Add v2 snapshot schema validation and import_snapshots command

apps/ingestion/snapshots/schema.py

@@ -0,0 +1,182 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from datetime import date
+from typing import Any
+
+from django.utils.dateparse import parse_date
+
+REQUIRED_RECORD_FIELDS = {
+    "competition_external_id",
+    "competition_name",
+    "season",
+    "team_external_id",
+    "team_name",
+    "player_external_id",
+    "full_name",
+    "first_name",
+    "last_name",
+    "birth_date",
+    "nationality",
+    "height_cm",
+    "weight_kg",
+    "position",
+    "games_played",
+    "minutes_per_game",
+    "points_per_game",
+    "rebounds_per_game",
+    "assists_per_game",
+    "steals_per_game",
+    "blocks_per_game",
+    "turnovers_per_game",
+    "fg_pct",
+    "three_pt_pct",
+    "ft_pct",
+}
+
+ALLOWED_TOP_LEVEL_FIELDS = {
+    "source_name",
+    "snapshot_date",
+    "records",
+    "source_metadata",
+    "raw_payload",
+}
+
+ALLOWED_RECORD_FIELDS = REQUIRED_RECORD_FIELDS | {
+    "role",
+    "source_metadata",
+    "raw_payload",
+}
+
+
+@dataclass
+class SnapshotValidationResult:
+    source_name: str
+    snapshot_date: date
+    records: list[dict[str, Any]]
+
+
+class SnapshotValidationError(ValueError):
+    pass
+
+
+class SnapshotSchemaValidator:
+    """Strict JSON schema validator for HoopScout v2 player-season snapshots."""
+
+    @staticmethod
+    def _require_string(value: Any, field: str) -> str:
+        if not isinstance(value, str) or not value.strip():
+            raise SnapshotValidationError(f"{field} must be a non-empty string")
+        return value.strip()
+
+    @staticmethod
+    def _require_non_negative_int(value: Any, field: str) -> int:
+        if isinstance(value, bool):
+            raise SnapshotValidationError(f"{field} must be a non-negative integer")
+        try:
+            parsed = int(value)
+        except (TypeError, ValueError) as exc:
+            raise SnapshotValidationError(f"{field} must be a non-negative integer") from exc
+        if parsed < 0:
+            raise SnapshotValidationError(f"{field} must be a non-negative integer")
+        return parsed
+
+    @staticmethod
+    def _require_float(value: Any, field: str) -> float:
+        try:
+            parsed = float(value)
+        except (TypeError, ValueError) as exc:
+            raise SnapshotValidationError(f"{field} must be numeric") from exc
+        return parsed
+
+    @classmethod
+    def _validate_record(cls, record: dict[str, Any], index: int) -> dict[str, Any]:
+        unknown = set(record.keys()) - ALLOWED_RECORD_FIELDS
+        if unknown:
+            raise SnapshotValidationError(
+                f"record[{index}] contains unknown fields: {', '.join(sorted(unknown))}"
+            )
+
+        missing = REQUIRED_RECORD_FIELDS - set(record.keys())
+        if missing:
+            raise SnapshotValidationError(
+                f"record[{index}] missing required fields: {', '.join(sorted(missing))}"
+            )
+
+        normalized = dict(record)
+        for field in (
+            "competition_external_id",
+            "competition_name",
+            "season",
+            "team_external_id",
+            "team_name",
+            "player_external_id",
+            "full_name",
+            "first_name",
+            "last_name",
+            "nationality",
+            "position",
+        ):
+            normalized[field] = cls._require_string(record.get(field), f"record[{index}].{field}")
+
+        if record.get("role") is not None:
+            normalized["role"] = cls._require_string(record.get("role"), f"record[{index}].role")
+
+        birth_date = parse_date(str(record.get("birth_date")))
+        if not birth_date:
+            raise SnapshotValidationError(f"record[{index}].birth_date must be YYYY-MM-DD")
+        normalized["birth_date"] = birth_date.isoformat()
+
+        normalized["height_cm"] = cls._require_non_negative_int(record.get("height_cm"), f"record[{index}].height_cm")
+        normalized["weight_kg"] = cls._require_non_negative_int(record.get("weight_kg"), f"record[{index}].weight_kg")
+        normalized["games_played"] = cls._require_non_negative_int(record.get("games_played"), f"record[{index}].games_played")
+
+        for field in (
+            "minutes_per_game",
+            "points_per_game",
+            "rebounds_per_game",
+            "assists_per_game",
+            "steals_per_game",
+            "blocks_per_game",
+            "turnovers_per_game",
+            "fg_pct",
+            "three_pt_pct",
+            "ft_pct",
+        ):
+            normalized[field] = cls._require_float(record.get(field), f"record[{index}].{field}")
+
+        return normalized
+
+    @classmethod
+    def validate(cls, payload: dict[str, Any]) -> SnapshotValidationResult:
+        if not isinstance(payload, dict):
+            raise SnapshotValidationError("Snapshot root must be an object")
+
+        unknown = set(payload.keys()) - ALLOWED_TOP_LEVEL_FIELDS
+        if unknown:
+            raise SnapshotValidationError(
+                f"Snapshot contains unknown top-level fields: {', '.join(sorted(unknown))}"
+            )
+
+        source_name = cls._require_string(payload.get("source_name"), "source_name")
+
+        snapshot_date_raw = payload.get("snapshot_date")
+        snapshot_date = parse_date(str(snapshot_date_raw))
+        if not snapshot_date:
+            raise SnapshotValidationError("snapshot_date must be YYYY-MM-DD")
+
+        records = payload.get("records")
+        if not isinstance(records, list) or not records:
+            raise SnapshotValidationError("records must be a non-empty array")
+
+        normalized_records: list[dict[str, Any]] = []
+        for index, record in enumerate(records):
+            if not isinstance(record, dict):
+                raise SnapshotValidationError(f"record[{index}] must be an object")
+            normalized_records.append(cls._validate_record(record, index=index))
+
+        return SnapshotValidationResult(
+            source_name=source_name,
+            snapshot_date=snapshot_date,
+            records=normalized_records,
+        )