Fix layers being merged on read when SVG groups share the same inkscape:label

[abey79]

Problem

vpype read input.svg splitdist -l 2 N write out.svg followed by read out.svg collapsed the split layers back together. For example, with a 3-layer input where layer 2 is split into 4 parts (6 layers in memory), re-reading the written SVG yielded only 3 layers, silently merging the four chunks of layer 2 back into one (62+79+92+30 → 263 paths).

The bug was on the read side, not write. read_multilayer_svg derived each layer's ID from the first digit group of its inkscape:label, taking priority over the always-unique id="layer{N}". splitdist stamps the source layer's name onto every overflow layer, so the chunks were all written with inkscape:label="2" and collapsed to a single layer on read. The same footgun hit any document with two layers sharing a digit-containing name (pen2, layer2, …), not just splitdist output.

Fix

Gate the label-based numbering on global distinctness: the inkscape:label is used to assign layer IDs only when it yields a distinct ID for every group. Otherwise it falls back to the id attribute (which vpype writes as layer{N}), then to appearance order.

This:

• makes the write→read round-trip lossless for splitdist output and any duplicate digit-names;
• preserves the Inkscape "rename a layer to a number" semantics for the common distinct case (verified against the existing multilayer_named_layers.svg fixture, whose ids are off-by-one — labels correctly still win there).

Tests

• test_read_layer_id_colliding_labels_fall_back_to_id — groups sharing a label fall back to id instead of merging.
• test_read_layer_id_distinct_labels_take_priority — distinct labels still win over the id attribute.
• test_splitdist_write_read_roundtrip — full splitdist → write → read round-trip preserves the layer count.

Each was confirmed to fail on the pre-fix code. Full test_files.py + test_commands.py suites pass (1024 passed, 67 skipped).

🤖 Generated with Claude Code

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud