Class: Fluent::Plugin::BigQueryOutput

Inherits:
Output
  • Object
show all
Defined in:
lib/fluent/plugin/out_bigquery.rb

Defined Under Namespace

Modules: InsertImplementation, LoadImplementation

Constant Summary collapse

REGEXP_MAX_NUM =

TODO: record field stream inserts doesn't works well? At table creation, table type json + field type record -> field type validation fails At streaming inserts, schema cannot be specified config_param :field_record, :string, defualt: nil config_param :optional_data_field, :string, default: nil

10

Instance Method Summary collapse

Constructor Details

#initialize ⇒ BigQueryOutput

Table types https://developers.google.com/bigquery/docs/tables

type - The following data types are supported; see Data Formats for details on each data type: STRING INTEGER FLOAT BOOLEAN RECORD A JSON object, used when importing nested records. This type is only available when using JSON source files.

mode - Whether a field can be null. The following values are supported: NULLABLE - The cell can be null. REQUIRED - The cell cannot be null. REPEATED - Zero or more repeated simple or nested subfields. This mode is only supported when using JSON source files.



175
176
177
178
179
180
181
182
183
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 175

def initialize
  super
  require 'multi_json'
  require 'google/apis/bigquery_v2'
  require 'googleauth'

  # MEMO: signet-0.6.1 depend on Farady.default_connection
  Faraday.default_connection.options.timeout = 60
end

Instance Method Details

#configure(conf) ⇒ Object 



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
254
255
256
257
258
259
260
261
262
263
264
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 185

def configure(conf)
  if conf["method"] == "load"
    configure_for_load(conf)
  else
    configure_for_insert(conf)
  end
  super

  case @method
  when :insert
    extend(InsertImplementation)
  when :load
    raise Fluent::ConfigError, "'template_suffix' is for only `insert` mode, instead use 'fetch_schema_table' and formatted table name" if @template_suffix
    extend(LoadImplementation)
  end

  case @auth_method
  when :private_key
    unless @email && @private_key_path
      raise Fluent::ConfigError, "'email' and 'private_key_path' must be specified if auth_method == 'private_key'"
    end
  when :compute_engine
    # Do nothing
  when :json_key
    unless @json_key
      raise Fluent::ConfigError, "'json_key' must be specified if auth_method == 'json_key'"
    end
  when :application_default
    # Do nothing
  else
    raise Fluent::ConfigError, "unrecognized 'auth_method': #{@auth_method}"
  end

  unless @table.nil? ^ @tables.nil?
    raise Fluent::ConfigError, "'table' or 'tables' must be specified, and both are invalid"
  end

  @tablelist = @tables ? @tables : [@table]

  legacy_schema_config_deprecation
  @table_schema = Fluent::BigQuery::RecordSchema.new('record')
  if @schema
    @table_schema.load_schema(@schema)
  end
  if @schema_path
    @table_schema.load_schema(MultiJson.load(File.read(@schema_path)))
  end

  types = %i(string integer float boolean timestamp)
  types.each do |type|
    fields = instance_variable_get("@field_#{type}")
    next unless fields
    fields.each do |field|
      @table_schema.register_field field, type
    end
  end

  @regexps = {}
  (1..REGEXP_MAX_NUM).each do |i|
    next unless conf["replace_record_key_regexp#{i}"]
    regexp, replacement = conf["replace_record_key_regexp#{i}"].split(/ /, 2)
    raise ConfigError, "replace_record_key_regexp#{i} does not contain 2 parameters" unless replacement
    raise ConfigError, "replace_record_key_regexp#{i} contains a duplicated key, #{regexp}" if @regexps[regexp]
    @regexps[regexp] = replacement
  end

  if @insert_id_field
    insert_id_keys = @insert_id_field.split('.')
    @get_insert_id = ->(record) {
      insert_id_keys.inject(record) {|h, k| h[k] }
    }
  else
    @get_insert_id = nil
  end

  placeholder_params = "project=#{@project}/dataset=#{@dataset}/table=#{@tablelist.join(",")}/fetch_schema_table=#{@fetch_schema_table}/template_suffix=#{@template_suffix}"
  placeholder_validate!(:bigquery, placeholder_params)

  warn "[DEPRECATION] `convert_hash_to_json` param is deprecated. If Hash value is inserted string field, plugin convert it to json automatically." if @convert_hash_to_json
end

#configure_for_insert(conf) ⇒ Object

default for insert

Raises:

  • (ConfigError) 


23
24
25
26
27
28
29
30
31
32
33
34
35
36
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 23

def configure_for_insert(conf)
  raise ConfigError unless conf["method"].nil? || conf["method"] == "insert"

  buffer_config = conf.elements("buffer")[0]
  return unless buffer_config
  buffer_config["@type"]                       = "memory"      unless buffer_config["@type"]
  buffer_config["flush_mode"]                  = :interval     unless buffer_config["flush_mode"]
  buffer_config["flush_interval"]              = 0.25          unless buffer_config["flush_interval"]
  buffer_config["flush_thread_interval"]       = 0.05          unless buffer_config["flush_thread_interval"]
  buffer_config["flush_thread_burst_interval"] = 0.05          unless buffer_config["flush_thread_burst_interval"]
  buffer_config["chunk_limit_size"]            = 1 * 1024 ** 2 unless buffer_config["chunk_limit_size"] # 1MB
  buffer_config["total_limit_size"]            = 1 * 1024 ** 3 unless buffer_config["total_limit_size"] # 1GB
  buffer_config["chunk_records_limit"]         = 500           unless buffer_config["chunk_records_limit"]
end

#configure_for_load(conf) ⇒ Object

default for loads

Raises:

  • (ConfigError) 


39
40
41
42
43
44
45
46
47
48
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 39

def configure_for_load(conf)
  raise ConfigError unless conf["method"] == "load"

  buffer_config = conf.elements("buffer")[0]
  return unless buffer_config
  buffer_config["@type"]                       = "file"         unless buffer_config["@type"]
  buffer_config["flush_mode"]                  = :interval      unless buffer_config["flush_mode"]
  buffer_config["chunk_limit_size"]            = 1 * 1024 ** 3  unless buffer_config["chunk_limit_size"] # 1GB
  buffer_config["total_limit_size"]            = 32 * 1024 ** 3 unless buffer_config["total_limit_size"] # 32GB
end

#convert_hash_to_json(record) ⇒ Object 



306
307
308
309
310
311
312
313
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 306

def convert_hash_to_json(record)
  record.each do |key, value|
    if value.class == Hash
      record[key] = MultiJson.dump(value)
    end
  end
  record
end

#fetch_schema(metadata) ⇒ Object 



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
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 366

def fetch_schema(metadata)
  table_id = nil
  project = extract_placeholders(@project, metadata)
  dataset = extract_placeholders(@dataset, metadata)
  table_id = fetch_schema_target_table(metadata)

  if Fluent::Engine.now - @last_fetch_schema_time["#{project}.#{dataset}.#{table_id}"] > @schema_cache_expire
    schema = writer.fetch_schema(project, dataset, table_id)

    if schema
      table_schema = Fluent::BigQuery::RecordSchema.new("record")
      table_schema.load_schema(schema)
      @fetched_schemas["#{project}.#{dataset}.#{table_id}"] = table_schema
    else
      if @fetched_schemas["#{project}.#{dataset}.#{table_id}"].empty?
        raise "failed to fetch schema from bigquery"
      else
        log.warn "#{table_id} uses previous schema"
      end
    end

    @last_fetch_schema_time["#{project}.#{dataset}.#{table_id}"] = Fluent::Engine.now
  end

  @fetched_schemas["#{project}.#{dataset}.#{table_id}"]
end

#fetch_schema_target_table(metadata) ⇒ Object 



393
394
395
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 393

def fetch_schema_target_table(metadata)
  extract_placeholders(@fetch_schema_table || @tablelist[0], metadata)
end

#format(tag, time, record) ⇒ Object 



315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 315

def format(tag, time, record)
  if @replace_record_key
    record = replace_record_key(record)
  end

  if @convert_hash_to_json
    record = convert_hash_to_json(record)
  end

  record = inject_values_to_record(tag, time, record)

  begin
    meta = metadata(tag, time, record)
    schema =
      if @fetch_schema
        fetch_schema(meta)
      else
        @table_schema
      end
  ensure
    @buffer.metadata_list.delete(meta)
  end

  begin
    buf = String.new
    row = schema.format(record)
    unless row.empty?
      buf << MultiJson.dump(row) + "\n"
    end
    buf
  rescue
    log.error("format error", record: record, schema: schema)
    raise
  end
end

#get_schema(project, dataset, metadata) ⇒ Object 



397
398
399
400
401
402
403
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 397

def get_schema(project, dataset, metadata)
  if @fetch_schema
    @fetched_schemas["#{project}.#{dataset}.#{fetch_schema_target_table(metadata)}"] || fetch_schema(metadata)
  else
    @table_schema
  end
end

#legacy_schema_config_deprecation ⇒ Object 



360
361
362
363
364
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 360

def legacy_schema_config_deprecation
  if [@field_string, @field_integer, @field_float, @field_boolean, @field_timestamp].any?
    warn "[DEPRECATION] `field_*` style schema config is deprecated. Instead of it, use `schema` config params that is array of json style."
  end
end

#replace_record_key(record) ⇒ Object 



293
294
295
296
297
298
299
300
301
302
303
304
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 293

def replace_record_key(record)
  new_record = {}
  record.each do |key, _|
    new_key = key
    @regexps.each do |regexp, replacement|
      new_key = new_key.gsub(/#{regexp}/, replacement)
    end
    new_key = new_key.gsub(/\W/, '')
    new_record.store(new_key, record[key])
  end
  new_record
end

#request_timeout_sec ⇒ Object

Timeout request_timeout_sec Bigquery API response timeout request_open_timeout_sec Bigquery API connection, and request timeout



153
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 153

config_param :request_timeout_sec, :time, default: nil

#start ⇒ Object 



266
267
268
269
270
271
272
273
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 266

def start
  super

  @tables_queue = @tablelist.shuffle
  @tables_mutex = Mutex.new
  @fetched_schemas = {}
  @last_fetch_schema_time = Hash.new(0)
end

#time_partitioning_type ⇒ Object

Partitioning



157
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 157

config_param :time_partitioning_type, :enum, list: [:day], default: nil

#write(chunk) ⇒ Object 



351
352
353
354
355
356
357
358
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 351

def write(chunk)
  table_id_format = @tables_mutex.synchronize do
    t = @tables_queue.shift
    @tables_queue.push t
    t
  end
  _write(chunk, table_id_format)
end

#writer ⇒ Object 



275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
 
# File 'lib/fluent/plugin/out_bigquery.rb', line 275

def writer
  @writer ||= Fluent::BigQuery::Writer.new(@log, @auth_method, {
    private_key_path: @private_key_path, private_key_passphrase: @private_key_passphrase,
    email: @email,
    json_key: @json_key,
    skip_invalid_rows: @skip_invalid_rows,
    ignore_unknown_values: @ignore_unknown_values,
    max_bad_records: @max_bad_records,
    allow_retry_insert_errors: @allow_retry_insert_errors,
    prevent_duplicate_load: @prevent_duplicate_load,
    auto_create_table: @auto_create_table,
    time_partitioning_type: @time_partitioning_type,
    time_partitioning_expiration: @time_partitioning_expiration,
    timeout_sec: @request_timeout_sec,
    open_timeout_sec: @request_open_timeout_sec,
  })
end