Fix footprint parser for KiCad 10: add duplicate_pad_numbers_are_jumpers + lenient unknown_sexp

[devin-ai-integration]

Fix footprint parser for KiCad 10: add missing token + lenient catch-all

Summary

KiCad 10 footprint files (.kicad_mod, version 20260206) fail to parse because the module grammar doesn't recognize the duplicate_pad_numbers_are_jumpers token. This causes the parser to error out entirely, resulting in all pads appearing stacked at the origin when imported into Flux.

Changes to module.pegjs:

1. Added duplicate_pad_numbers_are_jumpers as a new boolean rule in module_contents
2. Added an unknown_sexp catch-all rule (matching the pattern already used in symbol.pegjs) to 10 grammar rules: board, module_contents, pad_attr, fp_generics, gr_generics, module_property, fp_text, fp_text_box, effects, and font
3. Added .filter(function(v) { return v !== null; }) to all .map() calls in rules that now include unknown_sexp, since the catch-all returns null

The compiled dist/ files are regenerated from the grammar — the meaningful diff is entirely in src/module.pegjs (~40 lines changed).

Updates since last revision

• Fixed balanced_content quote bug — changed [^()]+ to [^()"']+ so quote characters are always handled by the string alternative. Without this, a value like (new_field "value with )") would have the " consumed by the catch-all character class, causing the ) inside the string to be treated as a structural paren and breaking the parse.
• Added KiCad 10 test fixtures from Thea:
  • Footprints: MSOP-10_3x3mm_P0.5mm.kicad_mod (10 pads) and QSOP-28_3.9x9.9mm_P0.635mm.kicad_mod (28 pads)
  • Symbols: ch334u.kicad_sym and ch340x.kicad_sym
• Updated .gitignore — changed data/module to data/module/* with !data/module/custom so custom footprint test files are tracked in git (matching the pattern used for data/symbols/custom/)

All test fixtures parse successfully with pnpm test.

Review & Testing Checklist for Human

• Verify unknown_sexp is always the last alternative in every rule it was added to — PEG grammars are ordered, so placing it last ensures known tokens are always tried first. If it appeared earlier, it could swallow valid tokens silently.
• Verify the null-filtering doesn't mask real issues — .filter(v => v !== null) was added to ~10 .map() calls. Confirm no existing grammar rule relied on returning null for a meaningful purpose.
• Test with real KiCad 10 footprint files — import MSOP-10_3x3mm_P0.5mm.kicad_mod and QSOP-28_3.9x9.9mm_P0.635mm.kicad_mod in Flux and verify pads are positioned correctly (not stacked at origin). MSOP-10 should have 10 pads spanning x={-2.1, 2.1}; QSOP-28 should have 28 pads spanning x={-2.6375, 2.6375}.
• Regression test with existing KiCad 5/6/7/8/9 footprints — the existing test suite passes, but spot-check a few older footprints to make sure the catch-all doesn't break anything.

Notes

• The dist/index.js diff is large but entirely auto-generated from pegjs. Only src/module.pegjs contains hand-written changes.
• This mirrors the approach already merged for the symbol parser (symbol.pegjs) in PR Fix boolean parsing, add lenient unknown-field handling, include dist/ #17.
• Link to Devin Session: https://app.devin.ai/sessions/103ae3c52c02438aabd3955415051fa5
• Requested by: @natarius

---

[devin-ai-integration]

Original prompt from Matthias

SYSTEM:
=== BEGIN THREAD HISTORY (in #dogfooding) ===
Thea (U0AFH6R3D8D): Given the speed the EAGLE, EasyEDA and other imports (seem) to have been implemented. Could we get a KiCad Import that works with the newest filetype?  No KiCad file that was created in the last 5 year works. The last supported version seems to be KiCad 5 and soon KiCad 10 will be launching.

Giulio (U091GFANVAT): Mmm weird since we are using kicad 9 internally 

Giulio (U091GFANVAT): I think it’s a matter of correctly mapping the kicad features into flux 

Thea (U0AFH6R3D8D): Its a biig Painpoint.

Thea (U0AFH6R3D8D): CC @matthias (U0852RD7LMP)

matthias (U0852RD7LMP): What exactly doesn't work?

Thea (U0AFH6R3D8D): Import of symbols and footprints created in a recent KiCad version just fail

matthias (U0852RD7LMP): Can you put together a folder of test files please

matthias (U0852RD7LMP): This is the first time I hear about this issue

matthias (U0852RD7LMP): @Jharwin (U091GF9SA5V) @Nicolas Tzovanis (U084Q3S45U3) is this a regressions?

Nicolas Tzovanis (U084Q3S45U3): Afaik we supported up to kicad v6

Nicolas Tzovanis (U084Q3S45U3): And kicad 9 uses the same format as kicad v6 for parts. Not sure what the deal is with Kicad 10 though.

Thea (U0AFH6R3D8D): 

ATTACHMENT:"https://app.devin.ai/attachments/a644df82-7cda-43c6-b42e-26affadf9653/image.png"

Thea (U0AFH6R3D8D): Currently making a Test folder.

Nicolas Tzovanis (U084Q3S45U3): Did the format change on Kicad 10?

matthias (U0852RD7LMP): @Devin please investigate
=== END THREAD HISTORY ===

Thread URL: https://fluxaigroup.slack.com/archives/C09286X4F7Y/p1772466774920479?thread_ts=1772466774.920479&cid=C09286X4F7Y

The latest message is the one right above that tagged you.

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring