Fix: #370 ダブルクオートを使った日本語検索が効かない (#448)

* Fix: #370 ダブルクオートを使った日本語検索が効かない

* Add test for multiple words

* Improve test

* 入れ替え

* 単語検索のテスト
This commit is contained in:
KMY(雪あすか) 2024-01-12 10:27:35 +09:00 committed by GitHub
parent b044cbeb63
commit d864399a09
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
5 changed files with 56 additions and 4 deletions

View file

@ -92,6 +92,6 @@ class AccountsIndex < Chewy::Index
field(:domain, type: 'keyword', value: ->(account) { account.domain || '' })
field(:display_name, type: 'text', analyzer: 'verbatim') { field :edge_ngram, type: 'text', analyzer: 'edge_ngram', search_analyzer: 'verbatim' }
field(:username, type: 'text', analyzer: 'verbatim', value: ->(account) { [account.username, account.domain].compact.join('@') }) { field :edge_ngram, type: 'text', analyzer: 'edge_ngram', search_analyzer: 'verbatim' }
field(:text, type: 'text', analyzer: 'sudachi_analyzer', value: ->(account) { account.searchable_text })
field(:text, type: 'text', analyzer: 'sudachi_analyzer', value: ->(account) { account.searchable_text }) { field(:stemmed, type: 'text', analyzer: 'natural') }
end
end

View file

@ -87,7 +87,7 @@ class PublicStatusesIndex < Chewy::Index
root date_detection: false do
field(:id, type: 'long')
field(:account_id, type: 'long')
field(:text, type: 'text', analyzer: 'sudachi_analyzer', value: ->(status) { status.searchable_text })
field(:text, type: 'text', analyzer: 'sudachi_analyzer', value: ->(status) { status.searchable_text }) { field(:stemmed, type: 'text', analyzer: 'content') }
field(:tags, type: 'text', analyzer: 'hashtag', value: ->(status) { status.tags.map(&:display_name) })
field(:language, type: 'keyword')
field(:domain, type: 'keyword', value: ->(status) { status.account.domain || '' })

View file

@ -107,7 +107,7 @@ class StatusesIndex < Chewy::Index
root date_detection: false do
field(:id, type: 'long')
field(:account_id, type: 'long')
field(:text, type: 'text', analyzer: 'sudachi_analyzer', value: ->(status) { status.searchable_text })
field(:text, type: 'text', analyzer: 'sudachi_analyzer', value: ->(status) { status.searchable_text }) { field(:stemmed, type: 'text', analyzer: 'content') }
field(:tags, type: 'text', analyzer: 'hashtag', value: ->(status) { status.tags.map(&:display_name) })
field(:searchable_by, type: 'long', value: ->(status) { status.searchable_by })
field(:mentioned_by, type: 'long', value: ->(status) { status.mentioned_by })

View file

@ -237,7 +237,7 @@ class SearchQueryTransformer < Parslet::Transform
else
# Memo for checking when manually merge
# { multi_match: { type: 'most_fields', query: @term, fields: ['text', 'text.stemmed'], operator: 'and' } }
{ match_phrase: { text: { query: @term } } }
{ multi_match: { type: 'most_fields', query: @term, fields: ['text', 'text.stemmed'], operator: 'and' } }
end
end
end
@ -251,6 +251,8 @@ class SearchQueryTransformer < Parslet::Transform
end
def to_query
# Memo for checking when manually merge
# { match_phrase: { text: { query: @phrase } } }
{ match_phrase: { text: { query: @phrase } } }
end
end

View file

@ -213,4 +213,54 @@ describe StatusesSearchService do
end
end
end
describe 'a local user posts with search keyword' do
subject do
described_class.new.call(search_keyword, account, limit: 10).map(&:id)
end
let(:search_keyword) { 'ohagi' }
let(:status_text) { 'りんごを食べました' }
let(:alice) { Fabricate(:user).account }
let(:account) { alice }
let!(:status) { Fabricate(:status, text: status_text, account: alice, searchability: :public) }
before do
alice.update!(username: 'alice')
StatusesIndex.import!
end
shared_examples 'hit status' do |name, keyword|
context name do
let(:search_keyword) { keyword }
it 'a status hits' do
expect(subject.count).to eq 1
expect(subject).to include status.id
end
end
end
shared_examples 'does not hit status' do |name, keyword|
context name do
let(:search_keyword) { keyword }
it 'no statuses hit' do
expect(subject.count).to eq 0
end
end
end
it_behaves_like 'hit status', 'when search with word', 'りんご'
it_behaves_like 'hit status', 'when search with multiple words', 'りんご 食べる'
it_behaves_like 'does not hit status', 'when search with multiple words but does not hit half', 'りんご 茹でる'
it_behaves_like 'hit status', 'when search with letter in word', 'ご'
it_behaves_like 'does not hit status', 'when double quote search with letter in word', '"ご"'
it_behaves_like 'hit status', 'when search with fixed word', '"りんご"'
it_behaves_like 'hit status', 'when double quote search with multiple letter in word', 'り ご'
it_behaves_like 'does not hit status', 'when double quote search with multiple letter in word but does not contain half', 'ず ご'
it_behaves_like 'hit status', 'when specify user name', 'りんご from:alice'
it_behaves_like 'does not hit status', 'when specify not existing user name', 'りんご from:ohagi'
end
end