The feed is now a class
This commit is contained in:
parent
fc55a4b9c0
commit
74ba0c896a
|
|
@ -1,438 +1,453 @@
|
||||||
<?php
|
<?php
|
||||||
|
/**
|
||||||
|
* @file src/Protocol/Feed.php
|
||||||
|
* @brief Imports RSS/RDF/Atom feeds
|
||||||
|
*
|
||||||
|
*/
|
||||||
|
namespace Friendica\Protocol;
|
||||||
|
|
||||||
use Friendica\Database\DBM;
|
use Friendica\Database\DBM;
|
||||||
use Friendica\Core\System;
|
use Friendica\Core\System;
|
||||||
|
use dba;
|
||||||
|
|
||||||
require_once("include/html2bbcode.php");
|
require_once("include/html2bbcode.php");
|
||||||
require_once("include/items.php");
|
require_once("include/items.php");
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* @brief Read a RSS/RDF/Atom feed and create an item entry for it
|
* @brief This class contain functions to import feeds
|
||||||
*
|
*
|
||||||
* @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) {
|
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();
|
$a = get_app();
|
||||||
|
|
||||||
if (!$simulate) {
|
if (!$simulate) {
|
||||||
logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
|
logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
|
||||||
} else {
|
} else {
|
||||||
logger("Test Atom/RSS feed", LOGGER_DEBUG);
|
logger("Test Atom/RSS feed", LOGGER_DEBUG);
|
||||||
}
|
}
|
||||||
if ($xml == "") {
|
if ($xml == "") {
|
||||||
logger('XML is empty.', LOGGER_DEBUG);
|
logger('XML is empty.', LOGGER_DEBUG);
|
||||||
return;
|
return;
|
||||||
}
|
|
||||||
|
|
||||||
if (!empty($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 = $xpath->query('/rdf:RDF/rss:item');
|
|
||||||
}
|
|
||||||
|
|
||||||
// Is it Atom?
|
if (!empty($contact['poll'])) {
|
||||||
if ($xpath->query('/atom:feed')->length > 0) {
|
$basepath = $contact['poll'];
|
||||||
$alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
|
} elseif (!empty($contact['url'])) {
|
||||||
if (is_object($alternate)) {
|
$basepath = $contact['url'];
|
||||||
foreach ($alternate AS $attributes) {
|
} else {
|
||||||
if ($attributes->name == "href") {
|
$basepath = '';
|
||||||
$author["author-link"] = $attributes->textContent;
|
}
|
||||||
}
|
|
||||||
|
$doc = new DOMDocument();
|
||||||
|
@$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 = $xpath->query('/rdf:RDF/rss:item');
|
||||||
}
|
}
|
||||||
|
|
||||||
if ($author["author-link"] == "") {
|
// Is it Atom?
|
||||||
$author["author-link"] = $author["author-id"];
|
if ($xpath->query('/atom:feed')->length > 0) {
|
||||||
}
|
$alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
|
||||||
if ($author["author-link"] == "") {
|
if (is_object($alternate)) {
|
||||||
$self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
|
foreach ($alternate AS $attributes) {
|
||||||
if (is_object($self)) {
|
|
||||||
foreach ($self AS $attributes) {
|
|
||||||
if ($attributes->name == "href") {
|
if ($attributes->name == "href") {
|
||||||
$author["author-link"] = $attributes->textContent;
|
$author["author-link"] = $attributes->textContent;
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
}
|
|
||||||
|
|
||||||
if ($author["author-link"] == "") {
|
if ($author["author-link"] == "") {
|
||||||
$author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
|
$author["author-link"] = $author["author-id"];
|
||||||
}
|
|
||||||
$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()', $context)->item(0)->nodeValue;
|
if ($author["author-link"] == "") {
|
||||||
if ($value != "") {
|
$self = $xpath->query("atom:link[@rel='self']")->item(0)->attributes;
|
||||||
$author["author-location"] = $value;
|
if (is_object($self)) {
|
||||||
}
|
foreach ($self AS $attributes) {
|
||||||
$value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
|
if ($attributes->name == "href") {
|
||||||
if ($value != "") {
|
$author["author-link"] = $attributes->textContent;
|
||||||
$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"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
|
if ($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["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
|
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
|
||||||
|
|
||||||
$entries = $xpath->query('/atom:feed/atom:entry');
|
if ($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;
|
||||||
|
|
||||||
// Is it RSS?
|
$value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
|
||||||
if ($xpath->query('/rss/channel')->length > 0) {
|
if ($value != "") {
|
||||||
$author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
|
$author["author-nick"] = $value;
|
||||||
|
}
|
||||||
$author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
|
$value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
|
||||||
$author["author-avatar"] = $xpath->evaluate('/rss/channel/image/url/text()')->item(0)->nodeValue;
|
if ($value != "") {
|
||||||
|
$author["author-location"] = $value;
|
||||||
if ($author["author-name"] == "") {
|
}
|
||||||
$author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue;
|
$value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
|
||||||
}
|
if ($value != "") {
|
||||||
if ($author["author-name"] == "") {
|
$author["author-about"] = $value;
|
||||||
$author["author-name"] = $xpath->evaluate('/rss/channel/description/text()')->item(0)->nodeValue;
|
}
|
||||||
}
|
$avatar = $xpath->evaluate("atom:author/atom:link[@rel='avatar']")->item(0)->attributes;
|
||||||
$author["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
|
if (is_object($avatar)) {
|
||||||
|
foreach ($avatar AS $attributes) {
|
||||||
$author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
|
if ($attributes->name == "href") {
|
||||||
|
$author["author-avatar"] = $attributes->textContent;
|
||||||
$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"] == "") {
|
$author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
|
||||||
$item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
|
|
||||||
}
|
$author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
|
||||||
if ($item["plink"] == "") {
|
|
||||||
$item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
|
$entries = $xpath->query('/atom:feed/atom:entry');
|
||||||
}
|
}
|
||||||
|
|
||||||
$item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
|
// Is it RSS?
|
||||||
|
if ($xpath->query('/rss/channel')->length > 0) {
|
||||||
|
$author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
|
||||||
|
|
||||||
if ($item["uri"] == "") {
|
$author["author-name"] = $xpath->evaluate('/rss/channel/title/text()')->item(0)->nodeValue;
|
||||||
$item["uri"] = $xpath->evaluate('guid/text()', $entry)->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 = $xpath->query('/rss/channel/item');
|
||||||
}
|
}
|
||||||
if ($item["uri"] == "") {
|
|
||||||
$item["uri"] = $item["plink"];
|
|
||||||
}
|
|
||||||
|
|
||||||
$orig_plink = $item["plink"];
|
|
||||||
|
|
||||||
$item["plink"] = original_url($item["plink"]);
|
|
||||||
|
|
||||||
$item["parent-uri"] = $item["uri"];
|
|
||||||
|
|
||||||
if (!$simulate) {
|
if (!$simulate) {
|
||||||
$r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')",
|
$author["author-link"] = $contact["url"];
|
||||||
intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN));
|
|
||||||
if (DBM::is_result($r)) {
|
if ($author["author-name"] == "") {
|
||||||
logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG);
|
$author["author-name"] = $contact["name"];
|
||||||
continue;
|
|
||||||
}
|
}
|
||||||
|
$author["author-avatar"] = $contact["thumb"];
|
||||||
|
|
||||||
|
$author["owner-link"] = $contact["url"];
|
||||||
|
$author["owner-name"] = $contact["name"];
|
||||||
|
$author["owner-avatar"] = $contact["thumb"];
|
||||||
}
|
}
|
||||||
|
|
||||||
$item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
|
$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;
|
||||||
|
$header["object-type"] = ACTIVITY_OBJ_NOTE;
|
||||||
|
|
||||||
if ($item["title"] == "") {
|
$header["contact-id"] = $contact["id"];
|
||||||
$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 == "") {
|
if (!strlen($contact["notify"])) {
|
||||||
$published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
|
// one way feed - no remote comment ability
|
||||||
|
$header["last-child"] = 0;
|
||||||
}
|
}
|
||||||
if ($published == "") {
|
|
||||||
$published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
|
|
||||||
}
|
|
||||||
$updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
|
|
||||||
|
|
||||||
if ($updated == "") {
|
if (!is_object($entries)) {
|
||||||
$updated = $published;
|
logger("There are no entries in this feed.", LOGGER_DEBUG);
|
||||||
|
return;
|
||||||
}
|
}
|
||||||
if ($published != "") {
|
|
||||||
$item["created"] = $published;
|
|
||||||
}
|
|
||||||
if ($updated != "") {
|
|
||||||
$item["edited"] = $updated;
|
|
||||||
}
|
|
||||||
$creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
|
|
||||||
|
|
||||||
if ($creator == "") {
|
$items = array();
|
||||||
$creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue;
|
|
||||||
|
$entrylist = array();
|
||||||
|
|
||||||
|
foreach ($entries AS $entry) {
|
||||||
|
$entrylist[] = $entry;
|
||||||
}
|
}
|
||||||
if ($creator == "") {
|
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;
|
$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 != "") {
|
if ($creator != "") {
|
||||||
$item["author-name"] = $creator;
|
$item["author-name"] = $creator;
|
||||||
}
|
}
|
||||||
|
|
||||||
/// @TODO ?
|
/// @TODO ?
|
||||||
// <category>Ausland</category>
|
// <category>Ausland</category>
|
||||||
// <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
|
// <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
|
||||||
|
|
||||||
$attachments = array();
|
$attachments = array();
|
||||||
|
|
||||||
$enclosures = $xpath->query("enclosure", $entry);
|
$enclosures = $xpath->query("enclosure", $entry);
|
||||||
foreach ($enclosures AS $enclosure) {
|
foreach ($enclosures AS $enclosure) {
|
||||||
$href = "";
|
$href = "";
|
||||||
$length = "";
|
$length = "";
|
||||||
$type = "";
|
$type = "";
|
||||||
$title = "";
|
$title = "";
|
||||||
|
|
||||||
foreach ($enclosure->attributes AS $attributes) {
|
foreach ($enclosure->attributes AS $attributes) {
|
||||||
if ($attributes->name == "url") {
|
if ($attributes->name == "url") {
|
||||||
$href = $attributes->textContent;
|
$href = $attributes->textContent;
|
||||||
} elseif ($attributes->name == "length") {
|
} elseif ($attributes->name == "length") {
|
||||||
$length = $attributes->textContent;
|
$length = $attributes->textContent;
|
||||||
} elseif ($attributes->name == "type") {
|
} elseif ($attributes->name == "type") {
|
||||||
$type = $attributes->textContent;
|
$type = $attributes->textContent;
|
||||||
|
}
|
||||||
}
|
}
|
||||||
}
|
if (strlen($item["attach"])) {
|
||||||
if (strlen($item["attach"])) {
|
$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"];
|
|
||||||
}
|
}
|
||||||
|
|
||||||
|
$attachments[] = array("link" => $href, "type" => $type, "length" => $length);
|
||||||
|
|
||||||
|
$item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
|
||||||
}
|
}
|
||||||
|
|
||||||
// Remove a possible link to the item itself
|
$tags = '';
|
||||||
$item["body"] = str_replace($item["plink"], '', $item["body"]);
|
$categories = $xpath->query("category", $entry);
|
||||||
$item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
|
foreach ($categories AS $category) {
|
||||||
|
$hashtag = $category->nodeValue;
|
||||||
|
if ($tags != '') {
|
||||||
|
$tags .= ', ';
|
||||||
|
}
|
||||||
|
|
||||||
// Replace the content when the title is longer than the body
|
$taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
|
||||||
$replace = (strlen($item["title"]) > strlen($item["body"]));
|
$tags .= $taglink;
|
||||||
|
|
||||||
// Replace it, when there is an image in the body
|
|
||||||
if (strstr($item["body"], '[/img]')) {
|
|
||||||
$replace = true;
|
|
||||||
}
|
}
|
||||||
|
|
||||||
// Replace it, when there is a link in the body
|
$body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
|
||||||
if (strstr($item["body"], '[/url]')) {
|
|
||||||
$replace = true;
|
if ($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()', $entry)->item(0)->nodeValue);
|
||||||
}
|
}
|
||||||
|
|
||||||
if ($replace) {
|
// 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["body"] = $item["title"];
|
||||||
|
$item["title"] = '';
|
||||||
}
|
}
|
||||||
// We always strip the title since it will be added in the page information
|
|
||||||
$item["title"] = "";
|
if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
|
||||||
$item["body"] = $item["body"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
|
$preview = "";
|
||||||
$item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
|
|
||||||
$item["object-type"] = ACTIVITY_OBJ_BOOKMARK;
|
// Handle enclosures and treat them as preview picture
|
||||||
unset($item["attach"]);
|
foreach ($attachments AS $attachment) {
|
||||||
} else {
|
if ($attachment["type"] == "image/jpeg") {
|
||||||
if ($contact["fetch_further_information"] == 3) {
|
$preview = $attachment["link"];
|
||||||
if (!empty($tags)) {
|
}
|
||||||
$item["tag"] = $tags;
|
}
|
||||||
} else {
|
|
||||||
$item["tag"] = add_page_keywords($item["plink"], false, $preview, true, $contact["ffi_keyword_blacklist"]);
|
// 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]";
|
||||||
}
|
}
|
||||||
$item["body"] .= "\n".$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 (!$simulate) {
|
|
||||||
logger("Stored feed: ".print_r($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) {
|
if ($simulate) {
|
||||||
break;
|
return array("header" => $author, "items" => $items);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
if ($simulate) {
|
|
||||||
return array("header" => $author, "items" => $items);
|
|
||||||
}
|
|
||||||
}
|
}
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue