string.cc revision c528dba35b5faece51ca658fc008b688f8b690ad
1/* 2 * Copyright (C) 2011 The Android Open Source Project 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 17#include "string.h" 18 19#include "array.h" 20#include "class-inl.h" 21#include "gc/accounting/card_table-inl.h" 22#include "intern_table.h" 23#include "object-inl.h" 24#include "runtime.h" 25#include "sirt_ref.h" 26#include "thread.h" 27#include "utf-inl.h" 28 29namespace art { 30namespace mirror { 31 32const CharArray* String::GetCharArray() const { 33 return GetFieldObject<const CharArray*>(ValueOffset(), false); 34} 35 36CharArray* String::GetCharArray() { 37 return GetFieldObject<CharArray*>(ValueOffset(), false); 38} 39 40void String::ComputeHashCode() SHARED_LOCKS_REQUIRED(Locks::mutator_lock_) { 41 SetHashCode(ComputeUtf16Hash(GetCharArray(), GetOffset(), GetLength())); 42} 43 44int32_t String::GetUtfLength() const { 45 return CountUtf8Bytes(GetCharArray()->GetData() + GetOffset(), GetLength()); 46} 47 48int32_t String::FastIndexOf(int32_t ch, int32_t start) const { 49 int32_t count = GetLength(); 50 if (start < 0) { 51 start = 0; 52 } else if (start > count) { 53 start = count; 54 } 55 const uint16_t* chars = GetCharArray()->GetData() + GetOffset(); 56 const uint16_t* p = chars + start; 57 const uint16_t* end = chars + count; 58 while (p < end) { 59 if (*p++ == ch) { 60 return (p - 1) - chars; 61 } 62 } 63 return -1; 64} 65 66void String::SetArray(CharArray* new_array) SHARED_LOCKS_REQUIRED(Locks::mutator_lock_) { 67 DCHECK(new_array != NULL); 68 SetFieldObject(OFFSET_OF_OBJECT_MEMBER(String, array_), new_array, false); 69} 70 71// TODO: get global references for these 72Class* String::java_lang_String_ = NULL; 73 74void String::SetClass(Class* java_lang_String) { 75 CHECK(java_lang_String_ == NULL); 76 CHECK(java_lang_String != NULL); 77 java_lang_String_ = java_lang_String; 78} 79 80void String::ResetClass() { 81 CHECK(java_lang_String_ != NULL); 82 java_lang_String_ = NULL; 83} 84 85String* String::Intern() { 86 return Runtime::Current()->GetInternTable()->InternWeak(this); 87} 88 89int32_t String::GetHashCode() { 90 int32_t result = GetField32(OFFSET_OF_OBJECT_MEMBER(String, hash_code_), false); 91 if (result == 0) { 92 ComputeHashCode(); 93 } 94 result = GetField32(OFFSET_OF_OBJECT_MEMBER(String, hash_code_), false); 95 DCHECK(result != 0 || ComputeUtf16Hash(GetCharArray(), GetOffset(), GetLength()) == 0) 96 << ToModifiedUtf8() << " " << result; 97 return result; 98} 99 100int32_t String::GetLength() const { 101 int32_t result = GetField32(OFFSET_OF_OBJECT_MEMBER(String, count_), false); 102 DCHECK(result >= 0 && result <= GetCharArray()->GetLength()); 103 return result; 104} 105 106uint16_t String::CharAt(int32_t index) const { 107 // TODO: do we need this? Equals is the only caller, and could 108 // bounds check itself. 109 DCHECK_GE(count_, 0); // ensures the unsigned comparison is safe. 110 if (UNLIKELY(static_cast<uint32_t>(index) >= static_cast<uint32_t>(count_))) { 111 Thread* self = Thread::Current(); 112 ThrowLocation throw_location = self->GetCurrentLocationForThrow(); 113 self->ThrowNewExceptionF(throw_location, "Ljava/lang/StringIndexOutOfBoundsException;", 114 "length=%i; index=%i", count_, index); 115 return 0; 116 } 117 return GetCharArray()->Get(index + GetOffset()); 118} 119 120String* String::AllocFromUtf16(Thread* self, 121 int32_t utf16_length, 122 const uint16_t* utf16_data_in, 123 int32_t hash_code) { 124 CHECK(utf16_data_in != NULL || utf16_length == 0); 125 String* string = Alloc(self, utf16_length); 126 if (UNLIKELY(string == nullptr)) { 127 return nullptr; 128 } 129 // TODO: use 16-bit wide memset variant 130 CharArray* array = const_cast<CharArray*>(string->GetCharArray()); 131 if (array == NULL) { 132 return NULL; 133 } 134 for (int i = 0; i < utf16_length; i++) { 135 array->Set(i, utf16_data_in[i]); 136 } 137 if (hash_code != 0) { 138 string->SetHashCode(hash_code); 139 } else { 140 string->ComputeHashCode(); 141 } 142 return string; 143} 144 145String* String::AllocFromModifiedUtf8(Thread* self, const char* utf) { 146 if (UNLIKELY(utf == nullptr)) { 147 return nullptr; 148 } 149 size_t char_count = CountModifiedUtf8Chars(utf); 150 return AllocFromModifiedUtf8(self, char_count, utf); 151} 152 153String* String::AllocFromModifiedUtf8(Thread* self, int32_t utf16_length, 154 const char* utf8_data_in) { 155 String* string = Alloc(self, utf16_length); 156 if (UNLIKELY(string == nullptr)) { 157 return nullptr; 158 } 159 uint16_t* utf16_data_out = 160 const_cast<uint16_t*>(string->GetCharArray()->GetData()); 161 ConvertModifiedUtf8ToUtf16(utf16_data_out, utf8_data_in); 162 string->ComputeHashCode(); 163 return string; 164} 165 166String* String::Alloc(Thread* self, int32_t utf16_length) { 167 SirtRef<CharArray> array(self, CharArray::Alloc(self, utf16_length)); 168 if (UNLIKELY(array.get() == nullptr)) { 169 return nullptr; 170 } 171 return Alloc(self, array); 172} 173 174String* String::Alloc(Thread* self, const SirtRef<CharArray>& array) { 175 // Hold reference in case AllocObject causes GC. 176 String* string = down_cast<String*>(GetJavaLangString()->AllocObject(self)); 177 if (LIKELY(string != nullptr)) { 178 string->SetArray(array.get()); 179 string->SetCount(array->GetLength()); 180 } 181 return string; 182} 183 184bool String::Equals(const String* that) const { 185 if (this == that) { 186 // Quick reference equality test 187 return true; 188 } else if (that == NULL) { 189 // Null isn't an instanceof anything 190 return false; 191 } else if (this->GetLength() != that->GetLength()) { 192 // Quick length inequality test 193 return false; 194 } else { 195 // Note: don't short circuit on hash code as we're presumably here as the 196 // hash code was already equal 197 for (int32_t i = 0; i < that->GetLength(); ++i) { 198 if (this->CharAt(i) != that->CharAt(i)) { 199 return false; 200 } 201 } 202 return true; 203 } 204} 205 206bool String::Equals(const uint16_t* that_chars, int32_t that_offset, int32_t that_length) const { 207 if (this->GetLength() != that_length) { 208 return false; 209 } else { 210 for (int32_t i = 0; i < that_length; ++i) { 211 if (this->CharAt(i) != that_chars[that_offset + i]) { 212 return false; 213 } 214 } 215 return true; 216 } 217} 218 219bool String::Equals(const char* modified_utf8) const { 220 for (int32_t i = 0; i < GetLength(); ++i) { 221 uint16_t ch = GetUtf16FromUtf8(&modified_utf8); 222 if (ch == '\0' || ch != CharAt(i)) { 223 return false; 224 } 225 } 226 return *modified_utf8 == '\0'; 227} 228 229bool String::Equals(const StringPiece& modified_utf8) const { 230 const char* p = modified_utf8.data(); 231 for (int32_t i = 0; i < GetLength(); ++i) { 232 uint16_t ch = GetUtf16FromUtf8(&p); 233 if (ch != CharAt(i)) { 234 return false; 235 } 236 } 237 return true; 238} 239 240// Create a modified UTF-8 encoded std::string from a java/lang/String object. 241std::string String::ToModifiedUtf8() const { 242 const uint16_t* chars = GetCharArray()->GetData() + GetOffset(); 243 size_t byte_count = GetUtfLength(); 244 std::string result(byte_count, static_cast<char>(0)); 245 ConvertUtf16ToModifiedUtf8(&result[0], chars, GetLength()); 246 return result; 247} 248 249#ifdef HAVE__MEMCMP16 250// "count" is in 16-bit units. 251extern "C" uint32_t __memcmp16(const uint16_t* s0, const uint16_t* s1, size_t count); 252#define MemCmp16 __memcmp16 253#else 254static uint32_t MemCmp16(const uint16_t* s0, const uint16_t* s1, size_t count) { 255 for (size_t i = 0; i < count; i++) { 256 if (s0[i] != s1[i]) { 257 return static_cast<int32_t>(s0[i]) - static_cast<int32_t>(s1[i]); 258 } 259 } 260 return 0; 261} 262#endif 263 264int32_t String::CompareTo(String* rhs) const { 265 // Quick test for comparison of a string with itself. 266 const String* lhs = this; 267 if (lhs == rhs) { 268 return 0; 269 } 270 // TODO: is this still true? 271 // The annoying part here is that 0x00e9 - 0xffff != 0x00ea, 272 // because the interpreter converts the characters to 32-bit integers 273 // *without* sign extension before it subtracts them (which makes some 274 // sense since "char" is unsigned). So what we get is the result of 275 // 0x000000e9 - 0x0000ffff, which is 0xffff00ea. 276 int lhsCount = lhs->GetLength(); 277 int rhsCount = rhs->GetLength(); 278 int countDiff = lhsCount - rhsCount; 279 int minCount = (countDiff < 0) ? lhsCount : rhsCount; 280 const uint16_t* lhsChars = lhs->GetCharArray()->GetData() + lhs->GetOffset(); 281 const uint16_t* rhsChars = rhs->GetCharArray()->GetData() + rhs->GetOffset(); 282 int otherRes = MemCmp16(lhsChars, rhsChars, minCount); 283 if (otherRes != 0) { 284 return otherRes; 285 } 286 return countDiff; 287} 288 289void String::VisitRoots(RootVisitor* visitor, void* arg) { 290 if (java_lang_String_ != nullptr) { 291 java_lang_String_ = down_cast<Class*>(visitor(java_lang_String_, arg)); 292 } 293} 294 295} // namespace mirror 296} // namespace art 297