test: add unit tests for canon/digest.go — Digest and PipelineDigest

[jpleva91]

Problem

canon/digest.go has zero test coverage. Digest() and PipelineDigest() are critical for deterministic command fingerprinting — if these produce non-deterministic hashes, doom-loop detection and event deduplication break silently.

Acceptance Criteria

• Test Digest() produces identical hashes for semantically equivalent commands (same tool/action/flags/args in different order)
• Test Digest() produces different hashes for different commands
• Test PipelineDigest() produces identical hashes for equivalent pipelines
• Test PipelineDigest() produces different hashes for pipelines with different operators or segments
• Test that flag ordering does not affect digest (flags are sorted)
• Test empty command edge case
• All tests pass with go test ./canon/...

Hints

• File: canon/digest.go
• Functions: Digest(cmd Command) string, PipelineDigest(p Pipeline) string
• Types in canon/types.go: Command, Pipeline, Segment, ChainOp
• Digest uses SHA-256, truncated to 16 hex chars
• Create test file: canon/digest_test.go

---

Generated by /forge fill

[jpleva91]

🧠 Triage complete

Tier B — Needs Planning. Issue requires scoping and decomposition before implementation.

Reason: triage error (defaulting to safe): API returned 404: {"type":"error","error":{"type":"not_found_error","message":"model: claude-3-haiku-20241022"},"request_id":"req_011CZrvD3YYLB5vVeLrHYpDW"}
Confidence: 0.00

Powered by Octi Pulpo pipeline

[jpleva91]

@claude Please implement this issue. Read the acceptance criteria carefully, check the source files mentioned in Hints, and create the test file. Run go test to verify all tests pass before finishing.

[jpleva91]

Test coverage for canon/digest.go hash determinism; low complexity, well-scoped, safe to automate.

---

Generated by Claude Code

[jpleva91]

Execution Plan — canon/digest_test.go

Agent: swarm-worker (intake phase)
Scope: Direct unit tests for Digest() and PipelineDigest() in canon/digest.go

---

Context

canon_test.go already exercises Digest() indirectly through ParseOne(), but the acceptance criteria require tests that call Digest() and PipelineDigest() directly via constructed Command/Pipeline structs. The key uncovered behaviours are:

• Flag-ordering independence (map iteration is random — sorted keys are the correctness guarantee)
• PipelineDigest() sensitivity to operator changes and segment reordering
• PipelineDigest() uses the pre-set Command.Digest field, not recomputing it

---

File to create: canon/digest_test.go

Package: canon (white-box — same package as source)

Tests for Digest()

Test	What it verifies
TestDigest_FlagOrderIndependence	Two Command objects with flags inserted in different order produce identical digests
TestDigest_EquivalentCommands	Same Tool/Action/Flags/Args → same digest
TestDigest_DifferentTool	Differing Tool → different digest
TestDigest_DifferentAction	Differing Action → different digest
TestDigest_DifferentArgs	Differing Args → different digest
TestDigest_DifferentFlagValue	Same flag key, different value → different digest
TestDigest_EmptyCommand	Empty Command{} → consistent 16-hex-char digest (no panic)
TestDigest_Length	Digest is always exactly 16 hex characters

Tests for PipelineDigest()

Test	What it verifies
TestPipelineDigest_Equivalent	Two pipelines with same segments/ops → same digest
TestPipelineDigest_DifferentOp	Same commands, OpAnd vs OpOr → different digest
TestPipelineDigest_DifferentSegmentDigest	Changing one segment's Command.Digest → different pipeline digest
TestPipelineDigest_SegmentOrderMatters	Reversing segment order → different digest
TestPipelineDigest_EmptyPipeline	Pipeline{} → consistent 16-hex-char digest (no panic)
TestPipelineDigest_Length	Always exactly 16 hex characters

---

Implementation notes

• Use table-driven tests (Go convention) where multiple cases share the same logic
• PipelineDigest reads seg.Command.Digest (the string field), so tests must pre-set that field explicitly — e.g. Command{Digest: "abc123"}
• No mocks, no external dependencies — pure struct construction
• All tests must pass: go test ./canon/...

---

Acceptance criteria checklist (from issue)

• Test Digest() produces identical hashes for semantically equivalent commands
• Test Digest() produces different hashes for different commands
• Test PipelineDigest() identical for equivalent pipelines
• Test PipelineDigest() differs for different operators or segments
• Test flag ordering does not affect digest
• Test empty command edge case
• All tests pass go test ./canon/...

---

Swarm intake run — 2026-04-10 — ready for execution queue

---

Generated by Claude Code