|
| 1 | +/* |
| 2 | + * Copyright (C) 2019-2022 The Kraken authors. All rights reserved. |
| 3 | + * Copyright (C) 2022-present The WebF authors. All rights reserved. |
| 4 | + */ |
| 5 | + |
| 6 | +#include "atomic_string.h" |
| 7 | +#include <algorithm> |
| 8 | +#include <vector> |
| 9 | +#include "built_in_string.h" |
| 10 | +#include "foundation/native_string.h" |
| 11 | + |
| 12 | +namespace webf { |
| 13 | + |
| 14 | +AtomicString AtomicString::Empty() { |
| 15 | + return built_in_string::kempty_string; |
| 16 | +} |
| 17 | + |
| 18 | +AtomicString AtomicString::Null() { |
| 19 | + return built_in_string::kNULL; |
| 20 | +} |
| 21 | + |
| 22 | +namespace { |
| 23 | + |
| 24 | +AtomicString::StringKind GetStringKind(const std::string& string, size_t length) { |
| 25 | + char first_char = string[0]; |
| 26 | + |
| 27 | + if (first_char < 0 || first_char > 255) { |
| 28 | + return AtomicString::StringKind::kUnknown; |
| 29 | + } |
| 30 | + |
| 31 | + AtomicString::StringKind predictKind = |
| 32 | + std::islower(string[0]) ? AtomicString::StringKind::kIsLowerCase : AtomicString::StringKind::kIsUpperCase; |
| 33 | + for (int i = 0; i < length; i++) { |
| 34 | + char c = string[i]; |
| 35 | + |
| 36 | + if (c < 0 || c > 255) { |
| 37 | + return AtomicString::StringKind::kUnknown; |
| 38 | + } |
| 39 | + |
| 40 | + if (predictKind == AtomicString::StringKind::kIsUpperCase && !std::isupper(c)) { |
| 41 | + return AtomicString::StringKind::kIsMixed; |
| 42 | + } else if (predictKind == AtomicString::StringKind::kIsLowerCase && !std::islower(c)) { |
| 43 | + return AtomicString::StringKind::kIsMixed; |
| 44 | + } |
| 45 | + } |
| 46 | + return predictKind; |
| 47 | +} |
| 48 | + |
| 49 | +AtomicString::StringKind GetStringKind(const SharedNativeString* native_string) { |
| 50 | + if (!native_string->length()) { |
| 51 | + return AtomicString::StringKind::kIsMixed; |
| 52 | + } |
| 53 | + |
| 54 | + AtomicString::StringKind predictKind = std::islower(native_string->string()[0]) |
| 55 | + ? AtomicString::StringKind::kIsLowerCase |
| 56 | + : AtomicString::StringKind::kIsUpperCase; |
| 57 | + for (int i = 0; i < native_string->length(); i++) { |
| 58 | + uint16_t c = native_string->string()[i]; |
| 59 | + if (predictKind == AtomicString::StringKind::kIsUpperCase && !std::isupper(c)) { |
| 60 | + return AtomicString::StringKind::kIsMixed; |
| 61 | + } else if (predictKind == AtomicString::StringKind::kIsLowerCase && !std::islower(c)) { |
| 62 | + return AtomicString::StringKind::kIsMixed; |
| 63 | + } |
| 64 | + } |
| 65 | + |
| 66 | + return predictKind; |
| 67 | +} |
| 68 | + |
| 69 | +} // namespace |
| 70 | + |
| 71 | +class AtomicStringOneByteResource : public v8::String::ExternalOneByteStringResource { |
| 72 | + public: |
| 73 | + AtomicStringOneByteResource(const std::string& string) : string_(string){}; |
| 74 | + |
| 75 | + const char* data() const override { return string_.data(); }; |
| 76 | + size_t length() const override { return string_.length(); }; |
| 77 | + |
| 78 | + private: |
| 79 | + std::string string_; |
| 80 | +}; |
| 81 | + |
| 82 | +class AtomicStringTwoByteResource : public v8::String::ExternalStringResource { |
| 83 | + public: |
| 84 | + AtomicStringTwoByteResource(std::unique_ptr<AutoFreeNativeString>&& native_string) |
| 85 | + : string_(std::move(native_string)) {} |
| 86 | + |
| 87 | + const uint16_t* data() const override { return string_->string(); } |
| 88 | + size_t length() const override { return string_->length(); } |
| 89 | + |
| 90 | + private: |
| 91 | + std::unique_ptr<AutoFreeNativeString> string_; |
| 92 | +}; |
| 93 | + |
| 94 | +AtomicString::AtomicString(v8::Isolate* isolate, const std::string& string) |
| 95 | + : kind_(GetStringKind(string, string.size())) { |
| 96 | + auto* external_string_resource = new AtomicStringOneByteResource(string); |
| 97 | + string_ = v8::String::NewExternalOneByte(isolate, external_string_resource).ToLocalChecked(); |
| 98 | +} |
| 99 | + |
| 100 | +AtomicString::AtomicString(v8::Isolate* isolate, const char* str, size_t length) : kind_(GetStringKind(str, length)) { |
| 101 | + auto* external_string_resource = new AtomicStringOneByteResource(std::string(str, length)); |
| 102 | + string_ = v8::String::NewExternalOneByte(isolate, external_string_resource).ToLocalChecked(); |
| 103 | +} |
| 104 | + |
| 105 | +AtomicString::AtomicString(v8::Isolate* isolate, std::unique_ptr<AutoFreeNativeString>&& native_string) { |
| 106 | + auto* external_resource = new AtomicStringTwoByteResource(std::move(native_string)); |
| 107 | + kind_ = GetStringKind(native_string.get()); |
| 108 | + string_ = v8::String::NewExternalTwoByte(isolate, external_resource).ToLocalChecked(); |
| 109 | +} |
| 110 | + |
| 111 | +AtomicString::AtomicString(v8::Isolate* isolate, const uint16_t* str, size_t length) { |
| 112 | + auto native_string = std::unique_ptr<AutoFreeNativeString>( |
| 113 | + reinterpret_cast<AutoFreeNativeString*>( new SharedNativeString(str, length))); |
| 114 | + kind_ = GetStringKind(native_string.get()); |
| 115 | + auto* external_resource = new AtomicStringTwoByteResource(std::move(native_string)); |
| 116 | + string_ = v8::String::NewExternalTwoByte(isolate, external_resource).ToLocalChecked(); |
| 117 | +} |
| 118 | + |
| 119 | +bool AtomicString::IsEmpty() const { |
| 120 | + return *this == built_in_string::kempty_string || IsNull(); |
| 121 | +} |
| 122 | + |
| 123 | +bool AtomicString::IsNull() const { |
| 124 | + return string_->IsNull(); |
| 125 | +} |
| 126 | + |
| 127 | +bool AtomicString::Is8Bit() const { |
| 128 | + return string_->IsExternalOneByte(); |
| 129 | +} |
| 130 | + |
| 131 | +const uint8_t* AtomicString::Character8() const { |
| 132 | + assert(string_->IsExternal()); |
| 133 | + return reinterpret_cast<const uint8_t*>(string_->GetExternalOneByteStringResource()->data()); |
| 134 | +} |
| 135 | + |
| 136 | +const uint16_t* AtomicString::Character16() const { |
| 137 | + assert(string_->IsExternal()); |
| 138 | + return string_->GetExternalStringResource()->data(); |
| 139 | +} |
| 140 | + |
| 141 | +int AtomicString::Find(bool (*CharacterMatchFunction)(char)) const { |
| 142 | +// return JS_FindCharacterInAtom(runtime_, atom_, CharacterMatchFunction); |
| 143 | +} |
| 144 | + |
| 145 | +int AtomicString::Find(bool (*CharacterMatchFunction)(uint16_t)) const { |
| 146 | +// return JS_FindWCharacterInAtom(runtime_, atom_, CharacterMatchFunction); |
| 147 | +} |
| 148 | + |
| 149 | +std::string AtomicString::ToStdString(v8::Isolate* isolate) const { |
| 150 | + if (IsEmpty()) |
| 151 | + return ""; |
| 152 | + |
| 153 | + if (string_->IsExternalOneByte()) { |
| 154 | + return {string_->GetExternalOneByteStringResource()->data()}; |
| 155 | + } |
| 156 | + |
| 157 | + std::string result; |
| 158 | + size_t length = string_->Utf8Length(isolate); |
| 159 | + result.reserve(length); |
| 160 | + |
| 161 | + string_->WriteUtf8(isolate, result.data(), length); |
| 162 | + return result; |
| 163 | +} |
| 164 | + |
| 165 | +std::unique_ptr<SharedNativeString> AtomicString::ToNativeString(v8::Isolate* isolate) const { |
| 166 | + if (IsNull()) { |
| 167 | + // Null string is same like empty string |
| 168 | + return built_in_string::kempty_string.ToNativeString(isolate); |
| 169 | + } |
| 170 | + |
| 171 | + if (string_->IsExternalTwoByte()) { |
| 172 | + auto* resource = string_->GetExternalStringResource(); |
| 173 | + return SharedNativeString::FromTemporaryString(resource->data(), resource->length()); |
| 174 | + } |
| 175 | + |
| 176 | + size_t length = string_->Length(); |
| 177 | + std::vector<uint16_t> buffer; |
| 178 | + buffer.reserve(length); |
| 179 | + |
| 180 | + string_->Write(isolate, buffer.data(), length); |
| 181 | + return SharedNativeString::FromTemporaryString(buffer.data(), buffer.size()); |
| 182 | +} |
| 183 | + |
| 184 | +StringView AtomicString::ToStringView() const { |
| 185 | + if (IsNull()) { |
| 186 | + return built_in_string::kempty_string.ToStringView(); |
| 187 | + } |
| 188 | + |
| 189 | + if (string_->IsExternalOneByte()) { |
| 190 | + auto* resource = string_->GetExternalOneByteStringResource(); |
| 191 | + return StringView((void*)(resource->data()), resource->length(), false); |
| 192 | + } |
| 193 | + |
| 194 | + auto* resource = string_->GetExternalStringResource(); |
| 195 | + |
| 196 | + return StringView((void*)(resource->data()), resource->length(), true); |
| 197 | +} |
| 198 | + |
| 199 | +AtomicString AtomicString::ToUpperIfNecessary(v8::Isolate* isolate) const { |
| 200 | + if (kind_ == StringKind::kIsUpperCase) { |
| 201 | + return *this; |
| 202 | + } |
| 203 | + if (!string_upper_->IsNull() || IsNull()) |
| 204 | + return *this; |
| 205 | + AtomicString upperString = ToUpperSlow(isolate); |
| 206 | + string_upper_ = v8::Local<v8::String>(upperString.string_); |
| 207 | + return upperString; |
| 208 | +} |
| 209 | + |
| 210 | +AtomicString AtomicString::ToUpperSlow(v8::Isolate* isolate) const { |
| 211 | + std::string str = ToStdString(isolate); |
| 212 | + std::transform(str.begin(), str.end(), str.begin(), toupper); |
| 213 | + return {isolate, str}; |
| 214 | +} |
| 215 | + |
| 216 | +AtomicString AtomicString::ToLowerIfNecessary(v8::Isolate* isolate) const { |
| 217 | + if (kind_ == StringKind::kIsLowerCase) { |
| 218 | + return *this; |
| 219 | + } |
| 220 | + if (!string_lower_->IsNull() || IsNull()) |
| 221 | + return *this; |
| 222 | + AtomicString lowerString = ToLowerSlow(isolate); |
| 223 | + string_lower_ = lowerString.string_; |
| 224 | + return lowerString; |
| 225 | +} |
| 226 | + |
| 227 | +AtomicString AtomicString::ToLowerSlow(v8::Isolate* isolate) const { |
| 228 | + std::string str = ToStdString(isolate); |
| 229 | + std::transform(str.begin(), str.end(), str.begin(), tolower); |
| 230 | + return {isolate, str}; |
| 231 | +} |
| 232 | + |
| 233 | +template <typename CharType> |
| 234 | +inline AtomicString RemoveCharactersInternal(v8::Isolate* isolate, |
| 235 | + const AtomicString& self, |
| 236 | + const CharType* characters, |
| 237 | + size_t len, |
| 238 | + CharacterMatchFunctionPtr find_match) { |
| 239 | + const CharType* from = characters; |
| 240 | + const CharType* fromend = from + len; |
| 241 | + |
| 242 | + // Assume the common case will not remove any characters |
| 243 | + while (from != fromend && !find_match(*from)) |
| 244 | + ++from; |
| 245 | + if (from == fromend) |
| 246 | + return self; |
| 247 | + |
| 248 | + auto* to = (CharType*)malloc(len); |
| 249 | + size_t outc = static_cast<size_t>(from - characters); |
| 250 | + |
| 251 | + if (outc) |
| 252 | + memcpy(to, characters, outc * sizeof(CharType)); |
| 253 | + |
| 254 | + while (true) { |
| 255 | + while (from != fromend && find_match(*from)) |
| 256 | + ++from; |
| 257 | + while (from != fromend && !find_match(*from)) |
| 258 | + to[outc++] = *from++; |
| 259 | + if (from == fromend) |
| 260 | + break; |
| 261 | + } |
| 262 | + |
| 263 | + AtomicString str; |
| 264 | + |
| 265 | + if (outc == 0) { |
| 266 | + return AtomicString::Empty(); |
| 267 | + } |
| 268 | + |
| 269 | + auto data = (CharType*)malloc(outc); |
| 270 | + memcpy(data, to, outc); |
| 271 | + free(to); |
| 272 | + if (self.Is8Bit()) { |
| 273 | + str = AtomicString(isolate, reinterpret_cast<const char*>(data), outc); |
| 274 | + } else { |
| 275 | + str = AtomicString(isolate, reinterpret_cast<const uint16_t*>(data), outc); |
| 276 | + } |
| 277 | + |
| 278 | + free(data); |
| 279 | + return str; |
| 280 | +} |
| 281 | + |
| 282 | +AtomicString AtomicString::RemoveCharacters(v8::Isolate* isolate, CharacterMatchFunctionPtr find_match) { |
| 283 | + if (IsEmpty()) |
| 284 | + return AtomicString::Empty(); |
| 285 | + if (Is8Bit()) |
| 286 | + return RemoveCharactersInternal(isolate, *this, Character8(), string_->Utf8Length(isolate), find_match); |
| 287 | + return RemoveCharactersInternal(isolate, *this, Character16(), string_->Length(), find_match); |
| 288 | +} |
| 289 | + |
| 290 | +AtomicString::AtomicString(const webf::AtomicString& value) { |
| 291 | + string_ = v8::Local<v8::String>(value.string_); |
| 292 | +} |
| 293 | +AtomicString& AtomicString::operator=(const webf::AtomicString& other) { |
| 294 | + string_ = v8::Local<v8::String>(other.string_); |
| 295 | +} |
| 296 | + |
| 297 | +AtomicString::AtomicString(webf::AtomicString&& value) noexcept { |
| 298 | + string_ = v8::Local<v8::String>(value.string_); |
| 299 | +} |
| 300 | +AtomicString& AtomicString::operator=(webf::AtomicString&& value) noexcept { |
| 301 | + string_ = v8::Local<v8::String>(value.string_); |
| 302 | +} |
| 303 | + |
| 304 | +} // namespace webf |
0 commit comments