def candidate_index_sets(self, found_indices):
candidates = []
abstractions = []
for index in found_indices:
abstractions = abstractions + self.all_abstractions(index)
for index in abstractions:
candidates = candidates + self.index_sets.get(index, [])
# print 'candidates for %s are %s' % (found_indices, candidates)
return utils.remove_duplicates(candidates)
def parse(self, text, debug=0):
"""Parses a string. Returns the list of valid parses."""
if not isinstance(text, basestring):
raise TypeError('%r is not a string.' % (text, ))
results = self.parse_tokens(self.tokenize(text), debug=debug)
if len(results) > 1:
results = utils.remove_duplicates(results)
for result in results:
result.text = text
return results
def parse(self, text, debug=0):
"""Parses a string. Returns the list of valid parses."""
if not isinstance(text, basestring):
raise TypeError('%r is not a string.' % (text,))
results = self.parse_tokens(self.tokenize(text), debug=debug)
if len(results) > 1:
results = utils.remove_duplicates(results)
for result in results:
result.text = text
return results
def candidate_index_sets(self, found_indices):
candidates = []
abstractions = []
for index in found_indices:
abstractions = abstractions + self.all_abstractions(index)
for index in abstractions:
candidates = candidates + self.index_sets.get(index, [])
# print 'candidates for %s are %s' % (found_indices, candidates)
return utils.remove_duplicates(candidates)
def parse_tokens(self, tokens, debug=0):
self.debug = debug
self.cp_parser.debug = debug
indices = self.find_indices(tokens, self.match_function)
if debug > 0:
print 'ICP parsing tokens %s' % (tokens,)
if debug > 1:
print 'ICP found indices %s' % (indices,)
results = self.score_index_sets(indices)
results.sort(key=lambda x: x.score, reverse=True)
results = [result for result in results if result.score > CUTOFF_ICP_SCORE]
results = utils.remove_duplicates(
results,
lambda r1, r2: r1.target_concept == r2.target_concept)
if debug > 0:
print 'ICP results: %s' % (results,)
return results
def parse_tokens(self, tokens, debug=0):
self.debug = debug
self.cp_parser.debug = debug
indices = self.find_indices(tokens, self.match_function)
if debug > 0:
print 'ICP parsing tokens %s' % (tokens, )
if debug > 1:
print 'ICP found indices %s' % (indices, )
results = self.score_index_sets(indices)
results.sort(key=lambda x: x.score, reverse=True)
results = [
result for result in results if result.score > CUTOFF_ICP_SCORE
]
results = utils.remove_duplicates(
results, lambda r1, r2: r1.target_concept == r2.target_concept)
if debug > 0:
print 'ICP results: %s' % (results, )
return results
