Merge pull request #4055 from annando/moved-feed
include/feed.php is now moved to /src as well
This commit is contained in:
GitHub
commit
f99fb2245e
438
include/feed.php
438
include/feed.php
|
|
@ -1,438 +0,0 @@
|
|||
<?php
|
||||
use Friendica\Database\DBM;
|
||||
use Friendica\Core\System;
|
||||
|
||||
require_once("include/html2bbcode.php");
|
||||
require_once("include/items.php");
|
||||
|
||||
/**
|
||||
* @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 $simulate If enabled, no data is imported
|
||||
*
|
||||
* @return array In simulation mode it returns the header and the first item
|
||||
*/
|
||||
function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
|
||||
|
||||
$a = get_app();
|
||||
|
||||
if (!$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;
|
||||
}
|
||||
|
||||
if (!empty($contact['poll'])) {
|
||||
$basepath = $contact['poll'];
|
||||
} elseif (!empty($contact['url'])) {
|
||||
$basepath = $contact['url'];
|
||||
} else {
|
||||
$basepath = '';
|
||||
}
|
||||
|
||||
$doc = new DOMDocument();
|
||||
@$doc->loadXML(trim($xml));
|
||||
$xpath = new DomXPath($doc);
|
||||
$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);
|
||||
|
||||
$author = array();
|
||||
|
||||
// Is it RDF?
|
||||
if ($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"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
|
||||
|
||||
if ($author["author-name"] == "") {
|
||||
$author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
|
||||
}
|
||||
$entries = $xpath->query('/rdf:RDF/rss:item');
|
||||
}
|
||||
|
||||
// Is it Atom?
|
||||
if ($xpath->query('/atom:feed')->length > 0) {
|
||||
$alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
|
||||
if (is_object($alternate)) {
|
||||
foreach ($alternate AS $attributes) {
|
||||
if ($attributes->name == "href") {
|
||||
$author["author-link"] = $attributes->textContent;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if ($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"] = $attributes->textContent;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if ($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;
|
||||
|
||||
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
|
||||
|
||||
if ($author["author-name"] == "") {
|
||||
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
|
||||
}
|
||||
if ($author["author-name"] == "") {
|
||||
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
|
||||
}
|
||||
$value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
|
||||
if ($value != "") {
|
||||
$author["author-name"] = $value;
|
||||
}
|
||||
if ($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()', $context)->item(0)->nodeValue;
|
||||
if ($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;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
$author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
|
||||
|
||||
$author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
|
||||
|
||||
$entries = $xpath->query('/atom:feed/atom:entry');
|
||||
}
|
||||
|
||||
// Is it RSS?
|
||||
if ($xpath->query('/rss/channel')->length > 0) {
|
||||
$author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
|
||||
|
||||
$author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
|
||||
$author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
|
||||
|
||||
if ($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;
|
||||
|
||||
$author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
|
||||
|
||||
$entries = $xpath->query('/rss/channel/item');
|
||||
}
|
||||
|
||||
if (!$simulate) {
|
||||
$author["author-link"] = $contact["url"];
|
||||
|
||||
if ($author["author-name"] == "") {
|
||||
$author["author-name"] = $contact["name"];
|
||||
}
|
||||
$author["author-avatar"] = $contact["thumb"];
|
||||
|
||||
$author["owner-link"] = $contact["url"];
|
||||
$author["owner-name"] = $contact["name"];
|
||||
$author["owner-avatar"] = $contact["thumb"];
|
||||
}
|
||||
|
||||
$header = array();
|
||||
$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;
|
||||
|
||||
$header["contact-id"] = $contact["id"];
|
||||
|
||||
if (!strlen($contact["notify"])) {
|
||||
// one way feed - no remote comment ability
|
||||
$header["last-child"] = 0;
|
||||
}
|
||||
|
||||
if (!is_object($entries)) {
|
||||
logger("There are no entries in this feed.", LOGGER_DEBUG);
|
||||
return;
|
||||
}
|
||||
|
||||
$items = array();
|
||||
|
||||
$entrylist = array();
|
||||
|
||||
foreach ($entries AS $entry) {
|
||||
$entrylist[] = $entry;
|
||||
}
|
||||
foreach (array_reverse($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"] = $attributes->textContent;
|
||||
}
|
||||
}
|
||||
}
|
||||
if ($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;
|
||||
}
|
||||
|
||||
$item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($item["uri"] == "") {
|
||||
$item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
|
||||
}
|
||||
if ($item["uri"] == "") {
|
||||
$item["uri"] = $item["plink"];
|
||||
}
|
||||
|
||||
$orig_plink = $item["plink"];
|
||||
|
||||
$item["plink"] = original_url($item["plink"]);
|
||||
|
||||
$item["parent-uri"] = $item["uri"];
|
||||
|
||||
if (!$simulate) {
|
||||
$r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
|
||||
intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
|
||||
if (DBM::is_result($r)) {
|
||||
logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
|
||||
continue;
|
||||
}
|
||||
}
|
||||
|
||||
$item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($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()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($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()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($updated == "") {
|
||||
$updated = $published;
|
||||
}
|
||||
if ($published != "") {
|
||||
$item["created"] = $published;
|
||||
}
|
||||
if ($updated != "") {
|
||||
$item["edited"] = $updated;
|
||||
}
|
||||
$creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($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"] = $creator;
|
||||
}
|
||||
if ($pubDate != "") {
|
||||
$item["edited"] = $item["created"] = $pubDate;
|
||||
}
|
||||
$creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($creator != "") {
|
||||
$item["author-name"] = $creator;
|
||||
}
|
||||
|
||||
/// @TODO ?
|
||||
// <category>Ausland</category>
|
||||
// <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
|
||||
|
||||
$attachments = array();
|
||||
|
||||
$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"] .= ',';
|
||||
}
|
||||
|
||||
$attachments[] = array("link" => $href, "type" => $type, "length" => $length);
|
||||
|
||||
$item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
|
||||
}
|
||||
|
||||
$tags = '';
|
||||
$categories = $xpath->query("category", $entry);
|
||||
foreach ($categories AS $category) {
|
||||
$hashtag = $category->nodeValue;
|
||||
if ($tags != '') {
|
||||
$tags .= ', ';
|
||||
}
|
||||
|
||||
$taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
|
||||
$tags .= $taglink;
|
||||
}
|
||||
|
||||
$body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
|
||||
|
||||
if ($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()', $entry)->item(0)->nodeValue);
|
||||
}
|
||||
|
||||
// remove the content of the title if it is identically to the body
|
||||
// This helps with auto generated titles e.g. from tumblr
|
||||
if (title_is_body($item["title"], $body)) {
|
||||
$item["title"] = "";
|
||||
}
|
||||
$item["body"] = html2bbcode($body, $basepath);
|
||||
|
||||
if (($item["body"] == '') && ($item["title"] != '')) {
|
||||
$item["body"] = $item["title"];
|
||||
$item["title"] = '';
|
||||
}
|
||||
|
||||
if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
|
||||
$preview = "";
|
||||
|
||||
// Handle enclosures and treat them as preview picture
|
||||
foreach ($attachments AS $attachment) {
|
||||
if ($attachment["type"] == "image/jpeg") {
|
||||
$preview = $attachment["link"];
|
||||
}
|
||||
}
|
||||
|
||||
// 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"]);
|
||||
|
||||
// Replace the content when the title is longer than the body
|
||||
$replace = (strlen($item["title"]) > strlen($item["body"]));
|
||||
|
||||
// Replace it, when there is an image in the body
|
||||
if (strstr($item["body"], '[/img]')) {
|
||||
$replace = true;
|
||||
}
|
||||
|
||||
// Replace it, when there is a link in the body
|
||||
if (strstr($item["body"], '[/url]')) {
|
||||
$replace = true;
|
||||
}
|
||||
|
||||
if ($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'];
|
||||
}
|
||||
if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
|
||||
$item["body"] .= "[hr][url]".$item['plink']."[/url]";
|
||||
}
|
||||
}
|
||||
|
||||
if (!$simulate) {
|
||||
logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
|
||||
|
||||
$notify = item_is_remote_self($contact, $item);
|
||||
|
||||
// Distributed items should have a well formatted URI.
|
||||
// Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
|
||||
if ($notify) {
|
||||
$item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
|
||||
unset($item['uri']);
|
||||
unset($item['parent-uri']);
|
||||
}
|
||||
|
||||
$id = item_store($item, false, $notify);
|
||||
|
||||
logger("Feed for contact ".$contact["url"]." stored under id ".$id);
|
||||
} else {
|
||||
$items[] = $item;
|
||||
}
|
||||
if ($simulate) {
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if ($simulate) {
|
||||
return array("header" => $author, "items" => $items);
|
||||
}
|
||||
}
|
||||
|
|
@ -17,6 +17,7 @@ use Friendica\Model\User;
|
|||
use Friendica\Object\Image;
|
||||
use Friendica\Protocol\DFRN;
|
||||
use Friendica\Protocol\OStatus;
|
||||
use Friendica\Protocol\Feed;
|
||||
|
||||
require_once 'include/bbcode.php';
|
||||
require_once 'include/oembed.php';
|
||||
|
|
@ -26,7 +27,6 @@ require_once 'include/files.php';
|
|||
require_once 'include/text.php';
|
||||
require_once 'include/threads.php';
|
||||
require_once 'include/plaintext.php';
|
||||
require_once 'include/feed.php';
|
||||
require_once 'mod/share.php';
|
||||
require_once 'include/enotify.php';
|
||||
|
||||
|
|
@ -1550,7 +1550,7 @@ function consume_feed($xml, $importer, &$contact, &$hub, $datedir = 0, $pass = 0
|
|||
if ($contact['network'] === NETWORK_FEED) {
|
||||
if ($pass < 2) {
|
||||
logger("Consume feeds", LOGGER_DEBUG);
|
||||
feed_import($xml, $importer, $contact, $hub);
|
||||
Feed::import($xml, $importer, $contact, $hub);
|
||||
}
|
||||
return;
|
||||
}
|
||||
|
|
|
|||
|
|
@ -16,13 +16,13 @@ use Friendica\Core\Config;
|
|||
use Friendica\Database\DBM;
|
||||
use Friendica\Model\Profile;
|
||||
use Friendica\Protocol\Email;
|
||||
use Friendica\Protocol\Feed;
|
||||
use Friendica\Util\XML;
|
||||
|
||||
use dba;
|
||||
use DomXPath;
|
||||
use DOMDocument;
|
||||
|
||||
require_once 'include/feed.php';
|
||||
require_once 'include/network.php';
|
||||
|
||||
/**
|
||||
|
|
@ -1267,7 +1267,7 @@ class Probe
|
|||
return false;
|
||||
}
|
||||
$feed = $ret['body'];
|
||||
$feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
|
||||
$feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
|
||||
if (!$feed_data) {
|
||||
return false;
|
||||
}
|
||||
|
|
@ -1451,7 +1451,7 @@ class Probe
|
|||
return false;
|
||||
}
|
||||
$feed = $ret['body'];
|
||||
$feed_data = feed_import($feed, $dummy1, $dummy2, $dummy3, true);
|
||||
$feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
|
||||
|
||||
if (!$feed_data) {
|
||||
if (!$probe) {
|
||||
|
|
|
|||
|
|
@ -1,6 +1,6 @@
|
|||
<?php
|
||||
/**
|
||||
* @file include/diaspora.php
|
||||
* @file src/Protocol/diaspora.php
|
||||
* @brief The implementation of the diaspora protocol
|
||||
*
|
||||
* The new protocol is described here: http://diaspora.github.io/diaspora_federation/index.html
|
||||
|
|
|
|||
455
src/Protocol/Feed.php
Normal file
455
src/Protocol/Feed.php
Normal file
|
|
@ -0,0 +1,455 @@
|
|||
<?php
|
||||
/**
|
||||
* @file src/Protocol/Feed.php
|
||||
* @brief Imports RSS/RDF/Atom feeds
|
||||
*
|
||||
*/
|
||||
namespace Friendica\Protocol;
|
||||
|
||||
use Friendica\Database\DBM;
|
||||
use Friendica\Core\System;
|
||||
use dba;
|
||||
use DOMDocument;
|
||||
use DomXPath;
|
||||
|
||||
require_once("include/html2bbcode.php");
|
||||
require_once("include/items.php");
|
||||
|
||||
/**
|
||||
* @brief This class contain functions to import feeds
|
||||
*
|
||||
*/
|
||||
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 $simulate If enabled, no data is imported
|
||||
*
|
||||
* @return array In simulation mode it returns the header and the first item
|
||||
*/
|
||||
function import($xml, $importer, &$contact, &$hub, $simulate = false) {
|
||||
|
||||
$a = get_app();
|
||||
|
||||
if (!$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;
|
||||
}
|
||||
|
||||
if (!empty($contact['poll'])) {
|
||||
$basepath = $contact['poll'];
|
||||
} elseif (!empty($contact['url'])) {
|
||||
$basepath = $contact['url'];
|
||||
} else {
|
||||
$basepath = '';
|
||||
}
|
||||
|
||||
$doc = new DOMDocument();
|
||||
@$doc->loadXML(trim($xml));
|
||||
$xpath = new DomXPath($doc);
|
||||
$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);
|
||||
|
||||
$author = array();
|
||||
|
||||
// Is it RDF?
|
||||
if ($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"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:title/text()')->item(0)->nodeValue;
|
||||
|
||||
if ($author["author-name"] == "") {
|
||||
$author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
|
||||
}
|
||||
$entries = $xpath->query('/rdf:RDF/rss:item');
|
||||
}
|
||||
|
||||
// Is it Atom?
|
||||
if ($xpath->query('/atom:feed')->length > 0) {
|
||||
$alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
|
||||
if (is_object($alternate)) {
|
||||
foreach ($alternate AS $attributes) {
|
||||
if ($attributes->name == "href") {
|
||||
$author["author-link"] = $attributes->textContent;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if ($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"] = $attributes->textContent;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if ($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;
|
||||
|
||||
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
|
||||
|
||||
if ($author["author-name"] == "") {
|
||||
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue;
|
||||
}
|
||||
if ($author["author-name"] == "") {
|
||||
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:author/atom:name/text()')->item(0)->nodeValue;
|
||||
}
|
||||
$value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
|
||||
if ($value != "") {
|
||||
$author["author-name"] = $value;
|
||||
}
|
||||
if ($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()', $context)->item(0)->nodeValue;
|
||||
if ($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;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
$author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
|
||||
|
||||
$author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
|
||||
|
||||
$entries = $xpath->query('/atom:feed/atom:entry');
|
||||
}
|
||||
|
||||
// Is it RSS?
|
||||
if ($xpath->query('/rss/channel')->length > 0) {
|
||||
$author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
|
||||
|
||||
$author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
|
||||
$author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
|
||||
|
||||
if ($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;
|
||||
|
||||
$author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
|
||||
|
||||
$entries = $xpath->query('/rss/channel/item');
|
||||
}
|
||||
|
||||
if (!$simulate) {
|
||||
$author["author-link"] = $contact["url"];
|
||||
|
||||
if ($author["author-name"] == "") {
|
||||
$author["author-name"] = $contact["name"];
|
||||
}
|
||||
$author["author-avatar"] = $contact["thumb"];
|
||||
|
||||
$author["owner-link"] = $contact["url"];
|
||||
$author["owner-name"] = $contact["name"];
|
||||
$author["owner-avatar"] = $contact["thumb"];
|
||||
}
|
||||
|
||||
$header = array();
|
||||
$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;
|
||||
|
||||
$header["contact-id"] = $contact["id"];
|
||||
|
||||
if (!strlen($contact["notify"])) {
|
||||
// one way feed - no remote comment ability
|
||||
$header["last-child"] = 0;
|
||||
}
|
||||
|
||||
if (!is_object($entries)) {
|
||||
logger("There are no entries in this feed.", LOGGER_DEBUG);
|
||||
return;
|
||||
}
|
||||
|
||||
$items = array();
|
||||
|
||||
$entrylist = array();
|
||||
|
||||
foreach ($entries AS $entry) {
|
||||
$entrylist[] = $entry;
|
||||
}
|
||||
foreach (array_reverse($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"] = $attributes->textContent;
|
||||
}
|
||||
}
|
||||
}
|
||||
if ($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;
|
||||
}
|
||||
|
||||
$item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($item["uri"] == "") {
|
||||
$item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
|
||||
}
|
||||
if ($item["uri"] == "") {
|
||||
$item["uri"] = $item["plink"];
|
||||
}
|
||||
|
||||
$orig_plink = $item["plink"];
|
||||
|
||||
$item["plink"] = original_url($item["plink"]);
|
||||
|
||||
$item["parent-uri"] = $item["uri"];
|
||||
|
||||
if (!$simulate) {
|
||||
$condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
|
||||
$importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
|
||||
$previous = dba::select('item', ['id'], $condition, ['limit' => 1]);
|
||||
if (DBM::is_result($previous)) {
|
||||
logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], LOGGER_DEBUG);
|
||||
continue;
|
||||
}
|
||||
}
|
||||
|
||||
$item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($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()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($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()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($updated == "") {
|
||||
$updated = $published;
|
||||
}
|
||||
if ($published != "") {
|
||||
$item["created"] = $published;
|
||||
}
|
||||
if ($updated != "") {
|
||||
$item["edited"] = $updated;
|
||||
}
|
||||
$creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($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"] = $creator;
|
||||
}
|
||||
if ($pubDate != "") {
|
||||
$item["edited"] = $item["created"] = $pubDate;
|
||||
}
|
||||
$creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
|
||||
|
||||
if ($creator != "") {
|
||||
$item["author-name"] = $creator;
|
||||
}
|
||||
|
||||
/// @TODO ?
|
||||
// <category>Ausland</category>
|
||||
// <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
|
||||
|
||||
$attachments = array();
|
||||
|
||||
$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"] .= ',';
|
||||
}
|
||||
|
||||
$attachments[] = array("link" => $href, "type" => $type, "length" => $length);
|
||||
|
||||
$item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
|
||||
}
|
||||
|
||||
$tags = '';
|
||||
$categories = $xpath->query("category", $entry);
|
||||
foreach ($categories AS $category) {
|
||||
$hashtag = $category->nodeValue;
|
||||
if ($tags != '') {
|
||||
$tags .= ', ';
|
||||
}
|
||||
|
||||
$taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
|
||||
$tags .= $taglink;
|
||||
}
|
||||
|
||||
$body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
|
||||
|
||||
if ($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()', $entry)->item(0)->nodeValue);
|
||||
}
|
||||
|
||||
// remove the content of the title if it is identically to the body
|
||||
// This helps with auto generated titles e.g. from tumblr
|
||||
if (title_is_body($item["title"], $body)) {
|
||||
$item["title"] = "";
|
||||
}
|
||||
$item["body"] = html2bbcode($body, $basepath);
|
||||
|
||||
if (($item["body"] == '') && ($item["title"] != '')) {
|
||||
$item["body"] = $item["title"];
|
||||
$item["title"] = '';
|
||||
}
|
||||
|
||||
if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
|
||||
$preview = "";
|
||||
|
||||
// Handle enclosures and treat them as preview picture
|
||||
foreach ($attachments AS $attachment) {
|
||||
if ($attachment["type"] == "image/jpeg") {
|
||||
$preview = $attachment["link"];
|
||||
}
|
||||
}
|
||||
|
||||
// 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"]);
|
||||
|
||||
// Replace the content when the title is longer than the body
|
||||
$replace = (strlen($item["title"]) > strlen($item["body"]));
|
||||
|
||||
// Replace it, when there is an image in the body
|
||||
if (strstr($item["body"], '[/img]')) {
|
||||
$replace = true;
|
||||
}
|
||||
|
||||
// Replace it, when there is a link in the body
|
||||
if (strstr($item["body"], '[/url]')) {
|
||||
$replace = true;
|
||||
}
|
||||
|
||||
if ($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'];
|
||||
}
|
||||
if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
|
||||
$item["body"] .= "[hr][url]".$item['plink']."[/url]";
|
||||
}
|
||||
}
|
||||
|
||||
if (!$simulate) {
|
||||
logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
|
||||
|
||||
$notify = item_is_remote_self($contact, $item);
|
||||
|
||||
// Distributed items should have a well formatted URI.
|
||||
// Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
|
||||
if ($notify) {
|
||||
$item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
|
||||
unset($item['uri']);
|
||||
unset($item['parent-uri']);
|
||||
}
|
||||
|
||||
$id = item_store($item, false, $notify);
|
||||
|
||||
logger("Feed for contact ".$contact["url"]." stored under id ".$id);
|
||||
} else {
|
||||
$items[] = $item;
|
||||
}
|
||||
if ($simulate) {
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if ($simulate) {
|
||||
return array("header" => $author, "items" => $items);
|
||||
}
|
||||
}
|
||||
}
|
||||
Loading…
Reference in a new issue