PMFlow/backend/app/services/signal/symbol_index/mixin_query_overview.py

from __future__ import annotations

from .shared import *

class SymbolIndexQueryOverviewMixin:
    async def search_symbols(
        self,
        *,
        query: str,
        language: Optional[str] = None,
        symbol_type: Optional[str] = None,
        limit: int = 30,
    ) -> dict[str, Any]:
        text = str(query or "").strip()
        if not text:
            return {"count": 0, "items": []}

        like = f"%{text}%"
        stmt = select(SymbolDefinition).where(
            and_(
                SymbolDefinition.tenant_id == self.tenant_id,
                SymbolDefinition.status == "active",
                or_(
                    SymbolDefinition.symbol_name.ilike(like),
                    SymbolDefinition.qualified_name.ilike(like),
                    SymbolDefinition.file_path.ilike(like),
                ),
            )
        )
        if language:
            stmt = stmt.where(SymbolDefinition.language == str(language).strip().lower()[:20])
        if symbol_type:
            stmt = stmt.where(SymbolDefinition.symbol_type == str(symbol_type).strip().lower()[:32])
        rows = (
            await self.db.execute(
                stmt.order_by(SymbolDefinition.updated_at.desc(), SymbolDefinition.id.desc()).limit(max(1, min(int(limit), 200)))
            )
        ).scalars().all()
        return {
            "count": len(rows),
            "items": [self._serialize_symbol(row) for row in rows],
        }

    async def get_symbol_detail(self, *, symbol_id: int) -> dict[str, Any]:
        symbol = (
            await self.db.execute(
                select(SymbolDefinition).where(
                    and_(
                        SymbolDefinition.tenant_id == self.tenant_id,
                        SymbolDefinition.id == int(symbol_id),
                    )
                )
            )
        ).scalar_one_or_none()
        if symbol is None:
            raise ValueError("symbol_not_found")

        refs = (
            await self.db.execute(
                select(SymbolReference).where(
                    and_(
                        SymbolReference.tenant_id == self.tenant_id,
                        SymbolReference.source_symbol_id == int(symbol.id),
                    )
                )
                .order_by(SymbolReference.id.desc())
                .limit(120)
            )
        ).scalars().all()

        inbound_refs = (
            await self.db.execute(
                select(SymbolReference)
                .where(
                    and_(
                        SymbolReference.tenant_id == self.tenant_id,
                        or_(
                            SymbolReference.target_symbol_name == str(symbol.symbol_name),
                            SymbolReference.target_qualified_name == str(symbol.qualified_name or ""),
                        ),
                    )
                )
                .order_by(SymbolReference.id.desc())
                .limit(120)
            )
        ).scalars().all()

        return {
            "symbol": self._serialize_symbol(symbol),
            "outbound_refs": [self._serialize_ref(row) for row in refs],
            "inbound_refs": [self._serialize_ref(row) for row in inbound_refs],
        }

    async def get_overview(
        self,
        *,
        top_n_files: int = 10,
        auto_sync_if_empty: bool = False,
        base_dir: Optional[str] = None,
    ) -> dict[str, Any]:
        total = int(
            (
                await self.db.execute(
                    select(func.count(SymbolDefinition.id)).where(SymbolDefinition.tenant_id == self.tenant_id)
                )
            ).scalar_one()
            or 0
        )
        if total == 0 and auto_sync_if_empty:
            await self.sync_codebase_symbols(base_dir=base_dir, max_files=300, force_reindex=False)
            total = int(
                (
                    await self.db.execute(
                        select(func.count(SymbolDefinition.id)).where(SymbolDefinition.tenant_id == self.tenant_id)
                    )
                ).scalar_one()
                or 0
            )

        language_groups = (
            await self.db.execute(
                select(SymbolDefinition.language, func.count(SymbolDefinition.id))
                .where(SymbolDefinition.tenant_id == self.tenant_id)
                .group_by(SymbolDefinition.language)
            )
        ).all()
        type_groups = (
            await self.db.execute(
                select(SymbolDefinition.symbol_type, func.count(SymbolDefinition.id))
                .where(SymbolDefinition.tenant_id == self.tenant_id)
                .group_by(SymbolDefinition.symbol_type)
            )
        ).all()
        file_groups = (
            await self.db.execute(
                select(SymbolDefinition.file_path, func.count(SymbolDefinition.id))
                .where(SymbolDefinition.tenant_id == self.tenant_id)
                .group_by(SymbolDefinition.file_path)
                .order_by(func.count(SymbolDefinition.id).desc())
                .limit(max(1, min(int(top_n_files), 50)))
            )
        ).all()
        ref_total = int(
            (
                await self.db.execute(
                    select(func.count(SymbolReference.id)).where(SymbolReference.tenant_id == self.tenant_id)
                )
            ).scalar_one()
            or 0
        )
        indexed_files_total = int(
            (
                await self.db.execute(
                    select(func.count(func.distinct(SymbolDefinition.file_path))).where(
                        SymbolDefinition.tenant_id == self.tenant_id
                    )
                )
            ).scalar_one()
            or 0
        )
        return {
            "tenant_id": self.tenant_id,
            "indexed_files": indexed_files_total,
            "symbol_total": total,
            "reference_total": ref_total,
            "python_symbols": int(dict(language_groups).get("python", 0) or 0),
            "typescript_symbols": int(dict(language_groups).get("typescript", 0) or 0),
            "symbols_by_language": {str(k): int(v or 0) for k, v in language_groups},
            "symbols_by_type": {str(k): int(v or 0) for k, v in type_groups},
            "top_modules": [
                {"module": str(path), "symbols": int(cnt or 0)}
                for path, cnt in file_groups
            ],
            "snapshot_at": now_cn_aware().isoformat(),
        }