feat(bom-export): attach a Supplier (JLCPCB by default) to PCB & SMT Stencil parts by peterus · Pull Request #33 · OE5XRX/HW-Module-CI

peterus · 2026-06-17T19:17:08Z

Summary

bom_export.py's create_pcb_part and create_stencil_part currently create the InvenTree Part only — no SupplierPart linking them to the fab. Re-order workflows in InvenTree break down because you can't add the PCB/Stencil to a JLCPCB PurchaseOrder via the normal flow.

This PR attaches a SupplierPart (default supplier: JLCPCB, configurable via --pcb-supplier) to every newly-created PCB and SMT Stencil Part. SKU is derived as f\"{full_name} rev {version}\", idempotent across re-runs. The Assembly Module is intentionally not touched (built in-house).

Trigger

After the recent v0.1 bulk-import we ended up with 4 PCBs (pks 1291/1294/1297/1301) and 4 Stencils (pks 1293/1296/1299/1303) with no SupplierPart. JLCPCB exists as Company pk=381 in InvenTree, just wasn't linked. Backfill of those 8 records will happen as a separate ad-hoc API run after merge.

What's in the diff

Spec (docs/superpowers/specs/2026-06-11-pcb-supplier-attachment-design.md)
Plan (docs/superpowers/plans/2026-06-11-pcb-supplier-attachment.md)
Code — scripts/bom_export.py:
- New _ensure_fab_supplier_part(api, part, supplier, sku) helper (idempotent, defensive list-then-post-filter on SKU)
- create_pcb_part / create_stencil_part accept keyword-only fab_supplier=None; call the helper after Part create OR reuse
- New constant DEFAULT_PCB_SUPPLIER_NAME = \"JLCPCB\"
- CLI flags --pcb-supplier NAME (default JLCPCB) and --no-pcb-supplier (opt-out)
- main() resolves supplier via get_or_create_supplier only in the non-dry-run branch (dry-run contract preserved)
Tests — scripts/tests/test_bom_export_fab_supplier.py (new): 13 unit tests covering the helper edge cases, create-function behavior on new+reuse+no-supplier paths, assembly-not-touched, and the dry-run no-probe guarantee.

Idempotency

Re-running the same release: existing Part found → existing SupplierPart found by SKU → no new write.
Re-running with --pcb-supplier=PCBWay: existing Part reused, new SupplierPart for PCBWay (different supplier).
Backfill old Parts (created before this PR): first call to modified create_pcb_part adds the SupplierPart on the reuse path.

Process artefacts

Spec brainstormed → plan written → 3 tasks executed via subagent-driven-development (sonnet implementer, atlas spec reviewer, audit code reviewer per task)
1 per-task review fix on Task 1 (caplog assertions, broad-except comments, ASCII arrow, 2 edge tests)
Final cross-branch review caught and fixed the dry-run side-effect leak (get_or_create_supplier was called before the dry-run gate — same bug class as PR fix(import-orders): make --dry-run actually dry on Part resolution #31, fixed by moving the resolution into the non-dry-run branch and adding a test that locks it in)

Test plan

225/225 unit tests pass (pytest scripts/tests/)
--help shows the new --pcb-supplier and --no-pcb-supplier flags
CLI tests unchanged (no regression in dry-run path)
Operator-side (post-merge): ad-hoc API run to backfill SupplierPart on the 8 existing v0.1 PCBs/Stencils (pks 1291/1293/1294/1296/1297/1299/1301/1303)

Backward compatibility

New default behavior: --pcb-supplier=JLCPCB IS active out of the box. This is intentional — the current no-supplier behavior is the bug.
Users without JLCPCB Company in their InvenTree: get_or_create_supplier auto-creates it on first non-dry-run.
Users who explicitly don't want fab linkage: --no-pcb-supplier opt-out preserves pre-PR behavior.
create_pcb_part / create_stencil_part add a keyword-only fab_supplier arg (default None) — no existing call site breaks.

Non-goals (deferred to future PRs)

ManufacturerPart linkage (bare-PCB substrate has no MPN worth tracking).
link / pricing / pack_quantity fields on the SupplierPart (no re-order planned near-term).
Multi-fab support (PCB from one fab, stencil from another).
Assembly-side supplier (JLCPCB-SMT-Assembly).

🤖 Generated with Claude Code

bom_export.py's create_pcb_part and create_stencil_part currently create the Part only, with no SupplierPart linking it to the fab (JLCPCB in our case). After v0.1 import we have 4 PCBs + 4 stencils without supplier metadata — re-order workflow breaks down at scale. Fix: attach a SupplierPart with a deterministic SKU ({full_name} rev {version}) on both create and reuse branches. CLI flag --pcb-supplier (default JLCPCB) selects the fab; --no-pcb-supplier opts out. Idempotent via SKU-existence check. Fab linkage is best- effort: a failure logs warning but never blocks the release CI. Assembly Module stays without supplier (built in-house). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Three TDD-strict tasks: Task 1 — _ensure_fab_supplier_part helper (idempotent SP linkage) Task 2 — create_pcb_part / create_stencil_part attach fab supplier (both create and reuse branches; assembly stays unchanged) Task 3 — CLI flags --pcb-supplier (default JLCPCB) and --no-pcb-supplier; main() resolves supplier via get_or_create_supplier and threads it through Each step has concrete code, exact pytest commands, expected output, and commit message ready to use. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Idempotently link a Part to a fab Supplier (JLCPCB by default) with a derived SKU. Same defensive list-then-post-filter pattern as ensure_supplier_parts in client.py. Failures during list / create are logged and swallowed — fab linkage is best-effort metadata, never blocks the release CI run. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

- Add caplog warning assertions to the two failure-path tests so future refactors can't silently drop the operator's only signal. - Add inline "Broad except is intentional" comments above the two except blocks to pre-empt Copilot review noise. - Replace Unicode arrow with ASCII -> in the success log message for consistency with the rest of the codebase. - Add two new tests: multi-SP early-return on match (after a non-matching item), and tolerance for SKU=None on existing SP. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…pplier Both create functions gain a keyword-only fab_supplier=None argument. When non-None, _ensure_fab_supplier_part is called after Part create OR reuse — the reuse path enables retroactive linkage by re-running bom_export.py on releases that predate this change. SKU is derived as f'{full_name} rev {version}', stable per design revision, idempotent across re-runs. create_assembly_part is intentionally NOT modified: the assembly is built in-house, no external supplier applies. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

main() resolves the fab supplier Company via get_or_create_supplier when --pcb-supplier is set (default: JLCPCB). The resolved Company is threaded into create_pcb_part and create_stencil_part as a keyword argument. create_assembly_part is intentionally not touched (assembly is built in-house). --no-pcb-supplier opts out: PCB+stencil Parts get no SupplierPart attached, matching pre-PR behaviour for users who don't want this. Dry-run path doesn't model SupplierPart decisions — the linkage is best-effort metadata, real-run logs cover the outcome. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…be supplier Move the get_or_create_supplier call from main()'s pre-dry-run section into the non-dry-run code path. The function calls Company.create when the named Company is missing, which violated the dry-run contract: `--dry-run --pcb-supplier=NewFab` would create a real Company in InvenTree despite the flag. Same bug class as PR #31 (--dry-run actually dry on Part resolution). New test test_dry_run_does_not_resolve_fab_supplier locks the behaviour in. Also tightened the dry-run code comment to make the absence of SupplierPart linkage in the report explicit and intentional (not a side-effect bug). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Copilot

Pull request overview

This PR enhances scripts/bom_export.py so newly created (or reused) PCB and SMT Stencil InvenTree Parts are automatically linked to a fab via a SupplierPart (default supplier: JLCPCB, configurable via CLI). This unblocks standard InvenTree purchasing/re-order flows for PCBs/stencils by ensuring they have supplier metadata.

Changes:

Add _ensure_fab_supplier_part(...) helper to idempotently create a SupplierPart for (part, supplier, SKU) using a derived SKU format.
Thread an optional fab_supplier through create_pcb_part / create_stencil_part and add CLI flags --pcb-supplier / --no-pcb-supplier (while preserving dry-run’s “no writes” contract).
Add a focused unit test suite covering helper edge cases, create/reuse paths, and dry-run non-resolution of the supplier.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
scripts/bom_export.py	Adds fab-supplier `SupplierPart` linkage helper, CLI options, and wiring into PCB/stencil part creation without impacting assembly behavior.
scripts/tests/test_bom_export_fab_supplier.py	Adds unit tests for supplier-part idempotency, error handling, create/reuse behavior, and dry-run guarantees.
docs/superpowers/specs/2026-06-11-pcb-supplier-attachment-design.md	Captures the design/specification for attaching a fab supplier to PCB/stencil parts.
docs/superpowers/plans/2026-06-11-pcb-supplier-attachment.md	Provides the implementation plan and test strategy for the change.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

peterus · 2026-06-17T19:24:00Z

+    # Resolve fab supplier now (not before the dry-run gate): get_or_create_supplier
+    # calls Company.create when the named Company is missing, which would violate
+    # the dry-run contract — same bug class as PR #31.
+    fab_supplier: Optional[Company] = None
+    if args.pcb_supplier is not None:
+        fab_supplier = get_or_create_supplier(api, name=args.pcb_supplier)
+        if fab_supplier is None:
+            log.error(
+                "Could not get or create fab supplier %r — proceeding "
+                "without SupplierPart linkage.", args.pcb_supplier)
+


Fixed. main() now normalises args.pcb_supplier via (args.pcb_supplier or "").strip() or None — empty/whitespace values are treated as opt-out, equivalent to --no-pcb-supplier. New parametrized test test_blank_pcb_supplier_treated_as_opt_out covers "", " ", "\t", "\n".

peterus · 2026-06-17T19:24:02Z

+    )
+
+
+def test_ensure_fab_supplier_part_returns_early_on_first_match():


Fixed. Renamed to test_ensure_fab_supplier_part_skips_create_when_any_existing_sku_matches. The matching SKU is the second item in the list, so the old name _returns_early_on_first_match was actively misleading.

- 3430855521 (important): --pcb-supplier with empty/whitespace value would have POSTed a Company with a blank name. Normalise in main() via `(args.pcb_supplier or "").strip() or None` and treat as opt-out (equivalent to --no-pcb-supplier). New parametrized test covers "", " ", "\t", "\n". - 3430855563 (minor): rename test_ensure_fab_supplier_part_returns_ early_on_first_match → _skips_create_when_any_existing_sku_matches. The matching SKU was the second item in the list, not the first; the old name read as a contradiction. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Copilot

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

peterus and others added 7 commits June 17, 2026 20:58

Copilot AI review requested due to automatic review settings June 17, 2026 19:17

Copilot started reviewing on behalf of peterus June 17, 2026 19:17 View session

Copilot AI reviewed Jun 17, 2026

View reviewed changes

peterus requested a review from Copilot June 17, 2026 19:24

Copilot started reviewing on behalf of peterus June 17, 2026 19:24 View session

Copilot AI reviewed Jun 17, 2026

View reviewed changes

peterus merged commit 0ebcd04 into main Jun 17, 2026
2 checks passed

peterus deleted the feat/pcb-stencil-supplier-attachment branch June 17, 2026 19:28

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

feat(bom-export): attach a Supplier (JLCPCB by default) to PCB & SMT Stencil parts#33

feat(bom-export): attach a Supplier (JLCPCB by default) to PCB & SMT Stencil parts#33
peterus merged 8 commits into
mainfrom
feat/pcb-stencil-supplier-attachment

peterus commented Jun 17, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

peterus Jun 17, 2026

Uh oh!

peterus Jun 17, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		)


		def test_ensure_fab_supplier_part_returns_early_on_first_match():

Uh oh!

Conversation

peterus commented Jun 17, 2026

Summary

Trigger

What's in the diff

Idempotency

Process artefacts

Test plan

Backward compatibility

Non-goals (deferred to future PRs)

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

peterus Jun 17, 2026

Choose a reason for hiding this comment

Uh oh!

peterus Jun 17, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants