@k-l-lambda/lilylet 0.1.60 → 0.1.62

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@k-l-lambda/lilylet",
-  "version": "0.1.60",
+  "version": "0.1.62",
   "description": "Lilylet is a lilyopnd-like sheet music language designed for Markdown rendering and symbolic music representation in AIGC applications.",
   "type": "module",
   "main": "lib/index.js",
@@ -27,12 +27,16 @@
     "build": "tsc -p tsconfig.build.json && node tools/convertGrammarToESM.cjs && node tools/fixEsmExtensions.cjs",
     "build:grammar": "npx tsx ./tools/buildJisonParser.ts && node tools/convertGrammarToESM.cjs",
     "prepublishOnly": "npm run build:grammar && npm run build",
-    "test": "npx tsx ./tests/parser.ts",
-    "test:mei": "npx tsx ./tests/mei.ts",
-    "test:unit": "npx tsx ./tests/unit/encodePitch.test.ts",
-    "test:decoder": "npx tsx ./tests/lilypondDecoder.ts",
-    "test:abc": "npx tsx ./tests/abc-decoder.ts",
-    "ts": "npx tsx"
+    "test": "tsx ./tests/parser.ts",
+    "test:mei": "tsx ./tests/mei.ts",
+    "test:unit": "tsx ./tests/unit/encodePitch.test.ts",
+    "test:partial": "tsx ./tests/unit/partialWarning.test.ts",
+    "test:decoder": "tsx ./tests/lilypondDecoder.ts",
+    "test:abc": "tsx ./tests/abc-decoder.ts",
+    "test:roundtrip": "tsx ./tests/lilypond-roundtrip.ts",
+    "build:tests": "tsc -p tsconfig.tests.json; cp source/lilylet/grammar.jison.js lib-tests/source/lilylet/ && cp source/abc/grammar.jison.js lib-tests/source/abc/ && node tools/fixEsmExtensions.cjs lib-tests && ln -sfn ../../tests/assets lib-tests/tests/assets",
+    "test:roundtrip:compiled": "node lib-tests/tests/lilypond-roundtrip.js",
+    "ts": "tsx"
   },
   "repository": {
     "type": "git",
@@ -51,6 +55,7 @@
     "jison": "^0.4.18",
     "sha1": "^1.1.1",
     "ts-node": "^10.9.2",
+    "tsx": "^4.21.0",
     "typescript": "^5.3.3",
     "verovio": "^5.7.0",
     "xmldom": "^0.6.0",

package/source/abc/TODO.md

@@ -0,0 +1,97 @@
+# ABC Grammar TODO
+
+Issues found by comparing `abc.jison` against the abcjs parser implementation.
+
+---
+
+## High Priority
+
+### 1. Church modes in key signatures
+
+`key_mode` only distinguishes `major` / `minor`. The `NAME` fallback does a naive
+`startsWith("ma")` check, which mis-classifies all church modes:
+
+- Dorian (`Dor`, `dorian`)
+- Phrygian (`Phr`, `phrygian`)
+- Lydian (`Lyd`, `lydian`)
+- Mixolydian (`Mix`, `mixolydian`)
+- Aeolian (`Aeo`, `aeolian`)
+- Locrian (`Loc`, `locrian`)
+- Scottish bagpipe: `HP`, `Hp`
+
+These should be recognized as distinct modes rather than silently falling back to
+major or minor. The `abcDecoder.ts` key-mapping logic will also need updating to
+handle these modes.
+
+### 2. Bare-number Q: tempo (`Q:120`)
+
+`numeric_tempo` only matches `frac '=' number` (e.g. `Q:1/4=120`). Two common
+forms are not parsed:
+
+- `Q:120` — plain BPM number (unit inferred from the current meter/L: value)
+- `Q:"Allegro"` — text-only tempo marking
+- `Q:"Allegro" 1/4=120` — combined text + numeric form
+
+`Q:120` is the most frequently seen form in real-world ABC files.
+
+---
+
+## Medium Priority
+
+### 3. Missing rest type: `y` (spacer)
+
+The `rest_phonet` rule covers `z`, `Z`, `x` but not `y` (an invisible spacer rest
+used for spacing/layout). Files containing `y` currently cause a parse error.
+
+### 4. Volta ending bracket (`endEnding`)
+
+The `bar` rule appends the ending number directly to the bar token string
+(`'|' + N → "|1"`). There is no representation of the *closing* bracket of a
+first-ending, so `[1 ... [2` style repeat structures cannot be round-tripped.
+abcjs tracks `startEnding` / `endEnding` flags on bar elements; consider a
+similar approach.
+
+---
+
+## Low Priority
+
+### 5. `~` maps to `mordent` instead of `irishroll`
+
+`abc.jison` line 507: `'~' → articulation("mordent")`.  
+The standard ABC specification defines `~` as an *Irish roll* ornament, not a
+mordent. abcjs uses the name `irishroll`. The mordent is correctly represented by
+`M`. This is a semantic mismatch that may affect downstream rendering/export.
+
+### 6. Microtonal accidentals (`^/`, `_/`)
+
+The `accidentals` rule does not cover quarter-tone accidentals:
+
+- `^/` → quarter sharp
+- `_/` → quarter flat
+
+These are recognised by abcjs. Files using microtonal notation will silently drop
+the accidental.
+
+### 7. Short trill decoration `t`
+
+The single-letter `t` (half/short trill, `trillh` in abcjs) is not handled. The
+current `P`/`PP` lexer patterns only match uppercase ornament letters
+(`HJLMOPRSTuv`), so lowercase `t` falls through as an unknown token.
+
+### 8. Overlay voices (`&`)
+
+The `&` operator (alternative voice within a single bar, same staff) is not
+supported. abcjs resolves overlays into separate voices via `resolveOverlays()`.
+This is a moderately common pattern in two-voice piano or lute transcriptions.
+
+---
+
+## Out of Scope (noted for awareness)
+
+- **Lyrics** (`w:` / `W:` fields): not handled at the ABC grammar level; notes
+  have no `lyric` property. Would require grammar additions and decoder support.
+- **Extended clef types**: only `treble`, `bass`, `tenor` are recognised. abcjs
+  also handles `alto`, `baritone`, `mezzo`, `soprano`, `tab`, `perc`, etc.
+- **Decoration name aliases**: abcjs normalises `tr`→`trill`, `emphasis`→`accent`,
+  `marcato`→`umarcato`, `<`/`>`→`accent`. The lilylet grammar passes `NAME`
+  through as-is; the decoder would need to handle the aliases explicitly.