Module: Rigor::Inference::ProjectPatchedScanner

Defined in:
lib/rigor/inference/project_patched_scanner.rb

Overview

ADR-17 slice 2 — pre-pass scanner. Walks every file the user listed under ‘pre_eval:` and harvests every `def` / `def self.` declaration inside a class / module body into a ProjectPatchedMethods registry the dispatcher consults below the plugin tier.

The walker is intentionally a strict subset of ScopeIndexer‘s machinery: it only needs `class C; def m; end; end` shape recognition, not full inference. Parse errors degrade to a fail-soft `:warning` `pre-eval.parse-error` diagnostic accumulated alongside the registry; per ADR-17 § “Failure modes” a parse failure in a pre-eval file MUST NOT abort the rest of the run.

Defined Under Namespace

Classes: Result

Class Method Summary collapse

Class Method Details

.scan(paths, buffer: nil) ⇒ Result

Returns the populated registry plus any per-file warnings.

Parameters:

  • paths (Array<String>) —

    absolute paths to the pre-eval files. The runner has already validated that each path exists (slice-1 ‘pre-eval.file-not-found` `:error` covers missing entries); the scanner does NOT re-check existence.

  • buffer (Rigor::Analysis::BufferBinding, nil) (defaults to: nil) —

    editor-mode buffer binding. When set, the scanner reads the buffer’s physical bytes if a pre-eval entry matches the logical path, so users editing a monkey-patch file see the in-flight version in their analysis.

Returns:

  • (Result) —

    the populated registry plus any per-file warnings.



49
50
51
52
53
54
55
56
57
58
 
# File 'lib/rigor/inference/project_patched_scanner.rb', line 49

def scan(paths, buffer: nil)
  entries = []
  diagnostics = []
  paths.each { |path| scan_file(path, entries, diagnostics, buffer) }
  diagnostics.concat(duplicate_declaration_diagnostics(entries))
  Result.new(
    registry: ProjectPatchedMethods.new(entries: entries),
    diagnostics: diagnostics
  )
end