Merge pull request #654 from AyuntamientoMadrid/search

Mejorar Búsquedas
This commit is contained in:
Juanjo Bazán
2015-10-29 16:46:32 +01:00
11 changed files with 407 additions and 37 deletions

View File

@@ -49,6 +49,7 @@ gem 'daemons'
gem 'devise-async'
gem 'newrelic_rpm', '~> 3.14'
gem 'whenever', require: false
gem 'pg_search'
gem 'ahoy_matey', '~> 1.2.1'
gem 'groupdate' # group temporary data

View File

@@ -263,6 +263,10 @@ GEM
paranoia (2.1.3)
activerecord (~> 4.0)
pg (0.18.3)
pg_search (1.0.5)
activerecord (>= 3.1)
activesupport (>= 3.1)
arel
poltergeist (1.7.0)
capybara (~> 2.1)
cliver (~> 0.3.1)
@@ -455,6 +459,7 @@ DEPENDENCIES
omniauth-twitter
paranoia
pg
pg_search
poltergeist
quiet_assets
rails (= 4.2.4)

View File

@@ -5,6 +5,7 @@ class Debate < ActiveRecord::Base
include Conflictable
include Measurable
include Sanitizable
include PgSearch
apply_simple_captcha
acts_as_votable
@@ -36,6 +37,22 @@ class Debate < ActiveRecord::Base
# Ahoy setup
visitable # Ahoy will automatically assign visit_id on create
pg_search_scope :pg_search, {
against: {
title: 'A',
description: 'B'
},
associated_against: {
tags: :name
},
using: {
tsearch: { dictionary: "spanish" },
trigram: { threshold: 0.1 },
},
ranked_by: '(:tsearch + debates.cached_votes_up)',
order_within_rank: "debates.created_at DESC"
}
def description
super.try :html_safe
end
@@ -102,12 +119,7 @@ class Debate < ActiveRecord::Base
end
def self.search(terms)
return none unless terms.present?
debate_ids = where("debates.title ILIKE ? OR debates.description ILIKE ?",
"%#{terms}%", "%#{terms}%").pluck(:id)
tag_ids = tagged_with(terms, wild: true, any: true).pluck(:id)
where(id: [debate_ids, tag_ids].flatten.compact)
self.pg_search(terms)
end
def after_hide

View File

@@ -4,6 +4,7 @@ class Proposal < ActiveRecord::Base
include Conflictable
include Measurable
include Sanitizable
include PgSearch
apply_simple_captcha
acts_as_votable
@@ -38,6 +39,24 @@ class Proposal < ActiveRecord::Base
scope :sort_by_random, -> { order("RANDOM()") }
scope :sort_by_flags, -> { order(flags_count: :desc, updated_at: :desc) }
pg_search_scope :pg_search, {
against: {
title: 'A',
question: 'B',
summary: 'C',
description: 'D'
},
associated_against: {
tags: :name
},
using: {
tsearch: { dictionary: "spanish" },
trigram: { threshold: 0.1 },
},
ranked_by: '(:tsearch + proposals.cached_votes_up)',
order_within_rank: "proposals.created_at DESC"
}
def description
super.try :html_safe
end
@@ -93,7 +112,7 @@ class Proposal < ActiveRecord::Base
end
def self.search(terms)
terms.present? ? where("title ILIKE ? OR description ILIKE ? OR question ILIKE ?", "%#{terms}%", "%#{terms}%", "%#{terms}%") : none
self.pg_search(terms)
end
def self.votes_needed_for_success

View File

@@ -0,0 +1,5 @@
class AddUnaccentExtension < ActiveRecord::Migration
def change
execute "create extension unaccent"
end
end

View File

@@ -0,0 +1,5 @@
class AddPgTrgmExtension < ActiveRecord::Migration
def change
execute "create extension pg_trgm"
end
end

View File

@@ -11,10 +11,12 @@
#
# It's strongly recommended that you check this file into your version control system.
ActiveRecord::Schema.define(version: 20151021113348) do
ActiveRecord::Schema.define(version: 20151028221647) do
# These are extensions that must be enabled in order to support this database
enable_extension "plpgsql"
enable_extension "unaccent"
enable_extension "pg_trgm"
create_table "activities", force: :cascade do |t|
t.integer "user_id"

View File

@@ -504,10 +504,12 @@ feature 'Debates' do
within("#debates") do
expect(page).to have_css('.debate', count: 4)
expect(page).to have_content(debate2.title)
expect(page).to have_content(debate4.title)
expect(page).to have_content(debate5.title)
expect(page).to have_content(debate6.title)
expect(page).to_not have_content(debate1.title)
expect(page).to_not have_content(debate3.title)
end

View File

@@ -574,9 +574,11 @@ feature 'Proposals' do
within("#proposals") do
expect(page).to have_css('.proposal', count: 3)
expect(page).to have_content(proposal2.title)
expect(page).to have_content(proposal4.title)
expect(page).to have_content(proposal5.title)
expect(page).to_not have_content(proposal1.title)
expect(page).to_not have_content(proposal3.title)
end

View File

@@ -318,35 +318,6 @@ describe Debate do
end
describe "self.search" do
it "find debates by title" do
debate1 = create(:debate, title: "Karpov vs Kasparov")
create(:debate, title: "Bird vs Magic")
search = Debate.search("Kasparov")
expect(search.size).to eq(1)
expect(search.first).to eq(debate1)
end
it "find debates by description" do
debate1 = create(:debate, description: "...chess masters...")
create(:debate, description: "...basket masters...")
search = Debate.search("chess")
expect(search.size).to eq(1)
expect(search.first).to eq(debate1)
end
it "find debates by title and description" do
create(:debate, title: "Karpov vs Kasparov", description: "...played like Gauss...")
create(:debate, title: "Euler vs Gauss", description: "...math masters...")
search = Debate.search("Gauss")
expect(search.size).to eq(2)
end
it "returns no results if no search term provided" do
expect(Debate.search(" ").size).to eq(0)
end
end
describe "cache" do
let(:debate) { create(:debate) }
@@ -457,4 +428,169 @@ describe Debate do
end
describe "search" do
context "attributes" do
it "searches by title" do
debate = create(:debate, title: 'save the world')
results = Debate.search('save the world')
expect(results).to eq([debate])
end
it "searches by description" do
debate = create(:debate, description: 'in order to save the world one must think about...')
results = Debate.search('one must think')
expect(results).to eq([debate])
end
end
context "stemming" do
it "searches word stems" do
debate = create(:debate, title: 'limpiar')
results = Debate.search('limpiará')
expect(results).to eq([debate])
results = Debate.search('limpiémos')
expect(results).to eq([debate])
results = Debate.search('limpió')
expect(results).to eq([debate])
end
end
context "accents" do
it "searches with accents" do
debate = create(:debate, title: 'difusión')
results = Debate.search('difusion')
expect(results).to eq([debate])
debate2 = create(:debate, title: 'estadisticas')
results = Debate.search('estadísticas')
expect(results).to eq([debate2])
end
end
context "case" do
it "searches case insensite" do
debate = create(:debate, title: 'SHOUT')
results = Debate.search('shout')
expect(results).to eq([debate])
debate2 = create(:debate, title: "scream")
results = Debate.search("SCREAM")
expect(results).to eq([debate2])
end
end
context "typos" do
it "searches with typos" do
debate = create(:debate, title: 'difusión')
results = Debate.search('difuon')
expect(results).to eq([debate])
debate2 = create(:debate, title: 'desarrollo')
results = Debate.search('desarolo')
expect(results).to eq([debate2])
end
end
context "order" do
it "orders by weight" do
debate_description = create(:debate, description: 'stop corruption')
debate_title = create(:debate, title: 'stop corruption')
results = Debate.search('stop corruption')
expect(results.first).to eq(debate_title)
expect(results.second).to eq(debate_description)
end
it "orders by weight and then votes" do
title_some_votes = create(:debate, title: 'stop corruption', cached_votes_up: 5)
title_least_voted = create(:debate, title: 'stop corruption', cached_votes_up: 2)
title_most_voted = create(:debate, title: 'stop corruption', cached_votes_up: 10)
description_most_voted = create(:debate, description: 'stop corruption', cached_votes_up: 10)
results = Debate.search('stop corruption')
expect(results.first).to eq(title_most_voted)
expect(results.second).to eq(description_most_voted)
expect(results.third).to eq(title_some_votes)
expect(results.fourth).to eq(title_least_voted)
end
it "orders by weight and then votes and then created_at" do
newest = create(:debate, title: 'stop corruption', cached_votes_up: 5, created_at: Time.now)
oldest = create(:debate, title: 'stop corruption', cached_votes_up: 5, created_at: 1.month.ago)
old = create(:debate, title: 'stop corruption', cached_votes_up: 5, created_at: 1.week.ago)
results = Debate.search('stop corruption')
expect(results.first).to eq(newest)
expect(results.second).to eq(old)
expect(results.third).to eq(oldest)
end
end
context "tags" do
it "searches by tags" do
debate = create(:debate, tag_list: 'Latina')
results = Debate.search('Latina')
expect(results.first).to eq(debate)
end
end
context "no results" do
it "no words match" do
debate = create(:debate, title: 'save world')
results = Debate.search('destroy planet')
expect(results).to eq([])
end
it "too many typos" do
debate = create(:debate, title: 'fantastic')
results = Debate.search('frantac')
expect(results).to eq([])
end
it "too much stemming" do
debate = create(:debate, title: 'reloj')
results = Debate.search('superrelojimetro')
expect(results).to eq([])
end
it "empty" do
debate = create(:debate, title: 'great')
results = Debate.search('')
expect(results).to eq([])
end
end
end
end

View File

@@ -355,4 +355,185 @@ describe Proposal do
end
end
describe "search" do
context "attributes" do
it "searches by title" do
proposal = create(:proposal, title: 'save the world')
results = Proposal.search('save the world')
expect(results).to eq([proposal])
end
it "searches by summary" do
proposal = create(:proposal, summary: 'basically...')
results = Proposal.search('basically')
expect(results).to eq([proposal])
end
it "searches by description" do
proposal = create(:proposal, description: 'in order to save the world one must think about...')
results = Proposal.search('one must think')
expect(results).to eq([proposal])
end
it "searches by question" do
proposal = create(:proposal, question: 'to be or not to be')
results = Proposal.search('to be or not to be')
expect(results).to eq([proposal])
end
end
context "stemming" do
it "searches word stems" do
proposal = create(:proposal, summary: 'limpiar')
results = Proposal.search('limpiará')
expect(results).to eq([proposal])
results = Proposal.search('limpiémos')
expect(results).to eq([proposal])
results = Proposal.search('limpió')
expect(results).to eq([proposal])
end
end
context "accents" do
it "searches with accents" do
proposal = create(:proposal, summary: 'difusión')
results = Proposal.search('difusion')
expect(results).to eq([proposal])
proposal2 = create(:proposal, summary: 'estadisticas')
results = Proposal.search('estadísticas')
expect(results).to eq([proposal2])
end
end
context "case" do
it "searches case insensite" do
proposal = create(:proposal, title: 'SHOUT')
results = Proposal.search('shout')
expect(results).to eq([proposal])
proposal2 = create(:proposal, title: "scream")
results = Proposal.search("SCREAM")
expect(results).to eq([proposal2])
end
end
context "typos" do
it "searches with typos" do
proposal = create(:proposal, summary: 'difusión')
results = Proposal.search('difuon')
expect(results).to eq([proposal])
proposal2 = create(:proposal, summary: 'desarrollo')
results = Proposal.search('desarolo')
expect(results).to eq([proposal2])
end
end
context "order" do
it "orders by weight" do
proposal_question = create(:proposal, question: 'stop corruption')
proposal_title = create(:proposal, title: 'stop corruption')
proposal_description = create(:proposal, description: 'stop corruption')
proposal_summary = create(:proposal, summary: 'stop corruption')
results = Proposal.search('stop corruption')
expect(results.first).to eq(proposal_title)
expect(results.second).to eq(proposal_question)
expect(results.third).to eq(proposal_summary)
expect(results.fourth).to eq(proposal_description)
end
it "orders by weight and then votes" do
title_some_votes = create(:proposal, title: 'stop corruption', cached_votes_up: 5)
title_least_voted = create(:proposal, title: 'stop corruption', cached_votes_up: 2)
title_most_voted = create(:proposal, title: 'stop corruption', cached_votes_up: 10)
summary_most_voted = create(:proposal, summary: 'stop corruption', cached_votes_up: 10)
results = Proposal.search('stop corruption')
expect(results.first).to eq(title_most_voted)
expect(results.second).to eq(summary_most_voted)
expect(results.third).to eq(title_some_votes)
expect(results.fourth).to eq(title_least_voted)
end
it "orders by weight and then votes and then created_at" do
newest = create(:proposal, title: 'stop corruption', cached_votes_up: 5, created_at: Time.now)
oldest = create(:proposal, title: 'stop corruption', cached_votes_up: 5, created_at: 1.month.ago)
old = create(:proposal, title: 'stop corruption', cached_votes_up: 5, created_at: 1.week.ago)
results = Proposal.search('stop corruption')
expect(results.first).to eq(newest)
expect(results.second).to eq(old)
expect(results.third).to eq(oldest)
end
end
context "tags" do
it "searches by tags" do
proposal = create(:proposal, tag_list: 'Latina')
results = Proposal.search('Latina')
expect(results.first).to eq(proposal)
end
end
context "no results" do
it "no words match" do
proposal = create(:proposal, title: 'save world')
results = Proposal.search('destroy planet')
expect(results).to eq([])
end
it "too many typos" do
proposal = create(:proposal, title: 'fantastic')
results = Proposal.search('frantac')
expect(results).to eq([])
end
it "too much stemming" do
proposal = create(:proposal, title: 'reloj')
results = Proposal.search('superrelojimetro')
expect(results).to eq([])
end
it "empty" do
proposal = create(:proposal, title: 'great')
results = Proposal.search('')
expect(results).to eq([])
end
end
end
end