notebook/app/models/concerns/has_parseable_text.rb
2019-02-27 14:58:04 -06:00

64 lines
1.5 KiB
Ruby

require 'active_support/concern'
module HasParseableText
extend ActiveSupport::Concern
included do
def plaintext
@plaintext ||= Documents::PlaintextService.from_html(self.body)
end
def characters
@characters ||= plaintext.chars
end
def paragraphs
@paragraphs ||= plaintext.split(/[\r\n\t]+/)
end
def sentences
@sentences ||= plaintext.strip.split(/[!\?\.]/).reject(&:empty?)
end
def words
@words ||= plaintext.downcase.gsub(/[^\s\w']/, '').split(' ').reject { |w| is_numeric?(w) }
end
def pages
@pages ||= plaintext.split("\n").each_slice(Documents::PlaintextService::PLAINTEXT_LINES_PER_PAGE)
end
def acronyms
@acroynyms ||= words
.select { |word| word == word.upcase && word.length > 1 && !is_numeric?(word) }
.uniq
.sort
end
# As defined by Robert Gunning in the GFI and SMOG
def complex_words
@complex_words ||= unique_words.select { |word| Documents::Analysis::SyllablesService.count(word) >= 3 }
end
def simple_words
@simple_words ||= unique_words - complex_words
end
def unique_words
words.map(&:downcase).uniq
end
def words_with_syllables syllable_count
words.select { |word| Documents::Analysis::SyllablesService.count(word) == syllable_count }
end
def word_syllables
words.map { |word| Documents::Analysis::SyllablesService.count(word) }
end
def is_numeric?(string)
true if Float(string) rescue false
end
end
end