def testParse(sentence_file):
"""Parse sentences from sentences.txt with preprocessing and postprocessing, printing them out one by one
Arguments: none
Returns: none, prints out list of trees
"""
grammar = nltk.data.load('file:gaelic.fcfg')
cp = nltk.FeatureChartParser(grammar)
sentences = gaelicProc.preprocessSentences(sentence_file)
parse_errors = []
single_parses = []
multiple_parses = 0
print "\n\n" + sentence_file
for sentence in sentences:
trees = cp.nbest_parse(sentence.split())
if not trees:
parse_errors = parse_errors + [sentence] #track parse errors
else:
if len(trees) > 1: #report sentences with multiple parses
print '\n' + gaelicProc.postprocessSentences(sentence).decode("utf8") #print out reverted orthographic form of the sentence
for tree in trees:
print tree
multiple_parses += 1
else:
single_parses = single_parses + [sentence] #track parse successes
if multiple_parses == 0:
print "CHA ROBH ROSGRAINN LE BARRACHD IS 1 TORADH PARSAIDH ANN"
else:
print "BARRACHD IS 1 TORADH PARSAIDH: " + str(multiple_parses)
#report parse statistics
print "\nROSGRAINN UILE: " + str(len(sentences))
#display sentences with single parses
print "\n1 TORADH PARSAIDH A-MHÀIN: " + str(len(single_parses))
for sentence in single_parses:
print gaelicProc.postprocessSentences(sentence).decode("utf8")
#display sentences with parse errors
print "\nCHA GHABH PARSADH: " + str(len(parse_errors))
for sentence in parse_errors:
print gaelicProc.postprocessSentences(sentence).decode("utf8")
def parseNow(sentence_file):
"""Parse sentences from sentences.txt with preprocessing and postprocessing, printing them out one by one
Arguments: none
Returns: none, prints out list of trees
"""
#[TODO] THIS ATTEMPTS TO CREATE A FEATURE GRAMMAR IN UTF8 - BUT IS GIVING PARSE ERRORS IN NLTK.GRAMMAR
#grammar = nltk.grammar.parse_fcfg(codecs.open('gaelic.fcfg', 'rb', encoding='utf8').read())
#THIS WORKS BUT WITHOUT THE UTF8 ENCODING
grammar = nltk.data.load('file:gaelic.fcfg')
cp = nltk.FeatureChartParser(grammar)
sentences = gaelicProc.preprocessSentences(sentence_file)
for sentence in sentences:
#this is where the first-pass correction could be applied i.e. sentence = filterOrthography(sentence)
rev = '\n' + gaelicProc.postprocessSentences(sentence) #or add .decode('utf8') if that's desired, the trees print in ASCII on a PC anyway
print rev #print out reverted orthographic form of the sentence
trees = cp.nbest_parse(sentence.split())
if not trees: print 'CHA GHABH AN ROSGRANN PARSADH.'
else:
for tree in trees:
print tree
