-- 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>
--   &gt;&gt;&gt; hyphenate english_US "supercalifragilisticexpialadocious"
--   ["su","per","cal","ifrag","ilis","tic","ex","pi","al","ado","cious"]
--   </pre>
--   
--   <pre>
--   &gt;&gt;&gt; hyphenate english_US "hyphenation"
--   ["hy","phen","ation"]
--   </pre>
--   
--   <pre>
--   &gt;&gt;&gt; hyphenate icelandic "va\240lahei\240avegavinnuverkf\230rageymslusk\250r"
--   ["va\240la","hei\240a","vega","vinnu","verk","f\230ra","geymslu","sk\250r"]
--   </pre>
@package hyphenation
@version 0.4


module Text.Hyphenation.Exception

-- | Hyphenation exceptions are special cases that should use the specified
--   hyphenation points.
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

-- | Try to find a matching hyphenation exception.
lookupException :: String -> Exceptions -> Maybe [Int]

-- | Convert an exception string to a score.
scoreException :: String -> [Int]

-- | Parse one exception per line from an input string
parseExceptions :: String -> Exceptions
instance Show Exceptions
instance Monoid Exceptions


module Text.Hyphenation.Pattern

-- | Hyphenation patterns
data Patterns

-- | Insert a Knuth-Liang hyphenation pattern into the trie
--   
--   <ol>
--   <li><tt>.</tt> denotes the start or end of the input</li>
--   <li><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).</li>
--   </ol>
--   
--   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]

-- | Convert a Pattern string to a Score
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

-- | A <tt>Hyphenator</tt> is combination of an alphabet normalization
--   scheme, a set of <a>Patterns</a>, a set of <a>Exceptions</a> to those
--   patterns and a number of characters at each end to skip hyphenating.
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>
--   &gt;&gt;&gt; import Text.Hyphenation
--   </pre>
--   
--   <pre>
--   &gt;&gt;&gt; hyphenate english_US "supercalifragilisticexpialadocious"
--   ["su","per","cal","ifrag","ilis","tic","ex","pi","al","ado","cious"]
--   </pre>
--   
--   <pre>
--   &gt;&gt;&gt; hyphenate english_US "hyphenation"
--   ["hy","phen","ation"]
--   </pre>
hyphenate :: Hyphenator -> String -> [String]

-- | By default, do not insert hyphens in the first two characters
--   
--   <pre>
--   &gt;&gt;&gt; defaultLeftMin
--   2
--   </pre>
defaultLeftMin :: Int

-- | By default, do not insert hyphens in the last three characters.
--   
--   <pre>
--   &gt;&gt;&gt; defaultRightMin
--   3
--   </pre>
defaultRightMin :: Int


module Text.Hyphenation.Language

-- | A strongly typed set of available languages you can use for
--   hyphenation.
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
Finnish :: Language
French :: Language
Friulan :: 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
Piedmontese :: Language
Polish :: Language
Portuguese :: Language
Romanian :: Language
Romansh :: 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
Thai :: Language
Turkish :: Language
Turkmen :: Language
Ukrainian :: Language
Uppersorbian :: Language
Welsh :: Language

-- | Load (and cache) the hyphenator for a given language.
languageHyphenator :: Language -> Hyphenator

-- | Hyphenators for a wide array of languages.
afrikaans :: Hyphenator

-- | Hyphenators for a wide array of languages.
basque :: Hyphenator

-- | Hyphenators for a wide array of languages.
bengali :: Hyphenator

-- | Hyphenators for a wide array of languages.
bulgarian :: Hyphenator

-- | Hyphenators for a wide array of languages.
catalan :: Hyphenator

-- | Hyphenators for a wide array of languages.
chinese :: Hyphenator

-- | Hyphenators for a wide array of languages.
coptic :: Hyphenator

-- | Hyphenators for a wide array of languages.
croatian :: Hyphenator

-- | Hyphenators for a wide array of languages.
czech :: Hyphenator

-- | Hyphenators for a wide array of languages.
danish :: Hyphenator

-- | Hyphenators for a wide array of languages.
dutch :: Hyphenator

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

-- | <pre>
--   &gt;&gt;&gt; hyphenate english_GB "supercalifragilisticexpialadocious"
--   ["su","per","cal","i","fra","gil","istic","ex","pi","alado","cious"]
--   </pre>
--   
--   favors UK hyphenation
english_GB :: Hyphenator

-- | Hyphenators for a wide array of languages.
esperanto :: Hyphenator

-- | Hyphenators for a wide array of languages.
estonian :: Hyphenator

-- | Hyphenators for a wide array of languages.
ethiopic :: Hyphenator

-- | Hyphenators for a wide array of languages.
finnish :: Hyphenator

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

-- | Hyphenators for a wide array of languages.
friulan :: Hyphenator

-- | Hyphenators for a wide array of languages.
galician :: Hyphenator

-- | Hyphenators for a wide array of languages.
german_1901 :: Hyphenator

-- | Hyphenators for a wide array of languages.
german_1996 :: Hyphenator

-- | Hyphenators for a wide array of languages.
german_Swiss :: Hyphenator

-- | Hyphenators for a wide array of languages.
greek_Ancient :: Hyphenator

-- | Hyphenators for a wide array of languages.
greek_Mono :: Hyphenator

-- | Hyphenators for a wide array of languages.
greek_Poly :: Hyphenator

-- | Hyphenators for a wide array of languages.
gujarati :: Hyphenator

-- | Hyphenators for a wide array of languages.
hindi :: Hyphenator

-- | Hyphenators for a wide array of languages.
hungarian :: Hyphenator

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

-- | Hyphenators for a wide array of languages.
indonesian :: Hyphenator

-- | Hyphenators for a wide array of languages.
interlingua :: Hyphenator

-- | Hyphenators for a wide array of languages.
irish :: Hyphenator

-- | Hyphenators for a wide array of languages.
italian :: Hyphenator

-- | Hyphenators for a wide array of languages.
kannada :: Hyphenator

-- | Hyphenators for a wide array of languages.
kurmanji :: Hyphenator

-- | Hyphenators for a wide array of languages.
lao :: Hyphenator

-- | Hyphenators for a wide array of languages.
latin :: Hyphenator

-- | Hyphenators for a wide array of languages.
latvian :: Hyphenator

-- | Hyphenators for a wide array of languages.
lithuanian :: Hyphenator

-- | Hyphenators for a wide array of languages.
malayalam :: Hyphenator

-- | Hyphenators for a wide array of languages.
marathi :: Hyphenator

-- | Hyphenators for a wide array of languages.
mongolian :: Hyphenator

-- | Hyphenators for a wide array of languages.
norwegian_Bokmal :: Hyphenator

-- | Hyphenators for a wide array of languages.
norwegian_Nynorsk :: Hyphenator

-- | Hyphenators for a wide array of languages.
oriya :: Hyphenator

-- | Hyphenators for a wide array of languages.
panjabi :: Hyphenator

-- | Hyphenators for a wide array of languages.
piedmontese :: Hyphenator

-- | Hyphenators for a wide array of languages.
polish :: Hyphenator

-- | Hyphenators for a wide array of languages.
portuguese :: Hyphenator

-- | Hyphenators for a wide array of languages.
romanian :: Hyphenator

-- | Hyphenators for a wide array of languages.
romansh :: Hyphenator

-- | Hyphenators for a wide array of languages.
russian :: Hyphenator

-- | Hyphenators for a wide array of languages.
sanskrit :: Hyphenator

-- | Hyphenators for a wide array of languages.
serbian_Cyrillic :: Hyphenator

-- | Hyphenators for a wide array of languages.
serbocroatian_Cyrillic :: Hyphenator

-- | Hyphenators for a wide array of languages.
serbocroatian_Latin :: Hyphenator

-- | Hyphenators for a wide array of languages.
slovak :: Hyphenator

-- | Hyphenators for a wide array of languages.
slovenian :: Hyphenator

-- | Hyphenators for a wide array of languages.
spanish :: Hyphenator

-- | Hyphenators for a wide array of languages.
swedish :: Hyphenator

-- | Hyphenators for a wide array of languages.
tamil :: Hyphenator

-- | Hyphenators for a wide array of languages.
telugu :: Hyphenator

-- | Hyphenators for a wide array of languages.
thai :: Hyphenator

-- | Hyphenators for a wide array of languages.
turkish :: Hyphenator

-- | Hyphenators for a wide array of languages.
turkmen :: Hyphenator

-- | Hyphenators for a wide array of languages.
ukrainian :: Hyphenator

-- | Hyphenators for a wide array of languages.
uppersorbian :: Hyphenator

-- | Hyphenators for a wide array of languages.
welsh :: 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