praiadeseselle/site/modules/ProcessRSSFetch/ProcessRSSFetch.module

<?php

/**
 * ProcessWire External RSS Feed Fetch
 *
 * For ProcessWire 3.x 
 * 
 * @author Laegnur
 *
 */

class ProcessRSSFetch extends Process
{
    protected $items = null;

    protected $replacements  = array(
        '<br />' => ' ',
        '<br>' => ' ',
        '</p>' => ' ',
        "\n" => ' ',
        "\r" => ' ',
    );

    public static function getModuleInfo()
    {
        return array(
            'title' => 'ProcessWire RSS Fetch',
            'summary' => 'ProcessWire External RSS Feed Fetch',
            'version' => 001,
            'author' => 'Laegnur',
            'href' => 'https://laegnur.artabro.org/',
            'icon'     => 'rss'
        );
    }

    public function ___execute()
    {
        $form = $this->modules->get("InputfieldForm");
        $form->action = "./";
        $form->method = "post";
        $form->attr("id+name",'rssfetch-form');

        $fieldText = $this->modules->get("InputfieldText");
        $fieldText->label = "RSS feed";
        $fieldText->name = 'rss';
        $fieldText->description = "URL del feed RSS";
        $fieldText->columnWidth = 100;
        $fieldText->required = 1;
        $form->append($fieldText);

        $submit = $this->modules->get("InputfieldSubmit");
        $submit->attr("value","Traer posts");
        $submit->attr("id+name","submit");
        $form->append($submit);

        $out = $form->render();

        if($this->input->post->submit)
        {
            set_time_limit(600);

            $form->processInput($this->input->post);
            if(!$form->getErrors())
            {
                $this->items = new WireArray();
                $xmlData = $this->loadXmlData($this->input->post->rss);
                $rss = simplexml_load_string($xmlData);
                if(!$rss)
                {
                    $msg = "Unable to load RSS feed at $url: \n";
                    foreach(libxml_get_errors() as $error) $msg .= $error . " \n";
                    $out .= $msg;
                    return $out;
                }

                foreach($rss->channel->category as $category)
                {
                    $name = $this->sanitizer->pageNameTranslate($category);

                    if(($this->pages->findOne("template=etiqueta,name=$name") instanceof NullPage))
                    {
                        $t = new Page();
                        $t->template = 'etiqueta';
                        $t->parent = wire('pages')->get('/noticias/etiquetas/');
                        $t->name = $name;
                        $t->title = $category;
                        $t->save();
                      
                        $this->log('Creada etiqueta: ' . $category, ['name' => 'rssfetch']);
                    }
                    else
                    {
                        $this->log('Etiqueta ya existente: ' . $category, ['name' => 'rssfetch']);
                    }
                }
                foreach($rss->channel->item as $item)
                {
                    $title = '';
                    $date = '';
                    $img = '';
                    $content = '';
                    $category = array();
                    $tags = array();

                    foreach($item as $key => $value)
                    {
                        switch ($key)
                        {
                            case 'pubDate':
                                $date = $value;
                                break;
                            case 'title':
                                $title = $value;
                                break;
                            case 'description':
                                $content = $this->cleanText($value);
                                break;
                            case 'media:thumbnail':
                                $attributes = $key->attributes();
                                $img = $attributes['url'];
                                break;
                            case 'category':
                                $tags[] = $value;
                                break;
                            default:
                                break;
                        }
                        
                    }

                    $out .= '[ '. "\n" . 'Data: ' . $date . "\n" . 'Titulo: ' . $title . "\n" . 'Contenido: ' . $content . "\n" . 'Img: ' . $img . "\n" . 'Etiquetas: ' . implode(', ', $tags) . "\n" . ']'. "\n";

                    /*$p = new Page();
                    $p->template = 'publicacion';
                    $p->parent = wire('pages')->get('/noticias/publicaciones/');
                    $p->addStatus(Page::statusUnpublished);

                    $p 

                    $p->save();*/
                }
            }
            return $out;
        }
        return $out;
    }

    public function install()
    {
        $page = new Page();

        $page->template = "admin";
        $page->name = "feed-fetch";
        $page->title = "RSS Feed Fetch";
        $page->save();

        $page->process = 'ProcessRSSFetch';

        $admin = $this->pages->get("id=2");
        $page->parent = $admin;

        $page->save();
    }

    public function uninstall()
    {
        $page = $this->pages->get("name=feed-fetch");
        if(count($page)) $this->pages->delete($page, true);
    }

    /**
     * Load XML data from either the RSS URL or the cache file
     * 
     * @param string $url
     * @return bool|string
     *
     */
    protected function loadXmlData($url)
    {
        $cacheFile = $this->config->paths->cache . $this->className() . '/' . md5($url) . '.xml.cache';

        if(!is_file($cacheFile) || time() - filemtime($cacheFile) > 120)
        {
            $http = new WireHttp();
            $this->wire($http);
            $xmlData = $http->get($url);
            if(empty($xmlData))
            {
                $this->error("Unable to load: $url");
                return false;
            }
            else
            {
                @file_put_contents($cacheFile, $xmlData, LOCK_EX);
            }
        }
        else
        {
            $xmlData = file_get_contents($cacheFile); 
            if($xmlData === false)
            {
                $this->error("Unable to load XML data cache");
                return false;
            }
        }

        return $xmlData; 	
    }

    /**
     * Cleans text string from a feed before storage in our object
     * 
     * @param string $value
     * @return string
     *
     */
    protected function cleanText($value)
    {
        if(count($this->replacements))
        { 
            $value = str_ireplace(array_keys($this->replacements), array_values($this->replacements), $value); 
        }
        $value = trim(strip_tags($value));

        if(mb_strlen($value, "UTF-8") > 2048)
        {
            $value = mb_substr($value, 0, 2048, "UTF-8");
        }

        $sanitizer = $this->wire('sanitizer');
        $value = $sanitizer->unentities($value);
        $value = $sanitizer->entities($value);

        return $value; 
    }
}