pkp/pkp-lib#10365 update parsing for onix xml format

classes/codelist/CodelistItem.php

@@ -42,7 +42,7 @@ public function getText(): string
      */
     public function setText(string $text): void
     {
-        return $this->setData('text', $text);
+        $this->setData('text', $text);
     }
 
     /**
@@ -56,9 +56,9 @@ public function getCode(): string
     /**
      * Set codelist code.
      */
-    public function setCode(string $code)
+    public function setCode(string $code): void
     {
-        return $this->setData('code', $code);
+        $this->setData('code', $code);
     }
 
     /**

classes/codelist/ONIXCodelistItem.php

@@ -40,9 +40,9 @@ public function getText()
      *
      * @param string $text
      */
-    public function setText($text)
+    public function setText($text): void
     {
-        return $this->setData('text', $text);
+        $this->setData('text', $text);
     }
 
     /**
@@ -60,9 +60,9 @@ public function getCode()
      *
      * @param string $code
      */
-    public function setCode($code)
+    public function setCode($code): void
     {
-        return $this->setData('code', $code);
+        $this->setData('code', $code);
     }
 }
 

classes/codelist/ONIXCodelistItemDAO.php

@@ -19,6 +19,7 @@
 
 use Illuminate\Support\Facades\Cache;
 use PKP\core\Registry;
+use PKP\db\DAO;
 use PKP\db\XMLDAO;
 use PKP\facades\Locale;
 use PKP\file\FileManager;
@@ -27,7 +28,7 @@
 use PKP\plugins\Hook;
 use PKP\xslt\XSLTransformer;
 
-class ONIXCodelistItemDAO extends \PKP\db\DAO
+class ONIXCodelistItemDAO extends DAO
 {
     /** @var string The name of the codelist we are interested in */
     public string $_list;
@@ -45,8 +46,7 @@ public function _getCache(?string $locale = null): array
 
             // Reload locale registry file
             $xmlDao = new XMLDAO();
-            $listName = $this->getListName(); // i.e., 'List30'
-            $handler = new ONIXParserDOMHandler($listName);
+            $listName = $this->getListName(); // i.e., '30'
 
             $temporaryFileManager = new TemporaryFileManager();
             $fileManager = new FileManager();
@@ -63,7 +63,13 @@ public function _getCache(?string $locale = null): array
             $xslTransformer->setRegisterPHPFunctions(true);
 
             $xslFile = 'lib/pkp/xml/onixFilter.xsl';
-            $filteredXml = $xslTransformer->transform($filename, XSLTransformer::XSL_TRANSFORMER_DOCTYPE_FILE, $xslFile, XSLTransformer::XSL_TRANSFORMER_DOCTYPE_FILE, XSLTransformer::XSL_TRANSFORMER_DOCTYPE_STRING);
+            $filteredXml = $xslTransformer->transform(
+                $filename,
+                XSLTransformer::XSL_TRANSFORMER_DOCTYPE_FILE,
+                $xslFile,
+                XSLTransformer::XSL_TRANSFORMER_DOCTYPE_FILE,
+                XSLTransformer::XSL_TRANSFORMER_DOCTYPE_STRING
+            );
             if (!$filteredXml) {
                 throw new \Exception('Unable to generate filtered XML!');
             }
@@ -74,6 +80,7 @@ public function _getCache(?string $locale = null): array
                 $fp = fopen($tmpName, 'wb');
                 fwrite($fp, $filteredXml);
                 fclose($fp);
+                $handler = new ONIXParserDOMHandler($listName);
                 $data = $xmlDao->parseWithHandler($tmpName, $handler);
                 $fileManager->deleteByPath($tmpName);
             } else {
@@ -101,13 +108,13 @@ public function _getCache(?string $locale = null): array
     public function getFilename(string $locale): string
     {
         $masterLocale = LocaleInterface::DEFAULT_LOCALE;
-        $localizedFile = "locale/{$locale}/ONIX_BookProduct_CodeLists.xsd";
+        $localizedFile = "locale/{$locale}/ONIX_BookProduct_CodeLists.xml";
         if (Locale::isLocaleValid($locale) && file_exists($localizedFile)) {
             return $localizedFile;
         }
 
         // Fall back on the version for the master locale.
-        return "locale/{$masterLocale}/ONIX_BookProduct_CodeLists.xsd";
+        return "locale/{$masterLocale}/ONIX_BookProduct_CodeLists.xml";
     }
 
     /**

classes/codelist/ONIXParserDOMHandler.php

@@ -11,12 +11,12 @@
  *
  * @see XMLParser
  *
- * @brief This parser extracts a specific xs:simpleType based on a name attribute
- * representing a code list within it. It returns the xs:enumeration values
- * within it along with the xs:documentation elements which serve as textual
- * descriptions of the Codelist values.
+ * @brief This parser extracts a specific CodeList based on a CodeListNumber
+ * representing a code list within it. It returns the Code values
+ * within it along with the CodeDescription elements which serve as textual
+ * descriptions of the Code values.
  *
- * Example:  <xs:simpleType name="List30">...</xs:simpleType>
+ * Example:  <CodeListNumber>28</CodeListNumber>
  */
 
 namespace APP\codelist;
@@ -28,27 +28,36 @@
 
 class ONIXParserDOMHandler extends XMLParserDOMHandler
 {
-    /** @var The list being searched for */
-    public string $_listName;
+    /** @var string The list being searched for */
+    public string $listName;
 
-    public bool $_foundRequestedList = false;
+    public bool $foundRequestedList = false;
 
-    /** @var List of items the parser eventually returns */
-    public ?array $_listItems = [];
+    /** @var ?array List of items the parser eventually returns */
+    public ?array $listItems = [];
 
-    /** @var string to store the current character data  */
-    public ?string $_currentValue = null;
+    /** @var ?string to store the current character data  */
+    public ?string $currentValue = null;
 
-    /** @var bool currently inside an xs:documentation element */
-    public bool $_insideDocumentation = false;
+    /** @var bool currently inside a CodeListNumber element */
+    public bool $inCodeListNumber = false;
+
+    /** @var bool currently inside a CodeValue element */
+    public bool $inCodeValue = false;
+
+    /** @var bool currently inside a CodeDescription element */
+    public bool $inDescription = false;
+
+    /** @var bool currently inside a DeprecatedNumber element */
+    public bool $inDeprecated = false;
 
     /**
      * Constructor.
      */
     public function __construct(string $listName)
     {
         parent::__construct();
-        $this->_listName = $listName;
+        $this->listName = $listName;
     }
 
     /**
@@ -59,21 +68,19 @@ public function startElement(PKPXMLParser|XMLParser $parser, string $tag, array
         $this->currentData = null;
 
         switch ($tag) {
-            case 'xs:simpleType':
-                if ($attributes['name'] == $this->_listName) {
-                    $this->_foundRequestedList = true;
-                }
+            case 'CodeListNumber':
+                $this->inCodeListNumber = true;
                 break;
-            case 'xs:enumeration':
-                if ($this->_foundRequestedList) {
-                    $this->_currentValue = $attributes['value'];
-                    $this->_listItems[$this->_currentValue] = []; // initialize the array cell
+            case 'CodeValue':
+                if ($this->foundRequestedList) {
+                    $this->inCodeValue = true;
                 }
                 break;
-            case 'xs:documentation':
-                if ($this->_foundRequestedList) {
-                    $this->_insideDocumentation = true;
-                }
+            case 'CodeDescription':
+                $this->inDescription = true;
+                break;
+            case 'DeprecatedNumber':
+                $this->inDeprecated = true;
                 break;
         }
 
@@ -92,38 +99,61 @@ public function startElement(PKPXMLParser|XMLParser $parser, string $tag, array
     /**
      * Callback function to act as the character data handler.
      */
-    public function characterData(PKPXMLParser|XMLParser $parser, string $data)
+    public function characterData(PKPXMLParser|XMLParser $parser, string $data): void
     {
-        if ($this->_insideDocumentation) {
-            if (count($this->_listItems[$this->_currentValue]) == 1) {
-                $this->_listItems[$this->_currentValue][0] .= $data;
+        if ($this->inCodeListNumber && $this->listName = $data) {
+            $this->foundRequestedList = true; // @TODO may not need this
+        }
+
+        if ($this->inCodeValue) {
+            $this->currentValue = $data;
+            $this->listItems[$data] = []; // initialize the array cell
+        }
+
+        if ($this->inDescription) {
+            if (count($this->listItems[$this->currentValue]) == 1) {
+                $this->listItems[$this->currentValue][0] .= $data;
             } else {
-                $this->_listItems[$this->_currentValue][0] = $data;
+                $this->listItems[$this->currentValue][0] = $data;
             }
         }
+
+        if ($this->inDeprecated) {
+            $this->listItems[$this->currentValue]['deprecated'] = 1;
+        }
     }
 
     /**
      * Callback function to act as the end element handler.
      */
-    public function endElement(PKPXMLParser|XMLParser $parser, string $tag)
+    public function endElement(PKPXMLParser|XMLParser $parser, string $tag): void
     {
         switch ($tag) {
-            case 'xs:simpleType':
-                $this->_foundRequestedList = false;
+            case 'CodeListNumber':
+                $this->inCodeListNumber = false;
                 break;
-            case 'xs:documentation':
-                $this->_insideDocumentation = false;
+            case 'CodeValue':
+                $this->inCodeValue = false;
+                break;
+            case 'CodeDescription':
+                $this->inDescription = false;
+                break;
+            case 'DeprecatedNumber':
+                $this->inDeprecated = false;
                 break;
         }
+
+        $this->currentNode->setValue($this->currentData);
+        $this->currentNode = & $this->currentNode->getParent();
+        $this->currentData = null;
     }
 
     /**
      * Returns the array of found list items
      */
     public function getResult(): mixed
     {
-        return [$this->_listName => $this->_listItems];
+        return [$this->listName => $this->listItems];
     }
 }
 

classes/components/forms/context/MastheadForm.php

@@ -32,7 +32,7 @@ public function __construct($action, $locales, $context, $imageUploadUrl)
 
         /** @var ONIXCodelistItemDAO */
         $onixCodelistItemDao = DAORegistry::getDAO('ONIXCodelistItemDAO');
-        $codeTypes = $onixCodelistItemDao->getCodes('List44');
+        $codeTypes = $onixCodelistItemDao->getCodes('44');
         $codeTypeOptions = array_map(function ($code, $name) {
             return ['value' => $code, 'label' => $name];
         }, array_keys($codeTypes), $codeTypes);

classes/components/forms/submission/AudienceForm.php

@@ -43,9 +43,9 @@ public function __construct($action, $submission)
 
         /** @var ONIXCodelistItemDAO */
         $onixCodelistItemDao = DAORegistry::getDAO('ONIXCodelistItemDAO');
-        $audienceCodes = $this->getOptions($onixCodelistItemDao->getCodes('List28'));
-        $audienceRangeQualifiers = $this->getOptions($onixCodelistItemDao->getCodes('List30'));
-        $audienceRanges = $this->getOptions($onixCodelistItemDao->getCodes('List77'));
+        $audienceCodes = $this->getOptions($onixCodelistItemDao->getCodes('28'));
+        $audienceRangeQualifiers = $this->getOptions($onixCodelistItemDao->getCodes('30'));
+        $audienceRanges = $this->getOptions($onixCodelistItemDao->getCodes('77'));
 
         $this->addField(new FieldSelect('audience', [
             'label' => __('monograph.audience'),

classes/monograph/Representative.php

@@ -38,15 +38,15 @@ public function getMonographId()
      *
      * @param int $monographId
      */
-    public function setMonographId($monographId)
+    public function setMonographId($monographId): void
     {
-        return $this->setData('monographId', $monographId);
+        $this->setData('monographId', $monographId);
     }
 
     /**
      * Set the ONIX code for this representative role (List93 for Suppliers, List69 for Agents)
      */
-    public function setRole($role)
+    public function setRole($role): void
     {
         $this->setData('role', $role);
     }
@@ -62,19 +62,19 @@ public function getRole()
     }
 
     /**
-     * Get the human readable name for this ONIX code
+     * Get the human-readable name for this ONIX code
      *
      * @return string
      */
     public function getNameForONIXCode()
     {
         $onixCodelistItemDao = DAORegistry::getDAO('ONIXCodelistItemDAO'); /** @var ONIXCodelistItemDAO $onixCodelistItemDao */
         if ($this->getIsSupplier()) {
-            $listName = 'List93'; // List93 -> Publisher to retailers, Wholesaler, etc
+            $listName = '93'; // List 93 -> Supplier role
         } else {
-            $listName = 'List69'; // List93 -> Local Publisher, Sales Agent, etc
+            $listName = '69'; // List 69 -> Agent role
         }
-        $codes = & $onixCodelistItemDao->getCodes($listName);
+        $codes = $onixCodelistItemDao->getCodes($listName);
         return $codes[$this->getRole()];
     }
 

classes/publicationFormat/IdentificationCode.php

@@ -39,7 +39,7 @@ public function getPublicationFormatId()
      */
     public function setPublicationFormatId($publicationFormatId)
     {
-        return $this->setData('publicationFormatId', $publicationFormatId);
+        $this->setData('publicationFormatId', $publicationFormatId);
     }
 
     /**
@@ -63,14 +63,14 @@ public function getCode()
     }
 
     /**
-     * Get the human readable name for this ONIX code
+     * Get the human-readable name for this ONIX code
      *
      * @return string
      */
     public function getNameForONIXCode()
     {
         $onixCodelistItemDao = DAORegistry::getDAO('ONIXCodelistItemDAO'); /** @var ONIXCodelistItemDAO $onixCodelistItemDao */
-        $codes = & $onixCodelistItemDao->getCodes('List5'); // List5 is for ISBN, GTIN-13, etc.
+        $codes = $onixCodelistItemDao->getCodes('5'); // List 5 is for product identifier type
         return $codes[$this->getCode()];
     }