refactor: FeedExpander (#3740)

* refactor: FeedExpander
This commit is contained in:
Dag 2023-10-12 22:14:04 +02:00 committed by GitHub
parent 6634291c67
commit 9bda9e246a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
9 changed files with 315 additions and 504 deletions

View file

@ -61,6 +61,10 @@ class CssSelectorFeedExpanderBridge extends CssSelectorBridge
$discard_thumbnail = $this->getInput('discard_thumbnail'); $discard_thumbnail = $this->getInput('discard_thumbnail');
$limit = $this->getInput('limit'); $limit = $this->getInput('limit');
//$xmlString = getContents($url);
//$feed = (new FeedParser())->parseFeed($xmlString);
//$items = $feed['items'];
$feed_expander = new CssSelectorFeedExpanderBridgeInternal(); $feed_expander = new CssSelectorFeedExpanderBridgeInternal();
$items = $feed_expander->collectExpandableDatas($url)->getItems(); $items = $feed_expander->collectExpandableDatas($url)->getItems();

View file

@ -46,21 +46,6 @@ class FeedExpanderExampleBridge extends FeedExpander
protected function parseItem($newsItem) protected function parseItem($newsItem)
{ {
switch ($this->getInput('version')) { return (array) $newsItem;
case 'rss_0_9_1':
return $this->parseRss091Item($newsItem);
break;
case 'rss_1_0':
return $this->parseRss1Item($newsItem);
break;
case 'rss_2_0':
return $this->parseRss2Item($newsItem);
break;
case 'atom_1_0':
return $this->parseATOMItem($newsItem);
break;
default:
returnClientError('Unknown version ' . $this->getInput('version') . '!');
}
} }
} }

View file

@ -82,14 +82,14 @@ class MastodonBridge extends BridgeAbstract
} }
$items = $content['orderedItems'] ?? $content['items']; $items = $content['orderedItems'] ?? $content['items'];
foreach ($items as $status) { foreach ($items as $status) {
$item = $this->parseItem($status); $item = $this->parseStatus($status);
if ($item) { if ($item) {
$this->items[] = $item; $this->items[] = $item;
} }
} }
} }
protected function parseItem($content) protected function parseStatus($content)
{ {
$item = []; $item = [];
switch ($content['type']) { switch ($content['type']) {

View file

@ -67,7 +67,7 @@ class NyaaTorrentsBridge extends FeedExpander
protected function parseItem($newItem) protected function parseItem($newItem)
{ {
$item = parent::parseRss2Item($newItem); $item = parent::parseItem($newItem);
$item['id'] = str_replace(['https://nyaa.si/download/', '.torrent'], '', $item['uri']); $item['id'] = str_replace(['https://nyaa.si/download/', '.torrent'], '', $item['uri']);
$nyaaFields = (array)($newItem->children('nyaa', true)); $nyaaFields = (array)($newItem->children('nyaa', true));

View file

@ -14,7 +14,7 @@ class RaceDepartmentBridge extends FeedExpander
protected function parseItem($feedItem) protected function parseItem($feedItem)
{ {
$item = parent::parseRss2Item($feedItem); $item = parent::parseItem($feedItem);
//fetch page //fetch page
$articlePage = getSimpleHTMLDOMCached($feedItem->link); $articlePage = getSimpleHTMLDOMCached($feedItem->link);

View file

@ -6,6 +6,56 @@ if (version_compare(\PHP_VERSION, '7.4.0') === -1) {
require_once __DIR__ . '/lib/bootstrap.php'; require_once __DIR__ . '/lib/bootstrap.php';
Configuration::verifyInstallation();
$customConfig = [];
if (file_exists(__DIR__ . '/config.ini.php')) {
$customConfig = parse_ini_file(__DIR__ . '/config.ini.php', true, INI_SCANNER_TYPED);
}
Configuration::loadConfiguration($customConfig, getenv());
// Consider: ini_set('error_reporting', E_ALL & ~E_DEPRECATED);
date_default_timezone_set(Configuration::getConfig('system', 'timezone'));
$rssBridge = new RssBridge(); $rssBridge = new RssBridge();
set_exception_handler(function (\Throwable $e) {
http_response_code(500);
print render(__DIR__ . '/templates/exception.html.php', ['e' => $e]);
RssBridge::getLogger()->error('Uncaught Exception', ['e' => $e]);
exit(1);
});
set_error_handler(function ($code, $message, $file, $line) {
if ((error_reporting() & $code) === 0) {
return false;
}
// In the future, uncomment this:
//throw new \ErrorException($message, 0, $code, $file, $line);
$text = sprintf(
'%s at %s line %s',
sanitize_root($message),
sanitize_root($file),
$line
);
RssBridge::getLogger()->warning($text);
});
// There might be some fatal errors which are not caught by set_error_handler() or \Throwable.
register_shutdown_function(function () {
$error = error_get_last();
if ($error) {
$message = sprintf(
'(shutdown) %s: %s in %s line %s',
$error['type'],
sanitize_root($error['message']),
sanitize_root($error['file']),
$error['line']
);
RssBridge::getLogger()->error($message);
if (Debug::isEnabled()) {
print sprintf("<pre>%s</pre>\n", e($message));
}
}
});
$rssBridge->main($argv ?? []); $rssBridge->main($argv ?? []);

View file

@ -1,95 +1,37 @@
<?php <?php
/** /**
* An abstract class for bridges that need to transform existing RSS or Atom * Expands an existing feed
* feeds.
*
* This class extends {@see BridgeAbstract} with functions to extract contents
* from existing RSS or Atom feeds. Bridges that need to transform existing feeds
* should inherit from this class instead of {@see BridgeAbstract}.
*
* Bridges that extend this class don't need to concern themselves with getting
* contents from existing feeds, but can focus on adding additional contents
* (i.e. by downloading additional data), filtering or just transforming a feed
* into another format.
*
* @link http://www.rssboard.org/rss-0-9-1 RSS 0.91 Specification
* @link http://web.resource.org/rss/1.0/spec RDF Site Summary (RSS) 1.0
* @link http://www.rssboard.org/rss-specification RSS 2.0 Specification
* @link https://tools.ietf.org/html/rfc4287 The Atom Syndication Format
*
* @todo The parsing functions should all be private. This class is complicated
* enough without having to consider children overriding functions.
*/ */
abstract class FeedExpander extends BridgeAbstract abstract class FeedExpander extends BridgeAbstract
{ {
/** Indicates an RSS 1.0 feed */
const FEED_TYPE_RSS_1_0 = 'RSS_1_0'; const FEED_TYPE_RSS_1_0 = 'RSS_1_0';
/** Indicates an RSS 2.0 feed */
const FEED_TYPE_RSS_2_0 = 'RSS_2_0'; const FEED_TYPE_RSS_2_0 = 'RSS_2_0';
/** Indicates an Atom 1.0 feed */
const FEED_TYPE_ATOM_1_0 = 'ATOM_1_0'; const FEED_TYPE_ATOM_1_0 = 'ATOM_1_0';
/** private string $feedType;
* Holds the title of the current feed private FeedParser $feedParser;
* private array $parsedFeed;
* @var string
*/
private $title;
/** public function __construct(CacheInterface $cache, Logger $logger)
* Holds the URI of the feed
*
* @var string
*/
private $uri;
/**
* Holds the icon of the feed
*
*/
private $icon;
/**
* Holds the feed type during internal operations.
*
* @var string
*/
private $feedType;
/**
* Collects data from an existing feed.
*
* Children should call this function in {@see BridgeAbstract::collectData()}
* to extract a feed.
*
* @param string $url URL to the feed.
* @param int $maxItems Maximum number of items to collect from the feed
* (`-1`: no limit).
* @return self
*/
public function collectExpandableDatas($url, $maxItems = -1)
{ {
if (empty($url)) { parent::__construct($cache, $logger);
$this->feedParser = new FeedParser();
}
public function collectExpandableDatas(string $url, $maxItems = -1)
{
if (!$url) {
throw new \Exception('There is no $url for this RSS expander'); throw new \Exception('There is no $url for this RSS expander');
} }
if ($maxItems === -1) {
Debug::log(sprintf('Loading from %s', $url)); $maxItems = 999;
}
/* Notice we do not use cache here on purpose: $accept = [MrssFormat::MIME_TYPE, AtomFormat::MIME_TYPE, '*/*'];
* we want a fresh view of the RSS stream each time $httpHeaders = ['Accept: ' . implode(', ', $accept)];
*/ // Notice we do not use cache here on purpose. We want a fresh view of the RSS stream each time
$xmlString = getContents($url, $httpHeaders);
$mimeTypes = [ if ($xmlString === '') {
MrssFormat::MIME_TYPE,
AtomFormat::MIME_TYPE,
'*/*',
];
$httpHeaders = ['Accept: ' . implode(', ', $mimeTypes)];
$xml = getContents($url, $httpHeaders);
if ($xml === '') {
throw new \Exception(sprintf('Unable to parse xml from `%s` because we got the empty string', $url), 10); throw new \Exception(sprintf('Unable to parse xml from `%s` because we got the empty string', $url), 10);
} }
// Maybe move this call earlier up the stack frames // Maybe move this call earlier up the stack frames
@ -97,8 +39,8 @@ abstract class FeedExpander extends BridgeAbstract
libxml_use_internal_errors(true); libxml_use_internal_errors(true);
// Consider replacing libxml with https://www.php.net/domdocument // Consider replacing libxml with https://www.php.net/domdocument
// Intentionally not using the silencing operator (@) because it has no effect here // Intentionally not using the silencing operator (@) because it has no effect here
$rssContent = simplexml_load_string(trim($xml)); $xml = simplexml_load_string(trim($xmlString));
if ($rssContent === false) { if ($xml === false) {
$xmlErrors = libxml_get_errors(); $xmlErrors = libxml_get_errors();
foreach ($xmlErrors as $xmlError) { foreach ($xmlErrors as $xmlError) {
Debug::log(trim($xmlError->message)); Debug::log(trim($xmlError->message));
@ -112,384 +54,62 @@ abstract class FeedExpander extends BridgeAbstract
// Restore previous behaviour in case other code relies on it being off // Restore previous behaviour in case other code relies on it being off
libxml_use_internal_errors(false); libxml_use_internal_errors(false);
// Commented out because it's spammy // Currently only feed metadata (not items) are plucked out
// Debug::log(sprintf("RSS content is ===========\n%s===========", var_export($rssContent, true))); $this->parsedFeed = $this->feedParser->parseFeed($xmlString);
switch (true) { if (isset($xml->item[0])) {
case isset($rssContent->item[0]): $this->feedType = self::FEED_TYPE_RSS_1_0;
Debug::log('Detected RSS 1.0 format'); $items = $xml->item;
$this->feedType = self::FEED_TYPE_RSS_1_0; } elseif (isset($xml->channel[0])) {
$this->collectRss1($rssContent, $maxItems); $this->feedType = self::FEED_TYPE_RSS_2_0;
break; $items = $xml->channel[0]->item;
case isset($rssContent->channel[0]): } elseif (isset($xml->entry[0])) {
Debug::log('Detected RSS 0.9x or 2.0 format'); $this->feedType = self::FEED_TYPE_ATOM_1_0;
$this->feedType = self::FEED_TYPE_RSS_2_0; $items = $xml->entry;
$this->collectRss2($rssContent, $maxItems); } else {
break; throw new \Exception(sprintf('Unable to detect feed format from `%s`', $url));
case isset($rssContent->entry[0]): }
Debug::log('Detected ATOM format'); foreach ($items as $item) {
$this->feedType = self::FEED_TYPE_ATOM_1_0; $parsedItem = $this->parseItem($item);
$this->collectAtom1($rssContent, $maxItems); if ($parsedItem) {
break; $this->items[] = $parsedItem;
default: }
Debug::log(sprintf('Unable to detect feed format from `%s`', $url)); if (count($this->items) >= $maxItems) {
throw new \Exception(sprintf('Unable to detect feed format from `%s`', $url)); break;
}
} }
return $this; return $this;
} }
/** /**
* Collect data from an RSS 1.0 compatible feed * @param \SimpleXMLElement $item The feed item to be parsed
*
* @link http://web.resource.org/rss/1.0/spec RDF Site Summary (RSS) 1.0
*
* @param string $rssContent The RSS content
* @param int $maxItems Maximum number of items to collect from the feed
* (`-1`: no limit).
* @return void
*
* @todo Instead of passing $maxItems to all functions, just add all items
* and remove excessive items later.
*/
protected function collectRss1($rssContent, $maxItems)
{
$this->loadRss2Data($rssContent->channel[0]);
foreach ($rssContent->item as $item) {
$tmp_item = $this->parseItem($item);
if (!empty($tmp_item)) {
$this->items[] = $tmp_item;
}
if ($maxItems !== -1 && count($this->items) >= $maxItems) {
break;
}
}
}
/**
* Collect data from a RSS 2.0 compatible feed
*
* @link http://www.rssboard.org/rss-specification RSS 2.0 Specification
*
* @param int $maxItems Maximum number of items to collect from the feed (`-1`: no limit).
* @return void
*
* @todo Instead of passing $maxItems to all functions, just add all items and remove excessive items later.
*/
protected function collectRss2(\SimpleXMLElement $rssContent, $maxItems)
{
$rssContent = $rssContent->channel[0];
$this->loadRss2Data($rssContent);
foreach ($rssContent->item as $item) {
$tmp_item = $this->parseItem($item);
if (!empty($tmp_item)) {
$this->items[] = $tmp_item;
}
if ($maxItems !== -1 && count($this->items) >= $maxItems) {
break;
}
}
}
/**
* Collect data from a Atom 1.0 compatible feed
*
* @link https://tools.ietf.org/html/rfc4287 The Atom Syndication Format
*
* @param object $content The Atom content
* @param int $maxItems Maximum number of items to collect from the feed
* (`-1`: no limit).
* @return void
*
* @todo Instead of passing $maxItems to all functions, just add all items
* and remove excessive items later.
*/
protected function collectAtom1($content, $maxItems)
{
$this->loadAtomData($content);
foreach ($content->entry as $item) {
$tmp_item = $this->parseItem($item);
if (!empty($tmp_item)) {
$this->items[] = $tmp_item;
}
if ($maxItems !== -1 && count($this->items) >= $maxItems) {
break;
}
}
}
/**
* Load RSS 2.0 feed data into RSS-Bridge
*
* @param object $rssContent The RSS content
* @return void
*
* @todo set title, link, description, language, and so on
*/
protected function loadRss2Data($rssContent)
{
$this->title = trim((string)$rssContent->title);
$this->uri = trim((string)$rssContent->link);
if (!empty($rssContent->image)) {
$this->icon = trim((string)$rssContent->image->url);
}
}
/**
* Load Atom feed data into RSS-Bridge
*
* @param object $content The Atom content
* @return void
*/
protected function loadAtomData($content)
{
$this->title = (string)$content->title;
// Find best link (only one, or first of 'alternate')
if (!isset($content->link)) {
$this->uri = '';
} elseif (count($content->link) === 1) {
$this->uri = (string)$content->link[0]['href'];
} else {
$this->uri = '';
foreach ($content->link as $link) {
if (strtolower($link['rel']) === 'alternate') {
$this->uri = (string)$link['href'];
break;
}
}
}
if (!empty($content->icon)) {
$this->icon = (string)$content->icon;
} elseif (!empty($content->logo)) {
$this->icon = (string)$content->logo;
}
}
/**
* Parse the contents of a single Atom feed item into a RSS-Bridge item for
* further transformation.
*
* @param object $feedItem A single feed item
* @return object The RSS-Bridge item
*
* @todo To reduce confusion, the RSS-Bridge item should maybe have a class
* of its own?
*/
protected function parseATOMItem($feedItem)
{
// Some ATOM entries also contain RSS 2.0 fields
$item = $this->parseRss2Item($feedItem);
if (isset($feedItem->id)) {
$item['uri'] = (string)$feedItem->id;
}
if (isset($feedItem->title)) {
$item['title'] = html_entity_decode((string)$feedItem->title);
}
if (isset($feedItem->updated)) {
$item['timestamp'] = strtotime((string)$feedItem->updated);
}
if (isset($feedItem->author)) {
$item['author'] = (string)$feedItem->author->name;
}
if (isset($feedItem->content)) {
$contentChildren = $feedItem->content->children();
if (count($contentChildren) > 0) {
$content = '';
foreach ($contentChildren as $contentChild) {
$content .= $contentChild->asXML();
}
$item['content'] = $content;
} else {
$item['content'] = (string)$feedItem->content;
}
}
//When "link" field is present, URL is more reliable than "id" field
if (count($feedItem->link) === 1) {
$item['uri'] = (string)$feedItem->link[0]['href'];
} else {
foreach ($feedItem->link as $link) {
if (strtolower($link['rel']) === 'alternate') {
$item['uri'] = (string)$link['href'];
}
if (strtolower($link['rel']) === 'enclosure') {
$item['enclosures'][] = (string)$link['href'];
}
}
}
return $item;
}
/**
* Parse the contents of a single RSS 0.91 feed item into a RSS-Bridge item
* for further transformation.
*
* @param object $feedItem A single feed item
* @return object The RSS-Bridge item
*
* @todo To reduce confusion, the RSS-Bridge item should maybe have a class
* of its own?
*/
protected function parseRss091Item($feedItem)
{
$item = [];
if (isset($feedItem->link)) {
$item['uri'] = (string)$feedItem->link;
}
if (isset($feedItem->title)) {
$item['title'] = html_entity_decode((string)$feedItem->title);
}
// rss 0.91 doesn't support timestamps
// rss 0.91 doesn't support authors
// rss 0.91 doesn't support enclosures
if (isset($feedItem->description)) {
$item['content'] = (string)$feedItem->description;
}
return $item;
}
/**
* Parse the contents of a single RSS 1.0 feed item into a RSS-Bridge item
* for further transformation.
*
* @param object $feedItem A single feed item
* @return object The RSS-Bridge item
*
* @todo To reduce confusion, the RSS-Bridge item should maybe have a class
* of its own?
*/
protected function parseRss1Item($feedItem)
{
// 1.0 adds optional elements around the 0.91 standard
$item = $this->parseRss091Item($feedItem);
$namespaces = $feedItem->getNamespaces(true);
if (isset($namespaces['dc'])) {
$dc = $feedItem->children($namespaces['dc']);
if (isset($dc->date)) {
$item['timestamp'] = strtotime((string)$dc->date);
}
if (isset($dc->creator)) {
$item['author'] = (string)$dc->creator;
}
}
return $item;
}
/**
* Parse the contents of a single RSS 2.0 feed item into a RSS-Bridge item
* for further transformation.
*
* @param object $feedItem A single feed item
* @return object The RSS-Bridge item
*
* @todo To reduce confusion, the RSS-Bridge item should maybe have a class
* of its own?
*/
protected function parseRss2Item($feedItem)
{
// Primary data is compatible to 0.91 with some additional data
$item = $this->parseRss091Item($feedItem);
$namespaces = $feedItem->getNamespaces(true);
if (isset($namespaces['dc'])) {
$dc = $feedItem->children($namespaces['dc']);
}
if (isset($namespaces['media'])) {
$media = $feedItem->children($namespaces['media']);
}
if (isset($feedItem->guid)) {
foreach ($feedItem->guid->attributes() as $attribute => $value) {
if (
$attribute === 'isPermaLink'
&& (
$value === 'true' || (
filter_var($feedItem->guid, FILTER_VALIDATE_URL)
&& (empty($item['uri']) || !filter_var($item['uri'], FILTER_VALIDATE_URL))
)
)
) {
$item['uri'] = (string)$feedItem->guid;
break;
}
}
}
if (isset($feedItem->pubDate)) {
$item['timestamp'] = strtotime((string)$feedItem->pubDate);
} elseif (isset($dc->date)) {
$item['timestamp'] = strtotime((string)$dc->date);
}
if (isset($feedItem->author)) {
$item['author'] = (string)$feedItem->author;
} elseif (isset($feedItem->creator)) {
$item['author'] = (string)$feedItem->creator;
} elseif (isset($dc->creator)) {
$item['author'] = (string)$dc->creator;
} elseif (isset($media->credit)) {
$item['author'] = (string)$media->credit;
}
if (isset($feedItem->enclosure) && !empty($feedItem->enclosure['url'])) {
$item['enclosures'] = [(string)$feedItem->enclosure['url']];
}
return $item;
}
/**
* Parse the contents of a single feed item, depending on the current feed
* type, into a RSS-Bridge item.
*
* @param object $item The current feed item
* @return object A RSS-Bridge item, with (hopefully) the whole content
*/ */
protected function parseItem($item) protected function parseItem($item)
{ {
switch ($this->feedType) { switch ($this->feedType) {
case self::FEED_TYPE_RSS_1_0: case self::FEED_TYPE_RSS_1_0:
return $this->parseRss1Item($item); return $this->feedParser->parseRss1Item($item);
case self::FEED_TYPE_RSS_2_0: case self::FEED_TYPE_RSS_2_0:
return $this->parseRss2Item($item); return $this->feedParser->parseRss2Item($item);
case self::FEED_TYPE_ATOM_1_0: case self::FEED_TYPE_ATOM_1_0:
return $this->parseATOMItem($item); return $this->feedParser->parseAtomItem($item);
default: default:
throw new \Exception(sprintf('Unknown version %s!', $this->getInput('version'))); throw new \Exception(sprintf('Unknown version %s!', $this->getInput('version')));
} }
} }
/** {@inheritdoc} */
public function getURI() public function getURI()
{ {
if (!empty($this->uri)) { return $this->parsedFeed['uri'] ?? parent::getURI();
return $this->uri;
}
return parent::getURI();
} }
/** {@inheritdoc} */
public function getName() public function getName()
{ {
if (!empty($this->title)) { return $this->parsedFeed['title'] ?? parent::getName();
return $this->title;
}
return parent::getName();
} }
/** {@inheritdoc} */
public function getIcon() public function getIcon()
{ {
if (!empty($this->icon)) { return $this->parsedFeed['icon'] ?? parent::getIcon();
return $this->icon;
}
return parent::getIcon();
} }
} }

205
lib/FeedParser.php Normal file
View file

@ -0,0 +1,205 @@
<?php
declare(strict_types=1);
final class FeedParser
{
public function parseFeed(string $xmlString): array
{
$xml = simplexml_load_string(trim($xmlString));
if ($xml === false) {
throw new \Exception('Unable to parse xml');
}
$feed = [
'title' => null,
'url' => null,
'icon' => null,
'items' => [],
];
if (isset($xml->item[0])) {
// rss 1.0
$channel = $xml->channel[0];
$feed['title'] = trim((string)$channel->title);
$feed['uri'] = trim((string)$channel->link);
if (!empty($channel->image)) {
$feed['icon'] = trim((string)$channel->image->url);
}
foreach ($xml->item as $item) {
$feed['items'][] = $this->parseRss1Item($item);
}
} elseif (isset($xml->channel[0])) {
// rss 2.0
$channel = $xml->channel[0];
$feed['title'] = trim((string)$channel->title);
$feed['uri'] = trim((string)$channel->link);
if (!empty($channel->image)) {
$feed['icon'] = trim((string)$channel->image->url);
}
foreach ($channel->item as $item) {
$feed['items'][] = $this->parseRss2Item($item);
}
} elseif (isset($xml->entry[0])) {
// atom 1.0
$feed['title'] = (string)$xml->title;
// Find best link (only one, or first of 'alternate')
if (!isset($xml->link)) {
$feed['uri'] = '';
} elseif (count($xml->link) === 1) {
$feed['uri'] = (string)$xml->link[0]['href'];
} else {
$feed['uri'] = '';
foreach ($xml->link as $link) {
if (strtolower((string) $link['rel']) === 'alternate') {
$feed['uri'] = (string)$link['href'];
break;
}
}
}
if (!empty($xml->icon)) {
$feed['icon'] = (string)$xml->icon;
} elseif (!empty($xml->logo)) {
$feed['icon'] = (string)$xml->logo;
}
foreach ($xml->entry as $item) {
$feed['items'][] = $this->parseAtomItem($item);
}
} else {
throw new \Exception(sprintf('Unable to detect feed format from `%s`', $url));
}
return $feed;
}
public function parseAtomItem(\SimpleXMLElement $feedItem): array
{
// Some ATOM entries also contain RSS 2.0 fields
$item = $this->parseRss2Item($feedItem);
if (isset($feedItem->id)) {
$item['uri'] = (string)$feedItem->id;
}
if (isset($feedItem->title)) {
$item['title'] = html_entity_decode((string)$feedItem->title);
}
if (isset($feedItem->updated)) {
$item['timestamp'] = strtotime((string)$feedItem->updated);
}
if (isset($feedItem->author)) {
$item['author'] = (string)$feedItem->author->name;
}
if (isset($feedItem->content)) {
$contentChildren = $feedItem->content->children();
if (count($contentChildren) > 0) {
$content = '';
foreach ($contentChildren as $contentChild) {
$content .= $contentChild->asXML();
}
$item['content'] = $content;
} else {
$item['content'] = (string)$feedItem->content;
}
}
// When "link" field is present, URL is more reliable than "id" field
if (count($feedItem->link) === 1) {
$item['uri'] = (string)$feedItem->link[0]['href'];
} else {
foreach ($feedItem->link as $link) {
if (strtolower((string) $link['rel']) === 'alternate') {
$item['uri'] = (string)$link['href'];
}
if (strtolower((string) $link['rel']) === 'enclosure') {
$item['enclosures'][] = (string)$link['href'];
}
}
}
return $item;
}
public function parseRss2Item(\SimpleXMLElement $feedItem): array
{
// Primary data is compatible to 0.91 with some additional data
$item = $this->parseRss091Item($feedItem);
$namespaces = $feedItem->getNamespaces(true);
if (isset($namespaces['dc'])) {
$dc = $feedItem->children($namespaces['dc']);
}
if (isset($namespaces['media'])) {
$media = $feedItem->children($namespaces['media']);
}
if (isset($feedItem->guid)) {
foreach ($feedItem->guid->attributes() as $attribute => $value) {
if (
$attribute === 'isPermaLink'
&& (
$value === 'true' || (
filter_var($feedItem->guid, FILTER_VALIDATE_URL)
&& (empty($item['uri']) || !filter_var($item['uri'], FILTER_VALIDATE_URL))
)
)
) {
$item['uri'] = (string)$feedItem->guid;
break;
}
}
}
if (isset($feedItem->pubDate)) {
$item['timestamp'] = strtotime((string)$feedItem->pubDate);
} elseif (isset($dc->date)) {
$item['timestamp'] = strtotime((string)$dc->date);
}
if (isset($feedItem->author)) {
$item['author'] = (string)$feedItem->author;
} elseif (isset($feedItem->creator)) {
$item['author'] = (string)$feedItem->creator;
} elseif (isset($dc->creator)) {
$item['author'] = (string)$dc->creator;
} elseif (isset($media->credit)) {
$item['author'] = (string)$media->credit;
}
if (isset($feedItem->enclosure) && !empty($feedItem->enclosure['url'])) {
$item['enclosures'] = [(string)$feedItem->enclosure['url']];
}
return $item;
}
public function parseRss1Item(\SimpleXMLElement $feedItem): array
{
// 1.0 adds optional elements around the 0.91 standard
$item = $this->parseRss091Item($feedItem);
$namespaces = $feedItem->getNamespaces(true);
if (isset($namespaces['dc'])) {
$dc = $feedItem->children($namespaces['dc']);
if (isset($dc->date)) {
$item['timestamp'] = strtotime((string)$dc->date);
}
if (isset($dc->creator)) {
$item['author'] = (string)$dc->creator;
}
}
return $item;
}
public function parseRss091Item(\SimpleXMLElement $feedItem): array
{
$item = [];
if (isset($feedItem->link)) {
$item['uri'] = (string)$feedItem->link;
}
if (isset($feedItem->title)) {
$item['title'] = html_entity_decode((string)$feedItem->title);
}
// rss 0.91 doesn't support timestamps
// rss 0.91 doesn't support authors
// rss 0.91 doesn't support enclosures
if (isset($feedItem->description)) {
$item['content'] = (string)$feedItem->description;
}
return $item;
}
}

View file

@ -8,66 +8,13 @@ final class RssBridge
public function __construct() public function __construct()
{ {
Configuration::verifyInstallation();
$customConfig = [];
if (file_exists(__DIR__ . '/../config.ini.php')) {
$customConfig = parse_ini_file(__DIR__ . '/../config.ini.php', true, INI_SCANNER_TYPED);
}
Configuration::loadConfiguration($customConfig, getenv());
// Consider: ini_set('error_reporting', E_ALL & ~E_DEPRECATED);
date_default_timezone_set(Configuration::getConfig('system', 'timezone'));
set_exception_handler(function (\Throwable $e) {
self::$logger->error('Uncaught Exception', ['e' => $e]);
http_response_code(500);
print render(__DIR__ . '/../templates/exception.html.php', ['e' => $e]);
exit(1);
});
set_error_handler(function ($code, $message, $file, $line) {
if ((error_reporting() & $code) === 0) {
return false;
}
// In the future, uncomment this:
//throw new \ErrorException($message, 0, $code, $file, $line);
$text = sprintf(
'%s at %s line %s',
sanitize_root($message),
sanitize_root($file),
$line
);
self::$logger->warning($text);
});
// There might be some fatal errors which are not caught by set_error_handler() or \Throwable.
register_shutdown_function(function () {
$error = error_get_last();
if ($error) {
$message = sprintf(
'(shutdown) %s: %s in %s line %s',
$error['type'],
sanitize_root($error['message']),
sanitize_root($error['file']),
$error['line']
);
self::$logger->error($message);
if (Debug::isEnabled()) {
print sprintf("<pre>%s</pre>\n", e($message));
}
}
});
self::$logger = new SimpleLogger('rssbridge'); self::$logger = new SimpleLogger('rssbridge');
if (Debug::isEnabled()) { if (Debug::isEnabled()) {
self::$logger->addHandler(new StreamHandler(Logger::DEBUG)); self::$logger->addHandler(new StreamHandler(Logger::DEBUG));
} else { } else {
self::$logger->addHandler(new StreamHandler(Logger::INFO)); self::$logger->addHandler(new StreamHandler(Logger::INFO));
} }
self::$httpClient = new CurlHttpClient(); self::$httpClient = new CurlHttpClient();
$cacheFactory = new CacheFactory(self::$logger); $cacheFactory = new CacheFactory(self::$logger);
if (Debug::isEnabled()) { if (Debug::isEnabled()) {
self::$cache = $cacheFactory->create('array'); self::$cache = $cacheFactory->create('array');