Class: TwitterCldr::Shared::Hyphenator

Inherits:
Object
  • Object
show all
Defined in:
lib/twitter_cldr/shared/hyphenator.rb

Defined Under Namespace

Classes: UnsupportedLocaleError

Constant Summary collapse

BASE_RESOURCE_PATH =
%w(shared hyphenation).freeze
DEFAULT_LEFT_HYPHEN_MIN =
2
DEFAULT_RIGHT_HYPHEN_MIN =
2
DEFAULT_NO_HYPHEN =
"-'’".freeze

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(rules, locale, options) ⇒ Hyphenator

Returns a new instance of Hyphenator.



71
72
73
74
75
76
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 71

def initialize(rules, locale, options)
  @rules = rules
  @locale = locale
  @options = options
  @trie = build_trie_from(rules)
end

Instance Attribute Details

#localeObject (readonly)

Returns the value of attribute locale.



69
70
71
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 69

def locale
  @locale
end

#optionsObject (readonly)

Returns the value of attribute options.



69
70
71
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 69

def options
  @options
end

#rulesObject (readonly)

Returns the value of attribute rules.



69
70
71
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 69

def rules
  @rules
end

#trieObject (readonly)

Returns the value of attribute trie.



69
70
71
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 69

def trie
  @trie
end

Class Method Details

.get(locale) ⇒ Object



19
20
21
22
23
24
25
26
27
28
29
30
31
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 19

def get(locale)
  locale = find_supported_locale(locale)

  unless locale
    raise UnsupportedLocaleError,
      "'#{locale}' is not a supported hyphenation locale"
  end

  cache[locale] ||= begin
    resource = resource_for(locale)
    new(resource[:rules], locale, resource[:options])
  end
end

.supported_locale?(locale) ⇒ Boolean

Returns:

  • (Boolean)


33
34
35
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 33

def supported_locale?(locale)
  !!find_supported_locale(locale)
end

.supported_localesObject



37
38
39
40
41
42
43
44
45
46
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 37

def supported_locales
  @supported_locales ||= begin
    absolute_resource_path = TwitterCldr.absolute_resource_path(
      File.join(BASE_RESOURCE_PATH)
    )

    files = Dir.glob(File.join(absolute_resource_path, '*.yml'))
    files.map { |f| File.basename(f).chomp('.yml') }
  end
end

Instance Method Details

#each_chunk(text) ⇒ Object



83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 83

def each_chunk(text)
  if block_given?
    last_pos = 0

    each_position(text) do |pos|
      yield text[last_pos...pos].tap { last_pos = pos }
    end

    if last_pos < text.size
      yield text[last_pos..text.size]
    end
  else
    to_enum(__method__, text)
  end
end

#each_position(text) ⇒ Object



99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 99

def each_position(text)
  if block_given?
    text = ".#{text}."
    break_weights = break_weights_for(text)

    left = left_hyphen_min
    right = text.size - right_hyphen_min - 2

    (left...right).each do |idx|
      yield idx if break_weights[idx].odd?
    end
  else
    to_enum(__method__, text)
  end
end

#hyphenate(text, hyphen = "\u00AD") ⇒ Object

0x00AD is a soft hyphen



79
80
81
# File 'lib/twitter_cldr/shared/hyphenator.rb', line 79

def hyphenate(text, hyphen = "\u00AD")
  each_chunk(text).to_a.join(hyphen)
end