Add human duration parser

xemlock · May 6, 2019 · 7f87930 · 7f87930
1 parent 5906a5a
commit 7f87930
Show file tree

Hide file tree

Showing 6 changed files with 99 additions and 31 deletions.
diff --git a/library/HTMLPurifier/AttrDef/HTML5/Datetime.php b/library/HTMLPurifier/AttrDef/HTML5/Datetime.php
@@ -9,8 +9,6 @@
  * be extracted from the input, or the extracted data is insufficient for
  * the desired format (with the exception of DatetimeGlobal format, which
  * uses server timezone offset if none is detected).
- *
- * @since 0.1.11
  */
 class HTMLPurifier_AttrDef_HTML5_Datetime extends HTMLPurifier_AttrDef
 {

diff --git a/library/HTMLPurifier/AttrDef/HTML5/Duration.php b/library/HTMLPurifier/AttrDef/HTML5/Duration.php
@@ -3,8 +3,6 @@
 /**
  * Validates HTML5 duration string according to spec
  * https://html.spec.whatwg.org/multipage/common-microsyntaxes.html#durations
- *
- * @since 0.1.11
  */
 class HTMLPurifier_AttrDef_HTML5_Duration extends HTMLPurifier_AttrDef
 {
@@ -20,9 +18,7 @@ class HTMLPurifier_AttrDef_HTML5_Duration extends HTMLPurifier_AttrDef
         )?
     $/xi';
 
-    const REGEX_HUMAN = '/^
-        (\s*\d+\s*[WDHMS]|\s*\d+\.\d+S)+
-    $/xi';
+    const REGEX_HUMAN = '/(\d+(\s*[WDHMS]|\.\d+\s*S))/i';
 
     /**
      * @param string $string
@@ -67,15 +63,15 @@ protected function validateISODuration($string)
             's' => 0,
         );
 
-        foreach ($parts as $part => $_) {
-            if (!isset($match[$part])) {
+        foreach ($parts as $unit => $_) {
+            if (!isset($match[$unit])) {
                 continue;
             }
 
-            $value = substr($match[$part], 0, -1);
-            $value = $part === 's' ? (float) $value : (int) $value;
+            $value = substr($match[$unit], 0, -1);
+            $value = $unit === 's' ? (float) $value : (int) $value;
 
-            $parts[$part] = $value;
+            $parts[$unit] = $value;
         }
 
         // The spec self-contradicts itself in disallowing weeks in ISO-8601
@@ -88,24 +84,15 @@ protected function validateISODuration($string)
         }
 
         $duration = 'P';
-
-        foreach ($parts as $part => $value) {
-            if ($part === 'h') {
+        foreach ($parts as $unit => $value) {
+            if ($unit === 'h') {
                 $duration .= 'T';
             }
 
             if ($value > 0) {
-                $duration .= (int) $value;
-                if ($part === 's') {
-                    $msec = round(($value - (int) $value) * 1000);
-                    if ($msec > 0) {
-                        $duration .= rtrim(sprintf('.%03d', $msec), '0');
-                    }
-                }
-                $duration .= strtoupper($part);
+                $duration .= ($unit === 's' ? $this->formatSeconds($value) : $value) . strtoupper($unit);
             }
         }
-
         $duration = rtrim($duration, 'T');
 
         // At least one element must be present, thus "P" is not a valid
@@ -122,11 +109,65 @@ protected function validateISODuration($string)
 
     /**
      * Validate human readable HTML5 duration string
+     *
      * @param string $string
      * @return boolean|string
      */
     protected function validateHumanDuration($string)
     {
-        return false;
+        if (!preg_match_all(self::REGEX_HUMAN, $string, $matches)) {
+            return false;
+        }
+
+        // One or more duration time components, each with a different duration
+        // time component scale, in any order.
+        $parts = array(
+            'w' => false,
+            'd' => false,
+            'h' => false,
+            'm' => false,
+            's' => false,
+        );
+
+        foreach ($matches[0] as $match) {
+            $unit = strtolower(substr($match, -1));
+
+            $value = rtrim(substr($match, 0, -1));
+            $value = $unit === 's' ? (float) $value : (int) $value;
+
+            if ($value > 0 && $parts[$unit] === false) {
+                $parts[$unit] = $value;
+            }
+        }
+
+        $duration = array();
+        foreach ($parts as $unit => $value) {
+            if ($value === false) {
+                continue;
+            }
+            $duration[] = ($unit === 's' ? $this->formatSeconds($value) : $value) . $unit;
+        }
+        $duration = implode(' ', $duration);
+
+        if ($duration === '') {
+            $duration = '0s';
+        }
+
+        return $duration;
+    }
+
+    /**
+     * Formats seconds without leading zero and at most 3 non-zero decimals
+     *
+     * @param float $sec
+     * @return string
+     */
+    protected function formatSeconds($sec)
+    {
+        $msec = round(($sec - (int) $sec) * 1000);
+        if ($msec > 0) {
+            return rtrim(sprintf('%d.%03d', $sec, $msec), '0');
+        }
+        return sprintf('%d', $sec);
     }
 }
diff --git a/library/HTMLPurifier/AttrDef/HTML5/Week.php b/library/HTMLPurifier/AttrDef/HTML5/Week.php
@@ -3,8 +3,6 @@
 /**
  * Validates HTML5 week string according to
  * https://html.spec.whatwg.org/multipage/common-microsyntaxes.html#weeks
- *
- * @since 0.1.11
  */
 class HTMLPurifier_AttrDef_HTML5_Week extends HTMLPurifier_AttrDef
 {

diff --git a/library/HTMLPurifier/AttrDef/HTML5/YearlessDate.php b/library/HTMLPurifier/AttrDef/HTML5/YearlessDate.php
@@ -3,8 +3,6 @@
 /**
  * Validates HTML5 yearless date string according to the spec
  * https://html.spec.whatwg.org/multipage/common-microsyntaxes.html#yearless-dates
- *
- * @since 0.1.11
  */
 class HTMLPurifier_AttrDef_HTML5_YearlessDate extends HTMLPurifier_AttrDef
 {

diff --git a/library/HTMLPurifier/ChildDef/HTML5/Time.php b/library/HTMLPurifier/ChildDef/HTML5/Time.php
@@ -7,7 +7,6 @@
  * is present if text content of the children is not a valid datetime string.
  *
  * @see https://html.spec.whatwg.org/multipage/text-level-semantics.html#the-time-element
- * @since 0.1.11
  */
 class HTMLPurifier_ChildDef_HTML5_Time extends HTMLPurifier_ChildDef_HTML5
 {

diff --git a/tests/HTMLPurifier/AttrDef/HTML5/DurationTest.php b/tests/HTMLPurifier/AttrDef/HTML5/DurationTest.php
@@ -17,7 +17,7 @@ protected function setUp()
      * @param string $input
      * @param string $expected OPTIONAL
      */
-    public function testDuration($input, $expected = null)
+    public function testISODuration($input, $expected = null)
     {
         $this->assertValidate($input, $expected);
     }
@@ -33,13 +33,47 @@ public function durationData()
             array('PT4H18M3.010S', 'PT4H18M3.01S'),
             array('PT4H18M3.100S', 'PT4H18M3.1S'),
 
+            array('1w 2d 3m 4s'),
+            array('1w 2d 3m 4.0s', '1w 2d 3m 4s'),
+            array('1w 2d 3m 4.010s', '1w 2d 3m 4.01s'),
+            array('1w 2d 3m 4.100s', '1w 2d 3m 4.1s'),
+
+            array('1w 4s'),
+            array('1w'),
+            array('2d'),
+            array('3m'),
+            array('4s'),
+
+            // empty duration
+            array('0d', '0s'),
+            array('0.000s', '0s'),
+
+            // any order
+            array('4s 3m 2d 1w', '1w 2d 3m 4s'),
+
+            // seconds
+            array('4.100s', '4.1s'),
+            array('4.010s', '4.01s'),
+            array('4.001s', '4.001s'),
+
+            // case sensitivity
+            array('1W 2D 3M 4S', '1w 2d 3m 4s'),
+            array('1W 2d 3M 4s', '1w 2d 3m 4s'),
+
             // invalid but fixable
             array('PT', 'PT0S'),
             array('P1W', 'P7D'),
             array('P1W1D', 'P8D'),
             array('PT4H18M3.6666S', 'PT4H18M3.667S'),
             array('pt4H18M3s', 'PT4H18M3S'),
 
+            // duplicated components, only first is relevant
+            array('1w 2w 3h 4h', '1w 3h'),
+
+            // round extra decimals
+            array('4.0001s', '4s'),
+            array('4.6666s', '4.667s'),
+
             // invalid
             array('', false),
             array('2010-04-10', false),