google-cloud-bigquery 1.21.2 → 1.26.0

data/lib/google/cloud/bigquery/external.rb

@@ -52,6 +52,24 @@ module Google
       #   # Retrieve the next page of results
       #   data = data.next if data.next?
       #
+      # @example Hive partitioning options:
+      #   require "google/cloud/bigquery"
+      #
+      #   bigquery = Google::Cloud::Bigquery.new
+      #
+      #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+      #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+      #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+      #     ext.hive_partitioning_mode = :auto
+      #     ext.hive_partitioning_require_partition_filter = true
+      #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+      #   end
+      #
+      #   external_data.hive_partitioning? #=> true
+      #   external_data.hive_partitioning_mode #=> "AUTO"
+      #   external_data.hive_partitioning_require_partition_filter? #=> true
+      #   external_data.hive_partitioning_source_uri_prefix #=> source_uri_prefix
+      #
       module External
         ##
         # @private New External from URLs and format
@@ -79,7 +97,8 @@ module Google
         # @private Determine source_format from inputs
         def self.source_format_for urls, format
           val = {
-            "csv" => "CSV",          "avro" => "AVRO",
+            "csv"                    => "CSV",
+            "avro"                   => "AVRO",
             "json"                   => "NEWLINE_DELIMITED_JSON",
             "newline_delimited_json" => "NEWLINE_DELIMITED_JSON",
             "sheets"                 => "GOOGLE_SHEETS",
@@ -87,7 +106,9 @@ module Google
             "datastore"              => "DATASTORE_BACKUP",
             "backup"                 => "DATASTORE_BACKUP",
             "datastore_backup"       => "DATASTORE_BACKUP",
-            "bigtable"               => "BIGTABLE"
+            "bigtable"               => "BIGTABLE",
+            "orc"                    => "ORC",
+            "parquet"                => "PARQUET"
           }[format.to_s.downcase]
           return val unless val.nil?
           Array(urls).each do |url|
@@ -110,7 +131,7 @@ module Google
           when "GOOGLE_SHEETS"          then External::SheetsSource
           when "BIGTABLE"               then External::BigtableSource
           else
-            # AVRO and DATASTORE_BACKUP
+            # AVRO, DATASTORE_BACKUP, PARQUET
             External::DataSource
           end
         end
@@ -148,6 +169,24 @@ module Google
         #   # Retrieve the next page of results
         #   data = data.next if data.next?
         #
+        # @example Hive partitioning options:
+        #   require "google/cloud/bigquery"
+        #
+        #   bigquery = Google::Cloud::Bigquery.new
+        #
+        #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+        #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+        #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+        #     ext.hive_partitioning_mode = :auto
+        #     ext.hive_partitioning_require_partition_filter = true
+        #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+        #   end
+        #
+        #   external_data.hive_partitioning? #=> true
+        #   external_data.hive_partitioning_mode #=> "AUTO"
+        #   external_data.hive_partitioning_require_partition_filter? #=> true
+        #   external_data.hive_partitioning_source_uri_prefix #=> source_uri_prefix
+        #
         class DataSource
           ##
           # @private The Google API Client object.
@@ -302,6 +341,52 @@ module Google
             @gapi.source_format == "BIGTABLE"
           end
 
+          ##
+          # Whether the data format is "ORC".
+          #
+          # @return [Boolean]
+          #
+          # @example
+          #   require "google/cloud/bigquery"
+          #
+          #   bigquery = Google::Cloud::Bigquery.new
+          #
+          #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+          #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+          #   external_data = bigquery.external gcs_uri, format: :orc do |ext|
+          #     ext.hive_partitioning_mode = :auto
+          #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+          #   end
+          #   external_data.format #=> "ORC"
+          #   external_data.orc? #=> true
+          #
+          def orc?
+            @gapi.source_format == "ORC"
+          end
+
+          ##
+          # Whether the data format is "PARQUET".
+          #
+          # @return [Boolean]
+          #
+          # @example
+          #   require "google/cloud/bigquery"
+          #
+          #   bigquery = Google::Cloud::Bigquery.new
+          #
+          #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+          #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+          #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+          #     ext.hive_partitioning_mode = :auto
+          #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+          #   end
+          #   external_data.format #=> "PARQUET"
+          #   external_data.parquet? #=> true
+          #
+          def parquet?
+            @gapi.source_format == "PARQUET"
+          end
+
           ##
           # The fully-qualified URIs that point to your data in Google Cloud.
           # For Google Cloud Storage URIs: Each URI can contain one '*' wildcard
@@ -536,6 +621,246 @@ module Google
             @gapi.max_bad_records = new_max_bad_records
           end
 
+          ###
+          # Checks if hive partitioning options are set.
+          #
+          # Not all storage formats support hive partitioning. Requesting hive partitioning on an unsupported format
+          # will lead to an error. Currently supported types include: `avro`, `csv`, `json`, `orc` and `parquet`.
+          # If your data is stored in ORC or Parquet on Cloud Storage, see [Querying columnar formats on Cloud
+          # Storage](https://cloud.google.com/bigquery/pricing#columnar_formats_pricing).
+          #
+          # @return [Boolean] `true` when hive partitioning options are set, or `false` otherwise.
+          #
+          # @example
+          #   require "google/cloud/bigquery"
+          #
+          #   bigquery = Google::Cloud::Bigquery.new
+          #
+          #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+          #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+          #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+          #     ext.hive_partitioning_mode = :auto
+          #     ext.hive_partitioning_require_partition_filter = true
+          #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+          #   end
+          #
+          #   external_data.hive_partitioning? #=> true
+          #   external_data.hive_partitioning_mode #=> "AUTO"
+          #   external_data.hive_partitioning_require_partition_filter? #=> true
+          #   external_data.hive_partitioning_source_uri_prefix #=> source_uri_prefix
+          #
+          def hive_partitioning?
+            !@gapi.hive_partitioning_options.nil?
+          end
+
+          ###
+          # The mode of hive partitioning to use when reading data. The following modes are supported:
+          #
+          #   1. `AUTO`: automatically infer partition key name(s) and type(s).
+          #   2. `STRINGS`: automatically infer partition key name(s). All types are interpreted as strings.
+          #   3. `CUSTOM`: partition key schema is encoded in the source URI prefix.
+          #
+          # @return [String, nil] The mode of hive partitioning, or `nil` if not set.
+          #
+          # @example
+          #   require "google/cloud/bigquery"
+          #
+          #   bigquery = Google::Cloud::Bigquery.new
+          #
+          #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+          #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+          #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+          #     ext.hive_partitioning_mode = :auto
+          #     ext.hive_partitioning_require_partition_filter = true
+          #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+          #   end
+          #
+          #   external_data.hive_partitioning? #=> true
+          #   external_data.hive_partitioning_mode #=> "AUTO"
+          #   external_data.hive_partitioning_require_partition_filter? #=> true
+          #   external_data.hive_partitioning_source_uri_prefix #=> source_uri_prefix
+          #
+          def hive_partitioning_mode
+            @gapi.hive_partitioning_options.mode if hive_partitioning?
+          end
+
+          ##
+          # Sets the mode of hive partitioning to use when reading data. The following modes are supported:
+          #
+          #   1. `auto`: automatically infer partition key name(s) and type(s).
+          #   2. `strings`: automatically infer partition key name(s). All types are interpreted as strings.
+          #   3. `custom`: partition key schema is encoded in the source URI prefix.
+          #
+          # Not all storage formats support hive partitioning. Requesting hive partitioning on an unsupported format
+          # will lead to an error. Currently supported types include: `avro`, `csv`, `json`, `orc` and `parquet`.
+          # If your data is stored in ORC or Parquet on Cloud Storage, see [Querying columnar formats on Cloud
+          # Storage](https://cloud.google.com/bigquery/pricing#columnar_formats_pricing).
+          #
+          # See {#format}, {#hive_partitioning_require_partition_filter=} and {#hive_partitioning_source_uri_prefix=}.
+          #
+          # @param [String, Symbol] mode The mode of hive partitioning to use when reading data.
+          #
+          # @example
+          #   require "google/cloud/bigquery"
+          #
+          #   bigquery = Google::Cloud::Bigquery.new
+          #
+          #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+          #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+          #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+          #     ext.hive_partitioning_mode = :auto
+          #     ext.hive_partitioning_require_partition_filter = true
+          #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+          #   end
+          #
+          #   external_data.hive_partitioning? #=> true
+          #   external_data.hive_partitioning_mode #=> "AUTO"
+          #   external_data.hive_partitioning_require_partition_filter? #=> true
+          #   external_data.hive_partitioning_source_uri_prefix #=> source_uri_prefix
+          #
+          def hive_partitioning_mode= mode
+            @gapi.hive_partitioning_options ||= Google::Apis::BigqueryV2::HivePartitioningOptions.new
+            @gapi.hive_partitioning_options.mode = mode.to_s.upcase
+          end
+
+          ###
+          # Whether queries over the table using this external data source require a partition filter that can be used
+          # for partition elimination to be specified. Note that this field should only be true when creating a
+          # permanent external table or querying a temporary external table.
+          #
+          # @return [Boolean] `true` when queries over this table require a partition filter, or `false` otherwise.
+          #
+          # @example
+          #   require "google/cloud/bigquery"
+          #
+          #   bigquery = Google::Cloud::Bigquery.new
+          #
+          #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+          #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+          #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+          #     ext.hive_partitioning_mode = :auto
+          #     ext.hive_partitioning_require_partition_filter = true
+          #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+          #   end
+          #
+          #   external_data.hive_partitioning? #=> true
+          #   external_data.hive_partitioning_mode #=> "AUTO"
+          #   external_data.hive_partitioning_require_partition_filter? #=> true
+          #   external_data.hive_partitioning_source_uri_prefix #=> source_uri_prefix
+          #
+          def hive_partitioning_require_partition_filter?
+            return false unless hive_partitioning?
+            !@gapi.hive_partitioning_options.require_partition_filter.nil?
+          end
+
+          ##
+          # Sets whether queries over the table using this external data source require a partition filter
+          # that can be used for partition elimination to be specified.
+          #
+          # See {#format}, {#hive_partitioning_mode=} and {#hive_partitioning_source_uri_prefix=}.
+          #
+          # @param [Boolean] require_partition_filter `true` if a partition filter must be specified, `false` otherwise.
+          #
+          # @example
+          #   require "google/cloud/bigquery"
+          #
+          #   bigquery = Google::Cloud::Bigquery.new
+          #
+          #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+          #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+          #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+          #     ext.hive_partitioning_mode = :auto
+          #     ext.hive_partitioning_require_partition_filter = true
+          #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+          #   end
+          #
+          #   external_data.hive_partitioning? #=> true
+          #   external_data.hive_partitioning_mode #=> "AUTO"
+          #   external_data.hive_partitioning_require_partition_filter? #=> true
+          #   external_data.hive_partitioning_source_uri_prefix #=> source_uri_prefix
+          #
+          def hive_partitioning_require_partition_filter= require_partition_filter
+            @gapi.hive_partitioning_options ||= Google::Apis::BigqueryV2::HivePartitioningOptions.new
+            @gapi.hive_partitioning_options.require_partition_filter = require_partition_filter
+          end
+
+          ###
+          # The common prefix for all source uris when hive partition detection is requested. The prefix must end
+          # immediately before the partition key encoding begins. For example, consider files following this data
+          # layout:
+          #
+          # ```
+          # gs://bucket/path_to_table/dt=2019-01-01/country=BR/id=7/file.avro
+          # gs://bucket/path_to_table/dt=2018-12-31/country=CA/id=3/file.avro
+          # ```
+          #
+          # When hive partitioning is requested with either `AUTO` or `STRINGS` mode, the common prefix can be either of
+          # `gs://bucket/path_to_table` or `gs://bucket/path_to_table/` (trailing slash does not matter).
+          #
+          # @return [String, nil] The common prefix for all source uris, or `nil` if not set.
+          #
+          # @example
+          #   require "google/cloud/bigquery"
+          #
+          #   bigquery = Google::Cloud::Bigquery.new
+          #
+          #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+          #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+          #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+          #     ext.hive_partitioning_mode = :auto
+          #     ext.hive_partitioning_require_partition_filter = true
+          #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+          #   end
+          #
+          #   external_data.hive_partitioning? #=> true
+          #   external_data.hive_partitioning_mode #=> "AUTO"
+          #   external_data.hive_partitioning_require_partition_filter? #=> true
+          #   external_data.hive_partitioning_source_uri_prefix #=> source_uri_prefix
+          #
+          def hive_partitioning_source_uri_prefix
+            @gapi.hive_partitioning_options.source_uri_prefix if hive_partitioning?
+          end
+
+          ##
+          # Sets the common prefix for all source uris when hive partition detection is requested. The prefix must end
+          # immediately before the partition key encoding begins. For example, consider files following this data
+          # layout:
+          #
+          # ```
+          # gs://bucket/path_to_table/dt=2019-01-01/country=BR/id=7/file.avro
+          # gs://bucket/path_to_table/dt=2018-12-31/country=CA/id=3/file.avro
+          # ```
+          #
+          # When hive partitioning is requested with either `AUTO` or `STRINGS` mode, the common prefix can be either of
+          # `gs://bucket/path_to_table` or `gs://bucket/path_to_table/` (trailing slash does not matter).
+          #
+          # See {#format}, {#hive_partitioning_mode=} and {#hive_partitioning_require_partition_filter=}.
+          #
+          # @param [String] source_uri_prefix The common prefix for all source uris.
+          #
+          # @example
+          #   require "google/cloud/bigquery"
+          #
+          #   bigquery = Google::Cloud::Bigquery.new
+          #
+          #   gcs_uri = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/*"
+          #   source_uri_prefix = "gs://cloud-samples-data/bigquery/hive-partitioning-samples/autolayout/"
+          #   external_data = bigquery.external gcs_uri, format: :parquet do |ext|
+          #     ext.hive_partitioning_mode = :auto
+          #     ext.hive_partitioning_require_partition_filter = true
+          #     ext.hive_partitioning_source_uri_prefix = source_uri_prefix
+          #   end
+          #
+          #   external_data.hive_partitioning? #=> true
+          #   external_data.hive_partitioning_mode #=> "AUTO"
+          #   external_data.hive_partitioning_require_partition_filter? #=> true
+          #   external_data.hive_partitioning_source_uri_prefix #=> source_uri_prefix
+          #
+          def hive_partitioning_source_uri_prefix= source_uri_prefix
+            @gapi.hive_partitioning_options ||= Google::Apis::BigqueryV2::HivePartitioningOptions.new
+            @gapi.hive_partitioning_options.source_uri_prefix = source_uri_prefix
+          end
+
           ##
           # @private Google API Client object.
           def to_gapi

data/lib/google/cloud/bigquery/extract_job.rb

@@ -20,15 +20,17 @@ module Google
       # # ExtractJob
       #
       # A {Job} subclass representing an export operation that may be performed
-      # on a {Table}. A ExtractJob instance is created when you call
-      # {Table#extract_job}.
+      # on a {Table} or {Model}. A ExtractJob instance is returned when you call
+      # {Project#extract_job}, {Table#extract_job} or {Model#extract_job}.
       #
       # @see https://cloud.google.com/bigquery/docs/exporting-data
-      #   Exporting Data From BigQuery
+      #   Exporting table data
+      # @see https://cloud.google.com/bigquery-ml/docs/exporting-models
+      #   Exporting models
       # @see https://cloud.google.com/bigquery/docs/reference/v2/jobs Jobs API
       #   reference
       #
-      # @example
+      # @example Export table data
       #   require "google/cloud/bigquery"
       #
       #   bigquery = Google::Cloud::Bigquery.new
@@ -40,6 +42,18 @@ module Google
       #   extract_job.wait_until_done!
       #   extract_job.done? #=> true
       #
+      # @example Export a model
+      #   require "google/cloud/bigquery"
+      #
+      #   bigquery = Google::Cloud::Bigquery.new
+      #   dataset = bigquery.dataset "my_dataset"
+      #   model = dataset.model "my_model"
+      #
+      #   extract_job = model.extract_job "gs://my-bucket/#{model.model_id}"
+      #
+      #   extract_job.wait_until_done!
+      #   extract_job.done? #=> true
+      #
       class ExtractJob < Job
         ##
         # The URI or URIs representing the Google Cloud Storage files to which
@@ -49,71 +63,126 @@ module Google
         end
 
         ##
-        # The table from which the data is exported. This is the table upon
-        # which {Table#extract_job} was called.
+        # The table or model which is exported.
         #
-        # @return [Table] A table instance.
+        # @return [Table, Model, nil] A table or model instance, or `nil`.
         #
         def source
-          table = @gapi.configuration.extract.source_table
-          return nil unless table
-          retrieve_table table.project_id, table.dataset_id, table.table_id
+          if (table = @gapi.configuration.extract.source_table)
+            retrieve_table table.project_id, table.dataset_id, table.table_id
+          elsif (model = @gapi.configuration.extract.source_model)
+            retrieve_model model.project_id, model.dataset_id, model.model_id
+          end
         end
 
         ##
-        # Checks if the export operation compresses the data using gzip. The
-        # default is `false`.
+        # Whether the source of the export job is a table. See {#source}.
         #
-        # @return [Boolean] `true` when `GZIP`, `false` otherwise.
+        # @return [Boolean] `true` when the source is a table, `false`
+        #   otherwise.
         #
-        def compression?
-          val = @gapi.configuration.extract.compression
-          val == "GZIP"
+        def table?
+          !@gapi.configuration.extract.source_table.nil?
         end
 
         ##
-        # Checks if the destination format for the data is [newline-delimited
-        # JSON](http://jsonlines.org/). The default is `false`.
+        # Whether the source of the export job is a model. See {#source}.
         #
-        # @return [Boolean] `true` when `NEWLINE_DELIMITED_JSON`, `false`
+        # @return [Boolean] `true` when the source is a model, `false`
         #   otherwise.
         #
+        def model?
+          !@gapi.configuration.extract.source_model.nil?
+        end
+
+        ##
+        # Checks if the export operation compresses the data using gzip. The
+        # default is `false`. Not applicable when extracting models.
+        #
+        # @return [Boolean] `true` when `GZIP`, `false` if not `GZIP` or not a
+        #   table extraction.
+        def compression?
+          return false unless table?
+          @gapi.configuration.extract.compression == "GZIP"
+        end
+
+        ##
+        # Checks if the destination format for the table data is [newline-delimited
+        # JSON](http://jsonlines.org/). The default is `false`. Not applicable when
+        # extracting models.
+        #
+        # @return [Boolean] `true` when `NEWLINE_DELIMITED_JSON`, `false` if not
+        #   `NEWLINE_DELIMITED_JSON` or not a table extraction.
+        #
         def json?
-          val = @gapi.configuration.extract.destination_format
-          val == "NEWLINE_DELIMITED_JSON"
+          return false unless table?
+          @gapi.configuration.extract.destination_format == "NEWLINE_DELIMITED_JSON"
         end
 
         ##
-        # Checks if the destination format for the data is CSV. Tables with
+        # Checks if the destination format for the table data is CSV. Tables with
         # nested or repeated fields cannot be exported as CSV. The default is
-        # `true`.
+        # `true` for tables. Not applicable when extracting models.
         #
-        # @return [Boolean] `true` when `CSV`, `false` otherwise.
+        # @return [Boolean] `true` when `CSV`, or `false` if not `CSV` or not a
+        #   table extraction.
         #
         def csv?
+          return false unless table?
           val = @gapi.configuration.extract.destination_format
           return true if val.nil?
           val == "CSV"
         end
 
         ##
-        # Checks if the destination format for the data is
-        # [Avro](http://avro.apache.org/). The default is `false`.
+        # Checks if the destination format for the table data is
+        # [Avro](http://avro.apache.org/). The default is `false`. Not applicable
+        # when extracting models.
         #
-        # @return [Boolean] `true` when `AVRO`, `false` otherwise.
+        # @return [Boolean] `true` when `AVRO`, `false` if not `AVRO` or not a
+        #   table extraction.
         #
         def avro?
+          return false unless table?
+          @gapi.configuration.extract.destination_format == "AVRO"
+        end
+
+        ##
+        # Checks if the destination format for the model is TensorFlow SavedModel.
+        # The default is `true` for models. Not applicable when extracting tables.
+        #
+        # @return [Boolean] `true` when `ML_TF_SAVED_MODEL`, `false` if not
+        #   `ML_TF_SAVED_MODEL` or not a model extraction.
+        #
+        def ml_tf_saved_model?
+          return false unless model?
           val = @gapi.configuration.extract.destination_format
-          val == "AVRO"
+          return true if val.nil?
+          val == "ML_TF_SAVED_MODEL"
+        end
+
+        ##
+        # Checks if the destination format for the model is XGBoost. The default
+        # is `false`. Not applicable when extracting tables.
+        #
+        # @return [Boolean] `true` when `ML_XGBOOST_BOOSTER`, `false` if not
+        #   `ML_XGBOOST_BOOSTER` or not a model extraction.
+        #
+        def ml_xgboost_booster?
+          return false unless model?
+          @gapi.configuration.extract.destination_format == "ML_XGBOOST_BOOSTER"
         end
 
         ##
         # The character or symbol the operation uses to delimit fields in the
-        # exported data. The default is a comma (,).
+        # exported data. The default is a comma (,) for tables. Not applicable
+        # when extracting models.
         #
-        # @return [String] A string containing the character, such as `","`.
+        # @return [String, nil] A string containing the character, such as `","`,
+        #   `nil` if not a table extraction.
         #
         def delimiter
+          return unless table?
           val = @gapi.configuration.extract.field_delimiter
           val = "," if val.nil?
           val
@@ -121,12 +190,13 @@ module Google
 
         ##
         # Checks if the exported data contains a header row. The default is
-        # `true`.
+        # `true` for tables. Not applicable when extracting models.
         #
         # @return [Boolean] `true` when the print header configuration is
-        #   present or `nil`, `false` otherwise.
+        #   present or `nil`, `false` if disabled or not a table extraction.
         #
         def print_header?
+          return false unless table?
           val = @gapi.configuration.extract.print_header
           val = true if val.nil?
           val
@@ -159,12 +229,14 @@ module Google
         # whether to enable extracting applicable column types (such as
         # `TIMESTAMP`) to their corresponding AVRO logical types
         # (`timestamp-micros`), instead of only using their raw types
-        # (`avro-long`).
+        # (`avro-long`). Not applicable when extracting models.
         #
         # @return [Boolean] `true` when applicable column types will use their
-        #   corresponding AVRO logical types, `false` otherwise.
+        #   corresponding AVRO logical types, `false` if not enabled or not a
+        #   table extraction.
         #
         def use_avro_logical_types?
+          return false unless table?
           @gapi.configuration.extract.use_avro_logical_types
         end
 
@@ -182,19 +254,24 @@ module Google
           #
           # @return [Google::Cloud::Bigquery::ExtractJob::Updater] A job
           #   configuration object for setting query options.
-          def self.from_options service, table, storage_files, options
+          def self.from_options service, source, storage_files, options
             job_ref = service.job_ref_from options[:job_id], options[:prefix]
             storage_urls = Array(storage_files).map do |url|
               url.respond_to?(:to_gs_url) ? url.to_gs_url : url
             end
             options[:format] ||= Convert.derive_source_format storage_urls.first
+            extract_config = Google::Apis::BigqueryV2::JobConfigurationExtract.new(
+              destination_uris: Array(storage_urls)
+            )
+            if source.is_a? Google::Apis::BigqueryV2::TableReference
+              extract_config.source_table = source
+            elsif source.is_a? Google::Apis::BigqueryV2::ModelReference
+              extract_config.source_model = source
+            end
             job = Google::Apis::BigqueryV2::Job.new(
               job_reference: job_ref,
               configuration: Google::Apis::BigqueryV2::JobConfiguration.new(
-                extract: Google::Apis::BigqueryV2::JobConfigurationExtract.new(
-                  destination_uris: Array(storage_urls),
-                  source_table:     table
-                ),
+                extract: extract_config,
                 dry_run: options[:dryrun]
               )
             )
@@ -253,7 +330,7 @@ module Google
           end
 
           ##
-          # Sets the compression type.
+          # Sets the compression type. Not applicable when extracting models.
           #
           # @param [String] value The compression type to use for exported
           #   files. Possible values include `GZIP` and `NONE`. The default
@@ -265,7 +342,7 @@ module Google
           end
 
           ##
-          # Sets the field delimiter.
+          # Sets the field delimiter. Not applicable when extracting models.
           #
           # @param [String] value Delimiter to use between fields in the
           #   exported data. Default is <code>,</code>.
@@ -276,14 +353,21 @@ module Google
           end
 
           ##
-          # Sets the destination file format. The default value is `csv`.
+          # Sets the destination file format. The default value for
+          # tables is `csv`. Tables with nested or repeated fields cannot be
+          # exported as CSV. The default value for models is `ml_tf_saved_model`.
           #
-          # The following values are supported:
+          # Supported values for tables:
           #
           # * `csv` - CSV
           # * `json` - [Newline-delimited JSON](http://jsonlines.org/)
           # * `avro` - [Avro](http://avro.apache.org/)
           #
+          # Supported values for models:
+          #
+          # * `ml_tf_saved_model` - TensorFlow SavedModel
+          # * `ml_xgboost_booster` - XGBoost Booster
+          #
           # @param [String] new_format The new source format.
           #
           # @!group Attributes
@@ -293,7 +377,8 @@ module Google
           end
 
           ##
-          # Print a header row in the exported file.
+          # Print a header row in the exported file. Not applicable when
+          # extracting models.
           #
           # @param [Boolean] value Whether to print out a header row in the
           #   results. Default is `true`.
@@ -307,12 +392,21 @@ module Google
           # Sets the labels to use for the job.
           #
           # @param [Hash] value A hash of user-provided labels associated with
-          #   the job. You can use these to organize and group your jobs. Label
-          #   keys and values can be no longer than 63 characters, can only
-          #   contain lowercase letters, numeric characters, underscores and
-          #   dashes. International characters are allowed. Label values are
-          #   optional. Label keys must start with a letter and each label in
-          #   the list must have a different key.
+          #   the job. You can use these to organize and group your jobs.
+          #
+          #   The labels applied to a resource must meet the following requirements:
+          #
+          #   * Each resource can have multiple labels, up to a maximum of 64.
+          #   * Each label must be a key-value pair.
+          #   * Keys have a minimum length of 1 character and a maximum length of
+          #     63 characters, and cannot be empty. Values can be empty, and have
+          #     a maximum length of 63 characters.
+          #   * Keys and values can contain only lowercase letters, numeric characters,
+          #     underscores, and dashes. All characters must use UTF-8 encoding, and
+          #     international characters are allowed.
+          #   * The key portion of a label must be unique. However, you can use the
+          #     same key with multiple resources.
+          #   * Keys must start with a lowercase letter or international character.
           #
           # @!group Attributes
           #
@@ -362,6 +456,16 @@ module Google
             @gapi
           end
         end
+
+        protected
+
+        def retrieve_model project_id, dataset_id, model_id
+          ensure_service!
+          gapi = service.get_project_model project_id, dataset_id, model_id
+          Model.from_gapi_json gapi, service
+        rescue Google::Cloud::NotFoundError
+          nil
+        end
       end
     end
   end