wayfarer 0.0.3
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +8 -0
- data/.rbenv-gemsets +1 -0
- data/.rspec +3 -0
- data/.rubocop.yml +21 -0
- data/.ruby-version +1 -0
- data/.travis.yml +5 -0
- data/.yardopts +3 -0
- data/Changelog.md +10 -0
- data/Gemfile +11 -0
- data/LICENSE +19 -0
- data/README.md +21 -0
- data/Rakefile +114 -0
- data/benchmark/frontiers.rb +143 -0
- data/bin/wayfarer +116 -0
- data/docs/.gitignore +2 -0
- data/docs/_config.yml +15 -0
- data/docs/_includes/base.html +7 -0
- data/docs/_includes/head.html +10 -0
- data/docs/_includes/navigation.html +187 -0
- data/docs/_layouts/default.html +42 -0
- data/docs/_sass/base.scss +439 -0
- data/docs/_sass/variables.scss +24 -0
- data/docs/_sass/vendor/bourbon/_bourbon-deprecate.scss +19 -0
- data/docs/_sass/vendor/bourbon/_bourbon-deprecated-upcoming.scss +425 -0
- data/docs/_sass/vendor/bourbon/_bourbon.scss +90 -0
- data/docs/_sass/vendor/bourbon/addons/_border-color.scss +29 -0
- data/docs/_sass/vendor/bourbon/addons/_border-radius.scss +48 -0
- data/docs/_sass/vendor/bourbon/addons/_border-style.scss +28 -0
- data/docs/_sass/vendor/bourbon/addons/_border-width.scss +28 -0
- data/docs/_sass/vendor/bourbon/addons/_buttons.scss +69 -0
- data/docs/_sass/vendor/bourbon/addons/_clearfix.scss +25 -0
- data/docs/_sass/vendor/bourbon/addons/_ellipsis.scss +30 -0
- data/docs/_sass/vendor/bourbon/addons/_font-stacks.scss +31 -0
- data/docs/_sass/vendor/bourbon/addons/_hide-text.scss +27 -0
- data/docs/_sass/vendor/bourbon/addons/_margin.scss +29 -0
- data/docs/_sass/vendor/bourbon/addons/_padding.scss +29 -0
- data/docs/_sass/vendor/bourbon/addons/_position.scss +51 -0
- data/docs/_sass/vendor/bourbon/addons/_prefixer.scss +66 -0
- data/docs/_sass/vendor/bourbon/addons/_retina-image.scss +27 -0
- data/docs/_sass/vendor/bourbon/addons/_size.scss +56 -0
- data/docs/_sass/vendor/bourbon/addons/_text-inputs.scss +118 -0
- data/docs/_sass/vendor/bourbon/addons/_timing-functions.scss +34 -0
- data/docs/_sass/vendor/bourbon/addons/_triangle.scss +63 -0
- data/docs/_sass/vendor/bourbon/addons/_word-wrap.scss +29 -0
- data/docs/_sass/vendor/bourbon/css3/_animation.scss +61 -0
- data/docs/_sass/vendor/bourbon/css3/_appearance.scss +5 -0
- data/docs/_sass/vendor/bourbon/css3/_backface-visibility.scss +5 -0
- data/docs/_sass/vendor/bourbon/css3/_background-image.scss +44 -0
- data/docs/_sass/vendor/bourbon/css3/_background.scss +57 -0
- data/docs/_sass/vendor/bourbon/css3/_border-image.scss +61 -0
- data/docs/_sass/vendor/bourbon/css3/_calc.scss +6 -0
- data/docs/_sass/vendor/bourbon/css3/_columns.scss +67 -0
- data/docs/_sass/vendor/bourbon/css3/_filter.scss +6 -0
- data/docs/_sass/vendor/bourbon/css3/_flex-box.scss +327 -0
- data/docs/_sass/vendor/bourbon/css3/_font-face.scss +29 -0
- data/docs/_sass/vendor/bourbon/css3/_font-feature-settings.scss +6 -0
- data/docs/_sass/vendor/bourbon/css3/_hidpi-media-query.scss +12 -0
- data/docs/_sass/vendor/bourbon/css3/_hyphens.scss +6 -0
- data/docs/_sass/vendor/bourbon/css3/_image-rendering.scss +15 -0
- data/docs/_sass/vendor/bourbon/css3/_keyframes.scss +38 -0
- data/docs/_sass/vendor/bourbon/css3/_linear-gradient.scss +40 -0
- data/docs/_sass/vendor/bourbon/css3/_perspective.scss +12 -0
- data/docs/_sass/vendor/bourbon/css3/_placeholder.scss +10 -0
- data/docs/_sass/vendor/bourbon/css3/_radial-gradient.scss +40 -0
- data/docs/_sass/vendor/bourbon/css3/_selection.scss +44 -0
- data/docs/_sass/vendor/bourbon/css3/_text-decoration.scss +27 -0
- data/docs/_sass/vendor/bourbon/css3/_transform.scss +21 -0
- data/docs/_sass/vendor/bourbon/css3/_transition.scss +81 -0
- data/docs/_sass/vendor/bourbon/css3/_user-select.scss +5 -0
- data/docs/_sass/vendor/bourbon/functions/_assign-inputs.scss +16 -0
- data/docs/_sass/vendor/bourbon/functions/_contains-falsy.scss +25 -0
- data/docs/_sass/vendor/bourbon/functions/_contains.scss +31 -0
- data/docs/_sass/vendor/bourbon/functions/_is-length.scss +16 -0
- data/docs/_sass/vendor/bourbon/functions/_is-light.scss +26 -0
- data/docs/_sass/vendor/bourbon/functions/_is-number.scss +16 -0
- data/docs/_sass/vendor/bourbon/functions/_is-size.scss +23 -0
- data/docs/_sass/vendor/bourbon/functions/_modular-scale.scss +74 -0
- data/docs/_sass/vendor/bourbon/functions/_px-to-em.scss +24 -0
- data/docs/_sass/vendor/bourbon/functions/_px-to-rem.scss +26 -0
- data/docs/_sass/vendor/bourbon/functions/_shade.scss +24 -0
- data/docs/_sass/vendor/bourbon/functions/_strip-units.scss +22 -0
- data/docs/_sass/vendor/bourbon/functions/_tint.scss +24 -0
- data/docs/_sass/vendor/bourbon/functions/_transition-property-name.scss +37 -0
- data/docs/_sass/vendor/bourbon/functions/_unpack.scss +32 -0
- data/docs/_sass/vendor/bourbon/helpers/_convert-units.scss +26 -0
- data/docs/_sass/vendor/bourbon/helpers/_directional-values.scss +108 -0
- data/docs/_sass/vendor/bourbon/helpers/_font-source-declaration.scss +53 -0
- data/docs/_sass/vendor/bourbon/helpers/_gradient-positions-parser.scss +24 -0
- data/docs/_sass/vendor/bourbon/helpers/_linear-angle-parser.scss +35 -0
- data/docs/_sass/vendor/bourbon/helpers/_linear-gradient-parser.scss +51 -0
- data/docs/_sass/vendor/bourbon/helpers/_linear-positions-parser.scss +77 -0
- data/docs/_sass/vendor/bourbon/helpers/_linear-side-corner-parser.scss +41 -0
- data/docs/_sass/vendor/bourbon/helpers/_radial-arg-parser.scss +74 -0
- data/docs/_sass/vendor/bourbon/helpers/_radial-gradient-parser.scss +55 -0
- data/docs/_sass/vendor/bourbon/helpers/_radial-positions-parser.scss +28 -0
- data/docs/_sass/vendor/bourbon/helpers/_render-gradients.scss +31 -0
- data/docs/_sass/vendor/bourbon/helpers/_shape-size-stripper.scss +15 -0
- data/docs/_sass/vendor/bourbon/helpers/_str-to-num.scss +55 -0
- data/docs/_sass/vendor/bourbon/settings/_asset-pipeline.scss +7 -0
- data/docs/_sass/vendor/bourbon/settings/_deprecation-warnings.scss +8 -0
- data/docs/_sass/vendor/bourbon/settings/_prefixer.scss +9 -0
- data/docs/_sass/vendor/bourbon/settings/_px-to-em.scss +1 -0
- data/docs/_sass/vendor/neat/_neat-helpers.scss +11 -0
- data/docs/_sass/vendor/neat/_neat.scss +23 -0
- data/docs/_sass/vendor/neat/functions/_new-breakpoint.scss +49 -0
- data/docs/_sass/vendor/neat/functions/_private.scss +114 -0
- data/docs/_sass/vendor/neat/grid/_box-sizing.scss +15 -0
- data/docs/_sass/vendor/neat/grid/_direction-context.scss +33 -0
- data/docs/_sass/vendor/neat/grid/_display-context.scss +28 -0
- data/docs/_sass/vendor/neat/grid/_fill-parent.scss +22 -0
- data/docs/_sass/vendor/neat/grid/_media.scss +92 -0
- data/docs/_sass/vendor/neat/grid/_omega.scss +87 -0
- data/docs/_sass/vendor/neat/grid/_outer-container.scss +34 -0
- data/docs/_sass/vendor/neat/grid/_pad.scss +25 -0
- data/docs/_sass/vendor/neat/grid/_private.scss +35 -0
- data/docs/_sass/vendor/neat/grid/_row.scss +52 -0
- data/docs/_sass/vendor/neat/grid/_shift.scss +50 -0
- data/docs/_sass/vendor/neat/grid/_span-columns.scss +94 -0
- data/docs/_sass/vendor/neat/grid/_to-deprecate.scss +97 -0
- data/docs/_sass/vendor/neat/grid/_visual-grid.scss +42 -0
- data/docs/_sass/vendor/neat/mixins/_clearfix.scss +25 -0
- data/docs/_sass/vendor/neat/settings/_disable-warnings.scss +13 -0
- data/docs/_sass/vendor/neat/settings/_grid.scss +51 -0
- data/docs/_sass/vendor/neat/settings/_visual-grid.scss +27 -0
- data/docs/_sass/vendor/normalize-3.0.2.scss +427 -0
- data/docs/_sass/vendor/pygments.scss +356 -0
- data/docs/automating_browsers/capybara.md +70 -0
- data/docs/css/screen.scss +7 -0
- data/docs/guides/callbacks.md +45 -0
- data/docs/guides/cli.md +52 -0
- data/docs/guides/configuration.md +184 -0
- data/docs/guides/error_handling.md +46 -0
- data/docs/guides/frontiers.md +93 -0
- data/docs/guides/halting.md +23 -0
- data/docs/guides/job_queues.md +26 -0
- data/docs/guides/locals.md +36 -0
- data/docs/guides/logging.md +22 -0
- data/docs/guides/page_objects.md +67 -0
- data/docs/guides/peeking.md +46 -0
- data/docs/guides/selenium_capybara.md +100 -0
- data/docs/guides/tutorial.md +452 -0
- data/docs/index.md +82 -0
- data/docs/js/navigation.js +11 -0
- data/docs/misc/contributing.md +20 -0
- data/docs/misc/testing.md +11 -0
- data/docs/recipes/authentication.md +23 -0
- data/docs/recipes/csv.md +29 -0
- data/docs/recipes/javascript.md +20 -0
- data/docs/recipes/multiple_uris.md +18 -0
- data/docs/recipes/screenshots.md +20 -0
- data/docs/routing/custom_rules.md +16 -0
- data/docs/routing/filetypes_rules.md +21 -0
- data/docs/routing/host_rules.md +24 -0
- data/docs/routing/path_rules.md +33 -0
- data/docs/routing/protocol_rules.md +17 -0
- data/docs/routing/query_rules.md +69 -0
- data/docs/routing/routes.md +96 -0
- data/docs/routing/uri_rules.md +18 -0
- data/examples/collect_github_issues.rb +65 -0
- data/examples/find_foobar_on_wikipedia.rb +23 -0
- data/lib/wayfarer/configuration.rb +86 -0
- data/lib/wayfarer/crawl.rb +79 -0
- data/lib/wayfarer/crawl_observer.rb +103 -0
- data/lib/wayfarer/dispatcher.rb +104 -0
- data/lib/wayfarer/finders.rb +61 -0
- data/lib/wayfarer/frontiers/frontier.rb +79 -0
- data/lib/wayfarer/frontiers/memory_bloomfilter.rb +32 -0
- data/lib/wayfarer/frontiers/memory_frontier.rb +76 -0
- data/lib/wayfarer/frontiers/memory_trie_frontier.rb +39 -0
- data/lib/wayfarer/frontiers/normalize_uris.rb +48 -0
- data/lib/wayfarer/frontiers/redis_bloomfilter.rb +34 -0
- data/lib/wayfarer/frontiers/redis_frontier.rb +83 -0
- data/lib/wayfarer/http_adapters/adapter_pool.rb +62 -0
- data/lib/wayfarer/http_adapters/net_http_adapter.rb +77 -0
- data/lib/wayfarer/http_adapters/selenium_adapter.rb +80 -0
- data/lib/wayfarer/job.rb +211 -0
- data/lib/wayfarer/locals.rb +40 -0
- data/lib/wayfarer/page.rb +94 -0
- data/lib/wayfarer/parsers/json_parser.rb +20 -0
- data/lib/wayfarer/parsers/xml_parser.rb +27 -0
- data/lib/wayfarer/processor.rb +103 -0
- data/lib/wayfarer/routing/custom_rule.rb +21 -0
- data/lib/wayfarer/routing/filetypes_rule.rb +20 -0
- data/lib/wayfarer/routing/host_rule.rb +19 -0
- data/lib/wayfarer/routing/path_rule.rb +54 -0
- data/lib/wayfarer/routing/protocol_rule.rb +21 -0
- data/lib/wayfarer/routing/query_rule.rb +59 -0
- data/lib/wayfarer/routing/router.rb +71 -0
- data/lib/wayfarer/routing/rule.rb +114 -0
- data/lib/wayfarer/routing/uri_rule.rb +21 -0
- data/lib/wayfarer.rb +68 -0
- data/spec/configuration_spec.rb +26 -0
- data/spec/crawl_spec.rb +48 -0
- data/spec/finders_spec.rb +49 -0
- data/spec/frontiers/memory_bloomfilter_spec.rb +6 -0
- data/spec/frontiers/memory_frontier_spec.rb +6 -0
- data/spec/frontiers/memory_trie_frontier_spec.rb +6 -0
- data/spec/frontiers/normalize_uris_spec.rb +59 -0
- data/spec/frontiers/redis_bloomfilter_spec.rb +6 -0
- data/spec/frontiers/redis_frontier_spec.rb +6 -0
- data/spec/http_adapters/adapter_pool_spec.rb +33 -0
- data/spec/http_adapters/net_http_adapter_spec.rb +83 -0
- data/spec/http_adapters/selenium_adapter_spec.rb +53 -0
- data/spec/integration/callbacks_spec.rb +42 -0
- data/spec/integration/locals_spec.rb +106 -0
- data/spec/integration/peeking_spec.rb +61 -0
- data/spec/job_spec.rb +122 -0
- data/spec/page_spec.rb +38 -0
- data/spec/parsers/json_parser_spec.rb +30 -0
- data/spec/parsers/xml_parser_spec.rb +24 -0
- data/spec/processor_spec.rb +31 -0
- data/spec/routing/custom_rule_spec.rb +26 -0
- data/spec/routing/filetypes_rule_spec.rb +40 -0
- data/spec/routing/host_rule_spec.rb +48 -0
- data/spec/routing/path_rule_spec.rb +66 -0
- data/spec/routing/protocol_rule_spec.rb +26 -0
- data/spec/routing/query_rule_spec.rb +124 -0
- data/spec/routing/router_spec.rb +67 -0
- data/spec/routing/rule_spec.rb +251 -0
- data/spec/routing/uri_rule_spec.rb +24 -0
- data/spec/shared/frontier.rb +96 -0
- data/spec/spec_helpers.rb +62 -0
- data/spec/wayfarer_spec.rb +24 -0
- data/support/static/finders.html +38 -0
- data/support/static/graph/details/a.html +10 -0
- data/support/static/graph/details/b.html +10 -0
- data/support/static/graph/index.html +20 -0
- data/support/static/json/dummy.json +13 -0
- data/support/static/links/links.html +28 -0
- data/support/static/xml/dummy.xml +120 -0
- data/support/test_app.rb +45 -0
- data/wayfarer-jruby.gemspec +49 -0
- data/wayfarer.gemspec +53 -0
- metadata +697 -0
@@ -0,0 +1,80 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "selenium-webdriver"
|
4
|
+
require "selenium/emulated_features"
|
5
|
+
require "capybara"
|
6
|
+
|
7
|
+
module Wayfarer
|
8
|
+
module HTTPAdapters
|
9
|
+
# An adapter for Selenium WebDrivers
|
10
|
+
# @api private
|
11
|
+
class SeleniumAdapter
|
12
|
+
# @!attribute [r] driver
|
13
|
+
# @return [URI] the Selenium WebDriver.
|
14
|
+
attr_reader :driver
|
15
|
+
|
16
|
+
def initialize(config = Wayfarer.config)
|
17
|
+
@config = config
|
18
|
+
end
|
19
|
+
|
20
|
+
# Fetches a page.
|
21
|
+
# @return [Page]
|
22
|
+
def fetch(uri)
|
23
|
+
driver.navigate.to(uri)
|
24
|
+
|
25
|
+
Page.new(
|
26
|
+
uri: @driver.current_url,
|
27
|
+
status_code: @driver.response_code,
|
28
|
+
body: @driver.page_source,
|
29
|
+
headers: @driver.response_headers
|
30
|
+
)
|
31
|
+
end
|
32
|
+
|
33
|
+
# Closes the driver.
|
34
|
+
def reload!
|
35
|
+
@driver&.close
|
36
|
+
@driver = nil
|
37
|
+
end
|
38
|
+
|
39
|
+
# Quits the browser.
|
40
|
+
def free
|
41
|
+
@driver&.quit
|
42
|
+
@driver = nil
|
43
|
+
end
|
44
|
+
|
45
|
+
# The WebDriver.
|
46
|
+
def driver
|
47
|
+
@driver ||= instantiate_driver
|
48
|
+
end
|
49
|
+
|
50
|
+
# A Capybara driver that wraps the {#driver}.
|
51
|
+
# @see https://github.com/teamcapybara/capybara Capybara
|
52
|
+
def browser
|
53
|
+
@browser ||= instantiate_capybara_driver
|
54
|
+
end
|
55
|
+
|
56
|
+
private
|
57
|
+
|
58
|
+
def instantiate_driver
|
59
|
+
driver = Selenium::WebDriver.for(*@config.selenium_argv)
|
60
|
+
driver.manage.window.size = Selenium::WebDriver::Dimension.new(
|
61
|
+
*@config.window_size
|
62
|
+
)
|
63
|
+
driver
|
64
|
+
end
|
65
|
+
|
66
|
+
def instantiate_capybara_driver
|
67
|
+
Capybara.run_server = false
|
68
|
+
Capybara.current_driver = :selenium
|
69
|
+
|
70
|
+
capybara_driver = Capybara::Selenium::Driver.new(nil)
|
71
|
+
capybara_driver.instance_variable_set(:@browser, driver)
|
72
|
+
|
73
|
+
session = Capybara::Session.new(:selenium, nil)
|
74
|
+
session.instance_variable_set(:@driver, capybara_driver)
|
75
|
+
|
76
|
+
session
|
77
|
+
end
|
78
|
+
end
|
79
|
+
end
|
80
|
+
end
|
data/lib/wayfarer/job.rb
ADDED
@@ -0,0 +1,211 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "forwardable"
|
4
|
+
require "hooks"
|
5
|
+
require "active_job"
|
6
|
+
|
7
|
+
# TODO: I only want deep_dup
|
8
|
+
require "active_support/all"
|
9
|
+
|
10
|
+
module Wayfarer
|
11
|
+
# A {Job} is a class that has a {Routing::Router} with many {Routing::Rule}s
|
12
|
+
# which are matched against a URI. Rules map URIs onto job instance methods.
|
13
|
+
# Under the hood, jobs are instantiated within separate threads by a
|
14
|
+
# {Processor}. Every instance gets its own thread. If a URI is matched, its
|
15
|
+
# {Page} is retrieved, and made available to instance methods via {#page}.
|
16
|
+
#
|
17
|
+
# Jobs implement ActiveJob's Job API and are therefore compatible with a wide
|
18
|
+
# range of job queues. To run a job immediately, call ::perform_now.
|
19
|
+
# enqueue a job, call ::perform_later.
|
20
|
+
#
|
21
|
+
# @see https://github.com/rails/rails/tree/master/activejob rails/activejob
|
22
|
+
# @see http://edgeguides.rubyonrails.org/active_job_basics.html ActiveJob Basics
|
23
|
+
class Job < ActiveJob::Base
|
24
|
+
extend Forwardable
|
25
|
+
|
26
|
+
include Hooks
|
27
|
+
include Locals
|
28
|
+
|
29
|
+
# @!group Callbacks
|
30
|
+
|
31
|
+
# Callback that fires __once__ before any pages are retrieved.
|
32
|
+
# @method before_crawl
|
33
|
+
# @scope class
|
34
|
+
define_hook :before_crawl
|
35
|
+
|
36
|
+
# Callback that fires __once__ after all pages have been retrieved and
|
37
|
+
# processing is done.
|
38
|
+
# @method after_crawl
|
39
|
+
# @scope class
|
40
|
+
define_hook :after_crawl
|
41
|
+
|
42
|
+
# Callback that fires when HTTP adapters are instantiated.
|
43
|
+
# @method setup_adapter
|
44
|
+
# @scope class
|
45
|
+
# @yield [[HTTPAdapters::NetHTTPAdapter, HTTPAdapters::SeleniumAdapter], [Selenium::WebDriver::Driver, nil], [Capybara::Selenium::Driver, nil]]
|
46
|
+
define_hooks :setup_adapter
|
47
|
+
|
48
|
+
# @!endgroup
|
49
|
+
|
50
|
+
class << self
|
51
|
+
extend Forwardable
|
52
|
+
|
53
|
+
# @!attribute [w] router
|
54
|
+
attr_writer :router
|
55
|
+
|
56
|
+
# @!attribute [w] config
|
57
|
+
attr_writer :config
|
58
|
+
|
59
|
+
# Returns a class copy.
|
60
|
+
def prepare
|
61
|
+
duplicate = dup
|
62
|
+
duplicate.router = router.dup
|
63
|
+
duplicate.locals = locals.deep_dup
|
64
|
+
duplicate.config = config.dup
|
65
|
+
|
66
|
+
duplicate.locals.each do |(key, val)|
|
67
|
+
duplicate.locals[key] = Locals.thread_safe_counterpart(val)
|
68
|
+
end
|
69
|
+
|
70
|
+
duplicate.locals.each do |(key, _)|
|
71
|
+
duplicate.send(:define_method, key) do duplicate.locals[key] end
|
72
|
+
duplicate.send(:define_singleton_method, key) do
|
73
|
+
duplicate.locals[key]
|
74
|
+
end
|
75
|
+
end
|
76
|
+
|
77
|
+
duplicate
|
78
|
+
end
|
79
|
+
|
80
|
+
# A configuration based off the global {Wayfarer.config}.
|
81
|
+
# @yield [Configuration]
|
82
|
+
# @return [Configuration]
|
83
|
+
def config
|
84
|
+
@config ||= Wayfarer.config.clone
|
85
|
+
yield(@config) if block_given?
|
86
|
+
@config
|
87
|
+
end
|
88
|
+
|
89
|
+
# A router.
|
90
|
+
# If a block is passed in, it is evaluated within the {Router}'s instance.
|
91
|
+
# @return [Routing::Router]
|
92
|
+
def router(&proc)
|
93
|
+
@router ||= Routing::Router.new
|
94
|
+
@router.instance_eval(&proc) if block_given?
|
95
|
+
@router
|
96
|
+
end
|
97
|
+
|
98
|
+
alias route router
|
99
|
+
alias routes router
|
100
|
+
|
101
|
+
# Overshadows ActiveJob::Base's own logger
|
102
|
+
delegate logger: :config
|
103
|
+
end
|
104
|
+
|
105
|
+
# @!attribute [r] staged_uris
|
106
|
+
# @return [Array<String>, Array<URI>] URIs to stage for the next cycle.
|
107
|
+
# @see #stage
|
108
|
+
attr_reader :staged_uris
|
109
|
+
|
110
|
+
# @!attribute [rw] page
|
111
|
+
attr_writer :page
|
112
|
+
|
113
|
+
# @!attribute [rw] adapter
|
114
|
+
attr_accessor :adapter
|
115
|
+
|
116
|
+
# @!attribute [rw] params
|
117
|
+
attr_accessor :params
|
118
|
+
|
119
|
+
def initialize(*argv)
|
120
|
+
@halts = false
|
121
|
+
@staged_uris = []
|
122
|
+
super(*argv)
|
123
|
+
end
|
124
|
+
|
125
|
+
# Whether this job will stop processing.
|
126
|
+
def halts?
|
127
|
+
@halts
|
128
|
+
end
|
129
|
+
|
130
|
+
# Performs this job.
|
131
|
+
# @note ActiveJob API
|
132
|
+
# @override
|
133
|
+
def perform(*uris)
|
134
|
+
Crawl.new(self.class, *uris).execute
|
135
|
+
end
|
136
|
+
|
137
|
+
protected
|
138
|
+
|
139
|
+
# All following instance methods are available within actions.
|
140
|
+
|
141
|
+
# Sets a halting flag that signals the processor to stop its work.
|
142
|
+
def halt
|
143
|
+
@halts = true
|
144
|
+
end
|
145
|
+
|
146
|
+
# Adds URIs to process in the next cycle.
|
147
|
+
# If a relative path is given, an absolute URI is constructed from the
|
148
|
+
# current {#page}'s URI.
|
149
|
+
# @param [String, URI, Array<String>, Array<URI>]
|
150
|
+
def stage(*uris)
|
151
|
+
expanded = uris.flatten.map do |u|
|
152
|
+
if (uri = URI(u)).absolute?
|
153
|
+
uri
|
154
|
+
else
|
155
|
+
# URI#join would discard the path of page.uri.path
|
156
|
+
current = page.uri.dup
|
157
|
+
current.path = File.join(page.uri.path, uri.path)
|
158
|
+
current
|
159
|
+
end
|
160
|
+
end
|
161
|
+
|
162
|
+
# This method has somewhat become the guard keeper for invalid URIs that
|
163
|
+
# would lead to exceptions otherwise down the line
|
164
|
+
supported = expanded.select do |uri|
|
165
|
+
HTTPAdapters::NetHTTPAdapter::RECOGNIZED_URI_TYPES.any? do |type|
|
166
|
+
uri.is_a?(type)
|
167
|
+
end
|
168
|
+
end
|
169
|
+
|
170
|
+
@staged_uris.push(*supported)
|
171
|
+
end
|
172
|
+
|
173
|
+
# The {Page} representing the URI currently processed by an action.
|
174
|
+
# When using the Selenium adapter, {Page#body} gets refreshed on every call.
|
175
|
+
# Otherwise, subsequent DOM updates (i.e. JavaScript-induced) would be
|
176
|
+
# invisible.
|
177
|
+
# @return Page
|
178
|
+
def page
|
179
|
+
return @page unless self.class.config.http_adapter == :selenium
|
180
|
+
|
181
|
+
Page.new(
|
182
|
+
uri: @page.uri,
|
183
|
+
status_code: @page.uri,
|
184
|
+
body: driver.page_source,
|
185
|
+
headers: @page.headers
|
186
|
+
)
|
187
|
+
end
|
188
|
+
|
189
|
+
# The parsed response body.
|
190
|
+
# When using the Selenium adapter, this parses the body again on every call.
|
191
|
+
# Otherwise, subsequent DOM updates (i.e. JavaScript-induced) would be
|
192
|
+
# invisible.
|
193
|
+
# @method doc
|
194
|
+
# @see Page#doc
|
195
|
+
delegate doc: :page
|
196
|
+
|
197
|
+
# The Selenium WebDriver.
|
198
|
+
# @method driver
|
199
|
+
# @see https://github.com/peterc/pismo Pismo
|
200
|
+
# @see Page#driver
|
201
|
+
delegate driver: :adapter
|
202
|
+
|
203
|
+
# A Capybara driver that wraps the {#driver}.
|
204
|
+
# @method browser
|
205
|
+
# @see HTTPAdapters::SeleniumAdapter#browser
|
206
|
+
delegate browser: :adapter
|
207
|
+
|
208
|
+
# @method logger
|
209
|
+
delegate logger: :"self.class"
|
210
|
+
end
|
211
|
+
end
|
@@ -0,0 +1,40 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "concurrent"
|
4
|
+
|
5
|
+
module Wayfarer
|
6
|
+
# @api private
|
7
|
+
module Locals
|
8
|
+
def self.thread_safe_counterpart(value)
|
9
|
+
case value
|
10
|
+
when Array then Concurrent::Array.new(value)
|
11
|
+
when Hash then Concurrent::Hash[value]
|
12
|
+
when TrueClass then Concurrent::AtomicBoolean.new(value)
|
13
|
+
when FalseClass then Concurrent::AtomicBoolean.new(value)
|
14
|
+
when Integer then Concurrent::AtomicFixnum.new(value)
|
15
|
+
else value
|
16
|
+
end
|
17
|
+
end
|
18
|
+
|
19
|
+
def self.included(base)
|
20
|
+
base.extend(ClassMethods)
|
21
|
+
end
|
22
|
+
|
23
|
+
module ClassMethods
|
24
|
+
attr_reader :locals
|
25
|
+
|
26
|
+
def let(key)
|
27
|
+
raise "#let called without a block" unless block_given?
|
28
|
+
locals[key] = yield
|
29
|
+
end
|
30
|
+
|
31
|
+
def locals
|
32
|
+
@locals ||= {}
|
33
|
+
end
|
34
|
+
|
35
|
+
def locals=(locals)
|
36
|
+
@locals = locals
|
37
|
+
end
|
38
|
+
end
|
39
|
+
end
|
40
|
+
end
|
@@ -0,0 +1,94 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "ostruct"
|
4
|
+
require "forwardable"
|
5
|
+
require "mime/types"
|
6
|
+
require "mime-types"
|
7
|
+
|
8
|
+
require "pismo" unless RUBY_PLATFORM == "java"
|
9
|
+
|
10
|
+
module Wayfarer
|
11
|
+
# The representation of fetched pages
|
12
|
+
class Page
|
13
|
+
extend Forwardable
|
14
|
+
|
15
|
+
include Finders
|
16
|
+
|
17
|
+
# @!attribute [r] uri
|
18
|
+
# @return [URI] the URI of the page.
|
19
|
+
attr_reader :uri
|
20
|
+
|
21
|
+
# @!attribute [r] status_code
|
22
|
+
# @return [Fixnum] the response status code.
|
23
|
+
attr_reader :status_code
|
24
|
+
|
25
|
+
# @!attribute [r] body
|
26
|
+
# @return [String] the response body.
|
27
|
+
attr_accessor :body
|
28
|
+
|
29
|
+
# @!attribute [r] headers
|
30
|
+
# @return [Hash] the response headers.
|
31
|
+
attr_reader :headers
|
32
|
+
|
33
|
+
def initialize(attrs = {})
|
34
|
+
@uri = attrs[:uri]
|
35
|
+
@status_code = attrs[:status_code]
|
36
|
+
@body = attrs[:body]
|
37
|
+
@headers = attrs[:headers]
|
38
|
+
end
|
39
|
+
|
40
|
+
# Returns a parsed representation of the fetched document depending on the
|
41
|
+
# Content-Type field.
|
42
|
+
# @return [OpenStruct] if the Content-Type field's sub-type is "json".
|
43
|
+
# @return [Nokogiri::XML::Document] if the Content-Type field's sub-type is "xml".
|
44
|
+
# @return [Nokogiri::HTML::Document] otherwise.
|
45
|
+
def doc
|
46
|
+
return @doc if @doc
|
47
|
+
|
48
|
+
# If no Content-Type field is present, assume HTML/XML
|
49
|
+
# TODO: Test
|
50
|
+
unless @headers["content-type"]
|
51
|
+
return @doc = Parsers::XMLParser.parse_html(@body)
|
52
|
+
end
|
53
|
+
|
54
|
+
content_type = @headers["content-type"].first
|
55
|
+
sub_type = MIME::Types[content_type].first.sub_type
|
56
|
+
|
57
|
+
# TODO: Tests
|
58
|
+
@doc = case sub_type
|
59
|
+
when "json"
|
60
|
+
Parsers::JSONParser.parse(@body)
|
61
|
+
when "xml"
|
62
|
+
Parsers::XMLParser.parse_xml(@body)
|
63
|
+
else
|
64
|
+
Parsers::XMLParser.parse_html(@body)
|
65
|
+
end
|
66
|
+
end
|
67
|
+
|
68
|
+
# Pismo is not supported on JRuby.
|
69
|
+
unless RUBY_PLATFORM == "java"
|
70
|
+
# `#images` is included from the Helpers module
|
71
|
+
# `#body` is an attribute reader defined above
|
72
|
+
delegate (Pismo::Document::ATTRIBUTE_METHODS - %i[images body]) => :pismo
|
73
|
+
end
|
74
|
+
|
75
|
+
private
|
76
|
+
|
77
|
+
# Returns a Pismo document.
|
78
|
+
# @note Not available on JRuby.
|
79
|
+
# @note Only succeeds when {#doc} returns a `Nokogiri::HTML::Document`.
|
80
|
+
# @return [Pismo::Document]
|
81
|
+
def pismo
|
82
|
+
@pismo_doc ||= instantiate_pismo_document
|
83
|
+
end
|
84
|
+
|
85
|
+
def instantiate_pismo_document
|
86
|
+
doc = Pismo::Document.allocate
|
87
|
+
doc.instance_variable_set(:@options, {})
|
88
|
+
doc.instance_variable_set(:@url, uri)
|
89
|
+
doc.instance_variable_set(:@html, body)
|
90
|
+
doc.instance_variable_set(:@doc, self.doc)
|
91
|
+
doc
|
92
|
+
end
|
93
|
+
end
|
94
|
+
end
|
@@ -0,0 +1,20 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "oj" unless RUBY_PLATFORM == "java"
|
4
|
+
|
5
|
+
module Wayfarer
|
6
|
+
module Parsers
|
7
|
+
# A wrapper class for parsing JSON.
|
8
|
+
# @private
|
9
|
+
module JSONParser
|
10
|
+
module_function
|
11
|
+
|
12
|
+
# Parses a JSON string.
|
13
|
+
# @param [String] json_str the JSON string to parse.
|
14
|
+
# @return [OpenStruct]
|
15
|
+
def parse(json_str)
|
16
|
+
RUBY_PLATFORM == "java" ? JSON.parse(json_str) : Oj.load(json_str)
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
@@ -0,0 +1,27 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "nokogiri"
|
4
|
+
|
5
|
+
module Wayfarer
|
6
|
+
module Parsers
|
7
|
+
# A wrapper class for parsing HTML/XML.
|
8
|
+
# @private
|
9
|
+
module XMLParser
|
10
|
+
module_function
|
11
|
+
|
12
|
+
# Parses an XML string.
|
13
|
+
# @param [String] xml_str the XML string to parse.
|
14
|
+
# @return [Nokogiri::XML::Document]
|
15
|
+
def parse_xml(xml_str)
|
16
|
+
Nokogiri::XML(xml_str)
|
17
|
+
end
|
18
|
+
|
19
|
+
# Parses a HTML string.
|
20
|
+
# @param [String] html_str the HTML string to parse.
|
21
|
+
# @return [Nokogiri::HTML::Document]
|
22
|
+
def parse_html(html_str)
|
23
|
+
Nokogiri::HTML(html_str)
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
27
|
+
end
|
@@ -0,0 +1,103 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "pp"
|
4
|
+
require "concurrent"
|
5
|
+
require "observer"
|
6
|
+
|
7
|
+
module Wayfarer
|
8
|
+
# Runs jobs.
|
9
|
+
class Processor
|
10
|
+
extend Forwardable
|
11
|
+
|
12
|
+
include Observable
|
13
|
+
include CrawlObserver::Events
|
14
|
+
include CrawlObserver::ObservableShortcuts
|
15
|
+
|
16
|
+
attr_reader :job
|
17
|
+
|
18
|
+
delegate config: :job
|
19
|
+
delegate logger: :config
|
20
|
+
|
21
|
+
def initialize(job, frontier, dispatcher)
|
22
|
+
@job = job
|
23
|
+
@frontier = frontier
|
24
|
+
@dispatcher = dispatcher
|
25
|
+
@halted = Concurrent::AtomicBoolean.new(false)
|
26
|
+
end
|
27
|
+
|
28
|
+
# Whether processing is done.
|
29
|
+
# @return [true, false]
|
30
|
+
def halted?
|
31
|
+
@halted.value
|
32
|
+
end
|
33
|
+
|
34
|
+
# Sets a halt flag.
|
35
|
+
def halt!
|
36
|
+
@halted.make_true
|
37
|
+
end
|
38
|
+
|
39
|
+
# Runs the job.
|
40
|
+
# @param [*Array<URI>, *Array<String>] uris
|
41
|
+
def run(*_uris)
|
42
|
+
notify_observers!(FirstCycle.new(@frontier))
|
43
|
+
|
44
|
+
while @halted.false? && @frontier.cycle
|
45
|
+
current_uris = @frontier.current_uris
|
46
|
+
queue = current_uris.inject(Queue.new, :push)
|
47
|
+
|
48
|
+
notify_observers!(NewCycle.new(current_uris.count))
|
49
|
+
|
50
|
+
@threads = Array.new(config.connection_count) do
|
51
|
+
Thread.new do
|
52
|
+
begin
|
53
|
+
loop do
|
54
|
+
uri = queue.pop(true)
|
55
|
+
break if uri.nil? || @halted.true?
|
56
|
+
handle_dispatch_result(@dispatcher.dispatch(@job, uri))
|
57
|
+
end
|
58
|
+
rescue ThreadError
|
59
|
+
notify_observers!(CycleFinished.new)
|
60
|
+
end
|
61
|
+
end
|
62
|
+
end
|
63
|
+
|
64
|
+
@threads.each(&:join)
|
65
|
+
|
66
|
+
notify_observers!(AboutToCycle.new(@frontier.staged_uris.count))
|
67
|
+
end
|
68
|
+
ensure
|
69
|
+
halt!
|
70
|
+
@frontier.free
|
71
|
+
@dispatcher.adapter_pool.free
|
72
|
+
end
|
73
|
+
|
74
|
+
private
|
75
|
+
|
76
|
+
def handle_dispatch_result(result)
|
77
|
+
case result
|
78
|
+
when Dispatcher::Mismatch then handle_mismatch(result)
|
79
|
+
when Dispatcher::Halt then handle_halt(result)
|
80
|
+
when Dispatcher::Stage then handle_stage(result)
|
81
|
+
when Dispatcher::Error then handle_error(result)
|
82
|
+
end
|
83
|
+
end
|
84
|
+
|
85
|
+
def handle_mismatch(mismatch)
|
86
|
+
notify_observers!(MismatchedURI.new(mismatch.uri))
|
87
|
+
end
|
88
|
+
|
89
|
+
def handle_halt(halt)
|
90
|
+
notify_observers!(HaltInitiated.new(halt.action, halt.uri))
|
91
|
+
halt!
|
92
|
+
end
|
93
|
+
|
94
|
+
def handle_stage(stage)
|
95
|
+
notify_observers!(StagingURIs.new(stage.uris.count))
|
96
|
+
@frontier.stage(*stage.uris) unless halted?
|
97
|
+
end
|
98
|
+
|
99
|
+
def handle_error(error)
|
100
|
+
notify_observers!(UnhandledError.new(error.exception))
|
101
|
+
end
|
102
|
+
end
|
103
|
+
end
|
@@ -0,0 +1,21 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "uri"
|
4
|
+
|
5
|
+
module Wayfarer
|
6
|
+
module Routing
|
7
|
+
# @private
|
8
|
+
class CustomRule < Rule
|
9
|
+
def initialize(delegate_or_block = proc, opts = {}, &proc)
|
10
|
+
@delegate_or_block = delegate_or_block
|
11
|
+
super(opts, &proc)
|
12
|
+
end
|
13
|
+
|
14
|
+
private
|
15
|
+
|
16
|
+
def match!(uri)
|
17
|
+
!!@delegate_or_block.call(uri)
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
21
|
+
end
|
@@ -0,0 +1,20 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
require "uri"
|
3
|
+
|
4
|
+
module Wayfarer
|
5
|
+
module Routing
|
6
|
+
# @private
|
7
|
+
class FiletypesRule < Rule
|
8
|
+
def initialize(types, opts = {}, &proc)
|
9
|
+
@types = types
|
10
|
+
super(opts, &proc)
|
11
|
+
end
|
12
|
+
|
13
|
+
private
|
14
|
+
|
15
|
+
def match!(uri)
|
16
|
+
@types.any? { |type| uri.path =~ /\.#{type}$/ }
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
@@ -0,0 +1,19 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module Wayfarer
|
4
|
+
module Routing
|
5
|
+
# @private
|
6
|
+
class HostRule < Rule
|
7
|
+
def initialize(str_or_regexp, opts = {}, &proc)
|
8
|
+
@str_or_regexp = str_or_regexp
|
9
|
+
super(opts, &proc)
|
10
|
+
end
|
11
|
+
|
12
|
+
# rubocop:disable Style/CaseEquality
|
13
|
+
def match!(uri)
|
14
|
+
@str_or_regexp === uri.host
|
15
|
+
end
|
16
|
+
# rubocop:enable Style/CaseEquality
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
@@ -0,0 +1,54 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "mustermann"
|
4
|
+
|
5
|
+
module Wayfarer
|
6
|
+
module Routing
|
7
|
+
# @private
|
8
|
+
class PathRule < Rule
|
9
|
+
attr_reader :matcher
|
10
|
+
|
11
|
+
def initialize(arg, opts = {}, &proc)
|
12
|
+
@matcher = if arg.is_a? String
|
13
|
+
Mustermann.new(arg, type: Wayfarer.config.mustermann_type)
|
14
|
+
else
|
15
|
+
arg
|
16
|
+
end
|
17
|
+
|
18
|
+
super(opts, &proc)
|
19
|
+
end
|
20
|
+
|
21
|
+
def params(uri)
|
22
|
+
return {} unless match!(uri)
|
23
|
+
|
24
|
+
path = uri.path
|
25
|
+
|
26
|
+
if @matcher.is_a? Mustermann
|
27
|
+
@matcher.params(path)
|
28
|
+
else
|
29
|
+
captures = @matcher.match(full_path(uri)).captures
|
30
|
+
|
31
|
+
captures.each.with_index.reduce({}) do |hash, (capture, i)|
|
32
|
+
hash.merge(i.to_s => capture)
|
33
|
+
end
|
34
|
+
end
|
35
|
+
end
|
36
|
+
|
37
|
+
private
|
38
|
+
|
39
|
+
# rubocop:disable Style/CaseEquality
|
40
|
+
def match!(uri)
|
41
|
+
if @matcher.is_a? Mustermann
|
42
|
+
@matcher === uri.path
|
43
|
+
else
|
44
|
+
@matcher =~ full_path(uri)
|
45
|
+
end
|
46
|
+
end
|
47
|
+
# rubocop:enable Style/CaseEquality
|
48
|
+
|
49
|
+
def full_path(uri)
|
50
|
+
"#{uri.path}?#{uri.query}##{uri.fragment}"
|
51
|
+
end
|
52
|
+
end
|
53
|
+
end
|
54
|
+
end
|