Patterns for LocalDate values

The LocalDate type supports the following patterns:

Standard Patterns

The following standard patterns are supported:

d: Short format pattern. This is the short date pattern as defined by the culture's DateTimeFormatInfo.ShortDatePattern. For example, in the invariant culture this is "MM/dd/yyyy".
D: Long format pattern. This is the long date pattern as defined by the culture's DateTimeFormatInfo.LongDatePattern. For example, in the invariant culture this is "dddd, dd MMMM yyyy". This is the default format pattern.
M: Month/day format pattern. This is the month/day pattern as defined by the culture's DateTimeFormatInfo.MonthDayPattern. For example, in the invariant culture this is "MMMM dd".
r: ISO format pattern plus calendar system, which is always "uuuu'-'MM'-'dd '('c')'" using the invariant culture.
R: ISO format pattern without calendar system, which is always "uuuu'-'MM'-'dd" using the invariant culture.

Custom Patterns

The following custom format pattern characters are supported for local dates. See custom pattern notes for general notes on custom patterns, including characters used for escaping and text literals.

For the meanings of "absolute" years and text handling, see later details.

Character	Meaning	Example
`yy`	Two digit year of era, in the range 0-99. When parsing, the "base century" is chosen from the template value; if the two-digit year is greater than 30, the corresponding year in the previous century is used, unless the century of the template value is already the first century of the era. If the template value is in the first century and the input is "00", an exception will be thrown on parsing. Note that when formatting, no checking is performed to ensure that the year will be parsed to the same value. (For example, 1725 would be formatted as 25 but parsed as 2025.) In general, use of this pattern specifier is discouraged, on the grounds of it leading to ambiguity.	Assuming a template value of 2000 (the default): 2012: `yy` => `12` 2040: `yy` => `40` - parsing "40" would give a date in 1940
`yyyy`	The year of era as 4 digits.	2000 A.D. (ISO calendar, en-US): `yyyy g` => `2000 A.D.` 13 B.C. (ISO calendar, en-US): `yyyy g` => `0013 B.C.`
`u`, `uu`, `uuu`, `uuuu`	The absolute year, zero-padded as necessary to the same number of characters as the number of 'u' characters, with an optional leading `-` sign. See notes below.	3 B.C.: `uuuu` => `-0002`
`g` or `gg`	The name of the era. This is calendar and culture specific. See notes below.	13 B.C. (ISO calendar, en-US): `y g` => `13 B.C.`
`M` or `MM`	Month of year specified as a number. `MM` is zero-padded; `M` is not.	June: `M` => `6` June: `MM` => `06` December: `M` => `12` December: `MM` => `12`
`MMM`	Abbreviated month name, parsed case-insensitively. This is culture-sensitive.	(In an English locale.) June: `MMM` => `Jun` (can parse from "jun" or "JUN" etc.) December: `MMM` => `Dec` (can parse from "dec" or "DEC" etc.)
`MMMM`	Full month name, parsed case-insensitively. This is culture-sensitive.	(In an English locale.) June: `MMMM` => `June` (can parse from "june" or "JUNE" etc.) December: `MMMM` => `December` (can parse from "december" or "DECEMBER" etc.)
`d` or `dd`	Day of month - `dd` is zero-padded; `d` is not.	1st of the month: `d` => `1` (would parse "01" as well) 1st of the month: `dd` => `01` 21st of the month: `d` => `21` 21st of the month: `dd` => `21`
`ddd`	Abbreviated day-of-week name, parsed case-insensitively. When parsing, the parsed day of week is validated against the computed date, but does not affect the calculations of that date. This value is culture-sensitive.	February 4th 2012 (a Saturday) en-US: `Sat` fr-FR: `sam.`
`dddd`	Full day-of-week name, parsed case-insensitively. When parsing, the parsed day of week is validated against the computed date, but does not affect the calculations of that date.	February 4th 2012 (a Saturday) en-US: `Saturday` fr-FR: `samedi`
`c`	The Noda-specific calendar system ID. This would rarely be appropriate for user-visible text, but allows exact round-tripping when serializing values via text.	`ISO` `Coptic 3` `Hijri Astronomical-Base16`
`/`	The date separator for the format provider; slash in the invariant culture.	en-US: `uuuu/MM/dd` => `2011/10/09` de-DE: `uuuu/MM/dd` => `2011.10.09` de-DE: `uuuu/MM/dd` => `2011.10.09`

Notes

Absolute and era years

Some calendars support multiple eras. For example, the ISO calendar supports the B.C. / B.C.E. and A.D. / C.E. eras. A mapping is provided between "year within era" and "absolute" year - where an absolute year uniquely identifies the date, and does not generally skip. In the ISO calendar, the absolute year 0 is deemed to be 1 B.C. and the absolute year 1 is deemed to be 1 A.D. thus giving a simplified arithmetic system.

Negative absolute years can be both parsed and formatted - so "13 B.C." would be formatted as "-0012" using the "uuuu" format.

Note that the meaning of the "y" specifier has changed over time: in Noda Time 1.x, this meant "absolute year"; it now means "year of era" to be consistent with the BCL. (This used to be handled by the "Y" specifier.) The "u" specifier is now used for "absolute year".

Text sources

Noda Time comes with its own limited set of era names, but month and day names are taken from the .NET framework. Unfortunately these are not available on a per-calendar basis, so the same names are used for all calendars, based solely on culture. It is hoped that future release of Noda Time may use information from the Unicode CLDR to provide a more comprehensive treatment.

Hebrew month names

The Hebrew calendar has two month numbering systems (scriptural and civil), each with their own benefits and drawbacks. Both have issues for text handling: as of Noda Time 1.3.0, the civil month numbering is assumed as that corresponds with the BCL month numbering... but due to the inclusion of a leap month, the month name/number correspondence changes in a leap year. Until this is fixed, it is strongly recommended that you only use month numbers in any textual representations of dates in the Hebrew calendar. Additionally, you may wish to consider how to best clarify whether that month number is in the scriptural or civil numbering system.

Parsing and dependent values

Text parsing consists of two phases.

The first phase obtains values from the text independently, from left to right. These values cannot affect each other during the first phase, which leads to a small limitation in date patterns: a date pattern cannot contain both a calendar specifier (c) and an era specifier (g or gg). The parser would not be able to determine which eras were valid without taking the calendar into account.

The second phase combines and validates values. For example, a "day of month" value may be valid in some year/month/calendar combinations but not in others. Likewise, a "day of week" value can only be validated when the complete date is known.

The ordering of these phases can lead to some potentially-unexpected results in the case of errors. For example, when parsing the text "2000-99-00" with an ISO date rule, the result will be a failure due to the "00" rather than due to the "99". When a two-digit month value is parsed, the first phase validates that it's in the range 1-99 inclusive, leaving more fine-grained validation to the second phase. In the particular case of an ISO pattern, the fine-grained validation could be performed earlier, but in the more general date-parsing case, the set of valid month values can depend on other values (the calendar and year). For the sake of code simplicity, Noda Time does not attempt to ensure that the error is reported on the earliest possible part of the text input, although that will usually be the case.