ utf_java_lang_NullPointerException =
+ utf_new_char(string_java_lang_NullPointerException);
utf_java_lang_Void = utf_new_char("java/lang/Void");
utf_java_lang_Boolean = utf_new_char("java/lang/Boolean");
utf_java_lang_Float = utf_new_char("java/lang/Float");
utf_java_lang_Double = utf_new_char("java/lang/Double");
+ utf_java_lang_StackTraceElement =
+ utf_new_char("java/lang/StackTraceElement");
+
+ utf_java_lang_reflect_Constructor =
+ utf_new_char("java/lang/reflect/Constructor");
+
+ utf_java_lang_reflect_Field = utf_new_char("java/lang/reflect/Field");
+ utf_java_lang_reflect_Method = utf_new_char("java/lang/reflect/Method");
utf_java_util_Vector = utf_new_char("java/util/Vector");
utf_InnerClasses = utf_new_char("InnerClasses");
utf_init = utf_new_char("<init>");
utf_clinit = utf_new_char("<clinit>");
+ utf_clone = utf_new_char("clone");
utf_finalize = utf_new_char("finalize");
+ utf_run = utf_new_char("run");
+
+ utf_add = utf_new_char("add");
+ utf_remove = utf_new_char("remove");
+ utf_put = utf_new_char("put");
+ utf_get = utf_new_char("get");
+ utf_value = utf_new_char("value");
utf_printStackTrace = utf_new_char("printStackTrace");
utf_fillInStackTrace = utf_new_char("fillInStackTrace");
utf_loadClass = utf_new_char("loadClass");
+ utf_getSystemClassLoader = utf_new_char("getSystemClassLoader");
+
+ utf_Z = utf_new_char("Z");
+ utf_B = utf_new_char("B");
+ utf_C = utf_new_char("C");
+ utf_S = utf_new_char("S");
+ utf_I = utf_new_char("I");
+ utf_J = utf_new_char("J");
+ utf_F = utf_new_char("F");
+ utf_D = utf_new_char("D");
utf_void__void = utf_new_char("()V");
utf_boolean__void = utf_new_char("(Z)V");
utf_long__void = utf_new_char("(J)V");
utf_float__void = utf_new_char("(F)V");
utf_double__void = utf_new_char("(D)V");
+ utf_void__java_lang_Object = utf_new_char("()Ljava/lang/Object;");
utf_void__java_lang_Throwable = utf_new_char("()Ljava/lang/Throwable;");
+
+ utf_void__java_lang_ClassLoader =
+ utf_new_char("()Ljava/lang/ClassLoader;");
+
+ utf_java_lang_Object__java_lang_Object =
+ utf_new_char("(Ljava/lang/Object;)Ljava/lang/Object;");
+
utf_java_lang_String__void = utf_new_char("(Ljava/lang/String;)V");
utf_java_lang_String__java_lang_Class =
utf_new_char("(Ljava/lang/String;)Ljava/lang/Class;");
utf_java_lang_Throwable__void = utf_new_char("(Ljava/lang/Throwable;)V");
+
+ utf_not_named_yet = utf_new_char("\t<not_named_yet>");
+
array_packagename = utf_new_char("\t<the array package>");
+
+ /* everything's ok */
+
+ return true;
}
*******************************************************************************/
-utf *utf_new_intern(const char *text, u2 length);
-
utf *utf_new(const char *text, u2 length)
-{
- utf *r;
-
-#if defined(USE_THREADS) && defined(NATIVE_THREADS)
- tables_lock();
-#endif
-
- r = utf_new_intern(text, length);
-
-#if defined(USE_THREADS) && defined(NATIVE_THREADS)
- tables_unlock();
-#endif
-
- return r;
-}
-
-
-utf *utf_new_intern(const char *text, u2 length)
{
u4 key; /* hashkey computed from utf-text */
u4 slot; /* slot in hashtable */
utf *u; /* hashtable element */
u2 i;
+ /* XXX REMOVE ME! after testing of course ;-) */
+ static int running = 0;
+ /* XXX REMOVE ME! */
+
+#if defined(USE_THREADS)
+ builtin_monitorenter(lock_utf_hashtable);
+#endif
+
+ /* XXX REMOVE ME! after testing of course ;-) */
+ assert(running == 0);
+ running = 1;
+ /* XXX REMOVE ME! */
+
#ifdef STATISTICS
if (opt_stat)
count_utf_new++;
u = utf_hash.ptr[slot];
/* search external hash chain for utf-symbol */
+
while (u) {
if (u->blength == length) {
-
/* compare text of hashtable elements */
+
for (i = 0; i < length; i++)
- if (text[i] != u->text[i]) goto nomatch;
+ if (text[i] != u->text[i])
+ goto nomatch;
-#ifdef STATISTICS
+#if defined(STATISTICS)
if (opt_stat)
count_utf_new_found++;
#endif
/* symbol found in hashtable */
+
+ /* XXX REMOVE ME! */
+ running = 0;
+ /* XXX REMOVE ME! */
+
+#if defined(USE_THREADS)
+ builtin_monitorexit(lock_utf_hashtable);
+#endif
+
return u;
}
+
nomatch:
u = u->hashlink; /* next element in external chain */
}
-#ifdef STATISTICS
+#if defined(STATISTICS)
if (opt_stat)
- count_utf_len += sizeof(utf) + length;
+ count_utf_len += sizeof(utf) + length + 1;
#endif
/* location in hashtable found, create new utf element */
utf_hash = newhash;
}
+ /* XXX REMOVE ME! */
+ running = 0;
+ /* XXX REMOVE ME! */
+
+#if defined(USE_THREADS)
+ builtin_monitorexit(lock_utf_hashtable);
+#endif
+
return u;
}
return 0;
}
- endpos = utf_end(u);
+ endpos = UTF_END(u);
utf_ptr = u->text;
while (utf_ptr < endpos) {
return;
}
- endpos = utf_end(u);
+ endpos = UTF_END(u);
utf_ptr = u->text;
while (utf_ptr < endpos) {
return;
}
- endpos = utf_end(u);
+ endpos = UTF_END(u);
utf_ptr = u->text;
while (utf_ptr < endpos) {
u2 pos = 0; /* position in c-string */
if (!u) {
- memcpy(buffer, "NULL", 5); /* 4 chars + terminating \0 */
+ strcpy(buffer, "NULL");
return;
}
- endpos = utf_end(u);
+ endpos = UTF_END(u);
utf_ptr = u->text;
while (utf_ptr < endpos)
u2 pos = 0; /* position in c-string */
if (!u) {
- memcpy(buffer, "NULL", 5); /* 4 chars + terminating \0 */
+ strcpy(buffer, "NULL");
return;
}
- endpos = utf_end(u);
+ endpos = UTF_END(u);
utf_ptr = u->text;
while (utf_ptr < endpos) {
}
+/* utf_strcat ******************************************************************
+
+ Like libc strcat, but uses an utf8 string.
+
+*******************************************************************************/
+
+void utf_strcat(char *buffer, utf *u)
+{
+ utf_sprint(buffer + strlen(buffer), u);
+}
+
+
+/* utf_strcat_classname ********************************************************
+
+ Like libc strcat, but uses an utf8 string.
+
+*******************************************************************************/
+
+void utf_strcat_classname(char *buffer, utf *u)
+{
+ utf_sprint_classname(buffer + strlen(buffer), u);
+}
+
+
/* utf_fprint ******************************************************************
Write utf symbol into file.
if (!u)
return;
- endpos = utf_end(u);
+ endpos = UTF_END(u);
utf_ptr = u->text;
while (utf_ptr < endpos) {
if (!u)
return;
- endpos = utf_end(u);
+ endpos = UTF_END(u);
utf_ptr = u->text;
while (utf_ptr < endpos) {
*******************************************************************************/
-static unsigned long min_codepoint[6] = {0,1L<<7,1L<<11,1L<<16,1L<<21,1L<<26};
+/* static unsigned long min_codepoint[6] = {0,1L<<7,1L<<11,1L<<16,1L<<21,1L<<26}; */
bool is_valid_utf(char *utf_ptr, char *end_pos)
{
} else {
/* Sun Java seems to allow overlong UTF-8 encodings */
- if (v < min_codepoint[len]) { /* overlong UTF-8 */
- if (!opt_liberalutf)
- fprintf(stderr,"WARNING: Overlong UTF-8 sequence found.\n");
- /* XXX change this to panic? */
- }
+ /* if (v < min_codepoint[len]) */
+ /* XXX throw exception? */
}
/* surrogates in UTF-8 seem to be allowed in Java classfiles */
bool is_valid_name_utf(utf *u)
{
- return is_valid_name(u->text,utf_end(u));
+ return is_valid_name(u->text, UTF_END(u));
}
projects / cacao.git / blobdiff