X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=src%2FProtocol%2FFeed.php;h=7bae9ef588894e6f7e467e7d823818933a33914a;hb=d60eb25ba789fa00c3f39e0a6337d0e11e22f449;hp=49ae8df5c5e29591cda5197dc0a514852de8dcd2;hpb=fc55a4b9c05123b806d30c228997397ca7897a11;p=friendica.git
diff --git a/src/Protocol/Feed.php b/src/Protocol/Feed.php
index 49ae8df5c5..7bae9ef588 100644
--- a/src/Protocol/Feed.php
+++ b/src/Protocol/Feed.php
@@ -1,438 +1,476 @@
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"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
+ if (!$simulate) {
+ logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
+ } else {
+ logger("Test Atom/RSS feed", LOGGER_DEBUG);
+ }
+ if (empty($xml)) {
+ logger('XML is empty.', LOGGER_DEBUG);
+ return;
}
- $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"] = $attributes->textContent;
- }
+ if (!empty($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 = [];
+ $entries = null;
+
+ // Is it RDF?
+ if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
+ $author["author-link"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
+ $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
+
+ if (empty($author["author-name"])) {
+ $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
}
+ $entries = $xpath->query('/rdf:RDF/rss:item');
}
- if ($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"] = $attributes->textContent;
+ // Is it Atom?
+ if ($xpath->query('/atom:feed')->length > 0) {
+ $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']");
+ if (is_object($alternate)) {
+ foreach ($alternate AS $attribute) {
+ if ($attribute->name == "href") {
+ $author["author-link"] = $attribute->textContent;
}
}
}
- }
- if ($author["author-link"] == "") {
- $author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue;
- }
- $author["author-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
+ if (empty($author["author-link"])) {
+ $self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']");
+ if (is_object($self)) {
+ foreach ($self AS $attribute) {
+ if ($attribute->name == "href") {
+ $author["author-link"] = $attribute->textContent;
+ }
+ }
+ }
+ }
- $author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/text()')->item(0)->nodeValue;
+ if (empty($author["author-link"])) {
+ $author["author-link"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()');
+ }
+ $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()');
- 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;
+ $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()');
- $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
- if ($value != "") {
- $author["author-nick"] = $value;
+ if (empty($author["author-name"])) {
+ $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
}
- $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
- if ($value != "") {
- $author["author-location"] = $value;
+ if (empty($author["author-name"])) {
+ $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
}
- $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
+ $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
if ($value != "") {
- $author["author-about"] = $value;
+ $author["author-name"] = $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"] = $attributes->textContent;
+ if ($simulate) {
+ $author["author-id"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:uri/text()');
+
+ $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()');
+ 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 $attribute) {
+ if ($attribute->name == "href") {
+ $author["author-avatar"] = $attribute->textContent;
+ }
}
}
}
- }
- $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
+ $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()');
- $author["app"] = $xpath->evaluate('/atom:feed/atom:generator/text()')->item(0)->nodeValue;
+ $author["app"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()');
- $entries = $xpath->query('/atom:feed/atom:entry');
- }
+ $entries = $xpath->query('/atom:feed/atom:entry');
+ }
- // Is it RSS?
- if ($xpath->query('/rss/channel')->length > 0) {
- $author["author-link"] = $xpath->evaluate('/rss/channel/link/text()')->item(0)->nodeValue;
+ // Is it RSS?
+ if ($xpath->query('/rss/channel')->length > 0) {
+ $author["author-link"] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
- $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;
+ $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
+ $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
- 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"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
+ if (empty($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"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
+ $author["app"] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
- $entries = $xpath->query('/rss/channel/item');
- }
+ $entries = $xpath->query('/rss/channel/item');
+ }
+
+ if (!$simulate) {
+ $author["author-link"] = $contact["url"];
- if (!$simulate) {
- $author["author-link"] = $contact["url"];
+ if (empty($author["author-name"])) {
+ $author["author-name"] = $contact["name"];
+ }
+ $author["author-avatar"] = $contact["thumb"];
- if ($author["author-name"] == "") {
- $author["author-name"] = $contact["name"];
+ $author["owner-link"] = $contact["url"];
+ $author["owner-name"] = $contact["name"];
+ $author["owner-avatar"] = $contact["thumb"];
}
- $author["author-avatar"] = $contact["thumb"];
- $author["owner-link"] = $contact["url"];
- $author["owner-name"] = $contact["name"];
- $author["owner-avatar"] = $contact["thumb"];
- }
+ $header = [];
+ $header["uid"] = $importer["uid"];
+ $header["network"] = NETWORK_FEED;
+ $header["wall"] = 0;
+ $header["origin"] = 0;
+ $header["gravity"] = GRAVITY_PARENT;
+ $header["private"] = 2;
+ $header["verb"] = ACTIVITY_POST;
+ $header["object-type"] = ACTIVITY_OBJ_NOTE;
- $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;
- }
+ $header["contact-id"] = $contact["id"];
- if (!is_object($entries)) {
- logger("There are no entries in this feed.", LOGGER_DEBUG);
- return;
- }
-
- $items = array();
+ if (!is_object($entries)) {
+ logger("There are no entries in this feed.", LOGGER_DEBUG);
+ return;
+ }
- $entrylist = array();
+ $items = [];
+ // Importing older entries first
+ for($i = $entries->length - 1; $i >= 0;--$i) {
+ $entry = $entries->item($i);
- foreach ($entries AS $entry) {
- $entrylist[] = $entry;
- }
- foreach (array_reverse($entrylist) AS $entry) {
- $item = array_merge($header, $author);
+ $item = array_merge($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;
+ $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry);
+ if (!is_object($alternate)) {
+ $alternate = XML::getFirstAttributes($xpath, "atom:link", $entry);
+ }
+ if (is_object($alternate)) {
+ foreach ($alternate AS $attribute) {
+ if ($attribute->name == "href") {
+ $item["plink"] = $attribute->textContent;
+ }
}
}
- }
- if ($item["plink"] == "") {
- $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue;
- }
- if ($item["plink"] == "") {
- $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
- }
+ if (empty($item["plink"])) {
+ $item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
+ }
+ if (empty($item["plink"])) {
+ $item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
+ }
- $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue;
+ $item["uri"] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry);
- if ($item["uri"] == "") {
- $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
- }
- if ($item["uri"] == "") {
- $item["uri"] = $item["plink"];
- }
+ if (empty($item["uri"])) {
+ $item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
+ }
+ if (empty($item["uri"])) {
+ $item["uri"] = $item["plink"];
+ }
- $orig_plink = $item["plink"];
+ $orig_plink = $item["plink"];
- $item["plink"] = original_url($item["plink"]);
+ $item["plink"] = Network::finalUrl($item["plink"]);
- $item["parent-uri"] = $item["uri"];
+ $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 ".$r[0]["id"], LOGGER_DEBUG);
- continue;
+ if (!$simulate) {
+ $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
+ $importer["uid"], $item["uri"], NETWORK_FEED, NETWORK_DFRN];
+ $previous = Item::selectFirst(['id'], $condition);
+ if (DBA::isResult($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;
+ $item["title"] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
- if ($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()', $entry)->item(0)->nodeValue;
+ if (empty($item["title"])) {
+ $item["title"] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
+ }
+ if (empty($item["title"])) {
+ $item["title"] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
+ }
+ $published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry);
- if ($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()', $entry)->item(0)->nodeValue;
+ if (empty($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()', $entry);
- if ($updated == "") {
- $updated = $published;
- }
- if ($published != "") {
- $item["created"] = $published;
- }
- if ($updated != "") {
- $item["edited"] = $updated;
- }
- $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue;
+ if (empty($updated)) {
+ $updated = $published;
+ }
+ if ($published != "") {
+ $item["created"] = $published;
+ }
+ if ($updated != "") {
+ $item["edited"] = $updated;
+ }
+ $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
- if ($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()', $entry)->item(0)->nodeValue;
+ if (empty($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 != "") {
- $item["author-name"] = $creator;
- }
+ if ($creator != "") {
+ $item["author-name"] = $creator;
+ }
- /// @TODO ?
- // Ausland
- //
-
- $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;
+ /// @TODO ?
+ // Ausland
+ //
+
+ $attachments = [];
+
+ $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
+ foreach ($enclosures AS $enclosure) {
+ $href = "";
+ $length = "";
+ $type = "";
+ $title = "";
+
+ foreach ($enclosure->attributes AS $attribute) {
+ if (in_array($attribute->name, ["url", "href"])) {
+ $href = $attribute->textContent;
+ } elseif ($attribute->name == "length") {
+ $length = $attribute->textContent;
+ } elseif ($attribute->name == "type") {
+ $type = $attribute->textContent;
+ }
+ }
+ if (!empty($item["attach"])) {
+ $item["attach"] .= ',';
+ } else {
+ $item["attach"] = '';
}
+
+ $attachments[] = ["link" => $href, "type" => $type, "length" => $length];
+
+ $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
}
- if (strlen($item["attach"])) {
- $item["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;
}
- $attachments[] = array("link" => $href, "type" => $type, "length" => $length);
+ $body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry));
- $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]';
- }
+ if (empty($body)) {
+ $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
+ }
+ if (empty($body)) {
+ $body = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
+ }
+ if (empty($body)) {
+ $body = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry));
+ }
- $tags = '';
- $categories = $xpath->query("category", $entry);
- foreach ($categories AS $category) {
- $hashtag = $category->nodeValue;
- if ($tags != '') {
- $tags .= ', ';
- }
+ // remove the content of the title if it is identically to the body
+ // This helps with auto generated titles e.g. from tumblr
+ if (self::titleIsBody($item["title"], $body)) {
+ $item["title"] = "";
+ }
+ $item["body"] = HTML::toBBCode($body, $basepath);
- $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]";
- $tags .= $taglink;
- }
+ if (($item["body"] == '') && ($item["title"] != '')) {
+ $item["body"] = $item["title"];
+ $item["title"] = '';
+ }
- $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
+ $preview = '';
+ if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
+ // Handle enclosures and treat them as preview picture
+ foreach ($attachments AS $attachment) {
+ if ($attachment["type"] == "image/jpeg") {
+ $preview = $attachment["link"];
+ }
+ }
- if ($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 a possible link to the item itself
+ $item["body"] = str_replace($item["plink"], '', $item["body"]);
+ $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
- // remove the content of the title if it is identically to the body
- // This helps with auto generated titles e.g. from tumblr
- if (title_is_body($item["title"], $body)) {
- $item["title"] = "";
- }
- $item["body"] = html2bbcode($body, $basepath);
+ // Replace the content when the title is longer than the body
+ $replace = (strlen($item["title"]) > strlen($item["body"]));
- if (($item["body"] == '') && ($item["title"] != '')) {
- $item["body"] = $item["title"];
- $item["title"] = '';
- }
+ // Replace it, when there is an image in the body
+ if (strstr($item["body"], '[/img]')) {
+ $replace = true;
+ }
- if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
- $preview = "";
+ // Replace it, when there is a link in the body
+ if (strstr($item["body"], '[/url]')) {
+ $replace = true;
+ }
- // Handle enclosures and treat them as preview picture
- foreach ($attachments AS $attachment) {
- if ($attachment["type"] == "image/jpeg") {
- $preview = $attachment["link"];
+ if ($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"], $preview, ($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 {
+ // @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'];
+ }
+ // 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]";
}
}
- // Remove a possible link to the item itself
- $item["body"] = str_replace($item["plink"], '', $item["body"]);
- $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]);
+ if (!$simulate) {
+ logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
- // Replace the content when the title is longer than the body
- $replace = (strlen($item["title"]) > strlen($item["body"]));
+ $notify = Item::isRemoteSelf($contact, $item);
- // Replace it, when there is an image in the body
- if (strstr($item["body"], '[/img]')) {
- $replace = true;
- }
-
- // Replace it, when there is a link in the body
- if (strstr($item["body"], '[/url]')) {
- $replace = true;
- }
+ // Distributed items should have a well formatted URI.
+ // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
+ if ($notify) {
+ $item['guid'] = Item::guidFromUri($orig_plink, $a->get_hostname());
+ unset($item['uri']);
+ unset($item['parent-uri']);
- if ($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"]);
+ // Set the delivery priority for "remote self" to "medium"
+ $notify = PRIORITY_MEDIUM;
}
- $item["body"] .= "\n".$item['tag'];
+
+ $id = Item::insert($item, false, $notify);
+
+ logger("Feed for contact ".$contact["url"]." stored under id ".$id);
+ } else {
+ $items[] = $item;
}
- if (!strstr($item["body"], '[url') && ($item['plink'] != '')) {
- $item["body"] .= "[hr][url]".$item['plink']."[/url]";
+ if ($simulate) {
+ break;
}
}
- if (!$simulate) {
- logger("Stored feed: ".print_r($item, true), LOGGER_DEBUG);
-
- $notify = item_is_remote_self($contact, $item);
+ if ($simulate) {
+ return ["header" => $author, "items" => $items];
+ }
+ }
- // Distributed items should have a well formatted URI.
- // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
- if ($notify) {
- $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname());
- unset($item['uri']);
- unset($item['parent-uri']);
- }
+ private static function titleIsBody($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);
- $id = item_store($item, false, $notify);
+ $body = strip_tags($body);
+ $body = trim($body);
+ $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
+ $body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
- logger("Feed for contact ".$contact["url"]." stored under id ".$id);
- } else {
- $items[] = $item;
- }
- if ($simulate) {
- break;
+ if (strlen($title) < strlen($body)) {
+ $body = substr($body, 0, strlen($title));
}
- }
- if ($simulate) {
- return array("header" => $author, "items" => $items);
+ if (($title != $body) && (substr($title, -3) == "...")) {
+ $pos = strrpos($title, "...");
+ if ($pos > 0) {
+ $title = substr($title, 0, $pos);
+ $body = substr($body, 0, $pos);
+ }
+ }
+ return ($title == $body);
}
}