def composeHangulJamoToSyllable(L0,V0,T0=None):
if not ((L0 == None or isUnicodeChar(L0)) and isUnicodeChar(V0) and (T0 == None or isUnicodeChar(T0))):
raise Exception('Jamo not unicode characters: %s %s %s' % (L0,V0,T0 if T0 != None else str(T0)))
if L0 == None:
L0 = emptyJamo_L
if isHangulCompatibilityJamo(L0):
L = hangulCompatibilityJamoToJamo_L(L0)
else:
L = L0
if isHangulCompatibilityJamo(V0):
V = hangulCompatibilityJamoToJamo_V(V0)
else:
V = V0
if T0 != None:
if isHangulCompatibilityJamo(T0):
T = hangulCompatibilityJamoToJamo_T(T0)
else:
T = T0
else:
T = None # or do we want empty T pseudo Jamo?
LIndex = _hangulJamoToIndex_L(L)
VIndex = _hangulJamoToIndex_V(V)
TIndex = _hangulJamoToIndex_T(T)
cpSyll = (LIndex * VCount + VIndex) * TCount + TIndex + SBase
if not isCombinedS(cpSyll):
raise Exception("Codepoint produced by composition not Hangul syllable: " + codePointToCodePointString(cpSyll))
return codePointToCharacter(cpSyll)
def joinHangulChars(hg1LastChar,hg2FirstChar):
if not (isUnicodeChar(hg1LastChar) and isUnicodeChar(hg2FirstChar)):
raise Exception('Not unicode chars: ' + hg1LastChar + ' ' + hg2FirstChar )
if isHangulSyllable(hg1LastChar) and isHangulSyllable(hg2FirstChar):
return hg1LastChar + hg2FirstChar
if not isHangulSyllable(hg1LastChar) and not isHangulSyllable(hg2FirstChar):
raise Exception('Cannot join two non-syllables: ' + hg1LastChar + ' ' + hg2FirstChar)
if isHangulSyllable(hg1LastChar):
return joinHangulSyllableAndJamo(hg1LastChar,hg2FirstChar)
else:
return joinHangulJamoAndSyllable(hg1LastChar,hg2FirstChar)
def test_isUnicodeChar(self):
for obj, expected in ((u'a', True), (u'ab', False), ('a', False),
(u'', False), (None, False), (u'ㄱ', True),
(u'가', True), (u'가자', False), (u'㐀', True),
(u'𠀀', True)):
result = cu.isUnicodeChar(obj)
self.assertEquals(result, expected)
def isHangulChar(obj):
if not isUnicodeChar(obj):
return False
if isAsciiChar(obj): # Most likely non-hangul char type
return False
if isHangulSyllable(obj) or isHangulCompatibilityJamo(obj) or isHangulJamo(obj):
return True
return False
def test_conversions(self):
for char,expected in ((u'a','U+0061'),('a','U+0061'),(u'ㄱ','U+3131'),(u'가','U+AC00'),(u'㐀','U+3400'),(u'𠀀','U+20000')):
cp = cu.characterToCodepoint(char) # ord(char)
self.assertTrue(cu.codePointExists(cp))
char2 = cu.codePointToCharacter(cp)
self.assertTrue(cu.isUnicodeChar(char2))
self.assertEquals(char,char2)
cpStr = cu.codePointToCodePointString(cp)
self.assertEquals(cpStr,expected)
cp2 = cu.codePointStringToCodePoint(cpStr)
self.assertEquals(cp,cp2)
char3 = cu.codePointStringToCharacter(cpStr)
self.assertEquals(char3,char)
def hangulJamoToTrans(jamo):
if jamo == None: # Assume empty T
return ''
if not isUnicodeChar(jamo):
raise Exception('Not a unicode char: ' + unicode(jamo))
cp = ord(jamo)
if isL(cp):
return jamoTransList_L[cp - LBase]
elif isV(cp):
return jamoTransList_V[cp - VBase]
elif isT(cp):
return jamoTransList_T[cp - TBase]
else:
raise Exception('Not a Hangul Jamo: ' + codePointToCodePointString(cp) + ' ' +unicode(jamo))
def test_conversions(self):
for char, expected in ((u'a', 'U+0061'), ('a', 'U+0061'),
(u'ㄱ', 'U+3131'), (u'가', 'U+AC00'),
(u'㐀', 'U+3400'), (u'𠀀', 'U+20000')):
cp = cu.characterToCodepoint(char) # ord(char)
self.assertTrue(cu.codePointExists(cp))
char2 = cu.codePointToCharacter(cp)
self.assertTrue(cu.isUnicodeChar(char2))
self.assertEquals(char, char2)
cpStr = cu.codePointToCodePointString(cp)
self.assertEquals(cpStr, expected)
cp2 = cu.codePointStringToCodePoint(cpStr)
self.assertEquals(cp, cp2)
char3 = cu.codePointStringToCharacter(cpStr)
self.assertEquals(char3, char)
def joinJamoSeparatedHangul(hangul1,jamo, hangul2):
if hangul1 and not isUnicodeString(hangul1):
raise Exception('Not a unicode string: ' + hangul1)
if hangul2 and not isUnicodeString(hangul2):
raise Exception('Not a unicode string: ' + hangul2)
if not isUnicodeChar(jamo):
raise Exception('Not a unicode char: ' + hangul2)
if not hangul1 and not hangul2:
raise Exception('Adjacent Hangul strings both empty')
# First try to join jamo to preceding syll
if hangul1:
hg1LastChar = hangul1[-1]
try:
joinedSyll = joinHangulSyllableAndJamo(hg1LastChar,jamo)
joinedHangul = hangul1[:-1] + joinedSyll
if hangul2:
joinedHangul += hangul2
return joinedHangul
except:
pass # fall through
# If can't to preceding hangul, try to join to following jangul
if hangul2:
hg2FirstChar = hangul2[0]
try:
joinedSyll = joinHangulJamoAndSyllable(jamo,hg2FirstChar)
joinedHangul = joinedSyll + hangul2[1:]
if hangul1:
joinedHangul = hangul1 + joinedHangul
return joinedHangul
except:
pass # fall through
raise Exception('Can\'t join hangul sequence: %s %s %s' % (str(hangul1),str(jamo),str(hangul2)))
def test_isUnicodeChar(self):
for obj,expected in ((u'a',True),(u'ab',False),('a',False),(u'',False),(None,False),(u'ㄱ',True),(u'가',True),(u'가자',False),(u'㐀',True),(u'𠀀',True)):
result = cu.isUnicodeChar(obj)
self.assertEquals(result,expected)
def isHangulCompatibilityJamo(obj):
if not isUnicodeChar(obj):
return False
cp = ord(obj)
return isC(cp)
def isHangulJamo_T(obj):
if not isUnicodeChar(obj):
return False
cp = ord(obj)
return isT(cp)
def isHangulSyllable(obj):
if not isUnicodeChar(obj):
return False
cp = ord(obj)
return isCombinedS(cp)