Standard stuff

This commit is contained in:
Michael 2017-03-15 06:00:22 +00:00
parent 35ec1c8f26
commit dc3fcf0487
1 changed files with 108 additions and 95 deletions

View File

@ -17,11 +17,11 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
$a = get_app(); $a = get_app();
if (!$simulate) if (!$simulate) {
logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG); logger("Import Atom/RSS feed '".$contact["name"]."' (Contact ".$contact["id"].") for user ".$importer["uid"], LOGGER_DEBUG);
else } else {
logger("Test Atom/RSS feed", LOGGER_DEBUG); logger("Test Atom/RSS feed", LOGGER_DEBUG);
}
if ($xml == "") { if ($xml == "") {
logger('XML is empty.', LOGGER_DEBUG); logger('XML is empty.', LOGGER_DEBUG);
return; return;
@ -45,63 +45,69 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
$author["author-link"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:link/text()')->item(0)->nodeValue; $author["author-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; $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; $author["author-name"] = $xpath->evaluate('/rdf:RDF/rss:channel/rss:description/text()')->item(0)->nodeValue;
}
$entries = $xpath->query('/rdf:RDF/rss:item'); $entries = $xpath->query('/rdf:RDF/rss:item');
} }
// Is it Atom? // Is it Atom?
if ($xpath->query('/atom:feed')->length > 0) { if ($xpath->query('/atom:feed')->length > 0) {
$alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes; $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
if (is_object($alternate)) if (is_object($alternate)) {
foreach($alternate AS $attributes) foreach($alternate AS $attributes) {
if ($attributes->name == "href") if ($attributes->name == "href") {
$author["author-link"] = $attributes->textContent; $author["author-link"] = $attributes->textContent;
}
if ($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"] = $attributes->textContent;
} }
if ($author["author-link"] == "") if ($author["author-link"] == "") {
$author["author-link"] = $xpath->evaluate('/atom:feed/atom:id/text()')->item(0)->nodeValue; $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"] = $attributes->textContent;
}
}
}
}
if ($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-avatar"] = $xpath->evaluate('/atom:feed/atom:logo/text()')->item(0)->nodeValue;
$author["author-name"] = $xpath->evaluate('/atom:feed/atom:title/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; $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; $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; $value = $xpath->evaluate('atom:author/poco:displayName/text()')->item(0)->nodeValue;
if ($value != "") if ($value != "") {
$author["author-name"] = $value; $author["author-name"] = $value;
}
if ($simulate) { if ($simulate) {
$author["author-id"] = $xpath->evaluate('/atom:feed/atom:author/atom:uri/text()')->item(0)->nodeValue; $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; $value = $xpath->evaluate('atom:author/poco:preferredUsername/text()')->item(0)->nodeValue;
if ($value != "") if ($value != "") {
$author["author-nick"] = $value; $author["author-nick"] = $value;
}
$value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue; $value = $xpath->evaluate('atom:author/poco:address/poco:formatted/text()', $context)->item(0)->nodeValue;
if ($value != "") if ($value != "") {
$author["author-location"] = $value; $author["author-location"] = $value;
}
$value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue; $value = $xpath->evaluate('atom:author/poco:note/text()')->item(0)->nodeValue;
if ($value != "") if ($value != "") {
$author["author-about"] = $value; $author["author-about"] = $value;
}
} }
$author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue; $author["edited"] = $author["created"] = $xpath->query('/atom:feed/atom:updated/text()')->item(0)->nodeValue;
@ -118,12 +124,12 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
$author["author-name"] = $xpath->evaluate('/rss/channel/title/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; $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; $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["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["edited"] = $author["created"] = $xpath->query('/rss/channel/pubDate/text()')->item(0)->nodeValue;
$author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue; $author["app"] = $xpath->evaluate('/rss/channel/generator/text()')->item(0)->nodeValue;
@ -134,9 +140,9 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
if (!$simulate) { if (!$simulate) {
$author["author-link"] = $contact["url"]; $author["author-link"] = $contact["url"];
if ($author["author-name"] == "") if ($author["author-name"] == "") {
$author["author-name"] = $contact["name"]; $author["author-name"] = $contact["name"];
}
$author["author-avatar"] = $contact["thumb"]; $author["author-avatar"] = $contact["thumb"];
$author["owner-link"] = $contact["url"]; $author["owner-link"] = $contact["url"];
@ -171,37 +177,39 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
$entrylist = array(); $entrylist = array();
foreach ($entries AS $entry) foreach ($entries AS $entry) {
$entrylist[] = $entry; $entrylist[] = $entry;
}
foreach (array_reverse($entrylist) AS $entry) { foreach (array_reverse($entrylist) AS $entry) {
$item = array_merge($header, $author); $item = array_merge($header, $author);
$alternate = $xpath->query("atom:link[@rel='alternate']", $entry)->item(0)->attributes; $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; $alternate = $xpath->query("atom:link", $entry)->item(0)->attributes;
}
if (is_object($alternate)) if (is_object($alternate)) {
foreach($alternate AS $attributes) foreach($alternate AS $attributes) {
if ($attributes->name == "href") if ($attributes->name == "href") {
$item["plink"] = $attributes->textContent; $item["plink"] = $attributes->textContent;
}
if ($item["plink"] == "") }
}
if ($item["plink"] == "") {
$item["plink"] = $xpath->evaluate('link/text()', $entry)->item(0)->nodeValue; $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"] = $xpath->evaluate('rss:link/text()', $entry)->item(0)->nodeValue;
}
$item["plink"] = original_url($item["plink"]); $item["plink"] = original_url($item["plink"]);
$item["uri"] = $xpath->evaluate('atom:id/text()', $entry)->item(0)->nodeValue; $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; $item["uri"] = $xpath->evaluate('guid/text()', $entry)->item(0)->nodeValue;
}
if ($item["uri"] == "") if ($item["uri"] == "") {
$item["uri"] = $item["plink"]; $item["uri"] = $item["plink"];
}
$item["parent-uri"] = $item["uri"]; $item["parent-uri"] = $item["uri"];
if (!$simulate) { if (!$simulate) {
@ -215,50 +223,50 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
$item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue; $item["title"] = $xpath->evaluate('atom:title/text()', $entry)->item(0)->nodeValue;
if ($item["title"] == "") if ($item["title"] == "") {
$item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue; $item["title"] = $xpath->evaluate('title/text()', $entry)->item(0)->nodeValue;
}
if ($item["title"] == "") if ($item["title"] == "") {
$item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue; $item["title"] = $xpath->evaluate('rss:title/text()', $entry)->item(0)->nodeValue;
}
$published = $xpath->query('atom:published/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; $published = $xpath->query('pubDate/text()', $entry)->item(0)->nodeValue;
}
if ($published == "") if ($published == "") {
$published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue; $published = $xpath->query('dc:date/text()', $entry)->item(0)->nodeValue;
}
$updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue; $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
if ($updated == "") if ($updated == "") {
$updated = $published; $updated = $published;
}
if ($published != "") if ($published != "") {
$item["created"] = $published; $item["created"] = $published;
}
if ($updated != "") if ($updated != "") {
$item["edited"] = $updated; $item["edited"] = $updated;
}
$creator = $xpath->query('author/text()', $entry)->item(0)->nodeValue; $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; $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; $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
}
if ($creator != "") if ($creator != "") {
$item["author-name"] = $creator; $item["author-name"] = $creator;
}
if ($pubDate != "") if ($pubDate != "") {
$item["edited"] = $item["created"] = $pubDate; $item["edited"] = $item["created"] = $pubDate;
}
$creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue; $creator = $xpath->query('dc:creator/text()', $entry)->item(0)->nodeValue;
if ($creator != "") if ($creator != "") {
$item["author-name"] = $creator; $item["author-name"] = $creator;
}
/// @TODO ? /// @TODO ?
// <category>Ausland</category> // <category>Ausland</category>
// <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/> // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
@ -273,12 +281,13 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
$title = ""; $title = "";
foreach($enclosure->attributes AS $attributes) { foreach($enclosure->attributes AS $attributes) {
if ($attributes->name == "url") if ($attributes->name == "url") {
$href = $attributes->textContent; $href = $attributes->textContent;
elseif ($attributes->name == "length") } elseif ($attributes->name == "length") {
$length = $attributes->textContent; $length = $attributes->textContent;
elseif ($attributes->name == "type") } elseif ($attributes->name == "type") {
$type = $attributes->textContent; $type = $attributes->textContent;
}
} }
if(strlen($item["attach"])) if(strlen($item["attach"]))
$item["attach"] .= ','; $item["attach"] .= ',';
@ -292,9 +301,11 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
$preview = ""; $preview = "";
// Handle enclosures and treat them as preview picture // Handle enclosures and treat them as preview picture
foreach ($attachments AS $attachment) foreach ($attachments AS $attachment) {
if ($attachment["type"] == "image/jpeg") if ($attachment["type"] == "image/jpeg") {
$preview = $attachment["link"]; $preview = $attachment["link"];
}
}
$item["body"] = $item["title"].add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]); $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"]); $item["tag"] = add_page_keywords($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
@ -304,20 +315,20 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
} else { } else {
$body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue); $body = trim($xpath->evaluate('atom:content/text()', $entry)->item(0)->nodeValue);
if ($body == "") if ($body == "") {
$body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue); $body = trim($xpath->evaluate('content:encoded/text()', $entry)->item(0)->nodeValue);
}
if ($body == "") if ($body == "") {
$body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue); $body = trim($xpath->evaluate('description/text()', $entry)->item(0)->nodeValue);
}
if ($body == "") if ($body == "") {
$body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue); $body = trim($xpath->evaluate('atom:summary/text()', $entry)->item(0)->nodeValue);
}
// remove the content of the title if it is identically to the body // remove the content of the title if it is identically to the body
// This helps with auto generated titles e.g. from tumblr // This helps with auto generated titles e.g. from tumblr
if (title_is_body($item["title"], $body)) if (title_is_body($item["title"], $body)) {
$item["title"] = ""; $item["title"] = "";
}
$item["body"] = html2bbcode($body); $item["body"] = html2bbcode($body);
} }
@ -336,14 +347,16 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
$id = item_store($item, false, $notify); $id = item_store($item, false, $notify);
logger("Feed for contact ".$contact["url"]." stored under id ".$id); logger("Feed for contact ".$contact["url"]." stored under id ".$id);
} else } else {
$items[] = $item; $items[] = $item;
}
if ($simulate) if ($simulate) {
break; break;
}
} }
if ($simulate) if ($simulate) {
return array("header" => $author, "items" => $items); return array("header" => $author, "items" => $items);
}
} }
?> ?>