Show matching context from the document body in ferret search results
ClosedPublic
Actions

Authored by mmodell on Feb 28 2020, 4:08 AM.

Details

Maniphest Tasks

T230787: Make search context highlights work with the ferret search engine

Reviewers

brennen
mmodell

Patch without arc

git checkout -b D1179 && curl -L https://phabricator.wikimedia.org/D1179?download=true | git apply

Summary

This is working fairly well, surprisingly.

Note: I've cheated, to get the context I re-build the search document using the indexing code
instead of querying mysql to get it out of the {object-type}_ffields table.
I _think_ this is more efficient than the alternative, however, I may be entirely mistaken.

Bug: T230787
Maniphest-Task: T230787

Test Plan

run a global-search query and look for snippets of the body text, complete with bolded text for any matching terms that appear within the body.

Diff Detail

Repository

rPHAB Phabricator

Branch

search-context (branched from wmf/stable)

Lint

Lint Errors

Severity	Location	Code	Message
Error	/srv/phab/libext/misc/scripts/init/init-script.php:3	PHL3	One Class Per File
Error	/srv/phab/libext/misc/src/oauth/PhabricatorMediaWikiAuthProvider.php:273	PHL1	Unknown Symbol
Error	/srv/phab/libext/misc/src/other/CustomLoginHandler.php:6	PHL1	Unknown Symbol
Error	src/applications/metamta/adapter/PhabricatorMailImplementationPHPMailerAdapter.php:4	PHL1	Unknown Symbol

Unit

Test Failures

Build Status

Buildable 3281
Build 5442: arc lint + arc unit

Unit TestsFailed
View All

	Time	Test
	177 ms	PhabricatorLibraryTestCase::testEverythingImplemented EXCEPTION (PhutilMissingSymbolException): Failed to load class/interface "PhabricatorMailImplementationPHPMailerAdapter". The symbol map for library "phabricator" (at "/srv/phab/phabricator/src") claims this symbol (of type "class/interface") is defined in "applications/metamta/adapter/PhabricatorMailImplementationPHPMailerAdapter.php", but loading that source file did not cause the symbol to become defined.
	363 ms	PhabricatorLibraryTestCase::testMethodVisibility EXCEPTION (PhutilMissingSymbolException): Failed to load class or interface "PhabricatorMailImplementationAdapter". The class or interface "PhabricatorMailImplementationAdapter" is not defined in the library map of any loaded library.
	154 ms	PhabricatorCelerityTestCase::testCelerityMaps 1 assertion passed.
	11 ms	PhabricatorConduitTestCase::testConduitMethods 1 assertion passed.
	0 ms	PhabricatorInfrastructureTestCase::testApplicationsInstalled 1 assertion passed.
		View Full Test Results (2 Failed · 5 Passed)

Event Timeline

mmodell requested review of this revision.Feb 28 2020, 4:08 AM

mmodell created this revision.

Harbormaster completed remote builds in B3279: Diff 3042.Feb 28 2020, 4:08 AM

mmodell added a task: T230787: Make search context highlights work with the ferret search engine.Feb 28 2020, 4:10 AM

mmodell mentioned this in T230787: Make search context highlights work with the ferret search engine.Feb 28 2020, 4:13 AM

A couple of nits commented inline. Still working on getting my head around the larger mechanics. What I understand seems pretty reasonable.

src/applications/search/fulltextstorage/PhabricatorFerretFulltextStorageEngine.php
89	`is_null($fulltext_engine)`?
src/applications/search/view/PhabricatorSearchResultView.php
56	Any case this would ever hit something that consists entirely of `0`? i.e., should this check something besides truthiness of the `$body` value itself?

Improved context display behavior
Improved some comments
Now searches the whole body for context instead of trimming to 1200 bytes.

Harbormaster completed remote builds in B3281: Diff 3044.Mar 4 2020, 1:17 AM

mmodell edited the summary of this revision. (Show Details)Mar 4 2020, 1:19 AM

mmodell retitled this revision from WIP: show matching context from the document body in ferret search results to Show matching context from the document body in ferret search results.Apr 8 2020, 5:35 PM

mmodell added a project: Phabricator.

mmodell accepted this revision.Jul 29 2020, 5:13 PM

This revision is now accepted and ready to land.Jul 29 2020, 5:13 PM

mmodell closed this revision.Jul 29 2020, 5:14 PM