Add synthetic sample library

.gitignore

@@ -4,6 +4,7 @@ __pycache__/
 .venv/
 *.egg-info/
 reports/
+samples/small-library/generated/
 *.sqlite
 *.db
 *.csv

ROADMAP.md

@@ -9,7 +9,7 @@
 - [x] Grouped missing reference report
 - [ ] HTML health dashboard
 - [x] Test suite
-- [ ] Sample library fixtures
+- [x] Sample library fixtures
 - [ ] Database V2 read-only parser
 
 ## v0.2 — Diagnostics

docs/design/sample-library.md

@@ -0,0 +1,26 @@
+# Synthetic Sample Library
+
+## Problem
+
+Real Serato libraries contain private paths, listening history, and copyrighted
+music. Contributors still need a repeatable library for tests and demonstrations.
+
+## Architecture
+
+`samples/small-library/manifest.json` describes a compact migration scenario.
+`generate.py` turns that manifest into fake audio files and UTF-16-LE crate files
+inside an ignored `generated` directory. The generated files are disposable and
+contain no real audio or user library data.
+
+## Edge Cases
+
+- Five manually maintained crates and two smart/dynamic crates.
+- Mixed supported audio extensions.
+- One reference whose file is absent.
+- One stale reference whose file has been renamed.
+- References shared by static and smart crates.
+
+## Verification
+
+`pytest` generates the sample in a temporary directory and verifies its crate,
+track, and missing-reference counts through the production parser and scanner.

samples/small-library/README.md

@@ -0,0 +1,25 @@
+# Small Synthetic Library
+
+This fixture models a small Serato migration without containing music or personal
+library data. It has 10 fake audio files, five static crates, two smart crates,
+one absent song, and one renamed song.
+
+Generate it with:
+
+```shell
+python3 samples/small-library/generate.py
+```
+
+Analyze it with:
+
+```shell
+python3 -m serato_doctor.cli \
+  --serato samples/small-library/generated/Serato/_Serato_ \
+  --music samples/small-library/generated/Music \
+  --out samples/small-library/generated/scan.csv \
+  --report samples/small-library/generated/report.txt
+```
+
+Expected CLI counts are 13 crate references, 10 disk tracks, and 2 references
+missing by filename. Everything under `generated/` is disposable and ignored by
+Git.

samples/small-library/generate.py

@@ -0,0 +1,50 @@
+"""Generate a disposable, synthetic Serato library from the sample manifest."""
+
+import argparse
+import json
+from pathlib import Path
+from typing import Optional
+
+
+SAMPLE_ROOT = Path(__file__).parent
+SERATO_PATH_PREFIX = "Users/djsplice/OneDrive/Jukebox/"
+
+
+def load_manifest() -> dict:
+    return json.loads((SAMPLE_ROOT / "manifest.json").read_text(encoding="utf-8"))
+
+
+def build_sample(output: Optional[Path] = None) -> Path:
+    root = output or SAMPLE_ROOT / "generated"
+    manifest = load_manifest()
+    music_root = root / "Music"
+    crate_root = root / "Serato" / "_Serato_" / "Subcrates"
+    crate_root.mkdir(parents=True, exist_ok=True)
+
+    for relative_path in manifest["tracks"]:
+        track_path = music_root / relative_path
+        track_path.parent.mkdir(parents=True, exist_ok=True)
+        track_path.write_bytes(
+            f"Synthetic Serato Doctor fixture: {relative_path}\n".encode("utf-8")
+        )
+
+    for crate in manifest["crates"]:
+        records = "".join(
+            f"{SERATO_PATH_PREFIX}{relative_path}otrk"
+            for relative_path in crate["references"]
+        )
+        (crate_root / crate["name"]).write_bytes(records.encode("utf-16-le"))
+
+    return root
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--output", type=Path)
+    args = parser.parse_args()
+    root = build_sample(args.output)
+    print(f"Generated synthetic library: {root}")
+
+
+if __name__ == "__main__":
+    main()

samples/small-library/manifest.json

@@ -0,0 +1,56 @@
+{
+  "tracks": [
+    "House/First.mp3",
+    "House/Second.m4a",
+    "Open Format/Third.wav",
+    "Open Format/Fourth.aif",
+    "Classics/Fifth.mp3",
+    "Classics/Sixth.flac",
+    "Warmup/Seventh.mp3",
+    "Warmup/Eighth.mp3",
+    "Renamed/New Name.mp3",
+    "Bonus/Ninth.MP3"
+  ],
+  "crates": [
+    {
+      "name": "House.crate",
+      "type": "static",
+      "references": ["House/First.mp3", "House/Second.m4a", "House/Missing.mp3"]
+    },
+    {
+      "name": "Open Format.crate",
+      "type": "static",
+      "references": ["Open Format/Third.wav", "Open Format/Fourth.aif"]
+    },
+    {
+      "name": "Classics.crate",
+      "type": "static",
+      "references": ["Classics/Fifth.mp3", "Classics/Sixth.flac"]
+    },
+    {
+      "name": "Renamed Tracks.crate",
+      "type": "static",
+      "references": ["Renamed/Old Name.mp3"]
+    },
+    {
+      "name": "Bonus.crate",
+      "type": "static",
+      "references": ["Bonus/Ninth.MP3"]
+    },
+    {
+      "name": "Compatible by key.crate",
+      "type": "smart",
+      "references": ["House/First.mp3", "House/Second.m4a"]
+    },
+    {
+      "name": "Smart Warmup.crate",
+      "type": "smart",
+      "references": ["Warmup/Seventh.mp3", "Warmup/Eighth.mp3"]
+    }
+  ],
+  "expected": {
+    "disk_tracks": 10,
+    "crate_references": 13,
+    "missing_by_filename": ["Missing.mp3", "Old Name.mp3"]
+  }
+}

tests/test_sample_library.py

@@ -0,0 +1,28 @@
+import runpy
+from pathlib import Path
+
+from serato_doctor.crate_parser import parse_crates
+from serato_doctor.models.library import Library
+from serato_doctor.scanner import scan_audio
+
+
+def test_generated_sample_library_has_expected_scenario(tmp_path):
+    generator_path = (
+        Path(__file__).parents[1] / "samples" / "small-library" / "generate.py"
+    )
+    build_sample = runpy.run_path(str(generator_path))["build_sample"]
+    sample_root = build_sample(tmp_path / "sample")
+
+    references = parse_crates(sample_root / "Serato" / "_Serato_" / "Subcrates")
+    tracks = scan_audio(sample_root / "Music")
+    results = Library.build(references, tracks).reconcile_by_filename()
+    missing = {
+        result.reference.filename
+        for result in results
+        if not result.exists_by_filename
+    }
+
+    assert len(list((sample_root / "Serato").rglob("*.crate"))) == 7
+    assert len(references) == 13
+    assert len(tracks) == 10
+    assert missing == {"Missing.mp3", "Old Name.mp3"}