Class: Embulk::Output::Bigquery::BigqueryClient

Inherits:
Object
  • Object
show all
Defined in:
lib/embulk/output/bigquery/bigquery_client.rb

Instance Method Summary collapse

Constructor Details

#initialize(task, schema, fields = nil) ⇒ BigqueryClient

Returns a new instance of BigqueryClient.



15
16
17
18
19
20
21
22
23
24
25
26
27
28
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 15

def initialize(task, schema, fields = nil)
  @task = task
  @schema = schema

  @auth_method = task['auth_method']
  @private_key_path = task['p12_keyfile']
  @private_key_passphrase = 'notasecret'
  @json_key = task['json_keyfile']

  @project = task['project']
  @dataset = task['dataset']

  reset_fields(fields) if fields
end

Instance Method Details

#client ⇒ Object 



30
31
32
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
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 30

def client
  return @cached_client if @cached_client && @cached_client_expiration > Time.now

  client = Google::Apis::BigqueryV2::BigqueryService.new
  client.client_options.application_name = @task['application_name']
  client.request_options.retries = @task['retries']
  client.request_options.timeout_sec = @task['timeout_sec']
  client.request_options.open_timeout_sec = @task['open_timeout_sec']
  Embulk.logger.debug { "embulk-output-bigquery: client_options: #{client.client_options.to_h}" }
  Embulk.logger.debug { "embulk-output-bigquery: request_options: #{client.request_options.to_h}" }

  scope = "https://www.googleapis.com/auth/bigquery"

  case @auth_method
  when 'private_key'
    key = Google::APIClient::KeyUtils.load_from_pkcs12(@private_key_path, @private_key_passphrase)
    auth = Signet::OAuth2::Client.new(
      token_credential_uri: "https://accounts.google.com/o/oauth2/token",
      audience: "https://accounts.google.com/o/oauth2/token",
      scope: scope,
      issuer: @email,
      signing_key: key)

  when 'compute_engine'
    auth = Google::Auth::GCECredentials.new

  when 'json_key'
    if File.exist?(@json_key)
      auth = File.open(@json_key) do |f|
        Google::Auth::ServiceAccountCredentials.make_creds(json_key_io: f, scope: scope)
      end
    else
      key = StringIO.new(@json_key)
      auth = Google::Auth::ServiceAccountCredentials.make_creds(json_key_io: key, scope: scope)
    end

  when 'application_default'
    auth = Google::Auth.get_application_default([scope])

  else
    raise ConfigError, "Unknown auth method: #{@auth_method}"
  end

  client.authorization = auth

  @cached_client_expiration = Time.now + 1800
  @cached_client = client
end

#copy(source_table, destination_table, destination_dataset = nil, write_disposition: 'WRITE_TRUNCATE') ⇒ Object 



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
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 194

def copy(source_table, destination_table, destination_dataset = nil, write_disposition: 'WRITE_TRUNCATE')
  begin
    destination_dataset ||= @dataset
    job_id = "embulk_copy_job_#{SecureRandom.uuid}"

    Embulk.logger.info {
      "embulk-output-bigquery: Copy job starting... job_id:[#{job_id}] " \
      "#{@project}:#{@dataset}.#{source_table} => #{@project}:#{destination_dataset}.#{destination_table}"
    }

    body = {
      job_reference: {
        project_id: @project,
        job_id: job_id,
      },
      configuration: {
        copy: {
          create_deposition: 'CREATE_IF_NEEDED',
          write_disposition: write_disposition,
          source_table: {
            project_id: @project,
            dataset_id: @dataset,
            table_id: source_table,
          },
          destination_table: {
            project_id: @project,
            dataset_id: destination_dataset,
            table_id: destination_table,
          },
        }
      }
    }

    opts = {}
    Embulk.logger.debug { "embulk-output-bigquery: insert_job(#{@project}, #{body}, #{opts})" }
    response = client.insert_job(@project, body, opts)
    wait_load('Copy', response)
  rescue Google::Apis::ServerError, Google::Apis::ClientError, Google::Apis::AuthorizationError => e
    response = {status_code: e.status_code, message: e.message, error_class: e.class}
    Embulk.logger.error {
      "embulk-output-bigquery: insert_job(#{@project}, #{body}, #{opts}), response:#{response}"
    }
    raise Error, "failed to copy #{@project}:#{@dataset}.#{source_table} " \
      "to #{@project}:#{destination_dataset}.#{destination_table}, response:#{response}"
  end
end

#create_dataset(dataset = nil, reference: nil) ⇒ Object 



289
290
291
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
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 289

def create_dataset(dataset = nil, reference: nil)
  dataset ||= @dataset
  begin
    Embulk.logger.info { "embulk-output-bigquery: Create dataset... #{@project}:#{dataset}" }
    hint = {}
    if reference
      response = get_dataset(reference)
      hint = { access: response.access }
    end
    body = {
      dataset_reference: {
        project_id: @project,
        dataset_id: dataset,
      },
    }.merge(hint)
    opts = {}
    Embulk.logger.debug { "embulk-output-bigquery: insert_dataset(#{@project}, #{dataset}, #{body}, #{opts})" }
    client.insert_dataset(@project, body, opts)
  rescue Google::Apis::ServerError, Google::Apis::ClientError, Google::Apis::AuthorizationError => e
    if e.status_code == 409 && /Already Exists:/ =~ e.message
      # ignore 'Already Exists' error
      return
    end

    response = {status_code: e.status_code, message: e.message, error_class: e.class}
    Embulk.logger.error {
      "embulk-output-bigquery: insert_dataset(#{@project}, #{body}, #{opts}), response:#{response}"
    }
    raise Error, "failed to create dataset #{@project}:#{dataset}, response:#{response}"
  end
end

#create_table(table) ⇒ Object 



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
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 339

def create_table(table)
  begin
    Embulk.logger.info { "embulk-output-bigquery: Create table... #{@project}:#{@dataset}.#{table}" }
    body = {
      table_reference: {
        table_id: table,
      },
      schema: {
        fields: fields,
      }
    }
    opts = {}
    Embulk.logger.debug { "embulk-output-bigquery: insert_table(#{@project}, #{@dataset}, #{body}, #{opts})" }
    client.insert_table(@project, @dataset, body, opts)
  rescue Google::Apis::ServerError, Google::Apis::ClientError, Google::Apis::AuthorizationError => e
    if e.status_code == 409 && /Already Exists:/ =~ e.message
      # ignore 'Already Exists' error
      return
    end

    response = {status_code: e.status_code, message: e.message, error_class: e.class}
    Embulk.logger.error {
      "embulk-output-bigquery: insert_table(#{@project}, #{@dataset}, #{body}, #{opts}), response:#{response}"
    }
    raise Error, "failed to create table #{@project}:#{@dataset}.#{table}, response:#{response}"
  end
end

#delete_table(table) ⇒ Object 



367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 367

def delete_table(table)
  begin
    Embulk.logger.info { "embulk-output-bigquery: Delete table... #{@project}:#{@dataset}.#{table}" }
    client.delete_table(@project, @dataset, table)
  rescue Google::Apis::ServerError, Google::Apis::ClientError, Google::Apis::AuthorizationError => e
    if e.status_code == 404 && /Not found:/ =~ e.message
      # ignore 'Not Found' error
      return
    end

    response = {status_code: e.status_code, message: e.message, error_class: e.class}
    Embulk.logger.error {
      "embulk-output-bigquery: delete_table(#{@project}, #{@dataset}, #{table}), response:#{response}"
    }
    raise Error, "failed to delete table #{@project}:#{@dataset}.#{table}, response:#{response}"
  end
end

#fields ⇒ Object 



79
80
81
82
83
84
85
86
87
88
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 79

def fields
  return @fields if @fields
  if @task['schema_file']
    @fields = Helper.deep_symbolize_keys(JSON.parse(File.read(@task['schema_file'])))
  elsif @task['template_table']
    @fields = fields_from_table(@task['template_table'])
  else
    @fields = Helper.fields_from_embulk_schema(@task, @schema)
  end
end

#fields_from_table(table) ⇒ Object 



90
91
92
93
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 90

def fields_from_table(table)
  response = get_table(table)
  response.schema.fields.map {|field| field.to_h }
end

#get_dataset(dataset = nil) ⇒ Object 



321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 321

def get_dataset(dataset = nil)
  dataset ||= @dataset
  begin
    Embulk.logger.info { "embulk-output-bigquery: Get dataset... #{@project}:#{@dataset}" }
    client.get_dataset(@project, dataset)
  rescue Google::Apis::ServerError, Google::Apis::ClientError, Google::Apis::AuthorizationError => e
    if e.status_code == 404
      raise NotFoundError, "Dataset #{@project}:#{dataset} is not found"
    end

    response = {status_code: e.status_code, message: e.message, error_class: e.class}
    Embulk.logger.error {
      "embulk-output-bigquery: get_dataset(#{@project}, #{dataset}), response:#{response}"
    }
    raise Error, "failed to get dataset #{@project}:#{dataset}, response:#{response}"
  end
end

#get_table(table) ⇒ Object 



385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 385

def get_table(table)
  begin
    Embulk.logger.info { "embulk-output-bigquery: Get table... #{@project}:#{@dataset}.#{table}" }
    client.get_table(@project, @dataset, table)
  rescue Google::Apis::ServerError, Google::Apis::ClientError, Google::Apis::AuthorizationError => e
    if e.status_code == 404
      raise NotFoundError, "Table #{@project}:#{@dataset}.#{table} is not found"
    end

    response = {status_code: e.status_code, message: e.message, error_class: e.class}
    Embulk.logger.error {
      "embulk-output-bigquery: get_table(#{@project}, #{@dataset}, #{table}), response:#{response}"
    }
    raise Error, "failed to get table #{@project}:#{@dataset}.#{table}, response:#{response}"
  end
end

#load(path, table) ⇒ Object 



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
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 129

def load(path, table)
  begin
    if File.exist?(path)
      # As https://cloud.google.com/bigquery/docs/managing_jobs_datasets_projects#managingjobs says,
      # we should generate job_id in client code, otherwise, retrying would cause duplication
      if @task['prevent_duplicate_insert'] and (@task['mode'] == 'append' or @task['mode'] == 'append_direct')
        job_id = Helper.create_load_job_id(@task, path, fields)
      else
        job_id = "embulk_load_job_#{SecureRandom.uuid}"
      end
      Embulk.logger.info { "embulk-output-bigquery: Load job starting... job_id:[#{job_id}] #{path} => #{@project}:#{@dataset}.#{table}" }
    else
      Embulk.logger.info { "embulk-output-bigquery: Load job starting... #{path} does not exist, skipped" }
      return
    end

    body = {
      job_reference: {
        project_id: @project,
        job_id: job_id,
      },
      configuration: {
        load: {
          destination_table: {
            project_id: @project,
            dataset_id: @dataset,
            table_id: table,
          },
          schema: {
            fields: fields,
          },
          write_disposition: 'WRITE_APPEND',
          source_format:         @task['source_format'],
          max_bad_records:       @task['max_bad_records'],
          field_delimiter:       @task['source_format'] == 'CSV' ? @task['field_delimiter'] : nil,
          encoding:              @task['encoding'],
          ignore_unknown_values: @task['ignore_unknown_values'],
          allow_quoted_newlines: @task['allow_quoted_newlines'],
        }
      }
    }

    opts = {
      upload_source: path,
      content_type: "application/octet-stream",
      # options: {
      #   retries: @task['retries'],
      #   timeout_sec: @task['timeout_sec'],
      #   open_timeout_sec: @task['open_timeout_sec']
      # },
    }
    Embulk.logger.debug { "embulk-output-bigquery: insert_job(#{@project}, #{body}, #{opts})" }
    response = client.insert_job(@project, body, opts)
    unless @task['is_skip_job_result_check']
      response = wait_load('Load', response)
    end
  rescue Google::Apis::ServerError, Google::Apis::ClientError, Google::Apis::AuthorizationError => e
    response = {status_code: e.status_code, message: e.message, error_class: e.class}
    Embulk.logger.error {
      "embulk-output-bigquery: insert_job(#{@project}, #{body}, #{opts}), response:#{response}"
    }
    raise Error, "failed to load #{path} to #{@project}:#{@dataset}.#{table}, response:#{response}"
  end
end

#load_in_parallel(paths, table) ⇒ Object 



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
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 100

def load_in_parallel(paths, table)
  return [] if paths.empty?
  # You may think as, load job is a background job, so sending requests in parallel
  # does not improve performance. However, with actual experiments, this parallel
  # loadings drastically shortened waiting time. It looks one jobs.insert takes about 50 sec.
  # NOTICE: parallel uploadings of files consumes network traffic. With 24 concurrencies
  # with 100MB files consumed about 500Mbps in the experimented environment at a peak.
  #
  # We before had a `max_load_parallels` option, but this was not extensible for map reduce executor
  # So, we dropped it. See https://github.com/embulk/embulk-output-bigquery/pull/35
  responses = []
  threads = []
  Embulk.logger.debug { "embulk-output-bigquery: LOAD IN PARALLEL #{paths}" }
  paths.each_with_index do |path, idx|
    threads << Thread.new do
      # I am not sure whether google-api-ruby-client is thread-safe,
      # so let me create new instances for each thread for safe
      bigquery = self.class.new(@task, @schema, fields)
      response = bigquery.load(path, table)
      [idx, response]
    end
  end
  ThreadsWait.all_waits(*threads) do |th|
    idx, response = th.value # raise errors occurred in threads
    responses[idx] = response
  end
  responses
end

#reset_fields(fields = nil) ⇒ Object 



95
96
97
98
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 95

def reset_fields(fields = nil)
  @fields = fields
  self.fields
end

#wait_load(kind, response) ⇒ Object 



241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
 
# File 'lib/embulk/output/bigquery/bigquery_client.rb', line 241

def wait_load(kind, response)
  started = Time.now

  wait_interval = @task['job_status_polling_interval']
  max_polling_time = @task['job_status_max_polling_time']
  _response = response

  while true
    job_id = _response.job_reference.job_id
    elapsed = Time.now - started
    status = _response.status.state
    if status == "DONE"
      Embulk.logger.info {
        "embulk-output-bigquery: #{kind} job completed... " \
        "job_id:[#{job_id}] elapsed_time:#{elapsed.to_f}sec status:[#{status}]"
      }
      break
    elsif elapsed.to_i > max_polling_time
      message = "embulk-output-bigquery: #{kind} job checking... " \
        "job_id:[#{job_id}] elapsed_time:#{elapsed.to_f}sec status:[TIMEOUT]"
      Embulk.logger.info { message }
      raise JobTimeoutError.new(message)
    else
      Embulk.logger.info {
        "embulk-output-bigquery: #{kind} job checking... " \
        "job_id:[#{job_id}] elapsed_time:#{elapsed.to_f}sec status:[#{status}]"
      }
      sleep wait_interval
      _response = client.get_job(@project, job_id)
    end
  end

  # cf. http://www.rubydoc.info/github/google/google-api-ruby-client/Google/Apis/BigqueryV2/JobStatus#errors-instance_method
  # `errors` returns Array<Google::Apis::BigqueryV2::ErrorProto> if any error exists.
  # Otherwise, this returns nil.
  if _errors = _response.status.errors
    Embulk.logger.error {
      "embulk-output-bigquery: get_job(#{@project}, #{job_id}), " \
      "errors:#{_errors.map(&:to_h)}"
    }
    raise Error, "failed during waiting a #{kind} job, errors:#{_errors.map(&:to_h)}"
  end

  Embulk.logger.info { "embulk-output-bigquery: #{kind} job response... job_id:[#{job_id}] response.statistics:#{_response.statistics.to_h}" }

  _response
end