-- Hoogle documentation, generated by Haddock
-- See Hoogle, http://www.haskell.org/hoogle/


-- | Configurable Knuth-Liang hyphenation
--   
--   Configurable Knuth-Liang hyphenation
--   
--   Uses the UTF8 encoded hyphenation patterns provided by hyph-utf8 from
--   <a>http://www.ctan.org/tex-archive/language/hyph-utf8</a>
--   
--   Usage:
--   
--   <pre>
--   ghci&gt; hyphenate english_US "supercalifragilisticexpialadocious"
--   ["su","per","cal","ifrag","ilis","tic","ex","pi","al","ado","cious"]
--   </pre>
--   
--   <pre>
--   ghci&gt; hyphenate english_US "hyphenation"
--   ["hy","phen","ation"]
--   </pre>
--   
--   <pre>
--   ghci&gt; hyphenate icelandic "vaðlaheiðavegavinnuverkfærageymsluskúr"
--   ["va\240la","hei\240a","vega","vinnu","verk","f\230ra","geymslu","sk\250r"]
--   </pre>
@package hyphenation
@version 0.2


module Text.Hyphenation.Exception
data Exceptions

-- | add an exception to the exception table. if it is already present,
--   this will restrict the set of hyphenations to the intersection of the
--   set provided and the set present.
addException :: String -> Exceptions -> Exceptions
lookupException :: String -> Exceptions -> Maybe [Int]
scoreException :: String -> [Int]
parseExceptions :: String -> Exceptions
instance Show Exceptions
instance Monoid Exceptions


module Text.Hyphenation.Pattern
data Patterns

-- | Insert a Knuth-Liang hyphenation pattern into the trie . * <tt>.</tt>
--   denotes the start or end of the input . * <tt>0-9</tt> are used to
--   denote hyphenation or dehyphenation depending on whether or not they
--   are even (no hyphen) or odd (hyphen allowed). . Patterns are overlaid
--   and the maximum value at each location is used. this allows you to
--   implement a finite number of precedences between hyphenation rules
--   
--   (e.g. <tt>3foo.</tt> indicates that the suffix '-foo' should be
--   hyphenated with precedence 3.)
insertPattern :: String -> Patterns -> Patterns

-- | Tallies the hyphenation scores for a word considering all tails.
lookupPattern :: String -> Patterns -> [Int]
scorePattern :: String -> [Int]

-- | Parse one pattern per line from an input string
--   
--   <tt>hyph-utf8</tt> supplies these files UTF-8 encoded in the
--   <tt>txt</tt> folder with a <tt>.pat.txt</tt> extension
parsePatterns :: String -> Patterns
instance Show Patterns
instance Monoid Patterns


-- | Hyphenation based on the Knuth-Liang algorithm as used by TeX.
module Text.Hyphenation.Hyphenator
data Hyphenator
Hyphenator :: (Char -> Char) -> Patterns -> Exceptions -> {-# UNPACK #-} !Int -> {-# UNPACK #-} !Int -> Hyphenator

-- | a normalization function applied to input characters before applying
--   patterns or exceptions
hyphenatorChars :: Hyphenator -> Char -> Char

-- | hyphenation patterns stored in a trie
hyphenatorPatterns :: Hyphenator -> Patterns

-- | exceptions to the general hyphenation rules, hyphenated manually
hyphenatorExceptions :: Hyphenator -> Exceptions

-- | the number of characters as the start of a word to skip hyphenating,
--   by default: 2
hyphenatorLeftMin :: Hyphenator -> {-# UNPACK #-} !Int

-- | the number of characters at the end of the word to skip hyphenating,
--   by default: 3
hyphenatorRightMin :: Hyphenator -> {-# UNPACK #-} !Int

-- | hyphenate a single word using the specified Hyphenator. Returns a set
--   of candidate breakpoints by decomposing the input into substrings.
--   
--   <pre>
--   ghci&gt; hyphenate english_US "supercalifragilisticexpialadocious"
--   ["su","per","cal","ifrag","ilis","tic","ex","pi","al","ado","cious"]
--   ghci&gt; hyphenate english_US "hyphenation"
--   ["hy","phen","ation"]
--   </pre>
hyphenate :: Hyphenator -> String -> [String]
defaultLeftMin, defaultRightMin :: Int


module Text.Hyphenation.Language
data Language
Afrikaans :: Language
Basque :: Language
Bengali :: Language
Bulgarian :: Language
Catalan :: Language
Chinese :: Language
Coptic :: Language
Croatian :: Language
Czech :: Language
Danish :: Language
Dutch :: Language
English_US :: Language
English_GB :: Language
Esperanto :: Language
Estonian :: Language
Ethiopic :: Language
Farsi :: Language
Finnish :: Language
French :: Language
Galician :: Language
German_1901 :: Language
German_1996 :: Language
German_Swiss :: Language
Greek_Ancient :: Language
Greek_Mono :: Language
Greek_Poly :: Language
Gujarati :: Language
Hindi :: Language
Hungarian :: Language
Icelandic :: Language
Indonesian :: Language
Interlingua :: Language
Irish :: Language
Italian :: Language
Kannada :: Language
Kurmanji :: Language
Lao :: Language
Latin :: Language
Latvian :: Language
Lithuanian :: Language
Malayalam :: Language
Marathi :: Language
Mongolian :: Language
Norwegian_Bokmal :: Language
Norwegian_Nynorsk :: Language
Oriya :: Language
Panjabi :: Language
Polish :: Language
Portuguese :: Language
Romanian :: Language
Russian :: Language
Sanskrit :: Language
Serbian_Cyrillic :: Language
Serbocroatian_Cyrillic :: Language
Serbocroatian_Latin :: Language
Slovak :: Language
Slovenian :: Language
Spanish :: Language
Swedish :: Language
Tamil :: Language
Telugu :: Language
Turkish :: Language
Turkmen :: Language
Ukrainian :: Language
Uppersorbian :: Language
Welsh :: Language
languageHyphenator :: Language -> Hyphenator
afrikaans, welsh, uppersorbian, ukrainian, turkmen, turkish, telugu, tamil, swedish, spanish, slovenian, slovak, serbocroatian_Latin, serbocroatian_Cyrillic, serbian_Cyrillic, sanskrit, russian, romanian, portuguese, polish, panjabi, oriya, norwegian_Nynorsk, norwegian_Bokmal, mongolian, marathi, malayalam, lithuanian, latvian, latin, lao, kurmanji, kannada, italian, irish, interlingua, indonesian, hungarian, hindi, gujarati, greek_Poly, greek_Mono, greek_Ancient, german_Swiss, german_1996, german_1901, galician, finnish, farsi, ethiopic, estonian, esperanto, dutch, danish, czech, croatian, coptic, chinese, catalan, bulgarian, bengali, basque :: Hyphenator

-- | <pre>
--   ghci&gt; hyphenate english_US "supercalifragilisticexpialadocious"
--   ["su","per","cal","ifrag","ilis","tic","ex","pi","al","ado","cious"]
--   </pre>
--   
--   favors US hyphenation
english_US :: Hyphenator

-- | <pre>
--   ghci&gt; hyphenate english_GB "supercalifragilisticexpialadocious"
--   ["su","per","cal","ifrag","ilis","tic","ex","pi","al","ado","cious"]
--   </pre>
--   
--   favors UK hyphenation
english_GB :: Hyphenator

-- | <pre>
--   ghci&gt; hyphenate french "anticonstitutionnellement"
--   ["an","ti","cons","ti","tu","tion","nel","le","ment"]
--   </pre>
french :: Hyphenator

-- | <pre>
--   ghci&gt; hyphenate icelandic "vaðlaheiðavegavinnuverkfærageymsluskúr"
--   ["va\240la","hei\240a","vega","vinnu","verk","f\230ra","geymslu","sk\250r"]
--   </pre>
icelandic :: Hyphenator

-- | Read a built-in language file from the data directory where cabal
--   installed this package.
--   
--   (e.g. <tt>hyphenateLanguage "en-us"</tt> opens
--   <tt>"/Users/ekmett/.cabal/share/hyphenation-0.2/ghc-7.4.1/hyph-en-us.hyp.txt"</tt>
--   among others when run on the author's local machine)
loadHyphenator :: String -> IO Hyphenator

-- | the infix portion of the data file names used for this language
languageAffix :: Language -> String
instance Eq Language
instance Ord Language
instance Show Language
instance Bounded Language
instance Enum Language


-- | Hyphenation based on the Knuth-Liang algorithm as used by TeX.
--   
--   The implementation is based on Ned Batchelder's public domain
--   <tt>hyphenate.py</tt> and simplified to remove the need for a manual
--   exception list.
module Text.Hyphenation