add performer schema validator and template

src/performers/schema.py

@@ -0,0 +1,160 @@
+#!/usr/bin/env python3
+"""
+schema.py
+---------
+Performer metadata schema and validator for Goondex.
+
+Ensures that all performer.json files follow the canonical Goondex structure
+and provides tools to create, validate, and repair metadata records.
+
+Used by:
+  - verifier.py
+  - trainer.py
+  - tpdb_bridge.py
+  - utils.py
+"""
+
+import json
+from pathlib import Path
+from datetime import datetime
+from typing import Any, Dict
+
+
+# ============================================================
+# Canonical schema definition
+# ============================================================
+
+PERFORMER_TEMPLATE: Dict[str, Any] = {
+    "name": "",
+    "normalized_id": "",
+    "aliases": [],
+    "gender": "",
+    "birth_date": "",
+    "age": None,
+    "country": "",
+    "ethnicity": "",
+    "hair_color": "",
+    "eye_color": "",
+    "height_cm": None,
+    "measurements": "",
+    "bust_type": "",
+    "tattoos": [],
+    "piercings": [],
+    "career": {
+        "start_year": None,
+        "end_year": None
+    },
+    "external_ids": {
+        "tpdb": None,
+        "stashdb": None,
+        "iafd": None
+    },
+    "studios": {},
+    "known_galleries": 0,
+    "faces": [],
+    "images": [],
+    "notes": "",
+    "biography": "",
+    "last_updated": ""
+}
+
+
+# ============================================================
+# Utility functions
+# ============================================================
+
+def _now_iso() -> str:
+    """Return current UTC time in ISO format."""
+    return datetime.utcnow().replace(microsecond=0).isoformat() + "Z"
+
+
+def normalize_id(name: str) -> str:
+    """Convert performer name to a filesystem-safe ID."""
+    return name.lower().replace(" ", "_").replace("/", "_")
+
+
+def validate_performer_data(data: Dict[str, Any]) -> Dict[str, Any]:
+    """
+    Validate a performer.json dictionary against the canonical schema.
+    Any missing keys are added with defaults. Unknown keys are preserved.
+    """
+    validated = dict(PERFORMER_TEMPLATE)
+
+    for key, default in PERFORMER_TEMPLATE.items():
+        if key not in data:
+            validated[key] = default
+        else:
+            validated[key] = data[key]
+
+    # Always refresh timestamp
+    validated["last_updated"] = _now_iso()
+    return validated
+
+
+def create_blank_performer(name: str) -> Dict[str, Any]:
+    """Return a fresh performer dictionary using the template."""
+    data = dict(PERFORMER_TEMPLATE)
+    data["name"] = name
+    data["normalized_id"] = normalize_id(name)
+    data["last_updated"] = _now_iso()
+    return data
+
+
+def load_performer_json(path: Path) -> Dict[str, Any]:
+    """Load performer.json, validate, and return clean data."""
+    if not path.exists():
+        raise FileNotFoundError(f"Missing performer file: {path}")
+
+    try:
+        data = json.loads(path.read_text(encoding="utf-8"))
+        return validate_performer_data(data)
+    except json.JSONDecodeError as e:
+        raise ValueError(f"Invalid JSON format in {path}: {e}")
+
+
+def save_performer_json(path: Path, data: Dict[str, Any]):
+    """Write performer.json back to disk with pretty formatting."""
+    path.parent.mkdir(parents=True, exist_ok=True)
+    data["last_updated"] = _now_iso()
+    path.write_text(json.dumps(data, indent=2, ensure_ascii=False), encoding="utf-8")
+
+
+def ensure_performer_schema(performer_name: str, base_dir: Path) -> Path:
+    """
+    Ensure performer.json exists and matches schema.
+    Returns path to the verified performer.json file.
+    """
+    normalized = normalize_id(performer_name)
+    performer_dir = base_dir / normalized
+    performer_file = performer_dir / "performer.json"
+
+    if performer_file.exists():
+        try:
+            data = load_performer_json(performer_file)
+        except (ValueError, FileNotFoundError):
+            data = create_blank_performer(performer_name)
+    else:
+        data = create_blank_performer(performer_name)
+
+    save_performer_json(performer_file, data)
+    return performer_file
+
+
+# ============================================================
+# CLI (for manual testing)
+# ============================================================
+
+if __name__ == "__main__":
+    import argparse
+
+    parser = argparse.ArgumentParser(description="Validate or create performer schema")
+    parser.add_argument("name", help="Performer name (e.g. 'Jane Doe')")
+    parser.add_argument(
+        "--dir", default="data/faces",
+        help="Base directory where performer folders live"
+    )
+    args = parser.parse_args()
+
+    base_dir = Path(args.dir)
+    path = ensure_performer_schema(args.name, base_dir)
+    print(f"[OK] Performer schema verified at {path}")