class Text::Hyphen

An object that knows how to perform hyphenation based on the TeX hyphenation algorithm with pattern files. Each object is constructed with a specific language's hyphenation patterns.

Constants

DEBUG
DEFAULT_MIN_LEFT
DEFAULT_MIN_RIGHT
VERSION

Attributes

iso_language[R]

Returns the language's ISO 639 ID, e.g., “en_us” or “pt”.

language[RW]

The name of the language to be used in hyphenating words. This will be a two or three character ISO 639 code, with the two character form being the canonical resource name. This will load the language hyphenation definitions from text/hyphen/language/<code> as a Ruby class. The resource 'text/hyphen/language/en_us' defines the language class Text::Hyphen::Language::EN_US. It also defines the secondary forms Text::Hyphen::Language::EN and Text::Hyphen::Language::ENG_US.

Minimal transformations will be performed on the language code provided, such that any dashes are converted to underscores (e.g., 'en-us' becomes 'en_us') and all characters are regularised. Resource names will be downcased and class names will be converted to uppercase (e.g., 'Pt' for the Portuguese language becomes 'pt' and 'PT', respectively).

The language may also be specified as an instance of Text::Hyphen::Language.

left[RW]

No fewer than this number of letters will show up to the left of the hyphen. The initial value for this will be specified by the language; setting this value will override the language's defaults.

right[RW]

No fewer than this number of letters will show up to the right of the hyphen. This overrides the default specified in the language.

Public Class Methods

new(options = {}) { |self| ... } click to toggle source

Creates a hyphenation object with the options requested. The options available are:

language

The language to perform hyphenation with. See language and iso_language.

left

The minimum number of characters to the left of a hyphenation point. See left.

right

The minimum number of characters to the right of a hyphenation point. See right.

The options can be provided either as hashed parameters or set as methods in an initialization block. The following initializations are all equivalent:

hyp = Text::Hyphenate.new(:language => 'en_us')
hyp = Text::Hyphenate.new(language: 'en_us') # under Ruby 1.9
hyp = Text::Hyphenate.new { |h| h.language = 'en_us' }
# File lib/text/hyphen.rb, line 75
def initialize(options = {}) # :yields self:
  @iso_language = options[:language]
  @left         = options[:left]
  @right        = options[:right]
  @language     = nil

  @cache        = {}
  @vcache       = {}

  @hyphen       = {}
  @begin_hyphen = {}
  @end_hyphen   = {}
  @both_hyphen  = {}
  @exception    = {}

  @first_load = true
  yield self if block_given?
  @first_load = false

  load_language

  @left  ||= DEFAULT_MIN_LEFT
  @right ||= DEFAULT_MIN_RIGHT
end

Public Instance Methods

clear_cache!() click to toggle source

Clears the per-instance hyphenation and visualization caches.

# File lib/text/hyphen.rb, line 173
def clear_cache!
  @cache.clear
  @vcache.clear
end
hyphenate(word) click to toggle source

Returns an array of character positions where a word can be hyphenated.

hyp.hyphenate('representation') #=> [3, 5, 8 10]

Because hyphenation can be expensive, if the word has been hyphenated previously, it will be returned from a per-instance cache.

# File lib/text/hyphen.rb, line 106
def hyphenate(word)
  word = word.downcase
  $stderr.puts "Hyphenating #{word}" if DEBUG
  return @cache[word] if @cache.has_key?(word)
  res = @language.exceptions[word]
  return @cache[word] = make_result_list(res) if res

  letters = word.scan(@language.scan_re)
  $stderr.puts letters.inspect if DEBUG
  word_size = letters.size

  result = [0] * (word_size + 1)
  right_stop = word_size - @right

  updater = Proc.new do |hash, str, pos|
    if hash.has_key?(str)
      $stderr.print "#{pos}: #{str}: #{hash[str]}" if DEBUG
      hash[str].scan(@language.scan_re).each_with_index do |cc, ii|
        cc = cc.to_i
        result[ii + pos] = cc if cc > result[ii + pos]
      end
      $stderr.print ": #{result.inspect}\n" if DEBUG
    end
  end

    # Walk the word
  (0..right_stop).each do |pos|
    rest_length = word_size - pos
    (1..rest_length).each do |length|
      substr = letters[pos, length].join('')
      updater[@language.hyphen, substr, pos]
      updater[@language.start, substr, pos] if pos.zero?
      updater[@language.stop, substr, pos] if (length == rest_length)
    end
  end

  updater[@language.both, word, 0] if @language.both[word]

  (0..@left).each { |i| result[i] = 0 }
  ((-1 - @right)..(-1)).each { |i| result[i] = 0 }
  @cache[word] = make_result_list(result)
end
hyphenate_to(word, size, hyphen = '-') click to toggle source

This function will hyphenate a word so that the first point is at most

NOTE: if hyphen is set to a string, it will still be counted as one character (since it represents a hyphen)

size characters.

# File lib/text/hyphen.rb, line 184
def hyphenate_to(word, size, hyphen = '-')
  point = hyphenate(word).delete_if { |e| e >= size }.max
  if point.nil?
    [nil, word]
  else
    [word[0 ... point] + hyphen, word[point .. -1]]
  end
end
stats() click to toggle source

Returns a string describing the structure of the patterns for the language of this hyphenation object.

# File lib/text/hyphen.rb, line 195
  def stats
    _b = @language.both.size
    _s = @language.start.size
    _e = @language.stop.size
    _h = @language.hyphen.size
    _x = @language.exceptions.size
    _T = _b + _s + _e + _h + _x

    s = <<-EOS

The language '%s' contains %d total hyphenation patterns.
    % 6d patterns are word start patterns.
    % 6d patterns are word stop patterns.
    % 6d patterns are word start/stop patterns.
    % 6d patterns are normal patterns.
    % 6d patterns are exceptions.

EOS
    s % [ @iso_language, _T, _s, _e, _b, _h, _x ]
  end
visualise(word, hyphen = '-') click to toggle source

Returns a visualization of the hyphenation points.

hyp.visualize('representation') #=> rep-re-sen-ta-tion

Any string can be set instead of the default hyphen:

hyp.visualize('example', '&shy;') #=> exam&shy;ple

Because hyphenation can be expensive, if the word has been visualised previously, it will be returned from a per-instance cache.

# File lib/text/hyphen.rb, line 159
def visualise(word, hyphen = '-')
  return @vcache[word] if @vcache.has_key?(word)
  w = word.dup
  s = hyphen.size
  hyphenate(w).each_with_index do |pos, n|
    # Insert the hyphen string at the ported position plus the offset of
    # the last hyphen string inserted.
    w[pos.to_i + (n * s), 0] = hyphen unless pos.zero?
  end
  @vcache[word] = w
end
Also aliased as: visualize
visualize(word, hyphen = '-')
Alias for: visualise

Private Instance Methods

load_language() click to toggle source
# File lib/text/hyphen.rb, line 235
def load_language
  return if @first_load

  @iso_language ||= "en_us"

  unless @language
    require "text/hyphen/language/#{@iso_language}"
    @language = Text::Hyphen::Language.const_get(@iso_language.upcase)
    @iso_language = @language.isocode if @language.isocode
  end

  @left ||= @language.left
  @right ||= @language.right

  @iso_language
end
make_result_list(res) click to toggle source
# File lib/text/hyphen.rb, line 228
def make_result_list(res)
  r = []
  res.each_with_index { |c, i| r <<  i * (c.to_i % 2) }
  r.reject { |i| i.to_i == 0 }
end
updateresult(hash, str, pos) click to toggle source
# File lib/text/hyphen.rb, line 216
def updateresult(hash, str, pos)
  if hash.has_key?(str)
    STDERR.print "#{pos}: #{str}: #{hash[str]}" if DEBUG
    hash[str].scan(@language.scan_re).each_with_index do |c, i|
      c = c.to_i
      @result[i + pos] = c if c > @result[i + pos]
    end
    STDERR.puts ": #{@result}" if DEBUG
  end
end