DCausse has uploaded a new change for review. (
https://gerrit.wikimedia.org/r/347824 )
Change subject: Fix problem when using the default similarity
......................................................................
Fix problem when using the default similarity
'default' is no longer a similarity provided by elastic
and cannot be used directly.
Cirrus will now throw an error if the similarity for a
specific field cannot be determined.
Fix and deprecated the default profile and set it to classic.
Add a new "classic" profile and make it the default.
Bug: T160301
Change-Id: I3ae80c142f1d14dc54e782319bdf10cd653dc2c1
---
M CirrusSearch.php
M docs/settings.txt
M includes/Search/TextIndexField.php
M profiles/SimilarityProfiles.php
4 files changed, 20 insertions(+), 7 deletions(-)
git pull ssh://gerrit.wikimedia.org:29418/mediawiki/extensions/CirrusSearch
refs/changes/24/347824/1
diff --git a/CirrusSearch.php b/CirrusSearch.php
index 31ef36c..6792629 100644
--- a/CirrusSearch.php
+++ b/CirrusSearch.php
@@ -365,7 +365,7 @@
// Configure the similarity module
// see profile/SimilarityProfiles.php for more details
-$wgCirrusSearchSimilarityProfile = 'default';
+$wgCirrusSearchSimilarityProfile = 'classic';
// Weight of fields. Must be integers not decimals. If
$wgCirrusSearchAllFields['use']
// is false this can be changed on the fly. If it is true then changes to
this require
diff --git a/docs/settings.txt b/docs/settings.txt
index 6902115..1968ffc 100644
--- a/docs/settings.txt
+++ b/docs/settings.txt
@@ -503,7 +503,7 @@
; $wgCirrusSearchSimilarityProfile
Default:
- $wgCirrusSearchSimilarityProfile = 'default';
+ $wgCirrusSearchSimilarityProfile = 'classic';
Configure the similarity module.
See profile/SimilarityProfiles.php for more details.
@@ -1487,4 +1487,4 @@
$wgCirrusSearchIndexDeletes = false;
Whether deletes are indexed for archive search when page is deleted. Note that
searching
-for archived pages can be done by manually indexing them too.
\ No newline at end of file
+for archived pages can be done by manually indexing them too.
diff --git a/includes/Search/TextIndexField.php
b/includes/Search/TextIndexField.php
index d13a46c..a059a08 100644
--- a/includes/Search/TextIndexField.php
+++ b/includes/Search/TextIndexField.php
@@ -221,7 +221,7 @@
'CirrusSearchSimilarityProfiles',
$config->get( 'CirrusSearchSimilarityProfile' )
);
- $fieldSimilarity = 'default';
+ $fieldSimilarity = null;
if ( isset( $similarity['fields'] ) ) {
if( isset( $similarity['fields'][$field] ) ) {
$fieldSimilarity =
$similarity['fields'][$field];
@@ -233,6 +233,9 @@
$fieldSimilarity =
$similarity['fields']["$field.$analyzer"];
}
}
+ if ( is_null ( $fieldSimilarity ) ) {
+ throw new \RuntimeException( "Invalid similarity
profile, unable to infer the similarity for the field $field, (defining a
__default__ field might solve the issue" );
+ }
return $fieldSimilarity;
}
}
diff --git a/profiles/SimilarityProfiles.php b/profiles/SimilarityProfiles.php
index 234deaf..13d96ed 100644
--- a/profiles/SimilarityProfiles.php
+++ b/profiles/SimilarityProfiles.php
@@ -26,9 +26,19 @@
*/
$wgCirrusSearchSimilarityProfiles = [
- // default profile, uses the classic TF/IDF from Lucene. With ES5
- // this will become BM25.
- 'default' => [],
+ // default profile, uses the classic TF/IDF from Lucene.
+ // deprecated the use of the name default is confusing
+ 'default' => [
+ 'fields' => [
+ '__default__' => 'classic'
+ ]
+ ],
+ // classic profile, uses the classic TF/IDF from Lucene.
+ 'classic' => [
+ 'fields' => [
+ '__default__' => 'classic'
+ ]
+ ],
// BM25 with default values for k and a for all fields
'bm25_with_defaults' => [
'similarity' => [
--
To view, visit https://gerrit.wikimedia.org/r/347824
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: newchange
Gerrit-Change-Id: I3ae80c142f1d14dc54e782319bdf10cd653dc2c1
Gerrit-PatchSet: 1
Gerrit-Project: mediawiki/extensions/CirrusSearch
Gerrit-Branch: master
Gerrit-Owner: DCausse <[email protected]>
_______________________________________________
MediaWiki-commits mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
