wayfarer 0.4.6 → 0.4.7
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.github/workflows/lint.yaml +25 -0
- data/.github/workflows/release.yaml +29 -0
- data/.github/workflows/tests.yaml +30 -0
- data/.gitignore +4 -0
- data/.rubocop.yml +5 -0
- data/.vale.ini +5 -0
- data/.yardopts +1 -3
- data/Dockerfile +5 -4
- data/Gemfile +3 -0
- data/Gemfile.lock +107 -102
- data/Rakefile +5 -56
- data/bin/wayfarer +1 -1
- data/docker-compose.yml +20 -9
- data/docs/cookbook/consent_screen.md +2 -2
- data/docs/cookbook/executing_javascript.md +3 -3
- data/docs/cookbook/navigation.md +12 -12
- data/docs/cookbook/querying_html.md +3 -3
- data/docs/cookbook/screenshots.md +2 -2
- data/docs/cookbook/user_agent.md +1 -1
- data/docs/design.md +36 -0
- data/docs/guides/callbacks.md +24 -126
- data/docs/guides/configuration.md +8 -8
- data/docs/guides/handlers.md +60 -0
- data/docs/guides/index.md +1 -0
- data/docs/guides/jobs/error_handling.md +40 -0
- data/docs/guides/jobs.md +99 -31
- data/docs/guides/navigation.md +1 -1
- data/docs/guides/networking/capybara.md +13 -22
- data/docs/guides/networking/custom_adapters.md +82 -41
- data/docs/guides/networking/ferrum.md +4 -4
- data/docs/guides/networking/http.md +9 -13
- data/docs/guides/networking/selenium.md +10 -11
- data/docs/guides/pages.md +76 -10
- data/docs/guides/redis.md +10 -0
- data/docs/guides/routing.md +74 -0
- data/docs/guides/tasks.md +33 -9
- data/docs/guides/tutorial.md +60 -0
- data/docs/guides/user_agents.md +113 -0
- data/docs/index.md +17 -40
- data/docs/reference/cli.md +35 -25
- data/docs/reference/configuration.md +36 -0
- data/lib/wayfarer/base.rb +124 -46
- data/lib/wayfarer/batch_completion.rb +56 -0
- data/lib/wayfarer/callbacks.rb +22 -48
- data/lib/wayfarer/cli/route_printer.rb +71 -57
- data/lib/wayfarer/cli.rb +121 -0
- data/lib/wayfarer/gc.rb +13 -6
- data/lib/wayfarer/handler.rb +15 -7
- data/lib/wayfarer/logging.rb +38 -0
- data/lib/wayfarer/middleware/base.rb +2 -0
- data/lib/wayfarer/middleware/batch_completion.rb +19 -0
- data/lib/wayfarer/middleware/content_type.rb +54 -0
- data/lib/wayfarer/middleware/controller.rb +19 -15
- data/lib/wayfarer/middleware/dedup.rb +16 -13
- data/lib/wayfarer/middleware/dispatch.rb +12 -4
- data/lib/wayfarer/middleware/normalize.rb +12 -11
- data/lib/wayfarer/middleware/redis.rb +15 -0
- data/lib/wayfarer/middleware/router.rb +33 -35
- data/lib/wayfarer/middleware/stage.rb +5 -5
- data/lib/wayfarer/middleware/uri_parser.rb +30 -0
- data/lib/wayfarer/middleware/user_agent.rb +49 -0
- data/lib/wayfarer/networking/capybara.rb +1 -1
- data/lib/wayfarer/networking/context.rb +2 -2
- data/lib/wayfarer/networking/ferrum.rb +2 -2
- data/lib/wayfarer/networking/follow.rb +12 -6
- data/lib/wayfarer/networking/http.rb +1 -1
- data/lib/wayfarer/networking/pool.rb +17 -12
- data/lib/wayfarer/networking/selenium.rb +3 -3
- data/lib/wayfarer/networking/strategy.rb +2 -2
- data/lib/wayfarer/page.rb +36 -14
- data/lib/wayfarer/parsing/xml.rb +6 -6
- data/lib/wayfarer/parsing.rb +24 -0
- data/lib/wayfarer/redis/barrier.rb +13 -21
- data/lib/wayfarer/redis/counter.rb +19 -9
- data/lib/wayfarer/redis/pool.rb +1 -1
- data/lib/wayfarer/redis/resettable.rb +19 -0
- data/lib/wayfarer/routing/dsl.rb +1 -0
- data/lib/wayfarer/routing/matchers/path.rb +4 -2
- data/lib/wayfarer/routing/root_route.rb +5 -1
- data/lib/wayfarer/routing/route.rb +4 -14
- data/lib/wayfarer/stringify.rb +22 -30
- data/lib/wayfarer/task.rb +12 -18
- data/lib/wayfarer.rb +28 -1
- data/mkdocs.yml +52 -7
- data/rake/docs.rake +26 -0
- data/rake/lint.rake +105 -0
- data/rake/release.rake +29 -0
- data/rake/tests.rake +28 -0
- data/requirements.txt +1 -1
- data/spec/base_spec.rb +140 -160
- data/spec/batch_completion_spec.rb +104 -0
- data/spec/cli/job_spec.rb +19 -23
- data/spec/cli/routing_spec.rb +101 -0
- data/spec/cli/version_spec.rb +1 -1
- data/spec/factories/task.rb +7 -1
- data/spec/fixtures/dummy_job.rb +5 -3
- data/spec/gc_spec.rb +8 -50
- data/spec/handler_spec.rb +1 -1
- data/spec/integration/callbacks_spec.rb +157 -45
- data/spec/integration/content_type_spec.rb +145 -0
- data/spec/integration/gc_spec.rb +44 -0
- data/spec/integration/handler_spec.rb +66 -0
- data/spec/integration/page_spec.rb +44 -29
- data/spec/integration/params_spec.rb +33 -25
- data/spec/integration/parsing_spec.rb +125 -0
- data/spec/integration/routing_spec.rb +18 -0
- data/spec/integration/stage_spec.rb +27 -20
- data/spec/middleware/batch_completion_spec.rb +34 -0
- data/spec/middleware/chain_spec.rb +8 -8
- data/spec/middleware/content_type_spec.rb +86 -0
- data/spec/middleware/controller_spec.rb +5 -5
- data/spec/middleware/dedup_spec.rb +38 -55
- data/spec/middleware/dispatch_spec.rb +23 -7
- data/spec/middleware/normalize_spec.rb +44 -13
- data/spec/middleware/router_spec.rb +29 -30
- data/spec/middleware/stage_spec.rb +8 -8
- data/spec/middleware/uri_parser_spec.rb +53 -0
- data/spec/middleware/{fetch_spec.rb → user_agent_spec.rb} +28 -27
- data/spec/networking/context_spec.rb +1 -1
- data/spec/networking/follow_spec.rb +2 -2
- data/spec/networking/pool_spec.rb +5 -5
- data/spec/networking/strategy.rb +2 -2
- data/spec/page_spec.rb +42 -20
- data/spec/parsing/xml_spec.rb +11 -12
- data/spec/redis/barrier_spec.rb +8 -48
- data/spec/redis/counter_spec.rb +13 -1
- data/spec/redis/pool_spec.rb +1 -1
- data/spec/spec_helpers.rb +27 -16
- data/spec/support/test_app.rb +8 -0
- data/spec/task_spec.rb +3 -24
- data/spec/wayfarer_spec.rb +1 -1
- data/wayfarer.gemspec +4 -3
- metadata +61 -51
- data/.github/workflows/ci.yaml +0 -32
- data/docs/guides/error_handling.md +0 -53
- data/docs/guides/networking.md +0 -94
- data/docs/guides/performance.md +0 -130
- data/docs/guides/reliability.md +0 -41
- data/docs/guides/routing/steering.md +0 -30
- data/docs/reference/api/base.md +0 -48
- data/docs/reference/configuration_keys.md +0 -43
- data/docs/reference/environment_variables.md +0 -83
- data/lib/wayfarer/cli/base.rb +0 -45
- data/lib/wayfarer/cli/generate.rb +0 -17
- data/lib/wayfarer/cli/job.rb +0 -56
- data/lib/wayfarer/cli/route.rb +0 -29
- data/lib/wayfarer/cli/runner.rb +0 -34
- data/lib/wayfarer/cli/templates/Gemfile.tt +0 -5
- data/lib/wayfarer/cli/templates/job.rb.tt +0 -10
- data/lib/wayfarer/config/capybara.rb +0 -10
- data/lib/wayfarer/config/ferrum.rb +0 -11
- data/lib/wayfarer/config/networking.rb +0 -29
- data/lib/wayfarer/config/redis.rb +0 -14
- data/lib/wayfarer/config/root.rb +0 -11
- data/lib/wayfarer/config/selenium.rb +0 -21
- data/lib/wayfarer/config/strconv.rb +0 -45
- data/lib/wayfarer/config/struct.rb +0 -72
- data/lib/wayfarer/middleware/fetch.rb +0 -56
- data/lib/wayfarer/redis/connection.rb +0 -13
- data/lib/wayfarer/redis/version.rb +0 -19
- data/lib/wayfarer/routing/router.rb +0 -28
- data/spec/callbacks_spec.rb +0 -102
- data/spec/cli/generate_spec.rb +0 -39
- data/spec/config/capybara_spec.rb +0 -18
- data/spec/config/ferrum_spec.rb +0 -24
- data/spec/config/networking_spec.rb +0 -73
- data/spec/config/redis_spec.rb +0 -32
- data/spec/config/root_spec.rb +0 -31
- data/spec/config/selenium_spec.rb +0 -56
- data/spec/config/strconv_spec.rb +0 -58
- data/spec/config/struct_spec.rb +0 -66
- data/spec/integration/steering_spec.rb +0 -57
- data/spec/redis/version_spec.rb +0 -13
- data/spec/routing/router_spec.rb +0 -24
@@ -3,85 +3,68 @@
|
|
3
3
|
require "spec_helpers"
|
4
4
|
|
5
5
|
describe Wayfarer::Middleware::Dedup, redis: true do
|
6
|
-
let(:task) { build(:task) }
|
6
|
+
let(:task) { build(:task, :redis_pool) }
|
7
|
+
let(:uri) { Addressable::URI.parse(task.url) }
|
8
|
+
let(:executions) { 1 }
|
9
|
+
let(:job) { double(executions: executions) }
|
7
10
|
subject { described_class.new }
|
8
11
|
|
9
|
-
before
|
12
|
+
before do
|
13
|
+
task[:job] = job
|
14
|
+
task[:controller] = job
|
15
|
+
task[:normalized_url] = task.url
|
16
|
+
end
|
10
17
|
|
11
18
|
describe "#call" do
|
12
|
-
|
13
|
-
|
19
|
+
it "assigns barrier" do
|
20
|
+
expect { subject.call(task) }.to change { task[:barrier] }.from(nil).to(instance_of(Wayfarer::Redis::Barrier))
|
21
|
+
end
|
14
22
|
|
15
|
-
|
16
|
-
|
17
|
-
|
23
|
+
it "assigns barrier for batch" do
|
24
|
+
subject.call(task)
|
25
|
+
|
26
|
+
expect(task[:barrier].task).to be(task)
|
27
|
+
end
|
18
28
|
|
19
|
-
|
29
|
+
context "with retry" do
|
30
|
+
let(:executions) { 2 }
|
31
|
+
|
32
|
+
specify do
|
20
33
|
expect { |spy| subject.call(task, &spy) }.to yield_control
|
21
34
|
end
|
22
|
-
end
|
23
35
|
|
24
|
-
|
25
|
-
|
26
|
-
|
36
|
+
specify do
|
37
|
+
expect(Wayfarer::Logging.logger).to receive(:add).with(Logger::DEBUG, "Not deduplicating retry")
|
38
|
+
|
39
|
+
subject.call(task)
|
27
40
|
end
|
28
41
|
end
|
29
42
|
|
30
|
-
context "
|
31
|
-
|
32
|
-
expect {
|
33
|
-
subject.call(task)
|
34
|
-
}.to change { task.barrier.seen?(task.url) }.to(true)
|
35
|
-
end
|
43
|
+
context "when rerouted" do
|
44
|
+
before { task[:controller] = Object.new }
|
36
45
|
|
37
|
-
|
46
|
+
specify do
|
38
47
|
expect { |spy| subject.call(task, &spy) }.to yield_control
|
39
48
|
end
|
40
|
-
end
|
41
49
|
|
42
|
-
|
43
|
-
|
50
|
+
specify do
|
51
|
+
expect(Wayfarer::Logging.logger).to receive(:add).with(Logger::DEBUG, "Not deduplicating rerouted task")
|
44
52
|
|
45
|
-
|
46
|
-
expect { |spy| subject.call(task, &spy) }.not_to yield_control
|
53
|
+
subject.call(task)
|
47
54
|
end
|
48
55
|
end
|
49
56
|
|
50
|
-
context "with
|
51
|
-
|
52
|
-
|
53
|
-
subject.call(task) { raise }
|
54
|
-
rescue StandardError
|
55
|
-
nil
|
56
|
-
end
|
57
|
-
expect(task.barrier.seen?(task.url)).to be(false)
|
58
|
-
end
|
59
|
-
|
60
|
-
it "re-raises the exception" do
|
61
|
-
expect {
|
62
|
-
subject.call(task) { raise }
|
63
|
-
}.to raise_error(RuntimeError)
|
57
|
+
context "with unchecked URL" do
|
58
|
+
specify do
|
59
|
+
expect { |spy| subject.call(task, &spy) }.to yield_control
|
64
60
|
end
|
65
61
|
end
|
66
62
|
|
67
|
-
|
68
|
-
|
69
|
-
let(:unseen_urls) { %w[https://w3c.org https://nasa.gov] }
|
63
|
+
context "with checked URL" do
|
64
|
+
before { Wayfarer::Redis::Barrier.new(task).check!(task[:normalized_url]) }
|
70
65
|
|
71
|
-
|
72
|
-
|
73
|
-
task.barrier.seen?(url)
|
74
|
-
end
|
75
|
-
|
76
|
-
[*seen_urls, *unseen_urls].each do |url|
|
77
|
-
task.metadata.staged_urls.add(url)
|
78
|
-
end
|
79
|
-
end
|
80
|
-
|
81
|
-
it "filters seen staged URLs" do
|
82
|
-
expect {
|
83
|
-
subject.call(task)
|
84
|
-
}.to change { task.metadata.staged_urls }.to(SortedSet.new(unseen_urls))
|
66
|
+
specify do
|
67
|
+
expect { |spy| subject.call(task, &spy) }.not_to yield_control
|
85
68
|
end
|
86
69
|
end
|
87
70
|
end
|
@@ -8,27 +8,27 @@ describe Wayfarer::Middleware::Dispatch do
|
|
8
8
|
subject(:chain) { described_class.new }
|
9
9
|
|
10
10
|
before do
|
11
|
-
task
|
12
|
-
task
|
11
|
+
task[:controller] = spy
|
12
|
+
task[:action] = action
|
13
13
|
|
14
|
-
allow(task
|
14
|
+
allow(task[:controller]).to receive(:run_callbacks).and_yield
|
15
15
|
end
|
16
16
|
|
17
17
|
describe "#call" do
|
18
18
|
it "runs callbacks" do
|
19
|
-
expect(task
|
19
|
+
expect(task[:controller]).to receive(:run_callbacks).with(action)
|
20
20
|
subject.call(task)
|
21
21
|
end
|
22
22
|
|
23
23
|
context "when action is a Symbol" do
|
24
24
|
it "calls the method" do
|
25
|
-
expect(task
|
25
|
+
expect(task[:controller]).to receive(action)
|
26
26
|
subject.call(task)
|
27
27
|
end
|
28
28
|
end
|
29
29
|
|
30
|
-
context "with
|
31
|
-
let(:action) { Class.new }
|
30
|
+
context "with handler" do
|
31
|
+
let(:action) { Class.new.include(Wayfarer::Handler) }
|
32
32
|
|
33
33
|
it "instantiates and calls" do
|
34
34
|
expect_any_instance_of(action).to receive(:call).with(task)
|
@@ -36,6 +36,22 @@ describe Wayfarer::Middleware::Dispatch do
|
|
36
36
|
end
|
37
37
|
end
|
38
38
|
|
39
|
+
context "without action" do
|
40
|
+
let(:action) { nil }
|
41
|
+
|
42
|
+
it "instantiates and calls" do
|
43
|
+
expect { subject.call(task) }.to raise_error(ArgumentError)
|
44
|
+
end
|
45
|
+
end
|
46
|
+
|
47
|
+
context "without other action" do
|
48
|
+
let(:action) { Class.new }
|
49
|
+
|
50
|
+
it "instantiates and calls" do
|
51
|
+
expect { subject.call(task) }.to raise_error(ArgumentError)
|
52
|
+
end
|
53
|
+
end
|
54
|
+
|
39
55
|
it "yields" do
|
40
56
|
expect { |spy| subject.call(task, &spy) }.to yield_control
|
41
57
|
end
|
@@ -4,26 +4,57 @@ require "spec_helpers"
|
|
4
4
|
|
5
5
|
describe Wayfarer::Middleware::Normalize do
|
6
6
|
let(:task) { build(:task) }
|
7
|
+
let(:uri) { Addressable::URI.parse(url) }
|
7
8
|
subject { described_class.new }
|
8
9
|
|
9
|
-
describe "
|
10
|
-
|
11
|
-
|
12
|
-
|
13
|
-
|
14
|
-
|
15
|
-
|
10
|
+
describe "::normalize" do
|
11
|
+
subject { described_class.normalize(uri).to_s }
|
12
|
+
|
13
|
+
context "without HTTP(S) URL" do
|
14
|
+
let(:url) { "localhost:3000" }
|
15
|
+
|
16
|
+
it { is_expected.to eq(url) }
|
16
17
|
end
|
17
18
|
|
18
|
-
|
19
|
+
context "with HTTP URL" do
|
20
|
+
let(:url) { "http://example.com" }
|
19
21
|
|
20
|
-
|
21
|
-
expect { |spy| subject.call(task, &spy) }.to yield_control
|
22
|
+
it { is_expected.to eq("http://example.com/") }
|
22
23
|
end
|
24
|
+
end
|
25
|
+
|
26
|
+
describe "#call" do
|
27
|
+
let(:url) { task.url }
|
28
|
+
before { task[:uri] = uri }
|
29
|
+
|
30
|
+
context "when already assigned" do
|
31
|
+
before { task[:normalized_url] = Object.new }
|
32
|
+
|
33
|
+
specify do
|
34
|
+
expect { |spy| subject.call(task, &spy) }.to yield_control
|
35
|
+
end
|
36
|
+
|
37
|
+
it "doesn't normalize URL" do
|
38
|
+
expect { subject.call(task) }.not_to(change { task[:normalized_url] })
|
39
|
+
end
|
40
|
+
end
|
41
|
+
|
42
|
+
context "with invalid URL" do
|
43
|
+
before do
|
44
|
+
# I can't come up with a URL that normalize_url raises on but
|
45
|
+
# Addressable::URI doesn't, hence the stub
|
46
|
+
allow(described_class).to receive(:normalize).with(uri).and_return(nil)
|
47
|
+
end
|
48
|
+
|
49
|
+
specify do
|
50
|
+
expect { |spy| subject.call(task, &spy) }.not_to yield_control
|
51
|
+
end
|
52
|
+
|
53
|
+
specify do
|
54
|
+
expect(Wayfarer::Logging.logger).to receive(:add).with(Logger::INFO, "Failed to normalize HTTP(S) URL")
|
23
55
|
|
24
|
-
|
25
|
-
|
26
|
-
expect(task.metadata.staged_urls).to eq(SortedSet[urls.first])
|
56
|
+
subject.call(task)
|
57
|
+
end
|
27
58
|
end
|
28
59
|
end
|
29
60
|
end
|
@@ -12,10 +12,11 @@ describe Wayfarer::Middleware::Router do
|
|
12
12
|
end
|
13
13
|
|
14
14
|
before do
|
15
|
-
allow(controller.class.
|
16
|
-
.with(Addressable::URI.parse(task.url)
|
15
|
+
allow(controller.class.route).to receive(:invoke)
|
16
|
+
.with(Addressable::URI.parse(task.url))
|
17
17
|
.and_return(result)
|
18
|
-
task
|
18
|
+
task[:controller] = controller
|
19
|
+
task[:uri] = Addressable::URI.parse(task.url)
|
19
20
|
end
|
20
21
|
|
21
22
|
context "with matching route" do
|
@@ -28,20 +29,26 @@ describe Wayfarer::Middleware::Router do
|
|
28
29
|
it "assigns the action" do
|
29
30
|
expect {
|
30
31
|
subject.call(task)
|
31
|
-
}.to change { task
|
32
|
+
}.to change { task[:action] }.to(action)
|
32
33
|
end
|
33
34
|
|
34
35
|
it "merges params" do
|
35
|
-
task
|
36
|
+
task[:params] = ActiveSupport::HashWithIndifferentAccess.new("bar" => "qux")
|
36
37
|
|
37
38
|
expect {
|
38
39
|
subject.call(task)
|
39
|
-
}.to change { task
|
40
|
+
}.to change { task[:params] }.to("foo" => "bar", "bar" => "qux")
|
40
41
|
end
|
41
42
|
|
42
|
-
|
43
|
+
specify do
|
43
44
|
expect { |spy| subject.call(task, &spy) }.to yield_control
|
44
45
|
end
|
46
|
+
|
47
|
+
specify do
|
48
|
+
expect(Wayfarer::Logging.logger).to receive(:add).with(Logger::INFO, kind_of(String))
|
49
|
+
|
50
|
+
subject.call(task)
|
51
|
+
end
|
45
52
|
end
|
46
53
|
|
47
54
|
context "without matching route" do
|
@@ -52,20 +59,26 @@ describe Wayfarer::Middleware::Router do
|
|
52
59
|
it "does not assign the action" do
|
53
60
|
expect {
|
54
61
|
subject.call(task)
|
55
|
-
}.not_to(change { task
|
62
|
+
}.not_to(change { task[:action] })
|
56
63
|
end
|
57
64
|
|
58
65
|
it "does not alter params" do
|
59
|
-
task
|
66
|
+
task[:params] = ActiveSupport::HashWithIndifferentAccess.new("bar" => "qux")
|
60
67
|
|
61
68
|
expect {
|
62
69
|
subject.call(task)
|
63
|
-
}.not_to(change { task
|
70
|
+
}.not_to(change { task[:params] })
|
64
71
|
end
|
65
72
|
|
66
|
-
|
73
|
+
specify do
|
67
74
|
expect { |spy| subject.call(task, &spy) }.not_to yield_control
|
68
75
|
end
|
76
|
+
|
77
|
+
specify do
|
78
|
+
expect(Wayfarer::Logging.logger).to receive(:add).with(Logger::INFO, kind_of(String))
|
79
|
+
|
80
|
+
subject.call(task)
|
81
|
+
end
|
69
82
|
end
|
70
83
|
end
|
71
84
|
|
@@ -74,31 +87,17 @@ describe Wayfarer::Middleware::Router do
|
|
74
87
|
Struct.new(:task).include(described_class).new(task)
|
75
88
|
end
|
76
89
|
|
77
|
-
describe "::
|
78
|
-
it "returns a
|
79
|
-
expect(controller.class.
|
90
|
+
describe "::route" do
|
91
|
+
it "returns a root route" do
|
92
|
+
expect(controller.class.route).to be_a(Wayfarer::Routing::RootRoute)
|
80
93
|
end
|
81
94
|
end
|
82
95
|
|
83
96
|
describe "::route" do
|
84
97
|
it "adds a routing block" do
|
85
98
|
expect {
|
86
|
-
controller.class.route
|
87
|
-
}.to change { controller.class.
|
88
|
-
end
|
89
|
-
end
|
90
|
-
|
91
|
-
describe "::steer" do
|
92
|
-
it "overrides #steer" do
|
93
|
-
expect {
|
94
|
-
controller.class.steer { :foobar }
|
95
|
-
}.to change { controller.steer }.from([]).to(:foobar)
|
96
|
-
end
|
97
|
-
end
|
98
|
-
|
99
|
-
describe "#steer" do
|
100
|
-
it "returns [] by default" do
|
101
|
-
expect(controller.steer).to eq([])
|
99
|
+
controller.class.route.to(:index)
|
100
|
+
}.to change { controller.class.route.children.count }.by(1)
|
102
101
|
end
|
103
102
|
end
|
104
103
|
end
|
@@ -9,7 +9,7 @@ describe Wayfarer::Middleware::Stage do
|
|
9
9
|
describe "#call" do
|
10
10
|
it "assigns an empty set" do
|
11
11
|
subject.call(task)
|
12
|
-
expect(task
|
12
|
+
expect(task[:staged_urls]).to eq(SortedSet.new)
|
13
13
|
end
|
14
14
|
|
15
15
|
it "yields" do
|
@@ -22,20 +22,20 @@ describe Wayfarer::Middleware::Stage do
|
|
22
22
|
spy.tap do |job|
|
23
23
|
expect(job).to receive(:crawl).with(urls.first, batch: task.batch).ordered
|
24
24
|
expect(job).to receive(:crawl).with(urls.second, batch: task.batch).ordered
|
25
|
-
task
|
25
|
+
task[:job] = double(class: job)
|
26
26
|
end
|
27
27
|
|
28
28
|
subject.call(task) do
|
29
|
-
task
|
29
|
+
task[:staged_urls] = SortedSet.new(urls)
|
30
30
|
end
|
31
31
|
end
|
32
32
|
|
33
33
|
it "resets staged URLs" do
|
34
|
-
task
|
34
|
+
task[:staged_urls] = SortedSet.new([test_app_path("/foo")])
|
35
35
|
|
36
36
|
expect {
|
37
37
|
subject.call(task)
|
38
|
-
}.to change { task
|
38
|
+
}.to change { task[:staged_urls].count }.to(0)
|
39
39
|
end
|
40
40
|
end
|
41
41
|
|
@@ -45,17 +45,17 @@ describe Wayfarer::Middleware::Stage do
|
|
45
45
|
end
|
46
46
|
|
47
47
|
describe "#stage" do
|
48
|
-
before { task
|
48
|
+
before { task[:staged_urls] = SortedSet.new }
|
49
49
|
|
50
50
|
it "stages URLs" do
|
51
51
|
expect {
|
52
52
|
controller.stage(test_app_path("/foo"))
|
53
|
-
}.to change { task
|
53
|
+
}.to change { task[:staged_urls].count }.by(1)
|
54
54
|
end
|
55
55
|
|
56
56
|
it "converts to strings" do
|
57
57
|
controller.stage(Addressable::URI.parse(test_app_path("/foo")))
|
58
|
-
expect(task
|
58
|
+
expect(task[:staged_urls].to_a.first).to be_a(String)
|
59
59
|
end
|
60
60
|
end
|
61
61
|
end
|
@@ -0,0 +1,53 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require "spec_helpers"
|
4
|
+
|
5
|
+
describe Wayfarer::Middleware::UriParser, "#call" do
|
6
|
+
let(:task) { build(:task) }
|
7
|
+
subject { described_class.new }
|
8
|
+
|
9
|
+
it "parses URLs" do
|
10
|
+
expect { subject.call(task) }.to change { task[:uri] }.to(Addressable::URI.parse(task.url))
|
11
|
+
end
|
12
|
+
|
13
|
+
it "normalizes URLs" do
|
14
|
+
task = build(:task, url: "http://example.com")
|
15
|
+
|
16
|
+
expect { subject.call(task) }.to change { task[:uri].to_s }.to("http://example.com/")
|
17
|
+
end
|
18
|
+
|
19
|
+
specify do
|
20
|
+
expect { |spy| subject.call(task, &spy) }.to yield_control
|
21
|
+
end
|
22
|
+
|
23
|
+
context "with invalid URL" do
|
24
|
+
let(:task) { build(:task, url: "ht%0atp://localhost/") }
|
25
|
+
|
26
|
+
specify do
|
27
|
+
expect { |spy| subject.call(task, &spy) }.not_to yield_control
|
28
|
+
end
|
29
|
+
|
30
|
+
specify do
|
31
|
+
expect(Wayfarer::Logging.logger)
|
32
|
+
.to receive(:add).with(Logger::INFO, "Not processing invalid URL (Invalid scheme format: 'ht%0atp')")
|
33
|
+
|
34
|
+
subject.call(task)
|
35
|
+
end
|
36
|
+
end
|
37
|
+
|
38
|
+
describe described_class::API do
|
39
|
+
subject(:controller) do
|
40
|
+
Struct.new(:task).include(described_class).new(task)
|
41
|
+
end
|
42
|
+
|
43
|
+
describe "#uri" do
|
44
|
+
let(:uri) { Addressable::URI.parse(task.url) }
|
45
|
+
|
46
|
+
before { task[:uri] = uri }
|
47
|
+
|
48
|
+
it "returns the agent" do
|
49
|
+
expect(controller.uri).to be(uri)
|
50
|
+
end
|
51
|
+
end
|
52
|
+
end
|
53
|
+
end
|
@@ -2,7 +2,7 @@
|
|
2
2
|
|
3
3
|
require "spec_helpers"
|
4
4
|
|
5
|
-
describe Wayfarer::Middleware::
|
5
|
+
describe Wayfarer::Middleware::UserAgent do
|
6
6
|
let(:task) { build(:task) }
|
7
7
|
let(:page) { Object.new }
|
8
8
|
let(:agent) { Object.new }
|
@@ -28,17 +28,17 @@ describe Wayfarer::Middleware::Fetch do
|
|
28
28
|
allow(subject).to receive(:pool).and_return(pool)
|
29
29
|
end
|
30
30
|
|
31
|
-
task
|
32
|
-
task
|
31
|
+
task[:staged_urls] = SortedSet.new
|
32
|
+
task[:controller] = controller
|
33
33
|
end
|
34
34
|
|
35
35
|
context "with page assigned" do
|
36
|
-
before { task
|
36
|
+
before { task[:page] = page }
|
37
37
|
|
38
38
|
it "does not alter the page" do
|
39
39
|
expect {
|
40
40
|
subject.call(task)
|
41
|
-
}.not_to(change { task
|
41
|
+
}.not_to(change { task[:page] })
|
42
42
|
end
|
43
43
|
|
44
44
|
it "yields" do
|
@@ -63,7 +63,7 @@ describe Wayfarer::Middleware::Fetch do
|
|
63
63
|
it "stages the redirect URL" do
|
64
64
|
expect {
|
65
65
|
subject.call(task)
|
66
|
-
}.to change { task
|
66
|
+
}.to change { task[:staged_urls].count }.by(1)
|
67
67
|
end
|
68
68
|
|
69
69
|
it "does not yield" do
|
@@ -75,13 +75,13 @@ describe Wayfarer::Middleware::Fetch do
|
|
75
75
|
it "assigns the context" do
|
76
76
|
expect {
|
77
77
|
subject.call(task)
|
78
|
-
}.to change { task
|
78
|
+
}.to change { task[:context] }.to(context)
|
79
79
|
end
|
80
80
|
|
81
81
|
it "assigns the page" do
|
82
82
|
expect {
|
83
83
|
subject.call(task)
|
84
|
-
}.to change { task
|
84
|
+
}.to change { task[:page] }.to(result.page)
|
85
85
|
end
|
86
86
|
|
87
87
|
it "yields" do
|
@@ -95,31 +95,23 @@ describe Wayfarer::Middleware::Fetch do
|
|
95
95
|
Struct.new(:task).include(described_class).new(task)
|
96
96
|
end
|
97
97
|
|
98
|
-
describe "#
|
99
|
-
before { task
|
98
|
+
describe "#user_agent" do
|
99
|
+
before { task[:context] = context }
|
100
100
|
|
101
101
|
it "returns the agent" do
|
102
|
-
expect(controller.
|
103
|
-
end
|
104
|
-
end
|
105
|
-
|
106
|
-
describe "#context" do
|
107
|
-
before { task.metadata.context = context }
|
108
|
-
|
109
|
-
it "returns the context" do
|
110
|
-
expect(controller.context).to be(task.metadata.context)
|
102
|
+
expect(controller.user_agent).to be(context.instance)
|
111
103
|
end
|
112
104
|
end
|
113
105
|
|
114
106
|
describe "#page" do
|
115
|
-
before { task
|
107
|
+
before { task[:page] = page }
|
116
108
|
|
117
109
|
it "returns the page" do
|
118
|
-
expect(controller.page).to be(task
|
110
|
+
expect(controller.page).to be(task[:page])
|
119
111
|
end
|
120
112
|
|
121
113
|
context "with live keyword" do
|
122
|
-
before { task
|
114
|
+
before { task[:context] = context }
|
123
115
|
|
124
116
|
context "with stateful agent" do
|
125
117
|
before do
|
@@ -130,7 +122,7 @@ describe Wayfarer::Middleware::Fetch do
|
|
130
122
|
it "replaces the page" do
|
131
123
|
expect {
|
132
124
|
controller.page(live: true)
|
133
|
-
}.to change { task
|
125
|
+
}.to change { task[:page] }.to(result.page)
|
134
126
|
end
|
135
127
|
end
|
136
128
|
|
@@ -140,15 +132,24 @@ describe Wayfarer::Middleware::Fetch do
|
|
140
132
|
it "does not alter the page" do
|
141
133
|
expect {
|
142
134
|
controller.page(live: true)
|
143
|
-
}.not_to(change { task
|
135
|
+
}.not_to(change { task[:page] })
|
144
136
|
end
|
145
137
|
end
|
146
138
|
end
|
147
139
|
end
|
148
140
|
|
149
|
-
describe "#
|
150
|
-
|
151
|
-
|
141
|
+
describe "#fetch" do
|
142
|
+
let(:url) { test_app_path("/redirect?times=3") }
|
143
|
+
subject { controller.fetch(url) }
|
144
|
+
|
145
|
+
it { is_expected.to be_a(Wayfarer::Page) }
|
146
|
+
|
147
|
+
context "with reries exhausted" do
|
148
|
+
let(:url) { test_app_path("/redirect?times=4") }
|
149
|
+
|
150
|
+
specify do
|
151
|
+
expect { subject }.to raise_error(Wayfarer::Networking::Follow::RedirectsExhaustedError)
|
152
|
+
end
|
152
153
|
end
|
153
154
|
end
|
154
155
|
end
|
@@ -49,7 +49,7 @@ describe Wayfarer::Networking::Context do
|
|
49
49
|
let(:other_error) { Class.new(StandardError) }
|
50
50
|
|
51
51
|
before do
|
52
|
-
Wayfarer.config
|
52
|
+
Wayfarer.config[:network][:renew_on] = [other_error]
|
53
53
|
allow(strategy).to receive(:fetch).and_raise(other_error)
|
54
54
|
end
|
55
55
|
|
@@ -16,7 +16,7 @@ describe Wayfarer::Networking::Follow do
|
|
16
16
|
before { allow(inner).to receive(:fetch).and_return(success) }
|
17
17
|
|
18
18
|
it "returns the page" do
|
19
|
-
expect(outer.fetch(url)).to be(page)
|
19
|
+
expect(outer.fetch(url, follow: 3)).to be(page)
|
20
20
|
end
|
21
21
|
end
|
22
22
|
|
@@ -34,7 +34,7 @@ describe Wayfarer::Networking::Follow do
|
|
34
34
|
|
35
35
|
expect {
|
36
36
|
outer.fetch(url, follow: 0)
|
37
|
-
}.to raise_error(
|
37
|
+
}.to raise_error(described_class::RedirectsExhaustedError)
|
38
38
|
end
|
39
39
|
end
|
40
40
|
end
|
@@ -17,7 +17,7 @@ describe Wayfarer::Networking::Pool do
|
|
17
17
|
end
|
18
18
|
|
19
19
|
context "when using Ferrum", ferrum: true do
|
20
|
-
before { Wayfarer.config
|
20
|
+
before { Wayfarer.config[:network][:agent] = :ferrum }
|
21
21
|
|
22
22
|
it "yields Ferrum" do
|
23
23
|
pool.with do |context|
|
@@ -27,7 +27,7 @@ describe Wayfarer::Networking::Pool do
|
|
27
27
|
end
|
28
28
|
|
29
29
|
context "when using Selenium", selenium: true do
|
30
|
-
before { Wayfarer.config
|
30
|
+
before { Wayfarer.config[:network][:agent] = :selenium }
|
31
31
|
|
32
32
|
it "yields Selenium" do
|
33
33
|
pool.with do |context|
|
@@ -38,8 +38,8 @@ describe Wayfarer::Networking::Pool do
|
|
38
38
|
|
39
39
|
context "when using Capybara", ferrum: true do
|
40
40
|
before do
|
41
|
-
Wayfarer.config
|
42
|
-
Wayfarer.config
|
41
|
+
Wayfarer.config[:network][:agent] = :capybara
|
42
|
+
Wayfarer.config[:capybara][:driver] = :cuprite
|
43
43
|
end
|
44
44
|
|
45
45
|
it "yields Capybara" do
|
@@ -55,7 +55,7 @@ describe Wayfarer::Networking::Pool do
|
|
55
55
|
|
56
56
|
before do
|
57
57
|
pool.class.registry[:foobar] = double(new: strategy)
|
58
|
-
Wayfarer.config
|
58
|
+
Wayfarer.config[:network][:agent] = :foobar
|
59
59
|
end
|
60
60
|
|
61
61
|
it "destroys the strategy" do
|