prop-data-guard/backend/app/services/metadata_service.py.bak

from typing import Optional, List, Tuple
from sqlalchemy.orm import Session
from fastapi import HTTPException, status

from app.models.metadata import DataSource, Database, DataTable, DataColumn
from app.services.datasource_service import get_datasource, _decrypt_password


def get_database(db: Session, db_id: int) -> Optional[Database]:
    return db.query(Database).filter(Database.id == db_id).first()


def get_table(db: Session, table_id: int) -> Optional[DataTable]:
    return db.query(DataTable).filter(DataTable.id == table_id).first()


def get_column(db: Session, column_id: int) -> Optional[DataColumn]:
    return db.query(DataColumn).filter(DataColumn.id == column_id).first()


def list_databases(db: Session, source_id: Optional[int] = None) -> List[Database]:
    query = db.query(Database)
    if source_id:
        query = query.filter(Database.source_id == source_id)
    return query.all()


def list_tables(db: Session, database_id: Optional[int] = None, keyword: Optional[str] = None) -> Tuple[List[DataTable], int]:
    query = db.query(DataTable)
    if database_id:
        query = query.filter(DataTable.database_id == database_id)
    if keyword:
        query = query.filter(
            (DataTable.name.contains(keyword)) | (DataTable.comment.contains(keyword))
        )
    return query.all(), query.count()


def list_columns(db: Session, table_id: Optional[int] = None, keyword: Optional[str] = None, page: int = 1, page_size: int = 50) -> Tuple[List[DataColumn], int]:
    query = db.query(DataColumn)
    if table_id:
        query = query.filter(DataColumn.table_id == table_id)
    if keyword:
        query = query.filter(
            (DataColumn.name.contains(keyword)) | (DataColumn.comment.contains(keyword))
        )
    total = query.count()
    items = query.offset((page - 1) * page_size).limit(page_size).all()
    return items, total


def build_tree(db: Session, source_id: Optional[int] = None) -> List[dict]:
    sources = db.query(DataSource)
    if source_id:
        sources = sources.filter(DataSource.id == source_id)
    sources = sources.all()

    result = []
    for s in sources:
        source_node = {
            "id": s.id,
            "name": s.name,
            "type": "source",
            "children": [],
            "meta": {"source_type": s.source_type, "status": s.status},
        }
        for d in s.databases:
            db_node = {
                "id": d.id,
                "name": d.name,
                "type": "database",
                "children": [],
                "meta": {"charset": d.charset, "table_count": d.table_count},
            }
            for t in d.tables:
                table_node = {
                    "id": t.id,
                    "name": t.name,
                    "type": "table",
                    "children": [],
                    "meta": {"comment": t.comment, "row_count": t.row_count, "column_count": t.column_count},
                }
                db_node["children"].append(table_node)
            source_node["children"].append(db_node)
        result.append(source_node)
    return result


def sync_metadata(db: Session, source_id: int, user_id: int) -> dict:
    from sqlalchemy import create_engine, inspect, text
    import json

    source = get_datasource(db, source_id)
    if not source:
        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="数据源不存在")

    driver_map = {
        "mysql": "mysql+pymysql",
        "postgresql": "postgresql+psycopg2",
        "oracle": "oracle+cx_oracle",
        "sqlserver": "mssql+pymssql",
    }
    driver = driver_map.get(source.source_type, source.source_type)

    if source.source_type == "dm":
        return {"success": True, "message": "达梦数据库同步成功（模拟）", "databases": 0, "tables": 0, "columns": 0}

    password = ""
    if source.encrypted_password:
        try:
            password = _decrypt_password(source.encrypted_password)
        except Exception:
            pass

    try:
        url = f"{driver}://{source.username}:{password}@{source.host}:{source.port}/{source.database_name}"
        engine = create_engine(url, pool_pre_ping=True)
        inspector = inspect(engine)

        db_names = inspector.get_schema_names() or [source.database_name]
        total_tables = 0
        total_columns = 0

        for db_name in db_names:
            db_obj = db.query(Database).filter(Database.source_id == source.id, Database.name == db_name).first()
            if not db_obj:
                db_obj = Database(source_id=source.id, name=db_name)
                db.add(db_obj)
                db.commit()
                db.refresh(db_obj)

            table_names = inspector.get_table_names(schema=db_name)
            for tname in table_names:
                table_obj = db.query(DataTable).filter(DataTable.database_id == db_obj.id, DataTable.name == tname).first()
                if not table_obj:
                    table_obj = DataTable(database_id=db_obj.id, name=tname)
                    db.add(table_obj)
                    db.commit()
                    db.refresh(table_obj)

                columns = inspector.get_columns(tname, schema=db_name)
                for col in columns:
                    col_obj = db.query(DataColumn).filter(DataColumn.table_id == table_obj.id, DataColumn.name == col["name"]).first()
                    if not col_obj:
                        sample = None
                        try:
                            with engine.connect() as conn:
                                result = conn.execute(text(f'SELECT "{col["name"]}" FROM "{db_name}"."{tname}" LIMIT 5'))
                                samples = [str(r[0]) for r in result if r[0] is not None]
                                sample = json.dumps(samples, ensure_ascii=False)
                        except Exception:
                            pass

                        col_obj = DataColumn(
                            table_id=table_obj.id,
                            name=col["name"],
                            data_type=str(col.get("type", "")),
                            length=col.get("max_length"),
                            comment=col.get("comment"),
                            is_nullable=col.get("nullable", True),
                            sample_data=sample,
                        )
                        db.add(col_obj)
                        total_columns += 1

                total_tables += 1

            db.commit()

        source.status = "active"
        db.commit()

        return {
            "success": True,
            "message": "元数据同步成功",
            "databases": len(db_names),
            "tables": total_tables,
            "columns": total_columns,
        }
    except Exception as e:
        source.status = "error"
        db.commit()
        return {"success": False, "message": f"同步失败: {str(e)}", "databases": 0, "tables": 0, "columns": 0}