Changeset fcf3700

Timestamp:

04/30/10 15:46:52 (2 years ago)

Author:

maparent <maparent@…>

Branches:

('master', '8434b8322dd6f8ff37c17bd45b94e7a826f86b53')('vitrine', 'afe109c61a407808bd54d88bcccb1cde726a010a')

Children:

a04edbd9894622947728b10e2d889e6975b35391

Parents:

ad3626832640a4cfafb8e5960b72d0e8dec9819b

git-author:

maparent <maparent@9a151f2f-a8df-0310-a5c5-9fb547312938>2010-04-30 19:46:52+00:00

git-committer:

maparent <maparent@9a151f2f-a8df-0310-a5c5-9fb547312938>2010-04-30 19:46:52+00:00

Message:

Basic XSLT for spiders

git-svn-id: ​http://projects.coeus.ca/svn/eureka@564 9a151f2f-a8df-0310-a5c5-9fb547312938

Location:

trunk/html

Files:

• admin/spider_admin.php (modified) (1 diff)
• classes/AbstractSpider.php (modified) (5 diffs)
• classes/SpiderFeed.php (modified) (1 diff)
• classes/SpiderHTML.php (modified) (1 diff)
• classes/SpiderOaiPmhClient.php (modified) (1 diff)
• include/schema_validate.php (modified) (1 diff)

trunk/html/admin/spider_admin.php

@@ -85 +85 @@
               }
 
-              if($spider){echo "<form method='POST' action=''>\n";
+              if($spider){echo "<form method='POST' action='' enctype='multipart/form-data'>\n";
               echo "<input type='hidden' name='spider_id' value='{$spider->getId()}'>\n";
               // Show admin UI

trunk/html/classes/AbstractSpider.php

@@ -154 +154 @@
         echo "<tr class='even'><td><b>"._("Feed URL")." : </b></td><td><input type='text' size='50' name='rep_base_url' value='{$this->getBaseUrl()}'></td></tr>\n";
 
-        echo "<tr class='ood'><td><b>"._("Base synchronization theme")." : </b></td><td>";
+        echo "<tr class='odd'><td><b>"._("Base synchronization theme")." : </b></td><td>";
         $vocabulary = new Vocabulary(VOC_EUREKA_THEME);
         if (($selected_theme = $this->getBaseTheme()) == null)
@@ -178 +178 @@
             echo sprintf(_("Last sync started %s and finised %s"), $this->getLastSyncStartTime(), $last_sync_date);
         }
+
+        echo "</tr>\n";
+        echo "<tr class='even'><td><b>"._("XSLT Transformation")."</b>";
+        $transform = $this->getXslTransform();
+        if (strlen($transform) == 0) {
+            echo ' (<i>'._("none").'</i>)';
+        }
+        echo "</td>\n<td><INPUT TYPE='hidden' name='MAX_FILE_SIZE' value='20000' />";
+        echo "<input name='xsl_transform_import' type='file' /><br/>";
+        echo "<textarea name='xsl_transform_content' rows='20' cols='80' wrap='off'>";
+        echo htmlspecialchars($transform);
+        echo "</textarea>";
+        echo "</td></tr>\n";
+
         echo $childUi;
         echo "</table>\n";
@@ -202 +216 @@
         {
             $this->setMetaDataExchangeFormat($_REQUEST['import_format']);
+        }
+        if (isset($_FILES['xsl_transform_import']['tmp_name'])
+            && $_FILES['xsl_transform_import']['tmp_name'] != '') {
+            $xsl = file_get_contents($_FILES['xsl_transform_import']['tmp_name']);
+        } else {
+            $xsl = $_REQUEST['xsl_transform_content'];
+        }
+        $xsl_dom = new DOMDocument();
+        if ($xsl_dom->loadXml($xsl)) {
+            // TODO: Also check it is valid XSLT
+            $this->setXslTransform($xsl);
+        } else {
+            // If not XML, clear it.
+            $this->setXslTransform('');
         }
 
@@ -387 +415 @@
     }
 
+    public function getXslTransform()
+    {
+        return $this->row['xsl_transform'];
+    }
+
+    public function setXslTransform($xsl)
+    {
+        $db=AbstractDb::getObject();
+        $xsl = $db->EscapeString($xsl);
+        $db->execSqlUpdate("UPDATE spiders SET xsl_transform = '{$xsl}' WHERE spider_id = {$this->getId()};", false);
+        $this->refresh();
+        return true;
+    }
+
     /** Run spider */
     abstract protected function harvest($verbose=false);
@@ -451 +493 @@
         }
     }
+    protected function transformDoc($xml_dom) {
+        if (!isset($this->xsl_proc_cache)) {
+            $this->xsl_proc_cache = null;
+            $xsl = $this->getXslTransform();
+            if (isset($xsl) && $xsl != '') {
+                $xsl_proc = new XSLTProcessor();
+                $xsl_dom = new DOMDocument(); 
+                if ($xsl_dom->loadXml( $xsl )) {
+                    $xsl_proc->importStylesheet( $xsl_dom );
+                    $this->xsl_proc_cache = $xsl_proc;
+                }
+            }
+        }
+        if ($this->xsl_proc_cache != null) {
+            $new_dom = $this->xsl_proc_cache->transformToDoc( $xml_dom );
+            if ($new_dom) {
+                return $new_dom;
+            }
+        }
+        return $xml_dom;
+    }
 }
 

trunk/html/classes/SpiderFeed.php

@@ -67 +67 @@
         else             {
             // Download the DOM document
-            $xml_dom = $xml_dom = DOMDocument :: load($this->getBaseURL());
+            $xml_dom = DOMDocument :: load($this->getBaseURL());
             if($xml_dom == null) {
                 $this->reportAppend(sprintf(_("unable to retrieve feed at %s"), $this->getBaseURL())."<br>\n");
             }
             else {
+                // TODO: Apply to items only? 
+                $document = $this->transformDoc($document);
                 // Create a XPath query object
                 $xpath = new DOMXPath($xml_dom);

trunk/html/classes/SpiderHTML.php

@@ -163 +163 @@
                     else {
                         $log .= sprintf("<strong>Processing <a href='%s'>%s</a></strong></td></tr><tr><td>\n", $link_url, $link["caption"]);
+                        $document = $this->transformDoc($document);
                         // Create an XPath for queries and register a default namespace for it.
                         $xpath = new DOMXPath($document);

trunk/html/classes/SpiderOaiPmhClient.php

@@ -265 +265 @@
                 // load the DOM in RAM
                 $xml_dom = DOMDocument :: loadXML($xml_file_content);
+                $xml_dom = $this->transformDoc($xml_dom);
                 if (OAI_PMH_VALIDATE_ENABLED && !$xml_dom->schemaValidate(OAI_PMH_VALIDATE_SCHEMA))
                 {

trunk/html/include/schema_validate.php

@@ -409 +409 @@
         }
 
+        $new_schema_version = 21;
+        if ($current_schema_version < $new_schema_version && $new_schema_version <= REQUIRED_SCHEMA_VERSION)
+        {
+            echo "<h2>"._("Preparing SQL statements to update schema to version:")." $new_schema_version</h2>";
+            $sql .= "UPDATE schema_info SET value='$new_schema_version' WHERE tag='schema_version';\n";
+            $sql .= "ALTER TABLE spiders ADD COLUMN xsl_transform text;\n";
+        }
+
         $db->execSqlUpdate("BEGIN;\n$sql\nCOMMIT;\n", true);
         $db->execSqlUpdate("VACUUM ANALYSE;\n", true);