File Coverage

blib/lib/Time/Format.pm

Criterion	Covered	Total	%
statement	78	83	93.9
branch	28	40	70.0
condition	4	6	66.6
subroutine	13	13	100.0
pod	2	2	100.0
total	125	144	86.8

=for gpg
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

=encoding utf8

=head1 NAME

Time::Format - Easy-to-use date/time formatting.

=head1 VERSION

This is version 1.15 of Time::Format, July 26, 2019.

=cut

use strict;
package Time::Format;
$Time::Format::VERSION  = '1.15';

# This module claims to be compatible with the following versions
# of Time::Format_XS.
%Time::Format::XSCOMPAT = map {$_ => 1} qw(1.01 1.02 1.03);

sub _croak
{
    require Carp;
    goto &Carp::croak;
}

# Store the file offset of the __DATA__ region.
my $data_pos = tell DATA;
close DATA;                     # so we don't hold a lock on this file.

# Here we go through a bunch of tests to decide whether we can use the
# XS module, or if we need to load and compile the perl-only
# subroutines (which are stored in __DATA__).
my $load_perlonly = 0;
$load_perlonly = 1  if defined $Time::Format::NOXS  &&  $Time::Format::NOXS;

if (!$load_perlonly)
{
    # Check whether the optional XS module is installed.
    eval { require Time::Format_XS };

    if ($@  ||  !defined $Time::Format_XS::VERSION)
    {
        $load_perlonly = 1;
    }
    else
    {
        # Check that we're compatible with them (backwards compatibility)
        # or they're compatible with us (forwards compatibility).
        unless ($Time::Format::XSCOMPAT{$Time::Format_XS::VERSION}
            ||  $Time::Format_XS::PLCOMPAT{$Time::Format::VERSION})
        {
            warn "Your Time::Format_XS version ($Time::Format_XS::VERSION) "
               . "is not compatible with Time::Format version ($Time::Format::VERSION).\n"
               . "Using Perl-only functions.\n";
            $load_perlonly = 1;
        }
    }

    # Okay to use the XS version?  Great.  Wrap it.
    if (!$load_perlonly)
    {
        *time_format = \&Time::Format_XS::time_format;
    }
}

if ($load_perlonly)
{
    # Time::Format_XS not installed, or version mismatch, or NOXS was set.
    # The perl routines will need to be loaded.
    # But let's defer this until someone actually calls time_format().
    *time_format = sub
    {
        if (not defined &time_format_perlonly)
        {
            open DATA, '<', __FILE__
                or die "Can't access code in " . __FILE__ . ": $!\n";;

            flock DATA, 1;      # LOCK_SH
            seek  DATA, $data_pos, 0;
            local $^W = 0;        # disable warning about subroutines redefined
            local $/ = undef;     # slurp
            my $code = <DATA>;
            flock DATA, 8;      # LOCK_UN
            close DATA;

            eval $code;
            die if $@;
        }
        goto &time_format_perlonly;
    };
    undef $Time::Format_XS::VERSION;    # Indicate that XS version is not available.
}


my @EXPORT      = qw(%time time_format);
my @EXPORT_OK   = qw(%time %strftime %manip time_format time_strftime time_manip);

# We don't need any of Exporter's fancy features, so it's quicker to
# do the import ourselves.
sub import
{
    my $pkg  = shift;
    my ($cpkg,$file,$line) = caller;
    my @symbols;
    if (@_)
    {
        if (grep $_ eq ':all', @_)
        {
            @symbols = (@EXPORT, @EXPORT_OK, grep $_ ne ':all', @_);
        } else {
            @symbols = @_;
        }
        my %seen;
        @symbols = grep !$seen{$_}++, @symbols;
    } else {
        @symbols = @EXPORT;
    }
    my %ok;
    @ok{@EXPORT_OK,@EXPORT} = ();
    my @badsym = grep !exists $ok{$_}, @symbols;
    if (@badsym)
    {
        my $s = @badsym>1? 's'   : '';
        my $v = @badsym>1? 'are' : 'is';
        _croak ("The symbol$s ", join(', ', @badsym), " $v not exported by Time::Format at $file line $line.\n");
    }

    no strict 'refs';
    foreach my $sym (@symbols)
    {
        $sym =~ s/^([\$\&\@\%])?//;
        my $pfx = $1 || '&';
        my $calsym = $cpkg . '::' . $sym;
        my $mysym  = $pkg  . '::' . $sym;
        if ($pfx eq '%')
        {
            *$calsym = \%$mysym;
        } elsif ($pfx eq '@') {
            *$calsym = \@$mysym;
        } elsif ($pfx eq '$') {
            *$calsym = \$$mysym;
        } else {
            *$calsym = \&$mysym;
        }
    }
}

# Simple tied-hash implementation.

# Each hash is simply tied to a subroutine reference.  "Fetching" a
# value from the hash invokes the subroutine.  If a hash (tied or
# otherwise) has multiple comma-separated values but the leading
# character is a $, then Perl joins the values with $;.  This makes it
# easy to simulate function calls with tied hashes -- we just split on
# $; to recreate the argument list.
#
# 2005/12/01: We must ensure that time_format gets two arguments, since
# the XS version cannot handle variable argument lists.

use vars qw(%time %strftime %manip);
tie %time,     'Time::Format', sub { push @_, 'time' if @_ == 1;  goto &time_format};
tie %strftime, 'Time::Format', \&time_strftime;
tie %manip,    'Time::Format', \&time_manip;

sub TIEHASH
{
    my $class = shift;
    my $func  = shift || die "Bad call to $class\::TIEHASH";
    bless $func, $class;
}

sub FETCH
{
    my $self = shift;
    my $key  = shift;
    my @args = split $;, $key, -1;
    $self->(@args);
}

use subs qw(
 STORE    EXISTS    CLEAR    FIRSTKEY    NEXTKEY  );
*STORE = *EXISTS = *CLEAR = *FIRSTKEY = *NEXTKEY = sub
{
    my ($pkg,$file,$line) = caller;
    _croak "Invalid call to Time::Format internal function at $file line $line.";
};


# Module finder -- do we have the specified module available?
{
    my %have;
    sub _have
    {
        my $module = shift || return;
        return $have{$module}  if exists $have{$module};

        my $incmod = $module;
        $incmod =~ s!::!/!g;
        return $have{$module} = 1  if exists $INC{"$incmod.pm"};

        $@ = '';
        eval "require $module";
        return $have{$module} = $@? 0 : 1;
    }
}


# POSIX strftime, for people who like those weird % formats.
sub time_strftime
{
    # Check if POSIX is available  (why wouldn't it be?)
    return 'NO_POSIX' unless _have('POSIX');

    my $fmt = shift;
    my @time;

    # If more than one arg, assume they're doing the whole arg list
    if (@_ > 1)
    {
        @time = @_;
    }
    else    # use unix time (current or passed)
    {
        my $time = @_? shift : time;
        @time = localtime $time;
    }

    return POSIX::strftime($fmt, @time);
}


# Date::Manip interface
sub time_manip
{
    return "NO_DATEMANIP" unless _have('Date::Manip');

    my $fmt  = shift;
    my $time = @_? shift : 'now';

    $time = $1 if $time =~ /^\s* (epoch \s+ \d+)/x;

    return Date::Manip::UnixDate($time, $fmt);
}


1;
__DATA__
# The following is only compiled if Time::Format_XS is not available.
#line 248 "Time/Format.pm"

use Time::Local;

# Default names for months, days
my %english_names =
(
 Month    => [qw[January February March April May June July August September October November December]],
 Weekday  => [qw[Sunday Monday Tuesday Wednesday Thursday Friday Saturday]],
 th       => [qw[/th st nd rd th th th th th th th th th th th th th th th th th st nd rd th th th th th th th st]],
);
my %names;
my $locale;
my %loc_cache;                # Cache for remembering times that have already been parsed out.
my $cache_size=0;             # Number of keys in %loc_cache
my $cache_size_limit = 1024;  # Max number of times to cache

# Internal function to initialize locale info.
# Returns true if the locale changed.
sub setup_locale
{
    # Do nothing if locale has not changed since %names was set up.
    my $locale_in_use;
    $locale_in_use = POSIX::setlocale(POSIX::LC_TIME()) if _have('POSIX');
    $locale_in_use = '' if  !defined $locale_in_use;
    return if defined $locale  &&  $locale eq $locale_in_use;

    my (@Month, @Mon, @Weekday, @Day);

    unless (eval {
        require I18N::Langinfo;
        I18N::Langinfo->import(qw(langinfo));
        @Month = map langinfo($_),   I18N::Langinfo::MON_1(),    I18N::Langinfo::MON_2(),    I18N::Langinfo::MON_3(),
                                     I18N::Langinfo::MON_4(),    I18N::Langinfo::MON_5(),    I18N::Langinfo::MON_6(),
                                     I18N::Langinfo::MON_7(),    I18N::Langinfo::MON_8(),    I18N::Langinfo::MON_9(),
                                     I18N::Langinfo::MON_10(),   I18N::Langinfo::MON_11(),   I18N::Langinfo::MON_12();
        @Mon   = map langinfo($_),   I18N::Langinfo::ABMON_1(),  I18N::Langinfo::ABMON_2(),  I18N::Langinfo::ABMON_3(),
                                     I18N::Langinfo::ABMON_4(),  I18N::Langinfo::ABMON_5(),  I18N::Langinfo::ABMON_6(),
                                     I18N::Langinfo::ABMON_7(),  I18N::Langinfo::ABMON_8(),  I18N::Langinfo::ABMON_9(),
                                     I18N::Langinfo::ABMON_10(), I18N::Langinfo::ABMON_11(), I18N::Langinfo::ABMON_12();
        @Weekday = map langinfo($_), I18N::Langinfo::DAY_1(),    I18N::Langinfo::DAY_2(),    I18N::Langinfo::DAY_3(),
            I18N::Langinfo::DAY_4(), I18N::Langinfo::DAY_5(),    I18N::Langinfo::DAY_6(),    I18N::Langinfo::DAY_7();
        @Day     = map langinfo($_), I18N::Langinfo::ABDAY_1(),  I18N::Langinfo::ABDAY_2(),  I18N::Langinfo::ABDAY_3(),
          I18N::Langinfo::ABDAY_4(), I18N::Langinfo::ABDAY_5(),  I18N::Langinfo::ABDAY_6(),  I18N::Langinfo::ABDAY_7();
        1;
        }
           )
    {    # Internationalization didn't work for some reason; go with English.
        @Month   = @{ $english_names{Month} };
        @Weekday = @{ $english_names{Weekday} };
        @Mon     = map substr($_,0,3), @Month;
        @Day     = map substr($_,0,3), @Weekday;
        $@ = '';
    }

    # Store in %names, setting proper case
    $names{Month}   = \@Month;
    $names{Weekday} = \@Weekday;
    $names{Mon}     = \@Mon;
    $names{Day}     = \@Day;
    $names{th}      = $english_names{th};
    $names{TH}      = [map uc, @{$names{th}}];

    foreach my $name (keys %names)
    {
        my $aref = $names{$name};              # locale-native case
        $names{uc $name} = [map uc, @$aref];   # upper=case
        $names{lc $name} = [map lc, @$aref];   # lower-case
    }

    %loc_cache = ();          # locale changes are rare.  Clear out cache.
    $cache_size = 0;
    $locale = $locale_in_use;

    return 1;
}

# Types of time values we can handle:
my $NUMERIC_TIME      = \&decode_epoch;
my $DATETIME_OBJECT   = \&decode_DateTime_object;
my $DATETIME_STRING   = \&decode_DateTime_string;
# my $DATEMANIP_STRING  = \&decode_DateManip_string;

# What kind of argument was passed to time_format?
# Returns (type, time, cache_time_key, milliseconds, microseconds)
sub _classify_time
{
    my $timeval = shift;
    $timeval = 'time' if !defined $timeval;

    my $frac;  # Fractional seconds, if any
    my $cache_value;    # 1/20 of 1 cent
    my $time_type;

    # DateTime object?
    if (UNIVERSAL::isa($timeval, 'DateTime'))
    {
        $cache_value = "$timeval";    # stringify
        $frac        = $timeval->nanosecond() / 1e9;
        $time_type   = $DATETIME_OBJECT;
    }
    # Numeric time?
    # 1 to 11 digits--  Epoch time should be <= 10 digits, and 12 digits might be YYYYMMDDHHMM.
    elsif ($timeval =~ /^\s* (  (\d{1,11}) (?:[.,](\d+))?  )  $/x)
    {
        $timeval     = $1;
        $cache_value = $2;
        $frac        = $3? '0.' . $3 : 0;
        $time_type   = $NUMERIC_TIME;
    }
    # Stringified DateTime object?
    # Except we make it more flexible by allowing the date OR the time to be specfied
    # This will also match Date::Manip strings, and many ISO-8601 strings.
    elsif ($timeval =~ m{\A(   (?!\d{6,8}\z)                 # string must not consist of only 6 or 8 digits.
                          (?:                                # year-month-day
                                   \d{4}                     #     year
                            [-/.]? (?:0[1-9]|1[0-2])         #     month
                            [-/.]? (?:0[1-9]|[12]\d|3[01])   #     day
                          )?                                 # ymd is optional
                          (?:  (?<=\d)  [T_ ]  (?=\d) )?     # separator: T or _ or space, but only if ymd and hms both present
                        )                                    # End of $1: YMD and separator
                          (?:                                # hms is optional
                        (
                                  (?:[01]\d|2[0-4])          #     hour
                            [:.]? (?:[0-5]\d)                #     minute
                            [:.]? (?:[0-5]\d|6[0-1])?        #     second
                        )                                    # End of $2: HMS
                            (?: [,.] (\d+))?                 # optional fraction
                            (Z?)                             # optional "zulu" (UTC) designator
                          )?                                 # end of optional (HMS.fraction)
                        \z
                      }x)
    {
        $cache_value = ($1 || q{}) . ($2 || q{}) . ($4 || q{});
        $frac        = $3? '0.' . $3 : 0;
        $time_type   = $DATETIME_STRING;
    }
    # Not set, or set to 'time' string?
    elsif ($timeval eq 'time'  ||  $timeval eq q{})
    {
        # Get numeric time
        $timeval     = _have('Time::HiRes')? Time::HiRes::time() : time;
        $cache_value = int $timeval;
        $frac        = $timeval - $cache_value;
        $time_type   = $NUMERIC_TIME;
    }
    # *Tiny* numeric time (very close to zero; exponential notation)?
    # (See bug 87484, https://rt.cpan.org/Ticket/Display.html?id=87484)
    elsif ($timeval =~ /^\s* -? \d\.\d+ e-\d+  \s*$/x)
    {
        $timeval     = sprintf '%8.6f', abs($timeval);
        $cache_value = int $timeval;
        $frac        = $timeval - $cache_value;
        $time_type   = $NUMERIC_TIME;
    }
    else
    {
        # User passed us something we don't know how to handle.
        _croak qq{Unrecognized time value: "$timeval"};
    }
    # We messed up.
    die qq{Illegal time type "$time_type"; programming error in Time::Format. Contact author.}
        if !defined &$time_type;

    # Calculate millisecond, microsecond from fraction
    # msec and usec are TRUNCATED, not ROUNDED, because rounding up
    # to the next higher second would be a nightmare.
    my $msec = sprintf '%03d', int (    1_000 * $frac);
    my $usec = sprintf '%06d', int (1_000_000 * $frac);

    return ($time_type, $timeval, $cache_value, $msec, $usec);
}

# Helper function -- returns localtime() hashref
sub _loctime
{
    my ($decode, $time, $cachekey, $msec, $usec) = _classify_time(@_);
    my $locale_changed = setup_locale;

    # Cached, because I expect this'll be called on the same time values frequently.
    die "Programming error: undefined cache value. Contact Time::Format author."
        if !defined $cachekey;

    # If locale has changed, can't use the cached value.
    if (!$locale_changed  &&  exists $loc_cache{$cachekey})
    {
        my $h = $loc_cache{$cachekey};
        ($h->{mmm}, $h->{uuuuuu}) = ($msec, $usec);
        return $h;
    }

    # Hour-12, time zone, localtime parts, decoded from input
    my ($h12, $tz, @time_parts) = $decode->($time);

    # Populate a whole mess o' data elements
    my %th;
    my $m0 = $time_parts[4] - 1;   # zero-based month

    # NOTE: When adding new codes, be wary of adding any that interfere
    # with the user's ability to use the words "at", "on", or "of" literally.

    # year, hour(12), month, day, hour, minute, second, millisecond, microsecond, time zone
    @th{qw[yyyy H  m{on}  d  h  m{in}  s  mmm uuuuuu tz]} = (    $time_parts[5],      $h12, @time_parts[4,3,2,1,0], $msec, $usec, $tz);
    @th{qw[yy  HH mm{on} dd hh mm{in} ss]} = map $_<10?"0$_":$_, $time_parts[5]%100,  $h12, @time_parts[4,3,2,1,0];
    @th{qw[    ?H ?m{on} ?d ?h ?m{in} ?s]} = map $_<10?" $_":$_,                      $h12, @time_parts[4,3,2,1,0];

    # AM/PM
    my ($h,$d,$wx) = @time_parts[2,3,6];  # Day, weekday index
    my $a = $h<12? 'a' : 'p';
    $th{am}     = $th{pm}     = $a . 'm';
    $th{'a.m.'} = $th{'p.m.'} = $a . '.m.';
    @th{qw/AM PM A.M. P.M./} = map uc, @th{qw/am pm a.m. p.m./};

    $th{$_} = $names{$_}[$wx] for qw/Weekday WEEKDAY weekday Day DAY day/;
    $th{$_} = $names{$_}[$m0] for qw/Month   MONTH   month   Mon MON mon/;
    $th{$_} = $names{$_}[$d]  for qw/th TH/;

    # Don't let the time cache grow boundlessly.
    if (++$cache_size == $cache_size_limit)
    {
        $cache_size = 0;
        %loc_cache = ();
    }
    return $loc_cache{$cachekey} = \%th;
}

sub decode_DateTime_object
{
    my $dt = shift;

    my @t = ($dt->hour_12, $dt->time_zone_short_name,
             $dt->second,  $dt->minute, $dt->hour,
             $dt->day,     $dt->month,  $dt->year,
             $dt->dow,     $dt->doy,    $dt->is_dst);
    $t[-3] = 0 if $t[-3] == 7;   # Convert 1-7 (Mon-Sun) to 0-6 (Sun-Sat).

    return @t;
}

# 2005-10-31T15:14:39
sub decode_DateTime_string
{
    my $dts = shift;
    unless ($dts =~ m{\A (?!>\d{6,8}\z)                        # string must not consist of only 6 or 8 digits.
                        (?:
                         (\d{4}) [-/.]? (\d{2}) [-/.]? (\d{2}) # year-month-day
                        )?                                     # ymd is optional, but next must not be digit
                      (?:  (?<=\d)  [T_ ]  (?=\d) )?           # separator: T or _ or space, but only if ymd and hms both present
                      (?:                                      # hms is optional
                         (\d{2}) [:.]? (\d{2}) [:.]? (\d{2})   # hour:minute:second
                         (?: [,.] \d+)?                        # optional fraction (ignored in this sub)
                         (Z?)                                  # optional "zulu" (UTC) indicator
                      )?   \z
                     }x)
    {
        # This "should" never happen, since we checked the format of
        # the string already.
        die qq{Unrecognized DateTime string "$dts": probable Time::Format bug};
    }

    my ($y,$mon,$d,$h,$min,$s,$tz) = ($1,$2,$3,$4,$5,$6,$7);
    my ($d_only, $t_only);
    my ($h12, $is_dst, $dow);
    if (!defined $y)
    {
        # Time only.  Set date to 1969-12-31.
        $y   = 1969;
        $mon =   12;
        $d   =   31;
        $h12  = $h == 0? 12
              : $h > 12? $h - 12
              :          $h;
        $is_dst = 0;   # (it's the dead of winter!)
        $dow = 3;      # 12/31/1969 is Wednesday.
        $t_only = 1;
    }
    if (!defined $h)
    {
        $h   =    0;
        $min =    0;
        $s   =    0;
        $d_only = 1;
    }

    if (!$t_only)
    {
        $h12  = $h == 0? 12
              : $h > 12? $h - 12
              :          $h;

        # DST?
        # If year is before 1970, use current year.
        my $tmp_year = $y > 1969? $y : (localtime)[5]+1900;
        my $ttime = timelocal(0, 0, 0, $d, $mon-1, $tmp_year);
        my @t = localtime $ttime;
        $is_dst = $t[8];
        $dow = _dow($y, $mon, $d);
    }

    # +0 is to force numeric (remove leading zeroes)
    my @t = map {$_+0} ($s,$min,$h,$d,$mon,$y);
    $h12 += 0;

    if ($tz  &&  $tz eq 'Z')
    {
        $tz = 'UTC';
    }
    elsif (_have('POSIX'))
    {
        $tz = POSIX::strftime('%Z', @t, $dow, -1, $is_dst);
    }

    return ($h12, $tz, @t, $dow, -1, $is_dst);
}

sub decode_epoch
{
    my $time = shift;   # Assumed to be an epoch time integer

    my @t = localtime $time;
    my $tz = _have('POSIX')? POSIX::strftime('%Z', @t) : '';
    my $h = $t[2];    # Hour (24), Month index
    $t[4]++;
    $t[5] += 1900;
    my $h12 = $h>12? $h-12 : ($h || 12);

    return ($h12, $tz, @t);
}

#  $int = dow ($year, $month, $day);
#
# Returns the day of the week (0=Sunday .. 6=Saturday).  Uses Zeller's
# congruence, so it isn't subject to the unix 2038 limitation.
#
#--->     $int = dow ($year, $month, $day);
sub _dow
{
    my ($Y, $M, $D) = @_;

    $M -= 2;
    if ($M < 1)
    {
        $M += 12;
        $Y--;
    }
    my $C = int($Y/100);
    $Y %= 100;

    return (int((26*$M - 2)/10) + $D + $Y + int($Y/4) + int($C/4) - 2*$C) % 7;
}


# The heart of the module.  Didja ever see so many wicked regexes in a row?

my %disam;    # Disambiguator for 'm' format.
$disam{$_} = "{on}" foreach qw/yy d dd ?d/;                # If year or day is nearby, it's 'month'
$disam{$_} = "{in}" foreach qw/h hh ?h H HH ?H s ss ?s/;   # If hour or second is nearby, it's 'minute'
sub time_format_perlonly
{
    my $fmt  = shift;
    my $time = _loctime(@_);

    # Remove \Q...\E sequences
    my $rc;
    if (index($fmt, '\Q') >= 0)
    {
        $rc = init_store($fmt);
        $fmt =~ s/\\Q(.*?)(?:\\E|$)/remember($1)/seg;
    }

    # "Guess" how to interpret ambiguous 'm'
    $fmt =~ s/
              (?<!\\)          # Must not follow a backslash
              (?=[ydhH])       # Must start with one of these
              (                # $1 begins
                (              # $2 begins.  Capture:
                    yy         #     a year
                  | [dhH]      #     a day or hour
                )
              [^?m\\]?         # Followed by something that's not part of a month
              )
              (?![?m]?m\{[io]n\})   # make sure it's not already unambiguous
              (?!mon)          # don't confuse "mon" with "m" "on"
              ([?m]?m)         # $3 is a month code
             /$1$3$disam{$2}/gx;

    # Ambiguous 'm', part 2.
    $fmt =~ s/(?<!\\)         # ignore things that begin with backslash
              ([?m]?m)        # $1 is a month code
              (               # $2 begins.
                 [^\\]?       #     0 or 1 characters
                 (?=[?dsy])   #     Next char must be one of these
                 (            #     $3 begins.  Capture:
                    \??[ds]   #         a day or a second
                  | yy        #         or a year
                 )
              )/$1$disam{$3}$2/gx;

    # The Big Date/Time Pattern of Doom
    $fmt =~ s/
              (?<!\\)                      # Don't expand something preceded by backslash
              (?=[dDy?hHsaApPMmWwutT])     # Jump to one of these characters
              (
                 [Dd]ay|DAY                # Weekday abbreviation
              |  yy(?:yy)?                 # Year
              |  [?m]?m\{[oi]n\}           # Unambiguous month-minute codes
              |  th | TH                   # day suffix
              |  [?d]?d                    # Day
              |  [?h]?h                    # Hour (24)
              |  [?H]?H                    # Hour (12)
              |  [?s]?s                    # Second
              |  [apAP]\.?[mM]\.?          # am and pm strings
              |  [Mm]on(?:th)?|MON(?:TH)?  # Month names and abbrev
              |  [Ww]eekday|WEEKDAY        # Weekday names
              |  mmm|uuuuuu                # millisecond and microsecond
              |  tz                        # time zone
             )/$time->{$1}/gx;

    # Simulate \U \L \u \l
    $fmt =~ s/((?:\\[UL])+)((?:\\[ul])+)/$2$1/g;
    $fmt =~ s/\\U(.*?)(?=\\[EULul]|$)/\U$1/gs;
    $fmt =~ s/\\L(.*?)(?=\\[EULul]|$)/\L$1/gs;
    $fmt =~ s/\\l(.)/\l$1/gs;
    $fmt =~ s/\\u(.)/\u$1/gs;
    $fmt =~ s/\\E//g;

    $fmt =~ tr/\\//d;    # Remove extraneous backslashes.

    if (defined $rc)    # Fixup \Q \E regions.
    {
        $fmt =~ s/$rc(..)/recall($1)/seg;
    }
    return $fmt;
}

# Code for remembering/restoring \Q...\E regions.
# init_store finds a sigil character that's not used within the format string.
# remember stores a string in the next slot in @store, and returns a coded replacement.
# recall looks up and returns a string from @store.
{
    my $rcode;
    my @store;
    my $stx;

    sub init_store
    {
        my $str = shift;
        $stx = 0;
        return $rcode = "\x01" unless index($str,"\x01") >= 0;

        for ($rcode="\x02"; $rcode<"\xFF"; $rcode=chr(1+ord $rcode))
        {
            return $rcode unless index($str, $rcode) >= 0;
        }
        _croak "Time::Format cannot process string: no unique characters left.";
    }

    sub remember
    {
        my $enc;
        do    # Must not return a code that contains a backslash
        {
            $enc = pack 'S', $stx++;
        } while index($enc, '\\') >= 0;

        $store[$stx-1] = shift;
        return join '', map "\\$_", split //, "$rcode$enc";    # backslash-escape it!
    }

    sub recall
    {
        return $store[unpack 'S', shift];
    }
}

__END__

=head1 SYNOPSIS

 use Time::Format qw(%time %strftime %manip);

 $time{$format}
 $time{$format, $unixtime}

 print "Today is $time{'yyyy/mm/dd'}\n";
 print "Yesterday was $time{'yyyy/mm/dd', time-24*60*60}\n";
 print "The time is $time{'hh:mm:ss'}\n";
 print "Another time is $time{'H:mm am tz', $another_time}\n";
 print "Timestamp: $time{'yyyymmdd.hhmmss.mmm'}\n";

C<%time> also accepts Date::Manip strings and DateTime objects:

 $dm = Date::Manip::ParseDate('last monday');
 print "Last monday was $time{'Month d, yyyy', $dm}";
 $dt = DateTime->new (....);
 print "Here's another date: $time{'m/d/yy', $dt}";

It also accepts most ISO-8601 date/time strings:

 $t = '2005/10/31T17:11:09';   # date separator: / or - or .
 $t = '2005-10-31 17.11.09';   # in-between separator: T or _ or space
 $t = '20051031_171109';       # time separator: : or .
 $t = '20051031171109';        # separators may be omitted
 $t = '2005/10/31';            # date-only is okay
 $t = '17:11:09';              # time-only is okay
 # But not:
 $t = '20051031';              # date-only without separators
 $t = '171109';                # time-only without separators
 # ...because those look like epoch time numbers.

C<%strftime> works like POSIX's C<strftime>, if you like those C<%>-formats.

 $strftime{$format}
 $strftime{$format, $unixtime}
 $strftime{$format, $sec,$min,$hour, $mday,$mon,$year, $wday,$yday,$isdst}

 print "POSIXish: $strftime{'%A, %B %d, %Y', 0,0,0,12,11,95,2}\n";
 print "POSIXish: $strftime{'%A, %B %d, %Y', 1054866251}\n";
 print "POSIXish: $strftime{'%A, %B %d, %Y'}\n";       # current time

C<%manip> works like Date::Manip's C<UnixDate> function.

 $manip{$format};
 $manip{$format, $when};

 print "Date::Manip: $manip{'%m/%d/%Y'}\n";            # current time
 print "Date::Manip: $manip{'%m/%d/%Y','last Tuesday'}\n";

These can also be used as standalone functions:

 use Time::Format qw(time_format time_strftime time_manip);

 print "Today is ", time_format('yyyy/mm/dd', $some_time), "\n";
 print "POSIXish: ", time_strftime('%A %B %d, %Y',$some_time), "\n";
 print "Date::Manip: ", time_manip('%m/%d/%Y',$some_time), "\n";

=head1 DESCRIPTION

This module creates global pseudovariables which format dates and
times, according to formatting codes you pass to them in strings.

The C<%time> formatting codes are designed to be easy to remember and
use, and to take up just as many characters as the output time value
whenever possible.  For example, the four-digit year code is
"C<yyyy>", the three-letter month abbreviation is "C<Mon>".

The nice thing about having a variable-like interface instead
of function calls is that the values can be used inside of strings (as
well as outside of strings in ordinary expressions).  Dates are
frequently used within strings (log messages, output, data records,
etc.), so having the ability to interpolate them directly is handy.

Perl allows arbitrary expressions within curly braces of a hash, even
when that hash is being interpolated into a string.  This allows you
to do computations on the fly while formatting times and inserting
them into strings.  See the "yesterday" example above.

The format strings are designed with programmers in mind.  What do you
need most frequently?  4-digit year, month, day, 24-based hour,
minute, second -- usually with leading zeroes.  These six are the
easiest formats to use and remember in Time::Format: C<yyyy>, C<mm>,
C<dd>, C<hh>, C<mm>, C<ss>.  Variants on these formats follow a simple
and consistent formula.  This module is for everyone who is weary of
trying to remember I<strftime(3)>'s arcane codes, or of endlessly
writing C<$t[4]++; $t[5]+=1900> as you manually format times or dates.

Note that C<mm> (and related codes) are used both for months and
minutes.  This is a feature.  C<%time> resolves the ambiguity by
examining other nearby formatting codes.  If it's in the context of a
year or a day, "month" is assumed.  If in the context of an hour or a
second, "minute" is assumed.

The format strings are not meant to encompass every date/time need
ever conceived.  But how often do you need the day of the year
(strftime's C<%j>) or the week number (strftime's C<%W>)?

For capabilities that C<%time> does not provide, C<%strftime> provides
an interface to POSIX's C<strftime>, and C<%manip> provides an
interface to the Date::Manip module's C<UnixDate> function.

If the companion module L<Time::Format_XS> is also installed,
Time::Format will detect and use it.  This will result in a
significant speed increase for C<%time> and C<time_format>.

=head1 VARIABLES

=over 4

=item time

 $time{$format}
 $time{$format,$time_value};

Formats a unix time number (seconds since the epoch), DateTime object,
stringified DateTime, Date::Manip string, or ISO-8601 string,
according to the specified format.  If the time expression is omitted,
the current time is used.  The format string may contain any of the
following:

    yyyy       4-digit year
    yy         2-digit year

    m          1- or 2-digit month, 1-12
    mm         2-digit month, 01-12
    ?m         month with leading space if < 10

    Month      full month name, mixed-case
    MONTH      full month name, uppercase
    month      full month name, lowercase
    Mon        3-letter month abbreviation, mixed-case
    MON  mon   ditto, uppercase and lowercase versions

    d          day number, 1-31
    dd         day number, 01-31
    ?d         day with leading space if < 10
    th         day suffix (st, nd, rd, or th)
    TH         uppercase suffix

    Weekday    weekday name, mixed-case
    WEEKDAY    weekday name, uppercase
    weekday    weekday name, lowercase
    Day        3-letter weekday name, mixed-case
    DAY  day   ditto, uppercase and lowercase versions

    h          hour, 0-23
    hh         hour, 00-23
    ?h         hour, 0-23 with leading space if < 10

    H          hour, 1-12
    HH         hour, 01-12
    ?H         hour, 1-12 with leading space if < 10

    m          minute, 0-59
    mm         minute, 00-59
    ?m         minute, 0-59 with leading space if < 10

    s          second, 0-59
    ss         second, 00-59
    ?s         second, 0-59 with leading space if < 10
    mmm        millisecond, 000-999
    uuuuuu     microsecond, 000000-999999

    am   a.m.  The string "am" or "pm" (second form with periods)
    pm   p.m.  same as "am" or "a.m."
    AM   A.M.  same as "am" or "a.m." but uppercase
    PM   P.M.  same as "AM" or "A.M."

    tz         time zone abbreviation

Millisecond and microsecond require Time::HiRes, otherwise they'll
always be zero.  Timezone requires POSIX, otherwise it'll be the empty
string.  The second codes (C<s>, C<ss>, C<?s>) can be 60 or 61 in rare
circumstances (leap seconds, if your system supports such).

Anything in the format string other than the above patterns is left
intact.  Any character preceded by a backslash is left alone and
not used for any part of a format code.  See the L</QUOTING> section
for more details.

For the most part, each of the above formatting codes takes up as much
space as the output string it generates.  The exceptions are the codes
whose output is variable length: C<Weekday>, C<Month>, time zone, and
the single-character codes.

The mixed-case "Month", "Mon", "Weekday", and "Day" codes return the
name of the month or weekday in the preferred case representation for
the locale currently in effect.  Thus in an English-speaking locale,
the seventh month would be "July" (uppercase first letter, lowercase
rest); while in a French-speaking locale, it would be "juillet" (all
lowercase).  See the L</QUOTING> section for ways to control the case
of month/weekday names.

Note that the "C<mm>", "C<m>", and "C<?m>" formats are ambiguous.
C<%time> tries to guess whether you meant "month" or "minute" based on
nearby characters in the format string.  Thus, a format of
"C<yyyy/mm/dd hh:mm:ss>" is correctly parsed as "year month day, hour
minute second".  If C<%time> cannot determine whether you meant
"month" or "minute", it leaves the C<mm>, C<m>, or C<?m> untranslated.
To remove the ambiguity, you can use the following codes:

    m{on}        month, 1-12
    mm{on}       month, 01-12
    ?m{on}       month, 1-12 with leading space if < 10

    m{in}        minute, 0-59
    mm{in}       minute, 00-59
    ?m{in}       minute, 0-59 with leading space if < 10

In other words, append "C<{on}>" or "C<{in}>" to make "C<m>", "C<mm>",
or "C<?m>" unambiguous.

=item strftime

 $strftime{$format, $sec,$min,$hour, $mday,$mon,$year, $wday,$yday,$isdst}
 $strftime{$format, $unixtime}
 $strftime{$format}

For those who prefer L<strftime|POSIX/strftime>'s weird % formats, or
who need POSIX compliance, or who need week numbers or other features
C<%time> does not provide.

=item manip

 $manip{$format};
 $manip{$format,$when};

Provides an interface to the Date::Manip module's C<UnixDate>
function.  This function is rather slow, but can parse a very wide
variety of date input.  See the L<Date::Manip> module for details
about the inputs accepted.

If you want to use the C<%time> codes, but need the input flexibility
of C<%manip>, you can use Date::Manip's C<ParseDate> function:

 print "$time{'yyyymmdd', ParseDate('last sunday')}";

=back

=head1 FUNCTIONS

=over 4

=item time_format

 time_format($format);
 time_format($format, $unix_time);

This is a function interface to C<%time>.  It accepts the same
formatting codes and everything.  This is provided for people who want
their function calls to I<look> like function calls, not hashes. :-)
The following two are equivalent:

 $x = $time{'yyyy/mm/dd'};
 $x = time_format('yyyy/mm/dd');

=item time_strftime

 time_strftime($format, $sec,$min,$hour, $mday,$mon,$year, $wday,$yday,$isdst);
 time_strftime($format, $unixtime);
 time_strftime($format);

This is a function interface to C<%strftime>.  It simply calls
POSIX::C<strftime>, but it does provide a bit of an advantage over
calling C<strftime> directly, in that you can pass the time as a unix
time (seconds since the epoch), or omit it in order to get the current
time.

=item time_manip

 manip($format);
 manip($format,$when);

This is a function interface to C<%manip>.  It calls
Date::Manip::C<UnixDate> under the hood.  It does not provide much of
an advantage over calling C<UnixDate> directly, except that you can
omit the C<$when> parameter in order to get the current time.

=back

=head1 QUOTING

This section applies to the format strings used by C<%time> and
C<time_format> only.

Sometimes it is necessary to suppress expansion of some format
characters in a format string.  For example:

    $time{'Hour: hh; Minute: mm{in}; Second: ss'};

In the above expression, the "H" in "Hour" would be expanded,
as would the "d" in "Second".  The result would be something like:

    8our: 08; Minute: 10; Secon17: 30

It would not be a good solution to break the above statement out
into three calls to %time:

    "Hour: $time{hh}; Minute: $time{'mm{in}'}; Second: $time{ss}"

because the time could change from one call to the next, which would
be a problem when the numbers roll over (for example, a split second
after 7:59:59).

For this reason, you can escape individual format codes with a
backslash:

    $time{'\Hour: hh; Minute: mm{in}; Secon\d: ss'};

Note that with double-quoted (and qq//) strings, the backslash must be
doubled, because Perl first interpolates the string:

    $time{"\\Hour: hh; Minute: mm{in}; Secon\\d: ss"};

For added convenience, Time::Format simulates Perl's built-in \Q and
\E inline quoting operators.  Anything in a string between a \Q and \E
will not be interpolated as any part of any formatting code:

    $time{'\QHour:\E hh; \QMinute:\E mm{in}; \QSecond:\E ss'};

Again, within interpolated strings, the backslash must be doubled, or
else Perl will interpret and remove the \Q...\E sequence before
Time::Format gets it:

    $time{"\\QHour:\\E hh; \\QMinute:\\E mm{in}; \\QSecond\\E: ss"};

Time::Format also recognizes and simulates the \U, \L, \u, and \l
sequences.  This is really only useful for finer control of the Month,
Mon, Weekday, and Day formats.  For example, in some locales, the
month names are all-lowercase by convention.  At the start of a
sentence, you may want to ensure that the first character is
uppercase:

    $time{'\uMonth \Qis the finest month of all.'};

Again, be sure to use \Q, and be sure to double the backslashes in
interpolated strings, otherwise you'll get something ugly like:

    July i37 ste fine37t july of all.

=head1 EXAMPLES

 $time{'Weekday Month d, yyyy'}   Thursday June 5, 2003
 $time{'Day Mon d, yyyy'}         Thu Jun 5, 2003
 $time{'dd/mm/yyyy'}              05/06/2003
 $time{yymmdd}                    030605
 $time{'yymmdd',time-86400}       030604
 $time{'dth of Month'}            5th of June

 $time{'H:mm:ss am'}              1:02:14 pm
 $time{'hh:mm:ss.uuuuuu'}         13:02:14.171447

 $time{'yyyy/mm{on}/dd hh:mm{in}:ss.mmm'}  2003/06/05 13:02:14.171
 $time{'yyyy/mm/dd hh:mm:ss.mmm'}          2003/06/05 13:02:14.171

 $time{"It's H:mm."}              It'14 1:02.    # OOPS!
 $time{"It'\\s H:mm."}            It's 1:02.     # Backslash fixes it.
                                                                               .
                                                                               .
 # Rename a file based on today's date:
 rename $file, "$file_$time{yyyymmdd}";

 # Rename a file based on its last-modify date:
 rename $file, "$file_$time{'yyyymmdd',(stat $file)[9]}";

 # stftime examples
 $strftime{'%A %B %d, %Y'}                 Thursday June 05, 2003
 $strftime{'%A %B %d, %Y',time+86400}      Friday June 06, 2003

 # manip examples
 $manip{'%m/%d/%Y'}                                   06/05/2003
 $manip{'%m/%d/%Y','yesterday'}                       06/04/2003
 $manip{'%m/%d/%Y','first monday in November 2000'}   11/06/2000

=head1 INTERNATIONALIZATION

If the I18N::Langinfo module is available, Time::Format will return
weekday and month names in the language appropriate for the current
locale.  If not, English names will be used.

Programmers in non-English locales may want to provide an alias to
C<%time> in their own preferred language.  This can be done by
assigning C<\%time> to a typeglob:

    # French
    use Time::Format;
    use vars '%temps';  *temps = \%time;
    print "C'est aujourd'hui le $temps{'d Month'}\n";

    # German
    use Time::Format;
    use vars '%zeit';   *zeit = \%time;
    print "Heutiger Tag ist $zeit{'d.m.yyyy'}\n";

=head1 EXPORTS

The following symbols are exported into your namespace by default:

 %time
 time_format

The following symbols are available for import into your namespace:

 %strftime
 %manip
 time_strftime
 time_manip

The C<:all> tag will import all of these into your namespace.
Example:

 use Time::Format ':all';

=head1 BUGS

The format string used by C<%time> must not have $; as a substring
anywhere.  $; (by default, ASCII character 28, or 1C hex) is used to
separate values passed to the tied hash, and thus Time::Format will
interpret your format string to be two or more arguments if it
contains $;.  The C<time_format> function does not have this
limitation.

=head1 REQUIREMENTS

 Time::Local
 I18N::Langinfo, if you want non-English locales to work.
 POSIX, if you choose to use %strftime or want the C<tz> format to work.
 Time::HiRes, if you want the C<mmm> and C<uuuuuu> time formats to work.
 Date::Manip, if you choose to use %manip.

 Time::Format_XS is optional but will make C<%time> and C<time_format>
     much faster.  The version of Time::Format_XS installed must match
     the version of Time::Format installed; otherwise Time::Format will
     not use it (and will issue a warning).

=head1 AUTHOR / COPYRIGHT

Copyright (c) 2003-2019 by Eric J. Roode, ROODE I<-at-> cpan I<-dot-> org

All rights reserved.

To avoid my spam filter, please include "Perl", "module", or this
module's name in the message's subject line, and/or GPG-sign your
message.

This module is copyrighted only to ensure proper attribution of
authorship and to ensure that it remains available to all.  This
module is free, open-source software.  This module may be freely used
for any purpose, commercial, public, or private, provided that proper
credit is given, and that no more-restrictive license is applied to
derivative (not dependent) works.

Substantial efforts have been made to ensure that this software meets
high quality standards; however, no guarantee can be made that there
are no undiscovered bugs, and no warranty is made as to suitability to
any given use, including merchantability.  Should this module cause
your house to burn down, your dog to collapse, your heart-lung machine
to fail, your spouse to desert you, or George Bush to be re-elected, I
can offer only my sincere sympathy and apologies, and promise to
endeavor to improve the software.


=begin gpg

-----BEGIN PGP SIGNATURE-----

iF0EARECAB0WIQTSmjxiQX/QfjsCVJLChJhzmpBWqgUCXTtSTQAKCRDChJhzmpBW
qqmjAKCO5fIeALvd5cOw6C+o+4nqamDxagCePM3JdKVIEYuSz9VOaTWiJGVbo54=
=drPW
-----END PGP SIGNATURE-----

=end gpg