Merge pull request neubig#8 from dbc148/master

shift reduce arc standard oracle

11-parsing/oracle.py

@@ -0,0 +1,61 @@
+import pdb
+class Word:
+	def __init__(self, word, location, head = -1):
+		self.word = word
+		self.location = location
+		self.head = head
+		self.rightmost_child = -1
+
+#abreviated conll file like gold.dev.txt
+def read_abrv(file_name):
+	f = [l.strip() for l in open(file_name)]
+	sents = []
+	sent = []
+	loc = 0
+	for line in f:
+		if line != '':
+			line = line.split()
+			word = line[0]
+			head = int(line[1])
+			sent.append(Word(word, loc, head))
+			loc += 1
+		else:
+			for word in sent:
+				if word.location > sent[word.head].rightmost_child:
+					sent[word.head].rightmost_child = word.location
+			sents.append(sent)
+			sent = []
+			loc = 0
+	return sents
+
+sentences = read_abrv('../data/parsing/gold.txt')
+actions_for_sents = []
+f = open('../data/parsing/output.txt', 'w')
+for sent in sentences:
+	stack, buffer = [], []
+	acts = []
+	for word in sent:
+		buffer.append(word)
+	buffer = list(reversed(buffer))
+	while len(buffer) > 0 or len(stack) > 1:
+		if len(stack) < 2:
+			stack.append(buffer.pop())
+			acts.append('SHIFT')
+		elif stack[-1].head == stack[-2].location  and (len(buffer) == 0 or stack[-1].rightmost_child < buffer[-1].location or stack[-2].rightmost_child == -1):
+			acts.append('REDUCE_R')
+			stack.pop()
+		elif stack[-2].head == stack[-1].location and (len(buffer) == 0 or stack[-2].rightmost_child < buffer[-1].location or stack[-2].rightmost_child == -1):
+			acts.append('REDUCE_L')
+			temp = stack.pop()
+			stack.pop()
+			stack.append(temp)
+		elif len(buffer) > 0:
+			stack.append(buffer.pop())
+			acts.append('SHIFT')
+		else:
+			break
+	actions_for_sents.append(acts)
+	f.write(' '.join([s.word for s in sent]) + ' ||| ' + ' '.join(acts) + '\n')
+
+f.close()
+