]> git.wh0rd.org - tt-rss.git/blobdiff - plugins/af_comics/init.php
Update af_comics to handle new GoComics site.
[tt-rss.git] / plugins / af_comics / init.php
index 88d5378141b57b17781dbb4658eed27598d61120..0c43f6aecb2453390a4032422d0421d2d509fac2 100644 (file)
@@ -2,9 +2,10 @@
 class Af_Comics extends Plugin {
 
        private $host;
+       private $filters = array();
 
        function about() {
-               return array(1.0,
+               return array(2.0,
                        "Fixes RSS feeds of assorted comic strips",
                        "fox");
        }
@@ -12,285 +13,128 @@ class Af_Comics extends Plugin {
        function init($host) {
                $this->host = $host;
 
+               $host->add_hook($host::HOOK_FETCH_FEED, $this);
                $host->add_hook($host::HOOK_ARTICLE_FILTER, $this);
                $host->add_hook($host::HOOK_PREFS_TAB, $this);
-       }
-
-       function hook_prefs_tab($args) {
-               if ($args != "prefPrefs") return;
-
-               print "<div dojoType=\"dijit.layout.AccordionPane\" title=\"".__('Feeds supported by af_comics')."\">";
 
-               print_notice("This plugin supports the following comics:");
-
-               print "<ul class=\"browseFeedList\" style=\"border-width : 1px\">";
-               print "<li>Buni</li>
-               <li>Buttersafe</li>
-               <li>CSection</li>
-               <li>Dilbert</li>
-               <li>Explosm</li>
-               <li>GoComics</li>
-               <li>Happy Jar</li>
-               <li>Penny Arcade</li>
-               <li>Three word phrase</li>
-               <li>Whomp</li>";
-               print "</ul>";
-
-               print "</div>";
-       }
+               require_once __DIR__ . "/filter_base.php";
 
-       function hook_article_filter($article) {
-               $owner_uid = $article["owner_uid"];
-
-               $found = false;
-
-               # div#comic - comicpress?
-
-               if (strpos($article["guid"], "bunicomic.com") !== FALSE ||
-                               strpos($article["guid"], "buttersafe.com") !== FALSE ||
-                               strpos($article["guid"], "whompcomic.com") !== FALSE ||
-                               strpos($article["guid"], "happyjar.com") !== FALSE ||
-                               strpos($article["guid"], "csectioncomics.com") !== FALSE) {
-
-                        if (strpos($article["plugin_data"], "af_comics,$owner_uid:") === FALSE) {
-
-
-                               // lol at people who block clients by user agent
-                               // oh noes my ad revenue Q_Q
-
-                               $res = fetch_file_contents($article["link"], false, false, false,
-                                        false, false, 0,
-                                        "Mozilla/5.0 (compatible; MSIE 10.0; Windows NT 6.1; WOW64; Trident/6.0)");
-
-                               $doc = new DOMDocument();
-                               @$doc->loadHTML($res);
+               $filters = glob(__DIR__ . "/filters/*.php");
 
-                               $basenode = false;
+               foreach ($filters as $file) {
+                       require_once $file;
+                       $filter_name = preg_replace("/\..*$/", "", basename($file));
 
-                               if ($doc) {
-                                       $xpath = new DOMXPath($doc);
-                                       $basenode = $xpath->query('//div[@id="comic"]')->item(0);
+                       $filter = new $filter_name();
 
-                                       if ($basenode) {
-                                               $article["content"] = $doc->saveXML($basenode);
-                                               $article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
-                                       }
-                               }
-                       } else if (isset($article["stored"]["content"])) {
-                               $article["content"] = $article["stored"]["content"];
+                       if (is_subclass_of($filter, "Af_ComicFilter")) {
+                               array_push($this->filters, $filter);
                        }
                }
 
-               if (strpos($article["guid"], "dilbert.com") !== FALSE) {
-                       if (strpos($article["plugin_data"], "af_comics,$owner_uid:") === FALSE) {
-                               $doc = new DOMDocument();
-                               @$doc->loadHTML(fetch_file_contents($article["link"]));
-
-                               $basenode = false;
-
-                               if ($doc) {
-                                       $xpath = new DOMXPath($doc);
-                                       $entries = $xpath->query('(//img[@src])'); // we might also check for img[@class='strip'] I guess...
+       }
 
-                                       $matches = array();
+       function hook_prefs_tab($args) {
+               if ($args != "prefPrefs") return;
 
-                                       foreach ($entries as $entry) {
+               print "<div dojoType=\"dijit.layout.AccordionPane\" title=\"".__('Feeds supported by af_comics')."\">";
 
-                                               if (preg_match("/dyn\/str_strip\/.*zoom\.gif$/", $entry->getAttribute("src"), $matches)) {
+               print "<p>" . __("The following comics are currently supported:") . "</p>";
 
-                                                       $entry->setAttribute("src",
-                                                               rewrite_relative_url("http://dilbert.com/",
-                                                               $matches[0]));
+               $comics = array("GoComics");
 
-                                                       $basenode = $entry;
-                                                       break;
-                                               }
-                                       }
-
-                                       if ($basenode) {
-                                               $article["content"] = $doc->saveXML($basenode);
-                                               $article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
-                                       }
-                               }
-                       } else if (isset($article["stored"]["content"])) {
-                               $article["content"] = $article["stored"]["content"];
+               foreach ($this->filters as $f) {
+                       foreach ($f->supported() as $comic) {
+                               array_push($comics, $comic);
                        }
                }
 
-               if (strpos($article["link"], "explosm.net/comics") !== FALSE) {
-                       if (strpos($article["plugin_data"], "af_comics,$owner_uid:") === FALSE) {
-
-                               $doc = new DOMDocument();
-                               @$doc->loadHTML(fetch_file_contents($article["link"]));
-
-                               $basenode = false;
-
-                               if ($doc) {
-                                       $xpath = new DOMXPath($doc);
-                                       $entries = $xpath->query('(//img[@src])'); // we might also check for img[@class='strip'] I guess...
-
-                                       $matches = array();
-
-                                       foreach ($entries as $entry) {
-
-                                               if (preg_match("/(http:\/\/.*\/db\/files\/Comics\/.*)/i", $entry->getAttribute("src"), $matches)) {
+               asort($comics);
 
-                                                       $basenode = $entry;
-                                                       break;
-                                               }
-                                       }
-
-                                       if ($basenode) {
-                                               $article["content"] = $doc->saveXML($basenode);
-                                               $article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
-                                       }
-                               }
-                       } else if (isset($article["stored"]["content"])) {
-                               $article["content"] = $article["stored"]["content"];
-                       }
+               print "<ul class=\"browseFeedList\" style=\"border-width : 1px\">";
+               foreach ($comics as $comic) {
+                       print "<li>$comic</li>";
                }
+               print "</ul>";
 
-               if (strpos($article["guid"], "gocomics.com") !== FALSE) {
-                       if (strpos($article["plugin_data"], "af_comics,$owner_uid:") === FALSE) {
-                               $doc = new DOMDocument();
-                               @$doc->loadHTML(fetch_file_contents($article["link"]));
-
-                               $basenode = false;
-
-                               if ($doc) {
-                                       $xpath = new DOMXPath($doc);
-                                       $entries = $xpath->query('(//img[@src])'); // we might also check for img[@class='strip'] I guess...
+               print "</div>";
+       }
 
-                                       $matches = array();
+       function hook_article_filter($article) {
+               $owner_uid = $article["owner_uid"];
 
-                                       foreach ($entries as $entry) {
+               foreach ($this->filters as $f) {
+                       if ($f->process($article))
+                               break;
+               }
 
-                                               if (preg_match("/(http:\/\/assets.amuniversal.com\/.*width.*)/i", $entry->getAttribute("src"), $matches)) {
+               return $article;
 
-                                                       $entry->setAttribute("src", $matches[0]);
-                                                       $basenode = $entry;
-                                                       break;
-                                               }
-                                       }
+       }
 
-                    if (!$basenode) {
-                        // fallback on the smaller version
-                        foreach ($entries as $entry) {
+       // GoComics dropped feed support so it needs to be handled when fetching the feed.
+       function hook_fetch_feed($feed_data, $fetch_url, $owner_uid, $feed, $last_article_timestamp, $auth_login, $auth_pass) {
+               if ($auth_login || $auth_pass)
+                       return $feed_data;
 
-                            if (preg_match("/(http:\/\/assets.amuniversal.com\/.*)/i", $entry->getAttribute("src"), $matches)) {
+               if (preg_match('#^https?://feeds.feedburner.com/uclick/([a-z]+)#', $fetch_url, $comic)) {
+                       $site_url = 'http://www.gocomics.com/' . $comic[1];
 
-                                $entry->setAttribute("src", $matches[0]);
-                                $basenode = $entry;
-                                break;
-                            }
-                        }
-                    }
+                       $article_link = $site_url . date('/Y/m/d');
 
-                                       if ($basenode) {
-                                               $article["content"] = $doc->saveXML($basenode);
-                                               $article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
-                                       }
-                               }
-                       } else if (isset($article["stored"]["content"])) {
-                               $article["content"] = $article["stored"]["content"];
-                       }
-               }
+                       $body = fetch_file_contents(array('url' => $article_link, 'type' => 'text/html', 'followlocation' => false));
 
-               if (strpos($article["link"], "penny-arcade.com") !== FALSE && strpos($article["title"], "Comic:") !== FALSE) {
-                       if (strpos($article["plugin_data"], "af_comics,$owner_uid:") === FALSE) {
+                       require_once 'lib/MiniTemplator.class.php';
 
-                               if ($debug_enabled) {
-                                       _debug("af_pennyarcade: Processing comic");
-                               }
+                       $feed_title = htmlspecialchars($comic[1]);
+                       $site_url = htmlspecialchars($site_url);
+                       $article_link = htmlspecialchars($article_link);
 
-                               $doc = new DOMDocument();
-                               $doc->loadHTML(fetch_file_contents($article["link"]));
+                       $tpl = new MiniTemplator();
 
-                               $basenode = false;
+                       $tpl->readTemplateFromFile('templates/generated_feed.txt');
 
-                               if ($doc) {
-                                       $xpath = new DOMXPath($doc);
-                                       $basenode = $xpath->query('(//div[@id="comicFrame"])')->item(0);
+                       $tpl->setVariable('FEED_TITLE', $feed_title, true);
+                       $tpl->setVariable('VERSION', VERSION, true);
+                       $tpl->setVariable('FEED_URL', htmlspecialchars($fetch_url), true);
+                       $tpl->setVariable('SELF_URL', $site_url, true);
 
-                                       if ($basenode) {
-                                               $article["content"] = $doc->saveXML($basenode);
-                                               $article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
-                                       }
-                               }
-                       } else if (isset($article["stored"]["content"])) {
-                               $article["content"] = $article["stored"]["content"];
-                       }
-               }
+                       $tpl->setVariable('ARTICLE_UPDATED_ATOM', date('c'), true);
+                       $tpl->setVariable('ARTICLE_UPDATED_RFC822', date(DATE_RFC822), true);
 
-               if (strpos($article["link"], "penny-arcade.com") !== FALSE && strpos($article["title"], "News Post:") !== FALSE) {
-                       if (strpos($article["plugin_data"], "af_comics,$owner_uid:") === FALSE) {
-                               if ($debug_enabled) {
-                                       _debug("af_pennyarcade: Processing news post");
-                               }
+                       if ($body) {
                                $doc = new DOMDocument();
-                               $doc->loadHTML(fetch_file_contents($article["link"]));
 
-                               if ($doc) {
+                               if (@$doc->loadHTML($body)) {
                                        $xpath = new DOMXPath($doc);
-                                       $entries = $xpath->query('(//div[@class="post"])');
-
-                                       $basenode = false;
-
-                                       foreach ($entries as $entry) {
-                                               $basenode = $entry;
-                                       }
-
-                                       $meta = $xpath->query('(//div[@class="meta"])')->item(0);
-                                       if ($meta->parentNode) { $meta->parentNode->removeChild($meta); }
 
-                                       $header = $xpath->query('(//div[@class="postBody"]/h2)')->item(0);
-                                       if ($header->parentNode) { $header->parentNode->removeChild($header); }
+                                       $node = $xpath->query('//picture[contains(@class, "item-comic-image")]/img')->item(0);
 
-                                       $header = $xpath->query('(//div[@class="postBody"]/div[@class="comicPost"])')->item(0);
-                                       if ($header->parentNode) { $header->parentNode->removeChild($header); }
+                                       if ($node) {
+                                               $tpl->setVariable('ARTICLE_ID', $article_link, true);
+                                               $tpl->setVariable('ARTICLE_LINK', $article_link, true);
+                                               $tpl->setVariable('ARTICLE_TITLE', date('l, F d, Y'), true);
+                                               $tpl->setVariable('ARTICLE_EXCERPT', '', true);
+                                               $tpl->setVariable('ARTICLE_CONTENT', $doc->saveXML($node), true);
 
-                                       $avatar = $xpath->query('(//div[@class="avatar"]//img)')->item(0);
-                                       $basenode->insertBefore($avatar, $basenode->firstChild);
+                                               $tpl->setVariable('ARTICLE_AUTHOR', '', true);
+                                               $tpl->setVariable('ARTICLE_SOURCE_LINK', $site_url, true);
+                                               $tpl->setVariable('ARTICLE_SOURCE_TITLE', $feed_title, true);
 
-                                       $uninteresting = $xpath->query('(//div[@class="avatar"])');
-                                       foreach ($uninteresting as $i) {
-                                               $i->parentNode->removeChild($i);
-                                       }
-
-                                       if ($basenode){
-                                               $article["content"] = $doc->saveXML($basenode);
-                                               $article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
+                                               $tpl->addBlock('entry');
                                        }
                                }
-                       } else if (isset($article["stored"]["content"])) {
-                               $article["content"] = $article["stored"]["content"];
                        }
-               }
-
-               if (strpos($article["link"], "threewordphrase.com") !== FALSE) {
-                       if (strpos($article["plugin_data"], "af_comics,$owner_uid:") === FALSE) {
-
-                               $doc = new DOMDocument();
-                               @$doc->loadHTML(fetch_file_contents($article["link"]));
 
-                               $basenode = false;
+                       $tpl->addBlock('feed');
 
-                               if ($doc) {
-                                       $xpath = new DOMXpath($doc);
+                       $tmp_data = '';
 
-                                       $basenode = $xpath->query("//td/center/img")->item(0);
-
-                                       if ($basenode) {
-                                               $article["content"] = $doc->saveXML($basenode);
-                                               $article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
-                                       }
-                               }
-                       } else if (isset($article["stored"]["content"])) {
-                               $article["content"] = $article["stored"]["content"];
-                       }
+                       if ($tpl->generateOutputToString($tmp_data))
+                               $feed_data = $tmp_data;
                }
 
-               return $article;
+               return $feed_data;
        }
 
        function api_version() {