Massive reworking of the export connectors using the plaintext function, improving of the tumblr export, option to prevent too short wordpress postings

This commit is contained in:
Michael Vogel 2014-05-29 12:42:25 +02:00
parent 8ce597cb4d
commit 021d6265cd
11 changed files with 191 additions and 880 deletions

View file

@ -168,10 +168,8 @@ function statusnet_settings_post ($a,$post) {
del_pconfig(local_user(), 'statusnet', 'oauthtoken');
del_pconfig(local_user(), 'statusnet', 'oauthsecret');
del_pconfig(local_user(), 'statusnet', 'baseapi');
del_pconfig(local_user(), 'statusnet', 'post_taglinks');
del_pconfig(local_user(), 'statusnet', 'lastid');
del_pconfig(local_user(), 'statusnet', 'mirror_posts');
del_pconfig(local_user(), 'statusnet', 'intelligent_shortening');
del_pconfig(local_user(), 'statusnet', 'import');
del_pconfig(local_user(), 'statusnet', 'create_user');
del_pconfig(local_user(), 'statusnet', 'own_id');
@ -249,9 +247,7 @@ function statusnet_settings_post ($a,$post) {
// to post a dent for every new __public__ posting to the wall
set_pconfig(local_user(),'statusnet','post',intval($_POST['statusnet-enable']));
set_pconfig(local_user(),'statusnet','post_by_default',intval($_POST['statusnet-default']));
set_pconfig(local_user(),'statusnet','post_taglinks',intval($_POST['statusnet-sendtaglinks']));
set_pconfig(local_user(), 'statusnet', 'mirror_posts', intval($_POST['statusnet-mirror']));
set_pconfig(local_user(), 'statusnet', 'intelligent_shortening', intval($_POST['statusnet-shortening']));
set_pconfig(local_user(), 'statusnet', 'import', intval($_POST['statusnet-import']));
set_pconfig(local_user(), 'statusnet', 'create_user', intval($_POST['statusnet-create_user']));
info( t('StatusNet settings updated.') . EOL);
@ -268,20 +264,16 @@ function statusnet_settings(&$a,&$s) {
* 3) Checkbox for "Send public notices (respect size limitation)
*/
$api = get_pconfig(local_user(), 'statusnet', 'baseapi');
$ckey = get_pconfig(local_user(), 'statusnet', 'consumerkey' );
$csecret = get_pconfig(local_user(), 'statusnet', 'consumersecret' );
$otoken = get_pconfig(local_user(), 'statusnet', 'oauthtoken' );
$osecret = get_pconfig(local_user(), 'statusnet', 'oauthsecret' );
$ckey = get_pconfig(local_user(), 'statusnet', 'consumerkey');
$csecret = get_pconfig(local_user(), 'statusnet', 'consumersecret');
$otoken = get_pconfig(local_user(), 'statusnet', 'oauthtoken');
$osecret = get_pconfig(local_user(), 'statusnet', 'oauthsecret');
$enabled = get_pconfig(local_user(), 'statusnet', 'post');
$checked = (($enabled) ? ' checked="checked" ' : '');
$defenabled = get_pconfig(local_user(),'statusnet','post_by_default');
$defchecked = (($defenabled) ? ' checked="checked" ' : '');
$linksenabled = get_pconfig(local_user(),'statusnet','post_taglinks');
$linkschecked = (($linksenabled) ? ' checked="checked" ' : '');
$mirrorenabled = get_pconfig(local_user(),'statusnet','mirror_posts');
$mirrorchecked = (($mirrorenabled) ? ' checked="checked" ' : '');
$shorteningenabled = get_pconfig(local_user(),'statusnet','intelligent_shortening');
$shorteningchecked = (($shorteningenabled) ? ' checked="checked" ' : '');
$importenabled = get_pconfig(local_user(),'statusnet','import');
$importchecked = (($importenabled) ? ' checked="checked" ' : '');
$create_userenabled = get_pconfig(local_user(),'statusnet','create_user');
@ -289,9 +281,6 @@ function statusnet_settings(&$a,&$s) {
$css = (($enabled) ? '' : '-disabled');
$globalshortening = get_config('statusnet','intelligent_shortening');
$s .= '<span id="settings_statusnet_inflated" class="settings-block fakelink" style="display: block;" onclick="openClose(\'settings_statusnet_expanded\'); openClose(\'settings_statusnet_inflated\');">';
$s .= '<img class="connector'.$css.'" src="images/gnusocial.png" /><h3 class="connector">'. t('StatusNet Import/Export/Mirror').'</h3>';
$s .= '</span>';
@ -394,16 +383,6 @@ function statusnet_settings(&$a,&$s) {
$s .= '<label id="statusnet-mirror-label" for="statusnet-mirror">'.t('Mirror all posts from statusnet that are no replies or repeated messages').'</label>';
$s .= '<input id="statusnet-mirror" type="checkbox" name="statusnet-mirror" value="1" '. $mirrorchecked . '/>';
$s .= '<div class="clear"></div>';
if (!$globalshortening) {
$s .= '<label id="statusnet-shortening-label" for="statusnet-shortening">'.t('Shortening method that optimizes the post').'</label>';
$s .= '<input id="statusnet-shortening" type="checkbox" name="statusnet-shortening" value="1" '. $shorteningchecked . '/>';
$s .= '<div class="clear"></div>';
$s .= '<label id="statusnet-sendtaglinks-label" for="statusnet-sendtaglinks">'.t('Send linked #-tags and @-names to StatusNet').'</label>';
$s .= '<input id="statusnet-sendtaglinks" type="checkbox" name="statusnet-sendtaglinks" value="1" '. $linkschecked . '/>';
$s .= '<div class="clear"></div>';
}
$s .= '</div>';
$s .= '<label id="statusnet-import-label" for="statusnet-import">'.t('Import the remote timeline').'</label>';
@ -447,192 +426,6 @@ function statusnet_post_local(&$a,&$b) {
}
}
if (! function_exists( 'short_link' )) {
function short_link($url) {
require_once('library/slinky.php');
$slinky = new Slinky( $url );
$yourls_url = get_config('yourls','url1');
if ($yourls_url) {
$yourls_username = get_config('yourls','username1');
$yourls_password = get_config('yourls', 'password1');
$yourls_ssl = get_config('yourls', 'ssl1');
$yourls = new Slinky_YourLS();
$yourls->set( 'username', $yourls_username );
$yourls->set( 'password', $yourls_password );
$yourls->set( 'ssl', $yourls_ssl );
$yourls->set( 'yourls-url', $yourls_url );
$slinky->set_cascade( array( $yourls, new Slinky_UR1ca(), new Slinky_Trim(), new Slinky_IsGd(), new Slinky_TinyURL() ) );
}
else {
// setup a cascade of shortening services
// try to get a short link from these services
// in the order ur1.ca, trim, id.gd, tinyurl
$slinky->set_cascade( array( new Slinky_UR1ca(), new Slinky_Trim(), new Slinky_IsGd(), new Slinky_TinyURL() ) );
}
return $slinky->short();
} };
function statusnet_shortenmsg($b, $max_char) {
require_once("include/api.php");
require_once("include/bbcode.php");
require_once("include/html2plain.php");
$b['body'] = bb_CleanPictureLinks($b['body']);
// Looking for the first image
$cleaned_body = api_clean_plain_items($b['body']);
$image = '';
if(preg_match("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/is",$cleaned_body,$matches))
$image = $matches[3];
if ($image == '')
if(preg_match("/\[img\](.*?)\[\/img\]/is",$cleaned_body,$matches))
$image = $matches[1];
$multipleimages = (strpos($cleaned_body, "[img") != strrpos($cleaned_body, "[img"));
// When saved into the database the content is sent through htmlspecialchars
// That means that we have to decode all image-urls
$image = htmlspecialchars_decode($image);
$body = $b["body"];
if ($b["title"] != "")
$body = $b["title"]."\n\n".$body;
if (strpos($body, "[bookmark") !== false) {
// splitting the text in two parts:
// before and after the bookmark
$pos = strpos($body, "[bookmark");
$body1 = substr($body, 0, $pos);
$body2 = substr($body, $pos);
// Removing all quotes after the bookmark
// they are mostly only the content after the bookmark.
$body2 = preg_replace("/\[quote\=([^\]]*)\](.*?)\[\/quote\]/ism",'',$body2);
$body2 = preg_replace("/\[quote\](.*?)\[\/quote\]/ism",'',$body2);
$body = $body1.$body2;
}
// Add some newlines so that the message could be cut better
$body = str_replace(array("[quote", "[bookmark", "[/bookmark]", "[/quote]"),
array("\n[quote", "\n[bookmark", "[/bookmark]\n", "[/quote]\n"), $body);
// remove the recycle signs and the names since they aren't helpful on statusnet
// recycle 1
$recycle = html_entity_decode("&#x2672; ", ENT_QUOTES, 'UTF-8');
$body = preg_replace( '/'.$recycle.'\[url\=(\w+.*?)\](\w+.*?)\[\/url\]/i', "\n", $body);
// recycle 2 (Test)
$recycle = html_entity_decode("&#x25CC; ", ENT_QUOTES, 'UTF-8');
$body = preg_replace( '/'.$recycle.'\[url\=(\w+.*?)\](\w+.*?)\[\/url\]/i', "\n", $body);
// remove the share element
//$body = preg_replace("/\[share(.*?)\](.*?)\[\/share\]/ism","\n\n$2\n\n",$body);
// At first convert the text to html
$html = bbcode(api_clean_plain_items($body), false, false, 2);
// Then convert it to plain text
//$msg = trim($b['title']." \n\n".html2plain($html, 0, true));
$msg = trim(html2plain($html, 0, true));
$msg = html_entity_decode($msg,ENT_QUOTES,'UTF-8');
// Removing multiple newlines
while (strpos($msg, "\n\n\n") !== false)
$msg = str_replace("\n\n\n", "\n\n", $msg);
// Removing multiple spaces
while (strpos($msg, " ") !== false)
$msg = str_replace(" ", " ", $msg);
$origmsg = $msg;
// Removing URLs
$msg = preg_replace('/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/i', "", $msg);
$msg = trim($msg);
$link = '';
// look for bookmark-bbcode and handle it with priority
if(preg_match("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/is",$b['body'],$matches))
$link = $matches[1];
$multiplelinks = (strpos($b['body'], "[bookmark") != strrpos($b['body'], "[bookmark"));
// If there is no bookmark element then take the first link
if ($link == '') {
$links = collecturls($html);
if (sizeof($links) > 0) {
reset($links);
$link = current($links);
}
$multiplelinks = (sizeof($links) > 1);
}
$msglink = "";
if ($multiplelinks)
$msglink = $b["plink"];
else if ($link != "")
$msglink = $link;
else if ($multipleimages)
$msglink = $b["plink"];
else if ($image != "")
$msglink = $image;
if (($msglink == "") and strlen($msg) > $max_char)
$msglink = $b["plink"];
// If the message is short enough then don't modify it. (if the link exists in the original message)
if ((strlen(trim($origmsg)) <= $max_char) AND (($msglink == "") OR strpos($origmsg, $msglink)))
return(array("msg"=>trim($origmsg), "image"=>""));
// If the message is short enough and contains a picture then post the picture as well
if ((strlen(trim($origmsg)) <= ($max_char - 20)) AND strpos($origmsg, $msglink))
return(array("msg"=>trim($origmsg), "image"=>$image));
// If the message is short enough and the link exists in the original message don't modify it as well
if ((strlen(trim($origmsg)) <= $max_char) AND strpos($origmsg, $msglink))
return(array("msg"=>trim($origmsg), "image"=>""));
// Preserve the unshortened link
$orig_link = $msglink;
if (strlen($msglink) > 20)
$msglink = short_link($msglink);
if (strlen(trim($msg." ".$msglink)) > $max_char) {
$msg = substr($msg, 0, $max_char - (strlen($msglink)));
$lastchar = substr($msg, -1);
$msg = substr($msg, 0, -1);
$pos = strrpos($msg, "\n");
if ($pos > 0)
$msg = substr($msg, 0, $pos);
else if ($lastchar != "\n")
$msg = substr($msg, 0, -3)."...";
}
//$msg = str_replace("\n", " ", $msg);
// Removing multiple spaces - again
while (strpos($msg, " ") !== false)
$msg = str_replace(" ", " ", $msg);
//return(array("msg"=>trim($msg."\n".$msglink), "image"=>$image));
// Looking if the link points to an image
$img_str = fetch_url($orig_link);
$tempfile = tempnam(get_config("system","temppath"), "cache");
file_put_contents($tempfile, $img_str);
$mime = image_type_to_mime_type(exif_imagetype($tempfile));
unlink($tempfile);
if (($image == $orig_link) OR (substr($mime, 0, 6) == "image/"))
return(array("msg"=>trim($msg), "image"=>$orig_link));
else if (($image != $orig_link) AND ($image != "") AND (strlen($msg." ".$msglink) <= ($max_char - 20)))
return(array("msg"=>trim($msg." ".$msglink)."\n", "image"=>$image));
else
return(array("msg"=>trim($msg." ".$msglink), "image"=>""));
}
function statusnet_action($a, $uid, $pid, $action) {
$api = get_pconfig($uid, 'statusnet', 'baseapi');
$ckey = get_pconfig($uid, 'statusnet', 'consumerkey');
@ -740,11 +533,6 @@ function statusnet_post_hook(&$a,&$b) {
$csecret = get_pconfig($b['uid'], 'statusnet', 'consumersecret');
$otoken = get_pconfig($b['uid'], 'statusnet', 'oauthtoken');
$osecret = get_pconfig($b['uid'], 'statusnet', 'oauthsecret');
$intelligent_shortening = get_pconfig($b['uid'], 'statusnet', 'intelligent_shortening');
// Global setting overrides this
if (get_config('statusnet','intelligent_shortening'))
$intelligent_shortening = get_config('statusnet','intelligent_shortening');
if($ckey && $csecret && $otoken && $osecret) {
@ -755,119 +543,34 @@ function statusnet_post_hook(&$a,&$b) {
require_once('include/bbcode.php');
$dent = new StatusNetOAuth($api,$ckey,$csecret,$otoken,$osecret);
$max_char = $dent->get_maxlength(); // max. length for a dent
// we will only work with up to two times the length of the dent
// we can later send to StatusNet. This way we can "gain" some
// information during shortening of potential links but do not
// shorten all the links in a 200000 character long essay.
$tempfile = "";
$intelligent_shortening = get_config('statusnet','intelligent_shortening');
if (!$intelligent_shortening) {
if (! $b['title']=='') {
$tmp = $b['title'].": \n".$b['body'];
// $tmp = substr($tmp, 0, 4*$max_char);
} else {
$tmp = $b['body']; // substr($b['body'], 0, 3*$max_char);
}
// if [url=bla][img]blub.png[/img][/url] get blub.png
$tmp = preg_replace( '/\[url\=(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)\]\[img\](\\w+.*?)\\[\\/img\]\\[\\/url\]/i', '$2', $tmp);
// preserve links to images, videos and audios
$tmp = preg_replace( '/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/ism', '$3', $tmp);
$tmp = preg_replace( '/\[\\/?img(\\s+.*?\]|\])/i', '', $tmp);
$tmp = preg_replace( '/\[\\/?video(\\s+.*?\]|\])/i', '', $tmp);
$tmp = preg_replace( '/\[\\/?youtube(\\s+.*?\]|\])/i', '', $tmp);
$tmp = preg_replace( '/\[\\/?vimeo(\\s+.*?\]|\])/i', '', $tmp);
$tmp = preg_replace( '/\[\\/?audio(\\s+.*?\]|\])/i', '', $tmp);
$linksenabled = get_pconfig($b['uid'],'statusnet','post_taglinks');
// if a #tag is linked, don't send the [url] over to SN
// that is, don't send if the option is not set in the
// connector settings
if ($linksenabled=='0') {
// #-tags
$tmp = preg_replace( '/#\[url\=(\w+.*?)\](\w+.*?)\[\/url\]/i', '#$2', $tmp);
// @-mentions
$tmp = preg_replace( '/@\[url\=(\w+.*?)\](\w+.*?)\[\/url\]/i', '@$2', $tmp);
// recycle 1
$recycle = html_entity_decode("&#x2672; ", ENT_QUOTES, 'UTF-8');
$tmp = preg_replace( '/'.$recycle.'\[url\=(\w+.*?)\](\w+.*?)\[\/url\]/i', $recycle.'$2', $tmp);
// recycle 2 (test)
$recycle = html_entity_decode("&#x25CC; ", ENT_QUOTES, 'UTF-8');
$tmp = preg_replace( '/'.$recycle.'\[url\=(\w+.*?)\](\w+.*?)\[\/url\]/i', $recycle.'$2', $tmp);
}
// preserve links to webpages
$tmp = preg_replace( '/\[url\=(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)\](\w+.*?)\[\/url\]/i', '$2 $1', $tmp);
$tmp = preg_replace( '/\[bookmark\=(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)\](\w+.*?)\[\/bookmark\]/i', '$2 $1', $tmp);
// find all http or https links in the body of the entry and
// apply the shortener if the link is longer then 20 characters
if (( strlen($tmp)>$max_char ) && ( $max_char > 0 )) {
preg_match_all ( '/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/i', $tmp, $allurls );
foreach ($allurls as $url) {
foreach ($url as $u) {
if (strlen($u)>20) {
$sl = short_link($u);
$tmp = str_replace( $u, $sl, $tmp );
}
}
}
}
// ok, all the links we want to send out are save, now strip
// away the remaining bbcode
//$msg = strip_tags(bbcode($tmp, false, false));
$msg = bbcode($tmp, false, false, true);
$msg = str_replace(array('<br>','<br />'),"\n",$msg);
$msg = strip_tags($msg);
require_once("include/plaintext.php");
require_once("include/network.php");
$msgarr = plaintext($a, $b, $max_char, true);
$msg = $msgarr["text"];
// quotes not working - let's try this
$msg = html_entity_decode($msg);
if (($msg == "") AND isset($msgarr["title"]))
$msg = shortenmsg($msgarr["title"], $max_char - 50);
if (( strlen($msg) > $max_char) && $max_char > 0) {
$shortlink = short_link( $b['plink'] );
// the new message will be shortened such that "... $shortlink"
// will fit into the character limit
$msg = nl2br(substr($msg, 0, $max_char-strlen($shortlink)-4));
$msg = str_replace(array('<br>','<br />'),' ',$msg);
$e = explode(' ', $msg);
// remove the last word from the cut down message to
// avoid sending cut words to the MicroBlog
array_pop($e);
$msg = implode(' ', $e);
$msg .= '... ' . $shortlink;
}
$image = "";
$msg = trim($msg);
$postdata = array('status' => $msg);
} else {
/* $msgarr = statusnet_shortenmsg($b, $max_char);
$msg = $msgarr["msg"];
if (isset($msgarr["url"])) {
if ((strlen($msgarr["url"]) > 20) AND
((strlen($msg." \n".$msgarr["url"]) > $max_char)))
$msg .= " \n".short_link($msgarr["url"]);
else
$msg .= " \n".$msgarr["url"];
} elseif (isset($msgarr["image"]))
$image = $msgarr["image"];
*/
require_once("include/plaintext.php");
require_once("include/network.php");
$msgarr = plaintext($a, $b, $max_char, true);
$msg = $msgarr["text"];
if (($msg == "") AND isset($msgarr["title"]))
$msg = shortenmsg($msgarr["title"], $max_char - 50);
$image = "";
if (isset($msgarr["url"])) {
if ((strlen($msgarr["url"]) > 20) AND
((strlen($msg." ".$msgarr["url"]) > $max_char)))
$msg .= " ".short_link($msgarr["url"]);
else
$msg .= " ".$msgarr["url"];
} elseif (isset($msgarr["image"]))
$image = $msgarr["image"];
if ($image != "") {
$img_str = fetch_url($image);
$tempfile = tempnam(get_config("system","temppath"), "cache");
file_put_contents($tempfile, $img_str);
$postdata = array("status" => $msg, "media[]" => $tempfile);
} else
$postdata = array("status"=>$msg);
}
if ($image != "") {
$img_str = fetch_url($image);
$tempfile = tempnam(get_config("system","temppath"), "cache");
file_put_contents($tempfile, $img_str);
$postdata = array("status" => $msg, "media[]" => $tempfile);
} else
$postdata = array("status"=>$msg);
// and now dent it :-)
if(strlen($msg)) {
@ -1712,6 +1415,7 @@ function statusnet_convertmsg($a, $body, $no_tags = false) {
$footer = "";
$footerurl = "";
$footerlink = "";
$type = "";
if ($links) {
@ -1720,14 +1424,18 @@ function statusnet_convertmsg($a, $body, $no_tags = false) {
$expanded_url = original_url($match[1]);
$oembed_data = oembed_fetch_url($expanded_url);
$oembed_data = oembed_fetch_url($expanded_url, true);
print_r($oembed_data);
if ($type == "")
$type = $oembed_data->type;
if ($oembed_data->type == "video") {
//$body = str_replace($search, "[video]".$expanded_url."[/video]", $body);
$type = $oembed_data->type;
$footerurl = $expanded_url;
$footerlink = "[url=".$expanded_url."]".$expanded_url."[/url]";
if ($oembed_data->type == "video")
$body = str_replace($search, "[video]".$expanded_url."[/video]", $body);
elseif (($oembed_data->type == "photo") AND isset($oembed_data->url) AND !$dontincludemedia)
$body = str_replace($search, $footerlink, $body);
} elseif (($oembed_data->type == "photo") AND isset($oembed_data->url) AND !$dontincludemedia)
$body = str_replace($search, "[url=".$expanded_url."][img]".$oembed_data->url."[/img][/url]", $body);
elseif ($oembed_data->type != "link")
$body = str_replace($search, "[url=".$expanded_url."]".$expanded_url."[/url]", $body);
@ -1747,7 +1455,7 @@ function statusnet_convertmsg($a, $body, $no_tags = false) {
$footerurl = $expanded_url;
$footerlink = "[url=".$expanded_url."]".$expanded_url."[/url]";
$body = str_replace($search, $expanded_url, $body);
$body = str_replace($search, $footerlink, $body);
}
}
}