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.
5 #ifndef V8_DATEPARSER_INL_H_
6 #define V8_DATEPARSER_INL_H_
8 #include "src/char-predicates-inl.h"
9 #include "src/dateparser.h"
10 #include "src/unicode-cache-inl.h"
15 template <typename Char>
16 bool DateParser::Parse(Vector<Char> str,
18 UnicodeCache* unicode_cache) {
19 DCHECK(out->length() >= OUTPUT_SIZE);
20 InputReader<Char> in(unicode_cache, str);
21 DateStringTokenizer<Char> scanner(&in);
27 // Accept ES6 ISO 8601 date-time-strings or legacy dates compatible
29 // ES6 ISO 8601 dates:
30 // [('-'|'+')yy]yyyy[-MM[-DD]][THH:mm[:ss[.sss]][Z|(+|-)hh:mm]]
31 // where yyyy is in the range 0000..9999 and
32 // +/-yyyyyy is in the range -999999..+999999 -
33 // but -000000 is invalid (year zero must be positive),
34 // MM is in the range 01..12,
35 // DD is in the range 01..31,
36 // MM and DD defaults to 01 if missing,,
37 // HH is generally in the range 00..23, but can be 24 if mm, ss
38 // and sss are zero (or missing), representing midnight at the
40 // mm and ss are in the range 00..59,
41 // sss is in the range 000..999,
42 // hh is in the range 00..23,
43 // mm, ss, and sss default to 00 if missing, and
44 // timezone defaults to local time if missing.
46 // We also allow sss to have more or less than three digits (but at
48 // We allow hh:mm to be specified as hhmm.
50 // Any unrecognized word before the first number is ignored.
51 // Parenthesized text is ignored.
52 // An unsigned number followed by ':' is a time value, and is
53 // added to the TimeComposer. A number followed by '::' adds a second
54 // zero as well. A number followed by '.' is also a time and must be
55 // followed by milliseconds.
56 // Any other number is a date component and is added to DayComposer.
57 // A month name (or really: any word having the same first three letters
58 // as a month name) is recorded as a named month in the Day composer.
59 // A word recognizable as a time-zone is recorded as such, as is
61 // Legacy dates don't allow extra signs ('+' or '-') or umatched ')'
62 // after a number has been read (before the first number, any garbage
64 // Intersection of the two:
65 // A string that matches both formats (e.g. 1970-01-01) will be
66 // parsed as an ES6 date-time string.
67 // After a valid "T" has been read while scanning an ES6 datetime string,
68 // the input can no longer be a valid legacy date, since the "T" is a
69 // garbage string after a number has been read.
71 // First try getting as far as possible with as ES6 Date Time String.
72 DateToken next_unhandled_token = ParseES6DateTime(&scanner, &day, &time, &tz);
73 if (next_unhandled_token.IsInvalid()) return false;
74 bool has_read_number = !day.IsEmpty();
75 // If there's anything left, continue with the legacy parser.
76 for (DateToken token = next_unhandled_token;
77 !token.IsEndOfInput();
78 token = scanner.Next()) {
79 if (token.IsNumber()) {
80 has_read_number = true;
81 int n = token.number();
82 if (scanner.SkipSymbol(':')) {
83 if (scanner.SkipSymbol(':')) {
85 if (!time.IsEmpty()) return false;
90 if (!time.Add(n)) return false;
91 if (scanner.Peek().IsSymbol('.')) scanner.Next();
93 } else if (scanner.SkipSymbol('.') && time.IsExpecting(n)) {
95 if (!scanner.Peek().IsNumber()) return false;
96 int n = ReadMilliseconds(scanner.Next());
97 if (n < 0) return false;
99 } else if (tz.IsExpecting(n)) {
100 tz.SetAbsoluteMinute(n);
101 } else if (time.IsExpecting(n)) {
103 // Require end, white space, "Z", "+" or "-" immediately after
105 DateToken peek = scanner.Peek();
106 if (!peek.IsEndOfInput() &&
107 !peek.IsWhiteSpace() &&
108 !peek.IsKeywordZ() &&
109 !peek.IsAsciiSign()) return false;
111 if (!day.Add(n)) return false;
112 scanner.SkipSymbol('-');
114 } else if (token.IsKeyword()) {
115 // Parse a "word" (sequence of chars. >= 'A').
116 KeywordType type = token.keyword_type();
117 int value = token.keyword_value();
118 if (type == AM_PM && !time.IsEmpty()) {
119 time.SetHourOffset(value);
120 } else if (type == MONTH_NAME) {
121 day.SetNamedMonth(value);
122 scanner.SkipSymbol('-');
123 } else if (type == TIME_ZONE_NAME && has_read_number) {
126 // Garbage words are illegal if a number has been read.
127 if (has_read_number) return false;
128 // The first number has to be separated from garbage words by
129 // whitespace or other separators.
130 if (scanner.Peek().IsNumber()) return false;
132 } else if (token.IsAsciiSign() && (tz.IsUTC() || !time.IsEmpty())) {
133 // Parse UTC offset (only after UTC or time).
134 tz.SetSign(token.ascii_sign());
135 // The following number may be empty.
137 if (scanner.Peek().IsNumber()) {
138 n = scanner.Next().number();
140 has_read_number = true;
142 if (scanner.Peek().IsSymbol(':')) {
143 tz.SetAbsoluteHour(n);
144 tz.SetAbsoluteMinute(kNone);
146 tz.SetAbsoluteHour(n / 100);
147 tz.SetAbsoluteMinute(n % 100);
149 } else if ((token.IsAsciiSign() || token.IsSymbol(')')) &&
151 // Extra sign or ')' is illegal if a number has been read.
154 // Ignore other characters and whitespace.
158 return day.Write(out) && time.Write(out) && tz.Write(out);
162 template<typename CharType>
163 DateParser::DateToken DateParser::DateStringTokenizer<CharType>::Scan() {
164 int pre_pos = in_->position();
165 if (in_->IsEnd()) return DateToken::EndOfInput();
166 if (in_->IsAsciiDigit()) {
167 int n = in_->ReadUnsignedNumeral();
168 int length = in_->position() - pre_pos;
169 return DateToken::Number(n, length);
171 if (in_->Skip(':')) return DateToken::Symbol(':');
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_->IsAsciiAlphaOrAbove()) {
177 DCHECK(KeywordTable::kPrefixLength == 3);
178 uint32_t buffer[3] = {0, 0, 0};
179 int length = in_->ReadWord(buffer, 3);
180 int index = KeywordTable::Lookup(buffer, length);
181 return DateToken::Keyword(KeywordTable::GetType(index),
182 KeywordTable::GetValue(index),
185 if (in_->SkipWhiteSpace()) {
186 return DateToken::WhiteSpace(in_->position() - pre_pos);
188 if (in_->SkipParentheses()) {
189 return DateToken::Unknown();
192 return DateToken::Unknown();
196 template <typename Char>
197 bool DateParser::InputReader<Char>::SkipWhiteSpace() {
198 if (unicode_cache_->IsWhiteSpaceOrLineTerminator(ch_)) {
206 template <typename Char>
207 bool DateParser::InputReader<Char>::SkipParentheses() {
208 if (ch_ != '(') return false;
211 if (ch_ == ')') --balance;
212 else if (ch_ == '(') ++balance;
214 } while (balance > 0 && ch_);
219 template <typename Char>
220 DateParser::DateToken DateParser::ParseES6DateTime(
221 DateStringTokenizer<Char>* scanner,
224 TimeZoneComposer* tz) {
225 DCHECK(day->IsEmpty());
226 DCHECK(time->IsEmpty());
227 DCHECK(tz->IsEmpty());
229 // Parse mandatory date string: [('-'|'+')yy]yyyy[':'MM[':'DD]]
230 if (scanner->Peek().IsAsciiSign()) {
231 // Keep the sign token, so we can pass it back to the legacy
232 // parser if we don't use it.
233 DateToken sign_token = scanner->Next();
234 if (!scanner->Peek().IsFixedLengthNumber(6)) return sign_token;
235 int sign = sign_token.ascii_sign();
236 int year = scanner->Next().number();
237 if (sign < 0 && year == 0) return sign_token;
238 day->Add(sign * year);
239 } else if (scanner->Peek().IsFixedLengthNumber(4)) {
240 day->Add(scanner->Next().number());
242 return scanner->Next();
244 if (scanner->SkipSymbol('-')) {
245 if (!scanner->Peek().IsFixedLengthNumber(2) ||
246 !DayComposer::IsMonth(scanner->Peek().number())) return scanner->Next();
247 day->Add(scanner->Next().number());
248 if (scanner->SkipSymbol('-')) {
249 if (!scanner->Peek().IsFixedLengthNumber(2) ||
250 !DayComposer::IsDay(scanner->Peek().number())) return scanner->Next();
251 day->Add(scanner->Next().number());
254 // Check for optional time string: 'T'HH':'mm[':'ss['.'sss]]Z
255 if (!scanner->Peek().IsKeywordType(TIME_SEPARATOR)) {
256 if (!scanner->Peek().IsEndOfInput()) return scanner->Next();
258 // ES6 Date Time String time part is present.
260 if (!scanner->Peek().IsFixedLengthNumber(2) ||
261 !Between(scanner->Peek().number(), 0, 24)) {
262 return DateToken::Invalid();
264 // Allow 24:00[:00[.000]], but no other time starting with 24.
265 bool hour_is_24 = (scanner->Peek().number() == 24);
266 time->Add(scanner->Next().number());
267 if (!scanner->SkipSymbol(':')) return DateToken::Invalid();
268 if (!scanner->Peek().IsFixedLengthNumber(2) ||
269 !TimeComposer::IsMinute(scanner->Peek().number()) ||
270 (hour_is_24 && scanner->Peek().number() > 0)) {
271 return DateToken::Invalid();
273 time->Add(scanner->Next().number());
274 if (scanner->SkipSymbol(':')) {
275 if (!scanner->Peek().IsFixedLengthNumber(2) ||
276 !TimeComposer::IsSecond(scanner->Peek().number()) ||
277 (hour_is_24 && scanner->Peek().number() > 0)) {
278 return DateToken::Invalid();
280 time->Add(scanner->Next().number());
281 if (scanner->SkipSymbol('.')) {
282 if (!scanner->Peek().IsNumber() ||
283 (hour_is_24 && scanner->Peek().number() > 0)) {
284 return DateToken::Invalid();
286 // Allow more or less than the mandated three digits.
287 time->Add(ReadMilliseconds(scanner->Next()));
290 // Check for optional timezone designation: 'Z' | ('+'|'-')hh':'mm
291 if (scanner->Peek().IsKeywordZ()) {
294 } else if (scanner->Peek().IsSymbol('+') ||
295 scanner->Peek().IsSymbol('-')) {
296 tz->SetSign(scanner->Next().symbol() == '+' ? 1 : -1);
297 if (scanner->Peek().IsFixedLengthNumber(4)) {
298 // hhmm extension syntax.
299 int hourmin = scanner->Next().number();
300 int hour = hourmin / 100;
301 int min = hourmin % 100;
302 if (!TimeComposer::IsHour(hour) || !TimeComposer::IsMinute(min)) {
303 return DateToken::Invalid();
305 tz->SetAbsoluteHour(hour);
306 tz->SetAbsoluteMinute(min);
308 // hh:mm standard syntax.
309 if (!scanner->Peek().IsFixedLengthNumber(2) ||
310 !TimeComposer::IsHour(scanner->Peek().number())) {
311 return DateToken::Invalid();
313 tz->SetAbsoluteHour(scanner->Next().number());
314 if (!scanner->SkipSymbol(':')) return DateToken::Invalid();
315 if (!scanner->Peek().IsFixedLengthNumber(2) ||
316 !TimeComposer::IsMinute(scanner->Peek().number())) {
317 return DateToken::Invalid();
319 tz->SetAbsoluteMinute(scanner->Next().number());
322 if (!scanner->Peek().IsEndOfInput()) return DateToken::Invalid();
324 // Successfully parsed ES6 Date Time String.
326 return DateToken::EndOfInput();
330 } // namespace internal
333 #endif // V8_DATEPARSER_INL_H_