af_comics: split filters to separate files to ease maintenance

This commit is contained in:
Andrew Dolgov 2014-02-17 13:00:25 +04:00
parent bf5e77a165
commit 901dd67a1a
10 changed files with 437 additions and 281 deletions

View File

@ -0,0 +1,14 @@
<?php
class Af_Comics_Template extends Af_ComicFilter {
function supported() {
return array("Example");
}
function process(&$article) {
$owner_uid = $article["owner_uid"];
return false;
}
}
?>

View File

@ -0,0 +1,6 @@
<?php
abstract class Af_ComicFilter {
public abstract function supported();
public abstract function process(&$article);
}
?>

View File

@ -0,0 +1,39 @@
<?php
class Af_Comics_Cad extends Af_ComicFilter {
function supported() {
return array("Ctrl+Alt+Del");
}
function process(&$article) {
$owner_uid = $article["owner_uid"];
if (strpos($article["link"], "cad-comic.com/cad/") !== 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);
$basenode = $xpath->query('(//img[contains(@src, "/comics/cad-")])')->item(0);
if ($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"];
}
return true;
}
return false;
}
}
?>

View File

@ -0,0 +1,50 @@
<?php
class Af_Comics_ComicPress extends Af_ComicFilter {
function supported() {
return array("Buni", "Buttersafe", "Whomp!", "Happy Jar", "CSection");
}
function process(&$article) {
$owner_uid = $article["owner_uid"];
if (strpos($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);
$basenode = false;
if ($doc) {
$xpath = new DOMXPath($doc);
$basenode = $xpath->query('//div[@id="comic"]')->item(0);
if ($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"];
}
return true;
}
return false;
}
}
?>

View File

@ -0,0 +1,52 @@
<?php
class Af_Comics_Dilbert extends Af_ComicFilter {
function supported() {
return array("Dilbert");
}
function process(&$article) {
$owner_uid = $article["owner_uid"];
if (strpos($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...
$matches = array();
foreach ($entries as $entry) {
if (preg_match("/dyn\/str_strip\/.*zoom\.gif$/", $entry->getAttribute("src"), $matches)) {
$entry->setAttribute("src",
rewrite_relative_url("http://dilbert.com/",
$matches[0]));
$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"];
}
return true;
}
return false;
}
}
?>

View File

@ -0,0 +1,49 @@
<?php
class Af_Comics_Explosm extends Af_ComicFilter {
function supported() {
return array("Cyanide and Happiness");
}
function process(&$article) {
$owner_uid = $article["owner_uid"];
if (strpos($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)) {
$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"];
}
return true;
}
return false;
}
}
?>

View File

@ -0,0 +1,62 @@
<?php
class Af_Comics_GoComics extends Af_ComicFilter {
function supported() {
return array("GoComics");
}
function process(&$article) {
$owner_uid = $article["owner_uid"];
if (strpos($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...
$matches = array();
foreach ($entries as $entry) {
if (preg_match("/(http:\/\/assets.amuniversal.com\/.*width.*)/i", $entry->getAttribute("src"), $matches)) {
$entry->setAttribute("src", $matches[0]);
$basenode = $entry;
break;
}
}
if (!$basenode) {
// fallback on the smaller version
foreach ($entries as $entry) {
if (preg_match("/(http:\/\/assets.amuniversal.com\/.*)/i", $entry->getAttribute("src"), $matches)) {
$entry->setAttribute("src", $matches[0]);
$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"];
}
return true;
}
return false;
}
}
?>

View File

@ -0,0 +1,89 @@
<?php
class Af_Comics_Pa extends Af_ComicFilter {
function supported() {
return array("Penny Arcade");
}
function process(&$article) {
$owner_uid = $article["owner_uid"];
if (strpos($article["link"], "penny-arcade.com") !== FALSE && strpos($article["title"], "Comic:") !== FALSE) {
if (strpos($article["plugin_data"], "af_comics,$owner_uid:") === FALSE) {
if ($debug_enabled) {
_debug("af_pennyarcade: Processing comic");
}
$doc = new DOMDocument();
$doc->loadHTML(fetch_file_contents($article["link"]));
$basenode = false;
if ($doc) {
$xpath = new DOMXPath($doc);
$basenode = $xpath->query('(//div[@id="comicFrame"])')->item(0);
if ($basenode) {
$article["content"] = $doc->saveXML($basenode);
$article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
}
}
} else if (isset($article["stored"]["content"])) {
$article["content"] = $article["stored"]["content"];
}
return true;
}
if (strpos($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");
}
$doc = new DOMDocument();
$doc->loadHTML(fetch_file_contents($article["link"]));
if ($doc) {
$xpath = new DOMXPath($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); }
$header = $xpath->query('(//div[@class="postBody"]/div[@class="comicPost"])')->item(0);
if ($header->parentNode) { $header->parentNode->removeChild($header); }
$avatar = $xpath->query('(//div[@class="avatar"]//img)')->item(0);
$basenode->insertBefore($avatar, $basenode->firstChild);
$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:" . $article["plugin_data"];
}
}
} else if (isset($article["stored"]["content"])) {
$article["content"] = $article["stored"]["content"];
}
return true;
}
return false;
}
}
?>

View File

@ -0,0 +1,39 @@
<?php
class Af_Comics_Twp extends Af_ComicFilter {
function supported() {
return array("Three Word Phrase");
}
function process(&$article) {
$owner_uid = $article["owner_uid"];
if (strpos($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"]));
$basenode = false;
if ($doc) {
$xpath = new DOMXpath($doc);
$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"];
}
return true;
}
return false;
}
}
?>

View File

@ -2,6 +2,7 @@
class Af_Comics extends Plugin {
private $host;
private $filters = array();
function about() {
return array(1.0,
@ -14,6 +15,22 @@ class Af_Comics extends Plugin {
$host->add_hook($host::HOOK_ARTICLE_FILTER, $this);
$host->add_hook($host::HOOK_PREFS_TAB, $this);
require_once __DIR__ . "/filter_base.php";
$filters = glob(__DIR__ . "/filters/*.php");
foreach ($filters as $file) {
require_once $file;
$filter_name = preg_replace("/\..*$/", "", basename($file));
$filter = new $filter_name();
if (is_subclass_of($filter, "Af_ComicFilter")) {
array_push($this->filters, $filter);
}
}
}
function hook_prefs_tab($args) {
@ -21,20 +38,24 @@ class Af_Comics extends Plugin {
print "<div dojoType=\"dijit.layout.AccordionPane\" title=\"".__('Feeds supported by af_comics')."\">";
print_notice("This plugin supports the following comics:");
print_notice("This plugin can be easily extended by adding filters to <code>plugins/af_comics/filters</code>.");
print "<p>" . __("The following comics are currently supported:") . "</p>";
$comics = array();
foreach ($this->filters as $f) {
foreach ($f->supported() as $comic) {
array_push($comics, $comic);
}
}
asort($comics);
print "<ul class=\"browseFeedList\" style=\"border-width : 1px\">";
print "<li>Buni</li>
<li>Buttersafe</li>
<li>CSection</li>
<li>Ctrl+Alt+Del</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>";
foreach ($comics as $comic) {
print "<li>$comic</li>";
}
print "</ul>";
print "</div>";
@ -43,278 +64,13 @@ class Af_Comics extends Plugin {
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);
$basenode = false;
if ($doc) {
$xpath = new DOMXPath($doc);
$basenode = $xpath->query('//div[@id="comic"]')->item(0);
if ($basenode) {
$article["content"] = $doc->saveXML($basenode);
$article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
}
}
} else if (isset($article["stored"]["content"])) {
$article["content"] = $article["stored"]["content"];
}
}
if (strpos($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...
$matches = array();
foreach ($entries as $entry) {
if (preg_match("/dyn\/str_strip\/.*zoom\.gif$/", $entry->getAttribute("src"), $matches)) {
$entry->setAttribute("src",
rewrite_relative_url("http://dilbert.com/",
$matches[0]));
$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"] = $article["stored"]["content"];
}
}
if (strpos($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)) {
$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"] = $article["stored"]["content"];
}
}
if (strpos($article["link"], "cad-comic.com/cad/") !== 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);
$basenode = $xpath->query('(//img[contains(@src, "/comics/cad-")])')->item(0);
if ($basenode) {
$article["content"] = $doc->saveXML($basenode);
$article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
}
}
} else if (isset($article["stored"]["content"])) {
$article["content"] = $article["stored"]["content"];
}
}
if (strpos($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...
$matches = array();
foreach ($entries as $entry) {
if (preg_match("/(http:\/\/assets.amuniversal.com\/.*width.*)/i", $entry->getAttribute("src"), $matches)) {
$entry->setAttribute("src", $matches[0]);
$basenode = $entry;
break;
}
}
if (!$basenode) {
// fallback on the smaller version
foreach ($entries as $entry) {
if (preg_match("/(http:\/\/assets.amuniversal.com\/.*)/i", $entry->getAttribute("src"), $matches)) {
$entry->setAttribute("src", $matches[0]);
$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"] = $article["stored"]["content"];
}
}
if (strpos($article["link"], "penny-arcade.com") !== FALSE && strpos($article["title"], "Comic:") !== FALSE) {
if (strpos($article["plugin_data"], "af_comics,$owner_uid:") === FALSE) {
if ($debug_enabled) {
_debug("af_pennyarcade: Processing comic");
}
$doc = new DOMDocument();
$doc->loadHTML(fetch_file_contents($article["link"]));
$basenode = false;
if ($doc) {
$xpath = new DOMXPath($doc);
$basenode = $xpath->query('(//div[@id="comicFrame"])')->item(0);
if ($basenode) {
$article["content"] = $doc->saveXML($basenode);
$article["plugin_data"] = "af_comics,$owner_uid:" . $article["plugin_data"];
}
}
} else if (isset($article["stored"]["content"])) {
$article["content"] = $article["stored"]["content"];
}
}
if (strpos($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");
}
$doc = new DOMDocument();
$doc->loadHTML(fetch_file_contents($article["link"]));
if ($doc) {
$xpath = new DOMXPath($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); }
$header = $xpath->query('(//div[@class="postBody"]/div[@class="comicPost"])')->item(0);
if ($header->parentNode) { $header->parentNode->removeChild($header); }
$avatar = $xpath->query('(//div[@class="avatar"]//img)')->item(0);
$basenode->insertBefore($avatar, $basenode->firstChild);
$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:" . $article["plugin_data"];
}
}
} else if (isset($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"]));
$basenode = false;
if ($doc) {
$xpath = new DOMXpath($doc);
$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"] = $article["stored"]["content"];
}
foreach ($this->filters as $f) {
if ($f->process($article))
break;
}
return $article;
}
function api_version() {