Module: Kumi::Analyzer
- Defined in:
- lib/kumi/analyzer.rb
Defined Under Namespace
Classes: Result
Constant Summary collapse
- Passes =
Core::Analyzer::Passes
- ERROR_THRESHOLD_PASS =
Passes::NormalizeToNASTPass
- DEFAULT_PASSES =
[ Passes::NameIndexerPass, Passes::ImportAnalysisPass, Passes::InputCollectorPass, Passes::InputFormSchemaPass, Passes::DeclarationValidatorPass, Passes::SemanticConstraintValidatorPass, Passes::DependencyResolverPass, Passes::ToposorterPass, Passes::InputAccessPlannerPass ].freeze
- LOWERING_PASSES =
Lowering pipeline: NAST -> SNAST -> DFIR -> VecIR -> LoopIR
[ Passes::NormalizeToNASTPass, Passes::ConstantFoldingPass, Passes::NASTDimensionalAnalyzerPass, Passes::SNASTPass, Passes::UnsatDetectorPass, Passes::OutputSchemaPass, Passes::AttachTerminalInfoPass, Passes::AttachAnchorsPass, Passes::PrecomputeAccessPathsPass, Passes::LowerToDFIRPass, Passes::DFValidatePass, Passes::Vec::LowerPass, Passes::VecValidatePass, Passes::Loop::LowerPass, Passes::LoopValidatePass ].freeze
- TARGET_PASSES =
[ Passes::Codegen::LoopRubyPass, Passes::Codegen::LoopJsPass ].freeze
Class Method Summary collapse
- .analyze!(schema, passes: DEFAULT_PASSES, registry: nil, **opts) ⇒ Object
- .create_analysis_result(state) ⇒ Object
-
.format_errors(errors) ⇒ Object
Handle both old and new error formats for backward compatibility.
- .handle_analysis_errors(errors) ⇒ Object
- .run_analysis_passes(schema, passes, state, errors) ⇒ Object
Class Method Details
.analyze!(schema, passes: DEFAULT_PASSES, registry: nil, **opts) ⇒ Object
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 |
# File 'lib/kumi/analyzer.rb', line 45 def self.analyze!(schema, passes: DEFAULT_PASSES, registry: nil, **opts) errors = [] schema_digest = schema.digest Core::Analyzer::Checkpoint.stop_after registry ||= Kumi::FunctionRegistry.load state = Core::Analyzer::AnalysisState.new(opts).with(:registry, registry).with(:schema_digest, schema_digest) state, stopped = run_analysis_passes(schema, passes, state, errors) return create_analysis_result(state) if stopped handle_analysis_errors(errors) unless errors.empty? state, stopped = run_analysis_passes(schema, LOWERING_PASSES, state, errors) return create_analysis_result(state) if stopped handle_analysis_errors(errors) unless errors.empty? state, = run_analysis_passes(schema, TARGET_PASSES, state, errors) handle_analysis_errors(errors) unless errors.empty? create_analysis_result(state) end |
.create_analysis_result(state) ⇒ Object
121 122 123 124 125 126 127 128 129 130 |
# File 'lib/kumi/analyzer.rb', line 121 def self.create_analysis_result(state) Result.new( definitions: state[:declarations], dependency_graph: state[:dependencies], leaf_map: state[:leaves], topo_order: state[:evaluation_order], decl_types: state[:inferred_types], state: state.to_h ) end |
.format_errors(errors) ⇒ Object
Handle both old and new error formats for backward compatibility
133 134 135 136 137 |
# File 'lib/kumi/analyzer.rb', line 133 def self.format_errors(errors) return "" if errors.empty? errors.map(&:to_s).join("\n") end |
.handle_analysis_errors(errors) ⇒ Object
110 111 112 113 114 115 116 117 118 119 |
# File 'lib/kumi/analyzer.rb', line 110 def self.handle_analysis_errors(errors) raise Kumi::Errors::AnalysisError, "\n" + errors.join("\n") if errors.first.is_a? String type_errors = errors.select { |e| e.type == :type } first_error_location = errors.first.location raise Errors::TypeError.new(format_errors(errors), first_error_location) if type_errors.any? raise Errors::SemanticError.new(format_errors(errors), first_error_location) end |
.run_analysis_passes(schema, passes, state, errors) ⇒ Object
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 |
# File 'lib/kumi/analyzer.rb', line 68 def self.run_analysis_passes(schema, passes, state, errors) # Resume from a saved state if configured state = Core::Analyzer::Checkpoint.load_initial_state(state) # Prepare options for PassManager debug_on = Core::Analyzer::Debug.enabled? resume_at = Core::Analyzer::Checkpoint.resume_at stop_after = Core::Analyzer::Checkpoint.stop_after # Filter passes based on checkpoint resume point filtered_passes = if resume_at passes.each_with_index do |pass_class, idx| pass_name = pass_class.name.split("::").last break passes[idx..] if pass_name == resume_at end.flatten.compact else passes end # Check for error threshold pass raise handle_analysis_errors(errors) if !errors.empty? && filtered_passes.include?(ERROR_THRESHOLD_PASS) # Use PassManager for orchestration manager = Core::Analyzer::PassManager.new(filtered_passes) = { checkpoint_enabled: true, debug_enabled: debug_on, profiling_enabled: true, stop_after: stop_after } result = manager.run(schema, state, errors, ) # `errors` is the canonical accumulator: passes append to it, and # PassManager appends any captured exception to it before deriving # result.errors (PassFailure objects) from the same array. Re-appending # result.errors here would surface every failure a second time, so we # don't — the originals are already in `errors`. [result.final_state, result.stopped || false] end |