SlatePermutate Changeset - 2a7a5fca904a · protofusion repositories

Changeset - 2a7a5fca904a

Parent rev.

Child rev.

[Not reviewed]

default

0 1 0

Nathan Brink (binki) - 15 years ago 2011-03-17 18:18:27
ohnobinki@ohnopublishing.net

cedarville: When crawling, detect a semester's start and end dates based on certain course meetings which have date information.

1 file changed with 24 insertions and 8 deletions:

school.d/cedarville.crawl.inc

0 comments (0 inline, 0 general)

school.d/cedarville.crawl.inc

➞

Show inline comments

@@ @@ -83,24 +83,26 @@ function cedarville_crawl(array &$semest @@
       $semester_href = $department_a_dom->getAttribute('href');
       $semester_href_parts = split('_', $semester_href);
       $semester_name = $department_a_dom->textContent;
       if (stripos($semester_name, 'graduate') !== FALSE
 	  || strpos($semester_href, 'index') === FALSE)
 	/* cedarville has about 1 graduate course, lol */
 	continue;
       $semester_name_parts = split(' ', $semester_name);
       $semester_year = $semester_name_parts[0];
       $semester_season = strtolower($semester_name_parts[1]);
       $semester_min_date_start = 0;
       $semester_max_date_end = 0;
       $semester = new Semester($semester_year, $semester_season);
       school_crawl_logf($school_crawl_log, 6, "Crawling semester: %s.",
 			$semester_name);
   /*
    * We need two passes because the first department's code name is
    * not accessible available in the first pageload.
    */
   $departments = array();
   if (cedarville_crawl_departments_get($basepath . $semester_href, $departments, $semester_href_parts[0], $school_crawl_log))
@@ @@ -194,57 +196,71 @@ function cedarville_crawl(array &$semest @@
 	   * different lab meetings and stuff...
 	   */
 	  $meetings_str = $course_table[6];
 	  if (strpos($meetings_str, 'TBA') !== FALSE)
+	    {
 	      school_crawl_logf($school_crawl_log, 8, "Skipping %s because its meeting time info has `TBA' in it.", implode('-', $section_parts));
 	      continue;
+	    }
 	  $meetings = array();
 	  $meeting_multiple_types = array();
 	  while (strlen($meetings_str) > 5)
+	    {
 	      if (!preg_match(';^ ([A-Z]+) +([A-Z]+[A-Z0-9]*) +([MTWRF]{1,5}) +([0-9:AP]+)-([0-9:AP]+);',
 			      $meetings_str, $meeting_matches))
 	      $meeting_start_regex = ';^';
 	      $meeting_base_regex = ' ([A-Z]+) +([A-Z]+[A-Z0-9]*) +([MTWRF]{1,5}) +([0-9:AP]+)-([0-9:AP]+)';
 	      $meeting_date_regex = 'Dates:[^0-9]+([/0-9]{8})-([/0-9]{8})';
 	      $meeting_end_regex = ';';
 	      if (!preg_match($meeting_start_regex . $meeting_base_regex . $meeting_date_regex . $meeting_end_regex,
 			      $meetings_str, $meeting_matches)
 		  && !preg_match($meeting_start_regex . $meeting_base_regex . $meeting_end_regex,
 				 $meetings_str, $meeting_matches))
+		{
-		  if (preg_match(';^Dates:[^0-9]+([/0-9]{8})-([/0-9]{8});',
+		  if (preg_match($meeting_start_regex . $meeting_date_regex . $meeting_end_regex,
 				 $meetings_str, $meeting_matches))
+		    {
 		      /**
 		       * \todo
 		       *   This is a perfect place to get Semester's
 		       *   time_start and time_end values.
 		       */
 		      school_crawl_logf($school_crawl_log, 8, "Skipping some meeting data for %s because it is a date range: `%s'.",
 					implode('-', $section_parts), $meeting_matches[0]);
 		      $meetings_str = substr($meetings_str, strlen($meeting_matches[0]));
 		      continue;
+		    }
 		  school_crawl_logf($school_crawl_log, 6, "Error parsing meeting time. Given `%s'. Skipping %s.", $meetings_str, implode('-', $section_parts));
 		  break;
+		}
 	      /* prepare for parsing the next meeting time */
 	      $meetings_str = substr($meetings_str, strlen($meeting_matches[0]));
 	      $days = school_crawl_days_str_format($meeting_matches[3]);
 	      $time_start = school_crawl_time_format(strptime($meeting_matches[4] . 'M', '%I:%M%p'));
 	      $time_end = school_crawl_time_format(strptime($meeting_matches[5] . 'M', '%I:%M%p'));
 	      $room = $meeting_matches[2];
 	      $type = $meeting_matches[1];
 	      while (isset($meeting_type_maps[$type]))
 		$type = $meeting_type_maps[$type];
 	      $type = strtolower($type);
 	      /* check for daterange information -- i.e., if the first regex successfully matched: */
 	      if (count($meeting_matches) > 7)
+		{
 		  $date_start = school_crawl_mktime(strptime($meeting_matches[6], '%m/%d/%y'));
 		  $date_end = school_crawl_mktime(strptime($meeting_matches[7], '%m/%d/%y'));
 		  if (!empty($date_start) && !empty($date_end))
+		    {
 		      $semester->time_start_set_test($date_start);
 		      $semester->time_end_set_test($date_end);
+		    }
+		}
 	      $meetings[] = new SectionMeeting($days, $time_start, $time_end,
 					       $room, $type);
+	    }
 	  $semester->section_add($section_parts['department'], $section_parts['course'],
 				 new Section($section_parts['section'], $meetings,
 					     $synonym, $instructor));
+	}
+    }
   $semesters[] = $semester;
+    }

0 comments (0 inline, 0 general)