primary
- Returns:
- names of English, Russian and the default locale
DateTimeScanner.
Lookups take a range of a CharSequence and allocate nothing: keys are stored
pre-normalized (lower-cased, without a trailing '.') in an open-addressed table and
compared character by character.
primary() covers English, Russian and the default locale and is built eagerly.
global() additionally covers one locale per language known to the JDK and is built on
first use, because that build walks a few hundred locales. Tokens that mean different months in
different languages are dropped from the global table unless the primary table defines them.
Based on sqlite-jdbc: FastDateParser.getDisplayNames and CaseInsensitiveTextStrategy, which collect every style a locale renders and lower-case the names once up front; the alternation regex they build is replaced here by a table that can be probed without allocating.
intamPmOf(CharSequence text,
int start,
int end) static CalendarNamesglobal()booleanisWeekday(CharSequence text,
int start,
int end) intmonthOf(CharSequence text,
int start,
int end) static CalendarNamesprimary()static intzoneOffsetSecondsOf(CharSequence text,
int start,
int end) NOT_FOUNDtrue when the range names a day of the weekAbbreviations are inherently ambiguous across regions (CST and IST name
several zones); the most widely used reading is applied, as other lenient parsers do.
Based on sqlite-jdbc:
FastDateParser.TimeZoneStrategy,
which resolves them from DateFormatSymbols.getZoneStrings() per locale instead.
NOT_FOUND