插入、更新或删除 Spanner Graph 数据

注意：此功能适用于 Spanner 企业版和企业 Plus 版。如需了解详情，请参阅 Spanner 版本概览。

本文档介绍了如何使用Google Cloud 控制台、Google Cloud CLI 和客户端库在 Spanner 图中更改数据。对图数据进行更改包括在 Spanner 图中插入、更新和删除节点和边。

Spanner Graph 会将表中的数据映射到图节点和边。如需在图表中更改数据，您必须更改相应输入表中的数据。如需了解详情，请参阅 Spanner 图架构概览。

准备工作

若要完成本文档中的 Google Cloud 控制台步骤和代码示例，您必须先按照设置和查询 Spanner 图中的步骤执行以下操作：

1. 创建实例。
2. 创建采用 Spanner 架构的数据库。
3. 插入图表数据。

插入节点或边缘

如需插入节点或边，请使用 Google Cloud 控制台、Google Cloud CLI 或 Spanner 客户端库将行插入节点表或边表。

在 Google Cloud 控制台和 Google Cloud CLI 中，您可以使用 GoogleSQL 数据操纵语言 (DML) 进行插入。在 Spanner 客户端库中，您可以使用 DML 或 Mutation API。

在插入边之前，请确保由边连接的源节点和目标节点存在。如果您在插入边时，边所连接的源节点或目标节点不存在，则可能会收到引用完整性违规错误。如需了解详情，请参阅缺少源节点违反了 INTERLEAVE IN 关系和缺少目的地节点违反了外键约束条件。

以下示例会将 Account 节点和 Transfer 边插入图表中：

控制台

1. 在 Google Cloud 控制台中运行语句。

2. 在 Google Cloud 控制台中，输入以下 DML 语句，然后点击Run Query（运行查询）：

-- Insert 2 Account nodes.
INSERT INTO Account (id, create_time, is_blocked)
VALUES (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false);
INSERT INTO Account (id, create_time, is_blocked)
VALUES (2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true);

-- Insert 2 Transfer edges.
INSERT INTO AccountTransferAccount (id, to_id, create_time, amount)
VALUES (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100);
INSERT INTO AccountTransferAccount (id, to_id, create_time, amount)
VALUES (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200);

gcloud

1. 使用 gcloud CLI 执行语句。
2. 在 gcloud CLI 中，运行以下命令：

gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="INSERT INTO Account (id, create_time, is_blocked) VALUES (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false)"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="INSERT INTO Account (id, create_time, is_blocked) VALUES (2, CAST('2000-08-12 07:13:16.463959-03:41'  AS TIMESTAMP), true)"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="INSERT INTO AccountTransferAccount (id, to_id, create_time, amount) VALUES (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100)"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="INSERT INTO AccountTransferAccount (id, to_id, create_time, amount) VALUES (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200)"

客户端库

Python

def insert_data_with_dml(instance_id, database_id):
    """Inserts sample data into the given database using a DML statement."""

    spanner_client = spanner.Client()
    instance = spanner_client.instance(instance_id)
    database = instance.database(database_id)

    def insert_accounts(transaction):
        row_ct = transaction.execute_update(
            "INSERT INTO Account (id, create_time, is_blocked) "
            "  VALUES"
            "    (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false),"
            "    (2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true)"
        )

        print("{} record(s) inserted into Account.".format(row_ct))

    def insert_transfers(transaction):
        row_ct = transaction.execute_update(
            "INSERT INTO AccountTransferAccount (id, to_id, create_time, amount) "
            "  VALUES"
            "    (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100),"
            "    (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200) "
        )

        print("{} record(s) inserted into AccountTransferAccount.".format(row_ct))

    database.run_in_transaction(insert_accounts)
    database.run_in_transaction(insert_transfers)

Java

static void insertUsingDml(DatabaseClient dbClient) {
  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql =
                "INSERT INTO Account (id, create_time, is_blocked) "
                    + "  VALUES"
                    + "    (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false),"
                    + "    (2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true)";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d record(s) inserted into Account.\n", rowCount);
            return null;
          });

  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql =
                "INSERT INTO AccountTransferAccount (id, to_id, create_time, amount) "
                    + "  VALUES"
                    + "    (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100),"
                    + "    (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200) ";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d record(s) inserted into AccountTransferAccount.\n", rowCount);
            return null;
          });
}

Go

import (
	"context"
	"fmt"
	"io"

	"cloud.google.com/go/spanner"
)

func insertGraphDataWithDml(w io.Writer, db string) error {
	ctx := context.Background()
	client, err := spanner.NewClient(ctx, db)
	if err != nil {
		return err
	}
	defer client.Close()

	// Execute a ReadWriteTransaction to insert values into the 'Account' table
	// underpinning 'Account' nodes in 'FinGraph'. The function run by ReadWriteTransaction
	// executes an 'INSERT' SQL DML statement. Graph queries run after this
	// transaction is committed will observe the effects of the new 'Account's
	// added to the graph.
	_, err1 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `INSERT INTO Account (id, create_time, is_blocked)
            		VALUES
            	    	(1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false),
            			(2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true)`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d Account record(s) inserted.\n", rowCount)
		return err
	})

	if err1 != nil {
		return err1
	}

	// Execute a ReadWriteTransaction to insert values into the 'AccountTransferAccount'
	// table underpinning 'AccountTransferAccount' edges in 'FinGraph'. The function run
	// by ReadWriteTransaction executes an 'INSERT' SQL DML statement.
	// Graph queries run after this transaction is committed will observe the effects
	// of the edges added to the graph.
	_, err2 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `INSERT INTO AccountTransferAccount (id, to_id, create_time, amount)
					VALUES
						(1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100),
						(1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200)`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d AccountTransferAccount record(s) inserted.\n", rowCount)
		return err
	})

	return err2
}

C++

void InsertDataWithDml(google::cloud::spanner::Client client) {
  using ::google::cloud::StatusOr;
  namespace spanner = ::google::cloud::spanner;

  std::int64_t rows_inserted;
  auto commit_result = client.Commit(
      [&client, &rows_inserted](
          spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto insert =
            client.ExecuteDml(std::move(txn), spanner::SqlStatement(R"""(
          INSERT INTO Account (id, create_time, is_blocked)
          VALUES
          (1, CAST('2000-08-10 08:18:48.463959-07:52' AS TIMESTAMP), false),
          (2, CAST('2000-08-12 07:13:16.463959-03:41' AS TIMESTAMP), true)
        )"""));
        if (!insert) return std::move(insert).status();
        rows_inserted = insert->RowsModified();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();
  std::cout << "Rows inserted into Account: " << rows_inserted << "\n";

  commit_result = client.Commit(
      [&client, &rows_inserted](
          spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto insert =
            client.ExecuteDml(std::move(txn), spanner::SqlStatement(R"""(
          INSERT INTO AccountTransferAccount (id, to_id, create_time, amount)
          VALUES
          (1, 2, CAST('2000-09-11 03:11:18.463959-06:36' AS TIMESTAMP), 100),
          (1, 1, CAST('2000-09-12 04:09:34.463959-05:12' AS TIMESTAMP), 200)
        )"""));
        if (!insert) return std::move(insert).status();
        rows_inserted = insert->RowsModified();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();
  std::cout << "Rows inserted into AccountTransferAccount: " << rows_inserted
            << "\n";

  std::cout << "Insert was successful [spanner_insert_graph_data_with_dml]\n";
}

更新节点或边

如需更新现有节点或边，请使用 Google Cloud 控制台、gcloud CLI 或 Spanner 客户端库。

您可以使用 GoogleSQL 数据操纵语言 (DML) 语句或带有 DML 语句的 Spanner 图查询更新现有节点或边。在 Spanner 客户端库中，您还可以使用 Mutation API。

使用 DML 更新节点或边

以下示例使用 DML 更新图中的 Account 节点和 Transfer 边：

控制台

1. 在 Google Cloud 控制台中运行语句。

2. 在 Google Cloud 控制台中，输入以下 DML 语句，然后点击Run Query（运行查询）：

-- Update Account node
UPDATE Account SET is_blocked = false WHERE id = 2;

-- Update Transfer edge
UPDATE AccountTransferAccount
SET amount = 300
WHERE id = 1 AND to_id = 2;

控制台

1. 使用 gcloud CLI 执行语句。
2. 在 gcloud CLI 中，运行以下命令：

gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="UPDATE Account SET is_blocked = false WHERE id = 2"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2"

客户端库

Python

def update_data_with_dml(instance_id, database_id):
    """Updates sample data from the database using a DML statement."""

    spanner_client = spanner.Client()
    instance = spanner_client.instance(instance_id)
    database = instance.database(database_id)

    def update_accounts(transaction):
        row_ct = transaction.execute_update(
            "UPDATE Account SET is_blocked = false WHERE id = 2"
        )

        print("{} Account record(s) updated.".format(row_ct))

    def update_transfers(transaction):
        row_ct = transaction.execute_update(
            "UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2"
        )

        print("{} AccountTransferAccount record(s) updated.".format(row_ct))

    database.run_in_transaction(update_accounts)
    database.run_in_transaction(update_transfers)

Java

static void updateUsingDml(DatabaseClient dbClient) {
  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql = "UPDATE Account SET is_blocked = false WHERE id = 2";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d Account record(s) updated.\n", rowCount);
            return null;
          });

  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql =
                "UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d AccountTransferAccount record(s) updated.\n", rowCount);
            return null;
          });
}

Go

import (
	"context"
	"fmt"
	"io"

	"cloud.google.com/go/spanner"
)

func updateGraphDataWithDml(w io.Writer, db string) error {
	ctx := context.Background()
	client, err := spanner.NewClient(ctx, db)
	if err != nil {
		return err
	}
	defer client.Close()

	// Execute a ReadWriteTransaction to update the 'Account' table underpinning
	// 'Account' nodes in 'FinGraph'. The function run by ReadWriteTransaction
	// executes an 'UPDATE' SQL DML statement. Graph queries run after this
	// transaction is committed will observe the effects of the update to 'Account'
	// with 'id' = 2.
	_, err1 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `UPDATE Account SET is_blocked = false WHERE id = 2`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d Account record(s) updated.\n", rowCount)
		return err
	})

	if err1 != nil {
		return err1
	}

	// Execute a ReadWriteTransaction to update the 'AccountTransferAccount' table
	// underpinning 'AccountTransferAccount' edges in 'FinGraph'. The function run
	// by ReadWriteTransaction executes an 'UPDATE' SQL DML statement.
	// Graph queries run after this transaction is committed will observe the effects
	// of the update to 'AccountTransferAccount' where the source of the transfer has
	// 'id' 1 and the destination has 'id' 2.
	_, err2 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d AccountTransferAccount record(s) updated.\n", rowCount)
		return err
	})

	return err2
}

C++

void UpdateDataWithDml(google::cloud::spanner::Client client) {
  using ::google::cloud::StatusOr;
  namespace spanner = ::google::cloud::spanner;

  auto commit_result = client.Commit(
      [&client](spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto update = client.ExecuteDml(
            std::move(txn),
            spanner::SqlStatement(
                "UPDATE Account SET is_blocked = false WHERE id = 2"));
        if (!update) return std::move(update).status();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();

  commit_result = client.Commit(
      [&client](spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto update =
            client.ExecuteDml(std::move(txn), spanner::SqlStatement(R"""(
          UPDATE AccountTransferAccount
            SET amount = 300 WHERE id = 1 AND to_id = 2)"""));
        if (!update) return std::move(update).status();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();

  std::cout << "Update was successful [spanner_update_graph_data_with_dml]\n";
}

使用图查询和 DML 更新节点或边

以下示例使用 DML 与 Spanner 图查询更新图中的 Account 节点和 Transfer 边：

控制台

1. 在 Google Cloud 控制台中运行语句。

2. 在 Google Cloud 控制台中，输入以下包含 DML 语句的 Spanner 图查询，然后点击运行查询：

-- Use Graph pattern matching to identify Account nodes to update:
UPDATE Account SET is_blocked = false
WHERE id IN {
  GRAPH FinGraph
  MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)
  RETURN b.id
}

gcloud

1. 使用 gcloud CLI 执行语句。
2. 在 gcloud CLI 中，运行以下命令：

gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="UPDATE Account SET is_blocked = false"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="UPDATE AccountTransferAccount SET amount = 300 WHERE id = 1 AND to_id = 2"
    --sql=" WHERE id IN { GRAPH FinGraph MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account) RETURN b.id }"

客户端库

Python

def update_data_with_graph_query_in_dml(instance_id, database_id):
    """Updates sample data from the database using a DML statement."""

    spanner_client = spanner.Client()
    instance = spanner_client.instance(instance_id)
    database = instance.database(database_id)

    def update_accounts(transaction):
        row_ct = transaction.execute_update(
            "UPDATE Account SET is_blocked = true "
            "WHERE id IN {"
            "  GRAPH FinGraph"
            "  MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)"
            "  RETURN b.id}"
        )

        print("{} Account record(s) updated.".format(row_ct))

    database.run_in_transaction(update_accounts)

Java

static void updateUsingGraphQueryInDml(DatabaseClient dbClient) {
  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql =
                "UPDATE Account SET is_blocked = true "
                    + "WHERE id IN {"
                    + "  GRAPH FinGraph"
                    + "  MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)"
                    + "  RETURN b.id}";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d Account record(s) updated.\n", rowCount);
            return null;
          });
}

Go

import (
	"context"
	"fmt"
	"io"

	"cloud.google.com/go/spanner"
)

func updateGraphDataWithGraphQueryInDml(w io.Writer, db string) error {
	ctx := context.Background()
	client, err := spanner.NewClient(ctx, db)
	if err != nil {
		return err
	}
	defer client.Close()

	// Execute a ReadWriteTransaction to update the 'Account' table underpinning
	// 'Account' nodes in 'FinGraph'. The function run by ReadWriteTransaction
	// executes an 'UPDATE' SQL DML statement. Graph queries run after this
	// transaction is committed will observe the effects of the updates to 'Account's
	//
	// The update is performed for all 'Account's whose 'id' is returned by
	// the graph query in the 'IN' subquery, i.e., all 'Account's that have
	// received transfers directly or via one intermediary from an 'Account'
	// whose 'id' is 1.
	_, err = client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{
			SQL: `UPDATE Account SET is_blocked = true 
            	  WHERE id IN {
            	    GRAPH FinGraph 
            	    MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)
            	    RETURN b.id}`,
		}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d Account record(s) updated.\n", rowCount)
		return err
	})

	return err
}

C++

void UpdateDataWithGraphQueryInDml(google::cloud::spanner::Client client) {
  using ::google::cloud::StatusOr;
  namespace spanner = ::google::cloud::spanner;
  auto commit_result = client.Commit(
      [&client](spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto update =
            client.ExecuteDml(std::move(txn), spanner::SqlStatement(R"""(
              UPDATE Account SET is_blocked = true
              WHERE id IN {
                GRAPH FinGraph
                MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->{1,2}(b:Account)
                RETURN b.id})"""));
        if (!update) return std::move(update).status();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();
  std::cout << "Update was successful "
            << "[spanner_update_graph_data_with_graph_query_in_dml]\n";
}

删除节点或边

如需删除现有节点或边，请使用 Google Cloud 控制台、gcloud CLI 或 Spanner 客户端库。

在 Google Cloud 控制台和 gcloud CLI 中，您可以使用 GoogleSQL 数据操纵语言 (DML) 进行删除。在 Spanner 客户端库中，您可以使用 DML 或 Mutation API。

在删除节点之前，请确保不存在指向该节点的边。如果存在这类边，您可能会收到引用完整性违规错误。如需了解详情，请参阅孤儿出边违反了父子关系和孤儿入边违反了父子关系。

以下示例会从图中删除 Transfer 边和 Account 节点。

控制台

1. 在 Google Cloud 控制台中运行语句。
2. 在 Google Cloud 控制台中，输入以下 DML 语句，然后点击Run Query（运行查询）：

-- Delete Transfer edge
DELETE FROM AccountTransferAccount
WHERE id = 1 AND to_id = 2;

-- Delete Account node
DELETE FROM Account WHERE id = 2;

gcloud

1. 使用 gcloud CLI 执行语句。
2. 在 gcloud CLI 中，运行以下命令：

gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2"
gcloud spanner databases execute-sql example-db --instance=test-instance \
    --sql="DELETE FROM Account WHERE id = 2"

客户端库

Python

def delete_data_with_dml(instance_id, database_id):
    """Deletes sample data from the database using a DML statement."""

    spanner_client = spanner.Client()
    instance = spanner_client.instance(instance_id)
    database = instance.database(database_id)

    def delete_transfers(transaction):
        row_ct = transaction.execute_update(
            "DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2"
        )

        print("{} AccountTransferAccount record(s) deleted.".format(row_ct))

    def delete_accounts(transaction):
        row_ct = transaction.execute_update("DELETE FROM Account WHERE id = 2")

        print("{} Account record(s) deleted.".format(row_ct))

    database.run_in_transaction(delete_transfers)
    database.run_in_transaction(delete_accounts)

Java

static void deleteUsingDml(DatabaseClient dbClient) {
  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql = "DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d AccountTransferAccount record(s) deleted.\n", rowCount);
            return null;
          });

  dbClient
      .readWriteTransaction()
      .run(
          transaction -> {
            String sql = "DELETE FROM Account WHERE id = 2";
            long rowCount = transaction.executeUpdate(Statement.of(sql));
            System.out.printf("%d Account record(s) deleted.\n", rowCount);
            return null;
          });
}

Go

// Copyright 2024 Google LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     https://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

package spanner


import (
	"context"
	"fmt"
	"io"

	"cloud.google.com/go/spanner"
)

func deleteGraphDataWithDml(w io.Writer, db string) error {
	ctx := context.Background()
	client, err := spanner.NewClient(ctx, db)
	if err != nil {
		return err
	}
	defer client.Close()

	// Execute a ReadWriteTransaction to update the 'AccountTransferAccount'
	// table underpinning 'AccountTransferAccount' edges in 'FinGraph'. The
	// function run by ReadWriteTransaction executes an 'DELETE' SQL DML
	// statement. This has the effect of deleting the 'AccountTransferAccount'
	// edge where the source 'id' is 1 and the destination 'id' is 2 from the graph.
	_, err1 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{SQL: `DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2`}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d AccountTransferAccount record(s) deleted.\n", rowCount)
		return nil
	})

	if err1 != nil {
		return err1
	}

	// Execute a ReadWriteTransaction to update the 'Account' table underpinning
	//'Account' nodes in 'FinGraph'. In 'FinGraph', nodes can only be deleted
	// after any edges referencing the nodes have been deleted first. The function
	// run by ReadWriteTransaction executes an 'DELETE' SQL DML statement. This has
	// the effect of deleting the 'Account' node whose 'id' is 1 from the graph.
	_, err2 := client.ReadWriteTransaction(ctx, func(ctx context.Context, txn *spanner.ReadWriteTransaction) error {
		stmt := spanner.Statement{SQL: `DELETE FROM Account WHERE id = 2`}
		rowCount, err := txn.Update(ctx, stmt)
		if err != nil {
			return err
		}
		fmt.Fprintf(w, "%d Account record(s) deleted.\n", rowCount)
		return nil
	})

	return err2
}

C++

void DeleteDataWithDml(google::cloud::spanner::Client client) {
  using ::google::cloud::StatusOr;
  namespace spanner = ::google::cloud::spanner;

  auto commit_result = client.Commit([&client](spanner::Transaction txn)
                                         -> StatusOr<spanner::Mutations> {
    auto deleted = client.ExecuteDml(
        std::move(txn),
        spanner::SqlStatement(
            "DELETE FROM AccountTransferAccount WHERE id = 1 AND to_id = 2"));
    if (!deleted) return std::move(deleted).status();
    return spanner::Mutations{};
  });
  if (!commit_result) throw std::move(commit_result).status();

  commit_result = client.Commit(
      [&client](spanner::Transaction txn) -> StatusOr<spanner::Mutations> {
        auto deleted = client.ExecuteDml(
            std::move(txn),
            spanner::SqlStatement("DELETE FROM Account WHERE id = 2"));
        if (!deleted) return std::move(deleted).status();
        return spanner::Mutations{};
      });
  if (!commit_result) throw std::move(commit_result).status();

  std::cout << "Delete was successful [spanner_delete_graph_data_with_dml]\n";
}

您可以将 Spanner Graph 查询与 DML 语句组合使用，如以下示例所示：

  -- Use Graph pattern matching to identify Account nodes to delete:
  DELETE FROM AccountTransferAccount
  WHERE id IN {
    GRAPH FinGraph
    MATCH (a:Account WHERE a.id = 1)-[:TRANSFERS]->(b:Account)
    RETURN b.id
  }

用于更改图数据的选项

您可以通过以下方式自动更改图数据：

• 使用 ON DELETE CASCADE 操作自动删除图中的边。
• 使用 TTL 政策自动删除图中的节点和边。如需了解详情，请参阅节点和边缘的 TTL。

如需高效地批量更新和删除图中的节点和边，请使用分区 DML。

后续步骤

• 阅读 Spanner 图查询概览。
• 了解调整 Spanner 图查询的最佳实践。