From 901dd67a1a756f5c4e03cc452c40b39584184474 Mon Sep 17 00:00:00 2001 From: Andrew Dolgov Date: Mon, 17 Feb 2014 13:00:25 +0400 Subject: af_comics: split filters to separate files to ease maintenance --- plugins/af_comics/filters/af_comics_cad.php | 39 ++++++++++ plugins/af_comics/filters/af_comics_comicpress.php | 50 ++++++++++++ plugins/af_comics/filters/af_comics_dilbert.php | 52 +++++++++++++ plugins/af_comics/filters/af_comics_explosm.php | 49 ++++++++++++ plugins/af_comics/filters/af_comics_gocomics.php | 62 +++++++++++++++ plugins/af_comics/filters/af_comics_pa.php | 89 ++++++++++++++++++++++ plugins/af_comics/filters/af_comics_twp.php | 39 ++++++++++ 7 files changed, 380 insertions(+) create mode 100644 plugins/af_comics/filters/af_comics_cad.php create mode 100644 plugins/af_comics/filters/af_comics_comicpress.php create mode 100644 plugins/af_comics/filters/af_comics_dilbert.php create mode 100644 plugins/af_comics/filters/af_comics_explosm.php create mode 100644 plugins/af_comics/filters/af_comics_gocomics.php create mode 100644 plugins/af_comics/filters/af_comics_pa.php create mode 100644 plugins/af_comics/filters/af_comics_twp.php (limited to 'plugins/af_comics/filters') diff --git a/plugins/af_comics/filters/af_comics_cad.php b/plugins/af_comics/filters/af_comics_cad.php new file mode 100644 index 000000000..6184977d1 --- /dev/null +++ b/plugins/af_comics/filters/af_comics_cad.php @@ -0,0 +1,39 @@ +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"] = $article["stored"]["content"]; + } + + return true; + } + + return false; + } +} +?> diff --git a/plugins/af_comics/filters/af_comics_comicpress.php b/plugins/af_comics/filters/af_comics_comicpress.php new file mode 100644 index 000000000..e60b2dfe5 --- /dev/null +++ b/plugins/af_comics/filters/af_comics_comicpress.php @@ -0,0 +1,50 @@ +loadHTML($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"] = $article["stored"]["content"]; + } + + return true; + } + + return false; + } +} +?> diff --git a/plugins/af_comics/filters/af_comics_dilbert.php b/plugins/af_comics/filters/af_comics_dilbert.php new file mode 100644 index 000000000..4c2233a3a --- /dev/null +++ b/plugins/af_comics/filters/af_comics_dilbert.php @@ -0,0 +1,52 @@ +loadHTML(fetch_file_contents($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"] = $article["stored"]["content"]; + } + + return true; + } + + return false; + } +} +?> diff --git a/plugins/af_comics/filters/af_comics_explosm.php b/plugins/af_comics/filters/af_comics_explosm.php new file mode 100644 index 000000000..7e0621852 --- /dev/null +++ b/plugins/af_comics/filters/af_comics_explosm.php @@ -0,0 +1,49 @@ +loadHTML(fetch_file_contents($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"] = $article["stored"]["content"]; + } + + return true; + } + + return false; + } +} +?> diff --git a/plugins/af_comics/filters/af_comics_gocomics.php b/plugins/af_comics/filters/af_comics_gocomics.php new file mode 100644 index 000000000..bd57c08ff --- /dev/null +++ b/plugins/af_comics/filters/af_comics_gocomics.php @@ -0,0 +1,62 @@ +loadHTML(fetch_file_contents($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"] = $article["stored"]["content"]; + } + + return true; + } + + return false; + } +} +?> diff --git a/plugins/af_comics/filters/af_comics_pa.php b/plugins/af_comics/filters/af_comics_pa.php new file mode 100644 index 000000000..b44de9cff --- /dev/null +++ b/plugins/af_comics/filters/af_comics_pa.php @@ -0,0 +1,89 @@ +loadHTML(fetch_file_contents($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"] = $article["stored"]["content"]; + } + + return true; + } + + return false; + } +} +?> diff --git a/plugins/af_comics/filters/af_comics_twp.php b/plugins/af_comics/filters/af_comics_twp.php new file mode 100644 index 000000000..156b2b687 --- /dev/null +++ b/plugins/af_comics/filters/af_comics_twp.php @@ -0,0 +1,39 @@ +loadHTML(fetch_file_contents($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; + } +} +?> -- cgit v1.2.3