Class: Embulk::Output::Bigquery

Inherits:
OutputPlugin
  • Object
show all
Defined in:
lib/embulk/output/bigquery.rb,
lib/embulk/output/bigquery/auth.rb,
lib/embulk/output/bigquery/helper.rb,
lib/embulk/output/bigquery/gcs_client.rb,
lib/embulk/output/bigquery/file_writer.rb,
lib/embulk/output/bigquery/google_client.rb,
lib/embulk/output/bigquery/bigquery_client.rb,
lib/embulk/output/bigquery/value_converter_factory.rb

Defined Under Namespace

Classes: Auth, BackendError, BigqueryClient, Error, FileWriter, GcsClient, GoogleClient, Helper, InternalError, JobTimeoutError, LocalFile, NotFoundError, RateLimitExceeded, ValueConverterFactory

Constant Summary collapse

FILE_WRITER_KEY = 
:embulk_output_bigquery_file_writer

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(task, schema, index) ⇒ Bigquery

instance is created on each task



472
473
474
475
476
477
478
 
# File 'lib/embulk/output/bigquery.rb', line 472

def initialize(task, schema, index)
  super

  if task['with_rehearsal'] and @index == 0
    @rehearsaled = false
  end
end

Class Method Details

.add_file_writer(file_writer) ⇒ Object 



448
449
450
451
452
 
# File 'lib/embulk/output/bigquery.rb', line 448

def self.add_file_writer(file_writer)
  @file_writers_mutex.synchronize do
    @file_writers << file_writer
  end
end

.auto_create(task, bigquery) ⇒ Object 



292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
 
# File 'lib/embulk/output/bigquery.rb', line 292

def self.auto_create(task, bigquery)
  if task['auto_create_dataset']
    bigquery.create_dataset(task['dataset'])
  else
    bigquery.get_dataset(task['dataset']) # raises NotFoundError
  end

  if task['mode'] == 'replace_backup' and task['dataset_old'] != task['dataset']
    if task['auto_create_dataset']
      bigquery.create_dataset(task['dataset_old'], reference: task['dataset'])
    else
      bigquery.get_dataset(task['dataset_old']) # raises NotFoundError
    end
  end

  temp_table_expiration = task['temporary_table_expiration']
  temp_options = {'expiration_time' => temp_table_expiration}

  case task['mode']
  when 'delete_in_advance'
    bigquery.delete_table_or_partition(task['table'])
    bigquery.create_table_if_not_exists(task['table'])
  when 'replace'
    bigquery.create_table_if_not_exists(task['temp_table'], options: temp_options)
    bigquery.create_table_if_not_exists(task['table']) # needs for when task['table'] is a partition
  when 'append'
    bigquery.create_table_if_not_exists(task['temp_table'], options: temp_options)
    bigquery.create_table_if_not_exists(task['table']) # needs for when task['table'] is a partition
  when 'replace_backup'
    bigquery.create_table_if_not_exists(task['temp_table'], options: temp_options)
    bigquery.create_table_if_not_exists(task['table'])

    bigquery.create_table_if_not_exists(task['table_old'], dataset: task['dataset_old']) # needs for when a partition
  else # append_direct
    if task['auto_create_table']
      bigquery.create_table_if_not_exists(task['table'])
    else
      bigquery.get_table(task['table']) # raises NotFoundError
    end
  end
end

.bigquery ⇒ Object 



255
256
257
 
# File 'lib/embulk/output/bigquery.rb', line 255

def self.bigquery
  @bigquery
end

.configure(config, schema, task_count) ⇒ Object 



33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
 
# File 'lib/embulk/output/bigquery.rb', line 33

def self.configure(config, schema, task_count)
  task = {
    'mode'                           => config.param('mode',                           :string,  :default => 'append'),
    'auth_method'                    => config.param('auth_method',                    :string,  :default => 'application_default'),
    'json_keyfile'                   => config.param('json_keyfile',                  LocalFile, :default => nil),
    'project'                        => config.param('project',                        :string,  :default => nil),
    'destination_project'            => config.param('destination_project',            :string,  :default => nil),
    'dataset'                        => config.param('dataset',                        :string),
    'location'                       => config.param('location',                       :string,  :default => nil),
    'table'                          => config.param('table',                          :string),
    'dataset_old'                    => config.param('dataset_old',                    :string,  :default => nil),
    'table_old'                      => config.param('table_old',                      :string,  :default => nil),
    'table_name_old'                 => config.param('table_name_old',                 :string,  :default => nil), # lower version compatibility
    'auto_create_dataset'            => config.param('auto_create_dataset',            :bool,    :default => false),
    'auto_create_table'              => config.param('auto_create_table',              :bool,    :default => true),
    'schema_file'                    => config.param('schema_file',                    :string,  :default => nil),
    'template_table'                 => config.param('template_table',                 :string,  :default => nil),

    'delete_from_local_when_job_end' => config.param('delete_from_local_when_job_end', :bool,    :default => true),
    'job_status_max_polling_time'    => config.param('job_status_max_polling_time',    :integer, :default => 3600),
    'job_status_polling_interval'    => config.param('job_status_polling_interval',    :integer, :default => 10),
    'is_skip_job_result_check'       => config.param('is_skip_job_result_check',       :bool,    :default => false),
    'with_rehearsal'                 => config.param('with_rehearsal',                 :bool,    :default => false),
    'rehearsal_counts'               => config.param('rehearsal_counts',               :integer, :default => 1000),
    'abort_on_error'                 => config.param('abort_on_error',                 :bool,    :default => nil),
    'progress_log_interval'          => config.param('progress_log_interval',          :float,   :default => nil),

    'column_options'                 => config.param('column_options',                 :array,   :default => []),
    'default_timezone'               => config.param('default_timezone',               :string,  :default => ValueConverterFactory::DEFAULT_TIMEZONE),
    'default_timestamp_format'       => config.param('default_timestamp_format',       :string,  :default => ValueConverterFactory::DEFAULT_TIMESTAMP_FORMAT),
    'payload_column'                 => config.param('payload_column',                 :string,  :default => nil),
    'payload_column_index'           => config.param('payload_column_index',           :integer, :default => nil),

    'description'                    => config.param('description',                    :string,  :default => nil),

    'open_timeout_sec'               => config.param('open_timeout_sec',               :integer, :default => nil),
    'timeout_sec'                    => config.param('timeout_sec',                    :integer, :default => nil), # google-api-ruby-client < v0.11.0
    'send_timeout_sec'               => config.param('send_timeout_sec',               :integer, :default => nil), # google-api-ruby-client >= v0.11.0
    'read_timeout_sec'               => config.param('read_timeout_sec',               :integer, :default => nil), # google-api-ruby-client >= v0.11.0
    'retries'                        => config.param('retries',                        :integer, :default => 5),
    'application_name'               => config.param('application_name',               :string,  :default => 'Embulk BigQuery plugin'),
    'sdk_log_level'                  => config.param('sdk_log_level',                  :string,  :default => nil),

    'path_prefix'                    => config.param('path_prefix',                    :string,  :default => nil),
    'sequence_format'                => config.param('sequence_format',                :string,  :default => '.%d.%d'),
    'file_ext'                       => config.param('file_ext',                       :string,  :default => nil),
    'skip_file_generation'           => config.param('skip_file_generation',           :bool,    :default => false),
    'compression'                    => config.param('compression',                    :string,  :default => 'NONE'),

    'gcs_bucket'                     => config.param('gcs_bucket',                     :string,  :default => nil),
    'auto_create_gcs_bucket'         => config.param('auto_create_gcs_bucket',         :bool,    :default => false),

    'source_format'                  => config.param('source_format',                  :string,  :default => 'CSV'),
    'max_bad_records'                => config.param('max_bad_records',                :integer, :default => 0),
    'field_delimiter'                => config.param('field_delimiter',                :string,  :default => ','),
    'encoding'                       => config.param('encoding',                       :string,  :default => 'UTF-8'),
    'ignore_unknown_values'          => config.param('ignore_unknown_values',          :bool,    :default => false),
    'allow_quoted_newlines'          => config.param('allow_quoted_newlines',          :bool,    :default => false),
    'time_partitioning'              => config.param('time_partitioning',              :hash,    :default => nil),
    'clustering'                     => config.param('clustering',                     :hash,    :default => nil), # google-api-ruby-client >= v0.21.0
    'schema_update_options'          => config.param('schema_update_options',          :array,   :default => nil),

    'temporary_table_expiration'     => config.param('temporary_table_expiration',     :integer, :default => nil),

    # for debug
    'skip_load'                      => config.param('skip_load',                      :bool,    :default => false),
    'temp_table'                     => config.param('temp_table',                     :string,  :default => nil),
    'rehearsal_table'                => config.param('rehearsal_table',                :string,  :default => nil),
  }

  now = Time.now

  task['mode'] = task['mode'].downcase
  unless %w[append append_direct replace delete_in_advance replace_backup].include?(task['mode'])
    raise ConfigError.new "`mode` must be one of append, append_direct, replace, delete_in_advance, replace_backup"
  end

  if %w[append replace delete_in_advance replace_backup].include?(task['mode']) and !task['auto_create_table']
    raise ConfigError.new "`mode: #{task['mode']}` requires `auto_create_table: true`"
  end

  if task['mode'] == 'replace_backup'
    task['table_old'] ||= task['table_name_old'] # for lower version compatibility
    if task['dataset_old'].nil? and task['table_old'].nil?
      raise ConfigError.new "`mode: replace_backup` requires either of `dataset_old` or `table_old`"
    end
    task['dataset_old'] ||= task['dataset']
    task['table_old']   ||= task['table']
  end

  if task['table_old']
    task['table_old'] = now.strftime(task['table_old'])
  end
  if task['table']
    task['table'] = now.strftime(task['table'])
  end

  task['auth_method'] = task['auth_method'].downcase
  unless %w[json_key service_account authorized_user compute_engine application_default].include?(task['auth_method'])
    raise ConfigError.new "`auth_method` must be one of service_account (or json_key), authorized_user, compute_engine, application_default"
  end
  if (task['auth_method'] == 'service_account' or task['auth_method'] == 'json_key') and task['json_keyfile'].nil?
    raise ConfigError.new "`json_keyfile` is required for auth_method: service_account (or json_key)"
  end

  if task['json_keyfile']
    begin
      json_key = JSON.parse(task['json_keyfile'])
      task['project'] ||= json_key['project_id']
    rescue => e
      raise ConfigError.new "Parsing 'json_keyfile' failed with error: #{e.class} #{e.message}"
    end
  end
  if task['project'].nil?
    raise ConfigError.new "Required field \"project\" is not set"
  end
  task['destination_project'] ||= task['project']

  if (task['payload_column'] or task['payload_column_index']) and task['auto_create_table']
    if task['schema_file'].nil? and task['template_table'].nil?
      raise ConfigError.new "Cannot guess table schema from Embulk schema with `payload_column` or `payload_column_index`. Either of `schema_file` or `template_table` is required for auto_create_table true"
    end
  end

  if task['payload_column_index']
    if task['payload_column_index'] < 0 || schema.size <= task['payload_column_index']
      raise ConfigError.new "payload_column_index #{task['payload_column_index']} is out of schema size"
    end
  elsif task['payload_column']
    task['payload_column_index'] = schema.find_index {|c| c[:name] == task['payload_column'] }
    if task['payload_column_index'].nil?
      raise ConfigError.new "payload_column #{task['payload_column']} does not exist in schema"
    end
  end

  if task['schema_file']
    unless File.exist?(task['schema_file'])
      raise ConfigError.new "schema_file #{task['schema_file']} is not found"
    end
    begin
      JSON.parse(File.read(task['schema_file']))
    rescue => e
      raise ConfigError.new "Parsing 'schema_file' #{task['schema_file']} failed with error: #{e.class} #{e.message}"
    end
  end

  if task['path_prefix'].nil?
    task['path_prefix'] = Tempfile.create('embulk_output_bigquery_') {|fp| fp.path }
  end

  task['source_format'] = task['source_format'].upcase
  if task['source_format'] == 'JSONL'
    task['source_format'] = 'NEWLINE_DELIMITED_JSON'
  end
  unless %w[CSV NEWLINE_DELIMITED_JSON].include?(task['source_format'])
    raise ConfigError.new "`source_format` must be CSV or NEWLINE_DELIMITED_JSON (JSONL)"
  end

  task['compression'] = task['compression'].upcase
  unless %w[GZIP NONE].include?(task['compression'])
    raise ConfigError.new "`compression` must be GZIP or NONE"
  end

  if task['file_ext'].nil?
    case task['source_format']
    when 'CSV'
      file_ext = '.csv'
    else # newline_delimited_json
      file_ext = '.jsonl'
    end
    case task['compression']
    when 'GZIP'
      file_ext << '.gz'
    end
    task['file_ext'] = file_ext
  end

  unique_name = SecureRandom.uuid.gsub('-', '_')

  if %w[replace replace_backup append].include?(task['mode'])
    task['temp_table'] ||= "LOAD_TEMP_#{unique_name}_#{task['table']}"
  else
    task['temp_table'] = nil
  end

  if task['with_rehearsal']
    task['rehearsal_table'] ||= "LOAD_REHEARSAL_#{unique_name}_#{task['table']}"
  end

  if task['sdk_log_level']
    Google::Apis.logger.level = eval("::Logger::#{task['sdk_log_level'].upcase}")
  end

  if task['abort_on_error'].nil?
    task['abort_on_error'] = (task['max_bad_records'] == 0)
  end

  if task['time_partitioning']
    unless task['time_partitioning']['type']
      raise ConfigError.new "`time_partitioning` must have `type` key"
    end
  elsif Helper.has_partition_decorator?(task['table'])
    task['time_partitioning'] = {'type' => 'DAY'}
  end

  if task['clustering']
    unless task['clustering']['fields']
      raise ConfigError.new "`clustering` must have `fields` key"
    end
  end

  if task['schema_update_options']
    task['schema_update_options'].each do |schema_update_option|
      unless %w[ALLOW_FIELD_ADDITION ALLOW_FIELD_RELAXATION].include?(schema_update_option)
        raise ConfigError.new "`schema_update_options` must contain either of ALLOW_FIELD_ADDITION or ALLOW_FIELD_RELAXATION or both"
      end
    end
  end

  task
end

.converters ⇒ Object 



259
260
261
 
# File 'lib/embulk/output/bigquery.rb', line 259

def self.converters
  @converters
end

.file_writers ⇒ Object 



444
445
446
 
# File 'lib/embulk/output/bigquery.rb', line 444

def self.file_writers
  @file_writers
end

.rehearsal_thread ⇒ Object 



263
264
265
 
# File 'lib/embulk/output/bigquery.rb', line 263

def self.rehearsal_thread
  @rehearsal_thread
end

.rehearsal_thread=(rehearsal_thread) ⇒ Object 



267
268
269
 
# File 'lib/embulk/output/bigquery.rb', line 267

def self.rehearsal_thread=(rehearsal_thread)
  @rehearsal_thread = rehearsal_thread
end

.reset_file_writers ⇒ Object 



440
441
442
 
# File 'lib/embulk/output/bigquery.rb', line 440

def self.reset_file_writers
  @file_writers = Array.new
end

.transaction(config, schema, task_count, &control) ⇒ Object 



334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
 
# File 'lib/embulk/output/bigquery.rb', line 334

def self.transaction(config, schema, task_count, &control)
  task = self.configure(config, schema, task_count)

  @task = task
  @schema = schema
  @bigquery = BigqueryClient.new(task, schema)
  @converters = ValueConverterFactory.create_converters(task, schema)

  self.auto_create(@task, @bigquery)

  begin
    paths = []
    if task['skip_file_generation']
      yield(task) # does nothing, but seems it has to be called
      path_pattern = "#{task['path_prefix']}*#{task['file_ext']}"
      Embulk.logger.info { "embulk-output-bigquery: Skip file generation. Get paths from `#{path_pattern}`" }
      paths = Dir.glob(path_pattern)
    else
      task_reports = yield(task) # generates local files

      ios = file_writers.map(&:io)
      paths = ios.map(&:path)
      ios.each do |io|
        Embulk.logger.debug { "close #{io.path}" }
        io.close rescue nil
      end
    end

    if rehearsal_thread
      rehearsal_thread.join
    end

    if task['skip_load'] # only for debug
      Embulk.logger.info { "embulk-output-bigquery: Skip load" }
    else
      if !paths.empty?
        target_table = task['temp_table'] ? task['temp_table'] : task['table']
        if bucket = task['gcs_bucket']
          gcs = GcsClient.new(task)
          gcs.insert_temporary_bucket(bucket) if task['auto_create_gcs_bucket']
          objects = paths.size.times.map { SecureRandom.uuid.to_s }
          gcs.insert_objects(paths, objects: objects, bucket: bucket)
          object_uris = objects.map {|object| URI.join("gs://#{bucket}", object).to_s }
          responses = bigquery.load_from_gcs(object_uris, target_table)
          objects.each {|object| gcs.delete_object(object, bucket: bucket) }
        else
          responses = bigquery.load_in_parallel(paths, target_table)
        end
      else
        responses = []
      end
      transaction_report = self.transaction_report(task, responses)
      Embulk.logger.info { "embulk-output-bigquery: transaction_report: #{transaction_report.to_json}" }

      if task['abort_on_error'] && !task['is_skip_job_result_check']
        if transaction_report['num_input_rows'] != transaction_report['num_output_rows']
          raise Error, "ABORT: `num_input_rows (#{transaction_report['num_input_rows']})` and " \
            "`num_output_rows (#{transaction_report['num_output_rows']})` does not match"
        end
      end

      if task['mode'] == 'replace_backup'
        begin
          bigquery.get_table_or_partition(task['table'])
          bigquery.copy(task['table'], task['table_old'], task['dataset_old'])
        rescue NotFoundError
        end
      end

      if task['temp_table']
        if task['mode'] == 'append'
          bigquery.copy(task['temp_table'], task['table'], write_disposition: 'WRITE_APPEND')
        else # replace or replace_backup
          bigquery.copy(task['temp_table'], task['table'], write_disposition: 'WRITE_TRUNCATE')
        end
      end
    end
  ensure
    begin
      if task['temp_table'] # append or replace or replace_backup
        bigquery.delete_table(task['temp_table'])
      end
    ensure
      if task['delete_from_local_when_job_end']
        paths.each do |path|
          Embulk.logger.info { "embulk-output-bigquery: delete #{path}" }
          File.unlink(path) rescue nil
        end
      else
        paths.each do |path|
          if File.exist?(path)
            Embulk.logger.info { "embulk-output-bigquery: keep #{path}" }
          end
        end
      end
    end
  end

  # this is for -c next_config option, add some paramters for next execution if wants
  next_config_diff = {}
  return next_config_diff
end

.transaction_report(task, responses) ⇒ Object 



271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
 
# File 'lib/embulk/output/bigquery.rb', line 271

def self.transaction_report(task, responses)
  num_input_rows = file_writers.empty? ? 0 : file_writers.map(&:num_rows).inject(:+)
  return {'num_input_rows' => num_input_rows} if task['is_skip_job_result_check']

  num_response_rows = responses.inject(0) do |sum, response|
    sum + (response ? response.statistics.load.output_rows.to_i : 0)
  end
  if task['temp_table']
    num_output_rows = bigquery.get_table_or_partition(task['temp_table']).num_rows.to_i
  else
    num_output_rows = num_response_rows
  end
  num_rejected_rows = num_input_rows - num_output_rows
  transaction_report = {
    'num_input_rows' => num_input_rows,
    'num_response_rows' => num_response_rows,
    'num_output_rows' => num_output_rows,
    'num_rejected_rows' => num_rejected_rows,
  }
end

Instance Method Details

#abort ⇒ Object 



524
525
 
# File 'lib/embulk/output/bigquery.rb', line 524

def abort
end

#add(page) ⇒ Object

called for each page in each task



485
486
487
488
489
490
491
492
493
494
495
 
# File 'lib/embulk/output/bigquery.rb', line 485

def add(page)
  return if task['skip_file_generation']
  num_rows = file_writer.add(page)

  if task['with_rehearsal'] and @index == 0 and !@rehearsaled
    if num_rows >= task['rehearsal_counts']
      load_rehearsal
      @rehearsaled = true
    end
  end
end

#close ⇒ Object

called for each page in each task



481
482
 
# File 'lib/embulk/output/bigquery.rb', line 481

def close
end

#commit ⇒ Object

called after processing all pages in each task, returns a task_report



528
529
530
 
# File 'lib/embulk/output/bigquery.rb', line 528

def commit
  {}
end

#file_writer ⇒ Object

Create one FileWriter object for one output thread, that is, share among tasks. Close theses shared objects in transaction. This is mainly to suppress (or control by -X max_threads) number of files, which equals to number of concurrency to load in parallel, when number of input tasks is many

#file_writer must be called at only #add because threads in other methods are different (called from non-output threads). Note also that #add method of the same task instance would be called in different output threads



464
465
466
467
468
469
 
# File 'lib/embulk/output/bigquery.rb', line 464

def file_writer
  return Thread.current[FILE_WRITER_KEY] if Thread.current[FILE_WRITER_KEY]
  file_writer = FileWriter.new(@task, @schema, @index, self.class.converters)
  self.class.add_file_writer(file_writer)
  Thread.current[FILE_WRITER_KEY] = file_writer
end

#finish ⇒ Object 



521
522
 
# File 'lib/embulk/output/bigquery.rb', line 521

def finish
end

#load_rehearsal ⇒ Object 



497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
 
# File 'lib/embulk/output/bigquery.rb', line 497

def load_rehearsal
  bigquery = self.class.bigquery
  Embulk.logger.info { "embulk-output-bigquery: Rehearsal started" }

  io = file_writer.close # need to close once for gzip
  rehearsal_path = "#{io.path}.rehearsal"
  Embulk.logger.debug { "embulk_output_bigquery: cp #{io.path} #{rehearsal_path}" }
  FileUtils.cp(io.path, rehearsal_path)
  file_writer.reopen

  self.class.rehearsal_thread = Thread.new do
    begin
      bigquery.create_table_if_not_exists(task['rehearsal_table'])
      response = bigquery.load(rehearsal_path, task['rehearsal_table'])
      num_output_rows = response ? response.statistics.load.output_rows.to_i : 0
      Embulk.logger.info { "embulk-output-bigquery: Loaded rehearsal #{num_output_rows}" }
    ensure
      Embulk.logger.debug { "embulk_output_bigquery: delete #{rehearsal_path}" }
      File.unlink(rehearsal_path) rescue nil
      bigquery.delete_table(task['rehearsal_table'])
    end
  end
end