# frozen_string_literal: true
require 'rails_helper'
describe SearchIndexer do
let(:post_id) { 99 }
before do
SearchIndexer.enable
end
after do
SearchIndexer.disable
end
def scrub(html, strip_diacritics: false)
SearchIndexer.scrub_html_for_search(html, strip_diacritics: strip_diacritics)
end
it 'can correctly inject if http or https links exist' do
val = "a https://cnn.com?bob=1, http://stuff.com.au?bill=1 b abc.net/xyz=1"
result = SearchIndexer.inject_extra_terms(val)
expected = "a https://cnn.com?bob=1, cnn com bob=1 http://stuff.com.au?bill=1 stuff com au bill=1 b abc.net/xyz=1 net xyz=1"
expect(result).to eq(expected)
end
it 'correctly indexes chinese' do
SiteSetting.default_locale = 'zh_CN'
data = "你好世界"
expect(data.split(" ").length).to eq(1)
SearchIndexer.update_posts_index(post_id, "你好世界", "", "", nil)
raw_data = PostSearchData.where(post_id: post_id).pluck(:raw_data)[0]
expect(raw_data.split(' ').length).to eq(2)
end
it 'extract youtube title' do
html = ""
scrubbed = scrub(html)
expect(scrubbed).to eq("Metallica Mixer Explains Missing Bass on 'And Justice for All' [Exclusive]")
end
it 'extract a link' do
html = "link"
scrubbed = scrub(html)
expect(scrubbed).to eq("http://meta.discourse.org/ link")
end
it 'extracts @username from mentions' do
html = '@狮子 @foo
'
scrubbed = scrub(html)
expect(scrubbed).to eq('@狮子 @foo')
end
it 'extracts @groupname from group mentions' do
html = '@автомобилист
'
scrubbed = scrub(html)
expect(scrubbed).to eq('@автомобилист')
end
it 'extracts emoji name from emoji image' do
html = %Q|
|
scrubbed = scrub(html)
expect(scrubbed).to eq(':wink:')
end
it 'uses ignore_accent setting to strip diacritics' do
html = "HELLO Hétérogénéité Здравствуйте هتاف للترحيب 你好
"
SiteSetting.search_ignore_accents = true
scrubbed = SearchIndexer.scrub_html_for_search(html)
expect(scrubbed).to eq("HELLO Heterogeneite Здравствуите هتاف للترحيب 你好")
SiteSetting.search_ignore_accents = false
scrubbed = SearchIndexer.scrub_html_for_search(html)
expect(scrubbed).to eq("HELLO Hétérogénéité Здравствуйте هتاف للترحيب 你好")
end
it "doesn't index local files" do
html = <<~HTML
HTML
scrubbed = scrub(html)
expect(scrubbed).to eq("Discourse 51%20PM Untitled%20design%20(21)")
end
it 'correctly indexes a post according to version' do
# Preparing so that they can be indexed to right version
SearchIndexer.update_posts_index(post_id, "dummy", "", nil, nil)
PostSearchData.find_by(post_id: post_id).update!(version: -1)
data = "This is a test"
SearchIndexer.update_posts_index(post_id, "", "", nil, data)
raw_data, locale, version = PostSearchData.where(post_id: post_id).pluck(:raw_data, :locale, :version)[0]
expect(raw_data).to eq("This is a test")
expect(locale).to eq(SiteSetting.default_locale)
expect(version).to eq(SearchIndexer::INDEX_VERSION)
SearchIndexer.update_posts_index(post_id, "tester", "", nil, nil)
raw_data = PostSearchData.where(post_id: post_id).pluck(:raw_data)[0]
expect(raw_data).to eq("tester")
end
describe '.index' do
let(:post) { Fabricate(:post) }
it 'should index posts correctly' do
expect { post }.to change { PostSearchData.count }.by(1)
expect { post.update!(raw: "this is new content") }
.to change { post.reload.post_search_data.raw_data }
expect { post.update!(topic_id: Fabricate(:topic).id) }
.to change { post.reload.post_search_data.raw_data }
end
it 'should not index posts with empty raw' do
expect do
post = Fabricate.build(:post, raw: "", post_type: Post.types[:small_action])
post.save!(validate: false)
end.to_not change { PostSearchData.count }
end
it "should not tokenize urls and duplicate title and href in " do
post = Fabricate(:post, raw: <<~RAW)
https://meta.discourse.org/some.png
RAW
post.rebake!
post.reload
topic = post.topic
expect(post.post_search_data.raw_data).to eq(
"#{topic.title} #{topic.category.name} https://meta.discourse.org/some.png meta discourse org some png"
)
end
it 'should not include lightbox in search' do
Jobs.run_immediately!
SiteSetting.crawl_images = true
SiteSetting.max_image_width = 1
stub_request(:get, "https://meta.discourse.org/some.png")
.to_return(status: 200, body: file_from_fixtures("logo.png").read)
src = "https://meta.discourse.org/some.png"
post = Fabricate(:post, raw: <<~RAW)
Let me see how I can fix this image
RAW
post.rebake!
post.reload
topic = post.topic
expect(post.cooked).to include(
CookedPostProcessor::LIGHTBOX_WRAPPER_CSS_CLASS
)
expect(post.post_search_data.raw_data).to eq(
"#{topic.title} #{topic.category.name} Let me see how I can fix this image white walkers GOT"
)
end
end
describe '.queue_post_reindex' do
let(:post) { Fabricate(:post) }
let(:topic) { post.topic }
it 'should reset the version of search data for all posts in the topic' do
post2 = Fabricate(:post)
SearchIndexer.queue_post_reindex(topic.id)
expect(post.reload.post_search_data.version).to eq(
SearchIndexer::REINDEX_VERSION
)
expect(post2.reload.post_search_data.version).to eq(
SearchIndexer::INDEX_VERSION
)
end
end
end
