google-cloud-vision 0.22.1 → 0.23.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/lib/google/cloud/vision.rb +133 -6
- data/lib/google/cloud/vision/annotate.rb +55 -24
- data/lib/google/cloud/vision/annotation.rb +93 -11
- data/lib/google/cloud/vision/annotation/crop_hint.rb +126 -0
- data/lib/google/cloud/vision/annotation/text.rb +789 -11
- data/lib/google/cloud/vision/annotation/web.rb +420 -0
- data/lib/google/cloud/vision/image.rb +153 -17
- data/lib/google/cloud/vision/project.rb +29 -8
- data/lib/google/cloud/vision/v1/doc/google/cloud/vision/v1/image_annotator.rb +148 -81
- data/lib/google/cloud/vision/v1/doc/google/cloud/vision/v1/text_annotation.rb +231 -0
- data/lib/google/cloud/vision/v1/doc/google/cloud/vision/v1/web_detection.rb +73 -0
- data/lib/google/cloud/vision/v1/image_annotator_client.rb +4 -4
- data/lib/google/cloud/vision/v1/image_annotator_pb.rb +24 -0
- data/lib/google/cloud/vision/v1/image_annotator_services_pb.rb +3 -3
- data/lib/google/cloud/vision/v1/text_annotation_pb.rb +88 -0
- data/lib/google/cloud/vision/v1/web_detection_pb.rb +40 -0
- data/lib/google/cloud/vision/version.rb +1 -1
- metadata +8 -2
@@ -0,0 +1,126 @@
|
|
1
|
+
# Copyright 2017 Google Inc. All rights reserved.
|
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
|
+
# http://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/cloud/vision/annotation/vertex"
|
17
|
+
|
18
|
+
module Google
|
19
|
+
module Cloud
|
20
|
+
module Vision
|
21
|
+
class Annotation
|
22
|
+
##
|
23
|
+
# # CropHint
|
24
|
+
#
|
25
|
+
# A single crop hint that is used to generate new crops when serving
|
26
|
+
# images.
|
27
|
+
#
|
28
|
+
# @example
|
29
|
+
# require "google/cloud/vision"
|
30
|
+
#
|
31
|
+
# vision = Google::Cloud::Vision.new
|
32
|
+
#
|
33
|
+
# image = vision.image "path/to/face.jpg"
|
34
|
+
#
|
35
|
+
# crop_hints = image.crop_hints
|
36
|
+
# crop_hints.count #=> 1
|
37
|
+
# crop_hint = crop_hints.first
|
38
|
+
#
|
39
|
+
# crop_hint.bounds.count #=> 4
|
40
|
+
# crop_hint.bounds[0].x #=> 1
|
41
|
+
# crop_hint.bounds[0].y #=> 0
|
42
|
+
# crop_hint.bounds[1].x #=> 511
|
43
|
+
# crop_hint.bounds[1].y #=> 0
|
44
|
+
# crop_hint.bounds[2].x #=> 511
|
45
|
+
# crop_hint.bounds[2].y #=> 383
|
46
|
+
# crop_hint.bounds[3].x #=> 0
|
47
|
+
# crop_hint.bounds[3].y #=> 383
|
48
|
+
#
|
49
|
+
# crop_hint.confidence #=> 1.0
|
50
|
+
# crop_hint.importance_fraction #=> 1.0399999618530273
|
51
|
+
#
|
52
|
+
class CropHint
|
53
|
+
##
|
54
|
+
# @private The Google::Cloud::Vision::V1::CropHint GRPC object.
|
55
|
+
attr_accessor :grpc
|
56
|
+
|
57
|
+
##
|
58
|
+
# @private Creates a new Entity instance.
|
59
|
+
def initialize
|
60
|
+
@grpc = nil
|
61
|
+
end
|
62
|
+
|
63
|
+
##
|
64
|
+
# The bounding polygon for the crop region. The coordinates of the
|
65
|
+
# bounding box are in the original image's scale.
|
66
|
+
#
|
67
|
+
# @return [Array<Vertex>] An array of vertices.
|
68
|
+
#
|
69
|
+
def bounds
|
70
|
+
return [] unless @grpc.bounding_poly
|
71
|
+
@bounds ||= Array(@grpc.bounding_poly.vertices).map do |v|
|
72
|
+
Vertex.from_grpc v
|
73
|
+
end
|
74
|
+
end
|
75
|
+
|
76
|
+
##
|
77
|
+
# The confidence of this being a salient region.
|
78
|
+
#
|
79
|
+
# @return [Float] A value in the range [0, 1].
|
80
|
+
#
|
81
|
+
def confidence
|
82
|
+
@grpc.confidence
|
83
|
+
end
|
84
|
+
|
85
|
+
##
|
86
|
+
# The fraction of importance of this salient region with respect to
|
87
|
+
# the original image.
|
88
|
+
#
|
89
|
+
# @return [Float]
|
90
|
+
#
|
91
|
+
def importance_fraction
|
92
|
+
@grpc.importance_fraction
|
93
|
+
end
|
94
|
+
|
95
|
+
##
|
96
|
+
# Deeply converts object to a hash. All keys will be symbolized.
|
97
|
+
#
|
98
|
+
# @return [Hash]
|
99
|
+
#
|
100
|
+
def to_h
|
101
|
+
{ bounds: bounds.map(&:to_h), confidence: confidence,
|
102
|
+
importance_fraction: importance_fraction }
|
103
|
+
end
|
104
|
+
|
105
|
+
# @private
|
106
|
+
def to_s
|
107
|
+
tmplt = "bounds: %i, confidence: %s, importance_fraction: %s"
|
108
|
+
format tmplt, bounds.count, confidence.inspect,
|
109
|
+
importance_fraction.inspect
|
110
|
+
end
|
111
|
+
|
112
|
+
# @private
|
113
|
+
def inspect
|
114
|
+
"#<#{self.class.name} #{self}>"
|
115
|
+
end
|
116
|
+
|
117
|
+
##
|
118
|
+
# @private New Annotation::Entity from a GRPC object.
|
119
|
+
def self.from_grpc grpc
|
120
|
+
new.tap { |f| f.instance_variable_set :@grpc, grpc }
|
121
|
+
end
|
122
|
+
end
|
123
|
+
end
|
124
|
+
end
|
125
|
+
end
|
126
|
+
end
|
@@ -22,7 +22,11 @@ module Google
|
|
22
22
|
##
|
23
23
|
# # Text
|
24
24
|
#
|
25
|
-
# The
|
25
|
+
# The results from either the `TEXT_DETECTION` feature (OCR for shorter
|
26
|
+
# documents with sparse text) or the `DOCUMENT_TEXT_DETECTION` feature
|
27
|
+
# (OCR for longer documents with dense text). Optional. Contains
|
28
|
+
# structured representations of OCR extracted text, as well as
|
29
|
+
# the entire UTF-8 text as a string.
|
26
30
|
#
|
27
31
|
# @example
|
28
32
|
# require "google/cloud/vision"
|
@@ -32,11 +36,20 @@ module Google
|
|
32
36
|
# image = vision.image "path/to/text.png"
|
33
37
|
#
|
34
38
|
# text = image.text
|
35
|
-
#
|
36
|
-
# text.words.count #=> 28
|
39
|
+
#
|
37
40
|
# text.text
|
38
41
|
# # "Google Cloud Client for Ruby an idiomatic, intuitive... "
|
39
42
|
#
|
43
|
+
# text.words[0].text #=> "Google"
|
44
|
+
# text.words[0].bounds.count #=> 4
|
45
|
+
# vertex = text.words[0].bounds.first
|
46
|
+
# vertex.x #=> 13
|
47
|
+
# vertex.y #=> 8
|
48
|
+
#
|
49
|
+
# # Use `pages` to access a full structural representation
|
50
|
+
# text.pages[0].blocks[0].paragraphs[0].words[0].symbols[0].text
|
51
|
+
# #=> "G"
|
52
|
+
#
|
40
53
|
class Text
|
41
54
|
##
|
42
55
|
# @private The EntityAnnotation GRPC object.
|
@@ -90,6 +103,20 @@ module Google
|
|
90
103
|
@words
|
91
104
|
end
|
92
105
|
|
106
|
+
##
|
107
|
+
# Each page in the detected text, with the metadata for each page.
|
108
|
+
# Contains a structured representation of OCR extracted text.
|
109
|
+
# The hierarchy of an OCR extracted text structure is like this:
|
110
|
+
# Page -> Block -> Paragraph -> Word -> Symbol
|
111
|
+
# Each structural component, starting from Page, may further have its
|
112
|
+
# own properties. Properties describe detected languages, breaks etc..
|
113
|
+
#
|
114
|
+
# @return [Array<Page>]
|
115
|
+
#
|
116
|
+
def pages
|
117
|
+
@pages
|
118
|
+
end
|
119
|
+
|
93
120
|
##
|
94
121
|
# Deeply converts object to a hash. All keys will be symbolized.
|
95
122
|
#
|
@@ -97,7 +124,7 @@ module Google
|
|
97
124
|
#
|
98
125
|
def to_h
|
99
126
|
{ text: text, locale: locale, bounds: bounds.map(&:to_h),
|
100
|
-
words: words.map(&:to_h) }
|
127
|
+
words: words.map(&:to_h), pages: pages.map(&:to_h) }
|
101
128
|
end
|
102
129
|
|
103
130
|
# @private
|
@@ -112,27 +139,35 @@ module Google
|
|
112
139
|
|
113
140
|
# @private
|
114
141
|
def inspect
|
115
|
-
format "#<Text text: %s, locale: %s, bounds: %i, words: %i
|
116
|
-
text.inspect, locale.inspect, bounds.count,
|
142
|
+
format "#<Text text: %s, locale: %s, bounds: %i, words: %i," \
|
143
|
+
" pages: %i>", text.inspect, locale.inspect, bounds.count,
|
144
|
+
words.count, pages.count
|
117
145
|
end
|
118
146
|
|
119
147
|
##
|
120
|
-
# @private
|
121
|
-
#
|
122
|
-
def self.from_grpc
|
123
|
-
text, *words = Array
|
148
|
+
# @private Create a new Annotation::Text by merging two GRPC models of
|
149
|
+
# text representation.
|
150
|
+
def self.from_grpc grpc_text_annotations, grpc_full_text_annotation
|
151
|
+
text, *words = Array grpc_text_annotations
|
124
152
|
return nil if text.nil?
|
153
|
+
|
154
|
+
# Since text is taken from grpc_text_annotations, do not use text
|
155
|
+
# from grpc_full_text_annotation in this merged model.
|
156
|
+
# Instead, just take the pages.
|
157
|
+
pages = grpc_full_text_annotation.pages
|
125
158
|
new.tap do |t|
|
126
159
|
t.instance_variable_set :@grpc, text
|
127
160
|
t.instance_variable_set :@words,
|
128
161
|
words.map { |w| Word.from_grpc w }
|
162
|
+
t.instance_variable_set :@pages,
|
163
|
+
pages.map { |p| Page.from_grpc p }
|
129
164
|
end
|
130
165
|
end
|
131
166
|
|
132
167
|
##
|
133
168
|
# # Word
|
134
169
|
#
|
135
|
-
# A word within a detected text (OCR). See {Text}.
|
170
|
+
# A word within a detected text (OCR). See {Text#words}.
|
136
171
|
#
|
137
172
|
# @example
|
138
173
|
# require "google/cloud/vision"
|
@@ -215,6 +250,749 @@ module Google
|
|
215
250
|
new.tap { |w| w.instance_variable_set :@grpc, grpc }
|
216
251
|
end
|
217
252
|
end
|
253
|
+
|
254
|
+
##
|
255
|
+
# # Page
|
256
|
+
#
|
257
|
+
# A page within a detected text (OCR). See {Text#pages}.
|
258
|
+
#
|
259
|
+
# @example
|
260
|
+
# require "google/cloud/vision"
|
261
|
+
#
|
262
|
+
# vision = Google::Cloud::Vision.new
|
263
|
+
#
|
264
|
+
# image = vision.image "path/to/text.png"
|
265
|
+
#
|
266
|
+
# text = image.text
|
267
|
+
#
|
268
|
+
# page = text.pages.first
|
269
|
+
#
|
270
|
+
# page.languages.first.code #=> "en"
|
271
|
+
# page.wont_be :prefix_break?
|
272
|
+
# page.width #=> 400
|
273
|
+
# page.height #=> 80
|
274
|
+
# page.blocks.count #=> 1
|
275
|
+
#
|
276
|
+
class Page
|
277
|
+
##
|
278
|
+
# @private The EntityAnnotation GRPC object.
|
279
|
+
attr_accessor :grpc
|
280
|
+
|
281
|
+
##
|
282
|
+
# @private Creates a new Page instance.
|
283
|
+
def initialize
|
284
|
+
@grpc = nil
|
285
|
+
end
|
286
|
+
|
287
|
+
##
|
288
|
+
# A list of detected languages together with confidence.
|
289
|
+
#
|
290
|
+
# @return [Array<Language>]
|
291
|
+
#
|
292
|
+
def languages
|
293
|
+
@languages ||= Array(@grpc.property.detected_languages).map do |l|
|
294
|
+
Language.from_grpc l
|
295
|
+
end
|
296
|
+
end
|
297
|
+
|
298
|
+
##
|
299
|
+
# The type of a detected break at the start or end of the page.
|
300
|
+
#
|
301
|
+
# @return [::Symbol]
|
302
|
+
#
|
303
|
+
def break_type
|
304
|
+
@grpc.property.detected_break &&
|
305
|
+
@grpc.property.detected_break.type.to_sym
|
306
|
+
end
|
307
|
+
|
308
|
+
##
|
309
|
+
# True if a detected break prepends the page.
|
310
|
+
#
|
311
|
+
# @return [Boolean]
|
312
|
+
#
|
313
|
+
def prefix_break?
|
314
|
+
@grpc.property.detected_break &&
|
315
|
+
@grpc.property.detected_break.is_prefix
|
316
|
+
end
|
317
|
+
|
318
|
+
##
|
319
|
+
# Page width in pixels.
|
320
|
+
#
|
321
|
+
# @return [Integer]
|
322
|
+
#
|
323
|
+
def width
|
324
|
+
@grpc.width
|
325
|
+
end
|
326
|
+
|
327
|
+
##
|
328
|
+
# Page height in pixels.
|
329
|
+
#
|
330
|
+
# @return [Integer]
|
331
|
+
#
|
332
|
+
def height
|
333
|
+
@grpc.height
|
334
|
+
end
|
335
|
+
|
336
|
+
##
|
337
|
+
# List of blocks of text, images etc on this page.
|
338
|
+
#
|
339
|
+
# @return [Array<Block>]
|
340
|
+
#
|
341
|
+
def blocks
|
342
|
+
@blocks ||= Array(@grpc.blocks).map do |b|
|
343
|
+
Block.from_grpc b
|
344
|
+
end
|
345
|
+
end
|
346
|
+
|
347
|
+
##
|
348
|
+
# Deeply converts object to a hash. All keys will be symbolized.
|
349
|
+
#
|
350
|
+
# @return [Hash]
|
351
|
+
#
|
352
|
+
def to_h
|
353
|
+
{ languages: languages.map(&:to_h), break_type: break_type,
|
354
|
+
prefix_break: prefix_break?, width: width, height: height,
|
355
|
+
blocks: blocks.map(&:to_h) }
|
356
|
+
end
|
357
|
+
|
358
|
+
# @private
|
359
|
+
def to_s
|
360
|
+
tmplt = "languages: %s, break_type: %s, prefix_break: %s," \
|
361
|
+
" width: %s, height: %s, blocks: %i"
|
362
|
+
format tmplt, languages.inspect, break_type, prefix_break?, width,
|
363
|
+
height, blocks.count
|
364
|
+
end
|
365
|
+
|
366
|
+
# @private
|
367
|
+
def inspect
|
368
|
+
"#<#{self.class.name} #{self}>"
|
369
|
+
end
|
370
|
+
|
371
|
+
##
|
372
|
+
# @private New Annotation::Text::Page from a GRPC
|
373
|
+
# object.
|
374
|
+
def self.from_grpc grpc
|
375
|
+
new.tap { |w| w.instance_variable_set :@grpc, grpc }
|
376
|
+
end
|
377
|
+
|
378
|
+
##
|
379
|
+
# # Block
|
380
|
+
#
|
381
|
+
# A logical element on the page. See {Page}.
|
382
|
+
#
|
383
|
+
# @example
|
384
|
+
# require "google/cloud/vision"
|
385
|
+
#
|
386
|
+
# vision = Google::Cloud::Vision.new
|
387
|
+
#
|
388
|
+
# image = vision.image "path/to/text.png"
|
389
|
+
# text = image.text
|
390
|
+
#
|
391
|
+
# block = text.pages[0].blocks.first
|
392
|
+
#
|
393
|
+
# block.languages.first.code #=> "en"
|
394
|
+
# block.bounds.count #=> 4
|
395
|
+
# block.paragraphs.count #=> 1
|
396
|
+
#
|
397
|
+
class Block
|
398
|
+
##
|
399
|
+
# @private The EntityAnnotation GRPC object.
|
400
|
+
attr_accessor :grpc
|
401
|
+
|
402
|
+
##
|
403
|
+
# @private Creates a new Block instance.
|
404
|
+
def initialize
|
405
|
+
@grpc = nil
|
406
|
+
end
|
407
|
+
|
408
|
+
##
|
409
|
+
# A list of detected languages together with confidence.
|
410
|
+
#
|
411
|
+
# @return [Array<Language>]
|
412
|
+
#
|
413
|
+
def languages
|
414
|
+
detected_languages = @grpc.property.detected_languages
|
415
|
+
@languages ||= Array(detected_languages).map do |l|
|
416
|
+
Language.from_grpc l
|
417
|
+
end
|
418
|
+
end
|
419
|
+
|
420
|
+
##
|
421
|
+
# The type of a detected break at the start or end of the page.
|
422
|
+
#
|
423
|
+
# @return [::Symbol]
|
424
|
+
#
|
425
|
+
def break_type
|
426
|
+
@grpc.property.detected_break &&
|
427
|
+
@grpc.property.detected_break.type.to_sym
|
428
|
+
end
|
429
|
+
|
430
|
+
##
|
431
|
+
# True if a detected break prepends the page.
|
432
|
+
#
|
433
|
+
# @return [Boolean]
|
434
|
+
#
|
435
|
+
def prefix_break?
|
436
|
+
@grpc.property.detected_break &&
|
437
|
+
@grpc.property.detected_break.is_prefix
|
438
|
+
end
|
439
|
+
|
440
|
+
##
|
441
|
+
# The bounding box for the block.
|
442
|
+
# The vertices are in the order of top-left, top-right,
|
443
|
+
# bottom-right, bottom-left. When a rotation of the bounding box
|
444
|
+
# is detected the rotation is represented as around the top-left
|
445
|
+
# corner as defined when the text is read in the 'natural'
|
446
|
+
# orientation.
|
447
|
+
# For example:
|
448
|
+
# * when the text is horizontal it might look like:
|
449
|
+
# 0----1
|
450
|
+
# | |
|
451
|
+
# 3----2
|
452
|
+
# * when rotated 180 degrees around the top-left corner it
|
453
|
+
# becomes:
|
454
|
+
# 2----3
|
455
|
+
# | |
|
456
|
+
# 1----0
|
457
|
+
# and the vertice order will still be (0, 1, 2, 3).
|
458
|
+
#
|
459
|
+
# @return [Array<Vertex>]
|
460
|
+
#
|
461
|
+
def bounds
|
462
|
+
return [] unless @grpc.bounding_box
|
463
|
+
@bounds ||= Array(@grpc.bounding_box.vertices).map do |v|
|
464
|
+
Vertex.from_grpc v
|
465
|
+
end
|
466
|
+
end
|
467
|
+
|
468
|
+
##
|
469
|
+
# List of paragraphs in this block (if this block is of type
|
470
|
+
# text).
|
471
|
+
#
|
472
|
+
# @return [Array<Paragraph>]
|
473
|
+
#
|
474
|
+
def paragraphs
|
475
|
+
@paragraphs ||= Array(@grpc.paragraphs).map do |b|
|
476
|
+
Paragraph.from_grpc b
|
477
|
+
end
|
478
|
+
end
|
479
|
+
|
480
|
+
##
|
481
|
+
# Detected block type (text, image etc) for the block.
|
482
|
+
#
|
483
|
+
# @return [::Symbol]
|
484
|
+
#
|
485
|
+
def block_type
|
486
|
+
@grpc.block_type.to_sym
|
487
|
+
end
|
488
|
+
|
489
|
+
##
|
490
|
+
# Deeply converts object to a hash. All keys will be symbolized.
|
491
|
+
#
|
492
|
+
# @return [Hash]
|
493
|
+
#
|
494
|
+
def to_h
|
495
|
+
{ languages: languages.map(&:to_h), break_type: break_type,
|
496
|
+
prefix_break: prefix_break?, bounds: bounds.map(&:to_h),
|
497
|
+
paragraphs: paragraphs.map(&:to_h), block_type: block_type }
|
498
|
+
end
|
499
|
+
|
500
|
+
# @private
|
501
|
+
def to_s
|
502
|
+
tmplt = "languages: %s, break_type: %s, prefix_break: %s," \
|
503
|
+
" bounds: %i, paragraphs: %i, block_type: %s"
|
504
|
+
format tmplt, languages.inspect, break_type, prefix_break?,
|
505
|
+
bounds.count, paragraphs.count, block_type
|
506
|
+
end
|
507
|
+
|
508
|
+
# @private
|
509
|
+
def inspect
|
510
|
+
"#<#{self.class.name} #{self}>"
|
511
|
+
end
|
512
|
+
|
513
|
+
##
|
514
|
+
# @private New Annotation::Text::Page::Block from a GRPC
|
515
|
+
# object.
|
516
|
+
def self.from_grpc grpc
|
517
|
+
new.tap { |w| w.instance_variable_set :@grpc, grpc }
|
518
|
+
end
|
519
|
+
end
|
520
|
+
|
521
|
+
##
|
522
|
+
# # Paragraph
|
523
|
+
#
|
524
|
+
# Structural unit of text representing a number of words in certain
|
525
|
+
# order. See {Block}.
|
526
|
+
#
|
527
|
+
# @example
|
528
|
+
# require "google/cloud/vision"
|
529
|
+
#
|
530
|
+
# vision = Google::Cloud::Vision.new
|
531
|
+
#
|
532
|
+
# image = vision.image "path/to/text.png"
|
533
|
+
# text = image.text
|
534
|
+
#
|
535
|
+
# paragraph = text.pages[0].blocks[0].paragraphs.first
|
536
|
+
#
|
537
|
+
# paragraph.languages.first.code #=> "en"
|
538
|
+
# paragraph.bounds.count #=> 4
|
539
|
+
# paragraph.words.count #=> 10
|
540
|
+
#
|
541
|
+
class Paragraph
|
542
|
+
##
|
543
|
+
# @private The EntityAnnotation GRPC object.
|
544
|
+
attr_accessor :grpc
|
545
|
+
|
546
|
+
##
|
547
|
+
# @private Creates a new Paragraph instance.
|
548
|
+
def initialize
|
549
|
+
@grpc = nil
|
550
|
+
end
|
551
|
+
|
552
|
+
##
|
553
|
+
# A list of detected languages together with confidence.
|
554
|
+
#
|
555
|
+
# @return [Array<Language>]
|
556
|
+
#
|
557
|
+
def languages
|
558
|
+
detected_languages = @grpc.property.detected_languages
|
559
|
+
@languages ||= Array(detected_languages).map do |l|
|
560
|
+
Language.from_grpc l
|
561
|
+
end
|
562
|
+
end
|
563
|
+
|
564
|
+
##
|
565
|
+
# The type of a detected break at the start or end of the page.
|
566
|
+
#
|
567
|
+
# @return [::Symbol]
|
568
|
+
#
|
569
|
+
def break_type
|
570
|
+
@grpc.property.detected_break &&
|
571
|
+
@grpc.property.detected_break.type.to_sym
|
572
|
+
end
|
573
|
+
|
574
|
+
##
|
575
|
+
# True if a detected break prepends the page.
|
576
|
+
#
|
577
|
+
# @return [Boolean]
|
578
|
+
#
|
579
|
+
def prefix_break?
|
580
|
+
@grpc.property.detected_break &&
|
581
|
+
@grpc.property.detected_break.is_prefix
|
582
|
+
end
|
583
|
+
|
584
|
+
##
|
585
|
+
# The bounding box for the paragraph.
|
586
|
+
# The vertices are in the order of top-left, top-right,
|
587
|
+
# bottom-right, bottom-left. When a rotation of the bounding box
|
588
|
+
# is detected the rotation is represented as around the top-left
|
589
|
+
# corner as defined when the text is read in the 'natural'
|
590
|
+
# orientation.
|
591
|
+
# For example:
|
592
|
+
# * when the text is horizontal it might look like:
|
593
|
+
# 0----1
|
594
|
+
# | |
|
595
|
+
# 3----2
|
596
|
+
# * when rotated 180 degrees around the top-left corner it
|
597
|
+
# becomes:
|
598
|
+
# 2----3
|
599
|
+
# | |
|
600
|
+
# 1----0
|
601
|
+
# and the vertice order will still be (0, 1, 2, 3).
|
602
|
+
#
|
603
|
+
# @return [Array<Vertex>]
|
604
|
+
#
|
605
|
+
def bounds
|
606
|
+
return [] unless @grpc.bounding_box
|
607
|
+
@bounds ||= Array(@grpc.bounding_box.vertices).map do |v|
|
608
|
+
Vertex.from_grpc v
|
609
|
+
end
|
610
|
+
end
|
611
|
+
|
612
|
+
##
|
613
|
+
# List of words in this paragraph.
|
614
|
+
#
|
615
|
+
# @return [Array<Word>]
|
616
|
+
#
|
617
|
+
def words
|
618
|
+
@words ||= Array(@grpc.words).map do |b|
|
619
|
+
Word.from_grpc b
|
620
|
+
end
|
621
|
+
end
|
622
|
+
|
623
|
+
##
|
624
|
+
# Deeply converts object to a hash. All keys will be symbolized.
|
625
|
+
#
|
626
|
+
# @return [Hash]
|
627
|
+
#
|
628
|
+
def to_h
|
629
|
+
{ languages: languages.map(&:to_h), break_type: break_type,
|
630
|
+
prefix_break: prefix_break?, bounds: bounds.map(&:to_h),
|
631
|
+
words: words.map(&:to_h) }
|
632
|
+
end
|
633
|
+
|
634
|
+
# @private
|
635
|
+
def to_s
|
636
|
+
tmplt = "languages: %s, break_type: %s, prefix_break: %s," \
|
637
|
+
" bounds: %i, words: %i"
|
638
|
+
format tmplt, languages.inspect, break_type, prefix_break?,
|
639
|
+
bounds.count, words.count
|
640
|
+
end
|
641
|
+
|
642
|
+
# @private
|
643
|
+
def inspect
|
644
|
+
"#<#{self.class.name} #{self}>"
|
645
|
+
end
|
646
|
+
|
647
|
+
##
|
648
|
+
# @private New Annotation::Text::Page::Paragraph from a GRPC
|
649
|
+
# object.
|
650
|
+
def self.from_grpc grpc
|
651
|
+
new.tap { |w| w.instance_variable_set :@grpc, grpc }
|
652
|
+
end
|
653
|
+
end
|
654
|
+
|
655
|
+
##
|
656
|
+
# # Word
|
657
|
+
#
|
658
|
+
# A word representation. See {Paragraph}.
|
659
|
+
#
|
660
|
+
# @example
|
661
|
+
# require "google/cloud/vision"
|
662
|
+
#
|
663
|
+
# vision = Google::Cloud::Vision.new
|
664
|
+
#
|
665
|
+
# image = vision.image "path/to/text.png"
|
666
|
+
# text = image.text
|
667
|
+
#
|
668
|
+
# word = text.pages[0].blocks[0].paragraphs[0].words.first
|
669
|
+
#
|
670
|
+
# word.languages.first.code #=> "en"
|
671
|
+
# word.bounds.count #=> 4
|
672
|
+
# word.symbols.count #=> 6
|
673
|
+
#
|
674
|
+
class Word
|
675
|
+
##
|
676
|
+
# @private The EntityAnnotation GRPC object.
|
677
|
+
attr_accessor :grpc
|
678
|
+
|
679
|
+
##
|
680
|
+
# @private Creates a new Word instance.
|
681
|
+
def initialize
|
682
|
+
@grpc = nil
|
683
|
+
end
|
684
|
+
|
685
|
+
##
|
686
|
+
# A list of detected languages together with confidence.
|
687
|
+
#
|
688
|
+
# @return [Array<Language>]
|
689
|
+
#
|
690
|
+
def languages
|
691
|
+
detected_languages = @grpc.property.detected_languages
|
692
|
+
@languages ||= Array(detected_languages).map do |l|
|
693
|
+
Language.from_grpc l
|
694
|
+
end
|
695
|
+
end
|
696
|
+
|
697
|
+
##
|
698
|
+
# The type of a detected break at the start or end of the page.
|
699
|
+
#
|
700
|
+
# @return [::Symbol]
|
701
|
+
#
|
702
|
+
def break_type
|
703
|
+
@grpc.property.detected_break &&
|
704
|
+
@grpc.property.detected_break.type.to_sym
|
705
|
+
end
|
706
|
+
|
707
|
+
##
|
708
|
+
# True if a detected break prepends the page.
|
709
|
+
#
|
710
|
+
# @return [Boolean]
|
711
|
+
#
|
712
|
+
def prefix_break?
|
713
|
+
@grpc.property.detected_break &&
|
714
|
+
@grpc.property.detected_break.is_prefix
|
715
|
+
end
|
716
|
+
|
717
|
+
##
|
718
|
+
# The bounding box for the word.
|
719
|
+
# The vertices are in the order of top-left, top-right,
|
720
|
+
# bottom-right, bottom-left. When a rotation of the bounding box
|
721
|
+
# is detected the rotation is represented as around the top-left
|
722
|
+
# corner as defined when the text is read in the 'natural'
|
723
|
+
# orientation.
|
724
|
+
# For example:
|
725
|
+
# * when the text is horizontal it might look like:
|
726
|
+
# 0----1
|
727
|
+
# | |
|
728
|
+
# 3----2
|
729
|
+
# * when rotated 180 degrees around the top-left corner it
|
730
|
+
# becomes:
|
731
|
+
# 2----3
|
732
|
+
# | |
|
733
|
+
# 1----0
|
734
|
+
# and the vertice order will still be (0, 1, 2, 3).
|
735
|
+
#
|
736
|
+
# @return [Array<Vertex>]
|
737
|
+
#
|
738
|
+
def bounds
|
739
|
+
return [] unless @grpc.bounding_box
|
740
|
+
@bounds ||= Array(@grpc.bounding_box.vertices).map do |v|
|
741
|
+
Vertex.from_grpc v
|
742
|
+
end
|
743
|
+
end
|
744
|
+
|
745
|
+
##
|
746
|
+
# List of symbols in the word. The order of the symbols follows
|
747
|
+
# the natural reading order.
|
748
|
+
#
|
749
|
+
# @return [Array<Symbol>]
|
750
|
+
#
|
751
|
+
def symbols
|
752
|
+
@symbols ||= Array(@grpc.symbols).map do |b|
|
753
|
+
Symbol.from_grpc b
|
754
|
+
end
|
755
|
+
end
|
756
|
+
|
757
|
+
##
|
758
|
+
# Deeply converts object to a hash. All keys will be symbolized.
|
759
|
+
#
|
760
|
+
# @return [Hash]
|
761
|
+
#
|
762
|
+
def to_h
|
763
|
+
{ languages: languages.map(&:to_h), break_type: break_type,
|
764
|
+
prefix_break: prefix_break?, bounds: bounds.map(&:to_h),
|
765
|
+
symbols: symbols.map(&:to_h) }
|
766
|
+
end
|
767
|
+
|
768
|
+
# @private
|
769
|
+
def to_s
|
770
|
+
tmplt = "languages: %s, break_type: %s, prefix_break: %s," \
|
771
|
+
" bounds: %i, symbols: %i"
|
772
|
+
format tmplt, languages.inspect, break_type, prefix_break?,
|
773
|
+
bounds.count, symbols.count
|
774
|
+
end
|
775
|
+
|
776
|
+
# @private
|
777
|
+
def inspect
|
778
|
+
"#<#{self.class.name} #{self}>"
|
779
|
+
end
|
780
|
+
|
781
|
+
##
|
782
|
+
# @private New Annotation::Text::Page::Word from a GRPC
|
783
|
+
# object.
|
784
|
+
def self.from_grpc grpc
|
785
|
+
new.tap { |w| w.instance_variable_set :@grpc, grpc }
|
786
|
+
end
|
787
|
+
end
|
788
|
+
|
789
|
+
##
|
790
|
+
# # Symbol
|
791
|
+
#
|
792
|
+
# A word representation. See {Paragraph}.
|
793
|
+
#
|
794
|
+
# @example
|
795
|
+
# require "google/cloud/vision"
|
796
|
+
#
|
797
|
+
# vision = Google::Cloud::Vision.new
|
798
|
+
#
|
799
|
+
# image = vision.image "path/to/text.png"
|
800
|
+
# text = image.text
|
801
|
+
# page = text.pages.first
|
802
|
+
#
|
803
|
+
# symbol = page.blocks[0].paragraphs[0].words[0].symbols[0]
|
804
|
+
#
|
805
|
+
# symbol.languages.first.code #=> "en"
|
806
|
+
# symbol.bounds.count #=> 4
|
807
|
+
# symbol.text #=> "G"
|
808
|
+
#
|
809
|
+
class Symbol
|
810
|
+
##
|
811
|
+
# @private The EntityAnnotation GRPC object.
|
812
|
+
attr_accessor :grpc
|
813
|
+
|
814
|
+
##
|
815
|
+
# @private Creates a new Symbol instance.
|
816
|
+
def initialize
|
817
|
+
@grpc = nil
|
818
|
+
end
|
819
|
+
|
820
|
+
##
|
821
|
+
# A list of detected languages together with confidence.
|
822
|
+
#
|
823
|
+
# @return [Array<Language>]
|
824
|
+
#
|
825
|
+
def languages
|
826
|
+
detected_languages = @grpc.property.detected_languages
|
827
|
+
@languages ||= Array(detected_languages).map do |l|
|
828
|
+
Language.from_grpc l
|
829
|
+
end
|
830
|
+
end
|
831
|
+
|
832
|
+
##
|
833
|
+
# The type of a detected break at the start or end of the page.
|
834
|
+
#
|
835
|
+
# @return [::Symbol]
|
836
|
+
#
|
837
|
+
def break_type
|
838
|
+
@grpc.property.detected_break &&
|
839
|
+
@grpc.property.detected_break.type.to_sym
|
840
|
+
end
|
841
|
+
|
842
|
+
##
|
843
|
+
# True if a detected break prepends the page.
|
844
|
+
#
|
845
|
+
# @return [Boolean]
|
846
|
+
#
|
847
|
+
def prefix_break?
|
848
|
+
@grpc.property.detected_break &&
|
849
|
+
@grpc.property.detected_break.is_prefix
|
850
|
+
end
|
851
|
+
|
852
|
+
##
|
853
|
+
# The bounding box for the symbol.
|
854
|
+
# The vertices are in the order of top-left, top-right,
|
855
|
+
# bottom-right, bottom-left. When a rotation of the bounding box
|
856
|
+
# is detected the rotation is represented as around the top-left
|
857
|
+
# corner as defined when the text is read in the 'natural'
|
858
|
+
# orientation.
|
859
|
+
# For example:
|
860
|
+
# * when the text is horizontal it might look like:
|
861
|
+
# 0----1
|
862
|
+
# | |
|
863
|
+
# 3----2
|
864
|
+
# * when rotated 180 degrees around the top-left corner it
|
865
|
+
# becomes:
|
866
|
+
# 2----3
|
867
|
+
# | |
|
868
|
+
# 1----0
|
869
|
+
# and the vertice order will still be (0, 1, 2, 3).
|
870
|
+
#
|
871
|
+
# @return [Array<Vertex>]
|
872
|
+
#
|
873
|
+
def bounds
|
874
|
+
return [] unless @grpc.bounding_box
|
875
|
+
@bounds ||= Array(@grpc.bounding_box.vertices).map do |v|
|
876
|
+
Vertex.from_grpc v
|
877
|
+
end
|
878
|
+
end
|
879
|
+
|
880
|
+
##
|
881
|
+
# The actual UTF-8 representation of the symbol.
|
882
|
+
#
|
883
|
+
# @return [String]
|
884
|
+
#
|
885
|
+
def text
|
886
|
+
@grpc.text
|
887
|
+
end
|
888
|
+
|
889
|
+
##
|
890
|
+
# Deeply converts object to a hash. All keys will be symbolized.
|
891
|
+
#
|
892
|
+
# @return [Hash]
|
893
|
+
#
|
894
|
+
def to_h
|
895
|
+
{ languages: languages.map(&:to_h), break_type: break_type,
|
896
|
+
prefix_break: prefix_break?, bounds: bounds.map(&:to_h),
|
897
|
+
text: text }
|
898
|
+
end
|
899
|
+
|
900
|
+
# @private
|
901
|
+
def to_s
|
902
|
+
tmplt = "languages: %s, break_type: %s, prefix_break: %s," \
|
903
|
+
" bounds: %i, text: %s"
|
904
|
+
format tmplt, languages.inspect, break_type, prefix_break?,
|
905
|
+
bounds.count, text
|
906
|
+
end
|
907
|
+
|
908
|
+
# @private
|
909
|
+
def inspect
|
910
|
+
"#<#{self.class.name} #{self}>"
|
911
|
+
end
|
912
|
+
|
913
|
+
##
|
914
|
+
# @private New Annotation::Text::Page::Symbol from a GRPC
|
915
|
+
# object.
|
916
|
+
def self.from_grpc grpc
|
917
|
+
new.tap { |w| w.instance_variable_set :@grpc, grpc }
|
918
|
+
end
|
919
|
+
end
|
920
|
+
|
921
|
+
##
|
922
|
+
# # Language
|
923
|
+
#
|
924
|
+
# A language within a detected text (OCR). See {Text#pages}.
|
925
|
+
#
|
926
|
+
# @example
|
927
|
+
# require "google/cloud/vision"
|
928
|
+
#
|
929
|
+
# vision = Google::Cloud::Vision.new
|
930
|
+
#
|
931
|
+
# image = vision.image "path/to/text.png"
|
932
|
+
# text = image.text
|
933
|
+
# page = text.pages.first
|
934
|
+
#
|
935
|
+
# language = page.languages.first
|
936
|
+
# language.code #=> "en"
|
937
|
+
#
|
938
|
+
class Language
|
939
|
+
##
|
940
|
+
# @private The EntityAnnotation GRPC object.
|
941
|
+
attr_accessor :grpc
|
942
|
+
|
943
|
+
##
|
944
|
+
# @private Creates a new Language instance.
|
945
|
+
def initialize
|
946
|
+
@grpc = nil
|
947
|
+
end
|
948
|
+
|
949
|
+
##
|
950
|
+
# The language code detected for a structural component.
|
951
|
+
#
|
952
|
+
# @return [String] The [ISO
|
953
|
+
# 639-1](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes)
|
954
|
+
# language code.
|
955
|
+
#
|
956
|
+
def code
|
957
|
+
@grpc.language_code
|
958
|
+
end
|
959
|
+
|
960
|
+
##
|
961
|
+
# Confidence of detected language.
|
962
|
+
#
|
963
|
+
# @return [Float] A value in the range [0,1].
|
964
|
+
#
|
965
|
+
def confidence
|
966
|
+
@grpc.confidence
|
967
|
+
end
|
968
|
+
|
969
|
+
##
|
970
|
+
# Deeply converts object to a hash. All keys will be symbolized.
|
971
|
+
#
|
972
|
+
# @return [Hash]
|
973
|
+
#
|
974
|
+
def to_h
|
975
|
+
{ code: code, confidence: confidence }
|
976
|
+
end
|
977
|
+
|
978
|
+
# @private
|
979
|
+
def to_s
|
980
|
+
format "code: %s, confidence: %s", code, confidence
|
981
|
+
end
|
982
|
+
|
983
|
+
# @private
|
984
|
+
def inspect
|
985
|
+
"#<#{self.class.name} #{self}>"
|
986
|
+
end
|
987
|
+
|
988
|
+
##
|
989
|
+
# @private New Annotation::Text::Page::Language from a GRPC
|
990
|
+
# object.
|
991
|
+
def self.from_grpc grpc
|
992
|
+
new.tap { |w| w.instance_variable_set :@grpc, grpc }
|
993
|
+
end
|
994
|
+
end
|
995
|
+
end
|
218
996
|
end
|
219
997
|
end
|
220
998
|
end
|