ParserVcalendar.inc 18.8 KB
Newer Older
Robert Rollins's avatar
Robert Rollins committed
1 2 3 4 5 6 7 8
<?php
/**
 * @file
 * Defines a class that parses iCalcreator vcalendar objects into
 * Feeds-compatible data arrays.
 */

class ParserVcalendar {
9 10 11
  /**
   * Variables used for parsing.
   */
Robert Rollins's avatar
Robert Rollins committed
12 13
  protected $calendar;
  protected $source;
14
  protected $fetcherResult;
Robert Rollins's avatar
Robert Rollins committed
15 16 17
  protected $config;
  protected $timezones = array();
  protected $xtimezone;
18 19 20 21 22 23 24
  
  /**
   * Variables used for batch processing.
   */
  protected $totalComponents = 0;
  protected $lastComponentParsed = 0;
  
25 26 27
  /**
   * Constructor.
   */
Robert Rollins's avatar
Robert Rollins committed
28 29 30
  public function __construct($calendar, $source, $fetcher_result, $config) {
    $this->calendar = $calendar;
    $this->source = $source;
31
    $this->mapping_sources = feeds_importer($source->id)->parser->getMappingSources();
32
    $this->fetcherResult = $fetcher_result;
Robert Rollins's avatar
Robert Rollins committed
33 34
    $this->config = $config;
  }
35
  
Robert Rollins's avatar
Robert Rollins committed
36 37 38
  /**
   * Parses the vcalendar object into an array of event data arrays.
   *
39
   * @param int $offset
40
   *   Specifies which section of the feed to start parsing at.
41 42
   *
   * @param int $limit
43
   *   Specifies how many components to parse on this run.
44
   *
Robert Rollins's avatar
Robert Rollins committed
45
   * @return array
46
   *   An array of parsed event data keyed by our mapping source property keys.
Robert Rollins's avatar
Robert Rollins committed
47
   */
48
  public function parse($offset, $limit) {
Robert Rollins's avatar
Robert Rollins committed
49 50 51 52 53 54 55 56 57 58 59 60
    // Sometimes, the feed will set a timezone for every event in the calendar
    // using the non-standard X-WR-TIMEZONE property. Date iCal uses this
    // timezone only if the date property is not in UTC and has no TZID.
    $xtimezone = $this->calendar->getProperty('X-WR-TIMEZONE');
    if (!empty($xtimezone[1])) {
      // Allow modules to alter the timezone string before it gets converted
      // into a DateTimeZone.
      $context = array(
        'property_key' => NULL,
        'calendar_component' => NULL,
        'calendar' => $this->calendar,
        'feeeds_source' => $this->source,
61
        'feeds_fetcher_result' => $this->fetcherResult,
Robert Rollins's avatar
Robert Rollins committed
62 63 64 65 66 67 68 69 70
      );
      drupal_alter('date_ical_import_timezone', $xtimezone[1], $context);
      $this->xtimezone = $this->_tzid_to_datetimezone($xtimezone[1]);
    }

    // Collect the timezones into an array, for easier access.
    while ($component = $this->calendar->getComponent('VTIMEZONE')) {
      $this->timezones[] = $component;
    }
71
    
72 73 74 75 76 77 78 79
    // This content array is used by date_ical_import_component_alter() and
    // date_ical_import_parsed_data_alter().
    $context2 = array(
      'calendar' => $this->calendar,
      'source' => $this->source,
      'fetcher_result' => $this->fetcherResult,
    );
    
80 81
    // Collect each component, so we can batch them properly in the next loop.
    $raw_components = array();
82 83 84
    $types = array('VEVENT', 'VTODO', 'VJOURNAL', 'VFREEBUSY', 'VALARM');
    foreach ($types as $type) {
      while ($vcalendar_component = $this->calendar->getComponent($type)) {
Robert Rollins's avatar
Robert Rollins committed
85 86
        // Allow modules to alter the vcalendar component before we parse it
        // into a Feeds-compatible data array.
87
        drupal_alter('date_ical_import_component', $vcalendar_component, $context2);
88
        $raw_components[] = $vcalendar_component;
Robert Rollins's avatar
Robert Rollins committed
89 90
      }
    }
91 92 93 94 95 96 97 98 99 100
    
    // Store this for use by DateiCalFeedsParser's batch processing code.
    $this->totalComponents = count($raw_components);
    
    // Parse each raw component in the current batch into a Feeds-compatible
    // event data array.
    $events = array();
    $batch = array_slice($raw_components, $offset, $limit, TRUE);
    foreach ($batch as $ndx => $raw_component) {
      $parsed_component = array();
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138
      foreach ($this->mapping_sources as $property_key => $data) {
        $handler = NULL;
        if (isset($data['date_ical_parse_handler'])) {
          $handler = $data['date_ical_parse_handler'];
        }
        else {
          // This is not one of our sources, so if we don't recognize and
          // support it, we'll have to pass a warning to the user.
          if ($property_key == 'geofield') {
            $handler = 'parseGeofield';
          }
          else {
            // We can safely ignore certain sources.
            $known_unknowns = array(
              // "Black Source 1" is from Feeds Tamper.
              'Blank source 1',
            );
            if (!in_array($property_key, $known_unknowns)) {
              // Only warn the user if this mapping source is in use.
              foreach ($this->source->importer->processor->config['mappings'] as $mapping) {
                if ($mapping['source'] == $property_key) {
                  drupal_set_message(t('Date iCal does not recognize the "@name" Mapping Source, and must skip it.', array('@name' => $data['name'])), 'warning', FALSE);
                  break;
                }
              }
            }
          }
        }
        if ($handler) {
          $parsed_component[$property_key] = $this->$handler($property_key, $raw_component);
        }
        
        if ($property_key == 'geofield' && !empty($parsed_component['geofield'])) {
          // To make our data readable by geofield_feeds_combined_source(), we
          // need to put it into the format output by Simplepie 1.3.
          $parsed_component['location_latitude'] = array($parsed_component['geofield']['lat']);
          $parsed_component['location_longitude'] = array($parsed_component['geofield']['lon']);
        }
139
      }
140
      
141 142 143 144
      // Allow modules to alter the final parsed data before we send it to the
      // Feeds processor.
      drupal_alter('date_ical_import_post_parse', $parsed_component, $context2);
      
145 146 147
      $events[] = $parsed_component;
      // The indices of the original $raw_components array are preserved in
      // $batch, so using the $ndx value here lets us communicate our progress
148
      // through the full collection of components.
149 150 151
      $this->lastComponentParsed = $ndx;
    }
    
Robert Rollins's avatar
Robert Rollins committed
152 153
    return $events;
  }
154
  
155 156 157
  /**
   * Getter for the protected totalComponents property.
   */
158 159 160 161
  public function getTotalComponents() {
    return $this->totalComponents;
  }
  
162 163 164
  /**
   * Getter for the protected lastComponentParsed property.
   */
165 166 167 168
  public function getLastComponentParsed() {
    return $this->lastComponentParsed;
  }
  
Robert Rollins's avatar
Robert Rollins committed
169 170 171 172
  /**
   * Parses text fields.
   *
   * @return string
173
   *   The parsed text property.
Robert Rollins's avatar
Robert Rollins committed
174 175 176 177 178 179 180 181 182 183 184 185 186 187 188
   */
  public function parseTextProperty($property_key, $vcalendar_component) {
    $text = $vcalendar_component->getProperty($property_key);
    if ($text === FALSE) {
      if ($property_key == 'SUMMARY') {
        $uid = $vcalendar_component->getProperty('UID');
        throw new DateIcalParseException(t('The component with UID %uid is invalid because it has no SUMMARY (nodes require a title).', array('%uid' => $uid)));
      }
      // If the component doesn't have this property, return NULL.
      return NULL;
    }
    // Convert literal \n and \N into newline characters.
    $text = str_replace(array('\n', '\N'), "\n", $text);
    return $text;
  }
189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204
  
  /**
   * Parses GEO fields.
   *
   * @return array
   *   The latitude and longitude values, keyed by 'lat' and 'lon'.
   */
  public function parseGeofield($property_key, $vcalendar_component) {
    if (!empty($vcalendar_component->geo['value'])) {
      return array(
        'lat' => $vcalendar_component->geo['value']['latitude'],
        'lon' => $vcalendar_component->geo['value']['longitude'],
      );
    }
  }
  
Robert Rollins's avatar
Robert Rollins committed
205 206 207 208
  /**
   * Parses field parameters.
   *
   * @return string
209
   *   The parsed field parameter.
Robert Rollins's avatar
Robert Rollins committed
210 211 212 213 214 215 216 217
   */
  public function parsePropertyParameter($property_key, $vcalendar_component) {
    list($key, $attr) = explode(':', $property_key);
    $property = $vcalendar_component->getProperty($key, FALSE, TRUE);
    if ($property === FALSE) {
      // If the component doesn't have this property, return NULL.
      return NULL;
    }
218
    return isset($property['params'][$attr]) ? $property['params'][$attr] : '';
Robert Rollins's avatar
Robert Rollins committed
219
  }
220
  
Robert Rollins's avatar
Robert Rollins committed
221
  /**
222
   * Parses DATE-TIME and DATE fields.
Robert Rollins's avatar
Robert Rollins committed
223 224
   *
   * @return FeedsDateTime
225
   *   The parsed datetime object.
Robert Rollins's avatar
Robert Rollins committed
226 227 228 229 230 231 232
   */
  public function parseDateTimeProperty($property_key, $vcalendar_component) {
    $property = $vcalendar_component->getProperty($property_key, FALSE, TRUE);
    // Gather all the other date properties, so we can work with them later.
    $duration = $vcalendar_component->getProperty('DURATION', FALSE, TRUE);
    $dtstart = $vcalendar_component->getProperty('DTSTART', FALSE, TRUE);
    $uid = $vcalendar_component->getProperty('UID');
233
    
234
    // DATE-type properties are treated as All Day events which can span over
235 236 237 238 239 240 241
    // multiple days.
    // The Date module's All Day event handling was never finalized
    // (http://drupal.org/node/874322), which requires us to do some some
    // special coddling later.
    $is_all_day = (isset($property['params']['VALUE']) && $property['params']['VALUE'] == 'DATE');
    
    // Cover various conditions in which either DTSTART or DTEND are not set.
Robert Rollins's avatar
Robert Rollins committed
242
    if ($property === FALSE) {
243
      // When DTEND isn't defined, we may need to emulate it.
Robert Rollins's avatar
Robert Rollins committed
244
      if ($property_key == 'DTEND') {
245 246 247 248 249
        // Unset DTENDs need to emulate the DATE type from DTSTART.
        $is_all_day = (isset($dtstart['params']['VALUE']) && $dtstart['params']['VALUE'] == 'DATE');
        
        if ($duration !== FALSE) {
          // If a DURATION is defined, emulate DTEND as DTSTART + DURATION.
250
          $property = array(
251 252 253 254
            'value' => iCalUtilityFunctions::_duration2date($dtstart['value'], $duration['value']),
            'params' => $dtstart['params'],
          );
        }
255
        elseif ($is_all_day) {
256 257
          // If this is an all-day event with no end or duration, treat this
          // as a single-day event by emulating DTEND as 1 day after DTSTART.
258
          $property = $dtstart;
259
          $property['value'] = iCalUtilityFunctions::_duration2date($property['value'], array('day' => 1));
260 261 262 263 264
        }
        else {
          // This event has no end date.
          return NULL;
        }
Robert Rollins's avatar
Robert Rollins committed
265
      }
266
      elseif ($property_key == 'DTSTART') {
267
        // DTSTART can only be legally unset in non-VEVENT components.
Robert Rollins's avatar
Robert Rollins committed
268 269 270 271 272 273 274 275
        if ($vcalendar_component->objName == 'vevent') {
          throw new DateIcalParseException(t('Feed import failed! The VEVENT with UID %uid is invalid: it has no DTSTART.', array('%uid' => $uid)));
        }
        else {
          return NULL;
        }
      }
    }
276 277 278 279 280
    
    // When iCalcreator parses a UTC date (one that ends with Z) from an iCal
    // feed, it stores that 'Z' into the $property['value']['tz'] value.
    if (isset($property['value']['tz'])) {
      $property['params']['TZID'] = 'UTC';
Robert Rollins's avatar
Robert Rollins committed
281
    }
282 283
    
    if ($is_all_day) {
Robert Rollins's avatar
Robert Rollins committed
284 285
      if ($property_key == 'DTEND') {
        if ($dtstart === FALSE) {
286 287
          // This will almost certainly never happen, but the error message
          // would be incomprehensible without this check.
Robert Rollins's avatar
Robert Rollins committed
288 289 290 291
          throw new DateIcalParseException(t('Feed import failed! The event with UID %uid is invalid: it has a DTEND but no DTSTART!', array('%uid' => $uid)));
        }

        if (module_exists('date_all_day')) {
292 293 294
          // If the Date All Day module is installed, we need to rewind the
          // DTEND by one day, because of the problem with FeedsDateTime
          // mentioned below.
Robert Rollins's avatar
Robert Rollins committed
295
          $prev_day = iCalUtilityFunctions::_duration2date($property['value'], array('day' => -1));
296
          $property['value'] = $prev_day;
Robert Rollins's avatar
Robert Rollins committed
297 298
        }
      }
299
      
Robert Rollins's avatar
Robert Rollins committed
300 301
      // FeedsDateTime->setTimezone() ignores timezone changes made to dates
      // with no time element, which means we can't compensate for the Date
302 303 304
      // module's automatic timezone conversion when it writes to the DB. To
      // get around that, we must add 00:00:00 explicitly, even though this
      // causes other problems (see above and below).
Robert Rollins's avatar
Robert Rollins committed
305 306
      $date_string = sprintf('%d-%d-%d 00:00:00', $property['value']['year'], $property['value']['month'], $property['value']['day']);
      // Use the server's timezone rather than letting it default to UTC.
307
      // This will help ensure that the date value doesn't get messed up when
308
      // Date converts its timezone as the value is read from the database.
309 310 311 312 313 314
      // This is *essential* for All Day events, because Date stores them as
      // '2013-10-03 00:00:00' in the database, rather than doing the sensible
      // thing and storing them as '2013-10-03'.
      // NOTE TO MAINTAINERS:
      // This will not work properly if the site is configured to allow users
      // to set their own timezone. Unfortunately, there isn't anything that
Robert Rollins's avatar
Robert Rollins committed
315
      // Date iCal can do about that, as far as I can tell.
Robert Rollins's avatar
Robert Rollins committed
316 317 318 319 320
      $datetimezone = new DateTimeZone(date_default_timezone_get());
    }
    else {
      // This is a DATE-TIME property.
      $date_string = iCalUtilityFunctions::_format_date_time($property['value']);
321 322 323 324 325 326 327 328 329 330 331 332 333 334 335
      
      // Allow modules to alter the timezone string. This also allows for
      // setting a TZID when one was not originally set for this property.
      $tzid = isset($property['params']['TZID']) ? $property['params']['TZID'] : NULL;
      $context = array(
        'property_key' => $property_key,
        'calendar_component' => $vcalendar_component,
        'calendar' => $this->calendar,
        'feeeds_source' => $this->source,
        'feeds_fetcher_result' => $this->fetcherResult,
      );
      drupal_alter('date_ical_import_timezone', $tzid, $context);
      
      if (isset($tzid)) {
        $datetimezone = $this->_tzid_to_datetimezone($tzid);
Robert Rollins's avatar
Robert Rollins committed
336
      }
337
      elseif (isset($this->xtimezone)) {
Robert Rollins's avatar
Robert Rollins committed
338 339 340 341 342
        // No timezone was set on the parsed date property, so if a timezone
        // was detected for the entire iCal feed, use it.
        $datetimezone = $this->xtimezone;
      }
      else {
343 344
        $msg = t("No timezone was detected for one or more of the events in this feed, forcing Date iCal to use this server's timezone as a fallback.<br>
            To make timezone-less events use a different timezone, implement hook_date_ical_import_timezone_alter() in a custom module.");
345 346 347
        drupal_set_message($msg, 'status', FALSE);
        $this->source->log('parse', $msg, array(), WATCHDOG_NOTICE);
        $datetimezone = new DateTimeZone(date_default_timezone_get());
Robert Rollins's avatar
Robert Rollins committed
348 349
      }
    }
350
    
Robert Rollins's avatar
Robert Rollins committed
351 352
    return new FeedsDateTime($date_string, $datetimezone);
  }
353
  
Robert Rollins's avatar
Robert Rollins committed
354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375
  /**
   * Parses multi-value fields, like the CATEGORIES component.
   *
   * @return array
   *   An array of strings contaning the individual values.
   */
  public function parseMultivalueProperty($property_key, $vcalendar_component) {
    // Since we're not telling it to give us the params data, $property will
    // be either FALSE, a string, or an array of strings.
    $property = $vcalendar_component->getProperty($property_key);
    if (empty($property)) {
      // If this multi-value property is being mapped to a Taxonomy field,
      // Feeds will interpret anything besides empty array as an array of
      // empty values (e.g. array('')). This will create a term for that
      // empty value, rather than leaving the field blank.
      return array();
    }
    if (!is_array($property)) {
      $property = array($property);
    }
    return $property;
  }
376
  
Robert Rollins's avatar
Robert Rollins committed
377 378 379 380 381 382 383 384 385 386 387
  /**
   * Format RRULEs, which specify when and how often the event is repeated.
   *
   * @return string
   *   An RRULE string, with EXDATE and RDATE values separated by \n.
   *   This is to make the RRULE compatible with date_repeat_split_rrule().
   */
  public function parseRepeatProperty($property_key, $vcalendar_component) {
    if ($vcalendar_component->getProperty($property_key) === FALSE) {
      return NULL;
    }
388
    
Robert Rollins's avatar
Robert Rollins committed
389 390 391 392 393
    // Due to a few bugs and limitations with Date Repeat, we need to massage
    // the RRULE a bit.
    if (count($vcalendar_component->rrule) > 1) {
      drupal_set_message(t('The event with UID %uid has multiple RRULEs, but the Date Repeat module only supports one. Only the first RRULE in the event will be used.<br>
          If your events need to have a complex repeat pattern, using RDATEs should help.',
394 395 396 397 398
        array('%uid' => $vcalendar_component->getProperty('UID'))), 'warning'
      );
      // Date Repeat will get extremely confused if it's sent multiple RRULE
      // values, so we need to manually pare it down to only the first one.
      $vcalendar_component->rrule = array($vcalendar_component->rrule[0]);
Robert Rollins's avatar
Robert Rollins committed
399
    }
Robert Rollins's avatar
Robert Rollins committed
400
    foreach ($vcalendar_component->rrule as &$rrule_data) {
Robert Rollins's avatar
Robert Rollins committed
401 402 403 404
      // RRULEs must have an INTERVAL, or Date Repeat will throw errors.
      if (!isset($rrule_data['value']['INTERVAL'])) {
        $rrule_data['value']['INTERVAL'] = '1';
      }
405
      
Robert Rollins's avatar
Robert Rollins committed
406
      if (!isset($rrule_data['value']['COUNT']) && !isset($rrule_data['value']['UNTIL'])) {
407 408
        drupal_set_message(t("The event with UID %uid has an indefinitely repeating RRULE, which the Date Repeat module doesn't support.<br>
            As a workaround, Date iCal set the repeat count to @count. This value can be customized in the iCal parser settings.",
Robert Rollins's avatar
Robert Rollins committed
409 410 411 412 413
          array('%uid' => $vcalendar_component->getProperty('UID'), '@count' => $this->config['indefinite_count'])), 'warning'
        );
        $rrule_data['value']['COUNT'] = $this->config['indefinite_count'];
      }
    }
414
    
Robert Rollins's avatar
Robert Rollins committed
415 416
    $rrule = trim($vcalendar_component->createRrule());
    $rdate = trim($vcalendar_component->createRdate());
417 418 419
    $exrule = trim($vcalendar_component->createExrule());
    $exdate = trim($vcalendar_component->createExdate());
    return "$rrule|$rdate|$exrule|$exdate";
Robert Rollins's avatar
Robert Rollins committed
420 421
  }

422 423 424
  /**
   * Internal helper function for creating DateTimeZone objects.
   */
Robert Rollins's avatar
Robert Rollins committed
425 426 427 428 429
  protected function _tzid_to_datetimezone($tzid) {
    try {
      $datetimezone = new DateTimeZone($tzid);
    }
    catch (Exception $e) {
430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453
      // In case this is a Windows TZID, read the mapping file to try and
      // convert it to a real TZID.
      $zones = file_get_contents(drupal_get_path('module', 'date_ical') . '/libraries/windowsZones.json');
      $zones_assoc = json_decode($zones, TRUE);
      $windows_to_olson_map = array();
      foreach ($zones_assoc['supplemental']['windowsZones']['mapTimezones'] as $mapTimezone) {
        if ($mapTimezone['mapZone']['_other'] == $tzid) {
          // $mapTimezone['mapZone']['_type'] is space-separated TZIDs.
          $tzids = preg_split('/\s/', $mapTimezone['mapZone']['_type']);
          try {
            // They all have the same UTC offset, so for our purposes we can
            // just take the first one.
            return new DateTimeZone($tzids[0]);
          }
          catch (Exception $e) {
            // If this one also fails, we're out of luck, so just fall through
            // to the regular error report code.
            break;
          }
        }
      }
      
      $tz_wiki = l(t('here'), 'http://en.wikipedia.org/wiki/List_of_tz_database_time_zones#List');
      $help = l(t('README'), 'admin/help/date_ical', array('absolute' => TRUE));
Robert Rollins's avatar
Robert Rollins committed
454
      $msg = t(
455 456 457
          '"@tz" is not a valid timezone (see the TZ column !here), so Date iCal had to fall back to UTC (which is probably wrong!).<br>
          Please read the Date iCal !readme for instructions on how to fix this.',
          array('@tz' => $tzid, '!here' => $tz_wiki, '!readme' => $help)
Robert Rollins's avatar
Robert Rollins committed
458 459 460
      );
      $this->source->log('parse', $msg, array(), WATCHDOG_WARNING);
      drupal_set_message($msg, 'warning', FALSE);
461
      $datetimezone = new DateTimeZone('UTC');
Robert Rollins's avatar
Robert Rollins committed
462 463 464 465
    }
    return $datetimezone;
  }
}