strftime

The C / POSIX strftime token grammar — the %-directives shared by C, Python, Ruby, and PHP. Imported from rosetta-date/dialects:

import { convert } from 'rosetta-date'
import { moment, strftime } from 'rosetta-date/dialects'
 
convert('%Y-%m-%d', { from: strftime, to: moment }) // 'YYYY-MM-DD'
convert('DD/MM/YYYY', { from: moment, to: strftime }) // '%d/%m/%Y'

• Spec: POSIX strftime
• Family: directive — every token is introduced by %
• Literals: any non-directive text is literal verbatim; a literal % is %% — see Literals & Adjacency

A directive grammar

strftime belongs to the directive family — the mirror image of the delimited family: every token is introduced by the % marker, and all other text — letters included — is literal. So Y on its own is the literal letter Y; only %Y is the year. Because each directive is self-delimiting, adjacent directives never need a separator (%H%M is unambiguous).

Tokens

Each directive and the canonical field it maps to. For how these compare across dialects, see Token Mapping.

Composite directives

A handful of directives are composites — a single spelling that stands for a whole sub-pattern. They expand on parse, so converting one out yields the expanded fields:

A composite is a parse-time macro, so rendering produces its expansion, never the composite: %T round-trips to %H:%M:%S (identical output, normalized spelling), the same way an alias normalizes — just across several tokens. %n and %t are the same mechanism applied to literal whitespace: they expand to a literal newline/tab, so they normalize to that raw character (the output is identical, only the %n/%t spelling is not preserved).

Padding flags

A directive may carry a glibc padding flag between the % and the field letter, overriding that field’s default padding:

%-X fills the numeric forms the bare directives lack. %d is always zero-padded, so %-d is how strftime writes a plain numeric day, matching moment’s D and ldml’s d. It is recognized for %-d %-m %-H %-I %-M %-S %-j %-V. %0X is redundant with the bare directive, so it parses but normalizes to %X on render.

%_X is recognized for the same fields (%_d %_m %_H %_I %_M %_S %_j %_V). %_d, %_H, and %_I are the flag form of the %e/%k/%l presets and normalize to them. The rest extend blank-padding to fields with no preset letter. No other built-in dialect renders a blank-padded field, so converting one out falls to the unsupported-token policy.

Aliases

%h is parsed as an abbreviated month, the same as %b; it normalizes to %b when rendered.

Caveats