SlatePermutate Changeset - 39cea00b7ea4 · protofusion repositories

Changeset - 39cea00b7ea4

Parent rev.

Child rev.

[Not reviewed]

default

0 1 0

Ethan Zonca (ethanzonca) - 15 years ago 2010-11-13 20:40:08
e@ethanzonca.com

Cedarville crawler: make output cleaner

1 file changed with 5 insertions and 0 deletions:

school.d/cedarville.inc

0 comments (0 inline, 0 general)

school.d/cedarville.inc

➞

Show inline comments

@@ @@ -69,12 +69,13 @@ function cedarville_default_classes() @@
+ *
  * \param $html
  *   HTML that PHP's DOM would willingly would eat.
  */
 function table_parse($html)
+{
   libxml_use_internal_errors(true); // Suppress warnings
   $arr = array();
   $dom = new DOMDocument;
   if(!$html)
     return NULL;
   $dom->loadHTML($html);
@@ @@ -90,23 +91,27 @@ function table_parse($html) @@
   return $arr;
+}
 /** Crawls Cedarville course listings. $season is "fa" or "sp", year is 4-digit year */
 function cedarville_crawl($semester, $verbosity = 1)
+{
   $season = strtolower(substr($semester->season_get(), 0, 2));
   $year = $semester->year_get();
   /* Current academic departments. Update as needed. */
   $departments = array('be','ba','ca','ed','eg','es','hg','id','ll','ms','mu','ns','ph','py','sm','sw');
   $basepath = "http://cedarville.edu/courses/schedule/";
   echo "cedarville_crawl(): Beginning crawl of Cedarville:\n";
   $season = strtolower($season);
   $tables = array();
   foreach($departments as $department)
+    {
       echo "cedarville_crawl(): Crawling department \"$department\"...\n";
       $html = file_get_contents($basepath . $year . $season . '_' . $department . '_' . 'all.htm');
       if (!$html)
 	continue;
       $tables[$department] = table_parse(cedarville_html_fix($html));
+    }

0 comments (0 inline, 0 general)