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

Fix capitalization of isPermaLink 

Also add support for DC Terms dates; tests to come
master
J. King 4 years ago
parent
9ab5616bc8
commit
915852bb2b
6 changed files with 24 additions and 19 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. 5
      lib/Parser/XML/Construct.php
  2. 13
      lib/Parser/XML/Entry.php
  3. 2
      lib/Parser/XML/Feed.php
  4. 1
      lib/Parser/XML/XPath.php
  5. 10
      tests/cases/XML/entry-rss2.yaml
  6. 12
      tests/cases/XML/feed-rss2.yaml

5
lib/Parser/XML/Construct.php
View File

@ -26,6 +26,9 @@ abstract class Construct {
protected const DATE_EARLIEST = 2;
protected const DATE_ALL = 3;
protected const QUERY_AMBIGUOUS_DATES = "rss2:pubDate|rss2:lastBuildDate|dc:date|dc:available|dc:dateAccepted|dc:dateCopyrighted|dc:dateSubmitted|dc:issued|dc:modified|dc:valid|dct:date|dct:available|dct:dateAccepted|dct:dateCopyrighted|dct:dateSubmitted|dct:issued|dct:modified|dct:valid";
protected const QUERY_RSS_PERMALINK = "rss2:guid[not(@isPermaLink) or @isPermaLink='true']";
/** @var \DOMDocument */
protected $document;
/** @var \DOMXPath */
@ -359,7 +362,7 @@ abstract class Construct {
}
protected function getLinkRss2(): ?Url {
return $this->fetchUrl("rss2:link") ?? $this->fetchUrl("rss2:guid[not(@isPermalink) or @isPermalink='true']");
return $this->fetchUrl("rss2:link") ?? $this->fetchUrl(self::QUERY_RSS_PERMALINK);
}
protected function getLinkRss1(): ?Url {

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

@ -88,8 +88,8 @@ class Entry extends Construct implements \MensBeam\Lax\Parser\Entry {
well-defined semantics here. Thus the semantics of all the other
formats are equal, and we want the latest date, whatever it is.
*/
return $this->fetchDate("atom:updated", self::DATE_LATEST) // Atom update date
?? $this->fetchDate("dc:date|rss2:pubDate|rss2:lastBuildDate", self::DATE_LATEST); // Latest other datee
return $this->fetchDate("atom:updated", self::DATE_LATEST) // Atom update date
?? $this->fetchDate(self::QUERY_AMBIGUOUS_DATES, self::DATE_LATEST); // Latest other datee
}
public function getDateCreated(): ?Date {
@ -98,8 +98,9 @@ class Entry extends Construct implements \MensBeam\Lax\Parser\Entry {
formats are equal, and we want the earliest date, but only if
there are at least two
*/
return $this->fetchDate("atom:created", self::DATE_EARLIEST) // Atom creation date
?? $this->getAssumedDateCreated(); // Earliest other date
return $this->fetchDate("atom:created", self::DATE_EARLIEST) // Atom creation date
?? $this->fetchDate("dct:created|dc:created", self::DATE_LATEST) // Dublin Core creation date
?? $this->getAssumedDateCreated(); // Earliest other date
}
public function getContent(): ?Text {
@ -146,7 +147,7 @@ class Entry extends Construct implements \MensBeam\Lax\Parser\Entry {
*/
protected function getLinkAndRelatedRss2(): array {
$link = $this->fetchUrl("rss2:link");
$guid = $this->fetchUrl("rss2:guid[not(@isPermalink) or @isPermalink='true']");
$guid = $this->fetchUrl(self::QUERY_RSS_PERMALINK);
if ($link && $guid) {
if ($link->getScheme() !== $guid->getScheme() || $link->getAuthority() !== $guid->getAuthority()) {
return [$guid, $link];
@ -156,7 +157,7 @@ class Entry extends Construct implements \MensBeam\Lax\Parser\Entry {
}
protected function getAssumedDateCreated(): ?Date {
$dates = $this->fetchDate("dc:date|rss2:pubDate|rss2:lastBuildDate", self::DATE_ALL);
$dates = $this->fetchDate(self::QUERY_AMBIGUOUS_DATES, self::DATE_ALL);
if (sizeof($dates) > 1) {
return $dates[0];
}

2
lib/Parser/XML/Feed.php
View File

@ -164,7 +164,7 @@ 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|rss2:pubDate|rss2:lastBuildDate", self::DATE_LATEST);
?? $this->fetchDate(self::QUERY_AMBIGUOUS_DATES, self::DATE_LATEST);
}
public function getIcon(): ?Url {

1
lib/Parser/XML/XPath.php
View File

@ -13,6 +13,7 @@ class XPath extends \DOMXpath {
'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/
'dct' => "http://purl.org/dc/terms/", // Dublin Core terms https://web.archive.org/web/20071222055924/http://web.resource.org/rss/1.0/modules/dcterms/
'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

10
tests/cases/XML/entry-rss2.yaml
View File

@ -10,7 +10,7 @@ GUID:
input: >
<rss><channel>
<item>
<guid isPermalink="false">blah</guid>
<guid isPermaLink="false">blah</guid>
</item>
</channel></rss>
output:
@ -22,7 +22,7 @@ Language:
input: >
<rss><channel>
<item>
<guid isPermalink="false">blah</guid>
<guid isPermaLink="false">blah</guid>
<language>fr</language>
</item>
</channel></rss>
@ -36,7 +36,7 @@ Entry link:
input: >
<rss><channel>
<item>
<guid isPermalink="true">http://example.com/</guid>
<guid isPermaLink="true">http://example.com/</guid>
</item>
<item>
<guid>http://example.com/</guid>
@ -58,7 +58,7 @@ Related link:
input: >
<rss><channel>
<item>
<guid isPermalink="true">http://example.com/</guid>
<guid isPermaLink="true">http://example.com/</guid>
<link>http://example.net/</link>
</item>
<item>
@ -66,7 +66,7 @@ Related link:
<link>http://example.net/</link>
</item>
<item>
<guid isPermalink="false">http://example.com/</guid>
<guid isPermaLink="false">http://example.com/</guid>
<link>http://example.net/</link>
</item>
<item>

12
tests/cases/XML/feed-rss2.yaml
View File

@ -9,7 +9,7 @@ Channel GUID:
input: >
<rss>
<channel>
<guid isPermalink="false">http://example.com/</guid>
<guid isPermaLink="false">http://example.com/</guid>
</channel>
</rss>
output:
@ -20,7 +20,7 @@ Channel GUID with whitespace:
input: >
<rss>
<channel>
<guid isPermalink="false">
<guid isPermaLink="false">
http://example.com/
</guid>
</channel>
@ -33,7 +33,7 @@ Root GUID: # Any elements on the RSS2 root element should be ignored
input: >
<rss>
<channel/>
<guid isPermalink="false">http://example.com/</guid>
<guid isPermaLink="false">http://example.com/</guid>
</rss>
output:
format: rss
@ -43,7 +43,7 @@ Bogus GUID before good:
<rss>
<channel>
<guid/>
<guid isPermalink="false">http://example.com/</guid>
<guid isPermaLink="false">http://example.com/</guid>
</channel>
</rss>
output:
@ -255,7 +255,7 @@ Feed link via GUID 1:
Feed link via GUID 2:
input: >
<rss><channel>
<guid isPermalink='true'>http://example.com/</guid>
<guid isPermaLink='true'>http://example.com/</guid>
</channel></rss>
output:
format: rss
@ -265,7 +265,7 @@ Feed link via GUID 2:
GUID not a link:
input: >
<rss><channel>
<guid isPermalink='maybe not'>http://example.com/</guid>
<guid isPermaLink='maybe not'>http://example.com/</guid>
</channel></rss>
output:
format: rss

Write Preview
Loading…
Cancel
Save