anewt-developers team mailing list archive
-
anewt-developers team
-
Mailing list archive
-
Message #00291
[Branch ~uws/anewt/anewt.uws] Rev 1802: [core] Return early when parsing date strings
------------------------------------------------------------
revno: 1802
committer: Wouter Bolsterlee <uws@xxxxxxxxx>
branch nick: anewt
timestamp: Tue 2010-10-26 21:27:26 +0200
message:
[core] Return early when parsing date strings
AnewtDateTime::parse_string() applied various regexes to
test date and time formats, even after a match was already
found. This change makes the function return early when a
match was found. Also put the most frequently encountered
date formats on top to avoid useless regex matching in
common cases.
And while we're at it: some code cleanups, including the use
of proper variable names (clearer and a bit more verbose).
modified:
core/datetime.lib.php
core/datetime.test.php
--
lp:anewt
https://code.launchpad.net/~uws/anewt/anewt.uws
Your team Anewt developers is subscribed to branch lp:anewt.
To unsubscribe from this branch go to https://code.launchpad.net/~uws/anewt/anewt.uws/+edit-subscription
=== modified file 'core/datetime.lib.php'
--- core/datetime.lib.php 2010-10-26 19:09:10 +0000
+++ core/datetime.lib.php 2010-10-26 19:27:26 +0000
@@ -201,8 +201,8 @@
{
assert('is_int($timestamp);');
- list ($y, $m, $d, $h, $i, $s) = explode(' ', date('Y m d H i s', $timestamp));
- $d = new AnewtDateTimeAtom((int) $y, (int) $m, (int) $d, (int) $h, (int) $i, (int) $s);
+ list ($year, $month, $day, $hour, $minute, $second) = explode(' ', date('Y m d H i s', $timestamp));
+ $d = new AnewtDateTimeAtom($year, $month, $day, $hour, $minute, $second);
return $d;
}
@@ -225,130 +225,194 @@
if (strlen($date) == 0)
return null;
-
- /* Defaults */
-
- $y = false; $m = 1; $d = 1; // date
- $h = 0; $i = 0; $s = 0; // time
-
-
- /* Date only */
-
- // ISO 8601: 2005
+ $date_parsed = AnewtDateTime::real_parse_string($date);
+
+ if (!$date_parsed)
+ return null;
+
+ list ($year, $month, $day, $hour, $minute, $second) = $date_parsed;
+
+ if (!AnewtDateTime::is_valid_date_ymd($year, $month, $day))
+ return null;
+
+ if (!AnewtDateTime::is_valid_time_hms($hour, $minute, $second))
+ return null;
+
+ return new AnewtDateTimeAtom($year, $month, $day, $hour, $minute, $second);
+ }
+
+ /**
+ * Real implementation of parse_string.
+ *
+ * \param $date
+ */
+ static private function real_parse_string($date)
+ {
+ /* Default values */
+
+ $year = 0; $month = 1; $day = 1;
+ $hour = 0; $minute = 0; $second = 0;
+
+ /* The most common formats (just a guess) are put first, so that no time
+ * is wasted testing uncommon regular expressions */
+
+ /* SQL92 DATETIME, e.g. 2005-10-31 12:00:00 */
+ $pattern = sprintf(
+ '/^(%s)-(%s)-(%s) (%s):(%s):(%s)(\.\d+)?(%s)?$/',
+ DATETIME_RE_YEAR, DATETIME_RE_MONTH, DATETIME_RE_DAY,
+ DATETIME_RE_HOUR, DATETIME_RE_MINUTE, DATETIME_RE_SECOND,
+ DATETIME_RE_TIMEZONE);
+ if (preg_match($pattern, $date, $matches) === 1)
+ {
+ $year = $matches[1];
+ $month = $matches[3];
+ $day = $matches[5];
+ $hour = $matches[7];
+ $minute = $matches[9];
+ $second = $matches[11];
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* ISO 8601 (year, month, and day), e.g. 2005-10-31 */
+ $pattern = sprintf(
+ '/^(%s)-?(%s)-?(%s)$/',
+ DATETIME_RE_YEAR, DATETIME_RE_MONTH, DATETIME_RE_DAY);
+ if (preg_match($pattern, $date, $matches) === 1)
+ {
+ $year = $matches[1];
+ $month = $matches[3];
+ $day = $matches[5];
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* Hour, minute, second */
+ $pattern = sprintf(
+ '/^(%s):(%s):(%s)$/',
+ DATETIME_RE_HOUR, DATETIME_RE_MINUTE, DATETIME_RE_SECOND);
+ if (preg_match($pattern, $date, $matches) === 1)
+ {
+ list ($year, $month, $day) = explode('-', strftime('%Y-%m-%d'));
+ $hour = $matches[1];
+ $minute = $matches[3];
+ $second = $matches[5];
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* Full ISO 8601, e.g. 2005-10-31T12:00:00 (with and without T) */
+ $pattern = sprintf(
+ '/^(%s)-?(%s)-?(%s)[T ](%s):?(%s):?(%s)$/',
+ DATETIME_RE_YEAR, DATETIME_RE_MONTH, DATETIME_RE_DAY,
+ DATETIME_RE_HOUR, DATETIME_RE_MINUTE, DATETIME_RE_SECOND);
+ if (preg_match($pattern, $date, $matches) === 1)
+ {
+ $year = $matches[1];
+ $month = $matches[3];
+ $day = $matches[5];
+ $hour = $matches[7];
+ $minute = $matches[9];
+ $second = $matches[11];
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* Hour, minute (without seconds) */
+ $pattern = sprintf(
+ '/^(%s):(%s)$/',
+ DATETIME_RE_HOUR, DATETIME_RE_MINUTE);
+ if (preg_match($pattern, $date, $matches) === 1)
+ {
+ list ($year, $month, $day) = explode('-', strftime('%Y-%m-%d'));
+ $hour = $matches[1];
+ $minute = $matches[3];
+ $second = 0;
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* ISO 8601 (year only), e.g. 2005 */
$pattern = sprintf('/^(%s)$/', DATETIME_RE_YEAR);
if (preg_match($pattern, $date, $matches) === 1)
- $y = $matches[1];
-
-
- // ISO 8601: 2005-10
- $pattern = sprintf('/^(%s)-?(%s)$/', DATETIME_RE_YEAR,
- DATETIME_RE_MONTH);
+ {
+ $year = $matches[1];
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* ISO 8601 (year and month), e.g. 2005-10 */
+ $pattern = sprintf(
+ '/^(%s)-?(%s)$/',
+ DATETIME_RE_YEAR, DATETIME_RE_MONTH);
if (preg_match($pattern, $date, $matches))
{
- $y = $matches[1];
- $m = $matches[3];
- }
-
- // ISO 8601: 2005-10-31
- $pattern = sprintf('/^(%s)-?(%s)-?(%s)$/', DATETIME_RE_YEAR,
- DATETIME_RE_MONTH, DATETIME_RE_DAY);
- if (preg_match($pattern, $date, $matches) === 1)
- {
- $y = $matches[1];
- $m = $matches[3];
- $d = $matches[5];
- }
-
-
- /* Time only */
-
- /* Time without seconds */
- $pattern = sprintf('/^(%s):(%s)$/', DATETIME_RE_HOUR, DATETIME_RE_MINUTE);
- if (preg_match($pattern, $date, $matches) === 1)
- {
- list ($y, $m, $d) = explode('-', strftime('%Y-%m-%d'));
- $y = (int) $y;
- $m = (int) $m;
- $d = (int) $d;
- $h = $matches[1];
- $i = $matches[3];
- $s = 0;
- }
-
- /* Time with seconds */
- $pattern = sprintf('/^(%s):(%s):(%s)$/', DATETIME_RE_HOUR,
- DATETIME_RE_MINUTE, DATETIME_RE_SECOND);
- if (preg_match($pattern, $date, $matches) === 1)
- {
- list ($y, $m, $d) = explode('-', strftime('%Y-%m-%d'));
- $y = (int) $y;
- $m = (int) $m;
- $d = (int) $d;
- $h = $matches[1];
- $i = $matches[3];
- $s = $matches[5];
- }
-
-
- /* Date and time */
-
- // SQL92 DATETIME: 2005-10-31 12:00:00
- $pattern = sprintf('/^(%s)-(%s)-(%s) (%s):(%s):(%s)(\.\d+)?(%s)?$/', DATETIME_RE_YEAR,
- DATETIME_RE_MONTH, DATETIME_RE_DAY, DATETIME_RE_HOUR, DATETIME_RE_MINUTE,
- DATETIME_RE_SECOND, DATETIME_RE_TIMEZONE);
- if (preg_match($pattern, $date, $matches) === 1)
- {
- $y = $matches[1];
- $m = $matches[3];
- $d = $matches[5];
- $h = $matches[7];
- $i = $matches[9];
- $s = $matches[11];
- }
-
- // ISO 8601: 2005-10-31T12:00:00 (with and without T)
- $pattern = sprintf('/^(%s)-?(%s)-?(%s)[T ](%s):?(%s):?(%s)$/',
- DATETIME_RE_YEAR, DATETIME_RE_MONTH, DATETIME_RE_DAY,
- DATETIME_RE_HOUR, DATETIME_RE_MINUTE, DATETIME_RE_SECOND);
- if (preg_match($pattern, $date, $matches) === 1)
- {
- $y = $matches[1];
- $m = $matches[3];
- $d = $matches[5];
- $h = $matches[7];
- $i = $matches[9];
- $s = $matches[11];
- }
-
- // ISO 8601: 2005-10-31T12:00 (with and without T)
- $pattern = sprintf('/^(%s)-?(%s)-?(%s)[T ](%s):?(%s)$/',
- DATETIME_RE_YEAR, DATETIME_RE_MONTH, DATETIME_RE_DAY,
- DATETIME_RE_HOUR, DATETIME_RE_MINUTE);
- if (preg_match($pattern, $date, $matches) === 1)
- {
- $y = $matches[1];
- $m = $matches[3];
- $d = $matches[5];
- $h = $matches[7];
- $i = $matches[9];
- }
-
- // ISO 8601: 2005-123
- $pattern = sprintf('/^(%s)-?(%s)$/', DATETIME_RE_YEAR, DATETIME_RE_DAY_OF_YEAR);
- if (preg_match($pattern, $date, $matches) === 1)
- {
- $y = $matches[1];
+ $year = $matches[1];
+ $month = $matches[3];
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* Full ISO 8601 without seconds, e.g. 2005-10-31T12:00 (with and without T) */
+ $pattern = sprintf(
+ '/^(%s)-?(%s)-?(%s)[T ](%s):?(%s)$/',
+ DATETIME_RE_YEAR, DATETIME_RE_MONTH, DATETIME_RE_DAY,
+ DATETIME_RE_HOUR, DATETIME_RE_MINUTE);
+ if (preg_match($pattern, $date, $matches) === 1)
+ {
+ $year = $matches[1];
+ $month = $matches[3];
+ $day = $matches[5];
+ $hour = $matches[7];
+ $minute = $matches[9];
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* RFC 2822 format, e.g. Tue, 10 Jun 2003 04:00:00 GMT */
+ $pattern = sprintf(
+ '/^%s,? %s %s %s %s:%s:%s %s$/',
+ DATETIME_RE_DAY_NAMES_ABBR,
+ DATETIME_RE_DAY, DATETIME_RE_MONTH_NAMES_ABBR, DATETIME_RE_YEAR,
+ DATETIME_RE_HOUR, DATETIME_RE_MINUTE, DATETIME_RE_SECOND,
+ DATETIME_RE_TIMEZONE);
+ if (preg_match($pattern, $date, $matches) === 1)
+ {
+ $year = $matches[4];
+ $month_names_abbr = array(1 => 'Jan', 'Feb', 'Mar', 'Apr', 'May',
+ 'Jun', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec');
+ $month_names_abbr = array_flip($month_names_abbr);
+ $month = $month_names_abbr[$matches[3]];
+ $day = $matches[2];
+ $hour = $matches[5];
+ $minute = $matches[6];
+ $second = $matches[7];
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* ISO 8601 year and day-of-year, e.g. 2005-123 */
+ $pattern = sprintf(
+ '/^(%s)-?(%s)$/',
+ DATETIME_RE_YEAR, DATETIME_RE_DAY_OF_YEAR);
+ if (preg_match($pattern, $date, $matches) === 1)
+ {
+ $year = $matches[1];
$timestamp = mktime(0, 0, 0, 1, $matches[3], $matches[1]);
- $m = (int) strftime('%m', $timestamp);
- $d = (int) strftime('%d', $timestamp);
+ $month = strftime('%m', $timestamp);
+ $day = strftime('%d', $timestamp);
+ return array($year, $month, $day, $hour, $minute, $second);
}
- // ISO 8601: 2005-W02
+
+ /* ISO 8601 year and week number, e.g. 2005-W02 */
/* TODO: Dates specified by year, week and day. */
- $pattern = sprintf('/^(%s)-?W(%s)$/', DATETIME_RE_YEAR, DATETIME_RE_WEEK_OF_YEAR);
+ $pattern = sprintf(
+ '/^(%s)-?W(%s)$/',
+ DATETIME_RE_YEAR, DATETIME_RE_WEEK_OF_YEAR);
if (preg_match($pattern, $date, $matches) === 1)
{
- $y_not_final = (int) $matches[1];
+ $year_not_final = (int) $matches[1];
$w = (int) $matches[3];
/* Too bad, it seems strptime() cannot parse %V correctly, so we
@@ -365,7 +429,7 @@
/* Find the highest possible week number for the year... */
$last_day_of_the_last_week_of_the_year = 31;
do {
- $w_max = (int) strftime('%V', mktime(0, 0, 0, 12, $last_day_of_the_last_week_of_the_year, $y_not_final));
+ $w_max = (int) strftime('%V', mktime(0, 0, 0, 12, $last_day_of_the_last_week_of_the_year, $year_not_final));
$last_day_of_the_last_week_of_the_year--;
} while ($w_max == 1);
@@ -379,59 +443,39 @@
while ((int) strftime('%V', $ts) != $w)
$ts += 24 * 60 * 60;
- list ($y, $m, $d) = explode('-', strftime('%Y-%m-%d', $ts));
+ list ($year, $month, $day) = explode('-', strftime('%Y-%m-%d', $ts));
}
- }
-
-
- // MS SQL default format: Mon Jan 23 00:00:00 2006
- $pattern = sprintf('/^%s,? %s %s %s:?%s:?%s %s$/',
- DATETIME_RE_DAY_NAMES_ABBR, DATETIME_RE_MONTH_NAMES_ABBR,
- DATETIME_RE_DAY, DATETIME_RE_HOUR, DATETIME_RE_MINUTE,
- DATETIME_RE_SECOND, DATETIME_RE_YEAR);
- if (preg_match($pattern, $date, $matches) === 1)
- {
- $y = $matches[7];
- $month_names_abbr = array(1 => 'Jan', 'Feb', 'Mar', 'Apr', 'May',
- 'Jun', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec');
- $month_names_abbr = array_flip($month_names_abbr);
- $m = $month_names_abbr[$matches[2]];
- $d = $matches[3];
-
- $h = $matches[4];
- $i = $matches[5];
- $s = $matches[6];
- }
-
- // RFC 2822 format: Tue, 10 Jun 2003 04:00:00 GMT
- $pattern = sprintf('/^%s,? %s %s %s %s:%s:%s %s$/',
- DATETIME_RE_DAY_NAMES_ABBR, DATETIME_RE_DAY,
- DATETIME_RE_MONTH_NAMES_ABBR, DATETIME_RE_YEAR,
- DATETIME_RE_HOUR, DATETIME_RE_MINUTE,
- DATETIME_RE_SECOND, DATETIME_RE_TIMEZONE);
- if (preg_match($pattern, $date, $matches) === 1)
- {
- $y = $matches[4];
- $month_names_abbr = array(1 => 'Jan', 'Feb', 'Mar', 'Apr', 'May',
- 'Jun', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec');
- $month_names_abbr = array_flip($month_names_abbr);
- $m = $month_names_abbr[$matches[3]];
- $d = $matches[2];
-
- $h = $matches[5];
- $i = $matches[6];
- $s = $matches[7];
- }
-
- /* Parsing done */
-
- if (!AnewtDateTime::is_valid_date_ymd($y, $m, $d))
- return null;
-
- if (!AnewtDateTime::is_valid_time_hms($h, $i, $s))
- return null;
-
- return new AnewtDateTimeAtom($y, $m, $d, $h, $i, $s);
+
+ /* Abort if no year was extracted ( */
+ if (!$year)
+ return null;
+
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* MS SQL default format, e.g. Mon Jan 23 00:00:00 2006 */
+ $pattern = sprintf(
+ '/^%s,? %s %s %s:?%s:?%s %s$/',
+ DATETIME_RE_DAY_NAMES_ABBR, DATETIME_RE_MONTH_NAMES_ABBR, DATETIME_RE_DAY,
+ DATETIME_RE_HOUR, DATETIME_RE_MINUTE, DATETIME_RE_SECOND,
+ DATETIME_RE_YEAR);
+ if (preg_match($pattern, $date, $matches) === 1)
+ {
+ $year = $matches[7];
+ $month_names_abbr = array(1 => 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec');
+ $month_names_abbr = array_flip($month_names_abbr);
+ $month = $month_names_abbr[$matches[2]];
+ $day = $matches[3];
+ $hour = $matches[4];
+ $minute = $matches[5];
+ $second = $matches[6];
+ return array($year, $month, $day, $hour, $minute, $second);
+ }
+
+
+ /* Nothing matched */
+ return null;
}
/**
@@ -965,7 +1009,7 @@
*
* \return
* A string in yyyy-mm-dd hh:mm:ss format.
- *
+ *
* \see AnewtDateTime::sql_date
* \see AnewtDateTime::sql_time
*/
@@ -1011,7 +1055,7 @@
*
* \return
* A string in hh:mm:ss format.
- *
+ *
* \see AnewtDateTime::sql
* \see AnewtDateTime::sql_date
*/
=== modified file 'core/datetime.test.php'
--- core/datetime.test.php 2010-03-27 20:50:26 +0000
+++ core/datetime.test.php 2010-10-26 19:27:26 +0000
@@ -39,8 +39,6 @@
f('1984-060');
f('1984060');
-f('This one should fail');
-
f('1983-01-15T18:30:00');
f('Mon Jan 23 02:04:06 2006');
@@ -56,12 +54,13 @@
f('2010-W01');
f('2010-W02');
f('2009W53');
-echo 'This one should be empty:', NL;
-f('2010-W53');
f('2013-W01');
echo 'There should only be empty lines below:', NL;
+f('this-is-not-a-date');
f(null);
+f('2010-W53');
+f('2013-W62');
assert('AnewtDateTime::date(AnewtDateTime::parse("2008-12-09")) === "2008-12-09";');
