Filter all output from views and artefacts to convert links into relative ones for the HTML export.

Uses a new helper class to do the gruntwork, and a lot of preg_replace stuff. It's a little bit hacky perhaps, but it's the simplest thing that could possibly work.

htdocs/artefact/blog/export/html/lib.php

@@ -48,7 +48,8 @@ class HtmlExportBlog extends HtmlExportArtefactPlugin {
                     array('text' => $blog->get('title'), 'path' => 'index.html'),
                 ));
                 $rendered = $blog->render_self(array());
-                $smarty->assign('rendered_blog', $rendered['html']);
+                $outputfilter = new HtmlExportOutputFilter('../../../');
+                $smarty->assign('rendered_blog', $outputfilter->filter($rendered['html']));
                 $content = $smarty->fetch('export:html/blog:index.tpl');
 
                 if (false === file_put_contents($this->fileroot . $dirname . '/index.html', $content)) {
@@ -61,7 +62,7 @@ class HtmlExportBlog extends HtmlExportArtefactPlugin {
                 if ($postcount > $perpage) {
                     for ($i = 2; $i <= ceil($postcount / $perpage); $i++) {
                         $rendered = $blog->render_self(array('page' => $i));
-                        $smarty->assign('rendered_blog', $rendered['html']);
+                        $smarty->assign('rendered_blog', $outputfilter->filter($rendered['html']));
                         $content = $smarty->fetch('export:html/blog:index.tpl');
 
                         if (false === file_put_contents($this->fileroot . $dirname . "/{$i}.html", $content)) {

htdocs/artefact/internal/export/html/lib.php

@@ -34,7 +34,8 @@ class HtmlExportInternal extends HtmlExportArtefactPlugin {
         // Profile page
         $smarty->assign('breadcrumbs', array(array('text' => 'Profile page', 'path' => 'profilepage.html')));
         $view = $this->exporter->get('user')->get_profile_view();
-        $smarty->assign('view', $view->build_columns());
+        $outputfilter = new HtmlExportOutputFilter('../../');
+        $smarty->assign('view', $outputfilter->filter($view->build_columns()));
 
         $content = $smarty->fetch('export:html/internal:profilepage.tpl');
         if (!file_put_contents($this->fileroot . 'profilepage.html', $content)) {

htdocs/export/html/TODO

 TODOs for export HTML plugin
 ============================
 
-* blogs - render without javascript
-* rewrite all links and images to point to the right places
- - blog attachments
- - smilies
 * views - render in proper structure
 * resume display tidyup
 * i18n

htdocs/export/html/lib.php

@@ -177,7 +177,8 @@ class PluginExportHtml extends PluginExport {
                     throw new SystemException("Could not create directory for view $viewid");
                 }
 
-                $smarty->assign('view', $view->build_columns());
+                $outputfilter = new HtmlExportOutputFilter('../../');
+                $smarty->assign('view', $outputfilter->filter($view->build_columns()));
                 $content = $smarty->fetch('export:html:view.tpl');
                 if (!file_put_contents("$directory/index.html", $content)) {
                     throw new SystemException("Could not write view page for view $viewid");
@@ -264,4 +265,152 @@ abstract class HtmlExportArtefactPlugin {
 
 }
 
+/**
+ * Provides a mechanism for converting the HTML generated by views and 
+ * artefacts for the HTML export.
+ *
+ * Mostly, this means rewriting links to artefacts to point the correct place 
+ * in the export.
+ */
+class HtmlExportOutputFilter {
+
+    /**
+     * The relative path to the root of the generated export - used for link munging
+     */
+    private $basepath = '';
+
+    /**
+     * A cache of view titles. See replace_view_link()
+     */
+    private $viewtitles = array();
+
+    /**
+     * A cache of folder data. See get_path_for_file()
+     */
+    private $folderdata = null;
+
+    /**
+     * @param string $basepath The relative path to the root of the generated export
+     */
+    public function __construct($basepath) {
+        $this->basepath = preg_replace('#/$#', '', $basepath);
+    }
+
+    /**
+     * Filters the given HTML for HTML export purposes
+     *
+     * @param string $html The HTML to filter
+     * @return string      The filtered HTML
+     */
+    public function filter($html) {
+        $wwwroot = preg_quote(get_config('wwwroot'));
+        $html = preg_replace(
+            array(
+                // We don't care about javascript
+                '#<script[^>]*>.*?</script>#si',
+                // Fix simlies from tinymce
+                '#<img src="(' . $wwwroot . ')?/?js/tinymce/plugins/emotions/images/([^"]+)"([^>]+)>#',
+            ),
+            array(
+                '',
+                '<img src="' . $this->basepath . '/static/smilies/$2"$3>',
+            ),
+            $html
+        );
+
+        // Links to views
+        $html = preg_replace_callback(
+            '#' . $wwwroot . 'view/view\.php\?id=(\d+)#',
+            array($this, 'replace_view_link'),
+            $html
+        );
+
+        // Links to artefacts
+        $html = preg_replace_callback(
+            '#<a[^>]+href="(' . preg_quote(get_config('wwwroot')) . ')?/?view/artefact\.php\?artefact=(\d+)(&amp;view=\d+)?(&amp;page=\d+)?"[^>]*>([^<]*)</a>#',
+            array($this, 'replace_artefact_link'),
+            $html
+        );
+
+        // Links to download files
+        $html = preg_replace_callback(
+            '#(' . preg_quote(get_config('wwwroot')) . ')?/?artefact/file/download\.php\?file=(\d+)(&amp;view=\d+)?#',
+            array($this, 'replace_download_link'),
+            $html
+        );
+
+
+        return $html;
+    }
+
+    /**
+     * Callback to replace links to views to point to the correct location in 
+     * the HTML export
+     */
+    private function replace_view_link($matches) {
+        $viewid = $matches[1];
+        if (!isset($this->viewtitles[$viewid])) {
+            $this->viewtitles[$viewid] = PluginExportHtml::text_to_path(get_field('view', 'title', 'id', $viewid));
+        }
+        return $this->basepath . '/views/' . $this->viewtitles[$viewid] . '/index.html';
+    }
+
+    /**
+     * Callback to replace links to artefact to point to the correct location 
+     * in the HTML export
+     */
+    private function replace_artefact_link($matches) {
+        $artefactid = $matches[2];
+        $artefact = artefact_instance_from_id($artefactid);
+
+        switch ($artefact->get('artefacttype')) {
+        case 'blog':
+            $page = ($matches[4]) ? intval(substr($matches[4], strlen('&amp;page='))) : 1;
+            $page = ($page == 1) ? 'index' : $page;
+            return '<a href="' . $this->basepath . '/files/blog/' . PluginExportHtml::text_to_path($artefact->get('title')) . '/' . $page . '.html">' . $matches[5] . '</a>';
+        case 'file':
+        case 'image':
+            $folderpath = $this->get_path_for_file($artefact);
+            return '<a href="' . $this->basepath . '/files/file/' . $folderpath . $artefact->get('title') . '">' . $matches[5] . '</a>';
+        default:
+            return $matches[5];
+        }
+    }
+
+    /**
+     * Callback to replace links to artefact/file/download.php to point to the 
+     * correct file in the HTML export
+     */
+    private function replace_download_link($matches) {
+        $artefactid = $matches[2];
+        $artefact = artefact_instance_from_id($artefactid);
+
+        // If artefact type not something that would be served by download.php, 
+        // replace link with nothing
+        if ($artefact->get_plugin_name() != 'file') {
+            return '';
+        }
+
+        $folderpath = $this->get_path_for_file($artefact);
+        return $this->basepath . '/files/file/' . $folderpath . $artefact->get('title');
+    }
+
+    /**
+     * Given a file, returns the folder path in the HTML export for it
+     *
+     * TODO: slash escaping in file/folder names
+     *
+     * @param ArtefactTypeFileBase $file The file to get the folder path for
+     * @return string
+     */
+    private function get_path_for_file(ArtefactTypeFileBase $file) {
+        if ($this->folderdata === null) {
+            $this->folderdata = get_records_select_assoc('artefact', "artefacttype = 'folder' AND owner = ?", array($file->get('owner')));
+        }
+        $folderpath = ArtefactTypeFileBase::get_full_path($file->get('parent'), $this->folderdata);
+        return $folderpath;
+    }
+
+}
+
 ?>