friendica-github/mod/parse_url.php

<?php
/** 
 * @file mod/parse_url.php
 * @brief The parse_url module
 * 
 * This module does parse an url for embedable content (audio, video, image files or link)
 * information and does format this information to BBCode or html (this depends
 * on the user settings - default is BBCode output).
 * If the user has enabled the richtext editor setting the output will be in html
 * (Note: This is not always possible and in some case not useful because
 * the richtext editor doesn't support all kind of html).
 * Otherwise the output will be constructed BBCode.
 * 
 * @todo https://developers.google.com/+/plugins/snippet/
 * 
 * @verbatim
 * <meta itemprop="name" content="Toller Titel">
 * <meta itemprop="description" content="Eine tolle Beschreibung">
 * <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
 * 
 * <body itemscope itemtype="http://schema.org/Product">
 *   <h1 itemprop="name">Shiny Trinket</h1>
 *   <img itemprop="image" src="{image-url}" />
 *   <p itemprop="description">Shiny trinkets are shiny.</p>
 * </body>
 * @endverbatim
*/

use \Friendica\ParseUrl;

require_once("include/items.php");

function parse_url_content(&$a) {

	$text = null;
	$str_tags = "";

	$textmode = false;

	if (local_user() && (!feature_enabled(local_user(), "richtext"))) {
		$textmode = true;
	}

	$br = (($textmode) ? "\n" : "<br />");

	if (x($_GET,"binurl")) {
		$url = trim(hex2bin($_GET["binurl"]));
	} else {
		$url = trim($_GET["url"]);
	}

	if ($_GET["title"]) {
		$title = strip_tags(trim($_GET["title"]));
	}

	if ($_GET["description"]) {
		$text = strip_tags(trim($_GET["description"]));
	}

	if ($_GET["tags"]) {
		$arr_tags = ParseUrl::convertTagsToArray($_GET["tags"]);
		if (count($arr_tags)) {
			$str_tags = $br . implode(" ", $arr_tags) . $br;
		}
	}

	// Add url scheme if it is missing
	$arrurl = parse_url($url);
	if (!x($arrurl, "scheme")) {
		if (x($arrurl, "host")) {
			$url = "http:".$url;
		} else {
			$url = "http://".$url;
		}
	}

	logger("prse_url: " . $url);

	// Check if the URL is an image, video or audio file. If so format
	// the URL with the corresponding BBCode media tag
	$redirects = 0;
	// Fetch the header of the URL
	$result = z_fetch_url($url, false, $redirects, array("novalidate" => true, "nobody" => true));
	if($result["success"]) {
		// Convert the header fields into an array
		$hdrs = array();
		$h = explode("\n", $result["header"]);
		foreach ($h as $l) {
			list($k,$v) = array_map("trim", explode(":", trim($l), 2));
			$hdrs[$k] = $v;
		}
		if (array_key_exists("Content-Type", $hdrs)) {
			$type = $hdrs["Content-Type"];
		}
		if ($type) {
			if(stripos($type, "image/") !== false) {
				echo $br . "[img]" . $url . "[/img]" . $br;
				killme();
			}
			if (stripos($type, "video/") !== false) {
				echo $br . "[video]" . $url . "[/video]" . $br;
				killme();
			}
			if (stripos($type, "audio/") !== false) {
				echo $br . "[audio]" . $url . "[/audio]" . $br;
				killme();
			}
		}
	}

	if ($textmode) {
		$template = "[bookmark=%s]%s[/bookmark]%s";
	} else {
		$template = "<a class=\"bookmark\" href=\"%s\" >%s</a>%s";
	}

	$arr = array("url" => $url, "text" => "");

	call_hooks("parse_link", $arr);

	if (strlen($arr["text"])) {
		echo $arr["text"];
		killme();
	}

	// If there is allready some content information submitted we don't
	// need to parse the url for content.
	if ($url && $title && $text) {

		$title = str_replace(array("\r","\n"),array("",""),$title);

		if ($textmode) {
			$text = "[quote]" . trim($text) . "[/quote]" . $br;
		} else {
			$text = "<blockquote>" . htmlspecialchars(trim($text)) . "</blockquote><br />";
			$title = htmlspecialchars($title);
		}

		$result = sprintf($template, $url, ($title) ? $title : $url, $text) . $str_tags;

		logger("parse_url (unparsed): returns: " . $result);

		echo $result;
		killme();
	}

	// Fetch the information directly from the webpage
	$siteinfo = ParseUrl::getSiteinfo($url);

	unset($siteinfo["keywords"]);

	// Format it as BBCode attachment
	$info = add_page_info_data($siteinfo);

	if (!$textmode) {
		// Replace ' with ’ - not perfect - but the richtext editor has problems otherwise
		$info = str_replace(array("&#039;"), array("&#8217;"), $info);
	}

	echo $info;

	killme();
}
-												ajaxify the wall poster

											
										
										
											2010-07-23 05:41:45 +00:00
+								<?php
-												Revert "Updated modules to allow for partial overrides without errors"

This reverts commit db949bb802448184bfe5164d8d3dd86ddf51b187.

											
										
										
											2016-02-07 14:11:34 +00:00
+								/**
-												reformat todo according to doxygen style
											
										
										
											2015-12-25 22:17:34 +00:00
+								 * @file mod/parse_url.php
-												parse_url: add some docu
											
										
										
											2016-11-25 15:59:31 +00:00
+								 * @brief The parse_url module
 								 *
 								 * This module does parse an url for embedable content (audio, video, image files or link)
 								 * information and does format this information to BBCode or html (this depends
 								 * on the user settings - default is BBCode output).
 								 * If the user has enabled the richtext editor setting the output will be in html
 								 * (Note: This is not always possible and in some case not useful because
-												parse_url: fix typo
											
										
										
											2016-11-26 03:10:50 +00:00
+								 * the richtext editor doesn't support all kind of html).
-												parse_url: add some docu
											
										
										
											2016-11-25 15:59:31 +00:00
+								 * Otherwise the output will be constructed BBCode.
-												Revert "Updated modules to allow for partial overrides without errors"

This reverts commit db949bb802448184bfe5164d8d3dd86ddf51b187.

											
										
										
											2016-02-07 14:11:34 +00:00
+								 *
-												reformat todo according to doxygen style
											
										
										
											2015-12-25 22:17:34 +00:00
+								 * @todo https://developers.google.com/+/plugins/snippet/
-												Revert "Updated modules to allow for partial overrides without errors"

This reverts commit db949bb802448184bfe5164d8d3dd86ddf51b187.

											
										
										
											2016-02-07 14:11:34 +00:00
+								 *
-												reformat todo according to doxygen style
											
										
										
											2015-12-25 22:17:34 +00:00
+								 * @verbatim
 								 * <meta itemprop="name" content="Toller Titel">
 								 * <meta itemprop="description" content="Eine tolle Beschreibung">
 								 * <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
-												Revert "Updated modules to allow for partial overrides without errors"

This reverts commit db949bb802448184bfe5164d8d3dd86ddf51b187.

											
										
										
											2016-02-07 14:11:34 +00:00
+								 *
-												reformat todo according to doxygen style
											
										
										
											2015-12-25 22:17:34 +00:00
+								 * <body itemscope itemtype="http://schema.org/Product">
 								 *   <h1 itemprop="name">Shiny Trinket</h1>
 								 *   <img itemprop="image" src="{image-url}" />
 								 *   <p itemprop="description">Shiny trinkets are shiny.</p>
 								 * </body>
 								 * @endverbatim
-												parse_url: Just added a comment for future elements that should be added as well.

											
										
										
											2012-07-18 19:06:38 +00:00
+								*/
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+								use \Friendica\ParseUrl;
-												update writable for dfrn contacts when using local delivery

											
										
										
											2012-09-05 23:26:11 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+								require_once("include/items.php");
-												parse_url: Further improvements of the new method to fetch page data

											
										
										
											2012-07-12 21:41:04 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+								function parse_url_content(&$a) {
-												OEmbed and parse_url are now cached in dedicated tables

											
										
										
											2016-01-14 22:59:51 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									$text = null;
 									$str_tags = "";
-												OEmbed and parse_url are now cached in dedicated tables

											
										
										
											2016-01-14 22:59:51 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									$textmode = false;
-												OEmbed and parse_url are now cached in dedicated tables

											
										
										
											2016-01-14 22:59:51 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									if (local_user() && (!feature_enabled(local_user(), "richtext"))) {
 										$textmode = true;
-												Poco: New fields for last contact and last failure of a contact.

											
										
										
											2015-07-16 23:08:28 +00:00
+									}
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									$br = (($textmode) ? "\n" : "<br />");
-												Support for twitter cards (in both directions)

											
										
										
											2014-02-22 14:46:19 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									if (x($_GET,"binurl")) {
 										$url = trim(hex2bin($_GET["binurl"]));
 									} else {
 										$url = trim($_GET["url"]);
-												Improved handling with wrong formatted urls and fetchinfo

											
										
										
											2016-11-12 20:23:00 +00:00
+									}
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									if ($_GET["title"]) {
 										$title = strip_tags(trim($_GET["title"]));
-												Preventing an endless loop while fetching site infos

											
										
										
											2014-07-14 06:37:40 +00:00
+									}
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									if ($_GET["description"]) {
 										$text = strip_tags(trim($_GET["description"]));
-												Suppoorting Open Graph and Dublin Core when displaying single item
Following link shorteners in parse_url

											
										
										
											2013-03-02 13:46:06 +00:00
+									}
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									if ($_GET["tags"]) {
 										$arr_tags = ParseUrl::convertTagsToArray($_GET["tags"]);
 										if (count($arr_tags)) {
 											$str_tags = $br . implode(" ", $arr_tags) . $br;
-												The order for the page parsing is changed.

											
										
										
											2015-04-05 18:40:31 +00:00
+										}
 									}
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									// Add url scheme if it is missing
 									$arrurl = parse_url($url);
 									if (!x($arrurl, "scheme")) {
 										if (x($arrurl, "host")) {
 											$url = "http:".$url;
 										} else {
 											$url = "http://".$url;
-												Added some more timestamp measuring

											
										
										
											2015-03-07 22:14:26 +00:00
+										}
-												community: Only show top postings, no comments
parse_url: Follow location in "meta"-header

											
										
										
											2013-02-24 11:54:53 +00:00
+									}
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									logger("prse_url: " . $url);
-												parse_url: add some docu
											
										
										
											2016-11-25 15:59:31 +00:00
+									// Check if the URL is an image, video or audio file. If so format
 									// the URL with the corresponding BBCode media tag
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									$redirects = 0;
 									// Fetch the header of the URL
 									$result = z_fetch_url($url, false, $redirects, array("novalidate" => true, "nobody" => true));
 									if($result["success"]) {
 										// Convert the header fields into an array
 										$hdrs = array();
 										$h = explode("\n", $result["header"]);
 										foreach ($h as $l) {
 											list($k,$v) = array_map("trim", explode(":", trim($l), 2));
 											$hdrs[$k] = $v;
 										}
 										if (array_key_exists("Content-Type", $hdrs)) {
 											$type = $hdrs["Content-Type"];
 										}
 										if ($type) {
 											if(stripos($type, "image/") !== false) {
 												echo $br . "[img]" . $url . "[/img]" . $br;
 												killme();
-												parse url: Emtpy values in meta headers are now ignored. Oembed data has priority. Videos are embedded with some further information.

											
										
										
											2014-04-04 08:58:31 +00:00
+											}
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+											if (stripos($type, "video/") !== false) {
 												echo $br . "[video]" . $url . "[/video]" . $br;
 												killme();
-												parse url: Emtpy values in meta headers are now ignored. Oembed data has priority. Videos are embedded with some further information.

											
										
										
											2014-04-04 08:58:31 +00:00
+											}
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+											if (stripos($type, "audio/") !== false) {
 												echo $br . "[audio]" . $url . "[/audio]" . $br;
 												killme();
-												parse_url: Fetch multiple pictures so that the user can decide what to take.

											
										
										
											2012-07-12 06:20:27 +00:00
+											}
-												Added some more timestamp measuring

											
										
										
											2015-03-07 22:14:26 +00:00
+										}
-												parse_url: Fetch multiple pictures so that the user can decide what to take.

											
										
										
											2012-07-12 06:20:27 +00:00
+									}
-												parse_url: complete new code for fetching website information

											
										
										
											2012-07-11 23:17:33 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									if ($textmode) {
 										$template = "[bookmark=%s]%s[/bookmark]%s";
 									} else {
-												Support for twitter cards (in both directions)

											
										
										
											2014-02-22 14:46:19 +00:00
+										$template = "<a class=\"bookmark\" href=\"%s\" >%s</a>%s";
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									}
-												ajaxify the wall poster

											
										
										
											2010-07-23 05:41:45 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									$arr = array("url" => $url, "text" => "");
-												more plugin hooks

											
										
										
											2010-12-25 23:01:02 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									call_hooks("parse_link", $arr);
-												more plugin hooks

											
										
										
											2010-12-25 23:01:02 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									if (strlen($arr["text"])) {
 										echo $arr["text"];
-												more plugin hooks

											
										
										
											2010-12-25 23:01:02 +00:00
+										killme();
 									}
-												parse_url: add some docu
											
										
										
											2016-11-25 15:59:31 +00:00
+									// If there is allready some content information submitted we don't
 									// need to parse the url for content.
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									if ($url && $title && $text) {
-												allow setting all the bookmarked URL fields via script/API

											
										
										
											2011-09-20 05:21:55 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+										$title = str_replace(array("\r","\n"),array("",""),$title);
-												parse url: Characters like < and > has to be escaped when showing the parsed output.

											
										
										
											2014-02-16 16:35:01 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+										if ($textmode) {
 											$text = "[quote]" . trim($text) . "[/quote]" . $br;
 										} else {
 											$text = "<blockquote>" . htmlspecialchars(trim($text)) . "</blockquote><br />";
-												parse url: Characters like < and > has to be escaped when showing the parsed output.

											
										
										
											2014-02-16 16:35:01 +00:00
+											$title = htmlspecialchars($title);
 										}
-												allow setting all the bookmarked URL fields via script/API

											
										
										
											2011-09-20 05:21:55 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+										$result = sprintf($template, $url, ($title) ? $title : $url, $text) . $str_tags;
-												allow setting all the bookmarked URL fields via script/API

											
										
										
											2011-09-20 05:21:55 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+										logger("parse_url (unparsed): returns: " . $result);
-												allow setting all the bookmarked URL fields via script/API

											
										
										
											2011-09-20 05:21:55 +00:00
 										echo $result;
 										killme();
 									}
-												parse_url: add some docu
											
										
										
											2016-11-25 15:59:31 +00:00
+									// Fetch the information directly from the webpage
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									$siteinfo = ParseUrl::getSiteinfo($url);
-												allow setting all the bookmarked URL fields via script/API

											
										
										
											2011-09-20 05:21:55 +00:00
-												Use the "attachment" element instead of a mixture of several elements

											
										
										
											2016-04-17 18:43:41 +00:00
+									unset($siteinfo["keywords"]);
-												Merge remote-tracking branch 'upstream/master'

Conflicts:
	mod/crepair.php

											
										
										
											2014-01-05 15:22:42 +00:00
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									// Format it as BBCode attachment
-												Bugfix: ' in titles were a problem.

											
										
										
											2016-04-18 18:57:01 +00:00
+									$info = add_page_info_data($siteinfo);
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									if (!$textmode) {
-												Bugfix: ' in titles were a problem.

											
										
										
											2016-04-18 18:57:01 +00:00
+										// Replace ' with ’ - not perfect - but the richtext editor has problems otherwise
 										$info = str_replace(array("&#039;"), array("&#8217;"), $info);
-												parse_url: recognize image/video/audio files + move functions into own class
											
										
										
											2016-11-24 00:11:22 +00:00
+									}
-												Bugfix: ' in titles were a problem.

											
										
										
											2016-04-18 18:57:01 +00:00
 									echo $info;
-												ajaxify the wall poster

											
										
										
											2010-07-23 05:41:45 +00:00
+									killme();
-												Add oembed user option for use oembed instead of standard yotube embedding.
Remove global option and parse_url use of oembed.

											
										
										
											2011-05-23 08:37:09 +00:00
+								}