<?php
/**
* @file src/Protocol/Feed.php
- * @brief Imports RSS/RDF/Atom feeds
+ * Imports RSS/RDF/Atom feeds
*
*/
namespace Friendica\Protocol;
use Friendica\Content\Text\HTML;
use Friendica\Core\Logger;
use Friendica\Core\Protocol;
-use Friendica\Core\System;
use Friendica\Database\DBA;
use Friendica\DI;
use Friendica\Model\Item;
-use Friendica\Protocol\ActivityNamespace;
-use Friendica\Util\ParseUrl;
use Friendica\Util\Network;
+use Friendica\Util\ParseUrl;
use Friendica\Util\XML;
/**
- * @brief This class contain functions to import feeds
+ * This class contain functions to import feeds
*
*/
class Feed {
/**
- * @brief Read a RSS/RDF/Atom feed and create an item entry for it
+ * 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 $simulate If enabled, no data is imported
*
- * @return array In simulation mode it returns the header and the first item
+ * @return array Returns the header and the first item in dry run mode
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
- public static function import($xml, $importer, &$contact, &$hub, $simulate = false) {
-
- $a = \get_app();
+ public static function import($xml, array $importer = [], array $contact = [])
+ {
+ $dryRun = empty($importer) && empty($contact);
- if (!$simulate) {
- Logger::log("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], Logger::DEBUG);
+ if ($dryRun) {
+ Logger::info("Test Atom/RSS feed");
} else {
- Logger::log("Test Atom/RSS feed", Logger::DEBUG);
+ Logger::info("Import Atom/RSS feed '" . $contact["name"] . "' (Contact " . $contact["id"] . ") for user " . $importer["uid"]);
}
+
if (empty($xml)) {
- Logger::log('XML is empty.', Logger::DEBUG);
- return;
+ Logger::info('XML is empty.');
+ return [];
}
if (!empty($contact['poll'])) {
if (empty($author["author-name"])) {
$author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
}
+
if (empty($author["author-name"])) {
$author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
}
+
$value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
if ($value != "") {
$author["author-name"] = $value;
}
- if ($simulate) {
+
+ if ($dryRun) {
$author["author-id"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:id/text()');
// See https://tools.ietf.org/html/rfc4287#section-3.2.2
if ($value != "") {
$author["author-nick"] = $value;
}
+
$value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()');
if ($value != "") {
$author["author-location"] = $value;
}
+
$value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()');
if ($value != "") {
$author["author-about"] = $value;
}
+
$avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']");
if (is_object($avatar)) {
foreach ($avatar AS $attribute) {
if (empty($author["author-name"])) {
$author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
}
+
if (empty($author["author-name"])) {
$author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
}
+
$author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
$author["app"] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
$entries = $xpath->query('/rss/channel/item');
}
- if (!$simulate) {
+ if (!$dryRun) {
$author["author-link"] = $contact["url"];
if (empty($author["author-name"])) {
$author["author-name"] = $contact["name"];
}
+
$author["author-avatar"] = $contact["thumb"];
$author["owner-link"] = $contact["url"];
}
$header = [];
- $header["uid"] = $importer["uid"];
+ $header["uid"] = $importer["uid"] ?? 0;
$header["network"] = Protocol::FEED;
$header["wall"] = 0;
$header["origin"] = 0;
$header["verb"] = Activity::POST;
$header["object-type"] = Activity\ObjectType::NOTE;
- $header["contact-id"] = $contact["id"];
+ $header["contact-id"] = $contact["id"] ?? 0;
if (!is_object($entries)) {
- Logger::log("There are no entries in this feed.", Logger::DEBUG);
- return;
+ Logger::info("There are no entries in this feed.");
+ return [];
}
$items = [];
// Importing older entries first
- for($i = $entries->length - 1; $i >= 0;--$i) {
+ for ($i = $entries->length - 1; $i >= 0; --$i) {
$entry = $entries->item($i);
$item = array_merge($header, $author);
}
}
}
+
if (empty($item["plink"])) {
$item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
}
+
if (empty($item["plink"])) {
$item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
}
if (empty($item["uri"])) {
$item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
}
+
if (empty($item["uri"])) {
$item["uri"] = $item["plink"];
}
$item["parent-uri"] = $item["uri"];
- if (!$simulate) {
+ if (!$dryRun) {
$condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
$importer["uid"], $item["uri"], Protocol::FEED, Protocol::DFRN];
$previous = Item::selectFirst(['id'], $condition);
if (DBA::isResult($previous)) {
- Logger::log("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$previous["id"], Logger::DEBUG);
+ Logger::info("Item with uri " . $item["uri"] . " for user " . $importer["uid"] . " already existed under id " . $previous["id"]);
continue;
}
}
if (empty($published)) {
$published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
}
+
if (empty($published)) {
$published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
}
+
$updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry);
if (empty($updated) && !empty($published)) {
if ($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()', $entry);
}
+
if (empty($creator)) {
$creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
}
+
if ($creator != "") {
$item["author-name"] = $creator;
}
+
$creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
if ($creator != "") {
$type = $attribute->textContent;
}
}
+
if (!empty($item["attach"])) {
$item["attach"] .= ',';
} else {
$attachments[] = ["link" => $href, "type" => $type, "length" => $length];
- $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
+ $item["attach"] .= '[attach]href="' . $href . '" length="' . $length . '" type="' . $type . '"[/attach]';
}
$tags = '';
$tags .= ', ';
}
- $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . $hashtag . "]" . $hashtag . "[/url]";
+ $taglink = "#[url=" . DI::baseUrl() . "/search?tag=" . $hashtag . "]" . $hashtag . "[/url]";
$tags .= $taglink;
}
$item["body"] = trim($item["title"]);
}
- $data = ParseUrl::getSiteinfoCached($item['plink'], true);
+ $data = ParseUrl::getSiteinfoCached($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
if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
// 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["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"]);
$item["object-type"] = Activity\ObjectType::BOOKMARK;
unset($item["attach"]);
$item["body"] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item["body"];
}
- if ($contact["fetch_further_information"] == 3) {
+ if (!empty($contact["fetch_further_information"]) && ($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"]);
}
- $item["body"] .= "\n".$item['tag'];
+ $item["body"] .= "\n" . $item['tag'];
}
+
// Add the link to the original feed entry if not present in feed
if (($item['plink'] != '') && !strstr($item["body"], $item['plink'])) {
- $item["body"] .= "[hr][url]".$item['plink']."[/url]";
+ $item["body"] .= "[hr][url]" . $item['plink'] . "[/url]";
}
}
- if (!$simulate) {
- Logger::log("Stored feed: ".print_r($item, true), Logger::DEBUG);
+ if ($dryRun) {
+ $items[] = $item;
+ break;
+ } else {
+ Logger::info("Stored feed: " . print_r($item, true));
$notify = Item::isRemoteSelf($contact, $item);
$id = Item::insert($item, false, $notify);
- Logger::log("Feed for contact ".$contact["url"]." stored under id ".$id);
- } else {
- $items[] = $item;
- }
- if ($simulate) {
- break;
+ Logger::info("Feed for contact " . $contact["url"] . " stored under id " . $id);
}
}
- if ($simulate) {
- return ["header" => $author, "items" => $items];
- }
+ return ["header" => $author, "items" => $items];
}
private static function titleIsBody($title, $body)
projects / friendica.git / blobdiff