merge: chore/cleanup-remove-bloat-and-secrets into main

app/domain/labels/sources/solana_labels_csv.py

@@ -0,0 +1,74 @@
+"""Solana wallet labels CSV source — 103K SOL labels via DuckDB."""
+from __future__ import annotations
+
+import asyncio
+import logging
+from datetime import UTC, datetime
+
+from app.domain.labels.models import Label, LabelSource
+
+log = logging.getLogger(__name__)
+
+CSV_PATH = "/home/dev/rmi/backend/data/wallet-labels-clean/wallet_labels_solana.csv"
+
+
+async def query_solana_labels_csv(address: str, chain: str = "solana") -> list[Label]:
+    """Query the Solana wallet labels CSV via DuckDB.
+
+    Schema (likely): address,source,name,entity_type,threat_group,risk_level
+
+    Only matches if chain == "solana" — returns [] for EVM chains.
+    """
+    if chain != "solana":
+        return []
+
+    def _query() -> list[Label]:
+        from app.core.duckdb_analytics import DuckDBAnalytics
+
+        try:
+            d = DuckDBAnalytics()
+            try:
+                rows = d.query_parquet(
+                    CSV_PATH,
+                    """
+                    SELECT *
+                    FROM data
+                    WHERE LOWER(address) = ?
+                    LIMIT 20
+                    """,
+                    [address.lower()],
+                )
+                labels = []
+                for row in rows:
+                    # Field names may differ — try common names
+                    label_value = (
+                        row.get("name")
+                        or row.get("label")
+                        or row.get("entity_type")
+                        or row.get("source")
+                    )
+                    if not label_value:
+                        continue
+                    labels.append(
+                        Label(
+                            address=row.get("address", address).lower(),
+                            chain="solana",
+                            label_type="entity",
+                            label_value=label_value,
+                            source=LabelSource.SOLANA_LABELS_CSV,
+                            confidence=0.8,
+                            verified_at=datetime.now(UTC),
+                            attributes={
+                                k: v for k, v in row.items()
+                                if k not in ("address", "name", "label") and v
+                            },
+                        )
+                    )
+                return labels
+            finally:
+                d.close()
+        except Exception as e:
+            log.warning("solana_labels_csv_query_fail address=%s err=%s", address[:10], e)
+            raise
+
+    return await asyncio.to_thread(_query)