NAME
Regexp::Common::time - Date and time regexps.
VERSION
This is version 0.01 of Regexp::Common::time, December 4, 2005.
SYNOPSIS
use Regexp::Common qw(time);
# Piecemeal, Time::Format-like patterns
$RE{time}{tf}{-pat => 'pattern'}
# Piecemeal, strftime-like patterns
$RE{time}{strftime}{-pat => 'pattern'}
# Match ISO8601-style date/time strings
$RE{time}{iso}
# Fuzzy date patterns
# YEAR/MONTH/DAY
$RE{time}{ymd} # Most flexible
$RE{time}{YMD} # Strictest (equivalent to y4m2d2)
# Other available patterns: y2md, y4md, y2m2d2, y4m2d2
# MONTH/DAY/YEAR (American style)
$RE{time}{mdy} # Most flexible
$RE{time}{MDY} # Strictest (equivalent to m2d2y4)
# Other available patterns: mdy2, mdy4, m2d2y2, m2d2y4
# DAY/MONTH/YEAR (European style)
$RE{time}{mdy} # Most flexible
$RE{time}{MDY} # Strictest (equivalent to d2m2y4)
# Other available patterns: dmy2, dmy4, d2m2y2, d2m2y4
# Fuzzy time pattern
# HOUR/MINUTE/SECOND
$RE{time}{hms} # H: matches 1 or 2 digits; 12 or 24 hours
# M: matches 2 digits.
# S: matches 2 digits; may be omitted
# May be followed by "a", "am", "p.m.", etc.DESCRIPTION
This module creates regular expressions that can be used for parsing dates and times. See Regexp::Common for a general description of how to use this interface.
Parsing dates is a dirty business. Dates are generally specified in one of three possible orders: year/month/day, month/day/year, and day/month/year. Years can be specified with four digits or with two digits (with assumptions made about the century). Months can be specified as one digit, two digits, as a spelled-out name, or as a three-letter abbreviation. Day numbers can be one digit or two digits, with limits depending on the month (and, in the case of February, even the year). Also, different people use different punctuation for separating the various elements.
A human can easily recognize that "October 21, 2005" and "21.10.05" refer to the same date, but it's tricky to get a program to come to the same conclusion. This module attempts to make it possible to do so, with a minimum of difficulty.
If you know the exact format of the data to be matched, use one of the specific, piecemeal pattern builders: tf or strftime. If there is some variability, use one of the fuzzy-matching patterns in the dmy, mdy, or ymd families. If the data are wildly variable, such as raw user input, give up and use the Date::Manip or Date::Parse module.
Time values are generally much simpler to parse than date values. Only one fuzzy pattern is provided, and it should suffice for most needs.
Time::Format PATTERNS
The Time::Format module uses simple, intuitive strings for specifying date and time formats. You can use these patterns here as well. See the Time::Format module for details about its format specifiers.
Example:
$str = 'Thu November 2, 2005';
$str =~ $RE{time}{tf}{-pat => 'Day Month d, yyyy'};The patterns can contain more complex regexp expressions as well:
$str =~ $RE{time}{tf}{-pat => '(Weekday|Day) (Month|Mon) d, yyyy'};Time zone matching (the tz format code) attempts to adhere to RFC2822 and ISO8601 as much as possible. The following time zones are matched:
Z
UT UTC
+hh:mm -hh:mm
+hhmm -hhmm
+hh -hh
GMT EST EDT CST CDT MST MDT PST PDTstrftime PATTERNS
The POSIX strftime function is a long-recognized standard for formatting dates and times. This module supports most of stftime's codes for matching; specifically, the aAbBcCDdeHIjmMnprRSTtuUVwWyxXYZ% codes. See above for how the %Z format matches time zones.
Also, this module provides the following nonstandard codes:
%_d 1- or 2-digit day number (1-31)
%_H 1- or 2-digit hour (0-23)
%_I 1- or 2-digit hour (1-12)
%_m 1- or 2-digit month number (1-12)
%_M 1- or 2-digit minute (0-59)
Example:
$str = 'Thu November 2, 2005';
$str =~ $RE{time}{strftime}{-pat => '%a %B %_d, %Y'};As above, the patterns can contain more complex regexp expressions as well:
$str =~ $RE{time}{strftime}{-pat => '(%A|%a)? (%B|%b) ?%_d, %Y'};ISO-8601 DATE/TIME MATCHING
The $RE{time}{iso} pattern will match many (most?) strings formatted as recommended by ISO-8601. The canonical ISO-8601 form is:
YYYY-MM-DDTHH:MM:SS(where "T" is a literal T character). The $RE{time}{iso} pattern will match this form, and some variants:
The date separator character may be a hyphen, slash (
/), period, or empty string (omitted). The two date separators must match.The time separator character may be a colon, a period, a space, or empty string (omitted). The two time separators must match.
The date-time separator may be a
T, an underscore, a space, or empty string (omitted).Either the date or the time may be omitted. But at least one must be there.
If the date is not omitted, all three of its components must be present.
If the time is not omitted, all three of its components must be present.
FUZZY PATTERN OVERVIEW
Fuzzy date patterns have the following properties in common:
The pattern names consist of the letters
y,m, andd, each optionally followed by a digit (2formandd;2or4fory).If a
yis followed by a 2 or a 4, it must match that many digits.If a
yhas no trailing digit, it can match either 2 or 4 digits, trying 4 first.If an
mis followed by a 2, then only two-digit matches for the month are considered, and month names are not matched.If an
mis not followed by a 2, then the month may be 1 or 2 digits, or a spelled-out name.Just like for months, if a
dis followed by a 2, then only two-digit matches for the day are considered.Just like for months, if a
dhas no trailing digit, then the day may be 1 or 2 digits, and a 1-digit match may not have any adjacent digits.The uppercase
DMY,MDY, andYMDpatterns are synonyms for the strictd2m2y4,m2d2y4, andy4m2d2patterns, respectively.If a one-digit match is considered for the month, then no adjacent digits are allowed. (e.g.: "1/23/45" in M/D/Y format has a valid one-digit month match, but "12345" does not. Nor does "91/23/45").
If a pattern begins with an digitless
d,m, ory, then, in the string to be matched, any leading digits will cause the pattern to fail. For example:"012/23/45"will not match$RE{time}{mdy}. However, it will match$RE{time}{m2d2y2}. If you specify an exact pattern by usingm2instead ofm, this module assumes you know what you're doing.Likewise, a pattern ending with a digitless
dorywill not match if there are trailing digits in the string.
FUZZY PATTERN DETAILS
Year-Month-Day order
- $RE{time}{ymd}
-
"05/4/2" =~ $RE{time}{ymd}; "2005-APR-02" =~ $RE{time}{ymd};This is the most flexible of the numeric-only year/month/day formats. It matches a date of the form "year/month/day", where the year may be 2 or 4 digits; the month may be 1 or 2 digits or a spelled-out name or name abbreviation, and the day may be 1 or 2 digits. The year/month/day elements may be directly adjacent to each other, or may be separated by a space, period, slash (
/), or hyphen. - $RE{time}{y4md}
-
"2005/4/2" =~ $RE{time}{y4md}; "2005 APR 02" =~ $RE{time}{y4md};This works as $RE{time}{ymd}, except that the year is restricted to be exactly 4 digits.
- $RE{time}{y4m2d2}
-
"2005/04/02" =~ $RE{time}{y4m2d2};This works as $RE{time}{ymd}, except that the year is restricted to be exactly 4 digits, and the month and day must be exactly 2 digits each.
- $RE{time}{y2md}
-
"05/4/2" =~ $RE{time}{y2md}; "05.APR.02" =~ $RE{time}{y2md};This works as $RE{time}{ymd}, except that the year is restricted to be exactly 2 digits.
- $RE{time}{y2m2d2}
-
"05/04/02" =~ $RE{time}{y2m2d2};This works as $RE{time}{ymd}, except that the year is restricted to be exactly 2 digits, and the month and day must be exactly 2 digits each.
- $RE{time}{YMD}
-
"2005/04/02" =~ $RE{time}{YMD};This is a shorthand for the "canonical" year/month/day format,
y4m2d2.
Month-Day-Year (American) order
- $RE{time}{mdy}
- $RE{time}{mdy4}
- $RE{time}{m2d2y4}
- $RE{time}{mdy2}
- $RE{time}{m2d2y2}
- $RE{time}{MDY}
-
These patterns function as the equivalent year/month/day patterns, above; the only difference is the order of the elements.
MDYis a synonym form2d2y4.
Day-Month-Year (European) order
- $RE{time}{dmy}
- $RE{time}{dmy4}
- $RE{time}{d2m2y4}
- $RE{time}{dmy2}
- $RE{time}{d2m2y2}
- $RE{time}{DMY}
-
These patterns function as the equivalent year/month/day patterns, above; the only difference is the order of the elements.
DMYis a synonym ford2m2y4.
Time pattern (Hour-minute-second)
- $RE{time}{hms}
-
"10:06:12a" =~ /$RE{time}{hms}/; "9:00 p.m." =~ /$RE{time}{hms}/;Matches a time value in a string.
The hour must be in the range 0 to 24. The minute and second values must be in the range 0 to 59, and must be two digits (i.e., they must have leading zeroes if less than 10).
The hour, minute, and second components may be separated by colons (
:), periods, or spaces.The "seconds" value may be omitted.
The time may be followed by an "am/pm" indicator; that is, one of the following values:
a am a.m. p pm p.m. A AM A.M. P PM P.M.There may be a space between the time and the am/pm indicator.
CAPTURES (-keep)
Under -keep, the tf and strftime patterns capture the entire match as $1, plus one capture variable for each format specifier. However, if your pattern contains any parentheses, tf and strftime will not capture anything additional beyond what you specify, -keep or not. In other words: if you use parentheses, you are responsible for all capturing.
The iso pattern captures:
$1 the entire match
$2 the year
$3 the month
$4 the day
$5 the hour
$6 the minute
$7 the second
The year, month, and day ($2, $3, and $4) will be undef if the matched string contains only a time value (e.g., "12:34:56"). The hour, minute, and second ($5, $6, and $7) will be undef if the matched string contains only a date value (e.g., "2005-01-23").
The fuzzy y/m/d patterns capture
$1 the entire match
$2 the year
$3 the month
$4 the day
The fuzzy m/d/y patterns capture
$1 the entire match
$2 the month
$3 the day
$4 the year
The fuzzy d/m/y patterns capture
$1 the entire match
$2 the day
$3 the month
$4 the year
The fuzzy h/m/s pattern captures
$1 the entire match
$2 the hour
$3 the minute
$4 the second (undef if omitted)
$5 the am/pm indicator (undef if omitted)
EXAMPLES
# Typical usage: parsing a data record.
#
$rec = "blah blah 2005/10/21 blah blarrrrrgh";
@date = $rec =~ m{^blah blah $RE{time}{YMD}{-keep}};
# or
@date = $rec =~ m{^blah blah $RE{time}{tf}{-pat=>'yyyy/mm/dd'}{-keep}};
# or
@date = $rec =~ m{^blah blah $RE{time}{strftime}{-pat=>'%Y/%m/%d'}{-keep}};
# Typical usage: parsing variable-format data.
#
use Time::Normalize;
$record = "10-SEP-2005";
# This block tries M-D-Y first, then D-M-Y, then Y-M-D
if ( ((undef,$m,$d,$y) = $record =~ /^$RE{time}{mdy}{-keep}/)
|| ((undef,$d,$m,$y) = $record =~ /^$RE{time}{dmy}{-keep}/)
|| ((undef,$y,$m,$d) = $record =~ /^$RE{time}{ymd}{-keep}/) )
{
eval {($year, $month, $day) = normalize_ymd($y, $m, $d)};
}
else # give up
{
... # handle bogus data (no match)
}
#
if ($@)
{
... # handle bogus data (erroneous)
}
#
# $day is now 10; $month is now 09; $year is now 2005.
# Time examples
$time = '9:10pm';
@time_data = $time =~ /$RE{time}{hms}{-keep}/;
# captures '9:10pm', '9', '10', undef, 'pm'
@time_data = $time =~ /$RE{time}{tf}{-pat => '(h):(mm)(:ss)?(am)?'}{-keep}/;
# captures '9', '10', undef, 'pm'EXPORTS
This module exports no symbols to the caller's namespace.
SEE ALSO
It's not enough that the date regexps can match various formats. You then have to parse those matched data values and translate them into useful values. The Time::Normalize module is highly recommended for performing this repetitive, error-prone task.
REQUIREMENTS
Requires Regexp::Common, of course.
Also requires POSIX.
If I18N::Langinfo is available, this module will use it; otherwise, it will use hardcoded English values for month and weekday names.
Test::More is required for the test suite.
AUTHOR / COPYRIGHT
Eric J. Roode, roode@cpan.org
Copyright (c) 2005 by Eric J. Roode. All Rights Reserved. This module is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
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.