mindee 2.2.1 → 3.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.gitignore +1 -0
- data/.rubocop.yml +2 -0
- data/.yardopts +1 -0
- data/CHANGELOG.md +40 -0
- data/README.md +29 -16
- data/bin/mindee.rb +23 -26
- data/docs/code_samples/bank_account_details_v1.txt +10 -5
- data/docs/code_samples/bank_account_details_v2.txt +19 -0
- data/docs/code_samples/bank_check_v1.txt +10 -5
- data/docs/code_samples/carte_vitale_v1.txt +10 -5
- data/docs/code_samples/custom_v1.txt +19 -10
- data/docs/code_samples/default.txt +10 -2
- data/docs/code_samples/expense_receipts_v4.txt +10 -5
- data/docs/code_samples/expense_receipts_v5.txt +11 -6
- data/docs/code_samples/financial_document_v1.txt +10 -5
- data/docs/code_samples/idcard_fr_v1.txt +10 -5
- data/docs/code_samples/invoice_splitter_v1_async.txt +66 -0
- data/docs/code_samples/invoices_v4.txt +10 -5
- data/docs/code_samples/license_plates_v1.txt +10 -5
- data/docs/code_samples/passport_v1.txt +10 -5
- data/docs/code_samples/proof_of_address_v1.txt +10 -5
- data/docs/code_samples/us_driver_license_v1.txt +19 -0
- data/docs/ruby-api-builder.md +30 -31
- data/docs/ruby-getting-started.md +64 -23
- data/docs/ruby-invoice-ocr.md +70 -59
- data/docs/ruby-passport-ocr.md +49 -40
- data/docs/ruby-receipt-ocr.md +45 -32
- data/lib/mindee/client.rb +150 -148
- data/lib/mindee/geometry/min_max.rb +23 -0
- data/lib/mindee/geometry/point.rb +35 -0
- data/lib/mindee/geometry/polygon.rb +23 -0
- data/lib/mindee/geometry/quadrilateral.rb +45 -0
- data/lib/mindee/geometry/utils.rb +81 -0
- data/lib/mindee/geometry.rb +5 -116
- data/lib/mindee/http/endpoint.rb +115 -16
- data/lib/mindee/http.rb +3 -0
- data/lib/mindee/input/sources.rb +90 -73
- data/lib/mindee/parsing/common/api_response.rb +109 -0
- data/lib/mindee/parsing/common/document.rb +48 -0
- data/lib/mindee/parsing/common/error.rb +24 -0
- data/lib/mindee/parsing/common/inference.rb +43 -0
- data/lib/mindee/parsing/common/ocr/mvision_v1.rb +34 -0
- data/lib/mindee/parsing/common/ocr/ocr.rb +169 -0
- data/lib/mindee/parsing/common/ocr.rb +3 -0
- data/lib/mindee/parsing/common/orientation.rb +26 -0
- data/lib/mindee/parsing/common/page.rb +40 -0
- data/lib/mindee/parsing/common/prediction.rb +15 -0
- data/lib/mindee/parsing/common/product.rb +19 -0
- data/lib/mindee/parsing/common.rb +10 -0
- data/lib/mindee/parsing/custom/classification_field.rb +28 -0
- data/lib/mindee/parsing/custom/list_field.rb +76 -0
- data/lib/mindee/parsing/custom.rb +4 -0
- data/lib/mindee/parsing/standard/amount_field.rb +26 -0
- data/lib/mindee/parsing/standard/base_field.rb +104 -0
- data/lib/mindee/parsing/standard/classification_field.rb +16 -0
- data/lib/mindee/parsing/standard/company_registration_field.rb +21 -0
- data/lib/mindee/parsing/standard/date_field.rb +34 -0
- data/lib/mindee/parsing/standard/locale_field.rb +50 -0
- data/lib/mindee/parsing/standard/payment_details_field.rb +42 -0
- data/lib/mindee/parsing/standard/position_field.rb +47 -0
- data/lib/mindee/parsing/standard/tax_field.rb +108 -0
- data/lib/mindee/parsing/standard/text_field.rb +16 -0
- data/lib/mindee/parsing/standard.rb +12 -0
- data/lib/mindee/parsing.rb +3 -2
- data/lib/mindee/{input → pdf}/pdf_processing.rb +4 -32
- data/lib/mindee/pdf/pdf_tools.rb +34 -0
- data/lib/mindee/pdf.rb +3 -0
- data/lib/mindee/product/.rubocop.yml +5 -0
- data/lib/mindee/product/custom/custom_v1.rb +35 -0
- data/lib/mindee/product/custom/custom_v1_document.rb +60 -0
- data/lib/mindee/product/custom/custom_v1_page.rb +32 -0
- data/lib/mindee/product/eu/license_plate/license_plate_v1.rb +38 -0
- data/lib/mindee/product/eu/license_plate/license_plate_v1_document.rb +37 -0
- data/lib/mindee/product/eu/license_plate/license_plate_v1_page.rb +34 -0
- data/lib/mindee/product/financial_document/financial_document_v1.rb +36 -0
- data/lib/mindee/product/financial_document/financial_document_v1_document.rb +202 -0
- data/lib/mindee/product/financial_document/financial_document_v1_line_item.rb +90 -0
- data/lib/mindee/product/financial_document/financial_document_v1_page.rb +32 -0
- data/lib/mindee/product/fr/bank_account_details/bank_account_details_v1.rb +38 -0
- data/lib/mindee/product/fr/bank_account_details/bank_account_details_v1_document.rb +43 -0
- data/lib/mindee/product/fr/bank_account_details/bank_account_details_v1_page.rb +34 -0
- data/lib/mindee/product/fr/bank_account_details/bank_account_details_v2.rb +38 -0
- data/lib/mindee/product/fr/bank_account_details/bank_account_details_v2_bban.rb +71 -0
- data/lib/mindee/product/fr/bank_account_details/bank_account_details_v2_document.rb +51 -0
- data/lib/mindee/product/fr/bank_account_details/bank_account_details_v2_page.rb +34 -0
- data/lib/mindee/product/fr/carte_vitale/carte_vitale_v1.rb +38 -0
- data/lib/mindee/product/fr/carte_vitale/carte_vitale_v1_document.rb +52 -0
- data/lib/mindee/product/fr/carte_vitale/carte_vitale_v1_page.rb +34 -0
- data/lib/mindee/product/fr/id_card/id_card_v1.rb +38 -0
- data/lib/mindee/product/fr/id_card/id_card_v1_document.rb +82 -0
- data/lib/mindee/product/fr/id_card/id_card_v1_page.rb +48 -0
- data/lib/mindee/product/invoice/invoice_v4.rb +37 -0
- data/lib/mindee/product/invoice/invoice_v4_document.rb +212 -0
- data/lib/mindee/product/invoice/invoice_v4_line_item.rb +66 -0
- data/lib/mindee/product/invoice/invoice_v4_page.rb +32 -0
- data/lib/mindee/product/invoice_splitter/invoice_splitter_v1.rb +36 -0
- data/lib/mindee/product/invoice_splitter/invoice_splitter_v1_document.rb +65 -0
- data/lib/mindee/product/invoice_splitter/invoice_splitter_v1_page.rb +32 -0
- data/lib/mindee/product/passport/passport_v1.rb +36 -0
- data/lib/mindee/{parsing/prediction/fr/id_card/id_card_v1.rb → product/passport/passport_v1_document.rb} +45 -45
- data/lib/mindee/product/passport/passport_v1_page.rb +32 -0
- data/lib/mindee/product/proof_of_address/proof_of_address_v1.rb +36 -0
- data/lib/mindee/product/proof_of_address/proof_of_address_v1_document.rb +83 -0
- data/lib/mindee/product/proof_of_address/proof_of_address_v1_page.rb +32 -0
- data/lib/mindee/product/receipt/receipt_v4.rb +36 -0
- data/lib/mindee/product/receipt/receipt_v4_document.rb +86 -0
- data/lib/mindee/product/receipt/receipt_v4_page.rb +32 -0
- data/lib/mindee/product/receipt/receipt_v5.rb +36 -0
- data/lib/mindee/product/receipt/receipt_v5_document.rb +149 -0
- data/lib/mindee/product/receipt/receipt_v5_line_item.rb +69 -0
- data/lib/mindee/product/receipt/receipt_v5_page.rb +32 -0
- data/lib/mindee/product/us/bank_check/bank_check_v1.rb +38 -0
- data/lib/mindee/product/us/bank_check/bank_check_v1_document.rb +62 -0
- data/lib/mindee/product/us/bank_check/bank_check_v1_page.rb +57 -0
- data/lib/mindee/product/us/driver_license/driver_license_v1.rb +38 -0
- data/lib/mindee/product/us/driver_license/driver_license_v1_document.rb +113 -0
- data/lib/mindee/product/us/driver_license/driver_license_v1_page.rb +53 -0
- data/lib/mindee/product.rb +17 -0
- data/lib/mindee/version.rb +2 -1
- data/lib/mindee.rb +3 -1
- metadata +91 -38
- data/docs/code_samples/shipping_containers_v1.txt +0 -14
- data/lib/mindee/document_config.rb +0 -60
- data/lib/mindee/parsing/document.rb +0 -31
- data/lib/mindee/parsing/error.rb +0 -22
- data/lib/mindee/parsing/inference.rb +0 -53
- data/lib/mindee/parsing/page.rb +0 -46
- data/lib/mindee/parsing/prediction/base.rb +0 -30
- data/lib/mindee/parsing/prediction/common_fields/amount.rb +0 -21
- data/lib/mindee/parsing/prediction/common_fields/base.rb +0 -72
- data/lib/mindee/parsing/prediction/common_fields/company_registration.rb +0 -17
- data/lib/mindee/parsing/prediction/common_fields/date.rb +0 -30
- data/lib/mindee/parsing/prediction/common_fields/locale.rb +0 -45
- data/lib/mindee/parsing/prediction/common_fields/payment_details.rb +0 -33
- data/lib/mindee/parsing/prediction/common_fields/position.rb +0 -39
- data/lib/mindee/parsing/prediction/common_fields/tax.rb +0 -44
- data/lib/mindee/parsing/prediction/common_fields/text.rb +0 -12
- data/lib/mindee/parsing/prediction/common_fields.rb +0 -11
- data/lib/mindee/parsing/prediction/custom/custom_v1.rb +0 -58
- data/lib/mindee/parsing/prediction/custom/fields.rb +0 -91
- data/lib/mindee/parsing/prediction/eu/license_plate/license_plate_v1.rb +0 -34
- data/lib/mindee/parsing/prediction/financial_document/financial_document_v1.rb +0 -237
- data/lib/mindee/parsing/prediction/financial_document/financial_document_v1_line_item.rb +0 -58
- data/lib/mindee/parsing/prediction/fr/bank_account_details/bank_account_details_v1.rb +0 -40
- data/lib/mindee/parsing/prediction/fr/carte_vitale/carte_vitale_v1.rb +0 -49
- data/lib/mindee/parsing/prediction/invoice/invoice_v4.rb +0 -212
- data/lib/mindee/parsing/prediction/invoice/invoice_v4_line_item.rb +0 -58
- data/lib/mindee/parsing/prediction/passport/passport_v1.rb +0 -121
- data/lib/mindee/parsing/prediction/proof_of_address/proof_of_address_v1.rb +0 -80
- data/lib/mindee/parsing/prediction/receipt/receipt_v4.rb +0 -87
- data/lib/mindee/parsing/prediction/receipt/receipt_v5.rb +0 -136
- data/lib/mindee/parsing/prediction/receipt/receipt_v5_line_item.rb +0 -37
- data/lib/mindee/parsing/prediction/shipping_container/shipping_container_v1.rb +0 -38
- data/lib/mindee/parsing/prediction/us/bank_check/bank_check_v1.rb +0 -70
- data/lib/mindee/parsing/prediction.rb +0 -15
@@ -0,0 +1,50 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module Mindee
|
4
|
+
module Parsing
|
5
|
+
module Standard
|
6
|
+
# Represents locale information
|
7
|
+
class Locale
|
8
|
+
# The confidence score, value will be between 0.0 and 1.0
|
9
|
+
# @return [Float]
|
10
|
+
attr_reader :confidence
|
11
|
+
# Language code in ISO 639-1 format.
|
12
|
+
# @return [String]
|
13
|
+
attr_reader :language
|
14
|
+
# Country code in ISO 3166-1 alpha-2 format.
|
15
|
+
# @return [String, nil]
|
16
|
+
attr_reader :country
|
17
|
+
# Currency code in ISO 4217 format.
|
18
|
+
# @return [String]
|
19
|
+
attr_reader :currency
|
20
|
+
# Language code, with country code when available.
|
21
|
+
# @return [String]
|
22
|
+
attr_reader :value
|
23
|
+
|
24
|
+
# @param prediction [Hash]
|
25
|
+
def initialize(prediction, _page_id = nil)
|
26
|
+
value_key = if prediction.include? 'value'
|
27
|
+
'value'
|
28
|
+
else
|
29
|
+
'language'
|
30
|
+
end
|
31
|
+
@confidence = prediction['confidence']
|
32
|
+
@value = prediction[value_key]
|
33
|
+
@language = prediction['language']
|
34
|
+
@country = prediction['country']
|
35
|
+
@currency = prediction['currency']
|
36
|
+
end
|
37
|
+
|
38
|
+
# @return [String]
|
39
|
+
def to_s
|
40
|
+
out_str = String.new
|
41
|
+
out_str << "#{@value}; " if @value
|
42
|
+
out_str << "#{@language}; " if @language
|
43
|
+
out_str << "#{@country}; " if @country
|
44
|
+
out_str << "#{@currency}; " if @currency
|
45
|
+
out_str.strip
|
46
|
+
end
|
47
|
+
end
|
48
|
+
end
|
49
|
+
end
|
50
|
+
end
|
@@ -0,0 +1,42 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative 'base_field'
|
4
|
+
|
5
|
+
module Mindee
|
6
|
+
module Parsing
|
7
|
+
module Standard
|
8
|
+
# Represents payment details for invoices and receipts
|
9
|
+
class PaymentDetails < Field
|
10
|
+
# @return [String, nil]
|
11
|
+
attr_reader :account_number
|
12
|
+
# @return [String, nil]
|
13
|
+
attr_reader :iban
|
14
|
+
# @return [String, nil]
|
15
|
+
attr_reader :routing_number
|
16
|
+
# @return [String, nil]
|
17
|
+
attr_reader :swift
|
18
|
+
|
19
|
+
# @param prediction [Hash]
|
20
|
+
# @param page_id [Integer, nil]
|
21
|
+
# @param reconstructed [Boolean]
|
22
|
+
def initialize(prediction, page_id, reconstructed: false)
|
23
|
+
super
|
24
|
+
@account_number = prediction['account_number']
|
25
|
+
@iban = prediction['iban']
|
26
|
+
@routing_number = prediction['routing_number']
|
27
|
+
@swift = prediction['swift']
|
28
|
+
end
|
29
|
+
|
30
|
+
# @return [String]
|
31
|
+
def to_s
|
32
|
+
out_str = String.new
|
33
|
+
out_str << "#{@account_number}; " if @account_number
|
34
|
+
out_str << "#{@iban}; " if @iban
|
35
|
+
out_str << "#{@routing_number}; " if @routing_number
|
36
|
+
out_str << "#{@swift}; " if @swift
|
37
|
+
out_str.strip
|
38
|
+
end
|
39
|
+
end
|
40
|
+
end
|
41
|
+
end
|
42
|
+
end
|
@@ -0,0 +1,47 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module Mindee
|
4
|
+
module Parsing
|
5
|
+
module Standard
|
6
|
+
# An element's position on the image
|
7
|
+
class PositionField
|
8
|
+
# @return [Mindee::Geometry::Polygon]
|
9
|
+
attr_reader :polygon
|
10
|
+
# @return [Mindee::Geometry::Polygon]
|
11
|
+
attr_reader :value
|
12
|
+
# @return [Mindee::Geometry::Quadrilateral]
|
13
|
+
attr_reader :quadrangle
|
14
|
+
# @return [Mindee::Geometry::Quadrilateral]
|
15
|
+
attr_reader :rectangle
|
16
|
+
# @return [Mindee::Geometry::Quadrilateral]
|
17
|
+
attr_reader :bounding_box
|
18
|
+
|
19
|
+
# @param prediction [Hash]
|
20
|
+
# @param page_id [Integer, nil]
|
21
|
+
def initialize(prediction, page_id)
|
22
|
+
unless prediction['polygon'].nil? || prediction['polygon'].empty?
|
23
|
+
@polygon = Geometry.polygon_from_prediction(prediction['polygon'])
|
24
|
+
end
|
25
|
+
@quadrangle = to_quadrilateral(prediction, 'quadrangle')
|
26
|
+
@rectangle = to_quadrilateral(prediction, 'rectangle')
|
27
|
+
@bounding_box = to_quadrilateral(prediction, 'bounding_box')
|
28
|
+
@page_id = page_id || prediction['page_id']
|
29
|
+
@value = @polygon
|
30
|
+
end
|
31
|
+
|
32
|
+
# @return [String]
|
33
|
+
def to_s
|
34
|
+
out_str = String.new
|
35
|
+
out_str << "Polygon with #{@polygon.size} points." if @polygon
|
36
|
+
out_str
|
37
|
+
end
|
38
|
+
|
39
|
+
private
|
40
|
+
|
41
|
+
def to_quadrilateral(prediction, key)
|
42
|
+
Geometry.quadrilateral_from_prediction(prediction[key]) unless prediction[key].nil? || prediction[key].empty?
|
43
|
+
end
|
44
|
+
end
|
45
|
+
end
|
46
|
+
end
|
47
|
+
end
|
@@ -0,0 +1,108 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative 'base_field'
|
4
|
+
|
5
|
+
module Mindee
|
6
|
+
module Parsing
|
7
|
+
module Standard
|
8
|
+
# Represents tax information.
|
9
|
+
class TaxField < Field
|
10
|
+
# Tax value as 3 decimal float
|
11
|
+
# @return [Float, nil]
|
12
|
+
attr_reader :value
|
13
|
+
# Tax rate percentage
|
14
|
+
# @return [Float]
|
15
|
+
attr_reader :rate
|
16
|
+
# Tax code
|
17
|
+
# @return [String]
|
18
|
+
attr_reader :code
|
19
|
+
# Tax base
|
20
|
+
# @return [Float]
|
21
|
+
attr_reader :base
|
22
|
+
|
23
|
+
# @param prediction [Hash]
|
24
|
+
# @param page_id [Integer, nil]
|
25
|
+
def initialize(prediction, page_id)
|
26
|
+
super
|
27
|
+
@value = prediction['value']&.round(3)
|
28
|
+
@rate = prediction['rate'].to_f unless prediction['rate'].nil?
|
29
|
+
@base = prediction['base'].to_f unless prediction['base'].nil?
|
30
|
+
@code = prediction['code'] unless prediction['code'] == 'None'
|
31
|
+
end
|
32
|
+
|
33
|
+
# @param value [Float]
|
34
|
+
def print_float(value)
|
35
|
+
format('%.2f', value)
|
36
|
+
end
|
37
|
+
|
38
|
+
# @return [String]
|
39
|
+
def to_s
|
40
|
+
printable = printable_values
|
41
|
+
out_str = String.new
|
42
|
+
out_str << ("Base: #{printable[:base]}")
|
43
|
+
out_str << (", Code: #{printable[:code]}")
|
44
|
+
out_str << (", Rate (%): #{printable[:rate]}")
|
45
|
+
out_str << (", Amount: #{printable[:value]}")
|
46
|
+
out_str.strip
|
47
|
+
end
|
48
|
+
|
49
|
+
# @return [Hash]
|
50
|
+
def printable_values
|
51
|
+
out_h = {}
|
52
|
+
out_h[:code] = @code.nil? ? '' : @code
|
53
|
+
out_h[:base] = @base.nil? ? '' : print_float(@base)
|
54
|
+
out_h[:rate] = @rate.nil? ? '' : print_float(@rate).to_s
|
55
|
+
out_h[:value] = @value.nil? ? '' : print_float(@value).to_s
|
56
|
+
out_h
|
57
|
+
end
|
58
|
+
|
59
|
+
# @return [String]
|
60
|
+
def to_table_line
|
61
|
+
printable = printable_values
|
62
|
+
out_str = String.new
|
63
|
+
out_str << ("| #{printable[:base].ljust(13, ' ')}")
|
64
|
+
out_str << (" | #{printable[:code].ljust(6, ' ')}")
|
65
|
+
out_str << (" | #{printable[:rate].ljust(8, ' ')}")
|
66
|
+
out_str << (" | #{printable[:value].ljust(13, ' ')} |")
|
67
|
+
out_str.strip
|
68
|
+
end
|
69
|
+
end
|
70
|
+
|
71
|
+
# Represents tax information, grouped as an array.
|
72
|
+
class Taxes < Array
|
73
|
+
# @param prediction [Hash]
|
74
|
+
# @param page_id [Integer, nil]
|
75
|
+
def initialize(prediction, page_id)
|
76
|
+
super(prediction.map { |entry| TaxField.new(entry, page_id) })
|
77
|
+
end
|
78
|
+
|
79
|
+
# @param char [String]
|
80
|
+
# @return [String]
|
81
|
+
def line_separator(char)
|
82
|
+
out_str = String.new
|
83
|
+
out_str << ' '
|
84
|
+
out_str << "+#{char * 15}"
|
85
|
+
out_str << "+#{char * 8}"
|
86
|
+
out_str << "+#{char * 10}"
|
87
|
+
out_str << "+#{char * 15}"
|
88
|
+
out_str << '+'
|
89
|
+
out_str
|
90
|
+
end
|
91
|
+
|
92
|
+
# @return [String]
|
93
|
+
def to_s
|
94
|
+
return '' if nil? || empty?
|
95
|
+
|
96
|
+
out_str = String.new
|
97
|
+
out_str << ("\n#{line_separator('-')}")
|
98
|
+
out_str << "\n | Base | Code | Rate (%) | Amount |"
|
99
|
+
out_str << "\n#{line_separator('=')}"
|
100
|
+
each do |entry|
|
101
|
+
out_str << "\n #{entry.to_table_line}\n#{line_separator('-')}"
|
102
|
+
end
|
103
|
+
out_str
|
104
|
+
end
|
105
|
+
end
|
106
|
+
end
|
107
|
+
end
|
108
|
+
end
|
@@ -0,0 +1,16 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative 'base_field'
|
4
|
+
|
5
|
+
module Mindee
|
6
|
+
module Parsing
|
7
|
+
module Standard
|
8
|
+
# Represents basic text information.
|
9
|
+
class TextField < Field
|
10
|
+
# Value as String
|
11
|
+
# @return [String, nil]
|
12
|
+
attr_reader :value
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
@@ -0,0 +1,12 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative 'standard/amount_field'
|
4
|
+
require_relative 'standard/base_field'
|
5
|
+
require_relative 'standard/classification_field'
|
6
|
+
require_relative 'standard/company_registration_field'
|
7
|
+
require_relative 'standard/date_field'
|
8
|
+
require_relative 'standard/payment_details_field'
|
9
|
+
require_relative 'standard/position_field'
|
10
|
+
require_relative 'standard/tax_field'
|
11
|
+
require_relative 'standard/text_field'
|
12
|
+
require_relative 'standard/locale_field'
|
data/lib/mindee/parsing.rb
CHANGED
@@ -2,41 +2,13 @@
|
|
2
2
|
|
3
3
|
require 'set'
|
4
4
|
require 'origami'
|
5
|
-
|
6
|
-
# Monkey-patching for Origami
|
7
|
-
module PDFTools
|
8
|
-
def to_io_stream(params = {})
|
9
|
-
options = {
|
10
|
-
delinearize: true,
|
11
|
-
recompile: true,
|
12
|
-
decrypt: false,
|
13
|
-
}
|
14
|
-
options.update(params)
|
15
|
-
|
16
|
-
if frozen? # incompatible flags with frozen doc (signed)
|
17
|
-
options[:recompile] = nil
|
18
|
-
options[:rebuild_xrefs] = nil
|
19
|
-
options[:noindent] = nil
|
20
|
-
options[:obfuscate] = false
|
21
|
-
end
|
22
|
-
load_all_objects unless @loaded
|
23
|
-
|
24
|
-
intents_as_pdfa1 if options[:intent] =~ %r{pdf[/-]?A1?/i}
|
25
|
-
delinearize! if options[:delinearize] && linearized?
|
26
|
-
compile(options) if options[:recompile]
|
27
|
-
|
28
|
-
io_stream = StringIO.new(output(options))
|
29
|
-
io_stream.set_encoding Encoding::BINARY
|
30
|
-
io_stream
|
31
|
-
end
|
32
|
-
end
|
33
|
-
|
34
|
-
Origami::PDF.class_eval { include PDFTools }
|
5
|
+
require_relative 'pdf_tools'
|
35
6
|
|
36
7
|
module Mindee
|
37
|
-
module
|
38
|
-
#
|
8
|
+
module PDF
|
9
|
+
# Module for PDF document handling
|
39
10
|
module PdfProcessor
|
11
|
+
Origami::PDF.class_eval { include PDFTools }
|
40
12
|
DEFAULT_OPTIONS = {
|
41
13
|
page_indexes: [0],
|
42
14
|
operation: :KEEP_ONLY,
|
@@ -0,0 +1,34 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module Mindee
|
4
|
+
module PDF
|
5
|
+
# Monkey-patching for Origami
|
6
|
+
module PDFTools
|
7
|
+
# @return [StringIO]
|
8
|
+
def to_io_stream(params = {})
|
9
|
+
options = {
|
10
|
+
delinearize: true,
|
11
|
+
recompile: true,
|
12
|
+
decrypt: false,
|
13
|
+
}
|
14
|
+
options.update(params)
|
15
|
+
|
16
|
+
if frozen? # incompatible flags with frozen doc (signed)
|
17
|
+
options[:recompile] = nil
|
18
|
+
options[:rebuild_xrefs] = nil
|
19
|
+
options[:noindent] = nil
|
20
|
+
options[:obfuscate] = false
|
21
|
+
end
|
22
|
+
load_all_objects unless @loaded
|
23
|
+
|
24
|
+
intents_as_pdfa1 if options[:intent] =~ %r{pdf[/-]?A1?/i}
|
25
|
+
delinearize! if options[:delinearize] && linearized?
|
26
|
+
compile(options) if options[:recompile]
|
27
|
+
|
28
|
+
io_stream = StringIO.new(output(options))
|
29
|
+
io_stream.set_encoding Encoding::BINARY
|
30
|
+
io_stream
|
31
|
+
end
|
32
|
+
end
|
33
|
+
end
|
34
|
+
end
|
data/lib/mindee/pdf.rb
ADDED
@@ -0,0 +1,35 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative 'custom_v1_document'
|
4
|
+
require_relative 'custom_v1_page'
|
5
|
+
|
6
|
+
module Mindee
|
7
|
+
module Product
|
8
|
+
module Custom
|
9
|
+
# Custom Document V1 prediction inference.
|
10
|
+
class CustomV1 < Mindee::Parsing::Common::Inference
|
11
|
+
@endpoint_name = ''
|
12
|
+
@endpoint_version = ''
|
13
|
+
|
14
|
+
# @param prediction [Hash]
|
15
|
+
def initialize(prediction)
|
16
|
+
super
|
17
|
+
@prediction = CustomV1Document.new(prediction['prediction'], nil)
|
18
|
+
@pages = []
|
19
|
+
prediction['pages'].each do |page|
|
20
|
+
@pages.push(CustomV1Page.new(page))
|
21
|
+
end
|
22
|
+
end
|
23
|
+
|
24
|
+
class << self
|
25
|
+
# Name of the endpoint for this product.
|
26
|
+
# @return [String]
|
27
|
+
attr_reader :endpoint_name
|
28
|
+
# Version for this product.
|
29
|
+
# @return [String]
|
30
|
+
attr_reader :endpoint_version
|
31
|
+
end
|
32
|
+
end
|
33
|
+
end
|
34
|
+
end
|
35
|
+
end
|
@@ -0,0 +1,60 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative '../../parsing'
|
4
|
+
|
5
|
+
module Mindee
|
6
|
+
module Product
|
7
|
+
module Custom
|
8
|
+
# Custom Document V1 prediction
|
9
|
+
class CustomV1Document < Mindee::Parsing::Common::Prediction
|
10
|
+
# All value fields in the document
|
11
|
+
# @return [Hash<Symbol, Mindee::Parsing::Custom::ListField>]
|
12
|
+
attr_reader :fields
|
13
|
+
# All classifications in the document
|
14
|
+
# @return [Hash<Symbol, Mindee::Parsing::Custom::ClassificationField>]
|
15
|
+
attr_reader :classifications
|
16
|
+
|
17
|
+
# @param prediction [Hash]
|
18
|
+
# @param page_id [Integer, nil]
|
19
|
+
def initialize(prediction, page_id)
|
20
|
+
super()
|
21
|
+
@fields = {}
|
22
|
+
@classifications = {}
|
23
|
+
prediction.each do |field_name, field_prediction|
|
24
|
+
field_sym = field_name.to_sym
|
25
|
+
set_field(field_sym, field_prediction, page_id)
|
26
|
+
end
|
27
|
+
end
|
28
|
+
|
29
|
+
# @return [String]
|
30
|
+
def to_s
|
31
|
+
out_str = String.new
|
32
|
+
@classifications.each do |name, info|
|
33
|
+
out_str << "\n:#{name}: #{info}".rstrip
|
34
|
+
end
|
35
|
+
@fields.each do |name, info|
|
36
|
+
out_str << "\n:#{name}: #{info}".rstrip
|
37
|
+
end
|
38
|
+
out_str[1..].to_s
|
39
|
+
end
|
40
|
+
|
41
|
+
private
|
42
|
+
|
43
|
+
# @param field_prediction [Hash]
|
44
|
+
def set_field(field_sym, field_prediction, page_id)
|
45
|
+
# Currently two types of fields possible in a custom API response:
|
46
|
+
# fields having a list of values, and classification fields.
|
47
|
+
# Here we use the fact that only value lists have the 'values' attribute.
|
48
|
+
|
49
|
+
if field_prediction.key? 'values'
|
50
|
+
@fields[field_sym] = Parsing::Custom::ListField.new(field_prediction, page_id)
|
51
|
+
elsif field_prediction.key? 'value'
|
52
|
+
@classifications[field_sym] = Parsing::Custom::ClassificationField.new(field_prediction)
|
53
|
+
else
|
54
|
+
throw 'Unknown API field type'
|
55
|
+
end
|
56
|
+
end
|
57
|
+
end
|
58
|
+
end
|
59
|
+
end
|
60
|
+
end
|
@@ -0,0 +1,32 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative '../../parsing'
|
4
|
+
require_relative 'custom_v1_document'
|
5
|
+
|
6
|
+
module Mindee
|
7
|
+
module Product
|
8
|
+
module Custom
|
9
|
+
# Custom Document V1 page.
|
10
|
+
class CustomV1Page < Mindee::Parsing::Common::Page
|
11
|
+
# @param prediction [Hash]
|
12
|
+
def initialize(prediction)
|
13
|
+
super(prediction)
|
14
|
+
@prediction = CustomV1PagePrediction.new(
|
15
|
+
prediction['prediction'],
|
16
|
+
prediction['id']
|
17
|
+
)
|
18
|
+
end
|
19
|
+
end
|
20
|
+
|
21
|
+
# Custom Document V1 page prediction.
|
22
|
+
class CustomV1PagePrediction < CustomV1Document
|
23
|
+
# @return [String]
|
24
|
+
def to_s
|
25
|
+
out_str = String.new
|
26
|
+
out_str << "\n#{super}"
|
27
|
+
out_str
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
31
|
+
end
|
32
|
+
end
|
@@ -0,0 +1,38 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative '../../../parsing'
|
4
|
+
require_relative 'license_plate_v1_document'
|
5
|
+
require_relative 'license_plate_v1_page'
|
6
|
+
|
7
|
+
module Mindee
|
8
|
+
module Product
|
9
|
+
module EU
|
10
|
+
module LicensePlate
|
11
|
+
# License Plate V1 prediction inference.
|
12
|
+
class LicensePlateV1 < Mindee::Parsing::Common::Inference
|
13
|
+
@endpoint_name = 'license_plates'
|
14
|
+
@endpoint_version = '1'
|
15
|
+
|
16
|
+
# @param prediction [Hash]
|
17
|
+
def initialize(prediction)
|
18
|
+
super
|
19
|
+
@prediction = LicensePlateV1Document.new(prediction['prediction'], nil)
|
20
|
+
@pages = []
|
21
|
+
prediction['pages'].each do |page|
|
22
|
+
@pages.push(LicensePlateV1Page.new(page))
|
23
|
+
end
|
24
|
+
end
|
25
|
+
|
26
|
+
class << self
|
27
|
+
# Name of the endpoint for this product.
|
28
|
+
# @return [String]
|
29
|
+
attr_reader :endpoint_name
|
30
|
+
# Version for this product.
|
31
|
+
# @return [String]
|
32
|
+
attr_reader :endpoint_version
|
33
|
+
end
|
34
|
+
end
|
35
|
+
end
|
36
|
+
end
|
37
|
+
end
|
38
|
+
end
|
@@ -0,0 +1,37 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative '../../../parsing'
|
4
|
+
|
5
|
+
module Mindee
|
6
|
+
module Product
|
7
|
+
module EU
|
8
|
+
module LicensePlate
|
9
|
+
# License Plate V1 document prediction.
|
10
|
+
class LicensePlateV1Document < Mindee::Parsing::Common::Prediction
|
11
|
+
include Mindee::Parsing::Standard
|
12
|
+
# List of all license plates found in the image.
|
13
|
+
# @return [Array<Mindee::Parsing::Standard::TextField>]
|
14
|
+
attr_reader :license_plates
|
15
|
+
|
16
|
+
# @param prediction [Hash]
|
17
|
+
# @param page_id [Integer, nil]
|
18
|
+
def initialize(prediction, page_id)
|
19
|
+
super()
|
20
|
+
@license_plates = []
|
21
|
+
prediction['license_plates'].each do |item|
|
22
|
+
@license_plates.push(TextField.new(item, page_id))
|
23
|
+
end
|
24
|
+
end
|
25
|
+
|
26
|
+
# @return [String]
|
27
|
+
def to_s
|
28
|
+
license_plates = @license_plates.join("\n #{' ' * 16}")
|
29
|
+
out_str = String.new
|
30
|
+
out_str << "\n:License Plates: #{license_plates}".rstrip
|
31
|
+
out_str[1..].to_s
|
32
|
+
end
|
33
|
+
end
|
34
|
+
end
|
35
|
+
end
|
36
|
+
end
|
37
|
+
end
|
@@ -0,0 +1,34 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative '../../../parsing'
|
4
|
+
require_relative 'license_plate_v1_document'
|
5
|
+
|
6
|
+
module Mindee
|
7
|
+
module Product
|
8
|
+
module EU
|
9
|
+
module LicensePlate
|
10
|
+
# License Plate V1 page.
|
11
|
+
class LicensePlateV1Page < Mindee::Parsing::Common::Page
|
12
|
+
# @param prediction [Hash]
|
13
|
+
def initialize(prediction)
|
14
|
+
super(prediction)
|
15
|
+
@prediction = LicensePlateV1PagePrediction.new(
|
16
|
+
prediction['prediction'],
|
17
|
+
prediction['id']
|
18
|
+
)
|
19
|
+
end
|
20
|
+
end
|
21
|
+
|
22
|
+
# License Plate V1 page prediction.
|
23
|
+
class LicensePlateV1PagePrediction < LicensePlateV1Document
|
24
|
+
# @return [String]
|
25
|
+
def to_s
|
26
|
+
out_str = String.new
|
27
|
+
out_str << "\n#{super}"
|
28
|
+
out_str
|
29
|
+
end
|
30
|
+
end
|
31
|
+
end
|
32
|
+
end
|
33
|
+
end
|
34
|
+
end
|
@@ -0,0 +1,36 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
require_relative '../../parsing'
|
4
|
+
require_relative 'financial_document_v1_document'
|
5
|
+
require_relative 'financial_document_v1_page'
|
6
|
+
|
7
|
+
module Mindee
|
8
|
+
module Product
|
9
|
+
module FinancialDocument
|
10
|
+
# Financial Document V1 prediction inference.
|
11
|
+
class FinancialDocumentV1 < Mindee::Parsing::Common::Inference
|
12
|
+
@endpoint_name = 'financial_document'
|
13
|
+
@endpoint_version = '1'
|
14
|
+
|
15
|
+
# @param prediction [Hash]
|
16
|
+
def initialize(prediction)
|
17
|
+
super
|
18
|
+
@prediction = FinancialDocumentV1Document.new(prediction['prediction'], nil)
|
19
|
+
@pages = []
|
20
|
+
prediction['pages'].each do |page|
|
21
|
+
@pages.push(FinancialDocumentV1Page.new(page))
|
22
|
+
end
|
23
|
+
end
|
24
|
+
|
25
|
+
class << self
|
26
|
+
# Name of the endpoint for this product.
|
27
|
+
# @return [String]
|
28
|
+
attr_reader :endpoint_name
|
29
|
+
# Version for this product.
|
30
|
+
# @return [String]
|
31
|
+
attr_reader :endpoint_version
|
32
|
+
end
|
33
|
+
end
|
34
|
+
end
|
35
|
+
end
|
36
|
+
end
|