Commit Graph

3 Commits

Author SHA1 Message Date
Levi Neuwirth 24f8c8099a Editor T4-pre W2: the glyph outlines become a shared typed seam
A canvas tessellator needs typed vector paths, and the outlines were private to
render-svg as SVG `d` strings. They move to a new epiphany-glyphs crate, which
implements layout-ir's GlyphCatalog with a real render_data.

The seam was already designed and merely unpopulated: PathCommand,
GlyphRenderData, and GlyphCatalog::render_data all existed, and BravuraCatalog
returned None by deliberate documented honesty -- reporting Some would claim
render data that does not exist. Like W1, this packet fills a structure rather
than building one.

Typed paths are parsed from the `d` strings rather than regenerated, because
Bravura.otf is not vendored -- tools/ carries only the extractor and its OFL
notice, and the generated header pins source hashes verified at extraction
time. Equivalence is therefore proven rather than assumed: parse every bundled
glyph, re-emit, compare byte-for-byte. All thirty-seven round-trip exactly, so
the sanctioned coordinate-comparison fallback was not needed. The real grammar
is wider than the contract's summary claimed -- absolute V/H lineto shorthand
appears in twenty-three of thirty-seven glyphs, and numbers carry nought to
three fractional digits with trailing zeros stripped, never a fixed four. The
parser lowers V/H to LineTo and the emitter reconstructs the shorthand from
geometry.

Nothing rendered moves. The outline data is byte-identical to its previous home
(the only edit is BravuraOutline's visibility), render-svg keeps emitting the
stored strings and re-exports bundled_glyph_count and smufl_codepoint
unchanged, and BRAVURA_METRICS is untouched -- it feeds metrics_hash and hence
GlyphCatalogIdentity, which is inside the resolved layout's canonical bytes.
Verified: all eight reference fixtures produce byte-identical canonical layouts,
and all five GUI goldens are unchanged.

The cross-table test locks what glyph.rs asserted only in prose: every glyph's
declared metrics bbox contains its drawn ink, re-derived through true
cubic-bezier extrema rather than control points. Worst-case deviation across the
whole table is exactly zero, at noteheadBlack's left edge. Stated honestly, this
is a rounding-chain lock rather than an independent cross-check -- the same
extractor emits both tables and deliberately rounds the metric box outward so it
always contains the ink -- but it catches a unit-conversion error, a
regeneration mismatch, or a future inward rounding, none of which had a test
before.

Five of six mutations killed, all coordinator-re-verified independently. The
sixth is an honest survivor and a defect in the contract rather than the work:
it asked for proof that re-emitting from the typed form breaks the goldens,
which the round-trip pin makes a no-op. The goldens were re-proven live a
different way -- perturbing one glyph's stored path by one digit flips one pixel
of 279,992 and fails all five.

Gate: 35 suites / 1351 tests / 0 failed, conformance 9/9 with golden-gate and
8/8 without, requirement labels 6/6, clippy 0, cargo tree showing exactly one
dependency and no third-party crate.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-24 18:54:30 -04:00
Levi Neuwirth 6293734fa0 Agent I-4c: embedded @font-face glyph mode (a second self-contained renderer)
A render mode that references each glyph by its SMuFL codepoint via `<text>`,
drawn from an `@font-face`-embedded Bravura subset, alongside the default inline
`<path>` outlines. The SVG stays self-contained (the font travels in it) and the
text is selectable, at a larger file size.

  - GlyphMode::EmbeddedFont: the SVG declares the font once via
    <defs><style>@font-face{...}</style></defs>, then emits one
    <text transform="translate(x y) scale(1 -1)" font-size="4" ...>&#xNNNN;</text>
    per glyph. Geometry is consistent with PathOutline by construction: same
    origin, a per-glyph counter-flip cancels the outer y-flip, and the SMuFL em is
    four staff spaces. A new `text_count` stat; unbundled glyphs still fall through
    to the visible bbox rect + diagnostic. The metadata comment declares which mode
    produced the SVG (and, on the empty canvas too, via a shared `glyph_note`).
    Path mode stays the byte-golden-locked, pixel-verified reference; the embedded
    mode is structurally tested (well-formed XML — also under `xmllint`; one
    @font-face; one <text>/codepoint per glyph; provenance preserved; determinism).

  - The subset is a GENERATED artifact, not a vendored binary: a deterministic
    base64 OTF emitted into src/font_subset_generated.rs by
    `tools/extract_bravura_outlines.py --font-out` (the same SHA-pinned 1.392 font
    the outlines come from). `recalcTimestamp=False` keeps the source font's fixed
    head.modified so the bytes are reproducible across runs, not just within one.

  - OFL compliance: the subset is a Modified Version, so its PRIMARY font name is
    renamed off the Reserved Font Name "Bravura" to "EpiphanyBravuraSubset" in BOTH
    naming structures an OTF carries — the SFNT `name` table AND the CFF (Name INDEX
    + top-dict FullName/FamilyName). The copyright/trademark/license records, which
    name Bravura as attribution, are kept; the renderer references the renamed
    family in @font-face and <text>. The generator reparses the saved bytes and
    fails if the reserved name leaks into a primary record, and validates the cmap
    covers every glyph.

  - Machine-locked payload: the generator emits the decoded length and a BLAKE3-256
    (the workspace's sole hash) of the font bytes; a render-svg test base64-decodes
    the payload (no new runtime dep) and asserts the length, the OTTO signature, the
    BLAKE3, and — parsing the SFNT name table and CFF Name INDEX — that neither
    primary name is the reserved name. Adds an epiphany-determinism dev-dep.

The demo example gains `--glyph-mode=path|embedded`; lib/README/DECISIONS document
the two modes, the subset's OFL rename, and the regeneration command (--font-out +
the blake3 dependency). PathOutline goldens are byte-unchanged; the outlines stay
byte-identical. Full gate green: build, fmt, clippy, 585 tests, conformance scale 1.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-27 11:09:34 -04:00
Levi Neuwirth ac7c076e73 Phase 2 (Agent I): visible-slice scaffold — Bravura SVG renderer vs stub
Lands the renderer-against-stub slice of Agent I's visible engraving work
(spec/PHASE2_QUICKSTART.md). Two new crates; prerequisites (G Pass 11, H
spelling/decomposition) are in place. Real engraving + Minimal-tier solver
follow next phase.

epiphany-render-svg (the deliverable this phase):
- Renders a ResolvedLayoutIR to well-formed SVG 1.1, drawing each glyph as a
  GENUINE Bravura SMuFL outline <path>. Outlines are extracted reproducibly
  from the official OFL Bravura.otf by a committed generator
  (tools/extract_bravura_outlines.py, OFL.txt); the font is not vendored, only
  the generated Rust (src/outlines_generated.rs). Staff-space/y-up coords with
  one global y-flip wrapper; viewBox in staff spaces, px scale on the root.
- Non-overreach: every element traces to a ResolvedGlyph (data-prov) or a
  declared wrapper; a glyph lacking an outline is surfaced as a diagnostic and
  drawn as a fallback rect, never silently dropped.
- Hand-rolled xml::check_well_formed (no XML dep); acceptance tests cross-check
  with system xmllint when present.
- examples/render_fixture.rs demo (fixture name -> SVG stdout, --solver=stub|real).
- Golden-locked machine acceptance snapshot + full-SVG golden for
  ten_measure_single_staff and valid_score_rich; deterministic output.

epiphany-engrave (honest scaffold):
- Engraver: a deterministic horizontal-spacing pass (first axis of the planned
  two-pass spring layout). Reports SolverTier::Stub — NOT Minimal — until it
  evaluates the declared hard constraints, guarded by a regression test. The
  demo's --solver=real exercises it end to end.

Honesty notes (recorded as Pass-12 candidates P12-I1..I3 in spec/PASS12_BATCH.md
and the crates' DECISIONS.md): the v0 to_logical/to_constrained pipeline is a
structural placeholder (arbitrary glyph per object, y=0), so stub output is not
yet recognizable notation and the QUICKSTART human visual gate is a next-phase
gate; MUSCLOID layout-id derivation stays unwired; bundled BRAVURA_METRICS are
approximations that disagree with the real outlines.

Gates: cargo fmt + clippy -D warnings clean; cargo test --workspace 504 passed,
0 failed, 0 ignored (criterion 6 layout round-trip still green).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-23 20:01:40 -04:00