[KoCo-CVS] [Commit] KoreanCodecs/korean/python hangul.py

[Chang <pe...@us...>]

perky       02/05/01 04:10:44

  Modified:    korean/python hangul.py
  Log:
  - Test long unicode string for ishangul, isJaeum, isMoeum
  
  Suggested by: Lee Gang-Seong <gs...@gw...>
  
  Revision  Changes    Path
  1.10      +36 -7     KoreanCodecs/korean/python/hangul.py
  
  Index: hangul.py
  ===================================================================
  RCS file: /cvsroot/koco/KoreanCodecs/korean/python/hangul.py,v
  retrieving revision 1.9
  retrieving revision 1.10
  diff -u -r1.9 -r1.10
  --- hangul.py	29 Apr 2002 14:24:25 -0000	1.9
  +++ hangul.py	1 May 2002 11:10:44 -0000	1.10
  @@ -15,7 +15,7 @@
   # Conjoining Jamo Behavior:
   #  http://www.unicode.org/unicode/uni2book/ch03.pdf  (section 3.11)
   #
  -# $Id: hangul.py,v 1.9 2002/04/29 14:24:25 perky Exp $
  +# $Id: hangul.py,v 1.10 2002/05/01 11:10:44 perky Exp $
   #
   
   class UnicodeHangulError(Exception):
  @@ -29,6 +29,11 @@
       __str__ = __repr__
   
   Null = u''
  +try:
  +    True
  +except:
  +    True = 1
  +    False = 0
   
   class Jaeum:
   
  @@ -85,9 +90,6 @@
           exec "%s = %s" % (name, repr(code))
   del name, code
   
  -isJaeum = lambda c: c in Jaeum.Codes
  -isMoeum = lambda c: c in Moeum.Codes
  -
   # Unicode Hangul Syllables Characteristics
   ZONE = (u'\uAC00', u'\uD7A3')
   NCHOSUNG  = len(Chosung)
  @@ -99,7 +101,7 @@
   CHOSUNG_FILLER = u'\u115F'
   JUNGSUNG_FILLER = u'\u1160'
   
  -ishangul = (
  +_ishangul = (
       lambda code:
           ZONE[0] <= code <= ZONE[1] or
           code in Jaeum.Codes or
  @@ -125,6 +127,33 @@
       u'(\uc778)': (4, u'\uc778'),       # (in)-
   }
   
  +def isJaeum(u):
  +    if u:
  +        for c in u:
  +            if c not in Jaeum.Codes:
  +                break
  +        else:
  +            return True
  +    return False
  +
  +def isMoeum(u):
  +    if u:
  +        for c in u:
  +            if c not in Moeum.Codes:
  +                break
  +        else:
  +            return True
  +    return False
  +
  +def ishangul(u):
  +    if u:
  +        for c in u:
  +            if not _ishangul(c):
  +                break
  +        else:
  +            return True
  +    return False
  +
   def join(codes):
       """ Join function which makes hangul syllable from jamos """
       if len(codes) is not 3:
  @@ -141,7 +170,7 @@
   
   def split(code):
       """ Split function which splits hangul syllable into jamos """
  -    if len(code) != 1 or not ishangul(code):
  +    if len(code) != 1 or not _ishangul(code):
           raise UnicodeHangulError("needs 1 hangul letter")
       if code in Jaeum.Codes:
           return (code, Null, Null)
  @@ -185,7 +214,7 @@
   def disjoint(s):
       obuff = []
       for c in s:
  -        if ishangul(c):
  +        if _ishangul(c):
               cho, jung, jong = split(c)
               if cho:
                   obuff.append( unichr(ord(JBASE_CHOSUNG) + Chosung.index(cho)) )