Merge pull request #54 from Copyleaks/bugfixes/1207271729983716-repor…

…t-not-generating 1207271729983716-report-not-generating
Copyleaks · Jul 10, 2024 · 1a6bf58 · 1a6bf58
2 parents f527798 + b2fed72
commit 1a6bf58
Show file tree

Hide file tree

Showing 5 changed files with 72 additions and 7 deletions.
diff --git a/classes/plagiarism_copyleaks_eventshandler.class.php b/classes/plagiarism_copyleaks_eventshandler.class.php
@@ -336,6 +336,8 @@ private function queue_files($data, $coursemodule, $authoruserid, $submitteruser
 
         foreach ($data['other']['pathnamehashes'] as $pathnamehash) {
 
+            $hashedcontent = $this->check_existing_file_identifier($pathnamehash, $coursemodule, $authoruserid);
+
             $filestorage = get_file_storage();
             $fileref = $filestorage->get_file_by_hash($pathnamehash);
 
@@ -368,13 +370,52 @@ private function queue_files($data, $coursemodule, $authoruserid, $submitteruser
                 $pathnamehash,
                 'file',
                 $data['objectid'],
-                $cmdata
+                $cmdata,
+                $hashedcontent
             );
         }
 
         return $result;
     }
 
+    /**
+     * Check if same identifier already exists in copyleaks_files table,
+     * in case the content is different it will be deleted.
+     * @param string $pathnamehash
+     * @return string hashed content
+     */
+    private function check_existing_file_identifier($pathnamehash, $coursemodule, $authoruserid) {
+        global $DB, $CFG;
+
+        $filerecord = $DB->get_record('files', array('pathnamehash' => $pathnamehash));
+        $hashedcontent = null;
+
+        if ($filerecord) {
+            $hashedcontent = $filerecord->contenthash;
+
+            $typefield = ($CFG->dbtype == "oci") ? " to_char(submissiontype) " : " submissiontype ";
+            $savedfile = $DB->get_record(
+                'plagiarism_copyleaks_files',
+                array(
+                    "cm" => $coursemodule->id,
+                    "userid" => $authoruserid,
+                    $typefield => 'file',
+                    "identifier" => $pathnamehash
+                ),
+            );
+
+            if (isset($savedfile->hashedcontent) && $savedfile->hashedcontent != $hashedcontent) {
+                $DB->delete_records('plagiarism_copyleaks_files', array(
+                    "cm" => $coursemodule->id,
+                    "userid" => $authoruserid,
+                    $typefield => 'file',
+                    "identifier" => $pathnamehash
+                ));
+            }
+        }
+        return $hashedcontent;
+    }
+
     /**
      * Queue submission to be submitted later to Copyleaks by submission task.
      * @param object $coursemodule
@@ -393,7 +434,8 @@ private function queue_submission_to_copyleaks(
         $identifier,
         $subtype,
         $itemid,
-        $cmdata
+        $cmdata,
+        $hashedcontent = null
     ) {
         global $DB, $CFG;
 
@@ -419,7 +461,13 @@ private function queue_submission_to_copyleaks(
             // Submission already exists, do not queue it again.
             return true;
         } else {
-            $submissionid = plagiarism_copyleaks_submissions::create($coursemodule, $authoruserid, $identifier, $subtype);
+            $submissionid = plagiarism_copyleaks_submissions::create(
+                $coursemodule,
+                $authoruserid,
+                $identifier,
+                $subtype,
+                $hashedcontent
+            );
         }
 
         // Check if file type is supported by Copyleaks.
@@ -443,7 +491,6 @@ private function queue_submission_to_copyleaks(
 
         // Scan immediately.
         $scheduledscandate = strtotime('- 1 minutes');
-
         if (isset($cmdata->duedate)) {
             // Get module settings.
             $clmoduleconfig = $DB->get_records_menu(

diff --git a/classes/plagiarism_copyleaks_submissions.class.php b/classes/plagiarism_copyleaks_submissions.class.php
@@ -39,7 +39,7 @@ class plagiarism_copyleaks_submissions {
      * @param string $identifier
      * @param string $submissiontype
      **/
-    public static function create($cm, $userid, $identifier, $submissiontype) {
+    public static function create($cm, $userid, $identifier, $submissiontype, $hashedcontent = null) {
         global $DB;
 
         $file = new stdClass();
@@ -49,6 +49,7 @@ public static function create($cm, $userid, $identifier, $submissiontype) {
         $file->statuscode = "queued";
         $file->similarityscore = null;
         $file->submissiontype = $submissiontype;
+        $file->hashedcontent = $hashedcontent;
 
         if (!$fileid = $DB->insert_record('plagiarism_copyleaks_files', $file)) {
             \plagiarism_copyleaks_logs::add(

diff --git a/db/install.xml b/db/install.xml
@@ -11,6 +11,7 @@
         <FIELD NAME="userid" TYPE="int" LENGTH="10" NOTNULL="true" DEFAULT="0" SEQUENCE="false"/>
         <FIELD NAME="submitter" TYPE="int" LENGTH="10" NOTNULL="false" SEQUENCE="false"/>
         <FIELD NAME="identifier" TYPE="char" LENGTH="255" NOTNULL="true" SEQUENCE="false"/>
+        <FIELD NAME="hashedcontent" TYPE="char" LENGTH="255" NOTNULL="false" SEQUENCE="false"/>
         <FIELD NAME="externalid" TYPE="char" LENGTH="255" NOTNULL="false" SEQUENCE="false"/>
         <FIELD NAME="itemid" TYPE="int" LENGTH="10" NOTNULL="false" SEQUENCE="false"/>
         <FIELD NAME="statuscode" TYPE="char" LENGTH="10" NOTNULL="false" SEQUENCE="false"/>

diff --git a/db/upgrade.php b/db/upgrade.php
@@ -412,7 +412,8 @@ function xmldb_plagiarism_copyleaks_upgrade($oldversion) {
             null,
             null,
             null,
-            'aiscore');
+            'aiscore'
+        );
         $grammarfield = new xmldb_field('grammarcases', XMLDB_TYPE_NUMBER, '10', null, null, null, null);
 
         if ($dbman->table_exists($table)) {
@@ -492,5 +493,20 @@ function xmldb_plagiarism_copyleaks_upgrade($oldversion) {
         upgrade_plugin_savepoint(true, 2024041700, 'plagiarism', 'copyleaks');
     }
 
+    if ($oldversion < 2024070700) {
+        $table = new xmldb_table('plagiarism_copyleaks_files');
+        $hashedcontentfield = new xmldb_field('hashedcontent', XMLDB_TYPE_CHAR, '255', null, false, null, null, 'identifier');
+
+        if ($dbman->table_exists($table)) {
+            // Add hashed content field to files table.
+            if (!$dbman->field_exists($table, $hashedcontentfield)) {
+                $dbman->add_field($table, $hashedcontentfield);
+            }
+        }
+
+        // Copyleaks savepoint reached.
+        upgrade_plugin_savepoint(true, 2024070700, 'plagiarism', 'copyleaks');
+    }
+
     return true;
 }
diff --git a/version.php b/version.php
@@ -23,7 +23,7 @@
 
 defined('MOODLE_INTERNAL') || die();
 
-$plugin->version = 2024051600;
+$plugin->version = 2024070700;
 $plugin->release = "3.5+";
 $plugin->requires = 2018051700;
 $plugin->component = 'plagiarism_copyleaks';