PartitionedDml

PartitionedDml

This type of transaction is used to execute a single Partitioned DML statement. Partitioned DML partitions the key space and runs the DML statement over each partition in parallel using separate, internal transactions that commit independently.

Chances are, you'll never need to create a partitioned DML transaction directly, instead you'll want to use Database#runPartitionedUpdate.

Constructor

new PartitionedDml()

See:
  • Database#runPartitionedUpdate

Extends

Members

ended

Whether or not the transaction has ended. If true, make no further requests, and discard the transaction.

Overrides:

id

The transaction ID.

Overrides:

metadata

The raw transaction response object. It is populated after Snapshot#begin is called.

Overrides:

readTimestamp

Snapshot only The timestamp at which all reads are performed.

Overrides:

readTimestampProto

Snapshot only The protobuf version of Snapshot#readTimestamp. This is useful if you require microsecond precision.

Overrides:

Methods

createReadStream(table, query) → {ReadableStream}

Create a readable object stream to receive rows from the database using key lookups and scans.

Wrapper around v1.SpannerClient#streamingRead.

Parameters:
Name Type Description
table string

The table to read from.
query ReadRequest

Configuration object. See official ReadRequest. API documentation.
Returns:
Type Description
ReadableStream

A readable stream that emits rows.
Overrides:
See:
Fires:
  • PartialResultStream#event:response
  • PartialResultStream#event:stats
Examples
```
transaction.createReadStream('Singers', {
    keys: ['1'],
    columns: ['SingerId', 'name']
  })
  .on('error', function(err) {})
  .on('data', function(row) {
    // row = [
    //   {
    //     name: 'SingerId',
    //     value: '1'
    //   },
    //   {
    //     name: 'Name',
    //     value: 'Eddie Wilson'
    //   }
    // ]
  })
  .on('end', function() {
    // All results retrieved.
  });

```
Provide an array for `query.keys` to read with a
composite key.
```
const query = {
  keys: [
    [
      'Id1',
      'Name1'
    ],
    [
      'Id2',
      'Name2'
    ]
  ],
  // ...
};
```
Rows are returned as an array of object arrays. Each
object has a `name` and `value` property. To get a serialized object, call
`toJSON()`.
```
transaction.createReadStream('Singers', {
    keys: ['1'],
    columns: ['SingerId', 'name']
  })
  .on('error', function(err) {})
  .on('data', function(row) {
    // row.toJSON() = {
    //   SingerId: '1',
    //   Name: 'Eddie Wilson'
    // }
  })
  .on('end', function() {
    // All results retrieved.
  });
```
Alternatively, set `query.json` to `true`, and this step
will perform automatically.
```
transaction.createReadStream('Singers', {
    keys: ['1'],
    columns: ['SingerId', 'name'],
    json: true,
  })
  .on('error', function(err) {})
  .on('data', function(row) {
    // row = {
    //   SingerId: '1',
    //   Name: 'Eddie Wilson'
    // }
  })
  .on('end', function() {
    // All results retrieved.
  });
```
If you anticipate many results, you can end a stream
early to prevent unnecessary processing and API requests.
```
transaction.createReadStream('Singers', {
    keys: ['1'],
    columns: ['SingerId', 'name']
  })
  .on('data', function(row) {
    this.end();
  });
```

end()

Let the client know you're done with a particular transaction. This should mainly be called for Snapshot objects, however in certain cases you may want to call them for Transaction objects as well.

Overrides:
Examples
Calling `end` on a read only snapshot
```
database.getSnapshot((err, transaction) => {
  if (err) {
    // Error handling omitted.
  }

  transaction.run('SELECT * FROM Singers', (err, rows) => {
    if (err) {
      // Error handling omitted.
    }

    // End the snapshot.
    transaction.end();
  });
});
```
Calling `end` on a read/write transaction
```
database.runTransaction((err, transaction) => {
  if (err) {
    // Error handling omitted.
  }

  const query = 'UPDATE Account SET Balance = 1000 WHERE Key = 1';

  transaction.runUpdate(query, err => {
    if (err) {
      // In the event of an error, there would be nothing to rollback,
so
      // instead of continuing, discard the
transaction. transaction.end(); return;
    }

    transaction.commit(err => {});
  });
});
```

excludeTxnFromChangeStreams()

Use option excludeTxnFromChangeStreams to exclude partitionedDml queries from being tracked in change streams.

Enabling this options to true will effectively disable change stream tracking for a specified partitionedDml query, allowing write queries to operate without being included in change streams.

runStream(query) → {ReadableStream}

Create a readable object stream to receive resulting rows from a SQL statement.

Wrapper around v1.SpannerClient#executeStreamingSql.

Parameters:
Name Type Description
query string | ExecuteSqlRequest

A SQL query or ExecuteSqlRequest object.
Returns:
Type Description
ReadableStream
Overrides:
See:
Fires:
  • PartialResultStream#event:response
  • PartialResultStream#event:stats
Examples
```
const query = 'SELECT * FROM Singers';

transaction.runStream(query)
  .on('error', function(err) {})
  .on('data', function(row) {
    // row = {
    //   SingerId: '1',
    //   Name: 'Eddie Wilson'
    // }
  })
  .on('end', function() {
    // All results retrieved.
  });

```
The SQL query string can contain parameter placeholders.
A parameter placeholder consists of '@' followed by the parameter name.
```
const query = {
  sql: 'SELECT * FROM Singers WHERE name = @name',
  params: {
    name: 'Eddie Wilson'
  }
};

transaction.runStream(query)
  .on('error', function(err) {})
  .on('data', function(row) {})
  .on('end', function() {});
```
If you anticipate many results, you can end a stream
early to prevent unnecessary processing and API requests.
```
transaction.runStream(query)
  .on('data', function(row) {
    this.end();
  });
```