Work on #463.

mik

@@ -247,7 +247,7 @@ foreach ($records as $record) {
     // Create metadata to return an XML file that can be passed
     // on to the writer.
     try {
-        $metadata = $parser->metadata($record_key) . "\n";
+        $metadata = $parser->metadata($record_key);
     } catch (Exception $exception) {
         $log->addError(
             'ErrorException',

src/metadataparsers/csv/Csv.php

@@ -0,0 +1,28 @@
+<?php
+// src/metadataparsers/csv/Csv.php
+
+namespace mik\metadataparsers\csv;
+
+use mik\metadataparsers\MetadataParser;
+
+/**
+ * Templated metadata parser - Generates CSV metadata.
+ */
+abstract class Csv extends MetadataParser
+{
+    public function __construct($settings)
+    {
+        parent::__construct($settings);
+        $fetcherClass = 'mik\\fetchers\\' . $settings['FETCHER']['class'];
+        $this->fetcher = new $fetcherClass($settings);
+
+        $this->outputFile = $this->settings['WRITER']['output_file'];
+
+        if (isset($this->settings['MANIPULATORS']['metadatamanipulators'])) {
+            $this->metadatamanipulators = $this->settings['MANIPULATORS']['metadatamanipulators'];
+        } else {
+            $this->metadatamanipulators = null;
+        }
+    }
+
+}

src/metadataparsers/csv/DcToCsv.php

@@ -0,0 +1,136 @@
+<?php
+// src/metadataparsers/csv/Csv.php
+
+/**
+ * Metadata parser class that writes metadata out to a CSV file.
+ */
+
+namespace mik\metadataparsers\csv;
+
+// use mik\metadataparsers\MetadataParser;
+use League\Csv\Writer;
+use Monolog\Logger;
+
+class DcToCsv extends Csv
+{
+    /**
+     * Create a new metadata parser instance
+     */
+    public function __construct($settings)
+    {
+        // Call Metadata.php contructor
+        parent::__construct($settings);
+
+        // The CSV writer that writes out object metadata is instantiated in the writer.
+        $headings = $this->settings['METADATA_PARSER']['dc_elements'];
+        array_unshift($headings, $this->settings['METADATA_PARSER']['record_key']);
+        $output_file_path = $this->settings['WRITER']['output_file'];
+        $output_csv_writer = Writer::createFromPath($output_file_path, 'a');
+        $output_csv_writer->insertOne($headings);
+    }
+
+  /**
+     * {@inheritdoc}
+     *
+     *  Returns the output of the template.
+     */
+    public function metadata($record_key)
+    {
+        $objectInfo = $this->fetcher->getItemInfo($record_key);
+        $metadata = $this->populateRow($record_key, $objectInfo);
+        return $metadata;
+    }
+
+    /**
+     * Converts the item's metadata into an array.
+     *
+     * @param string $record_key
+     *   The current item's record_key.
+     * @param object $objectInfo
+     *   The current item's metadata as generated by the fetcher.
+     *
+     * @return array
+     *   The row, as an array.
+     */
+    public function populateRow($record_key, $objectInfo)
+    {
+       $record = $this->getDcValues($objectInfo);
+       $row = array();
+
+       // Field will be named in metadata parser's 'record_key' config setting.
+       $row[] = $record_key;
+
+       foreach ($this->settings['METADATA_PARSER']['dc_elements'] as $element) {
+           // @todo: parse out multiple values and add them to the CSV separated
+           // by a character; what does Migrate Plus want?
+           foreach ($record as $record_element => $record_values) {
+               if ($element == $record_element) {
+                   if (count($record_values) > 0) {
+                       $row[] = $record_values[0];
+                   }
+                   else {
+                       $row[] = '';
+                   }
+               }
+           }
+       }
+
+       if (isset($this->metadatamanipulators)) {
+           $record = $this->applyMetadatamanipulators($record_key, $xml_from_template);
+       }
+
+        return $row;
+    }
+
+    /**
+     * Applies metadatamanipulators listed in the config to provided serialized XML document.
+     *
+     * @param string $record_key
+     *   The current item's record_key.
+     * @param array $row
+     *     An associative array containing the object's metadata.
+     *
+     * @return array
+     *     The modified associative array containing the object's metadata.
+     */
+    public function applyMetadatamanipulators($record_key, $row)
+    {
+        foreach ($this->metadatamanipulators as $metadatamanipulator) {
+            $metadatamanipulatorClassAndParams = explode('|', $metadatamanipulator);
+            $metadatamanipulatorClassName = array_shift($metadatamanipulatorClassAndParams);
+            $manipulatorParams = $metadatamanipulatorClassAndParams;
+            $metdataManipulatorClass = 'mik\\metadatamanipulators\\' . $metadatamanipulatorClassName;
+            $metadatamanipulator = new $metdataManipulatorClass($this->settings, $manipulatorParams, $record_key);
+            $modified_xml = $metadatamanipulator->manipulate($xml);
+        }
+
+        // return $modified_xml;
+        return $modified_row;
+    }
+
+    /**
+     * Parses a DC XML document into an array.
+     *
+     * @param string $xml
+     *   The DC XML document.
+     *
+     * @return array
+     *   An associative array containing element name => element values.
+     */
+    public function getDcValues($xml) {
+      $dc_values = array();
+      $dom = new \DomDocument();
+      $dom->loadXML($xml);
+      $elements = $dom->getElementsByTagNameNS('http://purl.org/dc/elements/1.1/', '*');
+      foreach ($elements as $e) {
+        if (!array_key_exists($e->localName, $dc_values)) {
+          $dc_values[$e->localName] = array();
+          $dc_values[$e->localName][] = $e->nodeValue;
+        }
+        else {
+          $dc_values[$e->localName][] = $e->nodeValue;
+        }
+      }
+      return $dc_values;
+    }
+}

src/writers/OaipmhCsv.php

@@ -0,0 +1,135 @@
+<?php
+
+namespace mik\writers;
+
+use GuzzleHttp\Client;
+use mik\exceptions\MikErrorException;
+use Monolog\Logger;
+use League\Csv\Writer;
+
+class OaipmhCsv extends \mik\writers\Writer
+{
+    /**
+     * @var array $settings - configuration settings from confugration class.
+     */
+    public $settings;
+
+    /**
+     * @var object $fetcher - Fetcher registered in .ini file.
+     */
+    private $fetcher;
+
+    /**
+     * @var object File getter registered in .ini file.
+     */
+    private $fileGetter;
+
+    /**
+     * Create a new OAI-PMH writer Instance
+     * @param array $settings configuration settings.
+     */
+    public function __construct($settings)
+    {
+        parent::__construct($settings);
+        $this->fetcher = new \mik\fetchers\Oaipmh($settings);
+        $fileGetterClass = 'mik\\filegetters\\' . $settings['FILE_GETTER']['class'];
+        $this->fileGetter = new $fileGetterClass($settings);
+        $this->output_directory = $settings['WRITER']['output_directory'];
+
+        $this->output_file_path = $this->settings['WRITER']['output_file'];
+        $this->output_csv_writer = Writer::createFromPath($this->output_file_path, 'a');
+
+        if (isset($this->settings['WRITER']['http_timeout'])) {
+            // Seconds.
+            $this->httpTimeout = $this->settings['WRITER']['http_timeout'];
+        } else {
+            $this->httpTimeout = 60;
+        }
+
+        if (isset($this->settings['WRITER']['metadata_only'])) {
+            // Seconds.
+            $this->metadata_only = $this->settings['WRITER']['metadata_only'];
+        } else {
+            $this->metadata_only = false;
+        }
+
+        // Default Mac PHP setups may use Apple's Secure Transport
+        // rather than OpenSSL, causing issues with CA verification.
+        // Allow configuration override of CA verification at users own risk.
+        if (isset($this->settings['SYSTEM']['verify_ca'])) {
+            if ($this->settings['SYSTEM']['verify_ca'] == false) {
+                $this->verifyCA = false;
+            }
+        } else {
+            $this->verifyCA = true;
+        }
+    }
+
+    /**
+     * Write folders and files.
+     */
+    public function writePackages($metadata, $pages, $record_id)
+    {
+        // Create root output folder
+        $this->createOutputDirectory();
+        $output_path = $this->outputDirectory . DIRECTORY_SEPARATOR;
+
+        $normalized_record_id = $this->normalizeFilename($record_id);
+        $this->writeMetadataFile($metadata);
+
+        if ($this->metadata_only) {
+              return;
+        }
+
+        // Retrieve the file associated with the document and write it to the output
+        // folder using the filename or record_id identifier
+        $source_file_url = $this->fileGetter->getFilePath($record_id);
+        // Retrieve the PDF, etc. using Guzzle.
+        if ($source_file_url) {
+            $client = new Client();
+            $response = $client->get(
+                $source_file_url,
+                ['stream' => true,
+                'timeout' => $this->httpTimeout,
+                'connect_timeout' => $this->httpTimeout,
+                'verify' => $this->verifyCA]
+            );
+
+            // Lazy MimeType => extension mapping: use the last part of the MimeType.
+            $content_types = $response->getHeader('Content-Type');
+            list($type, $extension) = explode('/', $content_types[0]);
+            $extension = preg_replace('/;.*$/', '', $extension);
+
+            $content_file_path = $output_path . $normalized_record_id . '.' . $extension;
+
+            $body = $response->getBody();
+            while (!$body->eof()) {
+                file_put_contents($content_file_path, $body->read(2048), FILE_APPEND);
+            }
+        } else {
+            $this->log->addWarning(
+                "No content file found in OAI-PMH record",
+                array('record' => $record_id)
+            );
+        }
+    }
+
+    /**
+     * Adds a row to CSV file (unlike other Writers' writeMetadataFile(),
+     * which writes out an entire metadata XML file.
+     */
+    public function writeMetadataFile($metadata, $output_file_path = '')
+    {
+        $this->output_csv_writer->insertOne($metadata);
+    }
+
+    /**
+     * Convert %3A (:) in filenames into underscores (_).
+     */
+    public function normalizeFilename($string)
+    {
+        $string = urldecode($string);
+        $string = preg_replace('/:/', '_', $string);
+        return $string;
+    }
+}