Merge pull request #104 from seth-shaw-asu/issue-103

fix/add tests for validation and ISO 8601

Author: alxp

src/EDTFConverter.php

@@ -30,7 +30,7 @@ public static function datetimeIso8601Value($data) {
   }
 
   /**
-   * Converts an EDTF text field into an ISO 8601 timestamp string.
+   * Converts an EDTF text field into an ISO 8601 date.
    *
    * It assumes the earliest valid date for approximations and intervals.
    *

src/EDTFUtils.php

@@ -183,7 +183,7 @@ public static function validateDate($datetime_str, $strict = FALSE) {
     $msgs = [];
 
     if (strpos($datetime_str, 'T') > -1) {
-      list($date, $time) = explode('T', $datetime_str);
+      [$date, $time] = explode('T', $datetime_str);
     }
     else {
       $date = (string) $datetime_str;
@@ -216,22 +216,29 @@ public static function validateDate($datetime_str, $strict = FALSE) {
     elseif (strlen(ltrim($parsed_date[self::YEAR_BASE], '-')) > 4) {
       $msgs[] = "Years longer than 4 digits must be prefixed with a 'Y'.";
     }
-    elseif (strlen($parsed_date[self::YEAR_BASE]) < 4) {
-      $msgs[] = "Years must be at least 4 characters long.";
-    }
     $strict_pattern = 'Y';
 
     // Month.
     if (array_key_exists(self::MONTH, $parsed_date) && !empty($parsed_date[self::MONTH])) {
       // Valid month values?
       if (
-        // Month doesn't exist in mapping or does exist in mapping, but is > 12
-        // and there is a day part.
-        (!array_key_exists($parsed_date[self::MONTH], self::MONTHS_MAP) ||
-          (array_key_exists($parsed_date[self::MONTH], self::MONTHS_MAP) &&
-          array_key_exists(self::DAY, $parsed_date) &&
-          $parsed_date[self::MONTH] > 12)) &&
-          strpos($parsed_date[self::MONTH], 'X') === FALSE) {
+            // Month doesn't exist in mapping
+            // and isn't a valid unspecified month value.
+            (
+              !array_key_exists($parsed_date[self::MONTH], self::MONTHS_MAP) &&
+              (intval(str_replace('X', '1', $parsed_date[self::MONTH])) > 12)
+            ) ||
+            // Sub-year groupings with day values.
+            (
+              array_key_exists($parsed_date[self::MONTH], self::MONTHS_MAP) &&
+              array_key_exists(self::DAY, $parsed_date) &&
+              $parsed_date[self::MONTH] > 12
+            ) ||
+            // Unspecifed character comes before number.
+            (
+              preg_match('/X+\d/', $parsed_date[self::MONTH])
+            )
+        ) {
         $msgs[] = "Provided month value '" . $parsed_date[self::MONTH] . "' is not valid.";
       }
       $strict_pattern = 'Y-m';
@@ -327,8 +334,28 @@ public static function expandYear($year_full, $year_base, $year_exponent) {
    */
   public static function iso8601Value(string $edtf) {
 
+    if (count(self::validate($edtf)) > 0) {
+      return '';
+    }
+    // Sets.
+    if (strpos($edtf, '[') !== FALSE || strpos($edtf, '{') !== FALSE) {
+      // Use first in set.
+      $dates = preg_split('/(,|\.\.)/', trim($edtf, '{}[]'));
+      return self::iso8601Value(array_shift($dates));
+    }
+    // Intervals.
+    if (str_contains($edtf, '/')) {
+      $dates = explode('/', $edtf);
+      return self::iso8601Value(array_shift($dates));
+    }
+
     $date_time = explode('T', $edtf);
 
+    // Valid EDTF values with time portions are already ISO 8601 timestamps.
+    if (array_key_exists(1, $date_time) && !empty($date_time[1])) {
+      return $edtf;
+    }
+
     preg_match(EDTFUtils::DATE_PARSE_REGEX, $date_time[0], $parsed_date);
 
     $year = '';
@@ -358,17 +385,7 @@ public static function iso8601Value(string $edtf) {
       $day = str_replace('X', '0', $day);
     }
 
-    $formatted_date = implode('-', array_filter([$year, $month, $day]));
-
-    // Time.
-    if (array_key_exists(1, $date_time) && !empty($date_time[1])) {
-      $formatted_date .= 'T' . $date_time[1];
-    }
-    else {
-      $formatted_date .= 'T00:00:00';
-    }
-
-    return $formatted_date;
+    return implode('-', array_filter([$year, $month, $day]));
 
   }
 

tests/src/Kernel/EdtfUtilsTest.php

@@ -19,25 +19,26 @@ class EdtfUtilsTest extends KernelTestBase {
    *
    * @var array
    */
-  private $testCases = [
+  private $singleDateValidations = [
     '1900' => [],
     '1900-01' => [],
     '1900-01-02' => [],
     '190X' => [],
     '1900-XX' => [],
     '1900-91' => ['Provided month value \'91\' is not valid.'],
     '1900-91-01' => ['Provided month value \'91\' is not valid.'],
+    '1900-X1' => ['Provided month value \'X1\' is not valid.'],
     // No validation for months with X.
-    '1900-3X' => [],
+    '1900-3X' => ['Provided month value \'3X\' is not valid.'],
     // Month 31 without a day matches summer so it's valid.
     '1900-31' => [],
     '1900-31-01' => ['Provided month value \'31\' is not valid.'],
-    '190X-5X-8X' => [],
+    '190X-5X-8X' => ['Provided month value \'5X\' is not valid.'],
     '19000' => ['Years longer than 4 digits must be prefixed with a \'Y\'.'],
     'Y19000' => [],
     '190u' => ['Could not parse the date \'190u\'.'],
-    '190' => ['Years must be at least 4 characters long.'],
-    '190-99-52' => ['Years must be at least 4 characters long.',
+    '190' => [],
+    '190-99-52' => [
       'Provided month value \'99\' is not valid.',
       'Provided day value \'52\' is not valid.',
     ],
@@ -53,9 +54,47 @@ class EdtfUtilsTest extends KernelTestBase {
    * @covers ::validate
    */
   public function testEdtfValidate() {
-    foreach ($this->testCases as $input => $expected) {
+    foreach ($this->singleDateValidations as $input => $expected) {
       $this->assertEquals($expected, EDTFUtils::validate($input, FALSE, FALSE, FALSE));
     }
   }
 
+  /**
+   * @covers ::iso8601Value
+   */
+  public function testIso8601() {
+    // EDTF value and ISO 8601 Timestamp results.
+    // Empty values are invalid dates which return blank.
+    $tests = [
+      '1900' => '1900',
+      '1900-01' => '1900-01',
+      '1900-01-02' => '1900-01-02',
+      '190X' => '1900',
+      '1900-XX' => '1900-01',
+      '1900-91' => '',
+      '1900-91-01' => '',
+      '1900-3X' => '',
+      '1900-31' => '1900-03',
+      '190X-5X-8X' => '',
+      '19000' => '',
+      'Y19000' => '19000',
+      '190u' => '',
+      '190' => '190',
+      '190-99-52' => '',
+      '1900-01-02T' => '',
+      '1900-01-02T1:1:1' => '',
+      '1900-01-02T01:22:33' => '1900-01-02T01:22:33',
+      '1900-01-02T01:22:33Z' => '1900-01-02T01:22:33Z',
+      '1900-01-02T01:22:33+' => '',
+      '1900-01-02T01:22:33+05:00' => '1900-01-02T01:22:33+05:00',
+      // Intervals and Sets should return the earliest value.
+      '1900/2023' => '1900',
+      '[1900,2023]' => '1900',
+      '[1900,2023}' => '1900',
+    ];
+    foreach ($tests as $date => $iso) {
+      $this->assertEquals($iso, EDTFUtils::iso8601Value($date));
+    }
+  }
+
 }