Class: Xberg::ExtractionConfig

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

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize ⇒ ExtractionConfig

Returns a new instance of ExtractionConfig.

Parameters:



981
 
# File 'sig/types.rbs', line 981

def initialize: (?mime_detection_policy: enum_MimeDetectionPolicy, ?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: enum_ResultFormat, ?security_limits: SecurityLimits, ?max_embedded_file_bytes: Integer, ?output_format: OutputFormat, ?escape_markdown: bool, ?table_anchors: bool, ?jupyter_cell_rendering: enum_JupyterCellRendering, ?apply_notebook_cell_tags: bool, ?layout: LayoutDetectionConfig, ?transcription: TranscriptionConfig, ?use_layout_for_markdown: bool, ?include_document_structure: bool, ?acceleration: AccelerationConfig, ?cache_namespace: String, ?cache_ttl_secs: Integer, ?email: EmailConfig, ?csv: CsvConfig, ?geojson: GeoJsonExtractionConfig, ?concurrency: ConcurrencyConfig, ?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? (readonly)

Returns the value of attribute acceleration.

Returns:



961
962
963
 
# File 'sig/types.rbs', line 961

def acceleration
  @acceleration
end

#apply_notebook_cell_tags ⇒ Boolean (readonly)

Returns the value of attribute apply_notebook_cell_tags.

Returns:

  • (Boolean) 


956
957
958
 
# File 'sig/types.rbs', line 956

def apply_notebook_cell_tags
  @apply_notebook_cell_tags
end

#cache_namespace ⇒ String? (readonly)

Returns the value of attribute cache_namespace.

Returns:

  • (String, nil) 


962
963
964
 
# File 'sig/types.rbs', line 962

def cache_namespace
  @cache_namespace
end

#cache_ttl_secs ⇒ Integer? (readonly)

Returns the value of attribute cache_ttl_secs.

Returns:

  • (Integer, nil) 


963
964
965
 
# File 'sig/types.rbs', line 963

def cache_ttl_secs
  @cache_ttl_secs
end

#captioning ⇒ CaptioningConfig? (readonly)

Returns the value of attribute captioning.

Returns:



978
979
980
 
# File 'sig/types.rbs', line 978

def captioning
  @captioning
end

#chunk_classification ⇒ ChunkClassificationConfig? (readonly)

Returns the value of attribute chunk_classification.

Returns:



977
978
979
 
# File 'sig/types.rbs', line 977

def chunk_classification
  @chunk_classification
end

#chunking ⇒ ChunkingConfig? (readonly)

Returns the value of attribute chunking.

Returns:



936
937
938
 
# File 'sig/types.rbs', line 936

def chunking
  @chunking
end

#concurrency ⇒ ConcurrencyConfig? (readonly)

Returns the value of attribute concurrency.

Returns:



967
968
969
 
# File 'sig/types.rbs', line 967

def concurrency
  @concurrency
end

#content_filter ⇒ ContentFilterConfig? (readonly)

Returns the value of attribute content_filter.

Returns:



937
938
939
 
# File 'sig/types.rbs', line 937

def content_filter
  @content_filter
end

#csv ⇒ CsvConfig? (readonly)

Returns the value of attribute csv.

Returns:



965
966
967
 
# File 'sig/types.rbs', line 965

def csv
  @csv
end

#disable_ocr ⇒ Boolean (readonly)

Returns the value of attribute disable_ocr.

Returns:

  • (Boolean) 


935
936
937
 
# File 'sig/types.rbs', line 935

def disable_ocr
  @disable_ocr
end

#email ⇒ EmailConfig? (readonly)

Returns the value of attribute email.

Returns:



964
965
966
 
# File 'sig/types.rbs', line 964

def email
  @email
end

#enable_quality_processing ⇒ Boolean (readonly)

Returns the value of attribute enable_quality_processing.

Returns:

  • (Boolean) 


930
931
932
 
# File 'sig/types.rbs', line 930

def enable_quality_processing
  @enable_quality_processing
end

#escape_markdown ⇒ Boolean (readonly)

Returns the value of attribute escape_markdown.

Returns:

  • (Boolean) 


953
954
955
 
# File 'sig/types.rbs', line 953

def escape_markdown
  @escape_markdown
end

#extraction_timeout_secs ⇒ Integer? (readonly)

Returns the value of attribute extraction_timeout_secs.

Returns:

  • (Integer, nil) 


947
948
949
 
# File 'sig/types.rbs', line 947

def extraction_timeout_secs
  @extraction_timeout_secs
end

#force_ocr ⇒ Boolean (readonly)

Returns the value of attribute force_ocr.

Returns:

  • (Boolean) 


932
933
934
 
# File 'sig/types.rbs', line 932

def force_ocr
  @force_ocr
end

#force_ocr_pages ⇒ Array[Integer]? (readonly)

Returns the value of attribute force_ocr_pages.

Returns:

  • (Array[Integer], nil) 


934
935
936
 
# File 'sig/types.rbs', line 934

def force_ocr_pages
  @force_ocr_pages
end

#geojson ⇒ GeoJsonExtractionConfig? (readonly)

Returns the value of attribute geojson.

Returns:



966
967
968
 
# File 'sig/types.rbs', line 966

def geojson
  @geojson
end

#html_options ⇒ ConversionOptions? (readonly)

Returns the value of attribute html_options.

Returns:



945
946
947
 
# File 'sig/types.rbs', line 945

def html_options
  @html_options
end

#html_output ⇒ HtmlOutputConfig? (readonly)

Returns the value of attribute html_output.

Returns:



946
947
948
 
# File 'sig/types.rbs', line 946

def html_output
  @html_output
end

#images ⇒ ImageExtractionConfig? (readonly)

Returns the value of attribute images.

Returns:



938
939
940
 
# File 'sig/types.rbs', line 938

def images
  @images
end

#include_document_structure ⇒ Boolean (readonly)

Returns the value of attribute include_document_structure.

Returns:

  • (Boolean) 


960
961
962
 
# File 'sig/types.rbs', line 960

def include_document_structure
  @include_document_structure
end

#jupyter_cell_rendering ⇒ enum_JupyterCellRendering (readonly)

Returns the value of attribute jupyter_cell_rendering.

Returns:

  • (enum_JupyterCellRendering) 


955
956
957
 
# File 'sig/types.rbs', line 955

def jupyter_cell_rendering
  @jupyter_cell_rendering
end

#keywords ⇒ KeywordConfig? (readonly)

Returns the value of attribute keywords.

Returns:



943
944
945
 
# File 'sig/types.rbs', line 943

def keywords
  @keywords
end

#language_detection ⇒ LanguageDetectionConfig? (readonly)

Returns the value of attribute language_detection.

Returns:



941
942
943
 
# File 'sig/types.rbs', line 941

def language_detection
  @language_detection
end

#layout ⇒ LayoutDetectionConfig? (readonly)

Returns the value of attribute layout.

Returns:



957
958
959
 
# File 'sig/types.rbs', line 957

def layout
  @layout
end

#max_archive_depth ⇒ Integer (readonly)

Returns the value of attribute max_archive_depth.

Returns:

  • (Integer) 


969
970
971
 
# File 'sig/types.rbs', line 969

def max_archive_depth
  @max_archive_depth
end

#max_concurrent_extractions ⇒ Integer? (readonly)

Returns the value of attribute max_concurrent_extractions.

Returns:

  • (Integer, nil) 


948
949
950
 
# File 'sig/types.rbs', line 948

def max_concurrent_extractions
  @max_concurrent_extractions
end

#max_embedded_file_bytes ⇒ Integer? (readonly)

Returns the value of attribute max_embedded_file_bytes.

Returns:

  • (Integer, nil) 


951
952
953
 
# File 'sig/types.rbs', line 951

def max_embedded_file_bytes
  @max_embedded_file_bytes
end

#mime_detection_policy ⇒ enum_MimeDetectionPolicy (readonly)

Returns the value of attribute mime_detection_policy.

Returns:

  • (enum_MimeDetectionPolicy) 


928
929
930
 
# File 'sig/types.rbs', line 928

def mime_detection_policy
  @mime_detection_policy
end

#ner ⇒ NerConfig? (readonly)

Returns the value of attribute ner.

Returns:



972
973
974
 
# File 'sig/types.rbs', line 972

def ner
  @ner
end

#ocr ⇒ OcrConfig? (readonly)

Returns the value of attribute ocr.

Returns:



931
932
933
 
# File 'sig/types.rbs', line 931

def ocr
  @ocr
end

#ocr_strategy ⇒ OcrStrategy (readonly)

Returns the value of attribute ocr_strategy.

Returns:



933
934
935
 
# File 'sig/types.rbs', line 933

def ocr_strategy
  @ocr_strategy
end

#output_format ⇒ OutputFormat (readonly)

Returns the value of attribute output_format.

Returns:



952
953
954
 
# File 'sig/types.rbs', line 952

def output_format
  @output_format
end

#page_classification ⇒ PageClassificationConfig? (readonly)

Returns the value of attribute page_classification.

Returns:



976
977
978
 
# File 'sig/types.rbs', line 976

def page_classification
  @page_classification
end

#pages ⇒ PageConfig? (readonly)

Returns the value of attribute pages.

Returns:



942
943
944
 
# File 'sig/types.rbs', line 942

def pages
  @pages
end

#pdf_options ⇒ PdfConfig? (readonly)

Returns the value of attribute pdf_options.

Returns:



939
940
941
 
# File 'sig/types.rbs', line 939

def pdf_options
  @pdf_options
end

#postprocessor ⇒ PostProcessorConfig? (readonly)

Returns the value of attribute postprocessor.

Returns:



944
945
946
 
# File 'sig/types.rbs', line 944

def postprocessor
  @postprocessor
end

#qr_codes ⇒ Boolean? (readonly)

Returns the value of attribute qr_codes.

Returns:

  • (Boolean, nil) 


979
980
981
 
# File 'sig/types.rbs', line 979

def qr_codes
  @qr_codes
end

#redaction ⇒ RedactionConfig? (readonly)

Returns the value of attribute redaction.

Returns:



973
974
975
 
# File 'sig/types.rbs', line 973

def redaction
  @redaction
end

#result_format ⇒ enum_ResultFormat (readonly)

Returns the value of attribute result_format.

Returns:

  • (enum_ResultFormat) 


949
950
951
 
# File 'sig/types.rbs', line 949

def result_format
  @result_format
end

#security_limits ⇒ SecurityLimits? (readonly)

Returns the value of attribute security_limits.

Returns:



950
951
952
 
# File 'sig/types.rbs', line 950

def security_limits
  @security_limits
end

#structured_extraction ⇒ StructuredExtractionConfig? (readonly)

Returns the value of attribute structured_extraction.

Returns:



971
972
973
 
# File 'sig/types.rbs', line 971

def structured_extraction
  @structured_extraction
end

#summarization ⇒ SummarizationConfig? (readonly)

Returns the value of attribute summarization.

Returns:



974
975
976
 
# File 'sig/types.rbs', line 974

def summarization
  @summarization
end

#table_anchors ⇒ Boolean (readonly)

Returns the value of attribute table_anchors.

Returns:

  • (Boolean) 


954
955
956
 
# File 'sig/types.rbs', line 954

def table_anchors
  @table_anchors
end

#token_reduction ⇒ TokenReductionOptions? (readonly)

Returns the value of attribute token_reduction.

Returns:



940
941
942
 
# File 'sig/types.rbs', line 940

def token_reduction
  @token_reduction
end

#transcription ⇒ TranscriptionConfig? (readonly)

Returns the value of attribute transcription.

Returns:



958
959
960
 
# File 'sig/types.rbs', line 958

def transcription
  @transcription
end

#translation ⇒ TranslationConfig? (readonly)

Returns the value of attribute translation.

Returns:



975
976
977
 
# File 'sig/types.rbs', line 975

def translation
  @translation
end

#tree_sitter ⇒ TreeSitterConfig? (readonly)

Returns the value of attribute tree_sitter.

Returns:



970
971
972
 
# File 'sig/types.rbs', line 970

def tree_sitter
  @tree_sitter
end

#url ⇒ UrlExtractionConfig (readonly)

Returns the value of attribute url.

Returns:



968
969
970
 
# File 'sig/types.rbs', line 968

def url
  @url
end

#use_cache ⇒ Boolean (readonly)

Returns the value of attribute use_cache.

Returns:

  • (Boolean) 


929
930
931
 
# File 'sig/types.rbs', line 929

def use_cache
  @use_cache
end

#use_layout_for_markdown ⇒ Boolean (readonly)

Returns the value of attribute use_layout_for_markdown.

Returns:

  • (Boolean) 


959
960
961
 
# File 'sig/types.rbs', line 959

def use_layout_for_markdown
  @use_layout_for_markdown
end

Instance Method Details

#needs_image_data ⇒ Boolean

Returns:

  • (Boolean) 


983
 
# File 'sig/types.rbs', line 983

def needs_image_data: () -> bool

#needs_image_processing ⇒ Boolean

Returns:

  • (Boolean) 


986
 
# File 'sig/types.rbs', line 986

def needs_image_processing: () -> bool

#runs_ocr_on_embedded_images ⇒ Boolean

Returns:

  • (Boolean) 


984
 
# File 'sig/types.rbs', line 984

def runs_ocr_on_embedded_images: () -> bool

#validate ⇒ void

This method returns an undefined value.



982
 
# File 'sig/types.rbs', line 982

def validate: () -> void

#wants_own_bytes_in_result ⇒ Boolean

Returns:

  • (Boolean) 


985
 
# File 'sig/types.rbs', line 985

def wants_own_bytes_in_result: () -> bool