WIP on #44.

sample_islandora_config.yml

@@ -32,6 +32,9 @@ gemini_auth_header: 'Bearer islandora'
 # Absolute or relative to the Riprap application directory.
 views_pager_data_file_path: 'var/fetchresourcelist.from.drupal.pager.txt'
 
+# Required. Non-standard or custom media need to have the fieldname they use to contain the file added to this list.
+drupal_file_fieldnames: ['field_media_audio', 'field_media_document', 'field_edited_text', 'field_media_file', 'field_media_image', 'field_media_video_file']
+
 plugins.fetchdigest: PluginFetchDigestFromFedoraAPI
 fedoraapi_method: HEAD
 fedoraapi_digest_header_leader_pattern: "^.+="

sample_islandora_config_fetch_digest_from_drupal.yml

@@ -0,0 +1,47 @@
+# Sample Riprap config file for using Islandora/database plugins.
+# Requires that the "Riprap resource list" View be enabled in
+# the Islandora instance. This View is bundled with the Islandora Riprap
+# module.
+#
+# This plugin is agnostic to which media have fixity event checks performed
+# on them. The filters in the "Riprap resource list" View determine that.
+# See the View's filter criteria GUI for examples.
+#
+# This plugin differs from sample_islandora_config.yml in that it gets the
+# media file's digest from Drupal, not Fedora. To do this, it calls the
+# /islandora_riprap/checksum/{file_hash}/{algorithm} endpoint added by the
+# Islandora Riprap module.
+
+####################
+# General settings #
+####################
+
+# thin: true
+# One of 'md5', 'sha1', or 'sha256'.
+fixity_algorithm: sha1
+
+# Absolute or relative to the Riprap application directory.
+failures_log_path: 'var/riprap_failed_events.log'
+
+###################
+# Plugin settings #
+###################
+
+# Use this plugin if you want to use the 'Riprap resource (media) list' View provided by Islandora Riprap.
+plugins.fetchresourcelist: ['PluginFetchResourceListFromDrupalView']
+drupal_baseurl: 'http://localhost:8000'
+drupal_user: admin
+drupal_password: islandora
+# Absolute or relative to the Riprap application directory.
+views_pager_data_file_path: 'var/fetchresourcelist.from.drupal.pager.txt'
+use_fedora_urls: false
+
+# Required. Non-standard or custom media need to have the fieldname they use to contain the file added to this list.
+drupal_file_fieldnames: ['field_media_audio', 'field_media_document', 'field_edited_text', 'field_media_file', 'field_media_image', 'field_media_video_file']
+
+plugins.fetchdigest: PluginFetchDigestFromDrupal
+
+plugins.persist: PluginPersistToDatabase
+
+plugins.postcheck: ['PluginPostCheckCopyFailures']
+

src/Plugin/PluginFetchDigestFromDrupal.php

@@ -19,33 +19,55 @@ class PluginFetchDigestFromDrupal extends AbstractFetchDigestPlugin
      * Gets the resource's digest from the REST endpoint provided by Islandora Riprap.
      *
      * @param string $resource_id
-     *    The file's UUID.
+     *    This is probably wrong.
+     *    The media's Drupal media ID. From this, we get the file's URL
+     *    using the following logic: with this media ID, we perform
+     *    a REST request to get the media. That request's response
+     *    contains an entry for the field that contains the file, e.g.
+     *    field_media_audio, field_media_document, field_edited_text,
+     *    field_media_file, field_media_image, field_media_video_file.
+     *    Within that entry, we get the file's URL, and from that, we
+     *    can get the file's digest.
      *
      * @return string $digest
      *   The digest value.
+     *
+     * @return string|bool
+     *   The digest value, false on error.     
      */
     public function execute($resource_id)
     {
+        if (isset($this->settings['drupal_baseurl'])) {
+            $this->drupal_base_url = $this->settings['drupal_baseurl'];
+        } else {
+            $this->drupal_base_url = 'http://localhost:8000';
+        }
+
         if (isset($this->settings['fixity_algorithm'])) {
             $this->fixity_algorithm = $this->settings['fixity_algorithm'];
         } else {
             $this->fixity_algorithm = 'sha1';
         }
 
+        $this->drupal_file_fieldnames = $this->settings['drupal_file_fieldnames'];
+
         $client = new \GuzzleHttp\Client();
-        // @todo: Wrap in try/catch.
 
-        // @todo: Request is to /islandora_riprap/checksum/{file_uuid}/{algorithm}, not to the resource ID as with Fedora.
-        $url = $resource_id;
-        if (!strlen($url)) {
+        if (!strlen($resource_id)) {
             if ($this->logger) {
                 $this->logger->info("PluginFetchDigestFromDrupal exited due to empty resource ID.");
             }
             return;
         }
 
-        $response = $client->request('GET', $url, [
+        // @todo: Request is to /islandora_riprap/checksum/{file_uuid}/{algorithm},
+        // not to the resource URI as with Fedora.
+        $get_digest_url = $this->drupal_base_url .
+            '/islandora_riprap/checksum/' . $resource_id . '/' . $this->fixity_algorithm;
+
+        $response = $client->request('GET', $get_digest_url, [
             'http_errors' => false,
+            'auth' => [$this->drupal_user, $this->drupal_password]
         ]);
         $status_code = $response->getStatusCode();
         $allowed_codes = array(200);

src/Plugin/PluginFetchResourceListFromDrupalView.php

@@ -1,5 +1,5 @@
 <?php
-// src/Plugin/PluginFetchResourceListFromDrupal.php
+// src/Plugin/PluginFetchResourceListFromDrupalView.php
 
 namespace App\Plugin;
 
@@ -12,6 +12,11 @@ public function execute()
     {
         $output = new ConsoleOutput();
 
+        if (isset($this->settings['fixity_algorithm'])) {
+            $this->fixity_algorithm = $this->settings['fixity_algorithm'];
+        } else {
+            $this->fixity_algorithm = 'sha1';
+        }
         if (isset($this->settings['drupal_baseurl'])) {
             $this->drupal_base_url = $this->settings['drupal_baseurl'];
         } else {
@@ -27,6 +32,9 @@ public function execute()
         } else {
             $this->drupal_password = 'islandora';
         }
+
+        $this->drupal_file_fieldnames = $this->settings['drupal_file_fieldnames'];
+
         if (isset($this->settings['use_fedora_urls'])) {
             $this->use_fedora_urls = $this->settings['use_fedora_urls'];
         } else {
@@ -86,17 +94,29 @@ public function execute()
         $output_resource_records = [];
         foreach ($media_list as $media) {
             if ($this->use_fedora_urls) {
-                // @todo: getFedoraUrl() returns false on failure, so build in logic here to log that
+                // @todo: getFileUrlFromFedora() returns false on failure, so build in logic here to log that
                 // the resource ID / URL cannot be found. (But, http responses are already logged in
-                // getFedoraUrl() so maybe we don't need to log here?)
-                $file_fedora_url = $this->getFedoraUrl($media['mid']);
+                // getFileUrlFromFedora() so maybe we don't need to log here?)
+                $file_fedora_url = $this->getFileUrlFromFedora($media['mid']);
                 if (strlen($file_fedora_url)) {
                     $resource_record_object = new \stdClass;
                     $resource_record_object->resource_id = $file_fedora_url;
                     $resource_record_object->last_modified_timestamp = $media['changed'];
                     $output_resource_records[] = $resource_record_object;
                 }
             }
+            if (!$this->use_fedora_urls) {
+                // @todo: getFileUrlFromDrupal() returns false on failure, so build in logic here to log that
+                // the resource ID / URL cannot be found. (But, http responses are already logged in
+                // getFileUrlFromFedora() so maybe we don't need to log here?)
+                $file_drupal_url = $this->getFileUuidFromDrupal($media['mid']);
+                if (strlen($file_drupal_url)) {
+                    $resource_record_object = new \stdClass;
+                    $resource_record_object->resource_id = $file_drupal_url;
+                    $resource_record_object->last_modified_timestamp = $media['changed'];
+                    $output_resource_records[] = $resource_record_object;
+                }
+            }            
         }
         $this->setPageNumber($page_number, $num_media);
 
@@ -115,9 +135,9 @@ public function execute()
     *   The media ID.
     *
     * @return string
-    *    The Fedora URL corresponding to the UUID, or false.
+    *    The Fedora URL corresponding to the media ID, or false.
     */
-    private function getFedoraUrl($mid)
+    private function getFileUrlFromFedora($mid)
     {
         // First, retrieve the media entity from Drupal.
         $media_url = $this->drupal_base_url . '/media/' . $mid . '?_format=json';
@@ -128,11 +148,12 @@ private function getFedoraUrl($mid)
         ]);
         $media_response_body = $media_response->getBody()->getContents();
         $media_response_body = json_decode($media_response_body, true);
-        if (isset($media_response_body['field_media_image'])) {
-            $file_field = 'field_media_image';
-        }
-        if (isset($media_response_body['field_media_file'])) {
-            $file_field = 'field_media_file';
+
+        foreach ($this->drupal_file_fieldnames as $file_fieldname) {
+            if (isset($media_response_body[$file_fieldname])) {
+                $file_field = $file_fieldname;
+                break;
+            }
         }
         $target_file_uuid = $media_response_body[$file_field][0]['target_uuid'];
 
@@ -155,7 +176,7 @@ private function getFedoraUrl($mid)
             } else {
                 if ($this->logger) {
                     $this->logger->error(
-                        "PluginFetchResourceListFromDrupal could not get Fedora URL from Gemini.",
+                        "PluginFetchResourceListFromDrupal could not get Fedora File URL from Gemini.",
                         array(
                             'HTTP response code' => $code
                         )
@@ -166,7 +187,7 @@ private function getFedoraUrl($mid)
         } catch (Exception $e) {
             if ($this->logger) {
                 $this->logger->error(
-                    "PluginFetchResourceListFromDrupal could not get Fedora URL from Gemini.",
+                    "PluginFetchResourceListFromDrupal could not get Fedora File URL from Gemini.",
                     array(
                         'HTTP response code' => $code,
                         'Exception message' => $e->getMessage()
@@ -177,6 +198,67 @@ private function getFedoraUrl($mid)
         }
     }
 
+   /**
+    * Get a URL for a File entity from Drupal.
+    *
+    * @param string $mid
+    *   The media ID.
+    *
+    * @return string
+    *    The UUID of the file, or false. We use the UUID to get the
+    *    digest and URL from the /islandora_riprap/checksum endpoint.
+    */
+    private function getFileUuidFromDrupal($mid)
+    {
+        // Retrieve the media entity from Drupal.
+        try {
+            $url = $this->drupal_base_url . '/media/' . $mid . '?_format=json';
+            $client = new \GuzzleHttp\Client();
+            $response = $client->request('GET', $url, [
+                'http_errors' => false,
+                'auth' => [$this->drupal_user, $this->drupal_password]
+            ]);
+            $code = $response->getStatusCode();
+            $body = $response->getBody()->getContents();
+            $body = json_decode($body, true);
+
+            foreach ($this->drupal_file_fieldnames as $file_fieldname) {
+                if (isset($body[$file_fieldname])) {
+                    $file_field = $file_fieldname;
+                    break;
+                }
+            }
+            if ($code == 200) {
+                $body = $response->getBody()->getContents();
+                $body_array = json_decode($body, true);
+                return $body_array[$file_field][0]['url'];
+            } elseif ($code == 404) {
+                return false;
+            } else {
+                if ($this->logger) {
+                    $this->logger->error(
+                        "PluginFetchResourceListFromDrupal could not get File URL from Drupal.",
+                        array(
+                            'HTTP response code' => $code
+                        )
+                    );
+                }
+                return false;
+            }
+        } catch (Exception $e) {
+            if ($this->logger) {
+                $this->logger->error(
+                    "PluginFetchResourceListFromDrupal could not get File URL from Drupal.",
+                    array(
+                        'HTTP response code' => $code,
+                        'Exception message' => $e->getMessage()
+                    )
+                );
+            }
+            return false;
+        }            
+    }
+
    /**
     * Sets the page offset to use in the next REST request to the Drupal View.
     *