X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;ds=sidebyside;f=include%2Ffeed.php;h=fb97c14ba126aa1875d1202df05a5caa8aad64e7;hb=57f114d7e1955eb22167b12a9d2453ab096531cc;hp=579ff7caae3652cce6c9fcb714720886bd32334f;hpb=45ebeba643a6d600e8778475f70f827f32bf3146;p=friendica.git diff --git a/include/feed.php b/include/feed.php index 579ff7caae..fb97c14ba1 100644 --- a/include/feed.php +++ b/include/feed.php @@ -17,18 +17,26 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) { $a = get_app(); - if (!$simulate) + if (!$simulate) { logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG); - else + } else { logger("Test Atom/RSS feed", LOGGER_DEBUG); - + } if ($xml == "") { logger('XML is empty.', LOGGER_DEBUG); return; } + if (!empty($contact['poll'])) { + $basepath = $contact['poll']; + } elseif (!empty($contact['url'])) { + $basepath = $contact['url']; + } else { + $basepath = ''; + } + $doc = new DOMDocument(); - @$doc->loadXML($xml); + @$doc->loadXML(trim($xml)); $xpath = new DomXPath($doc); $xpath->registerNamespace('atom', NAMESPACE_ATOM1); $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/"); @@ -45,63 +53,77 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) { $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"] == "") + if ($author["author-name"] == "") { $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue; - + } $entries = $xpath->query('/rdf:RDF/rss:item'); } // Is it Atom? if ($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") + if (is_object($alternate)) { + foreach ($alternate AS $attributes) { + if ($attributes->name == "href") { $author["author-link"] = $attributes->textContent; + } + } + } - if ($author["author-link"] == "") + 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") + if (is_object($self)) { + foreach ($self AS $attributes) { + if ($attributes->name == "href") { $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-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"] == "") + if ($author["author-name"] == "") { $author["author-name"] = $xpath->evaluate('/atom:feed/atom:subtitle/text()')->item(0)->nodeValue; - - if ($author["author-name"] == "") + } + 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 != "") + 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 != "") + if ($value != "") { $author["author-nick"] = $value; - + } $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue; - if ($value != "") + if ($value != "") { $author["author-location"] = $value; - + } $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue; - if ($value != "") + 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; + } + } + } } $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue; @@ -118,12 +140,12 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) { $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"] == "") + if ($author["author-name"] == "") { $author["author-name"] = $xpath->evaluate('/rss/channel/copyright/text()')->item(0)->nodeValue; - - if ($author["author-name"] == "") + } + 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"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue; @@ -134,9 +156,9 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) { if (!$simulate) { $author["author-link"] = $contact["url"]; - if ($author["author-name"] == "") + if ($author["author-name"] == "") { $author["author-name"] = $contact["name"]; - + } $author["author-avatar"] = $contact["thumb"]; $author["owner-link"] = $contact["url"]; @@ -157,7 +179,7 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) { $header["contact-id"] = $contact["id"]; - if(!strlen($contact["notify"])) { + if (!strlen($contact["notify"])) { // one way feed - no remote comment ability $header["last-child"] = 0; } @@ -171,92 +193,99 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) { $entrylist = array(); - foreach ($entries AS $entry) + foreach ($entries AS $entry) { $entrylist[] = $entry; - + } foreach (array_reverse($entrylist) AS $entry) { $item = array_merge($header, $author); - $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; - $alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes; - if (!is_object($alternate)) + 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") + } + if (is_object($alternate)) { + foreach ($alternate AS $attributes) { + if ($attributes->name == "href") { $item["plink"] = $attributes->textContent; - - if ($item["plink"] == "") + } + } + } + if ($item["plink"] == "") { $item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue; - - if ($item["plink"] == "") + } + if ($item["plink"] == "") { $item["plink"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue; - - $item["plink"] = original_url($item["plink"]); + } $item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue; - if ($item["uri"] == "") + if ($item["uri"] == "") { $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue; - - if ($item["uri"] == "") + } + 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; + } + } + + $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 == "") + if ($published == "") { $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue; - - if ($published == "") + } + 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 ($updated == "") { $updated = $published; - - if ($published != "") + } + if ($published != "") { $item["created"] = $published; - - if ($updated != "") + } + if ($updated != "") { $item["edited"] = $updated; - + } $creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue; - if ($creator == "") + if ($creator == "") { $creator = $xpath->query('atom:author/atom:name/text()', $entry)->item(0)->nodeValue; - - if ($creator == "") + } + if ($creator == "") { $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue; - - if ($creator != "") + } + if ($creator != "") { $item["author-name"] = $creator; - - if ($pubDate != "") + } + if ($pubDate != "") { $item["edited"] = $item["created"] = $pubDate; - + } $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue; - if ($creator != "") + if ($creator != "") { $item["author-name"] = $creator; - - if (!$simulate) { - $r = q("SELECT `id` FROM `item` WHERE `uid` = %d AND `uri` = '%s' AND `network` IN ('%s', '%s')", - intval($importer["uid"]), dbesc($item["uri"]), dbesc(NETWORK_FEED), dbesc(NETWORK_DFRN)); - if ($r) { - logger("Item with uri ".$item["uri"]." for user ".$importer["uid"]." already existed under id ".$r[0]["id"], LOGGER_DEBUG); - continue; - } } /// @TODO ? @@ -272,53 +301,88 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) { $type = ""; $title = ""; - foreach($enclosure->attributes AS $attributes) { - if ($attributes->name == "url") + foreach ($enclosure->attributes AS $attributes) { + if ($attributes->name == "url") { $href = $attributes->textContent; - elseif ($attributes->name == "length") + } elseif ($attributes->name == "length") { $length = $attributes->textContent; - elseif ($attributes->name == "type") + } elseif ($attributes->name == "type") { $type = $attributes->textContent; + } } - if(strlen($item["attach"])) + if (strlen($item["attach"])) { $item["attach"] .= ','; + } $attachments[] = array("link" => $href, "type" => $type, "length" => $length); $item["attach"] .= '[attach]href="'.$href.'" length="'.$length.'" type="'.$type.'"[/attach]'; } + $body = trim($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"]; + $item["title"] = ''; + } + if ($contact["fetch_further_information"]) { $preview = ""; // Handle enclosures and treat them as preview picture - foreach ($attachments AS $attachment) - if ($attachment["type"] == "image/jpeg") + foreach ($attachments AS $attachment) { + if ($attachment["type"] == "image/jpeg") { $preview = $attachment["link"]; + } + } - $item["body"] = $item["title"].add_page_info($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["title"] = ""; - $item["object-type"] = ACTIVITY_OBJ_BOOKMARK; - unset($item["attach"]); - } else { - $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue); - - if ($body == "") - $body = trim($xpath->evaluate('content:encoded/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', '', $item["body"]); - if ($body == "") - $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue); + // Replace the content when the title is longer than the body + $replace = (strlen($item["title"]) > strlen($item["body"])); - if ($body == "") - $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue); + // Replace it, when there is an image in the body + if (strstr($item["body"], '[/img]')) { + $replace = true; + } - // remove the content of the title if it is identically to the body - // This helps with auto generated titles e.g. from tumblr - if (title_is_body($item["title"], $body)) - $item["title"] = ""; + // Replace it, when there is a link in the body + if (strstr($item["body"], '[/url]')) { + $replace = true; + } - $item["body"] = html2bbcode($body); + 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 (!strstr($item["body"], '[url') && ($item['plink'] != '')) { + $item["body"] .= "[hr][url]".$item['plink']."[/url]"; + } } if (!$simulate) { @@ -329,6 +393,7 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) { // Distributed items should have a well formatted URI. // Additionally we have to avoid conflicts with identical URI between imported feeds and these items. if ($notify) { + $item['guid'] = uri_to_guid($orig_plink, $a->get_hostname()); unset($item['uri']); unset($item['parent-uri']); } @@ -336,14 +401,15 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) { $id = item_store($item, false, $notify); logger("Feed for contact ".$contact["url"]." stored under id ".$id); - } else + } else { $items[] = $item; - - if ($simulate) + } + if ($simulate) { break; + } } - if ($simulate) + if ($simulate) { return array("header" => $author, "items" => $items); + } } -?>