From fc55a4b9c05123b806d30c228997397ca7897a11 Mon Sep 17 00:00:00 2001 From: Michael Date: Wed, 13 Dec 2017 07:02:52 +0000 Subject: [PATCH 1/4] The feed moved into the new location --- include/items.php | 3 ++- src/Network/Probe.php | 5 +++-- src/Protocol/Diaspora.php | 2 +- include/feed.php => src/Protocol/Feed.php | 0 4 files changed, 6 insertions(+), 4 deletions(-) rename include/feed.php => src/Protocol/Feed.php (100%) diff --git a/include/items.php b/include/items.php index 3d03f78ca4..b028fcc38e 100644 --- a/include/items.php +++ b/include/items.php @@ -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'; @@ -1550,7 +1551,7 @@ 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; } diff --git a/src/Network/Probe.php b/src/Network/Probe.php index 320bb43cc1..5d5a78214f 100644 --- a/src/Network/Probe.php +++ b/src/Network/Probe.php @@ -16,6 +16,7 @@ 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; @@ -1267,7 +1268,7 @@ 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 +1452,7 @@ 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 (!$probe) { diff --git a/src/Protocol/Diaspora.php b/src/Protocol/Diaspora.php index b56f59dad6..5c6d1a2c06 100644 --- a/src/Protocol/Diaspora.php +++ b/src/Protocol/Diaspora.php @@ -1,6 +1,6 @@ Date: Wed, 13 Dec 2017 07:03:42 +0000 Subject: [PATCH 2/4] The feed is now a class --- src/Protocol/Feed.php | 741 +++++++++++++++++++++--------------------- 1 file changed, 378 insertions(+), 363 deletions(-) diff --git a/src/Protocol/Feed.php b/src/Protocol/Feed.php index 49ae8df5c5..9f93287370 100644 --- a/src/Protocol/Feed.php +++ b/src/Protocol/Feed.php @@ -1,438 +1,453 @@ loadXML(trim($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 ".$importer["uid"], LOGGER_DEBUG); + } else { + logger("Test Atom/RSS feed", LOGGER_DEBUG); + } + if ($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); + + $author = array(); + + // Is it RDF? + if ($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"] == "") { - $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 $attributes) { + // Is it Atom? + if ($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"] = $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"] = $value; + if ($author["author-link"] == "") { + $author["author-link"] = $author["author-id"]; } - $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"] = $attributes->textContent; + if ($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; + } } } } - } - $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"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue; - // Is it RSS? - if ($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; + $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"] = $attributes->textContent; + } + } } } - } - if ($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; + + $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'); } - $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"] == "") { - $item["uri"] = $xpath->evaluate('guid/text()', $entry)->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 = $xpath->query('/rss/channel/item'); } - if ($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; + $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"]; } - $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"] == "") { - $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; + $header["contact-id"] = $contact["id"]; - if ($published == "") { - $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue; + if (!strlen($contact["notify"])) { + // 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 == "") { - $updated = $published; + if (!is_object($entries)) { + 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 == "") { - $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue; + $items = array(); + + $entrylist = array(); + + foreach ($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()', $entry)->item(0)->nodeValue; - } - if ($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"] = $creator; - } + if ($creator != "") { + $item["author-name"] = $creator; + } - /// @TODO ? - // Ausland - // + /// @TODO ? + // Ausland + // - $attachments = array(); + $attachments = array(); - $enclosures = $xpath->query("enclosure", $entry); - foreach ($enclosures AS $enclosure) { - $href = ""; - $length = ""; - $type = ""; - $title = ""; + $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 = $attributes->textContent; + foreach ($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 = $attachment["link"]; + if (strlen($item["attach"])) { + $item["attach"] .= ','; } + + $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 - $item["body"] = str_replace($item["plink"], '', $item["body"]); - $item["body"] = preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]); + $tags = ''; + $categories = $xpath->query("category", $entry); + foreach ($categories AS $category) { + $hashtag = $category->nodeValue; + if ($tags != '') { + $tags .= ', '; + } - // 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; + $taglink = "#[url=" . System::baseUrl() . "/search?tag=" . rawurlencode($hashtag) . "]" . $hashtag . "[/url]"; + $tags .= $taglink; } - // Replace it, when there is a link in the body - if (strstr($item["body"], '[/url]')) { - $replace = true; + $body = trim($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()', $entry)->item(0)->nodeValue); } - if ($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["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, $contact["ffi_keyword_blacklist"]); + + if (!empty($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]"; } - $item["body"] .= "\n".$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) { - break; + return array("header" => $author, "items" => $items); } } - - if ($simulate) { - return array("header" => $author, "items" => $items); - } } From 84b85e91980583601fe5e5a14ce6db3cfd2c1f3b Mon Sep 17 00:00:00 2001 From: Michael Date: Wed, 13 Dec 2017 07:06:36 +0000 Subject: [PATCH 3/4] Removed old include --- include/items.php | 1 - src/Network/Probe.php | 1 - 2 files changed, 2 deletions(-) diff --git a/include/items.php b/include/items.php index b028fcc38e..275052c1fd 100644 --- a/include/items.php +++ b/include/items.php @@ -27,7 +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'; diff --git a/src/Network/Probe.php b/src/Network/Probe.php index 5d5a78214f..1393237c8e 100644 --- a/src/Network/Probe.php +++ b/src/Network/Probe.php @@ -23,7 +23,6 @@ use dba; use DomXPath; use DOMDocument; -require_once 'include/feed.php'; require_once 'include/network.php'; /** From 785fcd5916f5e0abf2915297c9fa6e28ab5fe361 Mon Sep 17 00:00:00 2001 From: Michael Date: Wed, 13 Dec 2017 07:09:20 +0000 Subject: [PATCH 4/4] Missing "use" added --- src/Protocol/Feed.php | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/Protocol/Feed.php b/src/Protocol/Feed.php index 9f93287370..c33852a7d5 100644 --- a/src/Protocol/Feed.php +++ b/src/Protocol/Feed.php @@ -9,6 +9,8 @@ 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");