def parseSurfaceSemantics(sss_str):
if '_' not in sss_str: return []
text,POS,senses = splitSurfaceSemantics(sss_str)
try:
return [pywordnet.getWord(text,POS).getSenses()[int(s)-1] for s in senses]
except (IndexError,KeyError):
sense = None
for altPOS in ('N','V','ADJ','ADV'):
if altPOS == POS: continue
try:
return [pywordnet.getWord(text,POS).getSenses()[int(s)-1] for s in senses]
except (IndexError,KeyError): pass
return []
def parseSurfaceSemantics(sss_str):
if "_" not in sss_str:
return []
text, POS, senses = splitSurfaceSemantics(sss_str)
try:
return [pywordnet.getWord(text, POS).getSenses()[int(s) - 1] for s in senses]
except (IndexError, KeyError):
sense = None
for altPOS in ("N", "V", "ADJ", "ADV"):
if altPOS == POS:
continue
try:
return [pywordnet.getWord(text, POS).getSenses()[int(s) - 1] for s in senses]
except (IndexError, KeyError):
pass
return []
def extractSurfaceSemantics(token,parent):
global Senses
POS=getPartOfSpeech(token,parent)
tokenSenses = {}
text = token['TEXT'].lower()
default = token['TEXT'].upper()
if POS in ['N', 'V', 'ADV', 'ADJ']:
try: #Redo as test = foo while not tokenSensesword: try: foo ; except KeyError: foo = next foo
tokenSenses = Senses[text]
except KeyError:
logger.warning('extractSurfaceSemantics : Text not in tagged senses: %s', text)
try:
#logger.warning('extractSurfaceSemantics : Previously unseen word but in WordNet?: %s', text)
# stringified range of possible senses without spaces
tokenSenses = {POS : range(1,len(pywordnet.getWord(text,POS).getSenses())+1)}
except KeyError:
try:
logger.warning('extractSurfaceSemantics : Inflected version of WordNet word? %s', text)
if text.endswith('s'):
text = text[:-1]
tokenSenses = Senses[text]
else:
stemmer = PorterStemmer() # Update WordNetStemmer to NLTK 1.4 API
stemmer.stem(token)
text = token['STEM']
tokenSenses = Senses[text]
except KeyError:
text = token['TEXT'].lower()
try:
logger.warning('extractSurfaceSemantics : Misspelling / typo of WordNet word? %s', text)
spellchecker = enchant.DictWithPWL('en_US', Lexicon)
s = ''
for s in spellchecker.suggest(text):
if s in Senses:
tokenSenses = Senses[s]
break
if not tokenSenses and spellchecker.suggest(text):
s = spellchecker.suggest(text)[0]
tokenSenses = {POS : range(1,len(pywordnet.getWord(s,POS).getSenses())+1)}
if s and Options.Spellcheck:
logger.warning('extractSurfaceSemantics : Found spelling correction %s for %s', s,text)
text = s
#logger.debug('*** extractSurfaceSemantics : Implement spelling correction. *** ')
#raise KeyError
except KeyError:
logger.error('extractSurfaceSemantics : Unknown token: %s', text)
return default
# Handle experienced typos.
if 'see' in tokenSenses:
### FIXME adding to dict for typos that are other words
text = tokenSenses['see']
try:
tokenSenses = Senses[text]
except: return default
# Handle morphology variants that wordnet understands.
elif isinstance(tokenSenses, tuple):
text,tokenSenses[POS] = tokenSenses[POS]
try:
return '_'.join([text,POS,','.join([str(i) for i in tokenSenses[POS]])])
except KeyError:
#logger.warning('extractSurfaceSemantics : Expected POS %s for token %s, Got %s, Using %s',
# POS, token, tokenSenses.keys(), tokenSenses.keys()[0])
if tokenSenses.keys():
POS = token['POS'] = tokenSenses.keys()[0]
return '_'.join([text,POS,','.join([str(i) for i in tokenSenses.values()[0]])])
except Exception,e:
logger.error('extractSurfaceSemantics: %s: Could not find sense %s for token %s',
e, POS, token) #tokenSenses, text
def extractSurfaceSemantics(token, parent):
global Senses
POS = getPartOfSpeech(token, parent)
tokenSenses = {}
text = token["TEXT"].lower()
default = token["TEXT"].upper()
if POS in ["N", "V", "ADV", "ADJ"]:
try: # Redo as test = foo while not tokenSensesword: try: foo ; except KeyError: foo = next foo
tokenSenses = Senses[text]
except KeyError:
logger.warning("extractSurfaceSemantics : Text not in tagged senses: %s", text)
try:
# logger.warning('extractSurfaceSemantics : Previously unseen word but in WordNet?: %s', text)
# stringified range of possible senses without spaces
tokenSenses = {POS: range(1, len(pywordnet.getWord(text, POS).getSenses()) + 1)}
except KeyError:
try:
logger.warning("extractSurfaceSemantics : Inflected version of WordNet word? %s", text)
if text.endswith("s"):
text = text[:-1]
tokenSenses = Senses[text]
else:
stemmer = PorterStemmer() # Update WordNetStemmer to NLTK 1.4 API
stemmer.stem(token)
text = token["STEM"]
tokenSenses = Senses[text]
except KeyError:
text = token["TEXT"].lower()
try:
logger.warning("extractSurfaceSemantics : Misspelling / typo of WordNet word? %s", text)
spellchecker = enchant.DictWithPWL("en_US", Lexicon)
s = ""
for s in spellchecker.suggest(text):
if s in Senses:
tokenSenses = Senses[s]
break
if not tokenSenses and spellchecker.suggest(text):
s = spellchecker.suggest(text)[0]
tokenSenses = {POS: range(1, len(pywordnet.getWord(s, POS).getSenses()) + 1)}
if s and Options.Spellcheck:
logger.warning("extractSurfaceSemantics : Found spelling correction %s for %s", s, text)
text = s
# logger.debug('*** extractSurfaceSemantics : Implement spelling correction. *** ')
# raise KeyError
except KeyError:
logger.error("extractSurfaceSemantics : Unknown token: %s", text)
return default
# Handle experienced typos.
if "see" in tokenSenses:
### FIXME adding to dict for typos that are other words
text = tokenSenses["see"]
try:
tokenSenses = Senses[text]
except:
return default
# Handle morphology variants that wordnet understands.
elif isinstance(tokenSenses, tuple):
text, tokenSenses[POS] = tokenSenses[POS]
try:
return "_".join([text, POS, ",".join([str(i) for i in tokenSenses[POS]])])
except KeyError:
# logger.warning('extractSurfaceSemantics : Expected POS %s for token %s, Got %s, Using %s',
# POS, token, tokenSenses.keys(), tokenSenses.keys()[0])
if tokenSenses.keys():
POS = token["POS"] = tokenSenses.keys()[0]
return "_".join([text, POS, ",".join([str(i) for i in tokenSenses.values()[0]])])
except Exception, e:
logger.error(
"extractSurfaceSemantics: %s: Could not find sense %s for token %s", e, POS, token
) # tokenSenses, text
