def nlp_nlidb(question):
'''STEP1: Extracting the keywords using Parser:'''
tree = stanford_client.to_tree(question)
top_node = penn_treebank_node.parse(tree)
extracted_words, Proper_Nouns = my_parser.key_words(top_node, question)
question_type = my_parser.questionType(top_node)
'''STEP2: Replace words with glossary terms:'''
uniqueWords = glossary.generalizedKeywords(question, extracted_words)
# '''STEP2.2: Extracting the related keywords:'''
uniqueWords.append(question_type)
'''STEP3: Adding some manually defined rules'''
allWords, conditions, target, component_value = answerGenerator(question, uniqueWords)
# To be handled differently, this is here only so the master branch can run.
allWords = filter(lambda x: x != None, allWords)
allWords = allWords + list(target)
allWords = set(allWords)
'''Creating Links between allWords and tables' entities'''
tables = set(semanticNet.tables(allWords))
'''required_values is not being used in our system anymore'''
required_values =''
debug.debug_statement([allWords, required_values, target, conditions, tables, question_type, Proper_Nouns, component_value])
return allWords, required_values, target, conditions, tables, question_type, Proper_Nouns, component_value
def extract_rules(self, tree):
tree_rules = []
tree = ptn.parse(tree)
queue = Queue()
queue.push(tree)
i = 0
while not queue.empty():
t = queue.pop()
rule = Rule(t.node_type, [])
for child in t.children:
queue.push(child)
rule.add_child(child.node_type)
tree_rules.append(rule)
if None in tree_rules:
raw_input('>>>')
return np.array(tree_rules)
