XML2HTMLPlugin.php

<?php
/**
 * Created by PhpStorm.
 * User: dje
 * Date: 04/10/22
 * Time: 17:36
 */

import('lib.pkp.classes.plugins.GenericPlugin');
import('lib.pkp.classes.file.PrivateFileManager');

class XML2HTMLPlugin extends GenericPlugin
{
    public function register($category, $path, $mainContextId = NULL)
    {

        // Register the plugin even when it is not enabled
        $success = parent::register($category, $path);

        if ($success && $this->getEnabled()) {
            HookRegistry::register('ArticleHandler::view::galley', array($this, 'articleHandler'));
            //HookRegistry::register('ArticleHandler::download', array($this, 'articleDownloadCallback'), HOOK_SEQUENCE_LATE);
            HookRegistry::register('LoadHandler', array($this, 'loadImageHandler'));
            HookRegistry::register('CatalogBookHandler::view', array($this, 'catalogHandler'));

        }

        return $success;
    }

    /**
     * @throws Exception
     */
    public function articleHandler($hookName, $params): bool
    {
        $galley =& $params[2];
        $submission = &$params[3];
        //handle only on zip galleys - supposed to be an xml archive (source + images)
        $zipMode = $galley && $galley->getFileType() === 'application/zip';
        if ($zipMode) {
            $this->archiveView($galley->getFile()->getData('path'), $submission->getId(), $galley->getId(),'article');
            return true;
        } else return false;
    }

    public function catalogHandler($hookName, $params): bool
    {

        $submission = &$params[1];
        $galley =& $params[2];
        $submissionFile =& $params[3];
        $zipMode = $submissionFile->getData('mimetype') === "application/zip";
        if ($zipMode) {
            $this->archiveView($submissionFile->getData('path'), $submission->getId(), $galley->getId(),'catalog');
            return true;
        } else
            return false;
    }


    private function archiveView(string $filePath, int $submissionId, int $galleyId, string $op)
    {
        $htmlFile = XML2HTMLPlugin::getHTMLPathFromZipPath($filePath);
        $html = file_get_contents($htmlFile);
        //make transfo only if html does not exist
        if(!$html) {
            $xmlFilePath = XML2HTMLPlugin::unzipXMLArchive($filePath);
            $html = $this->resolveZipImagePaths(
                XML2HTMLPlugin::transform($xmlFilePath),
                $submissionId,
                $galleyId,
                $xmlFilePath, $this->getRequest(),
                $op);
            //store transfo result
            file_put_contents($htmlFile, $html, LOCK_EX);
        }
        $templateMgr = TemplateManager::getManager($this->getRequest());
        $templateMgr->assign('assetsPath',$this->getPluginAssetsPath($this->getRequest()));
        $templateMgr->assign('text', $html);
        $templateMgr->display($this->getTemplateResource('HTMLGalleyView.tpl'));
    }

    /**
     * get HTML file path from its zip one
     * exemple : a/b/c/1234.zip -> ojs/files/dir/a/b/c/1234/1234.html
     * @param $zipPath
     * @return string
     */
    private static function getHTMLPathFromZipPath($zipPath){
        $fileMgr = new PrivateFileManager();
        $filesDir = $fileMgr->getBasePath() . DIRECTORY_SEPARATOR;
        $fullZipPath = $filesDir . $zipPath;
        $fullPrefixPath = str_replace('.zip', '', $fullZipPath);
        $pathInfo = pathinfo($fullZipPath);
        $zipName = $pathInfo['filename'];
        return str_replace($pathInfo['filename'], $zipName. DIRECTORY_SEPARATOR .  $zipName . '.html', $fullPrefixPath);
    }


    /**
     * Provide a name for this plugin
     *
     * The name will appear in the plugins list where editors can
     * enable and disable plugins.
     */
    public function getDisplayName()
    {
        return 'XML To HTML PLUGIN';
    }

    /**
     * Provide a description for this plugin
     *
     * The description will appear in the plugins list where editors can
     * enable and disable plugins.
     */
    public function getDescription()
    {
        return 'This plugin transform XML articles to HTML';
    }

    /**
     * @param $request
     * @param $verb
     * @return mixed
     */
    public function getActions($request, $verb) {
        $router = $request->getRouter();
        import('lib.pkp.classes.linkAction.request.AjaxModal');
        return array_merge(
            $this->getEnabled()?array(
                new LinkAction(
                    'settings',
                    new AjaxModal(
                        $router->url($request, null, null, 'manage', null, array('verb' => 'settings', 'plugin' => $this->getName(), 'category' => 'generic')),
                        $this->getDisplayName()
                    ),
                    __('manager.plugins.settings'),
                    null
                ),
            ):array(),
            parent::getActions($request, $verb)
        );
    }


    /**
     * Load a form when the `settings` button is clicked and
     * save the form when the user saves it.
     *
     * @param array $args
     * @param Request $request
     * @return JSONMessage
     */
    public function manage($args, $request)
    {
        switch ($request->getUserVar('verb')) {
            case 'settings':
                $this->import('XML2HTMLSettingsForm');
                // Load the custom form
                $form = new XML2HTMLSettingsForm($this);

                // Fetch the form the first time it loads, before
                // the user has tried to save it
                if (!$request->getUserVar('save')) {
                    $form->initData();
                    return new JSONMessage(true, $form->fetch($request));
                }

                // Validate and save the form data
                $form->readInputData();
                if ($form->validate()) {
                    $form->execute();
                    return new JSONMessage(true);
                }
        }
        return parent::manage($args, $request);
    }


    function getPluginAssetsPath($request) {
        return $request->getBaseUrl() . '/' . $this->getPluginPath() . '/resources/';
    }

    static function transform(
        string $xmlFilePath
    )
    {
        ////// xslt transfo
        $xsltPath = dirname(__FILE__) . DIRECTORY_SEPARATOR . '2html.xsl';
        // Load the XML source
        $xml = new DOMDocument;
        $xml->load($xmlFilePath);

        $xsl = new DOMDocument;
        $xsl->load($xsltPath, LIBXML_DTDATTR);

        // Configure the transformer
        $proc = new XSLTProcessor;
        $proc->importStyleSheet($xsl); // attach the xsl rules
        $locale = AppLocale::getLocale();
        $proc->setParameter('', 'lang', $locale);
        return $proc->transformToXML($xml);

    }


    static function unzipXMLArchive(string $path)
    {
        $fileManager = new FileManager();
        $submissionFileExtension = $fileManager->parseFileExtension($path);
        $fileInfo = pathinfo($path);
        if ($submissionFileExtension == 'zip') {
            $fileMgr = new PrivateFileManager();
            $filesDir = $fileMgr->getBasePath() . DIRECTORY_SEPARATOR;
            $zip = new ZipArchive;
            $zipPath = $filesDir . $path;
            $zipPathParts = pathinfo($zipPath);
            if (file_exists($zipPathParts['dirname'] . DIRECTORY_SEPARATOR . $zipPathParts['filename'])) {
                return $filesDir . $fileInfo['dirname'] . DIRECTORY_SEPARATOR . $fileInfo['filename'] . DIRECTORY_SEPARATOR . $fileInfo['filename'] . '.xml';
            }
            if ($zip->open($zipPath) === TRUE) {

                $archiveFolderName = NULL;
                if ($zip->numFiles > 0) {//gets zip top folder name
                    $archiveFolderName = $zip->getNameIndex(0);
                } else {
                    throw new \Exception("Bad zip Submission file.");
                }
                $zip->extractTo($zipPathParts['dirname']);
                rename(
                    $zipPathParts['dirname'] . DIRECTORY_SEPARATOR . $archiveFolderName,
                    $zipPathParts['dirname'] . DIRECTORY_SEPARATOR . $zipPathParts['filename']);
                $zip->close();

                $files = scandir($zipPathParts['dirname'] . DIRECTORY_SEPARATOR . $zipPathParts['filename']);
                foreach ($files as $file) {
                    $extension = $fileManager->parseFileExtension($file);
                    if ($extension == 'xml') {
                        $basePath = $zipPathParts['dirname'] . DIRECTORY_SEPARATOR . $zipPathParts['filename'] . '/';
                        rename(
                            $basePath . $file,
                            $basePath . $zipPathParts['filename'] . '.xml'
                        );
                        return $filesDir . $fileInfo['dirname'] . DIRECTORY_SEPARATOR . $fileInfo['filename'] . DIRECTORY_SEPARATOR . $fileInfo['filename'] . '.xml';
                    }
                }

            } else {
                return NULL;
            }
        } else return NULL;

    }

    /**
     * @param $hookName string
     * @param $args array
     * @brief Handle associated files of the full-text, only images are supported
     */
    function loadImageHandler($hookName, array $args)
    {
        $page = $args[0];
        $op = $args[1];

        if ($page == 'article' && (/*$op == 'image' || */$op == 'zimage')) {
            define('HANDLER_CLASS', 'ArticleImageDataHandler');
            import('plugins.generic.xml2html.ArticleImageDataHandler');
        } else if (($page == 'catalog') && (/*$op == 'image' || */$op == 'zimage')) {
            define('HANDLER_CLASS', 'BookImageDataHandler');
            import('plugins.generic.xml2html.BookImageDataHandler');
        }
    }


    static function resolveZipImagePaths(
        string  $htmlString,
        int     $submissionId,
        int     $galleyId,
        string  $xmlFilePath,
        Request $request,
        string  $op = 'article')
    {
        $pathInfo = pathinfo($xmlFilePath);
        $extractedZipPath = $pathInfo['dirname'];
        $files = scandir($extractedZipPath);

        foreach ($files as $file) {
            $info = pathinfo($file);

            if ($info["extension"] == 'jpg' || $info["extension"] == 'png') {
                $request = $request;
                $filePath = $request->url(
                    null, $op,
                    'zimage',
                    array(
                        $submissionId,
                        $galleyId,
                        $file
                    )
                );
                $pattern = preg_quote(rawurlencode($file));
                $htmlString = preg_replace(
                    '/([Ss][Rr][Cc]|[Hh][Rr][Ee][Ff]|[Dd][Aa][Tt][Aa])\s*=\s*"([^"]*' . $pattern . ')"/',
                    '\1="' . $filePath . '"',
                    $htmlString);

            }
        }
        return $htmlString;
    }

}