private void setupBigQueryTable(
String projectId, String datasetId, String tableId, TableSchema schema) throws IOException {
if (bigQueryClient == null) {
bigQueryClient = Transport.newBigQueryClient(options.as(BigQueryOptions.class)).build();
}
Datasets datasetService = bigQueryClient.datasets();
if (executeNullIfNotFound(datasetService.get(projectId, datasetId)) == null) {
Dataset newDataset =
new Dataset()
.setDatasetReference(
new DatasetReference().setProjectId(projectId).setDatasetId(datasetId));
datasetService.insert(projectId, newDataset).execute();
}
Tables tableService = bigQueryClient.tables();
Table table = executeNullIfNotFound(tableService.get(projectId, datasetId, tableId));
if (table == null) {
Table newTable =
new Table()
.setSchema(schema)
.setTableReference(
new TableReference()
.setProjectId(projectId)
.setDatasetId(datasetId)
.setTableId(tableId));
tableService.insert(projectId, datasetId, newTable).execute();
} else if (!table.getSchema().equals(schema)) {
throw new RuntimeException(
"Table exists and schemas do not match, expecting: "
+ schema.toPrettyString()
+ ", actual: "
+ table.getSchema().toPrettyString());
}
}
/**
* Retrieves or creates the table.
*
* <p>The table is checked to conform to insertion requirements as specified by WriteDisposition
* and CreateDisposition.
*
* <p>If table truncation is requested (WriteDisposition.WRITE_TRUNCATE), then this will re-create
* the table if necessary to ensure it is empty.
*
* <p>If an empty table is required (WriteDisposition.WRITE_EMPTY), then this will fail if the
* table exists and is not empty.
*
* <p>When constructing a table, a {@code TableSchema} must be available. If a schema is provided,
* then it will be used. If no schema is provided, but an existing table is being cleared
* (WRITE_TRUNCATE option above), then the existing schema will be re-used. If no schema is
* available, then an {@code IOException} is thrown.
*/
public Table getOrCreateTable(
WriteDisposition writeDisposition,
CreateDisposition createDisposition,
@Nullable TableSchema schema)
throws IOException {
// Check if table already exists.
Bigquery.Tables.Get get =
client.tables().get(ref.getProjectId(), ref.getDatasetId(), ref.getTableId());
Table table = null;
try {
table = get.execute();
} catch (IOException e) {
ApiErrorExtractor errorExtractor = new ApiErrorExtractor();
if (!errorExtractor.itemNotFound(e)
|| createDisposition != CreateDisposition.CREATE_IF_NEEDED) {
// Rethrow.
throw e;
}
}
// If we want an empty table, and it isn't, then delete it first.
if (table != null) {
if (writeDisposition == WriteDisposition.WRITE_APPEND) {
return table;
}
boolean empty = isEmpty();
if (empty) {
if (writeDisposition == WriteDisposition.WRITE_TRUNCATE) {
LOG.info("Empty table found, not removing {}", BigQueryIO.toTableSpec(ref));
}
return table;
} else if (writeDisposition == WriteDisposition.WRITE_EMPTY) {
throw new IOException("WriteDisposition is WRITE_EMPTY, " + "but table is not empty");
}
// Reuse the existing schema if none was provided.
if (schema == null) {
schema = table.getSchema();
}
// Delete table and fall through to re-creating it below.
LOG.info("Deleting table {}", BigQueryIO.toTableSpec(ref));
Bigquery.Tables.Delete delete =
client.tables().delete(ref.getProjectId(), ref.getDatasetId(), ref.getTableId());
delete.execute();
}
if (schema == null) {
throw new IllegalArgumentException("Table schema required for new table.");
}
// Create the table.
return tryCreateTable(schema);
}
/**
* Tries to create the BigQuery table. If a table with the same name already exists in the
* dataset, the table creation fails, and the function returns null. In such a case, the existing
* table doesn't necessarily have the same schema as specified by the parameter.
*
* @param schema Schema of the new BigQuery table.
* @return The newly created BigQuery table information, or null if the table with the same name
* already exists.
* @throws IOException if other error than already existing table occurs.
*/
@Nullable
public Table tryCreateTable(TableSchema schema) throws IOException {
LOG.info("Trying to create BigQuery table: {}", BigQueryIO.toTableSpec(ref));
Table content = new Table();
content.setTableReference(ref);
content.setSchema(schema);
try {
return client.tables().insert(ref.getProjectId(), ref.getDatasetId(), content).execute();
} catch (IOException e) {
if (new ApiErrorExtractor().itemAlreadyExists(e)) {
LOG.info("The BigQuery table already exists.");
return null;
}
throw e;
}
}
