Class: Ace::Test::EndToEndRunner::Molecules::PipelineExecutor

Inherits:
Object
  • Object
show all
Defined in:
lib/ace/test/end_to_end_runner/molecules/pipeline_executor.rb

Overview

Executes standalone scenarios using the deterministic pipeline.

Constant Summary collapse

AMBIENT_TMUX_ENV_VARS =
%w[TMUX TMUX_PANE].freeze

Instance Method Summary collapse

Constructor Details

#initialize(provider:, verifier_provider: nil, timeout:, sandbox_builder: nil, prompt_bundler: nil, report_generator: nil, sandbox_backend_factory: nil) ⇒ PipelineExecutor

Returns a new instance of PipelineExecutor.

Parameters:



23
24
25
26
27
28
29
30
31
32
33
34
# File 'lib/ace/test/end_to_end_runner/molecules/pipeline_executor.rb', line 23

def initialize(provider:, verifier_provider: nil, timeout:, sandbox_builder: nil, prompt_bundler: nil,
  report_generator: nil, sandbox_backend_factory: nil)
  @provider = provider
  @verifier_provider = verifier_provider || provider
  @timeout = timeout
  @sandbox_builder = sandbox_builder || PipelineSandboxBuilder.new
  @prompt_bundler = prompt_bundler || PipelinePromptBundler.new
  @report_generator = report_generator || PipelineReportGenerator.new
  @sandbox_backend_factory = sandbox_backend_factory || lambda { |sandbox_path, source_root: nil|
    Molecules::BwrapSandboxBackend.new(sandbox_root: sandbox_path, source_root: source_root)
  }
end

Instance Method Details

#execute(scenario:, cli_args:, sandbox_path:, report_dir:, env_vars: nil, test_cases: nil) ⇒ Models::TestResult

Parameters:

  • scenario (Models::TestScenario)
  • cli_args (String, nil)
  • sandbox_path (String)
  • report_dir (String)
  • env_vars (Hash, nil) (defaults to: nil)
  • test_cases (Array<String>, nil) (defaults to: nil)

Returns:



43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
# File 'lib/ace/test/end_to_end_runner/molecules/pipeline_executor.rb', line 43

def execute(scenario:, cli_args:, sandbox_path:, report_dir:, env_vars: nil, test_cases: nil)
  started_at = Time.now
  FileUtils.mkdir_p(report_dir)
  write_command_record(report_dir, "runner", provider: @provider, cli_args: cli_args)
  write_tc_manifests(report_dir, scenario, test_cases: test_cases)

  build_env = if prepared_sandbox?(sandbox_path, env_vars)
    @sandbox_builder.prepare_existing_sandbox(
      scenario: scenario,
      sandbox_path: sandbox_path,
      test_cases: test_cases
    )
  else
    @sandbox_builder.build(
      scenario: scenario,
      sandbox_path: sandbox_path,
      test_cases: test_cases
    )
  end
  merged_env = sanitize_subprocess_env((env_vars || {}).merge(build_env))
  sandbox_backend = @sandbox_backend_factory.call(
    sandbox_path,
    source_root: merged_env["ACE_E2E_SOURCE_ROOT"] || merged_env[:ACE_E2E_SOURCE_ROOT]
  )
  merged_env = sandbox_backend.prepared_env(merged_env)

  runner = @prompt_bundler.prepare_runner(
    scenario: scenario,
    sandbox_path: sandbox_path,
    test_cases: test_cases,
    artifact_contract: declared_artifact_contract(scenario, test_cases: test_cases)
  )
  runner_response = run_llm(
    prompt_path: runner[:prompt_path],
    system_path: runner[:system_path],
    output_path: runner[:output_path],
    cli_args: cli_args,
    env_vars: merged_env,
    subprocess_command_prefix: sandbox_backend.command_prefix(chdir: sandbox_path, env: merged_env),
    provider: @provider,
    fallback: false
  )
  runner_observations = extract_runner_observations(runner_response[:text])
  initial_artifact_contract = snapshot_artifacts(
    report_dir,
    sandbox_path,
    scenario,
    test_cases: test_cases,
    snapshot_name: "artifact-snapshot.initial.json"
  )
  artifact_contract = initial_artifact_contract

  if missing_required_artifacts?(artifact_contract)
    write_command_record(report_dir, "runner-repair", provider: @provider, cli_args: cli_args)
    repair_runner = @prompt_bundler.prepare_runner(
      scenario: scenario,
      sandbox_path: sandbox_path,
      test_cases: test_cases,
      artifact_contract: artifact_contract,
      repair_mode: true
    )
    repair_response = run_llm(
      prompt_path: repair_runner[:prompt_path],
      system_path: repair_runner[:system_path],
      output_path: repair_runner[:output_path],
      cli_args: cli_args,
      env_vars: merged_env,
      subprocess_command_prefix: sandbox_backend.command_prefix(chdir: sandbox_path, env: merged_env),
      provider: @provider,
      fallback: false
    )
    repair_observations = extract_runner_observations(repair_response[:text])
    runner_observations = merge_runner_observations(runner_observations, repair_observations)
    artifact_contract = snapshot_artifacts(report_dir, sandbox_path, scenario, test_cases: test_cases)
  else
    write_artifact_snapshot(report_dir, "artifact-snapshot.json", artifact_contract)
  end

  verifier = @prompt_bundler.prepare_verifier(
    scenario: scenario,
    sandbox_path: sandbox_path,
    test_cases: test_cases,
    runner_observations: runner_observations,
    artifact_contract: artifact_contract
  )
  write_command_record(report_dir, "verifier", provider: @verifier_provider, cli_args: cli_args)
  verifier_response = run_llm(
    prompt_path: verifier[:prompt_path],
    system_path: verifier[:system_path],
    output_path: verifier[:output_path],
    cli_args: cli_args,
    env_vars: merged_env,
    subprocess_command_prefix: sandbox_backend.command_prefix(chdir: sandbox_path, env: merged_env),
    provider: @verifier_provider,
    fallback: query_fallback_for(@verifier_provider)
  )

  @report_generator.generate(
    scenario: scenario,
    verifier_output: verifier_response[:text],
    report_dir: report_dir,
    provider: @verifier_provider,
    started_at: started_at,
    completed_at: Time.now,
    metadata: (
      report_dir,
      runner_observations: runner_observations,
      artifact_contract: artifact_contract,
      initial_artifact_contract: initial_artifact_contract
    )
  )
rescue => e
  begin
    @report_generator.write_failure_report(
      scenario: scenario,
      report_dir: report_dir,
      provider: @verifier_provider,
      started_at: started_at || Time.now,
      completed_at: Time.now,
      error_message: "#{e.class}: #{e.message}",
      failure_category: "runner-error",
      metadata: (report_dir)
    )
  rescue => write_error
    Models::TestResult.new(
      test_id: scenario.test_id,
      status: "error",
      summary: "Execution pipeline failed",
      error: "#{e.class}: #{e.message}; failed to write error report: #{write_error.class}: #{write_error.message}",
      started_at: started_at || Time.now,
      completed_at: Time.now
    )
  end
end