Class: Xberg::OcrQualityThresholds

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

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize ⇒ OcrQualityThresholds

Returns a new instance of OcrQualityThresholds.

Parameters:

  • min_total_non_whitespace: (Integer)
  • min_non_whitespace_per_page: (Float)
  • min_meaningful_word_len: (Integer)
  • min_meaningful_words: (Integer)
  • min_alnum_ratio: (Float)
  • min_garbage_chars: (Integer)
  • max_fragmented_word_ratio: (Float)
  • critical_fragmented_word_ratio: (Float)
  • max_ocr_output_fragmented_word_ratio: (Float)
  • min_ocr_mean_confidence: (Float)
  • min_words_for_ocr_output_check: (Integer)
  • max_ocr_output_dict_invalid_word_ratio: (Float)
  • discard_suspected_ocr_noise: (Boolean)
  • min_avg_word_length: (Float)
  • min_words_for_avg_length_check: (Integer)
  • min_consecutive_repeat_ratio: (Float)
  • min_words_for_repeat_check: (Integer)
  • substantive_min_chars: (Integer)
  • non_text_min_chars: (Integer)
  • alnum_ws_ratio_threshold: (Float)
  • pipeline_min_quality: (Float)
  • min_undecodable_ratio: (Float)
  • enable_provenance_ocr_routing: (Boolean)
  • min_provenance_fallback_ratio: (Float)
  • enable_plausibility_ocr_routing: (Boolean)
  • min_reliable_language_chunk_ratio: (Float) 


1792
 
# File 'sig/types.rbs', line 1792

def initialize: (?min_total_non_whitespace: Integer, ?min_non_whitespace_per_page: Float, ?min_meaningful_word_len: Integer, ?min_meaningful_words: Integer, ?min_alnum_ratio: Float, ?min_garbage_chars: Integer, ?max_fragmented_word_ratio: Float, ?critical_fragmented_word_ratio: Float, ?max_ocr_output_fragmented_word_ratio: Float, ?min_ocr_mean_confidence: Float, ?min_words_for_ocr_output_check: Integer, ?max_ocr_output_dict_invalid_word_ratio: Float, ?discard_suspected_ocr_noise: bool, ?min_avg_word_length: Float, ?min_words_for_avg_length_check: Integer, ?min_consecutive_repeat_ratio: Float, ?min_words_for_repeat_check: Integer, ?substantive_min_chars: Integer, ?non_text_min_chars: Integer, ?alnum_ws_ratio_threshold: Float, ?pipeline_min_quality: Float, ?min_undecodable_ratio: Float, ?enable_provenance_ocr_routing: bool, ?min_provenance_fallback_ratio: Float, ?enable_plausibility_ocr_routing: bool, ?min_reliable_language_chunk_ratio: Float) -> void

Instance Attribute Details

#alnum_ws_ratio_threshold ⇒ Float (readonly)

Returns the value of attribute alnum_ws_ratio_threshold.

Returns:

  • (Float) 


1784
1785
1786
 
# File 'sig/types.rbs', line 1784

def alnum_ws_ratio_threshold
  @alnum_ws_ratio_threshold
end

#critical_fragmented_word_ratio ⇒ Float (readonly)

Returns the value of attribute critical_fragmented_word_ratio.

Returns:

  • (Float) 


1772
1773
1774
 
# File 'sig/types.rbs', line 1772

def critical_fragmented_word_ratio
  @critical_fragmented_word_ratio
end

#discard_suspected_ocr_noise ⇒ Boolean (readonly)

Returns the value of attribute discard_suspected_ocr_noise.

Returns:

  • (Boolean) 


1777
1778
1779
 
# File 'sig/types.rbs', line 1777

def discard_suspected_ocr_noise
  @discard_suspected_ocr_noise
end

#enable_plausibility_ocr_routing ⇒ Boolean (readonly)

Returns the value of attribute enable_plausibility_ocr_routing.

Returns:

  • (Boolean) 


1789
1790
1791
 
# File 'sig/types.rbs', line 1789

def enable_plausibility_ocr_routing
  @enable_plausibility_ocr_routing
end

#enable_provenance_ocr_routing ⇒ Boolean (readonly)

Returns the value of attribute enable_provenance_ocr_routing.

Returns:

  • (Boolean) 


1787
1788
1789
 
# File 'sig/types.rbs', line 1787

def enable_provenance_ocr_routing
  @enable_provenance_ocr_routing
end

#max_fragmented_word_ratio ⇒ Float (readonly)

Returns the value of attribute max_fragmented_word_ratio.

Returns:

  • (Float) 


1771
1772
1773
 
# File 'sig/types.rbs', line 1771

def max_fragmented_word_ratio
  @max_fragmented_word_ratio
end

#max_ocr_output_dict_invalid_word_ratio ⇒ Float (readonly)

Returns the value of attribute max_ocr_output_dict_invalid_word_ratio.

Returns:

  • (Float) 


1776
1777
1778
 
# File 'sig/types.rbs', line 1776

def max_ocr_output_dict_invalid_word_ratio
  @max_ocr_output_dict_invalid_word_ratio
end

#max_ocr_output_fragmented_word_ratio ⇒ Float (readonly)

Returns the value of attribute max_ocr_output_fragmented_word_ratio.

Returns:

  • (Float) 


1773
1774
1775
 
# File 'sig/types.rbs', line 1773

def max_ocr_output_fragmented_word_ratio
  @max_ocr_output_fragmented_word_ratio
end

#min_alnum_ratio ⇒ Float (readonly)

Returns the value of attribute min_alnum_ratio.

Returns:

  • (Float) 


1769
1770
1771
 
# File 'sig/types.rbs', line 1769

def min_alnum_ratio
  @min_alnum_ratio
end

#min_avg_word_length ⇒ Float (readonly)

Returns the value of attribute min_avg_word_length.

Returns:

  • (Float) 


1778
1779
1780
 
# File 'sig/types.rbs', line 1778

def min_avg_word_length
  @min_avg_word_length
end

#min_consecutive_repeat_ratio ⇒ Float (readonly)

Returns the value of attribute min_consecutive_repeat_ratio.

Returns:

  • (Float) 


1780
1781
1782
 
# File 'sig/types.rbs', line 1780

def min_consecutive_repeat_ratio
  @min_consecutive_repeat_ratio
end

#min_garbage_chars ⇒ Integer (readonly)

Returns the value of attribute min_garbage_chars.

Returns:

  • (Integer) 


1770
1771
1772
 
# File 'sig/types.rbs', line 1770

def min_garbage_chars
  @min_garbage_chars
end

#min_meaningful_word_len ⇒ Integer (readonly)

Returns the value of attribute min_meaningful_word_len.

Returns:

  • (Integer) 


1767
1768
1769
 
# File 'sig/types.rbs', line 1767

def min_meaningful_word_len
  @min_meaningful_word_len
end

#min_meaningful_words ⇒ Integer (readonly)

Returns the value of attribute min_meaningful_words.

Returns:

  • (Integer) 


1768
1769
1770
 
# File 'sig/types.rbs', line 1768

def min_meaningful_words
  @min_meaningful_words
end

#min_non_whitespace_per_page ⇒ Float (readonly)

Returns the value of attribute min_non_whitespace_per_page.

Returns:

  • (Float) 


1766
1767
1768
 
# File 'sig/types.rbs', line 1766

def min_non_whitespace_per_page
  @min_non_whitespace_per_page
end

#min_ocr_mean_confidence ⇒ Float (readonly)

Returns the value of attribute min_ocr_mean_confidence.

Returns:

  • (Float) 


1774
1775
1776
 
# File 'sig/types.rbs', line 1774

def min_ocr_mean_confidence
  @min_ocr_mean_confidence
end

#min_provenance_fallback_ratio ⇒ Float (readonly)

Returns the value of attribute min_provenance_fallback_ratio.

Returns:

  • (Float) 


1788
1789
1790
 
# File 'sig/types.rbs', line 1788

def min_provenance_fallback_ratio
  @min_provenance_fallback_ratio
end

#min_reliable_language_chunk_ratio ⇒ Float (readonly)

Returns the value of attribute min_reliable_language_chunk_ratio.

Returns:

  • (Float) 


1790
1791
1792
 
# File 'sig/types.rbs', line 1790

def min_reliable_language_chunk_ratio
  @min_reliable_language_chunk_ratio
end

#min_total_non_whitespace ⇒ Integer (readonly)

Returns the value of attribute min_total_non_whitespace.

Returns:

  • (Integer) 


1765
1766
1767
 
# File 'sig/types.rbs', line 1765

def min_total_non_whitespace
  @min_total_non_whitespace
end

#min_undecodable_ratio ⇒ Float (readonly)

Returns the value of attribute min_undecodable_ratio.

Returns:

  • (Float) 


1786
1787
1788
 
# File 'sig/types.rbs', line 1786

def min_undecodable_ratio
  @min_undecodable_ratio
end

#min_words_for_avg_length_check ⇒ Integer (readonly)

Returns the value of attribute min_words_for_avg_length_check.

Returns:

  • (Integer) 


1779
1780
1781
 
# File 'sig/types.rbs', line 1779

def min_words_for_avg_length_check
  @min_words_for_avg_length_check
end

#min_words_for_ocr_output_check ⇒ Integer (readonly)

Returns the value of attribute min_words_for_ocr_output_check.

Returns:

  • (Integer) 


1775
1776
1777
 
# File 'sig/types.rbs', line 1775

def min_words_for_ocr_output_check
  @min_words_for_ocr_output_check
end

#min_words_for_repeat_check ⇒ Integer (readonly)

Returns the value of attribute min_words_for_repeat_check.

Returns:

  • (Integer) 


1781
1782
1783
 
# File 'sig/types.rbs', line 1781

def min_words_for_repeat_check
  @min_words_for_repeat_check
end

#non_text_min_chars ⇒ Integer (readonly)

Returns the value of attribute non_text_min_chars.

Returns:

  • (Integer) 


1783
1784
1785
 
# File 'sig/types.rbs', line 1783

def non_text_min_chars
  @non_text_min_chars
end

#pipeline_min_quality ⇒ Float (readonly)

Returns the value of attribute pipeline_min_quality.

Returns:

  • (Float) 


1785
1786
1787
 
# File 'sig/types.rbs', line 1785

def pipeline_min_quality
  @pipeline_min_quality
end

#substantive_min_chars ⇒ Integer (readonly)

Returns the value of attribute substantive_min_chars.

Returns:

  • (Integer) 


1782
1783
1784
 
# File 'sig/types.rbs', line 1782

def substantive_min_chars
  @substantive_min_chars
end