crabfarm 0.0.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/bin/crabfarm +9 -0
- data/lib/crabfarm.rb +24 -0
- data/lib/crabfarm/adapters.rb +23 -0
- data/lib/crabfarm/adapters/capybara_adapter.rb +7 -0
- data/lib/crabfarm/adapters/surfer_adapter.rb +7 -0
- data/lib/crabfarm/adapters/watir_adapter.rb +7 -0
- data/lib/crabfarm/base_parser.rb +26 -0
- data/lib/crabfarm/base_state.rb +41 -0
- data/lib/crabfarm/cli.rb +79 -0
- data/lib/crabfarm/configuration.rb +83 -0
- data/lib/crabfarm/context.rb +32 -0
- data/lib/crabfarm/default_driver_factory.rb +37 -0
- data/lib/crabfarm/driver_bucket.rb +50 -0
- data/lib/crabfarm/driver_bucket_pool.rb +48 -0
- data/lib/crabfarm/dsl/surfer.rb +22 -0
- data/lib/crabfarm/dsl/surfer/search_context.rb +134 -0
- data/lib/crabfarm/dsl/surfer/surf_context.rb +58 -0
- data/lib/crabfarm/engines/safe_state_loop.rb +96 -0
- data/lib/crabfarm/errors.rb +50 -0
- data/lib/crabfarm/loader.rb +83 -0
- data/lib/crabfarm/modes/console.rb +86 -0
- data/lib/crabfarm/modes/generator.rb +120 -0
- data/lib/crabfarm/modes/server.rb +78 -0
- data/lib/crabfarm/module_helper.rb +35 -0
- data/lib/crabfarm/phantom_driver_factory.rb +33 -0
- data/lib/crabfarm/phantom_runner.rb +74 -0
- data/lib/crabfarm/rspec.rb +39 -0
- data/lib/crabfarm/state_store.rb +24 -0
- data/lib/crabfarm/support/custom_puma.rb +64 -0
- data/lib/crabfarm/templates/Crabfile.erb +3 -0
- data/lib/crabfarm/templates/Gemfile.erb +7 -0
- data/lib/crabfarm/templates/boot.rb.erb +13 -0
- data/lib/crabfarm/templates/crabfarm_bin.erb +3 -0
- data/lib/crabfarm/templates/dot_gitignore.erb +1 -0
- data/lib/crabfarm/templates/dot_gitkeep.erb +0 -0
- data/lib/crabfarm/templates/dot_rspec.erb +4 -0
- data/lib/crabfarm/templates/parser.rb.erb +8 -0
- data/lib/crabfarm/templates/parser_spec.rb.erb +7 -0
- data/lib/crabfarm/templates/spec_helper.rb.erb +22 -0
- data/lib/crabfarm/templates/state.rb.erb +8 -0
- data/lib/crabfarm/templates/state_spec.rb.erb +7 -0
- data/lib/crabfarm/version.rb +3 -0
- metadata +359 -0
@@ -0,0 +1,48 @@
|
|
1
|
+
module Crabfarm
|
2
|
+
class DriverBucketPool
|
3
|
+
|
4
|
+
def initialize(_module)
|
5
|
+
@module = _module
|
6
|
+
@buckets = Hash.new
|
7
|
+
@phantom = nil
|
8
|
+
|
9
|
+
init_phantom_if_required
|
10
|
+
end
|
11
|
+
|
12
|
+
def driver(_session_id=nil)
|
13
|
+
_session_id ||= :default_driver
|
14
|
+
bucket = @buckets[_session_id.to_sym]
|
15
|
+
bucket = @buckets[_session_id.to_sym] = DriverBucket.new(@module, _session_id, build_driver_factory) if bucket.nil?
|
16
|
+
bucket
|
17
|
+
end
|
18
|
+
|
19
|
+
def reset
|
20
|
+
@buckets.values.each(&:reset)
|
21
|
+
@buckets = Hash.new
|
22
|
+
end
|
23
|
+
|
24
|
+
def release
|
25
|
+
reset
|
26
|
+
@phantom.stop unless @phantom.nil?
|
27
|
+
end
|
28
|
+
|
29
|
+
private
|
30
|
+
|
31
|
+
def init_phantom_if_required
|
32
|
+
if @module.settings.phantom_enabled?
|
33
|
+
@phantom = PhantomRunner.new @module.settings.phantom_config
|
34
|
+
@phantom.start
|
35
|
+
end
|
36
|
+
end
|
37
|
+
|
38
|
+
def build_driver_factory
|
39
|
+
if @module.settings.phantom_enabled?
|
40
|
+
PhantomDriverFactory.new @phantom, @module.settings.driver_config
|
41
|
+
else
|
42
|
+
return @module.settings.driver_factory if @module.settings.driver_factory
|
43
|
+
DefaultDriverFactory.new @module.settings.driver_config
|
44
|
+
end
|
45
|
+
end
|
46
|
+
|
47
|
+
end
|
48
|
+
end
|
@@ -0,0 +1,22 @@
|
|
1
|
+
require 'crabfarm/dsl/surfer/search_context'
|
2
|
+
require 'crabfarm/dsl/surfer/surf_context'
|
3
|
+
|
4
|
+
module Crabfarm
|
5
|
+
module Dsl
|
6
|
+
module Surfer
|
7
|
+
|
8
|
+
class Error < StandardError
|
9
|
+
attr_reader :source
|
10
|
+
|
11
|
+
def initialize(_message, _ctx)
|
12
|
+
super _message
|
13
|
+
@ctx = _ctx
|
14
|
+
@source = _ctx.root_context.page_source rescue nil # cache page source for future reference
|
15
|
+
end
|
16
|
+
end
|
17
|
+
|
18
|
+
class EmptySetError < Error; end
|
19
|
+
class WebdriverError < Error; end
|
20
|
+
end
|
21
|
+
end
|
22
|
+
end
|
@@ -0,0 +1,134 @@
|
|
1
|
+
module Crabfarm
|
2
|
+
module Dsl
|
3
|
+
module Surfer
|
4
|
+
class SearchContext
|
5
|
+
include Enumerable
|
6
|
+
extend Forwardable
|
7
|
+
|
8
|
+
TIMEOUT = 10.0 # Default timeout for waiting operations
|
9
|
+
|
10
|
+
def initialize(_elements, _parent)
|
11
|
+
@elements = _elements
|
12
|
+
@parent = _parent
|
13
|
+
end
|
14
|
+
|
15
|
+
# return the context's root context
|
16
|
+
def root_context
|
17
|
+
return @parent.root_context if @parent
|
18
|
+
self
|
19
|
+
end
|
20
|
+
|
21
|
+
# return the context's parent context
|
22
|
+
def parent_context
|
23
|
+
@parent
|
24
|
+
end
|
25
|
+
|
26
|
+
# forward read-only array methods to context
|
27
|
+
def_delegators :context, :each, :[], :length, :count, :empty?, :first, :last
|
28
|
+
|
29
|
+
# yield individual SearchContext for each element contained in this result
|
30
|
+
def explode(&_block)
|
31
|
+
return enum_for(__method__) if _block.nil?
|
32
|
+
context.each do |el|
|
33
|
+
_block.call SearchContext.new([el], self)
|
34
|
+
end
|
35
|
+
end
|
36
|
+
|
37
|
+
# searches for elements that match a given selector
|
38
|
+
def search(_selector=nil, _options={})
|
39
|
+
_options[:css] = _selector if _selector
|
40
|
+
|
41
|
+
wait_mode = _options.delete :wait
|
42
|
+
if wait_mode
|
43
|
+
|
44
|
+
# retrieve timeout
|
45
|
+
timeout = _options.delete :timeout
|
46
|
+
timeout = TIMEOUT if timeout.nil?
|
47
|
+
|
48
|
+
# use a selenium timeout
|
49
|
+
wait = Selenium::WebDriver::Wait.new(timeout: timeout)
|
50
|
+
wait.until do
|
51
|
+
new_elements = search_elements _options
|
52
|
+
|
53
|
+
# test wait condition
|
54
|
+
ok = case wait_mode
|
55
|
+
when :present then (new_elements.length > 0)
|
56
|
+
when :visible then (new_elements.length > 0 and new_elements.first.displayed?)
|
57
|
+
when :enabled then (new_elements.length > 0 and new_elements.first.displayed? and new_elements.first.enabled?)
|
58
|
+
when :not_present then (new_elements.length == 0)
|
59
|
+
when :not_visible then (not new_elements.any? { |e| e.displayed? })
|
60
|
+
else
|
61
|
+
raise SetupError.new "Invalid wait mode '#{wait_mode}'"
|
62
|
+
end
|
63
|
+
|
64
|
+
SearchContext.new new_elements, self if ok
|
65
|
+
end
|
66
|
+
else
|
67
|
+
SearchContext.new search_elements(_options), self
|
68
|
+
end
|
69
|
+
end
|
70
|
+
|
71
|
+
# clears and sends_keys to this context main element
|
72
|
+
def fill(_value)
|
73
|
+
raise EmptySetError.new('Cannot call \'fill\' on an empty set', self) if empty?
|
74
|
+
wrap_errors do
|
75
|
+
context.first.clear
|
76
|
+
context.first.send_keys _value
|
77
|
+
end
|
78
|
+
end
|
79
|
+
|
80
|
+
# Any methods missing are forwarded to the main element (first).
|
81
|
+
def method_missing(_method, *_args, &_block)
|
82
|
+
wrap_errors do
|
83
|
+
m = /^(.*)_all$/.match _method.to_s
|
84
|
+
if m then
|
85
|
+
return [] if empty?
|
86
|
+
context.map { |e| e.send(m[1], *_args, &_block) }
|
87
|
+
else
|
88
|
+
raise EmptySetError.new("Cannot call '#{_method}' on an empty set", self) if empty?
|
89
|
+
context.first.send(_method, *_args, &_block)
|
90
|
+
end
|
91
|
+
end
|
92
|
+
end
|
93
|
+
|
94
|
+
def respond_to?(_method, _include_all=false)
|
95
|
+
return true if super
|
96
|
+
m = /^.*_all$/.match _method.to_s
|
97
|
+
if m then
|
98
|
+
return true if empty?
|
99
|
+
context.first.respond_to? m[1], _include_all
|
100
|
+
else
|
101
|
+
return true if empty?
|
102
|
+
context.first.respond_to? _method, _include_all
|
103
|
+
end
|
104
|
+
end
|
105
|
+
|
106
|
+
private
|
107
|
+
|
108
|
+
# wrap every selenium errors that happen inside block.
|
109
|
+
def wrap_errors
|
110
|
+
begin
|
111
|
+
yield
|
112
|
+
rescue Selenium::WebDriver::Error::WebDriverError => e
|
113
|
+
raise WebdriverError.new e, self
|
114
|
+
end
|
115
|
+
end
|
116
|
+
|
117
|
+
# base filtering method, expands current context
|
118
|
+
def search_elements(_options)
|
119
|
+
wrap_errors do
|
120
|
+
context.inject([]) do |r, element|
|
121
|
+
r + element.find_elements(_options)
|
122
|
+
end
|
123
|
+
end
|
124
|
+
end
|
125
|
+
|
126
|
+
# returns the current context
|
127
|
+
def context
|
128
|
+
@elements
|
129
|
+
end
|
130
|
+
|
131
|
+
end
|
132
|
+
end
|
133
|
+
end
|
134
|
+
end
|
@@ -0,0 +1,58 @@
|
|
1
|
+
module Crabfarm
|
2
|
+
module Dsl
|
3
|
+
module Surfer
|
4
|
+
class SurfContext < SearchContext
|
5
|
+
extend Forwardable
|
6
|
+
|
7
|
+
def_delegators :@bucket, :parse, :setup
|
8
|
+
def_delegators 'driver.navigate', :back, :forward, :refresh
|
9
|
+
|
10
|
+
def initialize(_bucket)
|
11
|
+
super nil, nil
|
12
|
+
@bucket = _bucket
|
13
|
+
end
|
14
|
+
|
15
|
+
def driver
|
16
|
+
@bucket.original
|
17
|
+
end
|
18
|
+
|
19
|
+
def quit
|
20
|
+
@bucket.reset
|
21
|
+
end
|
22
|
+
|
23
|
+
def current_uri
|
24
|
+
URI.parse driver.current_url
|
25
|
+
end
|
26
|
+
|
27
|
+
def cookies
|
28
|
+
driver.manage.all_cookies
|
29
|
+
end
|
30
|
+
|
31
|
+
def goto(_url, _params=nil)
|
32
|
+
_url += "?#{_params.to_query}" if _params
|
33
|
+
retries = 0
|
34
|
+
|
35
|
+
loop do
|
36
|
+
begin
|
37
|
+
@bucket.reset if retries > 0
|
38
|
+
driver.get(_url)
|
39
|
+
break
|
40
|
+
rescue Timeout::Error #, Selenium::WebDriver::Error::UnknownError
|
41
|
+
# TODO: log this
|
42
|
+
raise if retries >= max_retries
|
43
|
+
retries += 1
|
44
|
+
sleep 1.0
|
45
|
+
end
|
46
|
+
end
|
47
|
+
end
|
48
|
+
|
49
|
+
private
|
50
|
+
|
51
|
+
def context
|
52
|
+
[driver]
|
53
|
+
end
|
54
|
+
|
55
|
+
end
|
56
|
+
end
|
57
|
+
end
|
58
|
+
end
|
@@ -0,0 +1,96 @@
|
|
1
|
+
require 'ostruct'
|
2
|
+
|
3
|
+
module Crabfarm
|
4
|
+
module Engines
|
5
|
+
class SafeStateLoop
|
6
|
+
|
7
|
+
def initialize(_loader)
|
8
|
+
@context = _loader.load_context
|
9
|
+
@running = true
|
10
|
+
@working = false
|
11
|
+
@lock = Mutex.new
|
12
|
+
@thread = Thread.new { crawl_loop }
|
13
|
+
end
|
14
|
+
|
15
|
+
def release
|
16
|
+
@running = false
|
17
|
+
@thread.join
|
18
|
+
@context.release
|
19
|
+
end
|
20
|
+
|
21
|
+
def change_state(_name, _params={}, _wait=nil)
|
22
|
+
@lock.synchronize {
|
23
|
+
raise StillWorkingError.new if @working
|
24
|
+
@next_state_name = _name
|
25
|
+
@next_state_params = _params
|
26
|
+
@working = true
|
27
|
+
|
28
|
+
wait_and_load_struct _wait
|
29
|
+
}
|
30
|
+
end
|
31
|
+
|
32
|
+
def wait_for_state(_wait=nil)
|
33
|
+
@lock.synchronize {
|
34
|
+
wait_and_load_struct _wait
|
35
|
+
}
|
36
|
+
end
|
37
|
+
|
38
|
+
def cancel
|
39
|
+
@lock.synchronize {
|
40
|
+
if @working
|
41
|
+
@thread.terminate
|
42
|
+
@thread.join
|
43
|
+
@thread = Thread.new { crawl_loop }
|
44
|
+
@working = false
|
45
|
+
end
|
46
|
+
}
|
47
|
+
end
|
48
|
+
|
49
|
+
private
|
50
|
+
|
51
|
+
def wait_and_load_struct(_wait)
|
52
|
+
# need to use this method because mutex are not reentrant and monitors are slow.
|
53
|
+
wait_while_working _wait unless _wait.nil?
|
54
|
+
raise TimeoutError.new if @working
|
55
|
+
state_as_struct
|
56
|
+
end
|
57
|
+
|
58
|
+
def wait_while_working(_wait)
|
59
|
+
# TODO: use condition variables instead of wait loops
|
60
|
+
start = Time.now
|
61
|
+
while @working and Time.now - start < _wait.seconds
|
62
|
+
@lock.sleep 0.25
|
63
|
+
end
|
64
|
+
end
|
65
|
+
|
66
|
+
def state_as_struct
|
67
|
+
raise CrawlerError.new @error if @error
|
68
|
+
|
69
|
+
OpenStruct.new({
|
70
|
+
name: @state_name,
|
71
|
+
params: @state_params,
|
72
|
+
doc: @doc
|
73
|
+
})
|
74
|
+
end
|
75
|
+
|
76
|
+
def crawl_loop
|
77
|
+
while @running
|
78
|
+
if @working
|
79
|
+
begin
|
80
|
+
last_state = @context.run_state @next_state_name, @next_state_params
|
81
|
+
@doc = last_state.output.attributes!
|
82
|
+
@error = nil
|
83
|
+
rescue Exception => e
|
84
|
+
@doc = nil
|
85
|
+
@error = e
|
86
|
+
end
|
87
|
+
|
88
|
+
@state_name = @next_state_name
|
89
|
+
@state_params = @next_state_params
|
90
|
+
@working = false
|
91
|
+
else sleep 0.2 end
|
92
|
+
end
|
93
|
+
end
|
94
|
+
end
|
95
|
+
end
|
96
|
+
end
|
@@ -0,0 +1,50 @@
|
|
1
|
+
module Crabfarm
|
2
|
+
|
3
|
+
class Error < StandardError; end
|
4
|
+
|
5
|
+
class ConfigurationError < Error; end
|
6
|
+
|
7
|
+
class EntityNotFoundError < Error
|
8
|
+
attr_accessor :role, :name
|
9
|
+
|
10
|
+
def initialize(_role, _name)
|
11
|
+
super("The required #{_role} was not found (#{_name})")
|
12
|
+
@role = _role
|
13
|
+
@name = _name
|
14
|
+
end
|
15
|
+
end
|
16
|
+
|
17
|
+
class ApiError < Error
|
18
|
+
def code; 500 end
|
19
|
+
def to_json; {} end
|
20
|
+
end
|
21
|
+
|
22
|
+
class StillWorkingError < ApiError
|
23
|
+
def code; 409 end
|
24
|
+
end
|
25
|
+
|
26
|
+
class TimeoutError < ApiError
|
27
|
+
def code; 408 end
|
28
|
+
end
|
29
|
+
|
30
|
+
class CrawlerBaseError < ApiError
|
31
|
+
def initialize(_msg, _trace)
|
32
|
+
@exc = _msg
|
33
|
+
@trace = _trace
|
34
|
+
end
|
35
|
+
|
36
|
+
def to_json
|
37
|
+
{
|
38
|
+
exception: @exc,
|
39
|
+
backtrace: @trace
|
40
|
+
}
|
41
|
+
end
|
42
|
+
end
|
43
|
+
|
44
|
+
class CrawlerError < CrawlerBaseError
|
45
|
+
def initialize(_exc)
|
46
|
+
super _exc.to_s, _exc.backtrace
|
47
|
+
end
|
48
|
+
end
|
49
|
+
|
50
|
+
end
|
@@ -0,0 +1,83 @@
|
|
1
|
+
require 'active_support'
|
2
|
+
|
3
|
+
module Crabfarm
|
4
|
+
|
5
|
+
class Loader
|
6
|
+
|
7
|
+
attr_reader :module
|
8
|
+
|
9
|
+
def initialize(_base_path, _module_name=nil, &_config_block)
|
10
|
+
@path = _base_path
|
11
|
+
@name = _module_name
|
12
|
+
@config_block = _config_block
|
13
|
+
@module = if is_wrapped? then
|
14
|
+
"::#{@name}".constantize rescue nil
|
15
|
+
else nil end
|
16
|
+
end
|
17
|
+
|
18
|
+
def is_wrapped?
|
19
|
+
@name.present?
|
20
|
+
end
|
21
|
+
|
22
|
+
def is_loaded?
|
23
|
+
not @module.nil?
|
24
|
+
end
|
25
|
+
|
26
|
+
def load
|
27
|
+
crabfile = load_crabfile(@path)
|
28
|
+
@module = load_module(@name, File.join(@path, 'app'))
|
29
|
+
@module.send(:remove_const, :CF_CONFIG) rescue nil
|
30
|
+
@module.const_set :CF_CONFIG, crabfile
|
31
|
+
end
|
32
|
+
|
33
|
+
def load_context
|
34
|
+
load unless is_loaded?
|
35
|
+
Context.new @module
|
36
|
+
end
|
37
|
+
|
38
|
+
def unload
|
39
|
+
Object.send(:remove_const, @name) if is_wrapped?
|
40
|
+
@module = nil
|
41
|
+
end
|
42
|
+
|
43
|
+
private
|
44
|
+
|
45
|
+
def load_crabfile(_path)
|
46
|
+
crabfile = File.read(File.join(_path, 'Crabfile'))
|
47
|
+
config = Configuration.new
|
48
|
+
config.instance_eval crabfile
|
49
|
+
@config_block.call(config) unless @config_block.nil?
|
50
|
+
return config
|
51
|
+
end
|
52
|
+
|
53
|
+
def load_module(_name, _path)
|
54
|
+
require_all_as(_name, _path)
|
55
|
+
if is_wrapped? then "::#{_name}".constantize else Object end
|
56
|
+
end
|
57
|
+
|
58
|
+
def require_all_as(_name, _src_path)
|
59
|
+
loader_code = "
|
60
|
+
pending = Dir.glob('#{File.join(_src_path, '**/*')}').select { |f| f.end_with? '.rb' }.map { |f| f[0...-3] }
|
61
|
+
|
62
|
+
while pending.size > 0
|
63
|
+
new_pending = []
|
64
|
+
pending.each do |file|
|
65
|
+
begin
|
66
|
+
require file
|
67
|
+
rescue NameError => e
|
68
|
+
new_pending << file
|
69
|
+
end
|
70
|
+
end
|
71
|
+
|
72
|
+
require new_pending.first if new_pending.size == pending.size
|
73
|
+
pending = new_pending
|
74
|
+
end
|
75
|
+
"
|
76
|
+
|
77
|
+
loader_code = "module ::#{_name}; #{loader_code}; end" if _name.present?
|
78
|
+
Object.instance_eval loader_code
|
79
|
+
end
|
80
|
+
|
81
|
+
end
|
82
|
+
|
83
|
+
end
|