1 // Copyright 2011 the V8 project authors. All rights reserved. 2 // Use of this source code is governed by a BSD-style license that can be 3 // found in the LICENSE file. 4 5 #ifndef V8_DATEPARSER_INL_H_ 6 #define V8_DATEPARSER_INL_H_ 7 8 #include "src/dateparser.h" 9 10 namespace v8 { 11 namespace internal { 12 13 template <typename Char> 14 bool DateParser::Parse(Vector<Char> str, 15 FixedArray* out, 16 UnicodeCache* unicode_cache) { 17 ASSERT(out->length() >= OUTPUT_SIZE); 18 InputReader<Char> in(unicode_cache, str); 19 DateStringTokenizer<Char> scanner(&in); 20 TimeZoneComposer tz; 21 TimeComposer time; 22 DayComposer day; 23 24 // Specification: 25 // Accept ES5 ISO 8601 date-time-strings or legacy dates compatible 26 // with Safari. 27 // ES5 ISO 8601 dates: 28 // [('-'|'+')yy]yyyy[-MM[-DD]][THH:mm[:ss[.sss]][Z|(+|-)hh:mm]] 29 // where yyyy is in the range 0000..9999 and 30 // +/-yyyyyy is in the range -999999..+999999 - 31 // but -000000 is invalid (year zero must be positive), 32 // MM is in the range 01..12, 33 // DD is in the range 01..31, 34 // MM and DD defaults to 01 if missing,, 35 // HH is generally in the range 00..23, but can be 24 if mm, ss 36 // and sss are zero (or missing), representing midnight at the 37 // end of a day, 38 // mm and ss are in the range 00..59, 39 // sss is in the range 000..999, 40 // hh is in the range 00..23, 41 // mm, ss, and sss default to 00 if missing, and 42 // timezone defaults to Z if missing 43 // (following Safari, ISO actually demands local time). 44 // Extensions: 45 // We also allow sss to have more or less than three digits (but at 46 // least one). 47 // We allow hh:mm to be specified as hhmm. 48 // Legacy dates: 49 // Any unrecognized word before the first number is ignored. 50 // Parenthesized text is ignored. 51 // An unsigned number followed by ':' is a time value, and is 52 // added to the TimeComposer. A number followed by '::' adds a second 53 // zero as well. A number followed by '.' is also a time and must be 54 // followed by milliseconds. 55 // Any other number is a date component and is added to DayComposer. 56 // A month name (or really: any word having the same first three letters 57 // as a month name) is recorded as a named month in the Day composer. 58 // A word recognizable as a time-zone is recorded as such, as is 59 // '(+|-)(hhmm|hh:)'. 60 // Legacy dates don't allow extra signs ('+' or '-') or umatched ')' 61 // after a number has been read (before the first number, any garbage 62 // is allowed). 63 // Intersection of the two: 64 // A string that matches both formats (e.g. 1970-01-01) will be 65 // parsed as an ES5 date-time string - which means it will default 66 // to UTC time-zone. That's unavoidable if following the ES5 67 // specification. 68 // After a valid "T" has been read while scanning an ES5 datetime string, 69 // the input can no longer be a valid legacy date, since the "T" is a 70 // garbage string after a number has been read. 71 72 // First try getting as far as possible with as ES5 Date Time String. 73 DateToken next_unhandled_token = ParseES5DateTime(&scanner, &day, &time, &tz); 74 if (next_unhandled_token.IsInvalid()) return false; 75 bool has_read_number = !day.IsEmpty(); 76 // If there's anything left, continue with the legacy parser. 77 for (DateToken token = next_unhandled_token; 78 !token.IsEndOfInput(); 79 token = scanner.Next()) { 80 if (token.IsNumber()) { 81 has_read_number = true; 82 int n = token.number(); 83 if (scanner.SkipSymbol(':')) { 84 if (scanner.SkipSymbol(':')) { 85 // n + "::" 86 if (!time.IsEmpty()) return false; 87 time.Add(n); 88 time.Add(0); 89 } else { 90 // n + ":" 91 if (!time.Add(n)) return false; 92 if (scanner.Peek().IsSymbol('.')) scanner.Next(); 93 } 94 } else if (scanner.SkipSymbol('.') && time.IsExpecting(n)) { 95 time.Add(n); 96 if (!scanner.Peek().IsNumber()) return false; 97 int n = ReadMilliseconds(scanner.Next()); 98 if (n < 0) return false; 99 time.AddFinal(n); 100 } else if (tz.IsExpecting(n)) { 101 tz.SetAbsoluteMinute(n); 102 } else if (time.IsExpecting(n)) { 103 time.AddFinal(n); 104 // Require end, white space, "Z", "+" or "-" immediately after 105 // finalizing time. 106 DateToken peek = scanner.Peek(); 107 if (!peek.IsEndOfInput() && 108 !peek.IsWhiteSpace() && 109 !peek.IsKeywordZ() && 110 !peek.IsAsciiSign()) return false; 111 } else { 112 if (!day.Add(n)) return false; 113 scanner.SkipSymbol('-'); 114 } 115 } else if (token.IsKeyword()) { 116 // Parse a "word" (sequence of chars. >= 'A'). 117 KeywordType type = token.keyword_type(); 118 int value = token.keyword_value(); 119 if (type == AM_PM && !time.IsEmpty()) { 120 time.SetHourOffset(value); 121 } else if (type == MONTH_NAME) { 122 day.SetNamedMonth(value); 123 scanner.SkipSymbol('-'); 124 } else if (type == TIME_ZONE_NAME && has_read_number) { 125 tz.Set(value); 126 } else { 127 // Garbage words are illegal if a number has been read. 128 if (has_read_number) return false; 129 // The first number has to be separated from garbage words by 130 // whitespace or other separators. 131 if (scanner.Peek().IsNumber()) return false; 132 } 133 } else if (token.IsAsciiSign() && (tz.IsUTC() || !time.IsEmpty())) { 134 // Parse UTC offset (only after UTC or time). 135 tz.SetSign(token.ascii_sign()); 136 // The following number may be empty. 137 int n = 0; 138 if (scanner.Peek().IsNumber()) { 139 n = scanner.Next().number(); 140 } 141 has_read_number = true; 142 143 if (scanner.Peek().IsSymbol(':')) { 144 tz.SetAbsoluteHour(n); 145 tz.SetAbsoluteMinute(kNone); 146 } else { 147 tz.SetAbsoluteHour(n / 100); 148 tz.SetAbsoluteMinute(n % 100); 149 } 150 } else if ((token.IsAsciiSign() || token.IsSymbol(')')) && 151 has_read_number) { 152 // Extra sign or ')' is illegal if a number has been read. 153 return false; 154 } else { 155 // Ignore other characters and whitespace. 156 } 157 } 158 159 return day.Write(out) && time.Write(out) && tz.Write(out); 160 } 161 162 163 template<typename CharType> 164 DateParser::DateToken DateParser::DateStringTokenizer<CharType>::Scan() { 165 int pre_pos = in_->position(); 166 if (in_->IsEnd()) return DateToken::EndOfInput(); 167 if (in_->IsAsciiDigit()) { 168 int n = in_->ReadUnsignedNumeral(); 169 int length = in_->position() - pre_pos; 170 return DateToken::Number(n, length); 171 } 172 if (in_->Skip(':')) return DateToken::Symbol(':'); 173 if (in_->Skip('-')) return DateToken::Symbol('-'); 174 if (in_->Skip('+')) return DateToken::Symbol('+'); 175 if (in_->Skip('.')) return DateToken::Symbol('.'); 176 if (in_->Skip(')')) return DateToken::Symbol(')'); 177 if (in_->IsAsciiAlphaOrAbove()) { 178 ASSERT(KeywordTable::kPrefixLength == 3); 179 uint32_t buffer[3] = {0, 0, 0}; 180 int length = in_->ReadWord(buffer, 3); 181 int index = KeywordTable::Lookup(buffer, length); 182 return DateToken::Keyword(KeywordTable::GetType(index), 183 KeywordTable::GetValue(index), 184 length); 185 } 186 if (in_->SkipWhiteSpace()) { 187 return DateToken::WhiteSpace(in_->position() - pre_pos); 188 } 189 if (in_->SkipParentheses()) { 190 return DateToken::Unknown(); 191 } 192 in_->Next(); 193 return DateToken::Unknown(); 194 } 195 196 197 template <typename Char> 198 DateParser::DateToken DateParser::ParseES5DateTime( 199 DateStringTokenizer<Char>* scanner, 200 DayComposer* day, 201 TimeComposer* time, 202 TimeZoneComposer* tz) { 203 ASSERT(day->IsEmpty()); 204 ASSERT(time->IsEmpty()); 205 ASSERT(tz->IsEmpty()); 206 207 // Parse mandatory date string: [('-'|'+')yy]yyyy[':'MM[':'DD]] 208 if (scanner->Peek().IsAsciiSign()) { 209 // Keep the sign token, so we can pass it back to the legacy 210 // parser if we don't use it. 211 DateToken sign_token = scanner->Next(); 212 if (!scanner->Peek().IsFixedLengthNumber(6)) return sign_token; 213 int sign = sign_token.ascii_sign(); 214 int year = scanner->Next().number(); 215 if (sign < 0 && year == 0) return sign_token; 216 day->Add(sign * year); 217 } else if (scanner->Peek().IsFixedLengthNumber(4)) { 218 day->Add(scanner->Next().number()); 219 } else { 220 return scanner->Next(); 221 } 222 if (scanner->SkipSymbol('-')) { 223 if (!scanner->Peek().IsFixedLengthNumber(2) || 224 !DayComposer::IsMonth(scanner->Peek().number())) return scanner->Next(); 225 day->Add(scanner->Next().number()); 226 if (scanner->SkipSymbol('-')) { 227 if (!scanner->Peek().IsFixedLengthNumber(2) || 228 !DayComposer::IsDay(scanner->Peek().number())) return scanner->Next(); 229 day->Add(scanner->Next().number()); 230 } 231 } 232 // Check for optional time string: 'T'HH':'mm[':'ss['.'sss]]Z 233 if (!scanner->Peek().IsKeywordType(TIME_SEPARATOR)) { 234 if (!scanner->Peek().IsEndOfInput()) return scanner->Next(); 235 } else { 236 // ES5 Date Time String time part is present. 237 scanner->Next(); 238 if (!scanner->Peek().IsFixedLengthNumber(2) || 239 !Between(scanner->Peek().number(), 0, 24)) { 240 return DateToken::Invalid(); 241 } 242 // Allow 24:00[:00[.000]], but no other time starting with 24. 243 bool hour_is_24 = (scanner->Peek().number() == 24); 244 time->Add(scanner->Next().number()); 245 if (!scanner->SkipSymbol(':')) return DateToken::Invalid(); 246 if (!scanner->Peek().IsFixedLengthNumber(2) || 247 !TimeComposer::IsMinute(scanner->Peek().number()) || 248 (hour_is_24 && scanner->Peek().number() > 0)) { 249 return DateToken::Invalid(); 250 } 251 time->Add(scanner->Next().number()); 252 if (scanner->SkipSymbol(':')) { 253 if (!scanner->Peek().IsFixedLengthNumber(2) || 254 !TimeComposer::IsSecond(scanner->Peek().number()) || 255 (hour_is_24 && scanner->Peek().number() > 0)) { 256 return DateToken::Invalid(); 257 } 258 time->Add(scanner->Next().number()); 259 if (scanner->SkipSymbol('.')) { 260 if (!scanner->Peek().IsNumber() || 261 (hour_is_24 && scanner->Peek().number() > 0)) { 262 return DateToken::Invalid(); 263 } 264 // Allow more or less than the mandated three digits. 265 time->Add(ReadMilliseconds(scanner->Next())); 266 } 267 } 268 // Check for optional timezone designation: 'Z' | ('+'|'-')hh':'mm 269 if (scanner->Peek().IsKeywordZ()) { 270 scanner->Next(); 271 tz->Set(0); 272 } else if (scanner->Peek().IsSymbol('+') || 273 scanner->Peek().IsSymbol('-')) { 274 tz->SetSign(scanner->Next().symbol() == '+' ? 1 : -1); 275 if (scanner->Peek().IsFixedLengthNumber(4)) { 276 // hhmm extension syntax. 277 int hourmin = scanner->Next().number(); 278 int hour = hourmin / 100; 279 int min = hourmin % 100; 280 if (!TimeComposer::IsHour(hour) || !TimeComposer::IsMinute(min)) { 281 return DateToken::Invalid(); 282 } 283 tz->SetAbsoluteHour(hour); 284 tz->SetAbsoluteMinute(min); 285 } else { 286 // hh:mm standard syntax. 287 if (!scanner->Peek().IsFixedLengthNumber(2) || 288 !TimeComposer::IsHour(scanner->Peek().number())) { 289 return DateToken::Invalid(); 290 } 291 tz->SetAbsoluteHour(scanner->Next().number()); 292 if (!scanner->SkipSymbol(':')) return DateToken::Invalid(); 293 if (!scanner->Peek().IsFixedLengthNumber(2) || 294 !TimeComposer::IsMinute(scanner->Peek().number())) { 295 return DateToken::Invalid(); 296 } 297 tz->SetAbsoluteMinute(scanner->Next().number()); 298 } 299 } 300 if (!scanner->Peek().IsEndOfInput()) return DateToken::Invalid(); 301 } 302 // Successfully parsed ES5 Date Time String. Default to UTC if no TZ given. 303 if (tz->IsEmpty()) tz->Set(0); 304 day->set_iso_date(); 305 return DateToken::EndOfInput(); 306 } 307 308 309 } } // namespace v8::internal 310 311 #endif // V8_DATEPARSER_INL_H_ 312