2
2
mirror of https://github.com/octoleo/restic.git synced 2024-11-29 16:23:59 +00:00
restic/vendor/cloud.google.com/go/bigquery/query.go
Alexander Neumann 61cb1cc6f8 Update vendored dependencies
This includes github.com/kurin/blazer 0.2.0, which resolves #1291
2017-10-01 10:13:39 +02:00

219 lines
7.4 KiB
Go

// Copyright 2015 Google Inc. All Rights Reserved.
//
// 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
//
// http://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 bigquery
import (
"errors"
"golang.org/x/net/context"
bq "google.golang.org/api/bigquery/v2"
)
// QueryConfig holds the configuration for a query job.
type QueryConfig struct {
// JobID is the ID to use for the job. If empty, a random job ID will be generated.
JobID string
// If AddJobIDSuffix is true, then a random string will be appended to JobID.
AddJobIDSuffix bool
// Dst is the table into which the results of the query will be written.
// If this field is nil, a temporary table will be created.
Dst *Table
// The query to execute. See https://cloud.google.com/bigquery/query-reference for details.
Q string
// DefaultProjectID and DefaultDatasetID specify the dataset to use for unqualified table names in the query.
// If DefaultProjectID is set, DefaultDatasetID must also be set.
DefaultProjectID string
DefaultDatasetID string
// TableDefinitions describes data sources outside of BigQuery.
// The map keys may be used as table names in the query string.
TableDefinitions map[string]ExternalData
// CreateDisposition specifies the circumstances under which the destination table will be created.
// The default is CreateIfNeeded.
CreateDisposition TableCreateDisposition
// WriteDisposition specifies how existing data in the destination table is treated.
// The default is WriteEmpty.
WriteDisposition TableWriteDisposition
// DisableQueryCache prevents results being fetched from the query cache.
// If this field is false, results are fetched from the cache if they are available.
// The query cache is a best-effort cache that is flushed whenever tables in the query are modified.
// Cached results are only available when TableID is unspecified in the query's destination Table.
// For more information, see https://cloud.google.com/bigquery/querying-data#querycaching
DisableQueryCache bool
// DisableFlattenedResults prevents results being flattened.
// If this field is false, results from nested and repeated fields are flattened.
// DisableFlattenedResults implies AllowLargeResults
// For more information, see https://cloud.google.com/bigquery/docs/data#nested
DisableFlattenedResults bool
// AllowLargeResults allows the query to produce arbitrarily large result tables.
// The destination must be a table.
// When using this option, queries will take longer to execute, even if the result set is small.
// For additional limitations, see https://cloud.google.com/bigquery/querying-data#largequeryresults
AllowLargeResults bool
// Priority specifies the priority with which to schedule the query.
// The default priority is InteractivePriority.
// For more information, see https://cloud.google.com/bigquery/querying-data#batchqueries
Priority QueryPriority
// MaxBillingTier sets the maximum billing tier for a Query.
// Queries that have resource usage beyond this tier will fail (without
// incurring a charge). If this field is zero, the project default will be used.
MaxBillingTier int
// MaxBytesBilled limits the number of bytes billed for
// this job. Queries that would exceed this limit will fail (without incurring
// a charge).
// If this field is less than 1, the project default will be
// used.
MaxBytesBilled int64
// UseStandardSQL causes the query to use standard SQL. The default.
UseStandardSQL bool
// UseLegacySQL causes the query to use legacy SQL.
UseLegacySQL bool
// Parameters is a list of query parameters. The presence of parameters
// implies the use of standard SQL.
// If the query uses positional syntax ("?"), then no parameter may have a name.
// If the query uses named syntax ("@p"), then all parameters must have names.
// It is illegal to mix positional and named syntax.
Parameters []QueryParameter
}
// QueryPriority specifies a priority with which a query is to be executed.
type QueryPriority string
const (
BatchPriority QueryPriority = "BATCH"
InteractivePriority QueryPriority = "INTERACTIVE"
)
// A Query queries data from a BigQuery table. Use Client.Query to create a Query.
type Query struct {
client *Client
QueryConfig
}
// Query creates a query with string q.
// The returned Query may optionally be further configured before its Run method is called.
func (c *Client) Query(q string) *Query {
return &Query{
client: c,
QueryConfig: QueryConfig{Q: q},
}
}
// Run initiates a query job.
func (q *Query) Run(ctx context.Context) (*Job, error) {
job := &bq.Job{
JobReference: createJobRef(q.JobID, q.AddJobIDSuffix, q.client.projectID),
Configuration: &bq.JobConfiguration{
Query: &bq.JobConfigurationQuery{},
},
}
if err := q.QueryConfig.populateJobQueryConfig(job.Configuration.Query); err != nil {
return nil, err
}
j, err := q.client.insertJob(ctx, &insertJobConf{job: job})
if err != nil {
return nil, err
}
j.isQuery = true
return j, nil
}
func (q *QueryConfig) populateJobQueryConfig(conf *bq.JobConfigurationQuery) error {
conf.Query = q.Q
if len(q.TableDefinitions) > 0 {
conf.TableDefinitions = make(map[string]bq.ExternalDataConfiguration)
}
for name, data := range q.TableDefinitions {
conf.TableDefinitions[name] = data.externalDataConfig()
}
if q.DefaultProjectID != "" || q.DefaultDatasetID != "" {
conf.DefaultDataset = &bq.DatasetReference{
DatasetId: q.DefaultDatasetID,
ProjectId: q.DefaultProjectID,
}
}
if tier := int64(q.MaxBillingTier); tier > 0 {
conf.MaximumBillingTier = &tier
}
conf.CreateDisposition = string(q.CreateDisposition)
conf.WriteDisposition = string(q.WriteDisposition)
conf.AllowLargeResults = q.AllowLargeResults
conf.Priority = string(q.Priority)
f := false
if q.DisableQueryCache {
conf.UseQueryCache = &f
}
if q.DisableFlattenedResults {
conf.FlattenResults = &f
// DisableFlattenResults implies AllowLargeResults.
conf.AllowLargeResults = true
}
if q.MaxBytesBilled >= 1 {
conf.MaximumBytesBilled = q.MaxBytesBilled
}
if q.UseStandardSQL && q.UseLegacySQL {
return errors.New("bigquery: cannot provide both UseStandardSQL and UseLegacySQL")
}
if len(q.Parameters) > 0 && q.UseLegacySQL {
return errors.New("bigquery: cannot provide both Parameters (implying standard SQL) and UseLegacySQL")
}
if q.UseLegacySQL {
conf.UseLegacySql = true
} else {
conf.UseLegacySql = false
conf.ForceSendFields = append(conf.ForceSendFields, "UseLegacySql")
}
if q.Dst != nil && !q.Dst.implicitTable() {
conf.DestinationTable = q.Dst.tableRefProto()
}
for _, p := range q.Parameters {
qp, err := p.toRaw()
if err != nil {
return err
}
conf.QueryParameters = append(conf.QueryParameters, qp)
}
return nil
}
// Read submits a query for execution and returns the results via a RowIterator.
// It is a shorthand for Query.Run followed by Job.Read.
func (q *Query) Read(ctx context.Context) (*RowIterator, error) {
job, err := q.Run(ctx)
if err != nil {
return nil, err
}
return job.Read(ctx)
}