clones/friendica
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Merge pull request #8050 from MrPetovan/bug/improve-feed-import

Browse Source 
Rework Feed::import parameters
This commit is contained in:
Tobias Diekershoff
2020-01-03 17:08:54 +01:00
committed by GitHub
parent 7420d07699 ebaf726744
commit d3f5c9f24c
4 changed files with 57 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
include/items.php
View File

@@ -258,7 +258,7 @@ function consume_feed($xml, array $importer, array $contact, &$hub)
if ($contact['network'] === Protocol::FEED) {
Logger::log("Consume feeds", Logger::DEBUG);
Feed::import($xml, $importer, $contact, $hub);
Feed::import($xml, $importer, $contact);
return;
}

5
src/Module/Debug/Feed.php
View File

@@ -29,15 +29,12 @@ class Feed extends BaseModule
if (!empty($_REQUEST['url'])) {
$url = $_REQUEST['url'];
$importer = Model\User::getById(local_user());
$contact_id = Model\Contact::getIdForURL($url, local_user(), true);
$contact = Model\Contact::getById($contact_id);
$xml = Network::fetchUrl($contact['poll']);
$dummy = null;
$import_result = Protocol\Feed::import($xml, $importer, $contact, $dummy, true);
$import_result = Protocol\Feed::import($xml);
$result = [
'input' => $xml,

8
src/Network/Probe.php
View File

@@ -1480,10 +1480,7 @@ class Probe
return false;
}
$feed = $curlResult->getBody();
$dummy1 = null;
$dummy2 = null;
$dummy2 = null;
$feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
$feed_data = Feed::import($feed);
if (!$feed_data) {
return false;
}
@@ -1763,8 +1760,7 @@ class Probe
return false;
}
$feed = $curlResult->getBody();
$dummy1 = $dummy2 = $dummy3 = null;
$feed_data = Feed::import($feed, $dummy1, $dummy2, $dummy3, true);
$feed_data = Feed::import($feed);
if (!$feed_data) {
if (!$probe) {

90
src/Protocol/Feed.php
View File

@@ -14,8 +14,8 @@ use Friendica\Core\Protocol;
use Friendica\Database\DBA;
use Friendica\DI;
use Friendica\Model\Item;
use Friendica\Util\ParseUrl;
use Friendica\Util\Network;
use Friendica\Util\ParseUrl;
use Friendica\Util\XML;
/**
@@ -29,24 +29,23 @@ class Feed {
* @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, $simulate = false) {
public static function import($xml, array $importer = [], array $contact = [])
{
$dryRun = empty($importer) && empty($contact);
$a = \get_app();
if (!$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($contact['poll'])) {
@@ -114,14 +113,17 @@ class Feed {
if (empty($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
@@ -134,14 +136,17 @@ class Feed {
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 $attribute) {
@@ -169,9 +174,11 @@ class Feed {
if (empty($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()');
@@ -179,12 +186,13 @@ class Feed {
$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"];
@@ -193,7 +201,7 @@ class Feed {
}
$header = [];
$header["uid"] = $importer["uid"];
$header["uid"] = $importer["uid"] ?? 0;
$header["network"] = Protocol::FEED;
$header["wall"] = 0;
$header["origin"] = 0;
@@ -202,16 +210,16 @@ class Feed {
$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, $author);
@@ -227,9 +235,11 @@ class Feed {
}
}
}
if (empty($item["plink"])) {
$item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
}
if (empty($item["plink"])) {
$item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
}
@@ -239,6 +249,7 @@ class Feed {
if (empty($item["uri"])) {
$item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
}
if (empty($item["uri"])) {
$item["uri"] = $item["plink"];
}
@@ -249,12 +260,12 @@ class Feed {
$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 " . $previous["id"]);
continue;
}
}
@@ -275,9 +286,11 @@ class Feed {
if (empty($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($published)) {
@@ -291,20 +304,25 @@ class Feed {
if ($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 != "") {
@@ -332,6 +350,7 @@ class Feed {
$type = $attribute->textContent;
}
}
if (!empty($item["attach"])) {
$item["attach"] .= ',';
} else {
@@ -340,7 +359,7 @@ class Feed {
$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 = '';
@@ -418,7 +437,7 @@ class Feed {
$item["body"] = trim($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)) {
@@ -432,7 +451,7 @@ class Feed {
// 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"]);
@@ -448,16 +467,20 @@ class Feed {
// @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);
@@ -474,18 +497,11 @@ class Feed {
$id = Item::insert($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)