MensBeam
/
lax
3
0
Fork 0
Code Issues 3 Pull Requests Releases Activity
Browse Source

Only match RSS2 elements in RSS2 documents

master
J. King 4 years ago
parent
dffedfb556
commit
e93234c3df
3 changed files with 22 additions and 13 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 10
      lib/Parser/XML/Construct.php
  2. 13
      lib/Parser/XML/Feed.php
  3. 12
      lib/Parser/XML/XPath.php

10
lib/Parser/XML/Construct.php
View File

@ -272,7 +272,7 @@ abstract class Construct {
* Using RSS' <guid> for feed identifiers is non-standard, but harmless
*/
protected function getIdRss2(): ?string {
return $this->fetchString("guid", ".+");
return $this->fetchString("rss2:guid", ".+");
}
/** Primitive to fetch a Dublin Core feed/entry identifier */
@ -295,7 +295,7 @@ abstract class Construct {
}
protected function getLangRss2(): ?string {
return $this->fetchString("language", ".+");
return $this->fetchString("rss2:language", ".+");
}
protected function getLinkAtom(): ?Url {
@ -303,7 +303,7 @@ abstract class Construct {
}
protected function getLinkRss2(): ?Url {
return $this->fetchUrl("link") ?? $this->fetchUrl("guid[not(@isPermalink) or @isPermalink='true']");
return $this->fetchUrl("rss2:link") ?? $this->fetchUrl("rss2:guid[not(@isPermalink) or @isPermalink='true']");
}
protected function getLinkRss1(): ?Url {
@ -319,7 +319,7 @@ abstract class Construct {
}
protected function getTitleRss2(): ?Text {
return $this->fetchText("title", self::TEXT_LOOSE);
return $this->fetchText("rss2:title", self::TEXT_LOOSE);
}
protected function getTitleDC(): ?Text {
@ -346,7 +346,7 @@ abstract class Construct {
protected function getCategoriesRss2(): ?CategoryCollection {
$out = new CategoryCollection;
foreach ($this->xpath->query("category") as $node) {
foreach ($this->xpath->query("rss2:category") as $node) {
$c = new Category;
$c->domain = $this->trimText($node->getAttribute("domain"));
$c->name = $this->trimText($node->textContent);

13
lib/Parser/XML/Feed.php
View File

@ -53,6 +53,7 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
$this->subject = $this->fetchElement("channel") ?? $this->subject;
$feed->format = "rss";
$feed->version = $this->document->documentElement->hasAttribute("version") ? $this->document->documentElement->getAttribute("version") : null;
$this->xpath->rss2 = true;
} elseif ($ns === XPath::NS['rdf'] && $name === "RDF") {
$feed->format = "rdf";
$channel = $this->fetchElement("rss1:channel|rss0:channel");
@ -150,7 +151,7 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
?? $this->fetchText("dc:description", self::TEXT_PLAIN) // Dublin Core description
?? $this->fetchText("rss1:description", self::TEXT_LOOSE) // RSS 1.0 description
?? $this->fetchText("rss0:description", self::TEXT_LOOSE) // RSS 0.90 description
?? $this->fetchText("description", self::TEXT_LOOSE) // RSS 2.0 description
?? $this->fetchText("rss2:description", self::TEXT_LOOSE) // RSS 2.0 description
?? $this->fetchText("gplay:description", self::TEXT_PLAIN) // Google Play podcast description
?? $this->fetchText("apple:summary", self::TEXT_PLAIN) // iTunes podcast summary
?? $this->fetchText("apple:subtitle", self::TEXT_PLAIN); // iTunes podcast subtitle
@ -162,7 +163,7 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
formats are equal, and we want the latest date, whatever it is.
*/
return $this->fetchDate("atom:updated", self::DATE_LATEST)
?? $this->fetchDate("dc:date|pubDate|lastBuildDate", self::DATE_LATEST);
?? $this->fetchDate("dc:date|rss2:pubDate|rss2:lastBuildDate", self::DATE_LATEST);
}
public function getIcon(): ?Url {
@ -178,7 +179,7 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
?? $this->fetchUrl("/rdf:RDF/rss1:image/rss1:url") // RSS 1.0 root image
?? $this->fetchUrl("rss0:image/rss0:url") // RSS 0.90 channel image
?? $this->fetchUrl("/rdf:RDF/rss0:image/rss0:url") // RSS 0.90 root image
?? $this->fetchUrl("image/url") // RSS 2.0 channel image
?? $this->fetchUrl("rss2:image/rss2:url") // RSS 2.0 channel image
?? $this->fetchUrl("gplay:image/@href") // Google Play podcast image
?? $this->fetchUrl("apple:image/@href"); // iTunes podcast image
}
@ -211,7 +212,7 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
/** Fetches the "time-to-live" value (a number of minutes before the feed should be re-fetched) from an RSS 2.0 feed */
protected function getSchedIntervalRss2(): ?\DateInterval {
$ttl = (int) $this->fetchString("ttl", "\d+");
$ttl = (int) $this->fetchString("rss2:ttl", "\d+");
if ($ttl) {
return new \DateInterval("PT{$ttl}M");
}
@ -247,7 +248,7 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
/** Computes the "skip-schedule" of an RSS feed, the set of days and hours during which a feed should not be fetched */
protected function getSchedSkipRss2(): ?int {
$out = 0;
foreach($this->fetchString("skipHours/hour", "\d+", true) ?? [] as $h) {
foreach($this->fetchString("rss2:skipHours/rss2:hour", "\d+", true) ?? [] as $h) {
$out |= [
Schedule::HOUR_0,
Schedule::HOUR_1,
@ -276,7 +277,7 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
Schedule::HOUR_0,
][(int) $h] ?? 0;
}
foreach($this->fetchString("skipDays/day", null, true) ?? [] as $d) {
foreach($this->fetchString("rss2:skipDays/rss2:day", null, true) ?? [] as $d) {
$out |= [
"monday" => Schedule::DAY_MON,
"tuesday" => Schedule::DAY_TUE,

12
lib/Parser/XML/XPath.php
View File

@ -9,24 +9,32 @@ namespace MensBeam\Lax\Parser\XML;
class XPath extends \DOMXpath {
public const NS = [
'atom' => "http://www.w3.org/2005/Atom", // Atom syndication format https://tools.ietf.org/html/rfc4287
'rss2' => "", // RSS 2.0 does not have a namespace // Really Simple Syndication 2.0.11 http://www.rssboard.org/rss-specification
'rss1' => "http://purl.org/rss/1.0/", // RDF site summary 1.0 http://purl.org/rss/1.0/spec
'rss0' => "http://channel.netscape.com/rdf/simple/0.9/", // RDF Site Summary 0.90 http://www.rssboard.org/rss-0-9-0
'dc' => "http://purl.org/dc/elements/1.1/", // Dublin Core metadata http://purl.org/rss/1.0/modules/dc/
'sched' => "http://purl.org/rss/1.0/modules/syndication/", // Syndication schedule extension http://purl.org/rss/1.0/modules/syndication/
'enc' => "http://purl.org/rss/1.0/modules/content/", // Explicitly encoded content extension http://purl.org/rss/1.0/modules/content/
'media' => "http://search.yahoo.com/mrss/", // Embedded media extension http://www.rssboard.org/media-rss
// RSS 2.0 does not have a namespace // Really Simple Syndication 2.0.11 http://www.rssboard.org/rss-specification
'rdf' => "http://www.w3.org/1999/02/22-rdf-syntax-ns#", // Resource Description Framework
'xhtml' => "http://www.w3.org/1999/xhtml", // XHTML
'apple' => "http://www.itunes.com/dtds/podcast-1.0.dtd", // iTunes podcasts https://help.apple.com/itc/podcasts_connect/#/itcb54353390
'gplay' => "http://www.google.com/schemas/play-podcasts/1.0", // Google Play podcasts https://support.google.com/googleplay/podcasts/answer/6260341
];
public $rss2 = false;
/** Returns an XPath processor with various necessary namespace prefixes defined */
public function __construct(\DOMDocument $doc) {
parent::__construct($doc);
foreach (XPath::NS as $prefix => $url) {
foreach (self::NS as $prefix => $url) {
$this->registerNamespace($prefix, $url);
}
}
/** {@inheritDoc} */
public function query($expression, $contextnode = null, $registerNS = true) {
$expression = $this->rss2 ? str_replace("rss2:", "", $expression) : $expression;
return parent::query($expression, $contextnode, $registerNS);
}
}

Write Preview
Loading…
Cancel
Save