From e7c53af6fa3ade7540f03ebc8d701bc5c70bf499 Mon Sep 17 00:00:00 2001
From: Hypolite Petovan <mrpetovan@gmail.com>
Date: Sun, 31 Dec 2017 20:58:09 -0500
Subject: [PATCH] Add Module\Oembed and Content\OEmbed

---
 include/bbcode.php     |   7 +-
 src/Content/OEmbed.php | 352 +++++++++++++++++++++++++++++++++++++++++
 src/Module/Oembed.php  |  53 +++++++
 src/ParseUrl.php       |   4 +-
 src/Protocol/DFRN.php  |   3 +-
 5 files changed, 413 insertions(+), 6 deletions(-)
 create mode 100644 src/Content/OEmbed.php
 create mode 100644 src/Module/Oembed.php

diff --git a/include/bbcode.php b/include/bbcode.php
index be59c18072..e672001937 100644
--- a/include/bbcode.php
+++ b/include/bbcode.php
@@ -2,6 +2,7 @@
 
 use Friendica\App;
 use Friendica\Content\Smilies;
+use Friendica\Content\OEmbed;
 use Friendica\Core\Cache;
 use Friendica\Core\System;
 use Friendica\Core\Config;
@@ -232,7 +233,7 @@ function tryoembed($match) {
 	$url = str_replace(array("http://www.youtube.com/", "http://player.vimeo.com/"),
 				array("https://www.youtube.com/", "https://player.vimeo.com/"), $url);
 
-	$o = oembed_fetch_url($url);
+	$o = OEmbed::fetchURL($url);
 
 	if (!is_object($o)) {
 		return $match[0];
@@ -246,7 +247,7 @@ function tryoembed($match) {
 		return $match[0];
 	}
 
-	$html = oembed_format_object($o);
+	$html = OEmbed::formatObject($o);
 
 	return $html;
 }
@@ -1263,7 +1264,7 @@ function bbcode($Text, $preserve_nl = false, $tryoembed = true, $simplehtml = fa
 //	$Text = preg_replace("/\[youtube\](.*?)\[\/youtube\]/", '<object width="425" height="350" type="application/x-shockwave-flash" data="http://www.youtube.com/v/$1" ><param name="movie" value="http://www.youtube.com/v/$1"></param><!--[if IE]><embed src="http://www.youtube.com/v/$1" type="application/x-shockwave-flash" width="425" height="350" /><![endif]--></object>', $Text);
 
 	// oembed tag
-	$Text = oembed_bbcode2html($Text);
+	$Text = OEmbed::BBCode2HTML($Text);
 
 	// Avoid triple linefeeds through oembed
 	$Text = str_replace("<br style='clear:left'></span><br /><br />", "<br style='clear:left'></span><br />", $Text);
diff --git a/src/Content/OEmbed.php b/src/Content/OEmbed.php
new file mode 100644
index 0000000000..db2c130eaf
--- /dev/null
+++ b/src/Content/OEmbed.php
@@ -0,0 +1,352 @@
+<?php
+
+/*
+ * To change this license header, choose License Headers in Project Properties.
+ * To change this template file, choose Tools | Templates
+ * and open the template in the editor.
+ */
+
+namespace Friendica\Content;
+
+use Friendica\Core\Cache;
+use Friendica\Core\System;
+use Friendica\ParseUrl;
+use Friendica\Core\Config;
+use Friendica\Database\DBM;
+use dba;
+use DOMDocument;
+use DOMXPath;
+use DOMNode;
+
+require_once 'include/dba.php';
+require_once 'mod/proxy.php';
+
+/**
+ * Description of OEmbed
+ *
+ * @author benlo
+ */
+class OEmbed
+{
+	public static function replaceCallback($matches)
+	{
+		$embedurl = $matches[1];
+		$j = OEmbed::fetchURL($embedurl);
+		$s = OEmbed::formatObject($j);
+
+		return $s;
+	}
+
+	/**
+	 * @brief Get data from an URL to embed its content.
+	 *
+	 * @param string $embedurl The URL from which the data should be fetched.
+	 * @param bool $no_rich_type If set to true rich type content won't be fetched.
+	 *
+	 * @return bool|object Returns object with embed content or false if no embedable
+	 * 	 content exists
+	 */
+	public static function fetchURL($embedurl, $no_rich_type = false)
+	{
+		$embedurl = trim($embedurl, "'");
+		$embedurl = trim($embedurl, '"');
+
+		$a = get_app();
+
+		$condition = array('url' => normalise_link($embedurl));
+		$r = dba::select('oembed', array('content'), $condition, array('limit' => 1));
+
+		if (DBM::is_result($r)) {
+			$txt = $r["content"];
+		} else {
+			$txt = Cache::get($a->videowidth . $embedurl);
+		}
+		// These media files should now be caught in bbcode.php
+		// left here as a fallback in case this is called from another source
+
+		$noexts = array("mp3", "mp4", "ogg", "ogv", "oga", "ogm", "webm");
+		$ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
+
+
+		if (is_null($txt)) {
+			$txt = "";
+
+			if (!in_array($ext, $noexts)) {
+				// try oembed autodiscovery
+				$redirects = 0;
+				$html_text = fetch_url($embedurl, false, $redirects, 15, "text/*");
+				if ($html_text) {
+					$dom = @DOMDocument::loadHTML($html_text);
+					if ($dom) {
+						$xpath = new DOMXPath($dom);
+						$entries = $xpath->query("//link[@type='application/json+oembed']");
+						foreach ($entries as $e) {
+							$href = $e->getAttributeNode("href")->nodeValue;
+							$txt = fetch_url($href . '&maxwidth=' . $a->videowidth);
+							break;
+						}
+						$entries = $xpath->query("//link[@type='text/json+oembed']");
+						foreach ($entries as $e) {
+							$href = $e->getAttributeNode("href")->nodeValue;
+							$txt = fetch_url($href . '&maxwidth=' . $a->videowidth);
+							break;
+						}
+					}
+				}
+			}
+
+			$txt = trim($txt);
+
+			if ($txt[0] != "{") {
+				$txt = '{"type":"error"}';
+			} else { //save in cache
+				$j = json_decode($txt);
+				if ($j->type != "error") {
+					dba::insert('oembed', array('url' => normalise_link($embedurl),
+						'content' => $txt, 'created' => datetime_convert()), true);
+				}
+
+				Cache::set($a->videowidth . $embedurl, $txt, CACHE_DAY);
+			}
+		}
+
+		$j = json_decode($txt);
+
+		if (!is_object($j)) {
+			return false;
+		}
+
+		// Always embed the SSL version
+		if (isset($j->html)) {
+			$j->html = str_replace(array("http://www.youtube.com/", "http://player.vimeo.com/"), array("https://www.youtube.com/", "https://player.vimeo.com/"), $j->html);
+		}
+
+		$j->embedurl = $embedurl;
+
+		// If fetching information doesn't work, then improve via internal functions
+		if (($j->type == "error") || ($no_rich_type && ($j->type == "rich"))) {
+			$data = ParseUrl::getSiteinfoCached($embedurl, true, false);
+			$j->type = $data["type"];
+
+			if ($j->type == "photo") {
+				$j->url = $data["url"];
+				//$j->width = $data["images"][0]["width"];
+				//$j->height = $data["images"][0]["height"];
+			}
+
+			if (isset($data["title"])) {
+				$j->title = $data["title"];
+			}
+
+			if (isset($data["text"])) {
+				$j->description = $data["text"];
+			}
+
+			if (is_array($data["images"])) {
+				$j->thumbnail_url = $data["images"][0]["src"];
+				$j->thumbnail_width = $data["images"][0]["width"];
+				$j->thumbnail_height = $data["images"][0]["height"];
+			}
+		}
+
+		call_hooks('oembed_fetch_url', $embedurl, $j);
+
+		return $j;
+	}
+
+	public static function formatObject($j)
+	{
+		$embedurl = $j->embedurl;
+		$jhtml = OEmbed::iframe($j->embedurl, (isset($j->width) ? $j->width : null), (isset($j->height) ? $j->height : null));
+		$ret = "<span class='oembed " . $j->type . "'>";
+		switch ($j->type) {
+			case "video":
+				if (isset($j->thumbnail_url)) {
+					$tw = (isset($j->thumbnail_width) && intval($j->thumbnail_width)) ? $j->thumbnail_width : 200;
+					$th = (isset($j->thumbnail_height) && intval($j->thumbnail_height)) ? $j->thumbnail_height : 180;
+					// make sure we don't attempt divide by zero, fallback is a 1:1 ratio
+					$tr = (($th) ? $tw / $th : 1);
+
+					$th = 120;
+					$tw = $th * $tr;
+					$tpl = get_markup_template('oembed_video.tpl');
+					$ret.=replace_macros($tpl, array(
+						'$baseurl' => System::baseUrl(),
+						'$embedurl' => $embedurl,
+						'$escapedhtml' => base64_encode($jhtml),
+						'$tw' => $tw,
+						'$th' => $th,
+						'$turl' => $j->thumbnail_url,
+					));
+				} else {
+					$ret = $jhtml;
+				}
+				//$ret.="<br>";
+				break;
+			case "photo":
+				$ret.= "<img width='" . $j->width . "' src='" . proxy_url($j->url) . "'>";
+				break;
+			case "link":
+				break;
+			case "rich":
+				// not so safe..
+				if (!Config::get("system", "no_oembed_rich_content")) {
+					$ret.= proxy_parse_html($jhtml);
+				}
+				break;
+		}
+
+		// add link to source if not present in "rich" type
+		if ($j->type != 'rich' || !strpos($j->html, $embedurl)) {
+			$ret .= "<h4>";
+			if (isset($j->title)) {
+				if (isset($j->provider_name)) {
+					$ret .= $j->provider_name . ": ";
+				}
+
+				$embedlink = (isset($j->title)) ? $j->title : $embedurl;
+				$ret .= "<a href='$embedurl' rel='oembed'>$embedlink</a>";
+				if (isset($j->author_name)) {
+					$ret.=" (" . $j->author_name . ")";
+				}
+			} elseif (isset($j->provider_name) || isset($j->author_name)) {
+				$embedlink = "";
+				if (isset($j->provider_name)) {
+					$embedlink .= $j->provider_name;
+				}
+
+				if (isset($j->author_name)) {
+					if ($embedlink != "") {
+						$embedlink .= ": ";
+					}
+
+					$embedlink .= $j->author_name;
+				}
+				if (trim($embedlink) == "") {
+					$embedlink = $embedurl;
+				}
+
+				$ret .= "<a href='$embedurl' rel='oembed'>$embedlink</a>";
+			}
+			//if (isset($j->author_name)) $ret.=" by ".$j->author_name;
+			//if (isset($j->provider_name)) $ret.=" on ".$j->provider_name;
+			$ret .= "</h4>";
+		} else {
+			// add <a> for html2bbcode conversion
+			$ret .= "<a href='$embedurl' rel='oembed'>$embedurl</a>";
+		}
+		$ret.="</span>";
+		$ret = str_replace("\n", "", $ret);
+		return mb_convert_encoding($ret, 'HTML-ENTITIES', mb_detect_encoding($ret));
+	}
+
+	public static function BBCode2HTML($text)
+	{
+		$stopoembed = Config::get("system", "no_oembed");
+		if ($stopoembed == true) {
+			return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
+		}
+		return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
+	}
+
+	/**
+	 * Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
+	 * and replace it with [embed]url[/embed]
+	 */
+	public static function HTML2BBCode($text)
+	{
+		// start parser only if 'oembed' is in text
+		if (strpos($text, "oembed")) {
+
+			// convert non ascii chars to html entities
+			$html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
+
+			// If it doesn't parse at all, just return the text.
+			$dom = @DOMDocument::loadHTML($html_text);
+			if (!$dom) {
+				return $text;
+			}
+			$xpath = new DOMXPath($dom);
+
+			$xattr = OEmbed::buildXPath("class", "oembed");
+			$entries = $xpath->query("//span[$xattr]");
+
+			$xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
+			foreach ($entries as $e) {
+				$href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
+				if (!is_null($href)) {
+					$e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
+				}
+			}
+			return OEmbed::getInnerHTML($dom->getElementsByTagName("body")->item(0));
+		} else {
+			return $text;
+		}
+	}
+
+	/**
+	 * @brief Generates the iframe HTML for an oembed attachment.
+	 *
+	 * Width and height are given by the remote, and are regularly too small for
+	 * the generated iframe.
+	 *
+	 * The width is entirely discarded for the actual width of the post, while fixed
+	 * height is used as a starting point before the inevitable resizing.
+	 *
+	 * Since the iframe is automatically resized on load, there are no need for ugly
+	 * and impractical scrollbars.
+	 *
+	 * @param string $src Original remote URL to embed
+	 * @param string $width
+	 * @param string $height
+	 * @return string formatted HTML
+	 *
+	 * @see oembed_format_object()
+	 */
+	private static function iframe($src, $width, $height)
+	{
+		$a = get_app();
+
+		if (!$height || strstr($height, '%')) {
+			$height = '200';
+		}
+		$width = '100%';
+
+		$s = System::baseUrl() . '/oembed/' . base64url_encode($src);
+		return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $s . '" allowfullscreen scrolling="no" frameborder="no">' . t('Embedded content') . '</iframe>';
+	}
+
+	/**
+	 * Generates an XPath query to select elements whose provided attribute contains
+	 * the provided value in a space-separated list.
+	 *
+	 * @brief Generates attribute search XPath string
+	 *
+	 * @param string $attr Name of the attribute to seach
+	 * @param string $value Value to search in a space-separated list
+	 * @return string
+	 */
+	private static function buildXPath($attr, $value)
+	{
+		// https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
+		return "contains( normalize-space( @$attr ), ' $value ' ) or substring( normalize-space( @$attr ), 1, string-length( '$value' ) + 1 ) = '$value ' or substring( normalize-space( @$attr ), string-length( @$attr ) - string-length( '$value' ) ) = ' $value' or @$attr = '$value'";
+	}
+
+	/**
+	 * Returns the inner XML string of a provided DOMNode
+	 *
+	 * @brief Returns the inner XML string of a provided DOMNode
+	 *
+	 * @param DOMNode $node
+	 * @return string
+	 */
+	private static function getInnerHTML(DOMNode $node)
+	{
+		$innerHTML = '';
+		$children = $node->childNodes;
+		foreach ($children as $child) {
+			$innerHTML .= $child->ownerDocument->saveXML($child);
+		}
+		return $innerHTML;
+	}
+}
diff --git a/src/Module/Oembed.php b/src/Module/Oembed.php
new file mode 100644
index 0000000000..f30fb861bc
--- /dev/null
+++ b/src/Module/Oembed.php
@@ -0,0 +1,53 @@
+<?php
+
+namespace Friendica\Module;
+
+use Friendica\BaseModule;
+use Friendica\Content;
+
+/**
+ * Oembed module
+ *
+ * Displays stored embed content based on a base64 hash of a remote URL
+ *
+ * Example: /oembed/aHR0cHM6Ly9...
+ *
+ * @author Hypolite Petovan <mrpetovan@gmail.com>
+ */
+class Oembed extends BaseModule
+{
+	public static function content()
+	{
+		$a = self::getApp();
+
+		// Unused form: /oembed/b2h?url=...
+		if ($a->argv[1] == 'b2h') {
+			$url = array("", trim(hex2bin($_REQUEST['url'])));
+			echo Content\OEmbed::replaceCallback($url);
+			killme();
+		}
+
+		// Unused form: /oembed/h2b?text=...
+		if ($a->argv[1] == 'h2b') {
+			$text = trim(hex2bin($_REQUEST['text']));
+			echo Content\OEmbed::HTML2BBCode($text);
+			killme();
+		}
+
+		if ($a->argc == 2) {
+			echo '<html><body>';
+			$url = base64url_decode($a->argv[1]);
+			$j = Content\OEmbed::fetchURL($url);
+
+			// workaround for media.ccc.de (and any other endpoint that return size 0)
+			if (substr($j->html, 0, 7) == "<iframe" && strstr($j->html, 'width="0"')) {
+				$j->html = '<style>html,body{margin:0;padding:0;} iframe{width:100%;height:100%;}</style>' . $j->html;
+				$j->html = str_replace('width="0"', '', $j->html);
+				$j->html = str_replace('height="0"', '', $j->html);
+			}
+			echo $j->html;
+			echo '</body></html>';
+		}
+		killme();
+	}
+}
diff --git a/src/ParseUrl.php b/src/ParseUrl.php
index 9e46281ec9..6252e0b68a 100644
--- a/src/ParseUrl.php
+++ b/src/ParseUrl.php
@@ -5,7 +5,7 @@
  */
 namespace Friendica;
 
-use Friendica\Core\Config;
+use Friendica\Content\OEmbed;
 use Friendica\Object\Image;
 use Friendica\Util\XML;
 
@@ -164,7 +164,7 @@ class ParseUrl
 		$body = $data["body"];
 
 		if ($do_oembed) {
-			$oembed_data = oembed_fetch_url($url);
+			$oembed_data = OEmbed::fetchURL($url);
 
 			if (!in_array($oembed_data->type, array("error", "rich", ""))) {
 				$siteinfo["type"] = $oembed_data->type;
diff --git a/src/Protocol/DFRN.php b/src/Protocol/DFRN.php
index 5e9c91645b..1ec0c792e0 100644
--- a/src/Protocol/DFRN.php
+++ b/src/Protocol/DFRN.php
@@ -8,6 +8,7 @@
  */
 namespace Friendica\Protocol;
 
+use Friendica\Content\OEmbed;
 use Friendica\Core\Config;
 use Friendica\Core\System;
 use Friendica\Core\Worker;
@@ -2502,7 +2503,7 @@ class DFRN
 
 			$item['body'] = html2bb_video($item['body']);
 
-			$item['body'] = oembed_html2bbcode($item['body']);
+			$item['body'] = OEmbed::HTML2BBCode($item['body']);
 
 			$config = \HTMLPurifier_Config::createDefault();
 			$config->set('Cache.DefinitionImpl', null);
-- 
2.39.5