def _generatePickledLexicalUnitsIndex(self):
"""
Initialises all the frames
"""
baseDir = FRAMENET_PATH + '/' + LU_DIR_ENV
pickledLUPath = FRAMENET_PATH + '/' + PICKLED_LU_FILE
lexicalUnitsIndexByName = {}
for _f in os.listdir(baseDir):
if _f.lower().startswith('lu') and _f.lower().endswith('.xml'):
print >> sys.stderr, 'Loading:', _f, ": ",
lu = LexicalUnit()
lu.loadXML(baseDir + '/' + _f)
print >> sys.stderr, lu['name'], '...',
print >> sys.stderr, 'done'
if lexicalUnitsIndexByName.has_key(lu['name']):
lexicalUnitsIndexByName[lu['name']][lu['ID']] = 1
else:
lexicalUnitsIndexByName[lu['name']] = {lu['ID']:1}
print >> sys.stderr, 'Saving the pickled lu files ...',
cPickle.dump(lexicalUnitsIndexByName, open(pickledLUPath, 'w'), cPickle.HIGHEST_PROTOCOL)
print >> sys.stderr, 'done'
pass
def _generatePickledLexicalUnitsIndex(self):
"""
Initialises all the frames
"""
baseDir = FRAMENET_PATH + '/' + LU_DIR_ENV
pickledLUPath = FRAMENET_PATH + '/' + PICKLED_LU_FILE
lexicalUnitsIndexByName = {}
for _f in os.listdir(baseDir):
if _f.lower().startswith('lu') and _f.lower().endswith('.xml'):
print >> sys.stderr, 'Loading:', _f, ": ",
lu = LexicalUnit()
lu.loadXML(baseDir + '/' + _f)
print >> sys.stderr, lu['name'], '...',
print >> sys.stderr, 'done'
if lexicalUnitsIndexByName.has_key(lu['name']):
lexicalUnitsIndexByName[lu['name']][lu['ID']] = 1
else:
lexicalUnitsIndexByName[lu['name']] = {lu['ID']: 1}
print >> sys.stderr, 'Saving the pickled lu files ...',
cPickle.dump(lexicalUnitsIndexByName, open(pickledLUPath, 'w'),
cPickle.HIGHEST_PROTOCOL)
print >> sys.stderr, 'done'
pass
def testLU(fileName):
a = LexicalUnit()
try:
if not a.loadXML(FRAMENET_PATH+"/"+LU_DIR_ENV+"/"+fileName):
print >> sys.stderr, 'loading:', fileName, 'failed'
except:
print >> sys.stderr, 'loading:', fileName, 'failed'
print a.getLexemes()
return a
def testLU(fileName):
a = LexicalUnit()
try:
if not a.loadXML(FRAMENET_PATH + "/" + LU_DIR_ENV + "/" + fileName):
print >> sys.stderr, 'loading:', fileName, 'failed'
except:
print >> sys.stderr, 'loading:', fileName, 'failed'
print a.getLexemes()
return a
def lookupLexicalUnit(self, headWord, pos):
"""
This function will look up a given word by its pos. The word must be already
lemmatised and in lower case. The pos must also be in lower case and it can be
one of the following:
(1) v -- for verb
(2) n -- for noun
(3) a -- for adjective
(4) adv -- for adverb
(5) prep -- for preposition
(6) num -- for numbers
(7) intj -- for interjections
This function will return a dictionary of lexical units which match the (headWord, pos)
pair. The keys to the dictionary will be the IDs of the lexical units, and the values of
the dictionary will be LexicalUnit objects.
"""
pickledLUPath = FRAMENET_PATH + '/' + LU_DIR_ENV
w = headWord + '.' + pos
if self['luCache'].has_key(w):
return self['luCache'][w]
if not self['luIndex'].has_key(w):
return {}
objects = {}
for _id in self['luIndex'][w].keys():
inputFile = pickledLUPath + '/lu' + str(_id) + '.xml'
lu = LexicalUnit()
lu.loadXML(inputFile)
objects[lu['ID']] = lu
self['luCache'][w] = objects
return objects
def lookupLexicalUnit(self, headWord, pos):
"""
This function will look up a given word by its pos. The word must be already
lemmatised and in lower case. The pos must also be in lower case and it can be
one of the following:
(1) v -- for verb
(2) n -- for noun
(3) a -- for adjective
(4) adv -- for adverb
(5) prep -- for preposition
(6) num -- for numbers
(7) intj -- for interjections
This function will return a dictionary of lexical units which match the (headWord, pos)
pair. The keys to the dictionary will be the IDs of the lexical units, and the values of
the dictionary will be LexicalUnit objects.
"""
pickledLUPath = FRAMENET_PATH + '/' + LU_DIR_ENV
w = headWord + '.' + pos
if self['luCache'].has_key(w):
return self['luCache'][w]
if not self['luIndex'].has_key(w):
return {}
objects = {}
for _id in self['luIndex'][w].keys():
inputFile = pickledLUPath + '/lu' + str(_id) + '.xml'
lu = LexicalUnit()
lu.loadXML(inputFile)
objects[lu['ID']] = lu
self['luCache'][w] = objects
return objects
