google-cloud-bigquery 1.31.0 → 1.32.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +17 -0
- data/lib/google/cloud/bigquery/external.rb +9 -2619
- data/lib/google/cloud/bigquery/external/bigtable_source.rb +230 -0
- data/lib/google/cloud/bigquery/external/bigtable_source/column.rb +404 -0
- data/lib/google/cloud/bigquery/external/bigtable_source/column_family.rb +945 -0
- data/lib/google/cloud/bigquery/external/csv_source.rb +481 -0
- data/lib/google/cloud/bigquery/external/data_source.rb +771 -0
- data/lib/google/cloud/bigquery/external/json_source.rb +170 -0
- data/lib/google/cloud/bigquery/external/parquet_source.rb +148 -0
- data/lib/google/cloud/bigquery/external/sheets_source.rb +166 -0
- data/lib/google/cloud/bigquery/load_job.rb +103 -0
- data/lib/google/cloud/bigquery/version.rb +1 -1
- metadata +23 -9
@@ -0,0 +1,945 @@
|
|
1
|
+
# Copyright 2021 Google LLC
|
2
|
+
#
|
3
|
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
4
|
+
# you may not use this file except in compliance with the License.
|
5
|
+
# You may obtain a copy of the License at
|
6
|
+
#
|
7
|
+
# https://www.apache.org/licenses/LICENSE-2.0
|
8
|
+
#
|
9
|
+
# Unless required by applicable law or agreed to in writing, software
|
10
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
11
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
12
|
+
# See the License for the specific language governing permissions and
|
13
|
+
# limitations under the License.
|
14
|
+
|
15
|
+
|
16
|
+
require "google/apis/bigquery_v2"
|
17
|
+
require "base64"
|
18
|
+
require "google/cloud/bigquery/external/bigtable_source/column"
|
19
|
+
|
20
|
+
module Google
|
21
|
+
module Cloud
|
22
|
+
module Bigquery
|
23
|
+
module External
|
24
|
+
class BigtableSource < External::DataSource
|
25
|
+
##
|
26
|
+
# # BigtableSource::ColumnFamily
|
27
|
+
#
|
28
|
+
# A Bigtable column family used to expose in the table schema along
|
29
|
+
# with its types and columns.
|
30
|
+
#
|
31
|
+
# @example
|
32
|
+
# require "google/cloud/bigquery"
|
33
|
+
#
|
34
|
+
# bigquery = Google::Cloud::Bigquery.new
|
35
|
+
#
|
36
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
37
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
38
|
+
# bt.rowkey_as_string = true
|
39
|
+
# bt.add_family "user" do |u|
|
40
|
+
# u.add_string "name"
|
41
|
+
# u.add_string "email"
|
42
|
+
# u.add_integer "age"
|
43
|
+
# u.add_boolean "active"
|
44
|
+
# end
|
45
|
+
# end
|
46
|
+
#
|
47
|
+
# data = bigquery.query "SELECT * FROM my_ext_table",
|
48
|
+
# external: { my_ext_table: bigtable_table }
|
49
|
+
#
|
50
|
+
# # Iterate over the first page of results
|
51
|
+
# data.each do |row|
|
52
|
+
# puts row[:name]
|
53
|
+
# end
|
54
|
+
# # Retrieve the next page of results
|
55
|
+
# data = data.next if data.next?
|
56
|
+
#
|
57
|
+
class ColumnFamily
|
58
|
+
##
|
59
|
+
# @private Create an empty BigtableSource::ColumnFamily object.
|
60
|
+
def initialize
|
61
|
+
@gapi = Google::Apis::BigqueryV2::BigtableColumnFamily.new
|
62
|
+
@columns = []
|
63
|
+
end
|
64
|
+
|
65
|
+
##
|
66
|
+
# The encoding of the values when the type is not `STRING`.
|
67
|
+
#
|
68
|
+
# @return [String]
|
69
|
+
#
|
70
|
+
# @example
|
71
|
+
# require "google/cloud/bigquery"
|
72
|
+
#
|
73
|
+
# bigquery = Google::Cloud::Bigquery.new
|
74
|
+
#
|
75
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
76
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
77
|
+
# bt.add_family "user" do |u|
|
78
|
+
# u.encoding = "UTF-8"
|
79
|
+
# end
|
80
|
+
# end
|
81
|
+
#
|
82
|
+
# bigtable_table.families[0].encoding #=> "UTF-8"
|
83
|
+
#
|
84
|
+
def encoding
|
85
|
+
@gapi.encoding
|
86
|
+
end
|
87
|
+
|
88
|
+
##
|
89
|
+
# Set the encoding of the values when the type is not `STRING`.
|
90
|
+
# Acceptable encoding values are:
|
91
|
+
#
|
92
|
+
# * `TEXT` - indicates values are alphanumeric text strings.
|
93
|
+
# * `BINARY` - indicates values are encoded using HBase
|
94
|
+
# `Bytes.toBytes` family of functions. This can be overridden on a
|
95
|
+
# column.
|
96
|
+
#
|
97
|
+
# @param [String] new_encoding New encoding value
|
98
|
+
#
|
99
|
+
# @example
|
100
|
+
# require "google/cloud/bigquery"
|
101
|
+
#
|
102
|
+
# bigquery = Google::Cloud::Bigquery.new
|
103
|
+
#
|
104
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
105
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
106
|
+
# bt.add_family "user" do |u|
|
107
|
+
# u.encoding = "UTF-8"
|
108
|
+
# end
|
109
|
+
# end
|
110
|
+
#
|
111
|
+
# bigtable_table.families[0].encoding #=> "UTF-8"
|
112
|
+
#
|
113
|
+
def encoding= new_encoding
|
114
|
+
frozen_check!
|
115
|
+
@gapi.encoding = new_encoding
|
116
|
+
end
|
117
|
+
|
118
|
+
##
|
119
|
+
# Identifier of the column family.
|
120
|
+
#
|
121
|
+
# @return [String]
|
122
|
+
#
|
123
|
+
# @example
|
124
|
+
# require "google/cloud/bigquery"
|
125
|
+
#
|
126
|
+
# bigquery = Google::Cloud::Bigquery.new
|
127
|
+
#
|
128
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
129
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
130
|
+
# bt.add_family "user"
|
131
|
+
# end
|
132
|
+
#
|
133
|
+
# bigtable_table.families[0].family_id #=> "user"
|
134
|
+
#
|
135
|
+
def family_id
|
136
|
+
@gapi.family_id
|
137
|
+
end
|
138
|
+
|
139
|
+
##
|
140
|
+
# Set the identifier of the column family.
|
141
|
+
#
|
142
|
+
# @param [String] new_family_id New family_id value
|
143
|
+
#
|
144
|
+
# @example
|
145
|
+
# require "google/cloud/bigquery"
|
146
|
+
#
|
147
|
+
# bigquery = Google::Cloud::Bigquery.new
|
148
|
+
#
|
149
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
150
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
151
|
+
# bt.add_family "user"
|
152
|
+
# end
|
153
|
+
#
|
154
|
+
# bigtable_table.families[0].family_id #=> "user"
|
155
|
+
# bigtable_table.families[0].family_id = "User"
|
156
|
+
# bigtable_table.families[0].family_id #=> "User"
|
157
|
+
#
|
158
|
+
def family_id= new_family_id
|
159
|
+
frozen_check!
|
160
|
+
@gapi.family_id = new_family_id
|
161
|
+
end
|
162
|
+
|
163
|
+
##
|
164
|
+
# Whether only the latest version of value are exposed for all
|
165
|
+
# columns in this column family.
|
166
|
+
#
|
167
|
+
# @return [Boolean]
|
168
|
+
#
|
169
|
+
# @example
|
170
|
+
# require "google/cloud/bigquery"
|
171
|
+
#
|
172
|
+
# bigquery = Google::Cloud::Bigquery.new
|
173
|
+
#
|
174
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
175
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
176
|
+
# bt.add_family "user" do |u|
|
177
|
+
# u.latest = true
|
178
|
+
# end
|
179
|
+
# end
|
180
|
+
#
|
181
|
+
# bigtable_table.families[0].latest #=> true
|
182
|
+
#
|
183
|
+
def latest
|
184
|
+
@gapi.only_read_latest
|
185
|
+
end
|
186
|
+
|
187
|
+
##
|
188
|
+
# Set whether only the latest version of value are exposed for all
|
189
|
+
# columns in this column family.
|
190
|
+
#
|
191
|
+
# @param [Boolean] new_latest New latest value
|
192
|
+
#
|
193
|
+
# @example
|
194
|
+
# require "google/cloud/bigquery"
|
195
|
+
#
|
196
|
+
# bigquery = Google::Cloud::Bigquery.new
|
197
|
+
#
|
198
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
199
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
200
|
+
# bt.add_family "user" do |u|
|
201
|
+
# u.latest = true
|
202
|
+
# end
|
203
|
+
# end
|
204
|
+
#
|
205
|
+
# bigtable_table.families[0].latest #=> true
|
206
|
+
#
|
207
|
+
def latest= new_latest
|
208
|
+
frozen_check!
|
209
|
+
@gapi.only_read_latest = new_latest
|
210
|
+
end
|
211
|
+
|
212
|
+
##
|
213
|
+
# The type to convert the value in cells of this column family. The
|
214
|
+
# values are expected to be encoded using HBase `Bytes.toBytes`
|
215
|
+
# function when using the `BINARY` encoding value. The following
|
216
|
+
# BigQuery types are allowed:
|
217
|
+
#
|
218
|
+
# * `BYTES`
|
219
|
+
# * `STRING`
|
220
|
+
# * `INTEGER`
|
221
|
+
# * `FLOAT`
|
222
|
+
# * `BOOLEAN`
|
223
|
+
#
|
224
|
+
# Default type is `BYTES`. This can be overridden on a column.
|
225
|
+
#
|
226
|
+
# @return [String]
|
227
|
+
#
|
228
|
+
# @example
|
229
|
+
# require "google/cloud/bigquery"
|
230
|
+
#
|
231
|
+
# bigquery = Google::Cloud::Bigquery.new
|
232
|
+
#
|
233
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
234
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
235
|
+
# bt.add_family "user" do |u|
|
236
|
+
# u.type = "STRING"
|
237
|
+
# end
|
238
|
+
# end
|
239
|
+
#
|
240
|
+
# bigtable_table.families[0].type #=> "STRING"
|
241
|
+
#
|
242
|
+
def type
|
243
|
+
@gapi.type
|
244
|
+
end
|
245
|
+
|
246
|
+
##
|
247
|
+
# Set the type to convert the value in cells of this column family.
|
248
|
+
# The values are expected to be encoded using HBase `Bytes.toBytes`
|
249
|
+
# function when using the `BINARY` encoding value. The following
|
250
|
+
# BigQuery types are allowed:
|
251
|
+
#
|
252
|
+
# * `BYTES`
|
253
|
+
# * `STRING`
|
254
|
+
# * `INTEGER`
|
255
|
+
# * `FLOAT`
|
256
|
+
# * `BOOLEAN`
|
257
|
+
#
|
258
|
+
# Default type is `BYTES`. This can be overridden on a column.
|
259
|
+
#
|
260
|
+
# @param [String] new_type New type value
|
261
|
+
#
|
262
|
+
# @example
|
263
|
+
# require "google/cloud/bigquery"
|
264
|
+
#
|
265
|
+
# bigquery = Google::Cloud::Bigquery.new
|
266
|
+
#
|
267
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
268
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
269
|
+
# bt.add_family "user" do |u|
|
270
|
+
# u.type = "STRING"
|
271
|
+
# end
|
272
|
+
# end
|
273
|
+
#
|
274
|
+
# bigtable_table.families[0].type #=> "STRING"
|
275
|
+
#
|
276
|
+
def type= new_type
|
277
|
+
frozen_check!
|
278
|
+
@gapi.type = new_type
|
279
|
+
end
|
280
|
+
|
281
|
+
##
|
282
|
+
# Lists of columns that should be exposed as individual fields.
|
283
|
+
#
|
284
|
+
# @return [Array<BigtableSource::Column>]
|
285
|
+
#
|
286
|
+
# @example
|
287
|
+
# require "google/cloud/bigquery"
|
288
|
+
#
|
289
|
+
# bigquery = Google::Cloud::Bigquery.new
|
290
|
+
#
|
291
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
292
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
293
|
+
# bt.rowkey_as_string = true
|
294
|
+
# bt.add_family "user" do |u|
|
295
|
+
# u.add_string "name"
|
296
|
+
# u.add_string "email"
|
297
|
+
# u.add_integer "age"
|
298
|
+
# u.add_boolean "active"
|
299
|
+
# end
|
300
|
+
# end
|
301
|
+
#
|
302
|
+
# bigtable_table.families[0].columns.count #=> 4
|
303
|
+
#
|
304
|
+
def columns
|
305
|
+
@columns
|
306
|
+
end
|
307
|
+
|
308
|
+
##
|
309
|
+
# Add a column to the column family to expose in the table schema
|
310
|
+
# along with its types.
|
311
|
+
#
|
312
|
+
# @param [String] qualifier Qualifier of the column. See
|
313
|
+
# {BigtableSource::Column#qualifier}.
|
314
|
+
# @param [String] as A valid identifier to be used as the column
|
315
|
+
# field name if the qualifier is not a valid BigQuery field
|
316
|
+
# identifier (i.e. does not match `[a-zA-Z][a-zA-Z0-9_]*`). See
|
317
|
+
# {BigtableSource::Column#field_name}.
|
318
|
+
# @param [String] type The type to convert the value in cells of
|
319
|
+
# this column. See {BigtableSource::Column#type}. The following
|
320
|
+
# BigQuery types are allowed:
|
321
|
+
#
|
322
|
+
# * `BYTES`
|
323
|
+
# * `STRING`
|
324
|
+
# * `INTEGER`
|
325
|
+
# * `FLOAT`
|
326
|
+
# * `BOOLEAN`
|
327
|
+
#
|
328
|
+
# @yield [column] a block for setting the column
|
329
|
+
# @yieldparam [BigtableSource::Column] column the column object
|
330
|
+
#
|
331
|
+
# @return [Array<BigtableSource::Column>]
|
332
|
+
#
|
333
|
+
# @example
|
334
|
+
# require "google/cloud/bigquery"
|
335
|
+
#
|
336
|
+
# bigquery = Google::Cloud::Bigquery.new
|
337
|
+
#
|
338
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
339
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
340
|
+
# bt.rowkey_as_string = true
|
341
|
+
# bt.add_family "user" do |u|
|
342
|
+
# u.add_column "name", type: "STRING"
|
343
|
+
# end
|
344
|
+
# end
|
345
|
+
#
|
346
|
+
def add_column qualifier, as: nil, type: nil
|
347
|
+
frozen_check!
|
348
|
+
col = BigtableSource::Column.new
|
349
|
+
col.qualifier = qualifier
|
350
|
+
col.field_name = as if as
|
351
|
+
col.type = type if type
|
352
|
+
yield col if block_given?
|
353
|
+
@columns << col
|
354
|
+
col
|
355
|
+
end
|
356
|
+
|
357
|
+
##
|
358
|
+
# Add a column to the column family to expose in the table schema
|
359
|
+
# that is specified as the `BYTES` type.
|
360
|
+
#
|
361
|
+
# @param [String] qualifier Qualifier of the column. See
|
362
|
+
# {BigtableSource::Column#qualifier}.
|
363
|
+
# @param [String] as A valid identifier to be used as the column
|
364
|
+
# field name if the qualifier is not a valid BigQuery field
|
365
|
+
# identifier (i.e. does not match `[a-zA-Z][a-zA-Z0-9_]*`). See
|
366
|
+
# {BigtableSource::Column#field_name}.
|
367
|
+
#
|
368
|
+
# @yield [column] a block for setting the column
|
369
|
+
# @yieldparam [BigtableSource::Column] column the column object
|
370
|
+
#
|
371
|
+
# @return [Array<BigtableSource::Column>]
|
372
|
+
#
|
373
|
+
# @example
|
374
|
+
# require "google/cloud/bigquery"
|
375
|
+
#
|
376
|
+
# bigquery = Google::Cloud::Bigquery.new
|
377
|
+
#
|
378
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
379
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
380
|
+
# bt.rowkey_as_string = true
|
381
|
+
# bt.add_family "user" do |u|
|
382
|
+
# u.add_bytes "avatar"
|
383
|
+
# end
|
384
|
+
# end
|
385
|
+
#
|
386
|
+
def add_bytes qualifier, as: nil
|
387
|
+
col = add_column qualifier, as: as, type: "BYTES"
|
388
|
+
yield col if block_given?
|
389
|
+
col
|
390
|
+
end
|
391
|
+
|
392
|
+
##
|
393
|
+
# Add a column to the column family to expose in the table schema
|
394
|
+
# that is specified as the `STRING` type.
|
395
|
+
#
|
396
|
+
# @param [String] qualifier Qualifier of the column. See
|
397
|
+
# {BigtableSource::Column#qualifier}.
|
398
|
+
# @param [String] as A valid identifier to be used as the column
|
399
|
+
# field name if the qualifier is not a valid BigQuery field
|
400
|
+
# identifier (i.e. does not match `[a-zA-Z][a-zA-Z0-9_]*`). See
|
401
|
+
# {BigtableSource::Column#field_name}.
|
402
|
+
#
|
403
|
+
# @yield [column] a block for setting the column
|
404
|
+
# @yieldparam [BigtableSource::Column] column the column object
|
405
|
+
#
|
406
|
+
# @return [Array<BigtableSource::Column>]
|
407
|
+
#
|
408
|
+
# @example
|
409
|
+
# require "google/cloud/bigquery"
|
410
|
+
#
|
411
|
+
# bigquery = Google::Cloud::Bigquery.new
|
412
|
+
#
|
413
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
414
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
415
|
+
# bt.rowkey_as_string = true
|
416
|
+
# bt.add_family "user" do |u|
|
417
|
+
# u.add_string "name"
|
418
|
+
# end
|
419
|
+
# end
|
420
|
+
#
|
421
|
+
def add_string qualifier, as: nil
|
422
|
+
col = add_column qualifier, as: as, type: "STRING"
|
423
|
+
yield col if block_given?
|
424
|
+
col
|
425
|
+
end
|
426
|
+
|
427
|
+
##
|
428
|
+
# Add a column to the column family to expose in the table schema
|
429
|
+
# that is specified as the `INTEGER` type.
|
430
|
+
#
|
431
|
+
# @param [String] qualifier Qualifier of the column. See
|
432
|
+
# {BigtableSource::Column#qualifier}.
|
433
|
+
# @param [String] as A valid identifier to be used as the column
|
434
|
+
# field name if the qualifier is not a valid BigQuery field
|
435
|
+
# identifier (i.e. does not match `[a-zA-Z][a-zA-Z0-9_]*`). See
|
436
|
+
# {BigtableSource::Column#field_name}.
|
437
|
+
#
|
438
|
+
# @yield [column] a block for setting the column
|
439
|
+
# @yieldparam [BigtableSource::Column] column the column object
|
440
|
+
#
|
441
|
+
# @return [Array<BigtableSource::Column>]
|
442
|
+
#
|
443
|
+
# @example
|
444
|
+
# require "google/cloud/bigquery"
|
445
|
+
#
|
446
|
+
# bigquery = Google::Cloud::Bigquery.new
|
447
|
+
#
|
448
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
449
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
450
|
+
# bt.rowkey_as_string = true
|
451
|
+
# bt.add_family "user" do |u|
|
452
|
+
# u.add_integer "age"
|
453
|
+
# end
|
454
|
+
# end
|
455
|
+
#
|
456
|
+
def add_integer qualifier, as: nil
|
457
|
+
col = add_column qualifier, as: as, type: "INTEGER"
|
458
|
+
yield col if block_given?
|
459
|
+
col
|
460
|
+
end
|
461
|
+
|
462
|
+
##
|
463
|
+
# Add a column to the column family to expose in the table schema
|
464
|
+
# that is specified as the `FLOAT` type.
|
465
|
+
#
|
466
|
+
# @param [String] qualifier Qualifier of the column. See
|
467
|
+
# {BigtableSource::Column#qualifier}.
|
468
|
+
# @param [String] as A valid identifier to be used as the column
|
469
|
+
# field name if the qualifier is not a valid BigQuery field
|
470
|
+
# identifier (i.e. does not match `[a-zA-Z][a-zA-Z0-9_]*`). See
|
471
|
+
# {BigtableSource::Column#field_name}.
|
472
|
+
#
|
473
|
+
# @yield [column] a block for setting the column
|
474
|
+
# @yieldparam [BigtableSource::Column] column the column object
|
475
|
+
#
|
476
|
+
# @return [Array<BigtableSource::Column>]
|
477
|
+
#
|
478
|
+
# @example
|
479
|
+
# require "google/cloud/bigquery"
|
480
|
+
#
|
481
|
+
# bigquery = Google::Cloud::Bigquery.new
|
482
|
+
#
|
483
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
484
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
485
|
+
# bt.rowkey_as_string = true
|
486
|
+
# bt.add_family "user" do |u|
|
487
|
+
# u.add_float "score"
|
488
|
+
# end
|
489
|
+
# end
|
490
|
+
#
|
491
|
+
def add_float qualifier, as: nil
|
492
|
+
col = add_column qualifier, as: as, type: "FLOAT"
|
493
|
+
yield col if block_given?
|
494
|
+
col
|
495
|
+
end
|
496
|
+
|
497
|
+
##
|
498
|
+
# Add a column to the column family to expose in the table schema
|
499
|
+
# that is specified as the `BOOLEAN` type.
|
500
|
+
#
|
501
|
+
# @param [String] qualifier Qualifier of the column. See
|
502
|
+
# {BigtableSource::Column#qualifier}.
|
503
|
+
# @param [String] as A valid identifier to be used as the column
|
504
|
+
# field name if the qualifier is not a valid BigQuery field
|
505
|
+
# identifier (i.e. does not match `[a-zA-Z][a-zA-Z0-9_]*`). See
|
506
|
+
# {BigtableSource::Column#field_name}.
|
507
|
+
#
|
508
|
+
# @yield [column] a block for setting the column
|
509
|
+
# @yieldparam [BigtableSource::Column] column the column object
|
510
|
+
#
|
511
|
+
# @return [Array<BigtableSource::Column>]
|
512
|
+
#
|
513
|
+
# @example
|
514
|
+
# require "google/cloud/bigquery"
|
515
|
+
#
|
516
|
+
# bigquery = Google::Cloud::Bigquery.new
|
517
|
+
#
|
518
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
519
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
520
|
+
# bt.rowkey_as_string = true
|
521
|
+
# bt.add_family "user" do |u|
|
522
|
+
# u.add_boolean "active"
|
523
|
+
# end
|
524
|
+
# end
|
525
|
+
#
|
526
|
+
def add_boolean qualifier, as: nil
|
527
|
+
col = add_column qualifier, as: as, type: "BOOLEAN"
|
528
|
+
yield col if block_given?
|
529
|
+
col
|
530
|
+
end
|
531
|
+
|
532
|
+
##
|
533
|
+
# @private Google API Client object.
|
534
|
+
def to_gapi
|
535
|
+
@gapi.columns = @columns.map(&:to_gapi)
|
536
|
+
@gapi
|
537
|
+
end
|
538
|
+
|
539
|
+
##
|
540
|
+
# @private Google API Client object.
|
541
|
+
def self.from_gapi gapi
|
542
|
+
new_fam = new
|
543
|
+
new_fam.instance_variable_set :@gapi, gapi
|
544
|
+
columns = Array(gapi.columns).map { |col_gapi| BigtableSource::Column.from_gapi col_gapi }
|
545
|
+
new_fam.instance_variable_set :@columns, columns
|
546
|
+
new_fam
|
547
|
+
end
|
548
|
+
|
549
|
+
##
|
550
|
+
# @private
|
551
|
+
def freeze
|
552
|
+
@columns.map(&:freeze!)
|
553
|
+
@columns.freeze!
|
554
|
+
super
|
555
|
+
end
|
556
|
+
|
557
|
+
protected
|
558
|
+
|
559
|
+
def frozen_check!
|
560
|
+
return unless frozen?
|
561
|
+
raise ArgumentError, "Cannot modify external data source when frozen"
|
562
|
+
end
|
563
|
+
end
|
564
|
+
|
565
|
+
##
|
566
|
+
# # BigtableSource::Column
|
567
|
+
#
|
568
|
+
# A Bigtable column to expose in the table schema along with its
|
569
|
+
# types.
|
570
|
+
#
|
571
|
+
# @example
|
572
|
+
# require "google/cloud/bigquery"
|
573
|
+
#
|
574
|
+
# bigquery = Google::Cloud::Bigquery.new
|
575
|
+
#
|
576
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
577
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
578
|
+
# bt.rowkey_as_string = true
|
579
|
+
# bt.add_family "user" do |u|
|
580
|
+
# u.add_string "name"
|
581
|
+
# u.add_string "email"
|
582
|
+
# u.add_integer "age"
|
583
|
+
# u.add_boolean "active"
|
584
|
+
# end
|
585
|
+
# end
|
586
|
+
#
|
587
|
+
# data = bigquery.query "SELECT * FROM my_ext_table",
|
588
|
+
# external: { my_ext_table: bigtable_table }
|
589
|
+
#
|
590
|
+
# # Iterate over the first page of results
|
591
|
+
# data.each do |row|
|
592
|
+
# puts row[:name]
|
593
|
+
# end
|
594
|
+
# # Retrieve the next page of results
|
595
|
+
# data = data.next if data.next?
|
596
|
+
#
|
597
|
+
class Column
|
598
|
+
##
|
599
|
+
# @private Create an empty BigtableSource::Column object.
|
600
|
+
def initialize
|
601
|
+
@gapi = Google::Apis::BigqueryV2::BigtableColumn.new
|
602
|
+
end
|
603
|
+
|
604
|
+
##
|
605
|
+
# Qualifier of the column. Columns in the parent column family that
|
606
|
+
# has this exact qualifier are exposed as `.` field. If the
|
607
|
+
# qualifier is valid UTF-8 string, it will be represented as a UTF-8
|
608
|
+
# string. Otherwise, it will represented as a ASCII-8BIT string.
|
609
|
+
#
|
610
|
+
# If the qualifier is not a valid BigQuery field identifier (does
|
611
|
+
# not match `[a-zA-Z][a-zA-Z0-9_]*`) a valid identifier must be
|
612
|
+
# provided as `field_name`.
|
613
|
+
#
|
614
|
+
# @return [String]
|
615
|
+
#
|
616
|
+
# @example
|
617
|
+
# require "google/cloud/bigquery"
|
618
|
+
#
|
619
|
+
# bigquery = Google::Cloud::Bigquery.new
|
620
|
+
#
|
621
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
622
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
623
|
+
# bt.add_family "user" do |u|
|
624
|
+
# u.add_string "name" do |col|
|
625
|
+
# col.qualifier # "user"
|
626
|
+
# col.qualifier = "User"
|
627
|
+
# col.qualifier # "User"
|
628
|
+
# end
|
629
|
+
# end
|
630
|
+
# end
|
631
|
+
#
|
632
|
+
def qualifier
|
633
|
+
@gapi.qualifier_string || Base64.strict_decode64(@gapi.qualifier_encoded.to_s)
|
634
|
+
end
|
635
|
+
|
636
|
+
##
|
637
|
+
# Set the qualifier of the column. Columns in the parent column
|
638
|
+
# family that has this exact qualifier are exposed as `.` field.
|
639
|
+
# Values that are valid UTF-8 strings will be treated as such. All
|
640
|
+
# other values will be treated as `BINARY`.
|
641
|
+
#
|
642
|
+
# @param [String] new_qualifier New qualifier value
|
643
|
+
#
|
644
|
+
# @example
|
645
|
+
# require "google/cloud/bigquery"
|
646
|
+
#
|
647
|
+
# bigquery = Google::Cloud::Bigquery.new
|
648
|
+
#
|
649
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
650
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
651
|
+
# bt.add_family "user" do |u|
|
652
|
+
# u.add_string "name" do |col|
|
653
|
+
# col.qualifier # "user"
|
654
|
+
# col.qualifier = "User"
|
655
|
+
# col.qualifier # "User"
|
656
|
+
# end
|
657
|
+
# end
|
658
|
+
# end
|
659
|
+
#
|
660
|
+
def qualifier= new_qualifier
|
661
|
+
frozen_check!
|
662
|
+
raise ArgumentError if new_qualifier.nil?
|
663
|
+
|
664
|
+
utf8_qualifier = new_qualifier.encode Encoding::UTF_8
|
665
|
+
if utf8_qualifier.valid_encoding?
|
666
|
+
@gapi.qualifier_string = utf8_qualifier
|
667
|
+
if @gapi.instance_variables.include? :@qualifier_encoded
|
668
|
+
@gapi.remove_instance_variable :@qualifier_encoded
|
669
|
+
end
|
670
|
+
else
|
671
|
+
@gapi.qualifier_encoded = Base64.strict_encode64 new_qualifier
|
672
|
+
if @gapi.instance_variables.include? :@qualifier_string
|
673
|
+
@gapi.remove_instance_variable :@qualifier_string
|
674
|
+
end
|
675
|
+
end
|
676
|
+
rescue EncodingError
|
677
|
+
@gapi.qualifier_encoded = Base64.strict_encode64 new_qualifier
|
678
|
+
@gapi.remove_instance_variable :@qualifier_string if @gapi.instance_variables.include? :@qualifier_string
|
679
|
+
end
|
680
|
+
|
681
|
+
##
|
682
|
+
# The encoding of the values when the type is not `STRING`.
|
683
|
+
#
|
684
|
+
# @return [String]
|
685
|
+
#
|
686
|
+
# @example
|
687
|
+
# require "google/cloud/bigquery"
|
688
|
+
#
|
689
|
+
# bigquery = Google::Cloud::Bigquery.new
|
690
|
+
#
|
691
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
692
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
693
|
+
# bt.add_family "user" do |u|
|
694
|
+
# u.add_bytes "name" do |col|
|
695
|
+
# col.encoding = "TEXT"
|
696
|
+
# col.encoding # "TEXT"
|
697
|
+
# end
|
698
|
+
# end
|
699
|
+
# end
|
700
|
+
#
|
701
|
+
def encoding
|
702
|
+
@gapi.encoding
|
703
|
+
end
|
704
|
+
|
705
|
+
##
|
706
|
+
# Set the encoding of the values when the type is not `STRING`.
|
707
|
+
# Acceptable encoding values are:
|
708
|
+
#
|
709
|
+
# * `TEXT` - indicates values are alphanumeric text strings.
|
710
|
+
# * `BINARY` - indicates values are encoded using HBase
|
711
|
+
# `Bytes.toBytes` family of functions. This can be overridden on a
|
712
|
+
# column.
|
713
|
+
#
|
714
|
+
# @param [String] new_encoding New encoding value
|
715
|
+
#
|
716
|
+
# @example
|
717
|
+
# require "google/cloud/bigquery"
|
718
|
+
#
|
719
|
+
# bigquery = Google::Cloud::Bigquery.new
|
720
|
+
#
|
721
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
722
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
723
|
+
# bt.add_family "user" do |u|
|
724
|
+
# u.add_bytes "name" do |col|
|
725
|
+
# col.encoding = "TEXT"
|
726
|
+
# col.encoding # "TEXT"
|
727
|
+
# end
|
728
|
+
# end
|
729
|
+
# end
|
730
|
+
#
|
731
|
+
def encoding= new_encoding
|
732
|
+
frozen_check!
|
733
|
+
@gapi.encoding = new_encoding
|
734
|
+
end
|
735
|
+
|
736
|
+
##
|
737
|
+
# If the qualifier is not a valid BigQuery field identifier (does
|
738
|
+
# not match `[a-zA-Z][a-zA-Z0-9_]*`) a valid identifier must be
|
739
|
+
# provided as the column field name and is used as field name in
|
740
|
+
# queries.
|
741
|
+
#
|
742
|
+
# @return [String]
|
743
|
+
#
|
744
|
+
# @example
|
745
|
+
# require "google/cloud/bigquery"
|
746
|
+
#
|
747
|
+
# bigquery = Google::Cloud::Bigquery.new
|
748
|
+
#
|
749
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
750
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
751
|
+
# bt.add_family "user" do |u|
|
752
|
+
# u.add_string "001_name", as: "user" do |col|
|
753
|
+
# col.field_name # "user"
|
754
|
+
# col.field_name = "User"
|
755
|
+
# col.field_name # "User"
|
756
|
+
# end
|
757
|
+
# end
|
758
|
+
# end
|
759
|
+
#
|
760
|
+
def field_name
|
761
|
+
@gapi.field_name
|
762
|
+
end
|
763
|
+
|
764
|
+
##
|
765
|
+
# Sets the identifier to be used as the column field name in queries
|
766
|
+
# when the qualifier is not a valid BigQuery field identifier (does
|
767
|
+
# not match `[a-zA-Z][a-zA-Z0-9_]*`).
|
768
|
+
#
|
769
|
+
# @param [String] new_field_name New field_name value
|
770
|
+
#
|
771
|
+
# @example
|
772
|
+
# require "google/cloud/bigquery"
|
773
|
+
#
|
774
|
+
# bigquery = Google::Cloud::Bigquery.new
|
775
|
+
#
|
776
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
777
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
778
|
+
# bt.add_family "user" do |u|
|
779
|
+
# u.add_string "001_name", as: "user" do |col|
|
780
|
+
# col.field_name # "user"
|
781
|
+
# col.field_name = "User"
|
782
|
+
# col.field_name # "User"
|
783
|
+
# end
|
784
|
+
# end
|
785
|
+
# end
|
786
|
+
#
|
787
|
+
def field_name= new_field_name
|
788
|
+
frozen_check!
|
789
|
+
@gapi.field_name = new_field_name
|
790
|
+
end
|
791
|
+
|
792
|
+
##
|
793
|
+
# Whether only the latest version of value in this column are
|
794
|
+
# exposed. Can also be set at the column family level. However, this
|
795
|
+
# value takes precedence when set at both levels.
|
796
|
+
#
|
797
|
+
# @return [Boolean]
|
798
|
+
#
|
799
|
+
# @example
|
800
|
+
# require "google/cloud/bigquery"
|
801
|
+
#
|
802
|
+
# bigquery = Google::Cloud::Bigquery.new
|
803
|
+
#
|
804
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
805
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
806
|
+
# bt.add_family "user" do |u|
|
807
|
+
# u.add_string "name" do |col|
|
808
|
+
# col.latest = true
|
809
|
+
# col.latest # true
|
810
|
+
# end
|
811
|
+
# end
|
812
|
+
# end
|
813
|
+
#
|
814
|
+
def latest
|
815
|
+
@gapi.only_read_latest
|
816
|
+
end
|
817
|
+
|
818
|
+
##
|
819
|
+
# Set whether only the latest version of value in this column are
|
820
|
+
# exposed. Can also be set at the column family level. However, this
|
821
|
+
# value takes precedence when set at both levels.
|
822
|
+
#
|
823
|
+
# @param [Boolean] new_latest New latest value
|
824
|
+
#
|
825
|
+
# @example
|
826
|
+
# require "google/cloud/bigquery"
|
827
|
+
#
|
828
|
+
# bigquery = Google::Cloud::Bigquery.new
|
829
|
+
#
|
830
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
831
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
832
|
+
# bt.add_family "user" do |u|
|
833
|
+
# u.add_string "name" do |col|
|
834
|
+
# col.latest = true
|
835
|
+
# col.latest # true
|
836
|
+
# end
|
837
|
+
# end
|
838
|
+
# end
|
839
|
+
#
|
840
|
+
def latest= new_latest
|
841
|
+
frozen_check!
|
842
|
+
@gapi.only_read_latest = new_latest
|
843
|
+
end
|
844
|
+
|
845
|
+
##
|
846
|
+
# The type to convert the value in cells of this column. The values
|
847
|
+
# are expected to be encoded using HBase `Bytes.toBytes` function
|
848
|
+
# when using the `BINARY` encoding value. The following BigQuery
|
849
|
+
# types are allowed:
|
850
|
+
#
|
851
|
+
# * `BYTES`
|
852
|
+
# * `STRING`
|
853
|
+
# * `INTEGER`
|
854
|
+
# * `FLOAT`
|
855
|
+
# * `BOOLEAN`
|
856
|
+
#
|
857
|
+
# Default type is `BYTES`. Can also be set at the column family
|
858
|
+
# level. However, this value takes precedence when set at both
|
859
|
+
# levels.
|
860
|
+
#
|
861
|
+
# @return [String]
|
862
|
+
#
|
863
|
+
# @example
|
864
|
+
# require "google/cloud/bigquery"
|
865
|
+
#
|
866
|
+
# bigquery = Google::Cloud::Bigquery.new
|
867
|
+
#
|
868
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
869
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
870
|
+
# bt.add_family "user" do |u|
|
871
|
+
# u.add_string "name" do |col|
|
872
|
+
# col.type # "STRING"
|
873
|
+
# end
|
874
|
+
# end
|
875
|
+
# end
|
876
|
+
#
|
877
|
+
def type
|
878
|
+
@gapi.type
|
879
|
+
end
|
880
|
+
|
881
|
+
##
|
882
|
+
# Set the type to convert the value in cells of this column. The
|
883
|
+
# values are expected to be encoded using HBase `Bytes.toBytes`
|
884
|
+
# function when using the `BINARY` encoding value. The following
|
885
|
+
# BigQuery types are allowed:
|
886
|
+
#
|
887
|
+
# * `BYTES`
|
888
|
+
# * `STRING`
|
889
|
+
# * `INTEGER`
|
890
|
+
# * `FLOAT`
|
891
|
+
# * `BOOLEAN`
|
892
|
+
#
|
893
|
+
# Default type is `BYTES`. Can also be set at the column family
|
894
|
+
# level. However, this value takes precedence when set at both
|
895
|
+
# levels.
|
896
|
+
#
|
897
|
+
# @param [String] new_type New type value
|
898
|
+
#
|
899
|
+
# @example
|
900
|
+
# require "google/cloud/bigquery"
|
901
|
+
#
|
902
|
+
# bigquery = Google::Cloud::Bigquery.new
|
903
|
+
#
|
904
|
+
# bigtable_url = "https://googleapis.com/bigtable/projects/..."
|
905
|
+
# bigtable_table = bigquery.external bigtable_url do |bt|
|
906
|
+
# bt.add_family "user" do |u|
|
907
|
+
# u.add_string "name" do |col|
|
908
|
+
# col.type # "STRING"
|
909
|
+
# col.type = "BYTES"
|
910
|
+
# col.type # "BYTES"
|
911
|
+
# end
|
912
|
+
# end
|
913
|
+
# end
|
914
|
+
#
|
915
|
+
def type= new_type
|
916
|
+
frozen_check!
|
917
|
+
@gapi.type = new_type
|
918
|
+
end
|
919
|
+
|
920
|
+
##
|
921
|
+
# @private Google API Client object.
|
922
|
+
def to_gapi
|
923
|
+
@gapi
|
924
|
+
end
|
925
|
+
|
926
|
+
##
|
927
|
+
# @private Google API Client object.
|
928
|
+
def self.from_gapi gapi
|
929
|
+
new_col = new
|
930
|
+
new_col.instance_variable_set :@gapi, gapi
|
931
|
+
new_col
|
932
|
+
end
|
933
|
+
|
934
|
+
protected
|
935
|
+
|
936
|
+
def frozen_check!
|
937
|
+
return unless frozen?
|
938
|
+
raise ArgumentError, "Cannot modify external data source when frozen"
|
939
|
+
end
|
940
|
+
end
|
941
|
+
end
|
942
|
+
end
|
943
|
+
end
|
944
|
+
end
|
945
|
+
end
|