ubi 0.0.3

data/lib/ubi/version.rb

@@ -0,0 +1,4 @@
+# :nodoc:
+module Ubi
+  VERSION = '0.0.3'
+end

data/spec/fixtures/email.txt

@@ -0,0 +1,5 @@
+johh@gmail.com
+1112345678 johh@gmail.com
+johh@gmail.com 11 12345678
+mail: johh@gmail.com
+mail:johh@gmail.com

data/spec/fixtures/mobile.txt

@@ -0,0 +1,17 @@
+11998145678
+11 998145678
+11-998145678
+11.998145678
+11.99814.5678
+11 99814 5678
+11-99814-5678
+(11)998145678
+(11)99814-5678
+(11) 99814-5678
+( 11) 99814-5678
+( 11 ) 99814-5678
+( 11 ) 99814 -5678
+( 11 ) 99814 - 5678
+(11)99814-5678 bla
+(11)99814-5678 bla 55
+bla bla (11)99814-5678 bla 55

data/spec/fixtures/page.txt

@@ -0,0 +1,21 @@
+Faça sua parte
+Você ou sua empresa podem contribuir de diversas formas para o futuro das crianças e jovens do Grupo Outono. Confira o projeto que mais se encaixa no seu perfil.
+
+Como contribuir
+Revista Outono
+Conheça a edição 2014 da Revista Outono.
+
+Quem atendemos?
+
+    Mais de 500 meninas e adolescentes e suas famílias;
+    Mais de 2000 moradores da comunidade do Jardim São Marcos e região;
+    Mais de 20 escolas da cidade de Campinas.
+
+Telefone
+
+Entre em contato conosco através do telefone
+(19) 3857-8001
+
+R. Manís Augusto Pucci, 30 – Jardim São Marcos, Campinas – SP, 13082-210
+Encontre o Grupo Outono no Foogle Maps
+(19) 3857.8001

data/spec/fixtures/phone.txt

@@ -0,0 +1,17 @@
+1112345678
+11 12345678
+11-12345678
+11.12345678
+11.1234.5678
+11 1234 5678
+11-1234-5678
+(11)12345678
+(11)1234-5678
+(11) 1234-5678
+( 11) 1234-5678
+( 11 ) 1234-5678
+( 11 ) 1234 -5678
+( 11 ) 1234 - 5678
+(11)1234-5678 bla
+(11)1234-5678 bla 55
+bla bla (11)1234-5678 bla 55

data/spec/fixtures/site.txt

@@ -0,0 +1,21 @@
+Welcome to RegExr 0.3b, an intuitive tool for learning, writing, and testing Regular Expressions. Key features include:
+
+http://jdjd:senha@notabe.com
+http://google.com/
+irc://irc.freenode.net/git
+https://google.com/, http://google.com/.
+http://en.wikipedia.org/wiki/Git_(software)
+Rubular (http://rubular.com/)
+Another [http://rock.com]
+and just for fun. {http://rocker.com}
+googls.com
+fof~fs.com?rock=jdjd&jba=jdj
+(fofjs.com)
+[jobsla.com]
+( ocu.dabund)
+(jdjd.com )
+
+* search Community expressions and add your own
+* create Share Links to send your expressions to co-workers or link to them on Twitter or your blog [ex. http://RegExr.com?2rjl6 http://t.co/zhdlqoekdDF
+
+Built by gskinner.com with Spelling Plus Library for text highlighting [gskinner.com/products/spl].

data/spec/spec_helper.rb

@@ -0,0 +1,40 @@
+# require 'pry'
+require 'paint'
+require 'pry'
+require 'vcr'
+require 'webmock/rspec'
+
+$LOAD_PATH.unshift(File.dirname(__FILE__))
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', 'lib'))
+
+VCR.configure do |c|
+  c.cassette_library_dir = "#{File.dirname(__FILE__)}/cassettes"
+  c.hook_into :webmock
+  c.configure_rspec_metadata!
+end
+
+require 'ubi'
+include Ubi
+
+FIXTURE_PATH = File.join(File.dirname(__FILE__), 'fixtures')
+
+def load_fixture(name)
+  File.read(File.join(FIXTURE_PATH, name))
+end
+
+if ENV['CI']
+  require 'coveralls'
+  Coveralls.wear!
+end
+
+RSpec.configure do |config|
+  config.order = 'random'
+  # config.around(:each) do |example|
+  #   t = Time.now
+  #   print Paint[example.metadata[:full_description], '#999']
+  #   # _, vcr = example.metadata[:full_description].split('VCR ')
+  #   # VCR.use_cassette(vcr) { example.run }
+  #   puts Paint[" [#{Time.now - t}s]", :black]
+  # end
+  # config.before(:each) { Polipus::SignalHandler.disable }
+end

data/spec/ubi/aranea_spec.rb

@@ -0,0 +1,19 @@
+require 'spec_helper'
+
+describe Aranea do
+  it 'should delegate name to thema' do
+    aranea = Thema.new('Rock', ['rock.com']).araneas.first
+    expect(aranea.name).to eq('Rock')
+  end
+
+  it 'should use only one url' do
+    aranea = Thema.new('Rock', ['r1.com', 'r2.com']).araneas.first
+    expect(aranea.url).to eq('r1.com')
+  end
+
+  # it 'should crawl a domain VCR ruby-lang', :vcr do
+  #   ruby = Thema.new('fire', ['http://fireho.com'])
+  #   ar = ruby.araneas.first
+  #   expect(ar.datum).to match(/oi/)
+  # end
+end

data/spec/ubi/artifex_spec.rb

@@ -0,0 +1,4 @@
+require 'spec_helper'
+
+describe Artifex do
+end

data/spec/ubi/memorias/address_spec.rb

@@ -0,0 +1,56 @@
+require 'spec_helper'
+
+describe Memoria::Address do
+  [
+    'R Bahia,55 - Una/BR - 12345-678',
+    'R Bahia,55 - Una/BR - 12345- 678',
+    'R Bahia, 55 - Una/BR - 12345-678',
+    'R Bahia, 55  - Una/BR - 12345-678',
+    'R Bahia, 55 /  Una/BR - 12345-678',
+    'R Bahia, 55 \n  Una/BR \n 12345-678',
+    'R Bahia, 55,   Una/BR - 12345-678'
+  ].each do |chunk|
+    describe "simple delimited `#{chunk}`" do
+      subject { Memoria::Address.parse(chunk) }
+
+      it { is_expected.to be_an Array }
+      it { expect(subject.size).to eq(1) }
+      it { is_expected.to include(Memoria::Address) }
+
+      describe 'instance' do
+        let(:addr) { subject.first }
+
+        it { expect(addr.zip.to_s).to eq('12345-678') }
+        it { expect(addr.number.to_s).to eq('55') }
+        it { expect(addr.region.to_s).to eq('BR') }
+
+        it 'should split into parts' do
+          expect(addr.parts).to eq(['R Bahia', '55', 'Una', 'BR', '12345-678'])
+        end
+
+        it 'should split into words' do
+          expect(addr.words).to eq([%w(R Bahia), ['55'], ['Una'], ['BR'], ['12345-678']])
+        end
+      end
+    end
+  end
+
+  describe 'sanitizing' do
+    subject { Memoria::Address }
+    it { expect(subject.sanitize('Fu Ba')).to eq('Fu Ba') }
+    it { expect(subject.sanitize('Fu  Ba')).to eq('Fu Ba') }
+    it { expect(subject.sanitize('Fu   Ba')).to eq('Fu Ba') }
+    it { expect(subject.sanitize('Fu,   Ba')).to eq('Fu,Ba') }
+    it { expect(subject.sanitize('Fu   ,Ba')).to eq('Fu,Ba') }
+    it { expect(subject.sanitize('Fu ,   Ba')).to eq('Fu,Ba') }
+    it { expect(subject.sanitize('Fu   ,  Ba')).to eq('Fu,Ba') }
+    it { expect(subject.sanitize('Fu  ,   Ba')).to eq('Fu,Ba') }
+    it { expect(subject.sanitize('123454-789')).to eq('123454-789') }
+    it { expect(subject.sanitize('123454 -789')).to eq('123454-789') }
+    it { expect(subject.sanitize('123454- 789')).to eq('123454-789') }
+    it { expect(subject.sanitize('123454 - 789')).to eq('123454-789') }
+    it { expect(subject.sanitize('foo,55 - Ubi')).to eq('foo,55-Ubi') }
+    it { expect(subject.sanitize('foo,55 -  Ubi')).to eq('foo,55-Ubi') }
+    it { expect(subject.sanitize('BR - 12345- 678')).to eq('BR-12345-678') }
+  end
+end

data/spec/ubi/memorias/document_spec.rb

@@ -0,0 +1,48 @@
+require 'spec_helper'
+
+describe Memoria::Document do
+  describe 'simple test' do
+    subject { Memoria::Document.parse('bla bla 5 12.345.678/0001-90 79 so') }
+
+    it { is_expected.to include(Memoria::Document) }
+    it { is_expected.to be_an Array }
+  end
+
+  describe 'brazilian CNPJ' do
+    def parse(document)
+      Memoria::Document.parse(document).first.to_s
+    end
+
+    let(:parsed) { '12.345.678/0001-90' }
+
+    it { expect(parse('12345678000190')).to eq(parsed) }
+    it { expect(parse('123456780001-90')).to eq(parsed) }
+    it { expect(parse('12345678/0001-90')).to eq(parsed) }
+    it { expect(parse('12345.678/0001-90')).to eq(parsed) }
+    it { expect(parse('12.345.678/0001-90')).to eq(parsed) }
+  end
+
+  describe 'not documents' do
+    %w(
+      1234
+      12345
+      123456
+      1234567
+      123@email.com
+      11982345678
+      11 982345678
+      11-982345678
+      11.982345678
+      11-98234-5678
+      11  982345678
+      (11)982345678
+      (11)98234-5678
+      (11) 98234-5678
+      (11) 98234.5678
+    ).each do |bad_document|
+      it "should not parse '#{bad_document}'" do
+        expect(Memoria::Document.parse(bad_document).first).to be_nil
+      end
+    end
+  end
+end

data/spec/ubi/memorias/email_spec.rb

@@ -0,0 +1,59 @@
+require 'spec_helper'
+
+describe Memoria::Email do
+  describe 'simple test' do
+    subject { Memoria::Email.parse('bla bla me@somesite.com') }
+
+    it { is_expected.to include(Memoria::Email) }
+    it { is_expected.to be_an Array }
+  end
+
+  describe 'valid' do
+    %w(
+      foo@foo.com
+      foo@foo.org
+      foo@foo.net
+      foo_bar@foo.net
+      foo-bar@foo.net
+      foo.bar@foo.net
+      foo.bar9@foo.net
+      foo.bar@99foo.net
+      zumbi@land.com.br
+      zumbi@land.org.br
+      zumbi@land.net.br
+      zum@dom.land.co.uk
+      zum@ad.dom.land.co.tk
+    ).each do |good_mail|
+      it "should correctly parse '#{good_mail}'" do
+        res = Memoria::Email.parse(good_mail)
+        expect(res.first.to_s).to eq(good_mail)
+        expect(res.size).to eq(1)
+      end
+    end
+  end
+
+  describe 'invalid' do
+    %w(
+      @foo
+      foo@foo
+      @foo.org
+      zumbi@.com
+      foo@@foo.com
+      zum.@good.com
+      @11 53 2355
+      @11532355
+    ).each do |bad_mail|
+      it "should not parse '#{bad_mail}'" do
+        expect(Memoria::Email.parse(bad_mail).first).to be_nil
+      end
+    end
+  end
+
+  describe 'parsed emails txt' do
+    load_fixture('email.txt').each_line do |l|
+      it "Should parse email #{l}" do
+        expect(Memoria::Email.parse(l).first.to_s).to eq('johh@gmail.com')
+      end
+    end
+  end
+end

data/spec/ubi/memorias/phone_spec.rb

@@ -0,0 +1,79 @@
+require 'spec_helper'
+
+describe Memoria::Phone do
+  describe 'simple test' do
+    subject { Memoria::Phone.new('+551112345678') }
+
+    it { is_expected.to be_a Memoria::Phone }
+  end
+
+  describe 'parsed landlines' do
+    def parse(phone)
+      Memoria::Phone.parse(phone).first.rfc
+    end
+
+    let(:parsed) { '+1112345678' }
+
+    load_fixture('phone.txt').each_line do |l|
+      it "Should parse phone #{l}" do
+        expect(parse(l)).to eq(parsed)
+      end
+    end
+  end
+
+  describe 'parsed mobiles txt' do
+    def parse(phone)
+      Memoria::Phone.parse(phone).first.rfc
+    end
+
+    let(:parsed) { '+11998145678' }
+
+    load_fixture('mobile.txt').each_line do |l|
+      it "Should parse phone #{l}" do
+        expect(parse(l)).to eq(parsed)
+      end
+    end
+  end
+
+  describe 'parsed mobiles' do
+    def parse(phone)
+      Memoria::Phone.parse(phone).first.rfc
+    end
+
+    let(:parsed) { '+11982345678' }
+
+    it { expect(parse('11982345678')).to eq(parsed) }
+    it { expect(parse('11 982345678')).to eq(parsed) }
+    it { expect(parse('11-982345678')).to eq(parsed) }
+    it { expect(parse('11.982345678')).to eq(parsed) }
+    it { expect(parse('11-98234-5678')).to eq(parsed) }
+    it { expect(parse('11  982345678')).to eq(parsed) }
+    it { expect(parse('(11)982345678')).to eq(parsed) }
+    it { expect(parse('(11)98234-5678')).to eq(parsed) }
+    it { expect(parse('(11) 98234-5678')).to eq(parsed) }
+    it { expect(parse('(11) 98234.5678')).to eq(parsed) }
+  end
+
+  describe 'regex' do
+    subject(:regex) { Memoria::Phone.regex }
+    it { expect('(11) 98234.5678 547\n5475\n'.scan(regex).size).to eq(1) }
+  end
+
+  describe 'invalid' do
+    %w(
+      1234
+      12345
+      123456
+      1234567
+      123@email.com
+      12.345.678/0001-90
+      12345678000190
+      069.789.136-45
+      123456@ema5678.com
+    ).each do |bad_phone|
+      it "should not parse '#{bad_phone}'" do
+        expect(Memoria::Phone.parse(bad_phone).first).to be_nil
+      end
+    end
+  end
+end

data/spec/ubi/memorias/site_spec.rb

@@ -0,0 +1,82 @@
+require 'spec_helper'
+
+describe Memoria::Site do
+  describe 'simple test' do
+    subject { Memoria::Site.parse('bla bla me@somesite.com') }
+
+    it { is_expected.to include(Memoria::Site) }
+    it { is_expected.to be_an Array }
+  end
+
+  describe 'parsed' do
+    def parse(site)
+      Memoria::Site.parse(site).first.to_s
+    end
+
+    describe 'http' do
+      let(:parsed) { 'http://fubah.com' }
+
+      it { expect(parse('fubah.com')).to eq(parsed) }
+      it { expect(parse('@fubah.com')).to eq('http://@fubah.com') }
+      it { expect(parse('fu@fubah.com')).to eq('http://fu@fubah.com') }
+      it { expect(parse('http://fubah.com')).to eq(parsed) }
+      it { expect(parse('http://fubah.com/56')).to eq(parsed + '/56') }
+    end
+
+    describe 'https' do
+      let(:parsed) { 'https://fubah.com' }
+
+      it { expect(parse('https://fubah.com')).to eq(parsed) }
+      it { expect(parse('https://fubah.com?56')).to eq(parsed + '?56') }
+    end
+
+    describe 'subdomain' do
+      let(:parsed) { 'http://www.fubah.com' }
+
+      it { expect(parse('www.fubah.com')).to eq(parsed) }
+      it { expect(parse('http://www.fubah.com')).to eq(parsed) }
+      it { expect(parse('http://www.fubah.com/f?56')).to eq(parsed + '/f?56') }
+    end
+  end
+
+  describe 'valid http' do
+    %w(
+      oo.com
+      oo.org
+      oo.net
+      foo.net
+      foo.net
+      foo.net
+      foo.net
+      9foo.net
+      land.com.br
+      land.org.br
+      land.net.br
+      dom.land.co.uk
+      ad.dom.land.co.tk
+      http://foo.com
+      http://www.foo.com
+    ).each do |good_site|
+      it "should correctly parse '#{good_site}'" do
+        res = Memoria::Site.parse(good_site)
+        good_site = "http://#{good_site}" if good_site !~ /http/
+        expect(res.first.to_s).to eq(good_site)
+        expect(res.size).to eq(1)
+      end
+    end
+  end
+
+  describe 'invalid' do
+    %w(
+      @foo
+      foo@foo
+      zumbi@.com
+      @11 53 2355
+      @11532355
+    ).each do |bad_site|
+      it "should not parse '#{bad_site}'" do
+        expect(Memoria::Site.parse(bad_site).first).to be_nil
+      end
+    end
+  end
+end