Merge pull request #168 from willtp87/ISLANDORA-624

Islandora 624

api/fedora_utils.inc

@@ -1,23 +1,23 @@
 <?php
-
-
-
 /**
  * @file
  * Base utilities used by the Islandora fedora module.
  */
+
 if (!function_exists('str_getcsv')) {
 
   /**
    * Functions that emulate php5.3 functionality for backwards compatiablity
+   *
    * @param type $input
    * @param type $delimiter
    * @param type $enclosure
    * @param type $escape
    * @param type $eol
-   * @return type 
+   *
+   * @return type
    */
-  function str_getcsv($input, $delimiter=',', $enclosure='"', $escape=NULL, $eol=NULL) {
+  function str_getcsv($input, $delimiter = ',', $enclosure = '"', $escape = NULL, $eol = NULL) {
     $temp = fopen("php://memory", "rw");
     fwrite($temp, $input);
     fseek($temp, 0);
@@ -29,16 +29,16 @@ if (!function_exists('str_getcsv')) {
 }
 
 /**
- * 
+ *
  * Utility method to get data from a url location using curl_exec and return
- * 
+ *
  * This method takes identical parameters to do_curl_ext and passes them
  * directly to do_curl_ext.  The output from do_curl_ext is processed into
  * either a TRUE, FALSE, or NULL value.
- * 
+ *
  * This method exists for historical reasons as existing code makes use of
  * do_curl directly and expects a simple return value and not an array.
- * 
+ *
  * @param $url
  *   URL to be accessed by the function
  * @param $return_to_variable
@@ -49,8 +49,8 @@ if (!function_exists('str_getcsv')) {
  *   Number of variable sot be posted
  * @param $post
  *   Whether the curl_exec is done as a "get" or a "post"
- *   
+ *
- * @return
+ * @return mixed
  *   TRUE, FALSE, NULL, or the data returned from accessing the URL
  */
 function do_curl($url, $return_to_variable = 1, $number_of_post_vars = 0, $post = NULL) {
@@ -59,18 +59,17 @@ function do_curl($url, $return_to_variable = 1, $number_of_post_vars = 0, $post
 }
 
 /**
- * 
  * Utility method to get data from a url location using curl_exec
- * 
+ *
  * This method takes a URL and three associated parameters and initializes the
  * structure required to make a call to curl_exec.  As a part of the
  * initialization the permissions associated with the (global) user are added
  * to the call.  This ensures access to any potentially user restricted URLs.
- * 
+ *
  * Various defaults are used for the parameters required by curl_exec including
  * the user agent and timeout.  These are hard-coded.
- * 
+ *
- * @param $url
+ * @param string $url
  *   URL to be accessed by the function
  * @param $return_to_variable
  *   Indicates whether the resource accessed by the curl call (HTML page,
@@ -80,8 +79,8 @@ function do_curl($url, $return_to_variable = 1, $number_of_post_vars = 0, $post
  *   Number of variable sot be posted
  * @param $post
  *   Whether the curl_exec is done as a "get" or a "post"
- *   
+ *
- * @return
+ * @return mixed
  *   an array that consists of three value or NULL if curl is not suported:
  *   - element 0:
  *     The value returned from the curl_exec function call.
@@ -112,11 +111,15 @@ function do_curl_ext($url, $return_to_variable = TRUE, $number_of_post_vars = 0,
     $user_agent = "Mozilla/4.0 pp(compatible; MSIE 5.01; Windows NT 5.0)";
     curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, FALSE);
     curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, FALSE);
-    curl_setopt($ch, CURLOPT_FAILONERROR, TRUE); // Fail on errors
+    // Fail on errors.
-    curl_setopt($ch, CURLOPT_FOLLOWLOCATION, TRUE); // allow redirects
+    curl_setopt($ch, CURLOPT_FAILONERROR, TRUE);
-    curl_setopt($ch, CURLOPT_TIMEOUT, 90); // times out after 90s
+    // Allow redirects.
+    curl_setopt($ch, CURLOPT_FOLLOWLOCATION, TRUE);
+    // Times out after 90s.
+    curl_setopt($ch, CURLOPT_TIMEOUT, 90);
     curl_setopt($ch, CURLOPT_USERAGENT, $user_agent);
-    curl_setopt($ch, CURLOPT_RETURNTRANSFER, $return_to_variable); // return into a variable
+    // Return into a variable.
+    curl_setopt($ch, CURLOPT_RETURNTRANSFER, $return_to_variable);
     curl_setopt($ch, CURLOPT_URL, $url);
     curl_setopt($ch, CURLOPT_USERPWD, "$fedora_user:$fedora_pass");
     //curl_setopt($ch, CURLOPT_HTTPAUTH, CURLAUTH_ANY);
@@ -141,7 +144,8 @@ function do_curl_ext($url, $return_to_variable = TRUE, $number_of_post_vars = 0,
 
 /**
  * Fedora available
- * @return type 
+ *
+ * @return mixed
  */
 function fedora_available() {
 
@@ -151,7 +155,8 @@ function fedora_available() {
 
 /**
  * Resource index search available
- * @return type 
+ *
+ * @return mixed
  */
 function risearch_available() {
 
@@ -161,7 +166,9 @@ function risearch_available() {
 
 /**
  * Returns a UTF-8-encoded transcripiton of the string given in $in_str.
+ *
  * @param string $in_str
+ *
  * @return string A UTF-8 encoded string.
  */
 function fix_encoding($in_str) {
@@ -176,8 +183,10 @@ function fix_encoding($in_str) {
 
 /**
  * valid pid ??
+ *
  * @param type $pid
- * @return boolean 
+ *
+ * @return boolean
  */
 function valid_pid($pid) {
   $valid = FALSE;
@@ -190,8 +199,10 @@ function valid_pid($pid) {
 
 /**
  * Valid Dsid ??
+ *
  * @param type $dsid
- * @return boolean 
+ *
+ * @return boolean
  */
 function valid_dsid($dsid) {
   $valid = FALSE;
@@ -204,8 +215,10 @@ function valid_dsid($dsid) {
 
 /**
  * fixDsid ??
+ *
  * @param type $dsid
- * @return string 
+ *
+ * @return string
  */
 function fix_dsid($dsid) {
   $new_dsid = trim($dsid);
@@ -214,14 +227,17 @@ function fix_dsid($dsid) {
   $replace = '';
   $new_dsid = preg_replace($find, $replace, $new_dsid);
 
-  if (strlen($new_dsid) > 63)
+  if (strlen($new_dsid) > 63) {
     $new_dsid = substr($new_dsid, -63);
+  }
 
-  if (preg_match('/^[^a-zA-Z]/', $dsid))
+  if (preg_match('/^[^a-zA-Z]/', $dsid)) {
     $new_dsid = 'x' . $new_dsid;
+  }
 
-  if (strlen($new_dsid) == 0)
+  if (strlen($new_dsid) == 0) {
     $new_dsid = 'item' . rand(1, 100);
+  }
 
   return $new_dsid;
 }
@@ -229,9 +245,8 @@ function fix_dsid($dsid) {
 /**
  * Function: get_collections_as_option_array
  *
- * Description: Returns an associative array of all collection objects in Fedora instance
- *
  * @return array
+ *   Returns an associative array of all collection objects in Fedora instance
  */
 function get_collections_as_option_array() {
   module_load_include('inc', 'fedora_repository', 'api/fedora_utils');
@@ -244,7 +259,7 @@ function get_collections_as_option_array() {
               and $object <fedora-model:state> <info:fedora/fedora-system:def/model#Active>)
             order by $title';
   $url = variable_get('fedora_repository_url', 'http://localhost:8080/fedora/risearch');
-  
+
   $options = array(
     'type' => 'tuples',
     'flush' => 'TRUE',
@@ -254,9 +269,9 @@ function get_collections_as_option_array() {
     'stream' => 'on',
     'query' => $query,
   );
-  //The url function will take care of URL encoding...
+  // The url function will take care of URL encoding.
   $content = do_curl(url($url, array('query' => $options)));
-  
+
   $list = explode("\n", $content);
   array_shift($list);
   $list = preg_replace('/info:fedora\//', '', $list);
@@ -264,7 +279,8 @@ function get_collections_as_option_array() {
     $trimmed_names[] = trim($namespace);
   }
   $options = array();
-  foreach ($list as $item) {   //removes  blanks
+  // Removes  blanks.
+  foreach ($list as $item) {
     if ($item) {
       $parts = explode(',', $item);
       $namespace = explode(':', $parts[0]);
@@ -283,46 +299,75 @@ function get_collections_as_option_array() {
 /**
  * Function: get_content_models_as_option_array
  *
- * Description: Returns an associative array of all available content models in Fedora instance
- *
  * @return array
+ *   associative array of all available content models in Fedora instance
  */
 function get_content_models_as_option_array() {
-    module_load_include('inc', 'fedora_repository', 'api/fedora_item');
+  module_load_include('inc', 'fedora_repository', 'api/fedora_item');
-    module_load_include('inc', 'fedora_repository', 'api/fedora_utils');
+  module_load_include('inc', 'fedora_repository', 'api/fedora_utils');
-    $restricted = variable_get('fedora_namespace_restriction_enforced', TRUE);
+  $restricted = variable_get('fedora_namespace_restriction_enforced', TRUE);
-    $allowed_string = variable_get('fedora_pids_allowed', 'default: demo: changeme: islandora:');
+  $allowed_string = variable_get('fedora_pids_allowed', 'default: demo: changeme: islandora:');
-    $namespaces = explode(':', $allowed_string);
+  $namespaces = explode(':', $allowed_string);
-    foreach ($namespaces as $namespace) {
+  foreach ($namespaces as $namespace) {
-        if ($namespace) {
+    if ($namespace) {
-            $allowed[] = trim($namespace);
+      $allowed[] = trim($namespace);
-        }
     }
-    $query = 'select $object $title from <#ri>
+  }
-          where ($object <fedora-model:label> $title
+  $query = 'select $object $title from <#ri>
-          and ($object <fedora-model:hasModel> <info:fedora/fedora-system:ContentModel-3.0>
+        where ($object <fedora-model:label> $title
-          or $object <fedora-rels-ext:isMemberOfCollection> <info:fedora/islandora:ContentModelsCollection>)
+        and ($object <fedora-model:hasModel> <info:fedora/fedora-system:ContentModel-3.0>
-          and $object <fedora-model:state> <info:fedora/fedora-system:def/model#Active>)
+        or $object <fedora-rels-ext:isMemberOfCollection> <info:fedora/islandora:ContentModelsCollection>)
-          order by $title';
+        and $object <fedora-model:state> <info:fedora/fedora-system:def/model#Active>)
-
+        order by $title';
-    $url = variable_get('fedora_repository_url', 'http://localhost:8080/fedora/risearch');
+
-    $url .= "?type=tuples&flush=TRUE&format=csv&limit=1000&lang=itql&stream=on&query=";
+  $url = variable_get('fedora_repository_url', 'http://localhost:8080/fedora/risearch');
-    $content = do_curl($url . htmlentities(urlencode($query)));
+  $url .= "?type=tuples&flush=TRUE&format=csv&limit=1000&lang=itql&stream=on&query=";
-    $list = explode("\n", $content);
+  $content = do_curl($url . htmlentities(urlencode($query)));
-    array_shift($list);
+  $list = explode("\n", $content);
-    $list = preg_replace('/info:fedora\//', '', $list);
+  array_shift($list);
-    foreach ($list as $item) {   //removes  blanks
+  $list = preg_replace('/info:fedora\//', '', $list);
-        if ($item) {
+  // Removes  blanks.
-            $parts = explode(',', $item);
+  foreach ($list as $item) {
-            $nameparts = explode(':', $parts[0]);
+    if ($item) {
-            if (!$restricted || in_array($nameparts[0], $allowed)) {
+      $parts = explode(',', $item);
-
+      $nameparts = explode(':', $parts[0]);
-                if (!preg_match('/fedora-system/', $nameparts[0])) {
+      if (!$restricted || in_array($nameparts[0], $allowed)) {
-                    $options[$parts[0]] = $parts[1] . ' ~ ' . $parts[0];
+        if (!preg_match('/fedora-system/', $nameparts[0])) {
-                }
+          $options[$parts[0]] = $parts[1] . ' ~ ' . $parts[0];
-            }
         }
+      }
     }
+  }
+
+  return $options;
+}
 
-    return $options;
+/**
-}
+ * This function will retrieve the collections that the given PID belongs to.
+ *
+ * @param string $PID
+ *   The PID to find the parents of.
+ *
+ * @return array
+ *   $object_PIDs the list of PIDs of the collections that the
+ *   indicated object is a member of.
+ */
+function get_parent_collections_from_pid($PID) {
+  $query_string = 'select  $parent  from <#ri>
+        where ($object <fedora-rels-ext:isMemberOf> $parent
+    or $object <fedora-rels-ext:isMemberOfCollection> $parent)
+        and $object <dc:identifier> \'' . $PID . '\'
+        order by $object';
+
+  $query_string = htmlentities(urlencode($query_string));
+  $url = variable_get('fedora_repository_url', 'http://localhost:8080/fedora/risearch');
+  $url .= '?type=tuples&flush=true&format=csv&limit=13000&lang=itql&stream=on&query=' . $query_string;
+
+  $content = do_curl($url, TRUE);
+
+  $results = explode("\n", $content);
+  $object_PIDs = preg_replace('/^info:fedora\/|"parent"| /', '', $results);
+  $object_PIDs = array_values(array_filter($object_PIDs));
+
+  return $object_PIDs;
+}