Kimurai Scraping Framework

Note about v1.0.0 version:

  • The code was massively refactored for a support to run spiders multiple times from inside a single process. Now it's possible to run Kimurai spiders using background jobs like Sidekiq.
  • require 'kimurai' doesn't require any gems except Active Support. Only when a particular spider starts, Capybara will be required with a specific driver.
  • Although Kimurai extends Capybara (all the magic happens inside extended Capybara::Session#visit method), session instances which were created manually will behave normally.
  • Small changes in design (check the readme again to see what was changed)
  • Stats database with a web dashboard were removed


Note: this readme is for 1.4.0 gem version. CHANGELOG here.

Kimurai is a modern web scraping framework written in Ruby which works out of box with Headless Chromium/Firefox, PhantomJS, or simple HTTP requests and allows to scrape and interact with JavaScript rendered websites.

Kimurai based on well-known Capybara and Nokogiri gems, so you don't have to learn anything new. Lets see:

# github_spider.rb
require 'kimurai'

class GithubSpider < Kimurai::Base
  @name = "github_spider"
  @engine = :selenium_chrome
  @start_urls = ["https://github.com/search?q=Ruby%20Web%20Scraping"]
  @config = {
    user_agent: "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/68.0.3440.84 Safari/537.36",
    before_request: { delay: 4..7 }
  }

  def parse(response, url:, data: {})
    response.xpath("//ul[@class='repo-list']/div//h3/a").each do |a|
      request_to :parse_repo_page, url: absolute_url(a[:href], base: url)
    end

    if next_page = response.at_xpath("//a[@class='next_page']")
      request_to :parse, url: absolute_url(next_page[:href], base: url)
    end
  end

  def parse_repo_page(response, url:, data: {})
    item = {}

    item[:owner] = response.xpath("//h1//a[@rel='author']").text
    item[:repo_name] = response.xpath("//h1/strong[@itemprop='name']/a").text
    item[:repo_url] = url
    item[:description] = response.xpath("//span[@itemprop='about']").text.squish
    item[:tags] = response.xpath("//div[@id='topics-list-container']/div/a").map { |a| a.text.squish }
    item[:watch_count] = response.xpath("//ul[@class='pagehead-actions']/li[contains(., 'Watch')]/a[2]").text.squish
    item[:star_count] = response.xpath("//ul[@class='pagehead-actions']/li[contains(., 'Star')]/a[2]").text.squish
    item[:fork_count] = response.xpath("//ul[@class='pagehead-actions']/li[contains(., 'Fork')]/a[2]").text.squish
    item[:last_commit] = response.xpath("//span[@itemprop='dateModified']/*").text

    save_to "results.json", item, format: :pretty_json
  end
end

GithubSpider.crawl!

Run: $ ruby github_spider.rb

I, [2018-08-22 13:08:03 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Spider: started: github_spider
D, [2018-08-22 13:08:03 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: BrowserBuilder (selenium_chrome): created browser instance
D, [2018-08-22 13:08:03 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: BrowserBuilder (selenium_chrome): enabled `browser before_request delay`
D, [2018-08-22 13:08:03 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: Browser: sleep 7 seconds before request...
D, [2018-08-22 13:08:10 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: BrowserBuilder (selenium_chrome): enabled custom user-agent
D, [2018-08-22 13:08:10 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: BrowserBuilder (selenium_chrome): enabled native headless_mode
I, [2018-08-22 13:08:10 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Browser: started get request to: https://github.com/search?q=Ruby%20Web%20Scraping
I, [2018-08-22 13:08:26 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Browser: finished get request to: https://github.com/search?q=Ruby%20Web%20Scraping
I, [2018-08-22 13:08:26 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Info: visits: requests: 1, responses: 1
D, [2018-08-22 13:08:27 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: Browser: driver.current_memory: 107968
D, [2018-08-22 13:08:27 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: Browser: sleep 5 seconds before request...
I, [2018-08-22 13:08:32 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Browser: started get request to: https://github.com/lorien/awesome-web-scraping
I, [2018-08-22 13:08:33 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Browser: finished get request to: https://github.com/lorien/awesome-web-scraping
I, [2018-08-22 13:08:33 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Info: visits: requests: 2, responses: 2
D, [2018-08-22 13:08:33 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: Browser: driver.current_memory: 212542
D, [2018-08-22 13:08:33 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: Browser: sleep 4 seconds before request...
I, [2018-08-22 13:08:37 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Browser: started get request to: https://github.com/jaimeiniesta/metainspector

...

I, [2018-08-22 13:23:07 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Browser: started get request to: https://github.com/preston/idclight
I, [2018-08-22 13:23:08 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Browser: finished get request to: https://github.com/preston/idclight
I, [2018-08-22 13:23:08 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Info: visits: requests: 140, responses: 140
D, [2018-08-22 13:23:08 +0400#15477] [M: 47377500980720] DEBUG -- github_spider: Browser: driver.current_memory: 204198
I, [2018-08-22 13:23:08 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Browser: driver selenium_chrome has been destroyed

I, [2018-08-22 13:23:08 +0400#15477] [M: 47377500980720]  INFO -- github_spider: Spider: stopped: {:spider_name=>"github_spider", :status=>:completed, :environment=>"development", :start_time=>2018-08-22 13:08:03 +0400, :stop_time=>2018-08-22 13:23:08 +0400, :running_time=>"15m, 5s", :visits=>{:requests=>140, :responses=>140}, :error=>nil}

results.json

[
  {
    "owner": "lorien",
    "repo_name": "awesome-web-scraping",
    "repo_url": "https://github.com/lorien/awesome-web-scraping",
    "description": "List of libraries, tools and APIs for web scraping and data processing.",
    "tags": [
      "awesome",
      "awesome-list",
      "web-scraping",
      "data-processing",
      "python",
      "javascript",
      "php",
      "ruby"
    ],
    "watch_count": "159",
    "star_count": "2,423",
    "fork_count": "358",
    "last_commit": "4 days ago",
    "position": 1
  },

  ...

  {
    "owner": "preston",
    "repo_name": "idclight",
    "repo_url": "https://github.com/preston/idclight",
    "description": "A Ruby gem for accessing the freely available IDClight (IDConverter Light) web service, which convert between different types of gene IDs such as Hugo and Entrez. Queries are screen scraped from http://idclight.bioinfo.cnio.es.",
    "tags": [

    ],
    "watch_count": "6",
    "star_count": "1",
    "fork_count": "0",
    "last_commit": "on Apr 12, 2012",
    "position": 127
  }
]


Okay, that was easy. How about javascript rendered websites with dynamic HTML? Lets scrape a page with infinite scroll:

# infinite_scroll_spider.rb
require 'kimurai'

class InfiniteScrollSpider < Kimurai::Base
  @name = "infinite_scroll_spider"
  @engine = :selenium_chrome
  @start_urls = ["https://infinite-scroll.com/demo/full-page/"]

  def parse(response, url:, data: {})
    posts_headers_path = "//article/h2"
    count = response.xpath(posts_headers_path).count

    loop do
      browser.execute_script("window.scrollBy(0,10000)") ; sleep 2
      response = browser.current_response

      new_count = response.xpath(posts_headers_path).count
      if count == new_count
        logger.info "> Pagination is done" and break
      else
        count = new_count
        logger.info "> Continue scrolling, current count is #{count}..."
      end
    end

    posts_headers = response.xpath(posts_headers_path).map(&:text)
    logger.info "> All posts from page: #{posts_headers.join('; ')}"
  end
end

InfiniteScrollSpider.crawl!

Run: $ ruby infinite_scroll_spider.rb

I, [2018-08-22 13:32:57 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: Spider: started: infinite_scroll_spider
D, [2018-08-22 13:32:57 +0400#23356] [M: 47375890851320] DEBUG -- infinite_scroll_spider: BrowserBuilder (selenium_chrome): created browser instance
D, [2018-08-22 13:32:57 +0400#23356] [M: 47375890851320] DEBUG -- infinite_scroll_spider: BrowserBuilder (selenium_chrome): enabled native headless_mode
I, [2018-08-22 13:32:57 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: Browser: started get request to: https://infinite-scroll.com/demo/full-page/
I, [2018-08-22 13:33:03 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: Browser: finished get request to: https://infinite-scroll.com/demo/full-page/
I, [2018-08-22 13:33:03 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: Info: visits: requests: 1, responses: 1
D, [2018-08-22 13:33:03 +0400#23356] [M: 47375890851320] DEBUG -- infinite_scroll_spider: Browser: driver.current_memory: 95463
I, [2018-08-22 13:33:05 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: > Continue scrolling, current count is 5...
I, [2018-08-22 13:33:18 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: > Continue scrolling, current count is 9...
I, [2018-08-22 13:33:20 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: > Continue scrolling, current count is 11...
I, [2018-08-22 13:33:26 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: > Continue scrolling, current count is 13...
I, [2018-08-22 13:33:28 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: > Continue scrolling, current count is 15...
I, [2018-08-22 13:33:30 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: > Pagination is done
I, [2018-08-22 13:33:30 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: > All posts from page: 1a - Infinite Scroll full page demo; 1b - RGB Schemes logo in Computer Arts; 2a - RGB Schemes logo; 2b - Masonry gets horizontalOrder; 2c - Every vector 2016; 3a - Logo Pizza delivered; 3b - Some CodePens; 3c - 365daysofmusic.com; 3d - Holograms; 4a - Huebee: 1-click color picker; 4b - Word is Flickity is good; Flickity v2 released: groupCells, adaptiveHeight, parallax; New tech gets chatter; Isotope v3 released: stagger in, IE8 out; Packery v2 released
I, [2018-08-22 13:33:30 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: Browser: driver selenium_chrome has been destroyed
I, [2018-08-22 13:33:30 +0400#23356] [M: 47375890851320]  INFO -- infinite_scroll_spider: Spider: stopped: {:spider_name=>"infinite_scroll_spider", :status=>:completed, :environment=>"development", :start_time=>2018-08-22 13:32:57 +0400, :stop_time=>2018-08-22 13:33:30 +0400, :running_time=>"33s", :visits=>{:requests=>1, :responses=>1}, :error=>nil}


Features

  • Scrape javascript rendered websites out of box
  • Supported engines: Headless Chrome, Headless Firefox, PhantomJS or simple HTTP requests (mechanize gem)
  • Write spider code once, and use it with any supported engine later
  • All the power of Capybara: use methods like click_on, fill_in, select, choose, set, go_back, etc. to interact with web pages
  • Rich configuration: set default headers, cookies, delay between requests, enable proxy/user-agents rotation
  • Built-in helpers to make scraping easy, like save_to (save items to JSON, JSON lines, or CSV formats) or unique? to skip duplicates
  • Automatically handle requests errors
  • Automatically restart browsers when reaching memory limit (memory control) or requests limit
  • Easily schedule spiders within cron using Whenever (no need to know cron syntax)
  • Parallel scraping using simple method in_parallel
  • Two modes: use single file for a simple spider, or generate Scrapy-like project
  • Convenient development mode with console, colorized logger and debugger (Pry, Byebug)
  • Automated server environment setup (for ubuntu 18.04) and deploy using commands kimurai setup and kimurai deploy (Ansible under the hood)
  • Command-line runner to run all project spiders one by one or in parallel

Table of Contents

Installation

Kimurai requires Ruby version >= 2.5.0. Supported platforms: Linux and Mac OS X.

1) If your system doesn't have appropriate Ruby version, install it:

Ubuntu 18.04

# Install required packages for ruby-build
sudo apt update
sudo apt install git-core curl zlib1g-dev build-essential libssl-dev libreadline-dev libreadline6-dev libyaml-dev libxml2-dev libxslt1-dev libcurl4-openssl-dev libffi-dev

# Install rbenv and ruby-build
cd && git clone https://github.com/rbenv/rbenv.git ~/.rbenv
echo 'export PATH="$HOME/.rbenv/bin:$PATH"' >> ~/.bashrc
echo 'eval "$(rbenv init -)"' >> ~/.bashrc
exec $SHELL

git clone https://github.com/rbenv/ruby-build.git ~/.rbenv/plugins/ruby-build
echo 'export PATH="$HOME/.rbenv/plugins/ruby-build/bin:$PATH"' >> ~/.bashrc
exec $SHELL

# Install latest Ruby
rbenv install 2.5.3
rbenv global 2.5.3

gem install bundler

Mac OS X

# Install homebrew if you don't have it https://brew.sh/
# Install rbenv and ruby-build:
brew install rbenv ruby-build

# Add rbenv to bash so that it loads every time you open a terminal
echo 'if which rbenv > /dev/null; then eval "$(rbenv init -)"; fi' >> ~/.bash_profile
source ~/.bash_profile

# Install latest Ruby
rbenv install 2.5.3
rbenv global 2.5.3

gem install bundler

2) Install Kimurai gem: $ gem install kimurai

3) Install browsers with webdrivers:

Ubuntu 18.04

Note: for Ubuntu 16.04-18.04 there is available automatic installation using setup command:

$ kimurai setup localhost --local --ask-sudo

It works using Ansible so you need to install it first: $ sudo apt install ansible. You can check using playbooks here.

If you chose automatic installation, you can skip following and go to "Getting To Know" part. In case if you want to install everything manually:

# Install basic tools
sudo apt install -q -y unzip wget tar openssl

# Install xvfb (for virtual_display headless mode, in additional to native)
sudo apt install -q -y xvfb

# Install chromium-browser and firefox
sudo apt install -q -y chromium-browser firefox

# Instal chromedriver (2.44 version)
# All versions located here https://sites.google.com/a/chromium.org/chromedriver/downloads
cd /tmp && wget https://chromedriver.storage.googleapis.com/2.44/chromedriver_linux64.zip
sudo unzip chromedriver_linux64.zip -d /usr/local/bin
rm -f chromedriver_linux64.zip

# Install geckodriver (0.23.0 version)
# All versions located here https://github.com/mozilla/geckodriver/releases/
cd /tmp && wget https://github.com/mozilla/geckodriver/releases/download/v0.23.0/geckodriver-v0.23.0-linux64.tar.gz
sudo tar -xvzf geckodriver-v0.23.0-linux64.tar.gz -C /usr/local/bin
rm -f geckodriver-v0.23.0-linux64.tar.gz

# Install PhantomJS (2.1.1)
# All versions located here http://phantomjs.org/download.html
sudo apt install -q -y chrpath libxft-dev libfreetype6 libfreetype6-dev libfontconfig1 libfontconfig1-dev
cd /tmp && wget https://bitbucket.org/ariya/phantomjs/downloads/phantomjs-2.1.1-linux-x86_64.tar.bz2
tar -xvjf phantomjs-2.1.1-linux-x86_64.tar.bz2
sudo mv phantomjs-2.1.1-linux-x86_64 /usr/local/lib
sudo ln -s /usr/local/lib/phantomjs-2.1.1-linux-x86_64/bin/phantomjs /usr/local/bin
rm -f phantomjs-2.1.1-linux-x86_64.tar.bz2

Mac OS X

# Install chrome and firefox
brew cask install google-chrome firefox

# Install chromedriver (latest)
brew cask install chromedriver

# Install geckodriver (latest)
brew install geckodriver

# Install PhantomJS (latest)
brew install phantomjs


Also, if you want to save scraped items to the database (using ActiveRecord, Sequel or MongoDB Ruby Driver/Mongoid), you need to install database clients/servers:

Ubuntu 18.04

SQlite: $ sudo apt -q -y install libsqlite3-dev sqlite3.

If you want to connect to a remote database, you don't need database server on a local machine (only client):

# Install MySQL client
sudo apt -q -y install mysql-client libmysqlclient-dev

# Install Postgres client
sudo apt install -q -y postgresql-client libpq-dev

# Install MongoDB client
sudo apt install -q -y mongodb-clients

But if you want to save items to a local database, database server required as well:

# Install MySQL client and server
sudo apt -q -y install mysql-server mysql-client libmysqlclient-dev

# Install  Postgres client and server
sudo apt install -q -y postgresql postgresql-contrib libpq-dev

# Install MongoDB client and server
# version 4.0 (check here https://docs.mongodb.com/manual/tutorial/install-mongodb-on-ubuntu/)
sudo apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv 9DA31620334BD75D9DCB49F368818C72E52529D4
# for 16.04:
# echo "deb [ arch=amd64,arm64 ] https://repo.mongodb.org/apt/ubuntu xenial/mongodb-org/4.0 multiverse" | sudo tee /etc/apt/sources.list.d/mongodb-org-4.0.list
# for 18.04:
echo "deb [ arch=amd64 ] https://repo.mongodb.org/apt/ubuntu bionic/mongodb-org/4.0 multiverse" | sudo tee /etc/apt/sources.list.d/mongodb-org-4.0.list
sudo apt update
sudo apt install -q -y mongodb-org
sudo service mongod start

Mac OS X

SQlite: $ brew install sqlite3

# Install MySQL client and server
brew install mysql
# Start server if you need it: brew services start mysql

# Install Postgres client and server
brew install postgresql
# Start server if you need it: brew services start postgresql

# Install MongoDB client and server
brew install mongodb
# Start server if you need it: brew services start mongodb

Getting to Know

Interactive console

Before you get to know all Kimurai features, there is $ kimurai console command which is an interactive console where you can try and debug your scraping code very quickly, without having to run any spider (yes, it's like Scrapy shell).

$ kimurai console --engine selenium_chrome --url https://github.com/vifreefly/kimuraframework

Show output

$ kimurai console --engine selenium_chrome --url https://github.com/vifreefly/kimuraframework

D, [2018-08-22 13:42:32 +0400#26079] [M: 47461994677760] DEBUG -- : BrowserBuilder (selenium_chrome): created browser instance
D, [2018-08-22 13:42:32 +0400#26079] [M: 47461994677760] DEBUG -- : BrowserBuilder (selenium_chrome): enabled native headless_mode
I, [2018-08-22 13:42:32 +0400#26079] [M: 47461994677760]  INFO -- : Browser: started get request to: https://github.com/vifreefly/kimuraframework
I, [2018-08-22 13:42:35 +0400#26079] [M: 47461994677760]  INFO -- : Browser: finished get request to: https://github.com/vifreefly/kimuraframework
D, [2018-08-22 13:42:35 +0400#26079] [M: 47461994677760] DEBUG -- : Browser: driver.current_memory: 201701

From: /home/victor/code/kimurai/lib/kimurai/base.rb @ line 189 Kimurai::Base#console:

    188: def console(response = nil, url: nil, data: {})
 => 189:   binding.pry
    190: end

[1] pry(#<Kimurai::Base>)> response.xpath("//title").text
=> "GitHub - vifreefly/kimuraframework: Modern web scraping framework written in Ruby which works out of box with Headless Chromium/Firefox, PhantomJS, or simple HTTP requests and allows to scrape and interact with JavaScript rendered websites"

[2] pry(#<Kimurai::Base>)> ls
Kimurai::Base#methods: browser  console  logger  request_to  save_to  unique?
instance variables: @browser  @config  @engine  @logger  @pipelines
locals: _  __  _dir_  _ex_  _file_  _in_  _out_  _pry_  data  response  url

[3] pry(#<Kimurai::Base>)> ls response
Nokogiri::XML::PP::Node#methods: inspect  pretty_print
Nokogiri::XML::Searchable#methods: %  /  at  at_css  at_xpath  css  search  xpath
Enumerable#methods:
  all?         collect         drop        each_with_index   find_all    grep_v    lazy    member?    none?      reject        slice_when  take_while  without
  any?         collect_concat  drop_while  each_with_object  find_index  group_by  many?   min        one?       reverse_each  sort        to_a        zip
  as_json      count           each_cons   entries           first       include?  map     min_by     partition  select        sort_by     to_h
  chunk        cycle           each_entry  exclude?          flat_map    index_by  max     minmax     pluck      slice_after   sum         to_set
  chunk_while  detect          each_slice  find              grep        inject    max_by  minmax_by  reduce     slice_before  take        uniq
Nokogiri::XML::Node#methods:
  <=>                   append_class       classes                 document?             has_attribute?      matches?          node_name=        processing_instruction?  to_str
  ==                    attr               comment?                each                  html?               name=             node_type         read_only?               to_xhtml
  >                     attribute          content                 elem?                 inner_html          namespace=        parent=           remove                   traverse
  []                    attribute_nodes    content=                element?              inner_html=         namespace_scopes  parse             remove_attribute         unlink
  []=                   attribute_with_ns  create_external_subset  element_children      inner_text          namespaced_key?   path              remove_class             values
  accept                before             create_internal_subset  elements              internal_subset     native_content=   pointer_id        replace                  write_html_to
  add_class             blank?             css_path                encode_special_chars  key?                next              prepend_child     set_attribute            write_to
  add_next_sibling      cdata?             decorate!               external_subset       keys                next=             previous          text                     write_xhtml_to
  add_previous_sibling  child              delete                  first_element_child   lang                next_element      previous=         text?                    write_xml_to
  after                 children           description             fragment?             lang=               next_sibling      previous_element  to_html                  xml?
  ancestors             children=          do_xinclude             get_attribute         last_element_child  node_name         previous_sibling  to_s
Nokogiri::XML::Document#methods:
  <<         canonicalize  collect_namespaces  create_comment  create_entity     decorate    document  encoding   errors   name        remove_namespaces!  root=  to_java  url       version
  add_child  clone         create_cdata        create_element  create_text_node  decorators  dup       encoding=  errors=  namespaces  root                slop!  to_xml   validate
Nokogiri::HTML::Document#methods: fragment  meta_encoding  meta_encoding=  serialize  title  title=  type
instance variables: @decorators  @errors  @node_cache

[4] pry(#<Kimurai::Base>)> exit
I, [2018-08-22 13:43:47 +0400#26079] [M: 47461994677760]  INFO -- : Browser: driver selenium_chrome has been destroyed
$


CLI options:

  • --engine (optional) engine to use. Default is mechanize
  • --url (optional) url to process. If url omitted, response and url objects inside the console will be nil (use browser object to navigate to any webpage).

Available engines

Kimurai has support for following engines and mostly can switch between them without need to rewrite any code:

  • :mechanize - pure Ruby fake http browser. Mechanize can't render javascript and don't know what DOM is it. It only can parse original HTML code of a page. Because of it, mechanize much faster, takes much less memory and in general much more stable than any real browser. Use mechanize if you can do it, and the website doesn't use javascript to render any meaningful parts of its structure. Still, because mechanize trying to mimic a real browser, it supports almost all Capybara's methods to interact with a web page (filling forms, clicking buttons, checkboxes, etc).
  • :poltergeist_phantomjs - PhantomJS headless browser, can render javascript. In general, PhantomJS still faster than Headless Chrome (and Headless Firefox). PhantomJS has memory leakage, but Kimurai has memory control feature so you shouldn't consider it as a problem. Also, some websites can recognize PhantomJS and block access to them. Like mechanize (and unlike selenium engines) :poltergeist_phantomjs can freely rotate proxies and change headers on the fly (see config section).
  • :selenium_chrome Chrome in headless mode driven by selenium. Modern headless browser solution with proper javascript rendering.
  • :selenium_firefox Firefox in headless mode driven by selenium. Usually takes more memory than other drivers, but sometimes can be useful.

Tip: add HEADLESS=false ENV variable before command ($ HEADLESS=false ruby spider.rb) to run browser in normal (not headless) mode and see it's window (only for selenium-like engines). It works for console command as well.

Minimum required spider structure

You can manually create a spider file, or use generator instead: $ kimurai generate spider simple_spider

require 'kimurai'

class SimpleSpider < Kimurai::Base
  @name = "simple_spider"
  @engine = :selenium_chrome
  @start_urls = ["https://example.com/"]

  def parse(response, url:, data: {})
  end
end

SimpleSpider.crawl!

Where:

  • @name name of a spider. You can omit name if use single-file spider
  • @engine engine for a spider
  • @start_urls array of start urls to process one by one inside parse method
  • Method parse is the start method, should be always present in spider class

Method arguments response, url and data

def parse(response, url:, data: {})
end
  • response (Nokogiri::HTML::Document object) Contains parsed HTML code of a processed webpage
  • url (String) url of a processed webpage
  • data (Hash) uses to pass data between requests

Example how to use data

Imagine that there is a product page which doesn't contain product category. Category name present only on category page with pagination. This is the case where we can use data to pass category name from parse to parse_product method:

class ProductsSpider < Kimurai::Base
  @engine = :selenium_chrome
  @start_urls = ["https://example-shop.com/example-product-category"]

  def parse(response, url:, data: {})
    category_name = response.xpath("//path/to/category/name").text
    response.xpath("//path/to/products/urls").each do |product_url|
      # Merge category_name with current data hash and pass it next to parse_product method
      request_to(:parse_product, url: product_url[:href], data: data.merge(category_name: category_name))
    end

    # ...
  end

  def parse_product(response, url:, data: {})
    item = {}
    # Assign item's category_name from data[:category_name]
    item[:category_name] = data[:category_name]

    # ...
  end
end


You can query response using XPath or CSS selectors. Check Nokogiri tutorials to understand how to work with response:

browser object

From any spider instance method there is available browser object, which is Capybara::Session object and uses to process requests and get page response (current_response method). Usually you don't need to touch it directly, because there is response (see above) which contains page response after it was loaded.

But if you need to interact with a page (like filling form fields, clicking elements, checkboxes, etc) browser is ready for you:

class GoogleSpider < Kimurai::Base
  @name = "google_spider"
  @engine = :selenium_chrome
  @start_urls = ["https://www.google.com/"]

  def parse(response, url:, data: {})
    browser.fill_in "q", with: "Kimurai web scraping framework"
    browser.click_button "Google Search"

    # Update response to current response after interaction with a browser
    response = browser.current_response

    # Collect results
    results = response.xpath("//div[@class='g']//h3/a").map do |a|
      { title: a.text, url: a[:href] }
    end

    # ...
  end
end

Check out Capybara cheat sheets where you can see all available methods to interact with browser:

request_to method

For making requests to a particular method there is request_to. It requires minimum two arguments: :method_name and url:. An optional argument is data: (see above what for is it). Example:

class Spider < Kimurai::Base
  @engine = :selenium_chrome
  @start_urls = ["https://example.com/"]

  def parse(response, url:, data: {})
    # Process request to `parse_product` method with `https://example.com/some_product` url:
    request_to :parse_product, url: "https://example.com/some_product"
  end

  def parse_product(response, url:, data: {})
    puts "From page https://example.com/some_product !"
  end
end

Under the hood request_to simply call #visit (browser.visit(url)) and then required method with arguments:

request_to

def request_to(handler, url:, data: {})
  request_data = { url: url, data: data }

  browser.visit(url)
  public_send(handler, browser.current_response, request_data)
end


request_to just makes things simpler, and without it we could do something like:

Check the code

class Spider < Kimurai::Base
  @engine = :selenium_chrome
  @start_urls = ["https://example.com/"]

  def parse(response, url:, data: {})
    url_to_process = "https://example.com/some_product"

    browser.visit(url_to_process)
    parse_product(browser.current_response, url: url_to_process)
  end

  def parse_product(response, url:, data: {})
    puts "From page https://example.com/some_product !"
  end
end

save_to helper

Sometimes all that you need is to simply save scraped data to a file format, like JSON or CSV. You can use save_to for it:

class ProductsSpider < Kimurai::Base
  @engine = :selenium_chrome
  @start_urls = ["https://example-shop.com/"]

  # ...

  def parse_product(response, url:, data: {})
    item = {}

    item[:title] = response.xpath("//title/path").text
    item[:description] = response.xpath("//desc/path").text.squish
    item[:price] = response.xpath("//price/path").text[/\d+/]&.to_f

    # Add each new item to the `scraped_products.json` file:
    save_to "scraped_products.json", item, format: :json
  end
end

Supported formats:

  • :json JSON
  • :pretty_json "pretty" JSON (JSON.pretty_generate)
  • :jsonlines JSON Lines
  • :csv CSV

Note: save_to requires data (item to save) to be a Hash.

By default save_to add position key to an item hash. You can disable it with position: false: save_to "scraped_products.json", item, format: :json, position: false.

How helper works:

Until spider stops, each new item will be appended to a file. At the next run, helper will clear the content of a file first, and then start again appending items to it.

If you don't want file to be cleared before each run, add option append: true: save_to "scraped_products.json", item, format: :json, append: true

Skip duplicates

It's pretty common when websites have duplicated pages. For example when an e-commerce shop has the same products in different categories. To skip duplicates, there is simple unique? helper:

class ProductsSpider < Kimurai::Base
  @engine = :selenium_chrome
  @start_urls = ["https://example-shop.com/"]

  def parse(response, url:, data: {})
    response.xpath("//categories/path").each do |category|
      request_to :parse_category, url: category[:href]
    end
  end

  # Check products for uniqueness using product url inside of parse_category:
  def parse_category(response, url:, data: {})
    response.xpath("//products/path").each do |product|
      # Skip url if it's not unique:
      next unless unique?(:product_url, product[:href])
      # Otherwise process it:
      request_to :parse_product, url: product[:href]
    end
  end

  # Or/and check products for uniqueness using product sku inside of parse_product:
  def parse_product(response, url:, data: {})
    item = {}
    item[:sku] = response.xpath("//product/sku/path").text.strip.upcase
    # Don't save product and return from method if there is already saved item with the same sku:
    return unless unique?(:sku, item[:sku])

    # ...
    save_to "results.json", item, format: :json
  end
end

unique? helper works pretty simple:

# Check string "http://example.com" in scope `url` for a first time:
unique?(:url, "http://example.com")
# => true

# Try again:
unique?(:url, "http://example.com")
# => false

To check something for uniqueness, you need to provide a scope:

# `product_url` scope
unique?(:product_url, "http://example.com/product_1")

# `id` scope
unique?(:id, 324234232)

# `custom` scope
unique?(:custom, "Lorem Ipsum")

Automatically skip all duplicated requests urls

It is possible to automatically skip all already visited urls while calling request_to method, using @config option skip_duplicate_requests: true. With this option, all already visited urls will be automatically skipped. Also check the @config for an additional options of this setting.

storage object

unique? method it's just an alias for storage#unique?. Storage has several methods:

  • #all - display storage hash where keys are existing scopes.
  • #include?(scope, value) - return true if value in the scope exists, and false if not
  • #add(scope, value) - add value to the scope
  • #unique?(scope, value) - method already described above, will return false if value in the scope exists, or return true + add value to the scope if value in the scope not exists.
  • #clear! - reset the whole storage by deleting all values from all scopes.

Handle request errors

It is quite common that some pages of crawling website can return different response code than 200 ok. In such cases, method request_to (or browser.visit) can raise an exception. Kimurai provides skip_request_errors and retry_request_errors config options to handle such errors:

skip_request_errors

You can automatically skip some of errors while requesting a page using skip_request_errors config option. If raised error matches one of the errors in the list, then this error will be caught, and request will be skipped. It is a good idea to skip errors like NotFound(404), etc.

Format for the option: array where elements are error classes or/and hashes. You can use hash format for more flexibility:

@config = {
  skip_request_errors: [{ error: RuntimeError, message: "404 => Net::HTTPNotFound" }]
}

In this case, provided message: will be compared with a full error message using String#include?. Also you can use regex instead: { error: RuntimeError, message: /404|403/ }.

retry_request_errors

You can automatically retry some of errors with a few attempts while requesting a page using retry_request_errors config option. If raised error matches one of the errors in the list, then this error will be caught and the request will be processed again within a delay.

There are 3 attempts: first: delay 15 sec, second: delay 30 sec, third: delay 45 sec. If after 3 attempts there is still an exception, then the exception will be raised. It is a good idea to try to retry errros like ReadTimeout, HTTPBadGateway, etc.

Format for the option: same like for skip_request_errors option.

If you would like to skip (not raise) error after all retries gone, you can specify skip_on_failure: true option:

@config = {
  retry_request_errors: [{ error: RuntimeError, skip_on_failure: true }]
}

Logging custom events

It is possible to save custom messages to the run_info hash using add_event('Some message') method. This feature helps you to keep track on important things which happened during crawling without checking the whole spider log (in case if you're logging these messages using logger). Example:

def parse_product(response, url:, data: {})
  unless response.at_xpath("//path/to/add_to_card_button")
    add_event("Product is sold") and return
  end

  # ...
end
...
I, [2018-11-28 22:20:19 +0400#7402] [M: 47156576560640]  INFO -- example_spider: Spider: new event (scope: custom): Product is sold
...
I, [2018-11-28 22:20:19 +0400#7402] [M: 47156576560640]  INFO -- example_spider: Spider: stopped: {:events=>{:custom=>{"Product is sold"=>1}}}

open_spider and close_spider callbacks

You can define .open_spider and .close_spider callbacks (class methods) to perform some action before spider started or after spider has been stopped:

require 'kimurai'

class ExampleSpider < Kimurai::Base
  @name = "example_spider"
  @engine = :selenium_chrome
  @start_urls = ["https://example.com/"]

  def self.open_spider
    logger.info "> Starting..."
  end

  def self.close_spider
    logger.info "> Stopped!"
  end

  def parse(response, url:, data: {})
    logger.info "> Scraping..."
  end
end

ExampleSpider.crawl!

Output

I, [2018-08-22 14:26:32 +0400#6001] [M: 46996522083840]  INFO -- example_spider: Spider: started: example_spider
I, [2018-08-22 14:26:32 +0400#6001] [M: 46996522083840]  INFO -- example_spider: > Starting...
D, [2018-08-22 14:26:32 +0400#6001] [M: 46996522083840] DEBUG -- example_spider: BrowserBuilder (selenium_chrome): created browser instance
D, [2018-08-22 14:26:32 +0400#6001] [M: 46996522083840] DEBUG -- example_spider: BrowserBuilder (selenium_chrome): enabled native headless_mode
I, [2018-08-22 14:26:32 +0400#6001] [M: 46996522083840]  INFO -- example_spider: Browser: started get request to: https://example.com/
I, [2018-08-22 14:26:34 +0400#6001] [M: 46996522083840]  INFO -- example_spider: Browser: finished get request to: https://example.com/
I, [2018-08-22 14:26:34 +0400#6001] [M: 46996522083840]  INFO -- example_spider: Info: visits: requests: 1, responses: 1
D, [2018-08-22 14:26:34 +0400#6001] [M: 46996522083840] DEBUG -- example_spider: Browser: driver.current_memory: 82415
I, [2018-08-22 14:26:34 +0400#6001] [M: 46996522083840]  INFO -- example_spider: > Scraping...
I, [2018-08-22 14:26:34 +0400#6001] [M: 46996522083840]  INFO -- example_spider: Browser: driver selenium_chrome has been destroyed
I, [2018-08-22 14:26:34 +0400#6001] [M: 46996522083840]  INFO -- example_spider: > Stopped!
I, [2018-08-22 14:26:34 +0400#6001] [M: 46996522083840]  INFO -- example_spider: Spider: stopped: {:spider_name=>"example_spider", :status=>:completed, :environment=>"development", :start_time=>2018-08-22 14:26:32 +0400, :stop_time=>2018-08-22 14:26:34 +0400, :running_time=>"1s", :visits=>{:requests=>1, :responses=>1}, :error=>nil}


Inside open_spider and close_spider class methods there is available run_info method which contains useful information about spider state:

    11: def self.open_spider
 => 12:   binding.pry
    13: end

[1] pry(example_spider)> run_info
=> {
  :spider_name=>"example_spider",
  :status=>:running,
  :environment=>"development",
  :start_time=>2018-08-05 23:32:00 +0400,
  :stop_time=>nil,
  :running_time=>nil,
  :visits=>{:requests=>0, :responses=>0},
  :error=>nil
}

Inside close_spider, run_info will be updated:

    15: def self.close_spider
 => 16:   binding.pry
    17: end

[1] pry(example_spider)> run_info
=> {
  :spider_name=>"example_spider",
  :status=>:completed,
  :environment=>"development",
  :start_time=>2018-08-05 23:32:00 +0400,
  :stop_time=>2018-08-05 23:32:06 +0400,
  :running_time=>6.214,
  :visits=>{:requests=>1, :responses=>1},
  :error=>nil
}

run_info[:status] helps to determine if spider was finished successfully or failed (possible values: :completed, :failed):

class ExampleSpider < Kimurai::Base
  @name = "example_spider"
  @engine = :selenium_chrome
  @start_urls = ["https://example.com/"]

  def self.close_spider
    puts ">>> run info: #{run_info}"
  end

  def parse(response, url:, data: {})
    logger.info "> Scraping..."
    # Let's try to strip nil:
    nil.strip
  end
end

Output

I, [2018-08-22 14:34:24 +0400#8459] [M: 47020523644400]  INFO -- example_spider: Spider: started: example_spider
D, [2018-08-22 14:34:25 +0400#8459] [M: 47020523644400] DEBUG -- example_spider: BrowserBuilder (selenium_chrome): created browser instance
D, [2018-08-22 14:34:25 +0400#8459] [M: 47020523644400] DEBUG -- example_spider: BrowserBuilder (selenium_chrome): enabled native headless_mode
I, [2018-08-22 14:34:25 +0400#8459] [M: 47020523644400]  INFO -- example_spider: Browser: started get request to: https://example.com/
I, [2018-08-22 14:34:26 +0400#8459] [M: 47020523644400]  INFO -- example_spider: Browser: finished get request to: https://example.com/
I, [2018-08-22 14:34:26 +0400#8459] [M: 47020523644400]  INFO -- example_spider: Info: visits: requests: 1, responses: 1
D, [2018-08-22 14:34:26 +0400#8459] [M: 47020523644400] DEBUG -- example_spider: Browser: driver.current_memory: 83351
I, [2018-08-22 14:34:26 +0400#8459] [M: 47020523644400]  INFO -- example_spider: > Scraping...
I, [2018-08-22 14:34:26 +0400#8459] [M: 47020523644400]  INFO -- example_spider: Browser: driver selenium_chrome has been destroyed

>>> run info: {:spider_name=>"example_spider", :status=>:failed, :environment=>"development", :start_time=>2018-08-22 14:34:24 +0400, :stop_time=>2018-08-22 14:34:26 +0400, :running_time=>2.01, :visits=>{:requests=>1, :responses=>1}, :error=>"#<NoMethodError: undefined method `strip' for nil:NilClass>"}

F, [2018-08-22 14:34:26 +0400#8459] [M: 47020523644400] FATAL -- example_spider: Spider: stopped: {:spider_name=>"example_spider", :status=>:failed, :environment=>"development", :start_time=>2018-08-22 14:34:24 +0400, :stop_time=>2018-08-22 14:34:26 +0400, :running_time=>"2s", :visits=>{:requests=>1, :responses=>1}, :error=>"#<NoMethodError: undefined method `strip' for nil:NilClass>"}
Traceback (most recent call last):
        6: from example_spider.rb:19:in `<main>'
        5: from /home/victor/code/kimurai/lib/kimurai/base.rb:127:in `crawl!'
        4: from /home/victor/code/kimurai/lib/kimurai/base.rb:127:in `each'
        3: from /home/victor/code/kimurai/lib/kimurai/base.rb:128:in `block in crawl!'
        2: from /home/victor/code/kimurai/lib/kimurai/base.rb:185:in `request_to'
        1: from /home/victor/code/kimurai/lib/kimurai/base.rb:185:in `public_send'
example_spider.rb:15:in `parse': undefined method `strip' for nil:NilClass (NoMethodError)


Usage example: if spider finished successfully, send JSON file with scraped items to a remote FTP location, otherwise (if spider failed), skip incompleted results and send email/notification to slack about it:

Example

Also you can use additional methods completed? or failed?

class Spider < Kimurai::Base
  @engine = :selenium_chrome
  @start_urls = ["https://example.com/"]

  def self.close_spider
    if completed?
      send_file_to_ftp("results.json")
    else
      send_error_notification(run_info[:error])
    end
  end

  def self.send_file_to_ftp(file_path)
    # ...
  end

  def self.send_error_notification(error)
    # ...
  end

  # ...

  def parse_item(response, url:, data: {})
    item = {}
    # ...

    save_to "results.json", item, format: :json
  end
end

KIMURAI_ENV

Kimurai has environments, default is development. To provide custom environment pass KIMURAI_ENV ENV variable before command: $ KIMURAI_ENV=production ruby spider.rb. To access current environment there is Kimurai.env method.

Usage example:

class Spider < Kimurai::Base
  @engine = :selenium_chrome
  @start_urls = ["https://example.com/"]

  def self.close_spider
    if failed? && Kimurai.env == "production"
      send_error_notification(run_info[:error])
    else
      # Do nothing
    end
  end

  # ...
end

Parallel crawling using in_parallel

Kimurai can process web pages concurrently in one single line: in_parallel(:parse_product, urls, threads: 3), where :parse_product is a method to process, urls is array of urls to crawl and threads: is a number of threads:

# amazon_spider.rb
require 'kimurai'

class AmazonSpider < Kimurai::Base
  @name = "amazon_spider"
  @engine = :mechanize
  @start_urls = ["https://www.amazon.com/"]

  def parse(response, url:, data: {})
    browser.fill_in "field-keywords", with: "Web Scraping Books"
    browser.click_on "Go"

    # Walk through pagination and collect products urls:
    urls = []
    loop do
      response = browser.current_response
      response.xpath("//li//a[contains(@class, 's-access-detail-page')]").each do |a|
        urls << a[:href].sub(/ref=.+/, "")
      end

      browser.find(:xpath, "//a[@id='pagnNextLink']", wait: 1).click rescue break
    end

    # Process all collected urls concurrently within 3 threads:
    in_parallel(:parse_book_page, urls, threads: 3)
  end

  def parse_book_page(response, url:, data: {})
    item = {}

    item[:title] = response.xpath("//h1/span[@id]").text.squish
    item[:url] = url
    item[:price] = response.xpath("(//span[contains(@class, 'a-color-price')])[1]").text.squish.presence
    item[:publisher] = response.xpath("//h2[text()='Product details']/following::b[text()='Publisher:']/following-sibling::text()[1]").text.squish.presence

    save_to "books.json", item, format: :pretty_json
  end
end

AmazonSpider.crawl!

Run: $ ruby amazon_spider.rb

I, [2018-08-22 14:48:37 +0400#13033] [M: 46982297486840]  INFO -- amazon_spider: Spider: started: amazon_spider
D, [2018-08-22 14:48:37 +0400#13033] [M: 46982297486840] DEBUG -- amazon_spider: BrowserBuilder (mechanize): created browser instance
I, [2018-08-22 14:48:37 +0400#13033] [M: 46982297486840]  INFO -- amazon_spider: Browser: started get request to: https://www.amazon.com/
I, [2018-08-22 14:48:38 +0400#13033] [M: 46982297486840]  INFO -- amazon_spider: Browser: finished get request to: https://www.amazon.com/
I, [2018-08-22 14:48:38 +0400#13033] [M: 46982297486840]  INFO -- amazon_spider: Info: visits: requests: 1, responses: 1

I, [2018-08-22 14:48:43 +0400#13033] [M: 46982297486840]  INFO -- amazon_spider: Spider: in_parallel: starting processing 52 urls within 3 threads
D, [2018-08-22 14:48:43 +0400#13033] [C: 46982320219020] DEBUG -- amazon_spider: BrowserBuilder (mechanize): created browser instance
I, [2018-08-22 14:48:43 +0400#13033] [C: 46982320219020]  INFO -- amazon_spider: Browser: started get request to: https://www.amazon.com/Practical-Web-Scraping-Data-Science/dp/1484235819/
D, [2018-08-22 14:48:44 +0400#13033] [C: 46982320189640] DEBUG -- amazon_spider: BrowserBuilder (mechanize): created browser instance
I, [2018-08-22 14:48:44 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Browser: started get request to: https://www.amazon.com/Python-Web-Scraping-Cookbook-scraping/dp/1787285219/
D, [2018-08-22 14:48:44 +0400#13033] [C: 46982319187320] DEBUG -- amazon_spider: BrowserBuilder (mechanize): created browser instance
I, [2018-08-22 14:48:44 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Browser: started get request to: https://www.amazon.com/Scraping-Python-Community-Experience-Distilled/dp/1782164367/
I, [2018-08-22 14:48:45 +0400#13033] [C: 46982320219020]  INFO -- amazon_spider: Browser: finished get request to: https://www.amazon.com/Practical-Web-Scraping-Data-Science/dp/1484235819/
I, [2018-08-22 14:48:45 +0400#13033] [C: 46982320219020]  INFO -- amazon_spider: Info: visits: requests: 4, responses: 2
I, [2018-08-22 14:48:45 +0400#13033] [C: 46982320219020]  INFO -- amazon_spider: Browser: started get request to: https://www.amazon.com/Web-Scraping-Python-Collecting-Modern/dp/1491910291/
I, [2018-08-22 14:48:46 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Browser: finished get request to: https://www.amazon.com/Python-Web-Scraping-Cookbook-scraping/dp/1787285219/
I, [2018-08-22 14:48:46 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Info: visits: requests: 5, responses: 3
I, [2018-08-22 14:48:46 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Browser: started get request to: https://www.amazon.com/Web-Scraping-Python-Collecting-Modern/dp/1491985577/
I, [2018-08-22 14:48:46 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Browser: finished get request to: https://www.amazon.com/Scraping-Python-Community-Experience-Distilled/dp/1782164367/
I, [2018-08-22 14:48:46 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Info: visits: requests: 6, responses: 4
I, [2018-08-22 14:48:46 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Browser: started get request to: https://www.amazon.com/Web-Scraping-Excel-Effective-Scrapes-ebook/dp/B01CMMJGZ8/

...

I, [2018-08-22 14:49:10 +0400#13033] [C: 46982320219020]  INFO -- amazon_spider: Info: visits: requests: 51, responses: 49
I, [2018-08-22 14:49:10 +0400#13033] [C: 46982320219020]  INFO -- amazon_spider: Browser: driver mechanize has been destroyed
I, [2018-08-22 14:49:11 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Browser: finished get request to: https://www.amazon.com/Scraping-Ice-Life-Bill-Rayburn-ebook/dp/B00C0NF1L8/
I, [2018-08-22 14:49:11 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Info: visits: requests: 51, responses: 50
I, [2018-08-22 14:49:11 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Browser: started get request to: https://www.amazon.com/Instant-Scraping-Jacob-Ward-2013-07-26/dp/B01FJ1G3G4/
I, [2018-08-22 14:49:11 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Browser: finished get request to: https://www.amazon.com/Php-architects-Guide-Scraping-Author/dp/B010DTKYY4/
I, [2018-08-22 14:49:11 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Info: visits: requests: 52, responses: 51
I, [2018-08-22 14:49:11 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Browser: started get request to: https://www.amazon.com/Ship-Tracking-Maritime-Domain-Awareness/dp/B001J5MTOK/
I, [2018-08-22 14:49:12 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Browser: finished get request to: https://www.amazon.com/Instant-Scraping-Jacob-Ward-2013-07-26/dp/B01FJ1G3G4/
I, [2018-08-22 14:49:12 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Info: visits: requests: 53, responses: 52
I, [2018-08-22 14:49:12 +0400#13033] [C: 46982320189640]  INFO -- amazon_spider: Browser: driver mechanize has been destroyed
I, [2018-08-22 14:49:12 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Browser: finished get request to: https://www.amazon.com/Ship-Tracking-Maritime-Domain-Awareness/dp/B001J5MTOK/
I, [2018-08-22 14:49:12 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Info: visits: requests: 53, responses: 53
I, [2018-08-22 14:49:12 +0400#13033] [C: 46982319187320]  INFO -- amazon_spider: Browser: driver mechanize has been destroyed

I, [2018-08-22 14:49:12 +0400#13033] [M: 46982297486840]  INFO -- amazon_spider: Spider: in_parallel: stopped processing 52 urls within 3 threads, total time: 29s
I, [2018-08-22 14:49:12 +0400#13033] [M: 46982297486840]  INFO -- amazon_spider: Browser: driver mechanize has been destroyed

I, [2018-08-22 14:49:12 +0400#13033] [M: 46982297486840]  INFO -- amazon_spider: Spider: stopped: {:spider_name=>"amazon_spider", :status=>:completed, :environment=>"development", :start_time=>2018-08-22 14:48:37 +0400, :stop_time=>2018-08-22 14:49:12 +0400, :running_time=>"35s", :visits=>{:requests=>53, :responses=>53}, :error=>nil}

books.json

[
  {
    "title": "Web Scraping with Python: Collecting More Data from the Modern Web2nd Edition",
    "url": "https://www.amazon.com/Web-Scraping-Python-Collecting-Modern/dp/1491985577/",
    "price": "$26.94",
    "publisher": "O'Reilly Media; 2 edition (April 14, 2018)",
    "position": 1
  },
  {
    "title": "Python Web Scraping Cookbook: Over 90 proven recipes to get you scraping with Python, micro services, Docker and AWS",
    "url": "https://www.amazon.com/Python-Web-Scraping-Cookbook-scraping/dp/1787285219/",
    "price": "$39.99",
    "publisher": "Packt Publishing - ebooks Account (February 9, 2018)",
    "position": 2
  },
  {
    "title": "Web Scraping with Python: Collecting Data from the Modern Web1st Edition",
    "url": "https://www.amazon.com/Web-Scraping-Python-Collecting-Modern/dp/1491910291/",
    "price": "$15.75",
    "publisher": "O'Reilly Media; 1 edition (July 24, 2015)",
    "position": 3
  },

  ...

  {
    "title": "Instant Web Scraping with Java by Ryan Mitchell (2013-08-26)",
    "url": "https://www.amazon.com/Instant-Scraping-Java-Mitchell-2013-08-26/dp/B01FEM76X2/",
    "price": "$35.82",
    "publisher": "Packt Publishing (2013-08-26) (1896)",
    "position": 52
  }
]


Note that save_to and unique? helpers are thread-safe (protected by Mutex) and can be freely used inside threads.

in_parallel can take additional options:

  • data: pass with urls custom data hash: in_parallel(:method, urls, threads: 3, data: { category: "Scraping" })
  • delay: set delay between requests: in_parallel(:method, urls, threads: 3, delay: 2). Delay can be Integer, Float or Range (2..5). In case of a Range, delay number will be chosen randomly for each request: rand (2..5) # => 3
  • engine: set custom engine than a default one: in_parallel(:method, urls, threads: 3, engine: :poltergeist_phantomjs)
  • config: pass custom options to config (see config section)

Active Support included

You can use all the power of familiar Rails core-ext methods for scraping inside Kimurai. Especially take a look at squish, truncate_words, titleize, remove, present? and presence.

Schedule spiders using Cron

1) Inside spider directory generate Whenever config: $ kimurai generate schedule.

schedule.rb

### Settings ###
require 'tzinfo'

# Export current PATH to the cron
env :PATH, ENV["PATH"]

# Use 24 hour format when using `at:` option
set :chronic_options, hours24: true

# Use local_to_utc helper to setup execution time using your local timezone instead
# of server's timezone (which is probably and should be UTC, to check run `$ timedatectl`).
# Also maybe you'll want to set same timezone in kimurai as well (use `Kimurai.configuration.time_zone =` for that),
# to have spiders logs in a specific time zone format.
# Example usage of helper:
# every 1.day, at: local_to_utc("7:00", zone: "Europe/Moscow") do
#   crawl "google_spider.com", output: "log/google_spider.com.log"
# end
def local_to_utc(time_string, zone:)
  TZInfo::Timezone.get(zone).local_to_utc(Time.parse(time_string))
end

# Note: by default Whenever exports cron commands with :environment == "production".
# Note: Whenever can only append log data to a log file (>>). If you want
# to overwrite (>) log file before each run, pass lambda:
# crawl "google_spider.com", output: -> { "> log/google_spider.com.log 2>&1" }

# Project job types
job_type :crawl,  "cd :path && KIMURAI_ENV=:environment bundle exec kimurai crawl :task :output"
job_type :runner, "cd :path && KIMURAI_ENV=:environment bundle exec kimurai runner --jobs :task :output"

# Single file job type
job_type :single, "cd :path && KIMURAI_ENV=:environment ruby :task :output"
# Single with bundle exec
job_type :single_bundle, "cd :path && KIMURAI_ENV=:environment bundle exec ruby :task :output"

### Schedule ###
# Usage (check examples here https://github.com/javan/whenever#example-schedulerb-file):
# every 1.day do
  # Example to schedule a single spider in the project:
  # crawl "google_spider.com", output: "log/google_spider.com.log"

  # Example to schedule all spiders in the project using runner. Each spider will write
  # it's own output to the `log/spider_name.log` file (handled by a runner itself).
  # Runner output will be written to log/runner.log file.
  # Argument number it's a count of concurrent jobs:
  # runner 3, output:"log/runner.log"

  # Example to schedule single spider (without project):
  # single "single_spider.rb", output: "single_spider.log"
# end

### How to set a cron schedule ###
# Run: `$ whenever --update-crontab --load-file config/schedule.rb`.
# If you don't have whenever command, install the gem: `$ gem install whenever`.

### How to cancel a schedule ###
# Run: `$ whenever --clear-crontab --load-file config/schedule.rb`.


2) Add at the bottom of schedule.rb following code:

every 1.day, at: "7:00" do
  single "example_spider.rb", output: "example_spider.log"
end

3) Run: $ whenever --update-crontab --load-file schedule.rb. Done!

You can check Whenever examples here. To cancel schedule, run: $ whenever --clear-crontab --load-file schedule.rb.

Configuration options

You can configure several options using configure block:

Kimurai.configure do |config|
  # Default logger has colored mode in development.
  # If you would like to disable it, set `colorize_logger` to false.
  # config.colorize_logger = false

  # Logger level for default logger:
  # config.log_level = :info

  # Custom logger:
  # config.logger = Logger.new(STDOUT)

  # Custom time zone (for logs):
  # config.time_zone = "UTC"
  # config.time_zone = "Europe/Moscow"

  # Provide custom chrome binary path (default is any available chrome/chromium in the PATH):
  # config.selenium_chrome_path = "/usr/bin/chromium-browser"
  # Provide custom selenium chromedriver path (default is "/usr/local/bin/chromedriver"):
  # config.chromedriver_path = "~/.local/bin/chromedriver"
end

Using Kimurai inside existing Ruby application

You can integrate Kimurai spiders (which are just Ruby classes) to an existing Ruby application like Rails or Sinatra, and run them using background jobs (for example). Check the following info to understand the running process of spiders:

.crawl! method

.crawl! (class method) performs a full run of a particular spider. This method will return run_info if run was successful, or an exception if something went wrong.

class ExampleSpider < Kimurai::Base
  @name = "example_spider"
  @engine = :mechanize
  @start_urls = ["https://example.com/"]

  def parse(response, url:, data: {})
    title = response.xpath("//title").text.squish
  end
end

ExampleSpider.crawl!
# => { :spider_name => "example_spider", :status => :completed, :environment => "development", :start_time => 2018-08-22 18:20:16 +0400, :stop_time => 2018-08-22 18:20:17 +0400, :running_time => 1.216, :visits => { :requests => 1, :responses => 1 }, :items => { :sent => 0, :processed => 0 }, :error => nil }

You can't .crawl! spider in different thread if it still running (because spider instances store some shared data in the @run_info class variable while crawling):

2.times do |i|
  Thread.new { p i, ExampleSpider.crawl! }
end # =>

# 1
# false

# 0
# {:spider_name=>"example_spider", :status=>:completed, :environment=>"development", :start_time=>2018-08-22 18:49:22 +0400, :stop_time=>2018-08-22 18:49:23 +0400, :running_time=>0.801, :visits=>{:requests=>1, :responses=>1}, :items=>{:sent=>0, :processed=>0}, :error=>nil}

So what if you're don't care about stats and just want to process request to a particular spider method and get the returning value from this method? Use .parse! instead:

.parse!(:method_name, url:) method

.parse! (class method) creates a new spider instance and performs a request to given method with a given url. Value from the method will be returned back:

class ExampleSpider < Kimurai::Base
  @name = "example_spider"
  @engine = :mechanize
  @start_urls = ["https://example.com/"]

  def parse(response, url:, data: {})
    title = response.xpath("//title").text.squish
  end
end

ExampleSpider.parse!(:parse, url: "https://example.com/")
# => "Example Domain"

Like .crawl!, .parse! method takes care of a browser instance and kills it (browser.destroy_driver!) before returning the value. Unlike .crawl!, .parse! method can be called from different threads at the same time:

urls = ["https://www.google.com/", "https://www.reddit.com/", "https://en.wikipedia.org/"]

urls.each do |url|
  Thread.new { p ExampleSpider.parse!(:parse, url: url) }
end # =>

# "Google"
# "Wikipedia, the free encyclopedia"
# "reddit: the front page of the internetHotHot"

Keep in mind, that save_to and unique? helpers are not thread-safe while using .parse! method.

Kimurai.list and Kimurai.find_by_name()

class GoogleSpider < Kimurai::Base
  @name = "google_spider"
end

class RedditSpider < Kimurai::Base
  @name = "reddit_spider"
end

class WikipediaSpider < Kimurai::Base
  @name = "wikipedia_spider"
end

# To get the list of all available spider classes:
Kimurai.list
# => {"google_spider"=>GoogleSpider, "reddit_spider"=>RedditSpider, "wikipedia_spider"=>WikipediaSpider}

# To find a particular spider class by it's name:
Kimurai.find_by_name("reddit_spider")
# => RedditSpider

Automated sever setup and deployment

EXPERIMENTAL

Setup

You can automatically setup required environment for Kimurai on the remote server (currently there is only Ubuntu Server 18.04 support) using $ kimurai setup command. setup will perform installation of: latest Ruby with Rbenv, browsers with webdrivers and in additional databases clients (only clients) for MySQL, Postgres and MongoDB (so you can connect to a remote database from ruby).

To perform remote server setup, Ansible is required on the desktop machine (to install: Ubuntu: $ sudo apt install ansible, Mac OS X: $ brew install ansible)

It's recommended to use regular user to setup the server, not root. To create a new user, login to the server $ ssh root@your_server_ip, type $ adduser username to create a user, and $ gpasswd -a username sudo to add new user to a sudo group.

Example:

$ kimurai setup [email protected] --ask-sudo --ssh-key-path path/to/private_key

CLI options:

  • --ask-sudo pass this option to ask sudo (user) password for system-wide installation of packages (apt install)
  • --ssh-key-path path/to/private_key authorization on the server using private ssh key. You can omit it if required key already added to keychain on your desktop (Ansible uses SSH agent forwarding)
  • --ask-auth-pass authorization on the server using user password, alternative option to --ssh-key-path.
  • -p port_number custom port for ssh connection (-p 2222)

You can check setup playbook here

Deploy

After successful setup you can deploy a spider to the remote server using $ kimurai deploy command. On each deploy there are performing several tasks: 1) pull repo from a remote origin to ~/repo_name user directory 2) run bundle install 3) Update crontab whenever --update-crontab (to update spider schedule from schedule.rb file).

Before deploy make sure that inside spider directory you have: 1) git repository with remote origin (bitbucket, github, etc.) 2) Gemfile 3) schedule.rb inside subfolder config (config/schedule.rb).

Example:

$ kimurai deploy [email protected] --ssh-key-path path/to/private_key --repo-key-path path/to/repo_private_key

CLI options: same like for setup command (except --ask-sudo), plus

  • --repo-url provide custom repo url (--repo-url [email protected]:username/repo_name.git), otherwise current origin/master will be taken (output from $ git remote get-url origin)
  • --repo-key-path if git repository is private, authorization is required to pull the code on the remote server. Use this option to provide a private repository SSH key. You can omit it if required key already added to keychain on your desktop (same like with --ssh-key-path option)

You can check deploy playbook here

Spider @config

Using @config you can set several options for a spider, like proxy, user-agent, default cookies/headers, delay between requests, browser memory control and so on:

class Spider < Kimurai::Base
  USER_AGENTS = ["Chrome", "Firefox", "Safari", "Opera"]
  PROXIES = ["2.3.4.5:8080:http:username:password", "3.4.5.6:3128:http", "1.2.3.4:3000:socks5"]

  @engine = :poltergeist_phantomjs
  @start_urls = ["https://example.com/"]
  @config = {
    headers: { "custom_header" => "custom_value" },
    cookies: [{ name: "cookie_name", value: "cookie_value", domain: ".example.com" }],
    user_agent: -> { USER_AGENTS.sample },
    proxy: -> { PROXIES.sample },
    window_size: [1366, 768],
    disable_images: true,
    restart_if: {
      # Restart browser if provided memory limit (in kilobytes) is exceeded:
      memory_limit: 350_000
    },
    before_request: {
      # Change user agent before each request:
      change_user_agent: true,
      # Change proxy before each request:
      change_proxy: true,
      # Clear all cookies and set default cookies (if provided) before each request:
      clear_and_set_cookies: true,
      # Process delay before each request:
      delay: 1..3
    }
  }

  def parse(response, url:, data: {})
    # ...
  end
end

All available @config options

@config = {
  # Custom headers, format: hash. Example: { "some header" => "some value", "another header" => "another value" }
  # Works only for :mechanize and :poltergeist_phantomjs engines (Selenium doesn't allow to set/get headers)
  headers: {},

  # Custom User Agent, format: string or lambda.
  # Use lambda if you want to rotate user agents before each run:
  # user_agent: -> { ARRAY_OF_USER_AGENTS.sample }
  # Works for all engines
  user_agent: "Mozilla/5.0 Firefox/61.0",

  # Custom cookies, format: array of hashes.
  # Format for a single cookie: { name: "cookie name", value: "cookie value", domain: ".example.com" }
  # Works for all engines
  cookies: [],

  # Proxy, format: string or lambda. Format of a proxy string: "ip:port:protocol:user:password"
  # `protocol` can be http or socks5. User and password are optional.
  # Use lambda if you want to rotate proxies before each run:
  # proxy: -> { ARRAY_OF_PROXIES.sample }
  # Works for all engines, but keep in mind that Selenium drivers doesn't support proxies
  # with authorization. Also, Mechanize doesn't support socks5 proxy format (only http)
  proxy: "3.4.5.6:3128:http:user:pass",

  # If enabled, browser will ignore any https errors. It's handy while using a proxy
  # with self-signed SSL cert (for example Crawlera or Mitmproxy)
  # Also, it will allow to visit webpages with expires SSL certificate.
  # Works for all engines
  ignore_ssl_errors: true,

  # Custom window size, works for all engines
  window_size: [1366, 768],

  # Skip images downloading if true, works for all engines
  disable_images: true,

  # Selenium engines only: headless mode, `:native` or `:virtual_display` (default is :native)
  # Although native mode has a better performance, virtual display mode
  # sometimes can be useful. For example, some websites can detect (and block)
  # headless chrome, so you can use virtual_display mode instead
  headless_mode: :native,

  # This option tells the browser not to use a proxy for the provided list of domains or IP addresses.
  # Format: array of strings. Works only for :selenium_firefox and selenium_chrome
  proxy_bypass_list: [],

  # Option to provide custom SSL certificate. Works only for :poltergeist_phantomjs and :mechanize
  ssl_cert_path: "path/to/ssl_cert",

  # Inject some JavaScript code to the browser.
  # Format: array of strings, where each string is a path to JS file.
  # Works only for poltergeist_phantomjs engine (Selenium doesn't support JS code injection)
  extensions: ["lib/code_to_inject.js"],

  # Automatically skip duplicated (already visited) urls when using `request_to` method.
  # Possible values: `true` or `hash` with options.
  # In case of `true`, all visited urls will be added to the storage's scope `:requests_urls`
  # and if url already contains in this scope, request will be skipped.
  # You can configure this setting by providing additional options as hash:
  # `skip_duplicate_requests: { scope: :custom_scope, check_only: true }`, where:
  # `scope:` - use custom scope than `:requests_urls`
  # `check_only:` - if true, then scope will be only checked for url, url will not
  # be added to the scope if scope doesn't contains it.
  # works for all drivers
  skip_duplicate_requests: true,

  # Automatically skip provided errors while requesting a page.
  # If raised error matches one of the errors in the list, then this error will be caught,
  # and request will be skipped.
  # It is a good idea to skip errors like NotFound(404), etc.
  # Format: array where elements are error classes or/and hashes. You can use hash format
  # for more flexibility: `{ error: "RuntimeError", message: "404 => Net::HTTPNotFound" }`.
  # Provided `message:` will be compared with a full error message using `String#include?`. Also
  # you can use regex instead: `{ error: "RuntimeError", message: /404|403/ }`.
  skip_request_errors: [{ error: RuntimeError, message: "404 => Net::HTTPNotFound" }],

  # Automatically retry provided errors with a few attempts while requesting a page.
  # If raised error matches one of the errors in the list, then this error will be caught
  # and the request will be processed again within a delay. There are 3 attempts:
  # first: delay 15 sec, second: delay 30 sec, third: delay 45 sec.
  # If after 3 attempts there is still an exception, then the exception will be raised.
  # It is a good idea to try to retry errros like `ReadTimeout`, `HTTPBadGateway`, etc.
  # Format: same like for `skip_request_errors` option.
  retry_request_errors: [Net::ReadTimeout],

  # Handle page encoding while parsing html response using Nokogiri. There are two modes:
  # Auto (`:auto`) (try to fetch correct encoding from <meta http-equiv="Content-Type"> or <meta charset> tags)
  # Set required encoding manually, example: `encoding: "GB2312"` (Set required encoding manually)
  # Default this option is unset.
  encoding: nil,

  # Restart browser if one of the options is true:
  restart_if: {
    # Restart browser if provided memory limit (in kilobytes) is exceeded (works for all engines)
    memory_limit: 350_000,

    # Restart browser if provided requests limit is exceeded (works for all engines)
    requests_limit: 100
  },

  # Perform several actions before each request:
  before_request: {
    # Change proxy before each request. The `proxy:` option above should be presented
    # and has lambda format. Works only for poltergeist and mechanize engines
    # (Selenium doesn't support proxy rotation).
    change_proxy: true,

    # Change user agent before each request. The `user_agent:` option above should be presented
    # and has lambda format. Works only for poltergeist and mechanize engines
    # (selenium doesn't support to get/set headers).
    change_user_agent: true,

    # Clear all cookies before each request, works for all engines
    clear_cookies: true,

    # If you want to clear all cookies + set custom cookies (`cookies:` option above should be presented)
    # use this option instead (works for all engines)
    clear_and_set_cookies: true,

    # Global option to set delay between requests.
    # Delay can be `Integer`, `Float` or `Range` (`2..5`). In case of a range,
    # delay number will be chosen randomly for each request: `rand (2..5) # => 3`
    delay: 1..3
  }
}

As you can see, most of the options are universal for any engine.

@config settings inheritance

Settings can be inherited:

class ApplicationSpider < Kimurai::Base
  @engine = :poltergeist_phantomjs
  @config = {
    user_agent: "Firefox",
    disable_images: true,
    restart_if: { memory_limit: 350_000 },
    before_request: { delay: 1..2 }
  }
end

class CustomSpider < ApplicationSpider
  @name = "custom_spider"
  @start_urls = ["https://example.com/"]
  @config = {
    before_request: { delay: 4..6 }
  }

  def parse(response, url:, data: {})
    # ...
  end
end

Here, @config of CustomSpider will be deep merged with ApplicationSpider config, so CustomSpider will keep all inherited options with only delay updated.

Project mode

Kimurai can work in project mode (Like Scrapy). To generate a new project, run: $ kimurai generate project web_spiders (where web_spiders is a name of project).

Structure of the project:

.
├── config/
│   ├── initializers/
│   ├── application.rb
│   ├── automation.yml
│   ├── boot.rb
│   └── schedule.rb
├── spiders/
│   └── application_spider.rb
├── db/
├── helpers/
│   └── application_helper.rb
├── lib/
├── log/
├── pipelines/
│   ├── validator.rb
│   └── saver.rb
├── tmp/
├── .env
├── Gemfile
├── Gemfile.lock
└── README.md

Description

  • config/ folder for configutation files
    • config/initializers Rails-like initializers to load custom code at start of framework
    • config/application.rb configuration settings for Kimurai (Kimurai.configure do block)
    • config/automation.yml specify some settings for setup and deploy
    • config/boot.rb loads framework and project
    • config/schedule.rb Cron schedule for spiders
  • spiders/ folder for spiders
    • spiders/application_spider.rb Base parent class for all spiders
  • db/ store here all database files (sqlite, json, csv, etc.)
  • helpers/ Rails-like helpers for spiders
    • helpers/application_helper.rb all methods inside ApplicationHelper module will be available for all spiders
  • lib/ put here custom Ruby code
  • log/ folder for logs
  • pipelines/ folder for Scrapy-like pipelines. One file = one pipeline
    • pipelines/validator.rb example pipeline to validate item
    • pipelines/saver.rb example pipeline to save item
  • tmp/ folder for temp. files
  • .env file to store ENV variables for project and load them using Dotenv
  • Gemfile dependency file
  • Readme.md example project readme

Generate new spider

To generate a new spider in the project, run:

$ kimurai generate spider example_spider
      create  spiders/example_spider.rb

Command will generate a new spider class inherited from ApplicationSpider:

class ExampleSpider < ApplicationSpider
  @name = "example_spider"
  @start_urls = []
  @config = {}

  def parse(response, url:, data: {})
  end
end

Crawl

To run a particular spider in the project, run: $ bundle exec kimurai crawl example_spider. Don't forget to add bundle exec before command to load required environment.

List

To list all project spiders, run: $ bundle exec kimurai list

Parse

For project spiders you can use $ kimurai parse command which helps to debug spiders:

$ bundle exec kimurai parse example_spider parse_product --url https://example-shop.com/product-1

where example_spider is a spider to run, parse_product is a spider method to process and --url is url to open inside processing method.

Pipelines, send_item method

You can use item pipelines to organize and store in one place item processing logic for all project spiders (also check Scrapy description of pipelines).

Imagine if you have three spiders where each of them crawls different e-commerce shop and saves only shoe positions. For each spider, you want to save items only with "shoe" category, unique sku, valid title/price and with existing images. To avoid code duplication between spiders, use pipelines:

Example

pipelines/validator.rb

class Validator < Kimurai::Pipeline
  def process_item(item, options: {})
    # Here you can validate item and raise `DropItemError`
    # if one of the validations failed. Examples:

    # Drop item if it's category is not "shoe":
    if item[:category] != "shoe"
      raise DropItemError, "Wrong item category"
    end

    # Check item sku for uniqueness using buit-in unique? helper:
    unless unique?(:sku, item[:sku])
      raise DropItemError, "Item sku is not unique"
    end

    # Drop item if title length shorter than 5 symbols:
    if item[:title].size < 5
      raise DropItemError, "Item title is short"
    end

    # Drop item if price is not present
    unless item[:price].present?
      raise DropItemError, "item price is not present"
    end

    # Drop item if it doesn't contains any images:
    unless item[:images].present?
      raise DropItemError, "Item images are not present"
    end

    # Pass item to the next pipeline (if it wasn't dropped):
    item
  end
end

pipelines/saver.rb

class Saver < Kimurai::Pipeline
  def process_item(item, options: {})
    # Here you can save item to the database, send it to a remote API or
    # simply save item to a file format using `save_to` helper:

    # To get the name of current spider: `spider.class.name`
    save_to "db/#{spider.class.name}.json", item, format: :json

    item
  end
end

spiders/application_spider.rb

class ApplicationSpider < Kimurai::Base
  @engine = :selenium_chrome
  # Define pipelines (by order) for all spiders:
  @pipelines = [:validator, :saver]
end

spiders/shop_spider_1.rb

class ShopSpiderOne < ApplicationSpider
  @name = "shop_spider_1"
  @start_urls = ["https://shop-1.com"]

  # ...

  def parse_product(response, url:, data: {})
    # ...

    # Send item to pipelines:
    send_item item
  end
end

spiders/shop_spider_2.rb

class ShopSpiderTwo < ApplicationSpider
  @name = "shop_spider_2"
  @start_urls = ["https://shop-2.com"]

  def parse_product(response, url:, data: {})
    # ...

    # Send item to pipelines:
    send_item item
  end
end

spiders/shop_spider_3.rb

class ShopSpiderThree < ApplicationSpider
  @name = "shop_spider_3"
  @start_urls = ["https://shop-3.com"]

  def parse_product(response, url:, data: {})
    # ...

    # Send item to pipelines:
    send_item item
  end
end


When you start using pipelines, there are stats for items appears:

Example pipelines/validator.rb ```ruby class Validator < Kimurai::Pipeline def process_item(item, options: {}) if item[:star_count] < 10 raise DropItemError, "Repository doesn't have enough stars" end item end end ``` spiders/github_spider.rb ```ruby class GithubSpider < ApplicationSpider @name = "github_spider" @engine = :selenium_chrome @pipelines = [:validator] @start_urls = ["https://github.com/search?q=Ruby%20Web%20Scraping"] @config = { user_agent: "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/68.0.3440.84 Safari/537.36", before_request: { delay: 4..7 } } def parse(response, url:, data: {}) response.xpath("//ul[@class='repo-list']/div//h3/a").each do |a| request_to :parse_repo_page, url: absolute_url(a[:href], base: url) end if next_page = response.at_xpath("//a[@class='next_page']") request_to :parse, url: absolute_url(next_page[:href], base: url) end end def parse_repo_page(response, url:, data: {}) item = {} item[:owner] = response.xpath("//h1//a[@rel='author']").text item[:repo_name] = response.xpath("//h1/strong[@itemprop='name']/a").text item[:repo_url] = url item[:description] = response.xpath("//span[@itemprop='about']").text.squish item[:tags] = response.xpath("//div[@id='topics-list-container']/div/a").map { |a| a.text.squish } item[:watch_count] = response.xpath("//ul[@class='pagehead-actions']/li[contains(., 'Watch')]/a[2]").text.squish.delete(",").to_i item[:star_count] = response.xpath("//ul[@class='pagehead-actions']/li[contains(., 'Star')]/a[2]").text.squish.delete(",").to_i item[:fork_count] = response.xpath("//ul[@class='pagehead-actions']/li[contains(., 'Fork')]/a[2]").text.squish.delete(",").to_i item[:last_commit] = response.xpath("//span[@itemprop='dateModified']/*").text send_item item end end ``` ``` $ bundle exec kimurai crawl github_spider I, [2018-08-22 15:56:35 +0400#1358] [M: 47347279209980] INFO -- github_spider: Spider: started: github_spider D, [2018-08-22 15:56:35 +0400#1358] [M: 47347279209980] DEBUG -- github_spider: BrowserBuilder (selenium_chrome): created browser instance I, [2018-08-22 15:56:40 +0400#1358] [M: 47347279209980] INFO -- github_spider: Browser: started get request to: https://github.com/search?q=Ruby%20Web%20Scraping I, [2018-08-22 15:56:44 +0400#1358] [M: 47347279209980] INFO -- github_spider: Browser: finished get request to: https://github.com/search?q=Ruby%20Web%20Scraping I, [2018-08-22 15:56:44 +0400#1358] [M: 47347279209980] INFO -- github_spider: Info: visits: requests: 1, responses: 1 D, [2018-08-22 15:56:44 +0400#1358] [M: 47347279209980] DEBUG -- github_spider: Browser: driver.current_memory: 116182 D, [2018-08-22 15:56:44 +0400#1358] [M: 47347279209980] DEBUG -- github_spider: Browser: sleep 5 seconds before request... I, [2018-08-22 15:56:49 +0400#1358] [M: 47347279209980] INFO -- github_spider: Browser: started get request to: https://github.com/lorien/awesome-web-scraping I, [2018-08-22 15:56:50 +0400#1358] [M: 47347279209980] INFO -- github_spider: Browser: finished get request to: https://github.com/lorien/awesome-web-scraping I, [2018-08-22 15:56:50 +0400#1358] [M: 47347279209980] INFO -- github_spider: Info: visits: requests: 2, responses: 2 D, [2018-08-22 15:56:50 +0400#1358] [M: 47347279209980] DEBUG -- github_spider: Browser: driver.current_memory: 217432 D, [2018-08-22 15:56:50 +0400#1358] [M: 47347279209980] DEBUG -- github_spider: Pipeline: starting processing item through 1 pipeline... I, [2018-08-22 15:56:50 +0400#1358] [M: 47347279209980] INFO -- github_spider: Pipeline: processed: of libraries, tools and APIs for web scraping and data processing.","tags":["awesome","awesome-list","web-scraping","data-processing","python","javascript","php","ruby"],"watch_count":159,"star_count":2423,"fork_count":358,"last_commit":"4 days ago" I, [2018-08-22 15:56:50 +0400#1358] [M: 47347279209980] INFO -- github_spider: Info: items: sent: 1, processed: 1 D, [2018-08-22 15:56:50 +0400#1358] [M: 47347279209980] DEBUG -- github_spider: Browser: sleep 6 seconds before request... ... I, [2018-08-22 16:11:50 +0400#1358] [M: 47347279209980] INFO -- github_spider: Browser: started get request to: https://github.com/preston/idclight I, [2018-08-22 16:11:51 +0400#1358] [M: 47347279209980] INFO -- github_spider: Browser: finished get request to: https://github.com/preston/idclight I, [2018-08-22 16:11:51 +0400#1358] [M: 47347279209980] INFO -- github_spider: Info: visits: requests: 140, responses: 140 D, [2018-08-22 16:11:51 +0400#1358] [M: 47347279209980] DEBUG -- github_spider: Browser: driver.current_memory: 211713 D, [2018-08-22 16:11:51 +0400#1358] [M: 47347279209980] DEBUG -- github_spider: Pipeline: starting processing item through 1 pipeline... E, [2018-08-22 16:11:51 +0400#1358] [M: 47347279209980] ERROR -- github_spider: Pipeline: dropped: #, item: :repo_name=>"idclight", :repo_url=>"https://github.com/preston/idclight", :description=>"A Ruby gem for accessing the freely available IDClight (IDConverter Light) web service, which convert between different types of gene IDs such as Hugo and Entrez. Queries are screen scraped from http://idclight.bioinfo.cnio.es.", :tags=>[], :watch_count=>6, :star_count=>1, :fork_count=>0, :last_commit=>"on Apr 12, 2012" I, [2018-08-22 16:11:51 +0400#1358] [M: 47347279209980] INFO -- github_spider: Info: items: sent: 127, processed: 12 I, [2018-08-22 16:11:51 +0400#1358] [M: 47347279209980] INFO -- github_spider: Browser: driver selenium_chrome has been destroyed I, [2018-08-22 16:11:51 +0400#1358] [M: 47347279209980] INFO -- github_spider: Spider: stopped: :status=>:completed, :environment=>"development", :start_time=>2018-08-22 15:56:35 +0400, :stop_time=>2018-08-22 16:11:51 +0400, :running_time=>"15m, 16s", :visits=>{:requests=>140, :responses=>140, :items=>:processed=>12, :error=>nil} ```

Also, you can pass custom options to pipeline from a particular spider if you want to change pipeline behavior for this spider:
Example spiders/custom_spider.rb ```ruby class CustomSpider < ApplicationSpider @name = "custom_spider" @start_urls = ["https://example.com"] @pipelines = [:validator] # ... def parse_item(response, url:, data: {}) # ... # Pass custom option `skip_uniq_checking` for Validator pipeline: send_item item, validator: { skip_uniq_checking: true } end end ``` pipelines/validator.rb ```ruby class Validator < Kimurai::Pipeline def process_item(item, options: {}) # Do not check item sku for uniqueness if options[:skip_uniq_checking] is true if options[:skip_uniq_checking] != true raise DropItemError, "Item sku is not unique" unless unique?(:sku, item[:sku]) end end end ```

Runner

You can run project spiders one by one or in parallel using $ kimurai runner command:

$ bundle exec kimurai list
custom_spider
example_spider
github_spider

$ bundle exec kimurai runner -j 3
>>> Runner: started: {:id=>1533727423, :status=>:processing, :start_time=>2018-08-08 15:23:43 +0400, :stop_time=>nil, :environment=>"development", :concurrent_jobs=>3, :spiders=>["custom_spider", "github_spider", "example_spider"]}
> Runner: started spider: custom_spider, index: 0
> Runner: started spider: github_spider, index: 1
> Runner: started spider: example_spider, index: 2
< Runner: stopped spider: custom_spider, index: 0
< Runner: stopped spider: example_spider, index: 2
< Runner: stopped spider: github_spider, index: 1
<<< Runner: stopped: {:id=>1533727423, :status=>:completed, :start_time=>2018-08-08 15:23:43 +0400, :stop_time=>2018-08-08 15:25:11 +0400, :environment=>"development", :concurrent_jobs=>3, :spiders=>["custom_spider", "github_spider", "example_spider"]}

Each spider runs in a separate process. Spiders logs available at log/ folder. Pass -j option to specify how many spiders should be processed at the same time (default is 1).

You can provide additional arguments like --include or --exclude to specify which spiders to run:

# Run only custom_spider and example_spider:
$ bundle exec kimurai runner --include custom_spider example_spider

# Run all except github_spider:
$ bundle exec kimurai runner --exclude github_spider

Runner callbacks

You can perform custom actions before runner starts and after runner stops using config.runner_at_start_callback and config.runner_at_stop_callback. Check config/application.rb to see example.

Chat Support and Feedback

Will be updated

License

The gem is available as open source under the terms of the MIT License.