Class: Xberg::ExtractionConfig

Inherits:
Object
  • Object
show all
Defined in:
sig/types.rbs

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize ⇒ ExtractionConfig

Returns a new instance of ExtractionConfig.

Parameters:



128
 
# File 'sig/types.rbs', line 128

def initialize: (?use_cache: bool, ?enable_quality_processing: bool, ?ocr: OcrConfig, ?force_ocr: bool, ?ocr_strategy: OcrStrategy, ?force_ocr_pages: Array[Integer], ?disable_ocr: bool, ?chunking: ChunkingConfig, ?content_filter: ContentFilterConfig, ?images: ImageExtractionConfig, ?pdf_options: PdfConfig, ?token_reduction: TokenReductionOptions, ?language_detection: LanguageDetectionConfig, ?pages: PageConfig, ?keywords: KeywordConfig, ?postprocessor: PostProcessorConfig, ?html_options: ConversionOptions, ?html_output: HtmlOutputConfig, ?extraction_timeout_secs: Integer, ?max_concurrent_extractions: Integer, ?result_format: ResultFormat, ?security_limits: SecurityLimits, ?max_embedded_file_bytes: Integer, ?output_format: OutputFormat, ?escape_markdown: bool, ?table_anchors: bool, ?jupyter_cell_rendering: JupyterCellRendering, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?use_layout_for_markdown: bool, ?include_document_structure: bool, ?acceleration: AccelerationConfig, ?cache_namespace: String, ?cache_ttl_secs: Integer, ?email: EmailConfig, ?url: UrlExtractionConfig, ?max_archive_depth: Integer, ?tree_sitter: TreeSitterConfig, ?structured_extraction: StructuredExtractionConfig, ?ner: NerConfig, ?redaction: RedactionConfig, ?summarization: SummarizationConfig, ?translation: TranslationConfig, ?page_classification: PageClassificationConfig, ?chunk_classification: ChunkClassificationConfig, ?captioning: CaptioningConfig, ?qr_codes: bool) -> void

Instance Attribute Details

#acceleration ⇒ AccelerationConfig?

Returns the value of attribute acceleration.

Returns:



111
112
113
 
# File 'sig/types.rbs', line 111

def acceleration
  @acceleration
end

#cache_namespace ⇒ String?

Returns the value of attribute cache_namespace.

Returns:

  • (String, nil) 


112
113
114
 
# File 'sig/types.rbs', line 112

def cache_namespace
  @cache_namespace
end

#cache_ttl_secs ⇒ Integer?

Returns the value of attribute cache_ttl_secs.

Returns:

  • (Integer, nil) 


113
114
115
 
# File 'sig/types.rbs', line 113

def cache_ttl_secs
  @cache_ttl_secs
end

#captioning ⇒ CaptioningConfig?

Returns the value of attribute captioning.

Returns:



125
126
127
 
# File 'sig/types.rbs', line 125

def captioning
  @captioning
end

#chunk_classification ⇒ ChunkClassificationConfig?

Returns the value of attribute chunk_classification.

Returns:



124
125
126
 
# File 'sig/types.rbs', line 124

def chunk_classification
  @chunk_classification
end

#chunking ⇒ ChunkingConfig?

Returns the value of attribute chunking.

Returns:



87
88
89
 
# File 'sig/types.rbs', line 87

def chunking
  @chunking
end

#content_filter ⇒ ContentFilterConfig?

Returns the value of attribute content_filter.

Returns:



88
89
90
 
# File 'sig/types.rbs', line 88

def content_filter
  @content_filter
end

#disable_ocr ⇒ Boolean?

Returns the value of attribute disable_ocr.

Returns:

  • (Boolean, nil) 


86
87
88
 
# File 'sig/types.rbs', line 86

def disable_ocr
  @disable_ocr
end

#email ⇒ EmailConfig?

Returns the value of attribute email.

Returns:



114
115
116
 
# File 'sig/types.rbs', line 114

def email
  @email
end

#enable_quality_processing ⇒ Boolean?

Returns the value of attribute enable_quality_processing.

Returns:

  • (Boolean, nil) 


81
82
83
 
# File 'sig/types.rbs', line 81

def enable_quality_processing
  @enable_quality_processing
end

#escape_markdown ⇒ Boolean?

Returns the value of attribute escape_markdown.

Returns:

  • (Boolean, nil) 


104
105
106
 
# File 'sig/types.rbs', line 104

def escape_markdown
  @escape_markdown
end

#extraction_timeout_secs ⇒ Integer?

Returns the value of attribute extraction_timeout_secs.

Returns:

  • (Integer, nil) 


98
99
100
 
# File 'sig/types.rbs', line 98

def extraction_timeout_secs
  @extraction_timeout_secs
end

#force_ocr ⇒ Boolean?

Returns the value of attribute force_ocr.

Returns:

  • (Boolean, nil) 


83
84
85
 
# File 'sig/types.rbs', line 83

def force_ocr
  @force_ocr
end

#force_ocr_pages ⇒ Array[Integer]?

Returns the value of attribute force_ocr_pages.

Returns:

  • (Array[Integer], nil) 


85
86
87
 
# File 'sig/types.rbs', line 85

def force_ocr_pages
  @force_ocr_pages
end

#html_options ⇒ ConversionOptions?

Returns the value of attribute html_options.

Returns:



96
97
98
 
# File 'sig/types.rbs', line 96

def html_options
  @html_options
end

#html_output ⇒ HtmlOutputConfig?

Returns the value of attribute html_output.

Returns:



97
98
99
 
# File 'sig/types.rbs', line 97

def html_output
  @html_output
end

#images ⇒ ImageExtractionConfig?

Returns the value of attribute images.

Returns:



89
90
91
 
# File 'sig/types.rbs', line 89

def images
  @images
end

#include_document_structure ⇒ Boolean?

Returns the value of attribute include_document_structure.

Returns:

  • (Boolean, nil) 


110
111
112
 
# File 'sig/types.rbs', line 110

def include_document_structure
  @include_document_structure
end

#jupyter_cell_rendering ⇒ JupyterCellRendering?

Returns the value of attribute jupyter_cell_rendering.

Returns:



106
107
108
 
# File 'sig/types.rbs', line 106

def jupyter_cell_rendering
  @jupyter_cell_rendering
end

#keywords ⇒ KeywordConfig?

Returns the value of attribute keywords.

Returns:



94
95
96
 
# File 'sig/types.rbs', line 94

def keywords
  @keywords
end

#language_detection ⇒ LanguageDetectionConfig?

Returns the value of attribute language_detection.

Returns:



92
93
94
 
# File 'sig/types.rbs', line 92

def language_detection
  @language_detection
end

#layout ⇒ LayoutDetectionConfig?

Returns the value of attribute layout.

Returns:



107
108
109
 
# File 'sig/types.rbs', line 107

def layout
  @layout
end

#max_archive_depth ⇒ Integer?

Returns the value of attribute max_archive_depth.

Returns:

  • (Integer, nil) 


116
117
118
 
# File 'sig/types.rbs', line 116

def max_archive_depth
  @max_archive_depth
end

#max_concurrent_extractions ⇒ Integer?

Returns the value of attribute max_concurrent_extractions.

Returns:

  • (Integer, nil) 


99
100
101
 
# File 'sig/types.rbs', line 99

def max_concurrent_extractions
  @max_concurrent_extractions
end

#max_embedded_file_bytes ⇒ Integer?

Returns the value of attribute max_embedded_file_bytes.

Returns:

  • (Integer, nil) 


102
103
104
 
# File 'sig/types.rbs', line 102

def max_embedded_file_bytes
  @max_embedded_file_bytes
end

#ner ⇒ NerConfig?

Returns the value of attribute ner.

Returns:



119
120
121
 
# File 'sig/types.rbs', line 119

def ner
  @ner
end

#ocr ⇒ OcrConfig?

Returns the value of attribute ocr.

Returns:



82
83
84
 
# File 'sig/types.rbs', line 82

def ocr
  @ocr
end

#ocr_strategy ⇒ OcrStrategy?

Returns the value of attribute ocr_strategy.

Returns:



84
85
86
 
# File 'sig/types.rbs', line 84

def ocr_strategy
  @ocr_strategy
end

#output_format ⇒ OutputFormat?

Returns the value of attribute output_format.

Returns:



103
104
105
 
# File 'sig/types.rbs', line 103

def output_format
  @output_format
end

#page_classification ⇒ PageClassificationConfig?

Returns the value of attribute page_classification.

Returns:



123
124
125
 
# File 'sig/types.rbs', line 123

def page_classification
  @page_classification
end

#pages ⇒ PageConfig?

Returns the value of attribute pages.

Returns:



93
94
95
 
# File 'sig/types.rbs', line 93

def pages
  @pages
end

#pdf_options ⇒ PdfConfig?

Returns the value of attribute pdf_options.

Returns:



90
91
92
 
# File 'sig/types.rbs', line 90

def pdf_options
  @pdf_options
end

#postprocessor ⇒ PostProcessorConfig?

Returns the value of attribute postprocessor.

Returns:



95
96
97
 
# File 'sig/types.rbs', line 95

def postprocessor
  @postprocessor
end

#qr_codes ⇒ Boolean?

Returns the value of attribute qr_codes.

Returns:

  • (Boolean, nil) 


126
127
128
 
# File 'sig/types.rbs', line 126

def qr_codes
  @qr_codes
end

#redaction ⇒ RedactionConfig?

Returns the value of attribute redaction.

Returns:



120
121
122
 
# File 'sig/types.rbs', line 120

def redaction
  @redaction
end

#result_format ⇒ ResultFormat?

Returns the value of attribute result_format.

Returns:



100
101
102
 
# File 'sig/types.rbs', line 100

def result_format
  @result_format
end

#security_limits ⇒ SecurityLimits?

Returns the value of attribute security_limits.

Returns:



101
102
103
 
# File 'sig/types.rbs', line 101

def security_limits
  @security_limits
end

#structured_extraction ⇒ StructuredExtractionConfig?

Returns the value of attribute structured_extraction.

Returns:



118
119
120
 
# File 'sig/types.rbs', line 118

def structured_extraction
  @structured_extraction
end

#summarization ⇒ SummarizationConfig?

Returns the value of attribute summarization.

Returns:



121
122
123
 
# File 'sig/types.rbs', line 121

def summarization
  @summarization
end

#table_anchors ⇒ Boolean?

Returns the value of attribute table_anchors.

Returns:

  • (Boolean, nil) 


105
106
107
 
# File 'sig/types.rbs', line 105

def table_anchors
  @table_anchors
end

#token_reduction ⇒ TokenReductionOptions?

Returns the value of attribute token_reduction.

Returns:



91
92
93
 
# File 'sig/types.rbs', line 91

def token_reduction
  @token_reduction
end

#transcription ⇒ TranscriptionConfig?

Returns the value of attribute transcription.

Returns:



108
109
110
 
# File 'sig/types.rbs', line 108

def transcription
  @transcription
end

#translation ⇒ TranslationConfig?

Returns the value of attribute translation.

Returns:



122
123
124
 
# File 'sig/types.rbs', line 122

def translation
  @translation
end

#tree_sitter ⇒ TreeSitterConfig?

Returns the value of attribute tree_sitter.

Returns:



117
118
119
 
# File 'sig/types.rbs', line 117

def tree_sitter
  @tree_sitter
end

#url ⇒ UrlExtractionConfig?

Returns the value of attribute url.

Returns:



115
116
117
 
# File 'sig/types.rbs', line 115

def url
  @url
end

#use_cache ⇒ Boolean?

Returns the value of attribute use_cache.

Returns:

  • (Boolean, nil) 


80
81
82
 
# File 'sig/types.rbs', line 80

def use_cache
  @use_cache
end

#use_layout_for_markdown ⇒ Boolean?

Returns the value of attribute use_layout_for_markdown.

Returns:

  • (Boolean, nil) 


109
110
111
 
# File 'sig/types.rbs', line 109

def use_layout_for_markdown
  @use_layout_for_markdown
end

Class Method Details

.default ⇒ ExtractionConfig

Returns:



131
 
# File 'sig/types.rbs', line 131

def self.default: () -> ExtractionConfig

Instance Method Details

#needs_image_data ⇒ Boolean

Returns:

  • (Boolean) 


129
 
# File 'sig/types.rbs', line 129

def needs_image_data: () -> bool

#needs_image_processing ⇒ Boolean

Returns:

  • (Boolean) 


130
 
# File 'sig/types.rbs', line 130

def needs_image_processing: () -> bool