SlatePermutate Changeset - 3487a5b3cbfd · protofusion repositories

Changeset - 3487a5b3cbfd

Parent rev.

Child rev.

[Not reviewed]

default

0 1 0

Nathan Brink (binki) - 13 years ago 2012-10-01 20:19:09
ohnobinki@ohnopublishing.net

Fix some crawler utility functions to handle relative URI resolving correctly and be more flexible when parsing days.

1 file changed with 13 insertions and 4 deletions:

inc/school.crawl.inc

0 comments (0 inline, 0 general)

inc/school.crawl.inc

➞

Show inline comments

@@ @@ -197,53 +197,60 @@ function school_crawl_days_format(array @@
 	school_crawl_logf($school_crawl_log, 5, "school_crawl_days_format() got invalid day specifier: `%s' => `%s'.",
 			  $day_orig, $day);
+    }
   $day_str = '';
   foreach ($my_days as $day_val => $junk)
     $day_str .= $day_val;
   return $day_str;
+}
 /**
  * \brief
  *   Take a string of day initials and format it.
+ *
  * \param $school_crawl_log
  *   The school_crawl_log handle to write errors out to.
  * \param $days_str
  *   Example input: 'mwf', 'TR'.
  * \return
  *   Same as school_crawl_days_format()
  */
 function school_crawl_days_str_format(array $school_crawl_log, $days_str)
+{
   $day_initials = array();
   for ($i = 0; $i < strlen($days_str); $i ++)
     $day_initials[] = $days_str[$i];
   $days_str = preg_replace('/[^[:alnum:]]+/', '', $days_str);
   return school_crawl_days_format($school_crawl_log, $day_initials);
   /*
    * str_split() will produce an array with an empty string in it if
    * the input string is empty. We just want an empty array in that
    * case, but also if there are no input days we can just shortcut
    * and return no days.
    */
   if (empty($days_str))
     return '';
   return school_crawl_days_format($school_crawl_log, str_split($days_str));
+}
 /**
  * \brief
  *   Try to guess a more standardized section_meeting type.
+ *
  * \param $meeting_type
  *   The upstream's meeting_type, such as 'LEC', 'lec', 'LAB',
  *   etc. New mappings should be added to this function as long as
  *   they are general enough.
  */
 function school_crawl_meeting_type($meeting_type = 'lecture')
+{
   static $meeting_type_maps =
     array(
 	  'lec' => 'lecture',
 	  'lab' => 'lab',
 	  'dis' => 'discussion',
 	  );
   if (empty($meeting_type))
     $meeting_type = 'lecture';
   $meeting_type = strtolower(trim($meeting_type));
@@ @@ -583,48 +590,50 @@ function school_crawl_url($orig_url, $ur @@
+{
   /*
    * This accounts for both if the $url is already an absolute, fully
    * qualified URL. It falls back to the original URL if it fails to
    * match.
    */
   foreach (array($url, $orig_url) as $aurl)
     if (preg_match(';^(https?)://([^/]+)(/.*)$;', $aurl, $matches))
+      {
 	$new_url['schema'] = $matches[1];
 	$new_url['hostname'] = $matches[2];
 	$new_url['path'] = $matches[3];
+      }
   /* check if we have an absolute relative path */
   if (!strncmp($url, '/', 1))
     $new_url['path'] = $url;
   /* relative */
   while (!strncmp($url, '../', 3))
+    {
       $new_url['path'] = preg_replace(';[^/]+/[^/]+$;', '/', dirname($new_url['path']));
       $url = substr($url, 3);
+    }
   if (strncmp($url, '/', 1) && strpos($url, '://') === FALSE)
     $new_url['path'] .= $url;
   return $new_url['schema'] . '://' . $new_url['hostname'] . $new_url['path'];
+}
 /**
  * \brief
  *   Map a name onto a column of the table with the help of <th />.
+ *
  * This should be a quite reliable way of matching the data that a
  * user sees onto the actual data because, in most cases, HTML writers
  * are forced to properly align <th /> and the following hundreds of
  * <td />s for there to be a visual alignment.
+ *
  * \param $tr_node
  *   The <tr /> with the <th /> elements to resolve.
  * \param $column_name
  *   The name of the column to search for.
  * \param $strcmp
  *   The function to use with a strcmp($text_content, $column_name) interface when judging
  *   whether or not a <th />'s textContent matches $column_name.
  * \param $trim
  *   The function to apply to the <th />'s textContent before

0 comments (0 inline, 0 general)