friendica/src/Protocol/Feed.php

518 lines
17 KiB
PHP
Raw Normal View History

<?php
2017-12-13 08:03:42 +01:00
/**
* @file src/Protocol/Feed.php
* @brief Imports RSS/RDF/Atom feeds
*
*/
namespace Friendica\Protocol;
use DOMDocument;
use DOMXPath;
use Friendica\Content\Text\HTML;
2018-10-29 22:20:46 +01:00
use Friendica\Core\Logger;
use Friendica\Core\Protocol;
use Friendica\Core\System;
use Friendica\Database\DBA;
use Friendica\Model\Item;
2019-10-25 00:34:46 +02:00
use Friendica\Protocol\ActivityNamespace;
use Friendica\Util\ParseUrl;
use Friendica\Util\Network;
2018-07-08 15:39:48 +02:00
use Friendica\Util\XML;
2018-03-08 20:58:35 +01:00
2016-02-16 08:06:55 +01:00
/**
2017-12-13 08:03:42 +01:00
* @brief This class contain functions to import feeds
2016-02-16 08:06:55 +01:00
*
*/
2017-12-13 08:03:42 +01:00
class Feed {
/**
* @brief Read a RSS/RDF/Atom feed and create an item entry for it
*
2019-01-06 22:06:53 +01:00
* @param string $xml The feed data
* @param array $importer The user record of the importer
* @param array $contact The contact record of the feed
* @param string $hub Unused dummy value for compatibility reasons
* @param bool $simulate If enabled, no data is imported
2017-12-13 08:03:42 +01:00
*
* @return array In simulation mode it returns the header and the first item
2019-01-06 22:06:53 +01:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
2017-12-13 08:03:42 +01:00
*/
public static function import($xml, $importer, &$contact, &$hub, $simulate = false) {
2017-12-13 08:03:42 +01:00
$a = \get_app();
2017-12-13 08:03:42 +01:00
if (!$simulate) {
Logger::log("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], Logger::DEBUG);
2017-12-13 08:03:42 +01:00
} else {
Logger::log("Test Atom/RSS feed", Logger::DEBUG);
2017-12-13 08:03:42 +01:00
}
if (empty($xml)) {
Logger::log('XML is empty.', Logger::DEBUG);
2017-12-13 08:03:42 +01:00
return;
}
2017-12-13 08:03:42 +01:00
if (!empty($contact['poll'])) {
$basepath = $contact['poll'];
} elseif (!empty($contact['url'])) {
$basepath = $contact['url'];
} else {
$basepath = '';
}
2017-12-13 08:03:42 +01:00
$doc = new DOMDocument();
@$doc->loadXML(trim($xml));
$xpath = new DOMXPath($doc);
2019-10-25 00:32:35 +02:00
$xpath->registerNamespace('atom', ActivityNamespace::ATOM1);
2017-12-13 08:03:42 +01:00
$xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
$xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
$xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
$xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
$xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
2019-10-25 00:32:35 +02:00
$xpath->registerNamespace('poco', ActivityNamespace::POCO);
$author = [];
$entries = null;
2017-12-13 08:03:42 +01:00
// Is it RDF?
if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
$author["author-link"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
$author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
if (empty($author["author-name"])) {
$author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
2017-03-15 07:00:22 +01:00
}
2017-12-13 08:03:42 +01:00
$entries = $xpath->query('/rdf:RDF/rss:item');
2017-03-15 07:00:22 +01:00
}
2017-12-13 08:03:42 +01:00
// Is it Atom?
if ($xpath->query('/atom:feed')->length > 0) {
$alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']");
2017-12-13 08:03:42 +01:00
if (is_object($alternate)) {
foreach ($alternate AS $attribute) {
if ($attribute->name == "href") {
$author["author-link"] = $attribute->textContent;
2017-03-15 07:00:22 +01:00
}
}
}
if (empty($author["author-link"])) {
$self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']");
2017-12-13 08:03:42 +01:00
if (is_object($self)) {
foreach ($self AS $attribute) {
if ($attribute->name == "href") {
$author["author-link"] = $attribute->textContent;
2017-12-13 08:03:42 +01:00
}
}
}
}
if (empty($author["author-link"])) {
$author["author-link"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()');
2017-12-13 08:03:42 +01:00
}
$author["author-avatar"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()');
$author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()');
if (empty($author["author-name"])) {
$author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
2017-03-15 07:00:22 +01:00
}
if (empty($author["author-name"])) {
$author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
2017-03-15 07:00:22 +01:00
}
$value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
2017-03-15 07:00:22 +01:00
if ($value != "") {
2017-12-13 08:03:42 +01:00
$author["author-name"] = $value;
2017-03-15 07:00:22 +01:00
}
2017-12-13 08:03:42 +01:00
if ($simulate) {
$author["author-id"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:id/text()');
// See https://tools.ietf.org/html/rfc4287#section-3.2.2
$value = XML::getFirstNodeValue($xpath, 'atom:author/atom:uri/text()');
if ($value != "") {
$author["author-link"] = $value;
}
2017-12-13 08:03:42 +01:00
$value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()');
2017-12-13 08:03:42 +01:00
if ($value != "") {
$author["author-nick"] = $value;
}
$value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()');
2017-12-13 08:03:42 +01:00
if ($value != "") {
$author["author-location"] = $value;
}
$value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()');
2017-12-13 08:03:42 +01:00
if ($value != "") {
$author["author-about"] = $value;
}
$avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']");
2017-12-13 08:03:42 +01:00
if (is_object($avatar)) {
foreach ($avatar AS $attribute) {
if ($attribute->name == "href") {
$author["author-avatar"] = $attribute->textContent;
2017-12-13 08:03:42 +01:00
}
2017-08-21 22:21:04 +02:00
}
}
}
$author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()');
$author["app"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()');
2017-12-13 08:03:42 +01:00
$entries = $xpath->query('/atom:feed/atom:entry');
}
2017-12-13 08:03:42 +01:00
// Is it RSS?
if ($xpath->query('/rss/channel')->length > 0) {
$author["author-link"] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
$author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
$author["author-avatar"] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
if (empty($author["author-name"])) {
$author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
2017-12-13 08:03:42 +01:00
}
if (empty($author["author-name"])) {
$author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
2017-12-13 08:03:42 +01:00
}
$author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
$author["app"] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
2017-12-13 08:03:42 +01:00
$entries = $xpath->query('/rss/channel/item');
}
if (!$simulate) {
$author["author-link"] = $contact["url"];
if (empty($author["author-name"])) {
2017-12-13 08:03:42 +01:00
$author["author-name"] = $contact["name"];
}
$author["author-avatar"] = $contact["thumb"];
2017-12-13 08:03:42 +01:00
$author["owner-link"] = $contact["url"];
$author["owner-name"] = $contact["name"];
$author["owner-avatar"] = $contact["thumb"];
2017-03-15 07:00:22 +01:00
}
$header = [];
2017-12-13 08:03:42 +01:00
$header["uid"] = $importer["uid"];
$header["network"] = Protocol::FEED;
2017-12-13 08:03:42 +01:00
$header["wall"] = 0;
$header["origin"] = 0;
$header["gravity"] = GRAVITY_PARENT;
$header["private"] = 2;
$header["verb"] = Activity::POST;
2019-10-25 00:10:20 +02:00
$header["object-type"] = Activity\ObjectType::NOTE;
2017-12-13 08:03:42 +01:00
$header["contact-id"] = $contact["id"];
2015-10-03 13:58:10 +02:00
2017-12-13 08:03:42 +01:00
if (!is_object($entries)) {
Logger::log("There are no entries in this feed.", Logger::DEBUG);
2017-12-13 08:03:42 +01:00
return;
}
$items = [];
2018-03-10 18:40:21 +01:00
// Importing older entries first
for($i = $entries->length - 1; $i >= 0;--$i) {
$entry = $entries->item($i);
2017-12-13 08:03:42 +01:00
$item = array_merge($header, $author);
$alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry);
2017-12-13 08:03:42 +01:00
if (!is_object($alternate)) {
$alternate = XML::getFirstAttributes($xpath, "atom:link", $entry);
2017-12-13 08:03:42 +01:00
}
if (is_object($alternate)) {
foreach ($alternate AS $attribute) {
if ($attribute->name == "href") {
$item["plink"] = $attribute->textContent;
2017-12-13 08:03:42 +01:00
}
2017-03-15 07:00:22 +01:00
}
}
if (empty($item["plink"])) {
$item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
2017-12-13 08:03:42 +01:00
}
if (empty($item["plink"])) {
$item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
2017-12-13 08:03:42 +01:00
}
$item["uri"] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry);
if (empty($item["uri"])) {
$item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
2017-12-13 08:03:42 +01:00
}
if (empty($item["uri"])) {
2017-12-13 08:03:42 +01:00
$item["uri"] = $item["plink"];
}
2017-12-13 08:03:42 +01:00
$orig_plink = $item["plink"];
$item["plink"] = Network::finalUrl($item["plink"]);
2017-12-13 08:03:42 +01:00
$item["parent-uri"] = $item["uri"];
2017-12-13 08:03:42 +01:00
if (!$simulate) {
$condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
$importer["uid"], $item["uri"], Protocol::FEED, Protocol::DFRN];
$previous = Item::selectFirst(['id'], $condition);
2018-07-21 14:46:04 +02:00
if (DBA::isResult($previous)) {
Logger::log("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], Logger::DEBUG);
2017-12-13 08:03:42 +01:00
continue;
}
}
$item["title"] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
if (empty($item["title"])) {
$item["title"] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
2017-12-13 08:03:42 +01:00
}
if (empty($item["title"])) {
$item["title"] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
2017-12-13 08:03:42 +01:00
}
$item["title"] = html_entity_decode($item["title"], ENT_QUOTES, 'UTF-8');
$published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry);
if (empty($published)) {
$published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
2017-12-13 08:03:42 +01:00
}
if (empty($published)) {
$published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
2017-12-13 08:03:42 +01:00
}
$updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry);
if (empty($updated) && !empty($published)) {
2017-12-13 08:03:42 +01:00
$updated = $published;
}
if (empty($published) && !empty($updated)) {
$published = $updated;
}
2017-12-13 08:03:42 +01:00
if ($published != "") {
$item["created"] = $published;
}
if ($updated != "") {
$item["edited"] = $updated;
}
$creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
if (empty($creator)) {
$creator = XML::getFirstNodeValue($xpath, 'atom:author/atom:name/text()', $entry);
2017-12-13 08:03:42 +01:00
}
if (empty($creator)) {
$creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
2017-12-13 08:03:42 +01:00
}
if ($creator != "") {
$item["author-name"] = $creator;
}
$creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
2017-12-13 08:03:42 +01:00
if ($creator != "") {
$item["author-name"] = $creator;
}
2017-12-13 08:03:42 +01:00
/// @TODO ?
// <category>Ausland</category>
// <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
$attachments = [];
2017-12-13 08:03:42 +01:00
2018-03-10 18:40:21 +01:00
$enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
2017-12-13 08:03:42 +01:00
foreach ($enclosures AS $enclosure) {
$href = "";
$length = "";
$type = "";
foreach ($enclosure->attributes AS $attribute) {
if (in_array($attribute->name, ["url", "href"])) {
$href = $attribute->textContent;
} elseif ($attribute->name == "length") {
$length = $attribute->textContent;
} elseif ($attribute->name == "type") {
$type = $attribute->textContent;
2017-12-13 08:03:42 +01:00
}
}
if (!empty($item["attach"])) {
2017-12-13 08:03:42 +01:00
$item["attach"] .= ',';
} else {
$item["attach"] = '';
2017-03-15 07:00:22 +01:00
}
$attachments[] = ["link" => $href, "type" => $type, "length" => $length];
2017-12-13 08:03:42 +01:00
$item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
}
2017-12-13 08:03:42 +01:00
$tags = '';
$categories = $xpath->query("category", $entry);
foreach ($categories AS $category) {
$hashtag = $category->nodeValue;
if ($tags != '') {
$tags .= ', ';
}
$taglink = "#[url=" . System::baseUrl() . "/search?tag=" . $hashtag . "]" . $hashtag . "[/url]";
2017-12-13 08:03:42 +01:00
$tags .= $taglink;
}
$body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry));
2017-10-17 11:10:19 +02:00
if (empty($body)) {
$body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
2017-12-13 08:03:42 +01:00
}
$summary = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry));
if (empty($summary)) {
$summary = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
2017-12-13 08:03:42 +01:00
}
if (empty($body)) {
$body = $summary;
$summary = '';
}
if ($body == $summary) {
$summary = '';
2017-12-13 08:03:42 +01:00
}
2017-10-17 11:10:19 +02:00
2017-12-13 08:03:42 +01:00
// remove the content of the title if it is identically to the body
// This helps with auto generated titles e.g. from tumblr
if (self::titleIsBody($item["title"], $body)) {
2017-12-13 08:03:42 +01:00
$item["title"] = "";
}
2018-03-08 20:58:35 +01:00
$item["body"] = HTML::toBBCode($body, $basepath);
2017-10-17 11:10:19 +02:00
2017-12-13 08:03:42 +01:00
if (($item["body"] == '') && ($item["title"] != '')) {
$item["body"] = $item["title"];
$item["title"] = '';
}
2017-10-17 11:10:19 +02:00
$preview = '';
2017-12-13 08:03:42 +01:00
if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
// Handle enclosures and treat them as preview picture
foreach ($attachments AS $attachment) {
if ($attachment["type"] == "image/jpeg") {
$preview = $attachment["link"];
}
2017-03-15 07:00:22 +01:00
}
2017-12-13 08:03:42 +01:00
// Remove a possible link to the item itself
$item["body"] = str_replace($item["plink"], '', $item["body"]);
$item["body"] = trim(preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]));
2017-10-17 13:39:09 +02:00
2017-12-13 08:03:42 +01:00
// Replace the content when the title is longer than the body
$replace = (strlen($item["title"]) > strlen($item["body"]));
2017-10-17 11:58:29 +02:00
2017-12-13 08:03:42 +01:00
// Replace it, when there is an image in the body
if (strstr($item["body"], '[/img]')) {
$replace = true;
}
2017-10-17 11:58:29 +02:00
2017-12-13 08:03:42 +01:00
// Replace it, when there is a link in the body
if (strstr($item["body"], '[/url]')) {
$replace = true;
}
2017-10-17 11:58:29 +02:00
2017-12-13 08:03:42 +01:00
if ($replace) {
$item["body"] = trim($item["title"]);
2017-12-13 08:03:42 +01:00
}
$data = ParseUrl::getSiteinfoCached($item['plink'], true);
if (!empty($data['text']) && !empty($data['title']) && (mb_strlen($item['body']) < mb_strlen($data['text']))) {
// When the fetched page info text is longer than the body, we do try to enhance the body
2019-11-18 19:09:21 +01:00
if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
// The body is not part of the fetched page info title or page info text. So we add the text to the body
$item['body'] .= "\n\n" . $data['text'];
} else {
// Else we replace the body with the page info text
$item['body'] = $data['text'];
}
}
2017-12-13 08:03:42 +01:00
// We always strip the title since it will be added in the page information
$item["title"] = "";
$item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
$item["tag"] = add_page_keywords($item["plink"], $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
2019-10-25 00:10:20 +02:00
$item["object-type"] = Activity\ObjectType::BOOKMARK;
2017-12-13 08:03:42 +01:00
unset($item["attach"]);
} else {
if (!empty($summary)) {
2019-03-16 12:18:36 +01:00
$item["body"] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item["body"];
}
2017-12-13 08:03:42 +01:00
if ($contact["fetch_further_information"] == 3) {
if (!empty($tags)) {
$item["tag"] = $tags;
} else {
// @todo $preview is never set in this case, is it intended? - @MrPetovan 2018-02-13
$item["tag"] = add_page_keywords($item["plink"], $preview, true, $contact["ffi_keyword_blacklist"]);
2017-12-13 08:03:42 +01:00
}
$item["body"] .= "\n".$item['tag'];
}
2018-01-12 06:55:14 +01:00
// Add the link to the original feed entry if not present in feed
2018-01-29 07:03:39 +01:00
if (($item['plink'] != '') && !strstr($item["body"], $item['plink'])) {
2017-12-13 08:03:42 +01:00
$item["body"] .= "[hr][url]".$item['plink']."[/url]";
}
}
2017-12-13 08:03:42 +01:00
if (!$simulate) {
Logger::log("Stored feed: ".print_r($item, true), Logger::DEBUG);
$notify = Item::isRemoteSelf($contact, $item);
2017-12-13 08:03:42 +01:00
// Distributed items should have a well formatted URI.
// Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
if ($notify) {
$item['guid'] = Item::guidFromUri($orig_plink, $a->getHostName());
2017-12-13 08:03:42 +01:00
unset($item['uri']);
unset($item['parent-uri']);
2018-05-15 21:29:14 +02:00
// Set the delivery priority for "remote self" to "medium"
$notify = PRIORITY_MEDIUM;
2017-12-13 08:03:42 +01:00
}
$id = Item::insert($item, false, $notify);
2018-10-29 22:20:46 +01:00
Logger::log("Feed for contact ".$contact["url"]." stored under id ".$id);
2017-12-13 08:03:42 +01:00
} else {
$items[] = $item;
}
if ($simulate) {
break;
}
2017-03-15 07:00:22 +01:00
}
2017-12-13 08:03:42 +01:00
2017-03-15 07:00:22 +01:00
if ($simulate) {
return ["header" => $author, "items" => $items];
2017-03-15 07:00:22 +01:00
}
}
private static function titleIsBody($title, $body)
{
$title = strip_tags($title);
$title = trim($title);
$title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
$title = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $title);
$body = strip_tags($body);
$body = trim($body);
$body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
$body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
if (strlen($title) < strlen($body)) {
$body = substr($body, 0, strlen($title));
}
if (($title != $body) && (substr($title, -3) == "...")) {
$pos = strrpos($title, "...");
if ($pos > 0) {
$title = substr($title, 0, $pos);
$body = substr($body, 0, $pos);
}
}
return ($title == $body);
}
}