class RTesseract

Constants

VERSION

Attributes

config[R]

errors[R]

source[R]

Public Class Methods

check_version!() click to toggle source

# File lib/rtesseract/check.rb, line 11
def check_version!
  raise RTesseract::Error, 'Tesseract OCR 3.5 or later not installed' if RTesseract.tesseract_version < 3.05
end

config() click to toggle source

# File lib/rtesseract/configuration.rb, line 17
def config
  @config ||= RTesseract::Configuration.new(
    command: 'tesseract',
    debug_file: '/dev/null'
  )
end

configure() { |config| ... } click to toggle source

# File lib/rtesseract/configuration.rb, line 24
def configure
  yield(config) if block_given?
end

new(src = '', options = {}) click to toggle source

# File lib/rtesseract.rb, line 17
def initialize(src = '', options = {})
  @source = src
  @config = RTesseract.config.merge(options)
  @errors = []
end

reset_config!() click to toggle source

# File lib/rtesseract/configuration.rb, line 28
def reset_config!
  @config = nil
end

tesseract_version() click to toggle source

# File lib/rtesseract/check.rb, line 5
def tesseract_version
  Open3.capture2e(RTesseract.config.command, '--version').first.to_s.match(/\d+.\d+/)[0].to_f
rescue Errno::ENOENT
  0
end

Public Instance Methods

to_box() click to toggle source

# File lib/rtesseract.rb, line 23
def to_box
  Box.run(@source, @errors, @config)
end

to_pdf() click to toggle source

# File lib/rtesseract.rb, line 31
def to_pdf
  Pdf.run(@source, @errors, @config)
end

to_s() click to toggle source

Output value

# File lib/rtesseract.rb, line 40
def to_s
  Text.run(@source, @errors, @config)
end

to_s_without_spaces() click to toggle source

Remove spaces and break-lines

# File lib/rtesseract.rb, line 45
def to_s_without_spaces
  to_s.gsub(/\s/, '')
end

to_tsv() click to toggle source

# File lib/rtesseract.rb, line 35
def to_tsv
  Tsv.run(@source, @errors, @config)
end

words() click to toggle source

# File lib/rtesseract.rb, line 27
def words
  to_box.map { |word| word[:word] }
end