Physikerwelt has uploaded a new change for review.
https://gerrit.wikimedia.org/r/255079
Change subject: Improve snippet generation
......................................................................
Improve snippet generation
Change-Id: Ib201ba5a2693f516ac377ac28df6cb100bb6a93e
---
M includes/special/SpecialMlpEval.php
1 file changed, 48 insertions(+), 7 deletions(-)
git pull ssh://gerrit.wikimedia.org:29418/mediawiki/extensions/MathSearch
refs/changes/79/255079/1
diff --git a/includes/special/SpecialMlpEval.php
b/includes/special/SpecialMlpEval.php
index a2dc2fe..7b1eaba 100644
--- a/includes/special/SpecialMlpEval.php
+++ b/includes/special/SpecialMlpEval.php
@@ -13,13 +13,14 @@
*/
class SpecialMlpEval extends SpecialPage {
const MAX_ATTEMPTS = 10;
- const WINDOW_SIZE = 500;
+ const WINDOW_SIZE = 1200;
private $step = 1;
/**
* @var Title
*/
private $title;
private $wikitext;
+ private $snippet;
private $mathTags;
private $revison;
private $lastError = false;
@@ -179,6 +180,7 @@
return LoggerFactory::getInstance( 'MathSearch' );
}
+
/**
* @return array
* @throws MWException
@@ -190,23 +192,62 @@
$tag = $this->mathTags[$unique];
$formDescriptor = array();
$this->getOutput()->addWikiText( $tag[3] );
-
$tagPos = strpos( $this->wikitext, $unique );
$wikiText = $this->wikitext;
- $wikiText = substr( $wikiText,
- max( $tagPos - self::WINDOW_SIZE, 0 ),
- min( 2 * self::WINDOW_SIZE, strlen( $wikiText ) -
$tagPos ) );
+ $startPos = $this->getStartPos( $tagPos, $wikiText );
+ $length = $this->getEndPos( $tagPos, $wikiText ) - $startPos;
+ $wikiText = substr( $wikiText, $startPos, $length );
$wikiText = str_replace( $unique,
'<span id="theelement" style="background-color:
yellow">' . $tag[3] . '</span>',
$wikiText );
foreach ( $this->mathTags as $key => $content ) {
$wikiText = str_replace( $key, $content[3], $wikiText );
}
-
- $this->getOutput()->addWikiText( "== Extract
==\n...\n$wikiText\n..." );
+ $this->snippet = "== Extract ==\nStart of the
extract...\n\n$wikiText\n\n...end of the extract";
+ $this->getOutput()->addWikiText( $this->snippet );
$url = $this->title->getLinkURL();
$this->getOutput()
->addHTML( "<a href=\"$url\" target=\"_blank\">Full
article (new Window)</a>" );
return array( $tagCount, $formDescriptor, $wikiText );
}
+
+ /**
+ * @param $tagPos
+ * @param $wikiText
+ * @return array
+ */
+ private function getStartPos( $tagPos, $wikiText ) {
+ $startPos = max( $tagPos - round( self::WINDOW_SIZE / 2 ), 0 );
+ if ( $startPos > 0 ) {
+ // Heuristics to find a reasonable cutting point
+ $newPos = strpos( $wikiText, "\n", $startPos );
+ if ( $newPos !== false && ( $newPos - $startPos ) <
round( self::WINDOW_SIZE / 4 ) ) {
+ // only change startPos, if it seems reasonable
+ $startPos = $newPos;
+ }
+ }
+ return $startPos;
+ }
+
+ /**
+ * @param $wikiText
+ * @param $tagPos
+ * @return bool|int|mixed
+ */
+ private function getEndPos( $tagPos, $wikiText ) {
+ $halfWindow = round( self::WINDOW_SIZE / 2 );
+ $distance2End = strlen( $wikiText ) - $tagPos;
+ if ( $distance2End > $halfWindow ) {
+ $newPos = strpos( $wikiText, "\n", $tagPos +
$halfWindow );
+ if ( $newPos !== false && ( $newPos - $tagPos ) <
round( 3 / 4 * self::WINDOW_SIZE ) ) {
+ // only change startPos, if it seems reasonable
+ return $newPos;
+ } else {
+ return $tagPos + $halfWindow;
+ }
+ } else {
+ return strlen( $wikiText );
+ }
+ }
+
}
--
To view, visit https://gerrit.wikimedia.org/r/255079
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: newchange
Gerrit-Change-Id: Ib201ba5a2693f516ac377ac28df6cb100bb6a93e
Gerrit-PatchSet: 1
Gerrit-Project: mediawiki/extensions/MathSearch
Gerrit-Branch: master
Gerrit-Owner: Physikerwelt <[email protected]>
_______________________________________________
MediaWiki-commits mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
