friendica/src/Protocol/Feed.php

478 lines
15 KiB
PHP
Raw Normal View History

<?php
2017-12-13 02:03:42 -05:00
/**
* @file src/Protocol/Feed.php
* @brief Imports RSS/RDF/Atom feeds
*
*/
namespace Friendica\Protocol;
2017-11-07 22:57:46 -05:00
use Friendica\Database\DBM;
use Friendica\Core\System;
use Friendica\Model\Item;
use Friendica\Util\Network;
2017-12-13 02:03:42 -05:00
use dba;
2017-12-13 02:09:20 -05:00
use DOMDocument;
use DOMXPath;
require_once 'include/dba.php';
require_once 'include/html2bbcode.php';
require_once 'include/items.php';
2016-02-16 02:06:55 -05:00
/**
2017-12-13 02:03:42 -05:00
* @brief This class contain functions to import feeds
2016-02-16 02:06:55 -05:00
*
*/
2017-12-13 02:03:42 -05:00
class Feed {
/**
* @brief Read a RSS/RDF/Atom feed and create an item entry for it
*
* @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 $simulate If enabled, no data is imported
*
* @return array In simulation mode it returns the header and the first item
*/
public static function import($xml, $importer, &$contact, &$hub, $simulate = false) {
2017-12-13 02:03:42 -05:00
$a = get_app();
if (!$simulate) {
logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
} else {
logger("Test Atom/RSS feed", LOGGER_DEBUG);
}
if ($xml == "") {
logger('XML is empty.', LOGGER_DEBUG);
return;
}
2017-12-13 02:03:42 -05:00
if (!empty($contact['poll'])) {
$basepath = $contact['poll'];
} elseif (!empty($contact['url'])) {
$basepath = $contact['url'];
} else {
$basepath = '';
}
2017-12-13 02:03:42 -05:00
$doc = new DOMDocument();
@$doc->loadXML(trim($xml));
$xpath = new DOMXPath($doc);
2017-12-13 02:03:42 -05:00
$xpath->registerNamespace('atom', NAMESPACE_ATOM1);
$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/");
$xpath->registerNamespace('poco', NAMESPACE_POCO);
$author = [];
2017-12-13 02:03:42 -05:00
// Is it RDF?
if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
$author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue;
$author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($author["author-name"] == "") {
$author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
2017-03-15 02:00:22 -04:00
}
2017-12-13 02:03:42 -05:00
$entries = $xpath->query('/rdf:RDF/rss:item');
2017-03-15 02:00:22 -04:00
}
2017-12-13 02:03:42 -05:00
// Is it Atom?
if ($xpath->query('/atom:feed')->length > 0) {
$alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
if (is_object($alternate)) {
foreach ($alternate AS $attributes) {
2017-03-15 02:00:22 -04:00
if ($attributes->name == "href") {
$author["author-link"] = $attributes->textContent;
2017-03-15 02:00:22 -04:00
}
}
}
2017-12-13 02:03:42 -05:00
if ($author["author-link"] == "") {
$author["author-link"] = $author["author-id"];
}
if ($author["author-link"] == "") {
$self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
if (is_object($self)) {
foreach ($self AS $attributes) {
if ($attributes->name == "href") {
$author["author-link"] = $attributes->textContent;
}
}
}
}
2017-12-13 02:03:42 -05:00
if ($author["author-link"] == "") {
$author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
}
$author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($author["author-name"] == "") {
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
2017-03-15 02:00:22 -04:00
}
2017-12-13 02:03:42 -05:00
if ($author["author-name"] == "") {
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
2017-03-15 02:00:22 -04:00
}
2017-12-13 02:03:42 -05:00
$value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
2017-03-15 02:00:22 -04:00
if ($value != "") {
2017-12-13 02:03:42 -05:00
$author["author-name"] = $value;
2017-03-15 02:00:22 -04:00
}
2017-12-13 02:03:42 -05:00
if ($simulate) {
$author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue;
$value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
if ($value != "") {
$author["author-nick"] = $value;
}
$value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($value != "") {
$author["author-location"] = $value;
}
$value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
if ($value != "") {
$author["author-about"] = $value;
}
$avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
if (is_object($avatar)) {
foreach ($avatar AS $attributes) {
if ($attributes->name == "href") {
$author["author-avatar"] = $attributes->textContent;
}
2017-08-21 16:21:04 -04:00
}
}
}
2017-12-13 02:03:42 -05:00
$author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
$author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
$entries = $xpath->query('/atom:feed/atom:entry');
}
2017-12-13 02:03:42 -05:00
// Is it RSS?
if ($xpath->query('/rss/channel')->length > 0) {
$author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
$author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
$author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($author["author-name"] == "") {
$author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
}
if ($author["author-name"] == "") {
$author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
}
$author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
$author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
$entries = $xpath->query('/rss/channel/item');
}
if (!$simulate) {
$author["author-link"] = $contact["url"];
2017-12-13 02:03:42 -05:00
if ($author["author-name"] == "") {
$author["author-name"] = $contact["name"];
}
$author["author-avatar"] = $contact["thumb"];
2017-12-13 02:03:42 -05:00
$author["owner-link"] = $contact["url"];
$author["owner-name"] = $contact["name"];
$author["owner-avatar"] = $contact["thumb"];
2017-03-15 02:00:22 -04:00
}
$header = [];
2017-12-13 02:03:42 -05:00
$header["uid"] = $importer["uid"];
$header["network"] = NETWORK_FEED;
$header["type"] = "remote";
$header["wall"] = 0;
$header["origin"] = 0;
$header["gravity"] = GRAVITY_PARENT;
$header["private"] = 2;
$header["verb"] = ACTIVITY_POST;
$header["object-type"] = ACTIVITY_OBJ_NOTE;
2017-12-13 02:03:42 -05:00
$header["contact-id"] = $contact["id"];
2015-10-03 07:58:10 -04:00
2017-12-13 02:03:42 -05:00
if (!is_object($entries)) {
logger("There are no entries in this feed.", LOGGER_DEBUG);
return;
}
$items = [];
$entrylist = [];
2017-12-13 02:03:42 -05:00
foreach ($entries AS $entry) {
$entrylist[] = $entry;
2017-03-15 02:00:22 -04:00
}
2017-12-13 02:03:42 -05:00
foreach (array_reverse($entrylist) AS $entry) {
$item = array_merge($header, $author);
$alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes;
if (!is_object($alternate)) {
$alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
}
if (is_object($alternate)) {
foreach ($alternate AS $attributes) {
if ($attributes->name == "href") {
$item["plink"] = $attributes->textContent;
}
2017-03-15 02:00:22 -04:00
}
}
2017-12-13 02:03:42 -05:00
if ($item["plink"] == "") {
$item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
}
if ($item["plink"] == "") {
$item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
}
2017-12-13 02:03:42 -05:00
$item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($item["uri"] == "") {
$item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
}
if ($item["uri"] == "") {
$item["uri"] = $item["plink"];
}
2017-12-13 02:03:42 -05:00
$orig_plink = $item["plink"];
$item["plink"] = Network::finalUrl($item["plink"]);
2017-12-13 02:03:42 -05:00
$item["parent-uri"] = $item["uri"];
2017-12-13 02:03:42 -05:00
if (!$simulate) {
$condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
$importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
2018-01-10 08:36:02 -05:00
$previous = dba::selectFirst('item', ['id'], $condition);
2017-12-13 02:03:42 -05:00
if (DBM::is_result($previous)) {
logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG);
continue;
}
}
2017-12-13 02:03:42 -05:00
$item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($item["title"] == "") {
$item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
}
if ($item["title"] == "") {
$item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
}
$published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($published == "") {
$published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
}
if ($published == "") {
$published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
}
$updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($updated == "") {
$updated = $published;
}
if ($published != "") {
$item["created"] = $published;
}
if ($updated != "") {
$item["edited"] = $updated;
}
$creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($creator == "") {
$creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
}
if ($creator == "") {
$creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
}
if ($creator != "") {
$item["author-name"] = $creator;
}
$creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
2017-12-13 02:03:42 -05:00
if ($creator != "") {
$item["author-name"] = $creator;
}
2017-12-13 02:03:42 -05:00
/// @TODO ?
// <category>Ausland</category>
// <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
$attachments = [];
2017-12-13 02:03:42 -05:00
$enclosures = $xpath->query("enclosure", $entry);
foreach ($enclosures AS $enclosure) {
$href = "";
$length = "";
$type = "";
$title = "";
foreach ($enclosure->attributes AS $attributes) {
if ($attributes->name == "url") {
$href = $attributes->textContent;
} elseif ($attributes->name == "length") {
$length = $attributes->textContent;
} elseif ($attributes->name == "type") {
$type = $attributes->textContent;
}
}
if (strlen($item["attach"])) {
$item["attach"] .= ',';
2017-03-15 02:00:22 -04:00
}
$attachments[] = ["link" => $href, "type" => $type, "length" => $length];
2017-12-13 02:03:42 -05:00
$item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
}
2017-12-13 02:03:42 -05:00
$tags = '';
$categories = $xpath->query("category", $entry);
foreach ($categories AS $category) {
$hashtag = $category->nodeValue;
if ($tags != '') {
$tags .= ', ';
}
2017-12-13 02:03:42 -05:00
$taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
$tags .= $taglink;
}
2017-12-13 02:03:42 -05:00
$body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
2017-10-17 05:10:19 -04:00
2017-12-13 02:03:42 -05:00
if ($body == "") {
$body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
}
if ($body == "") {
$body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
}
if ($body == "") {
$body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
}
2017-10-17 05:10:19 -04:00
2017-12-13 02:03:42 -05:00
// remove the content of the title if it is identically to the body
// This helps with auto generated titles e.g. from tumblr
if (self::titleIsBody($item["title"], $body)) {
2017-12-13 02:03:42 -05:00
$item["title"] = "";
}
$item["body"] = html2bbcode($body, $basepath);
2017-10-17 05:10:19 -04:00
2017-12-13 02:03:42 -05:00
if (($item["body"] == '') && ($item["title"] != '')) {
$item["body"] = $item["title"];
$item["title"] = '';
}
2017-10-17 05:10:19 -04:00
2017-12-13 02:03:42 -05:00
if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
$preview = "";
2017-12-13 02:03:42 -05:00
// Handle enclosures and treat them as preview picture
foreach ($attachments AS $attachment) {
if ($attachment["type"] == "image/jpeg") {
$preview = $attachment["link"];
}
2017-03-15 02:00:22 -04:00
}
2017-12-13 02:03:42 -05:00
// Remove a possible link to the item itself
$item["body"] = str_replace($item["plink"], '', $item["body"]);
$item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
2017-10-17 07:39:09 -04:00
2017-12-13 02:03:42 -05:00
// Replace the content when the title is longer than the body
$replace = (strlen($item["title"]) > strlen($item["body"]));
2017-10-17 05:58:29 -04:00
2017-12-13 02:03:42 -05:00
// Replace it, when there is an image in the body
if (strstr($item["body"], '[/img]')) {
$replace = true;
}
2017-10-17 05:58:29 -04:00
2017-12-13 02:03:42 -05:00
// Replace it, when there is a link in the body
if (strstr($item["body"], '[/url]')) {
$replace = true;
}
2017-10-17 05:58:29 -04:00
2017-12-13 02:03:42 -05:00
if ($replace) {
$item["body"] = $item["title"];
}
// 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"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
$item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
unset($item["attach"]);
} else {
if ($contact["fetch_further_information"] == 3) {
if (!empty($tags)) {
$item["tag"] = $tags;
} else {
$item["tag"] = add_page_keywords($item["plink"], false, $preview, true, $contact["ffi_keyword_blacklist"]);
}
$item["body"] .= "\n".$item['tag'];
}
2018-01-12 00:55:14 -05:00
// Add the link to the original feed entry if not present in feed
if (!strstr($item["body"], $item['plink']) && ($item['plink'] != '')) {
2017-12-13 02:03:42 -05:00
$item["body"] .= "[hr][url]".$item['plink']."[/url]";
}
}
2017-12-13 02:03:42 -05:00
if (!$simulate) {
logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
$notify = Item::isRemoteSelf($contact, $item);
2017-12-13 02:03:42 -05:00
// Distributed items should have a well formatted URI.
// Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
if ($notify) {
2018-01-20 19:18:31 -05:00
$item['guid'] = Item::guidFromUri($orig_plink, $a->get_hostname());
2017-12-13 02:03:42 -05:00
unset($item['uri']);
unset($item['parent-uri']);
}
$id = Item::insert($item, false, $notify);
2017-12-13 02:03:42 -05:00
logger("Feed for contact ".$contact["url"]." stored under id ".$id);
} else {
$items[] = $item;
}
if ($simulate) {
break;
}
2017-03-15 02:00:22 -04:00
}
2017-12-13 02:03:42 -05:00
2017-03-15 02:00:22 -04:00
if ($simulate) {
return ["header" => $author, "items" => $items];
2017-03-15 02:00:22 -04:00
}
}
private static function titleIsBody($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);
}
}