Physikerwelt has submitted this change and it was merged.
Change subject: Identification of identifiers
......................................................................
Identification of identifiers
Change-Id: I2400975895fca5d61babe8703a237d5038f156dc
---
M MathObject.php
M MathSearch.hooks.php
A db/mathsemantics.sql
R db/snippets/CosProd.sql
A db/snippets/createSentenceHash.sql
R db/snippets/getNorm.sql
A db/snippets/mathidentifier.view.sql
A db/snippets/most_common_meanings.sql
A runTestsOnVagrant
9 files changed, 74 insertions(+), 20 deletions(-)
Approvals:
Physikerwelt: Verified; Looks good to me, approved
diff --git a/MathObject.php b/MathObject.php
index a078c74..6034c95 100644
--- a/MathObject.php
+++ b/MathObject.php
@@ -72,7 +72,7 @@
$dbr = wfGetDB( DB_SLAVE );
try {
$res = $dbr->select( 'mathpagesimilarity', array(
'pagesimilarity_A as A', 'pagesimilarity_B as B', 'pagesimilarity_Value as V'
), "pagesimilarity_A=$pid OR pagesimilarity_B=$pid", __METHOD__, array(
- "ORDER BY" => 'V DESC', "LIMIT" => 10 )
+ "ORDER BY" => 'V DESC', "LIMIT" => 10 )
);
foreach ( $res as $row ) {
if ( $row->A == $pid ) {
@@ -82,7 +82,7 @@
}
$article = WikiPage::newFromId( $other );
$out .= '# [[' . $article->getTitle() . ']]
similarity ' .
- $row->V * 100 . "%\n";
+ $row->V * 100 . "%\n";
// .' ( pageid'.$other.'/'.$row->A.')' );
}
$wgOut->addWikiText( $out );
@@ -96,18 +96,18 @@
$dbr = wfGetDB( DB_SLAVE );
try {
$res = $dbr->select( array( "mathobservation",
"mathvarstat", 'mathpagestat' )
- , array( "mathobservation_featurename",
"mathobservation_featuretype", 'varstat_featurecount',
- 'pagestat_featurecount', "count(*) as localcnt"
), array( "mathobservation_inputhash" => $this->getInputHash(),
- 'varstat_featurename =
mathobservation_featurename',
- 'varstat_featuretype =
mathobservation_featuretype',
- 'pagestat_pageid' => $this->getPageID(),
- 'pagestat_featureid = varstat_id'
- )
- , __METHOD__, array( 'GROUP BY' =>
'mathobservation_featurename',
- 'ORDER BY' => 'varstat_featurecount' )
+ , array( "mathobservation_featurename",
"mathobservation_featuretype", 'varstat_featurecount',
+ 'pagestat_featurecount', "count(*) as
localcnt" ), array( "mathobservation_inputhash" => $this->getInputHash(),
+ 'varstat_featurename =
mathobservation_featurename',
+ 'varstat_featuretype =
mathobservation_featuretype',
+ 'pagestat_pageid' => $this->getPageID(),
+ 'pagestat_featureid = varstat_id'
+ )
+ , __METHOD__, array( 'GROUP BY' =>
'mathobservation_featurename',
+ 'ORDER BY' => 'varstat_featurecount' )
);
} catch ( Exception $e ) {
- return "Databaseproblem";
+ return "Database problem";
}
$wgOut->addWikiText($res->numRows(). 'results');
if ($res->numRows() == 0){
@@ -117,10 +117,31 @@
if ( $res ) {
foreach ( $res as $row ) {
$wgOut->addWikiText( '*' .
$row->mathobservation_featuretype . ' <code>' .
- utf8_decode(
$row->mathobservation_featurename ) . '</code> (' . $row->localcnt . '/'
- . $row->pagestat_featurecount .
"/" . $row->varstat_featurecount . ')' );
+ utf8_decode(
$row->mathobservation_featurename ) . '</code> (' . $row->localcnt . '/'
+ . $row->pagestat_featurecount . "/" .
$row->varstat_featurecount . ')' );
+ $identifiers = $this->getNouns(utf8_decode(
$row->mathobservation_featurename )) ;
+ if ( $identifiers ){
+ foreach($identifiers as $identifier){
+
$wgOut->addWikiText('**'.$identifier->noun .'('.$identifier->evidence.')');
+ }
+ } else {
+ $wgOut->addWikiText('** not found');
+ }
}
}
+ }
+ public function getNouns($identifier){
+ $dbr = wfGetDB( DB_SLAVE );
+ $article = Article::newFromId( $this->pageID );
+ $pagename = (string)$article->getTitle();;
+ $identifiers = $dbr->select('math_identifier',
+ array( 'noun', 'evidence', 'sentence' ),
+ array( 'pageTitle' => $pagename, 'identifier' =>
$identifier),
+ __METHOD__ ,
+ array('ORDER BY' => 'evidence DESC', 'LIMIT' => 5)
+ );
+
+
}
public function updateObservations( $dbw = null ) {
@@ -147,7 +168,7 @@
}
/**
- *
+ *
* @param int $pid
* @param int $eid
* @return self instance
@@ -155,8 +176,8 @@
public static function constructformpage( $pid, $eid ) {
$dbr = wfGetDB( DB_SLAVE );
$res = $dbr->selectRow(
- array( 'mathindex' ),
self::dbIndexFieldsArray(), 'mathindex_page_id = ' . $pid
- . ' AND mathindex_anchor= ' . $eid
+ array( 'mathindex' ), self::dbIndexFieldsArray(),
'mathindex_page_id = ' . $pid
+ . ' AND mathindex_anchor= ' . $eid
);
//self::DebugPrint( var_export( $res, true ) );
$start = microtime(true);
@@ -173,7 +194,7 @@
$out = array( );
$dbr = wfGetDB( DB_SLAVE );
$res = $dbr->select(
- 'mathindex', self::dbIndexFieldsArray(), array(
'mathindex_inputhash' => $this->getInputHash() )
+ 'mathindex', self::dbIndexFieldsArray(), array(
'mathindex_inputhash' => $this->getInputHash() )
);
foreach ( $res as $row ) {
@@ -197,7 +218,7 @@
$wgOut->addHtml( " " );
$pageString = $hidePage ? "" : $this->getPageTitle() . " ";
$wgOut->addWikiText( "[[" . $this->getPageTitle() . "#math" .
$this->getAnchorID()
- . "|" . $pageString . "Eq: " .
$this->getAnchorID() . "]] ", false );
+ . "|" . $pageString . "Eq: " . $this->getAnchorID() .
"]] ", false );
// $wgOut->addHtml( MathLaTeXML::embedMathML( $this->mathml ) );
$wgOut->addHtml( "<br />" );
}
diff --git a/MathSearch.hooks.php b/MathSearch.hooks.php
index a59a40a..e56501a 100644
--- a/MathSearch.hooks.php
+++ b/MathSearch.hooks.php
@@ -33,6 +33,7 @@
$updater->addExtensionTable( 'mathobservation', $dir .
'mathobservation.sql' );
$updater->addExtensionTable( 'mathvarstat', $dir .
'mathvarstat.sql' );
$updater->addExtensionTable( 'mathpagestat', $dir .
'mathpagestat.sql' );
+ $updater->addExtensionTable( 'mathsemantics', $dir .
'mathsemantics.sql' );
} else {
//throw new MWException( "Math extension does not
currently support $type database." );
}
@@ -111,4 +112,4 @@
$files = array_merge( $files, glob( "$testDir/*Test.php" ) );
return true;
}
-}
\ No newline at end of file
+}
diff --git a/db/mathsemantics.sql b/db/mathsemantics.sql
new file mode 100644
index 0000000..3c0dcc2
--- /dev/null
+++ b/db/mathsemantics.sql
@@ -0,0 +1,8 @@
+CREATE TABLE `mathsemantics` (
+ `pageId` int(5) NOT NULL,
+ `identifier` varchar(4) NOT NULL,
+ `evidence` double NOT NULL,
+ `noun` varchar(20) NOT NULL,
+ `sentence` text NOT NULL,
+ KEY `pageId` (`pageId`)
+) ENGINE=MyISAM DEFAULT CHARSET=utf8;
diff --git a/db/CosProd.sql b/db/snippets/CosProd.sql
similarity index 100%
rename from db/CosProd.sql
rename to db/snippets/CosProd.sql
diff --git a/db/snippets/createSentenceHash.sql
b/db/snippets/createSentenceHash.sql
new file mode 100644
index 0000000..50d2af6
--- /dev/null
+++ b/db/snippets/createSentenceHash.sql
@@ -0,0 +1,5 @@
+UPDATE math_semantics
+SET
+ sentenceHash = MD5(sentence)
+WHERE
+ sentenceHash is null
\ No newline at end of file
diff --git a/db/getNorm.sql b/db/snippets/getNorm.sql
similarity index 100%
rename from db/getNorm.sql
rename to db/snippets/getNorm.sql
diff --git a/db/snippets/mathidentifier.view.sql
b/db/snippets/mathidentifier.view.sql
new file mode 100644
index 0000000..7300777
--- /dev/null
+++ b/db/snippets/mathidentifier.view.sql
@@ -0,0 +1,16 @@
+CREATE
+ ALGORITHM = UNDEFINED
+ DEFINER = `root`@`localhost`
+ SQL SECURITY DEFINER
+VIEW `math_identifier` AS
+ select
+ `S`.`identifier` AS `identifier`,
+ `S`.`noun` AS `noun`,
+ `S`.`evidence` AS `evidence`,
+ `S`.`sentence` AS `sentence`,
+ `S`.`sentenceHash` AS `sentenceHash`,
+ `M`.`pageTitle` AS `pageTitle`,
+ `M`.`pageId` AS `pageID`
+ from
+ (`mathsemantics` `S`
+ join `mathIdMap` `M` ON ((`S`.`pageId` = `M`.`pageId`)))
\ No newline at end of file
diff --git a/db/snippets/most_common_meanings.sql
b/db/snippets/most_common_meanings.sql
new file mode 100644
index 0000000..94fe3b1
--- /dev/null
+++ b/db/snippets/most_common_meanings.sql
@@ -0,0 +1 @@
+select identifier, noun, count(*) FROM wikienmath.math_identifier group by
identifier, noun order by count(noun) DESC;
\ No newline at end of file
diff --git a/runTestsOnVagrant b/runTestsOnVagrant
new file mode 100644
index 0000000..ba9a10a
--- /dev/null
+++ b/runTestsOnVagrant
@@ -0,0 +1,2 @@
+#/bin/bash
+/vagrant/mediawiki/tests/phpunit/phpunit.php ./tests/
\ No newline at end of file
--
To view, visit https://gerrit.wikimedia.org/r/115107
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: merged
Gerrit-Change-Id: I2400975895fca5d61babe8703a237d5038f156dc
Gerrit-PatchSet: 3
Gerrit-Project: mediawiki/extensions/MathSearch
Gerrit-Branch: master
Gerrit-Owner: Physikerwelt <w...@physikerwelt.de>
Gerrit-Reviewer: Physikerwelt <w...@physikerwelt.de>
Gerrit-Reviewer: jenkins-bot <>
_______________________________________________
MediaWiki-commits mailing list
MediaWiki-commits@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
