add stats.py script to check containment of names in chunk_np

scripts/chunker_scripts/stats/stats.py

+#!/usr/bin/python
+# -*- coding: utf-8 -*-
+import corpus2
+
+files = '/home/macias/kpwr-1.1-nm-chunk-ccl-folds/ccl-test10.xml'
+
+def main():
+	tagset = corpus2.get_named_tagset('nkjp')
+	reader = corpus2.TokenReader.create_path_reader('ccl', tagset, files)
+	sent = reader.get_next_sentence()
+	wyniki_o = []	
+	while sent:
+		asent = corpus2.AnnotatedSentence.wrap_sentence(sent)
+#		print sent.id(), sent.size()
+		sent_size = sent.size()
+		tab = []		
+		for chan_name in asent.all_channels():
+			if chan_name == 'chunk_np' or chan_name.endswith("_nam") :		
+				chan = asent.get_channel(chan_name)				
+				ann_vec = chan.make_annotation_vector()
+				row = [chan_name, ann_vec]
+				tab.append(row)			
+#				for ann in ann_vec:
+#					inds = sorted(ann.indices)				
+#					print inds
+		if len(tab) > 0 :
+			wyniki_o.append(check_sentence(tab, sent_size))
+		sent = reader.get_next_sentence()
+ 	ok = 0
+	err = 0
+	no_np = 0
+	for results in wyniki_o:
+		ok = ok + results[0]
+		if results[2]:
+			err = err + results[1]
+		else:
+			no_np = no_np + results[1]
+	print files	
+	print "Nazwy własne znajdujące się poprawnie w chunk_np: ",ok 
+	print "Nazwy własne nie zawierające sie w chunk_np: ",err
+	print "Brak chunk_np w zdaniu:", no_np		
+
+def check_sentence(tab, size):
+	chunknp = []	
+	otherchunks = []	
+	wyniki = []	
+	ok = 0
+	err = 0
+	np = False
+	for ann in tab:
+		if ann[0] == "chunk_np":
+			chunknp = create_row(ann, size)
+			
+		else:
+			otherchunks.append(create_row(ann, size))
+
+	if len(chunknp) > 0:
+		for chunk in otherchunks:
+			wyniki.append(isOK(chunknp, chunk))
+	
+		ok = 0
+		err = 0
+	
+		for results in wyniki:
+			ok = ok + results[0]
+			err = err + results[1]
+		np = True
+	else:
+		for chunk in otherchunks:
+			wyniki.append(count_all_nonzero_nam(chunk))
+		np = False
+	ok = 0
+	err = 0
+	
+	for results in wyniki:
+		ok = ok + results[0]
+		err = err + results[1]
+
+	
+	return [ok, err, np]	
+
+def count_all_nonzero_nam(chunk):
+	tab = []
+	err = 0
+	for toks in chunk:
+		if toks != 0 and have_value(toks, tab) and isinstance( toks, int ):	
+			err += 1
+			tab.append(toks)	
+	return [0, err]
+	
+def create_row(ann, size):
+	row = []
+	x = 0	
+	while x < size:
+		row.append(0) 
+		x += 1
+	row.append(ann[0])
+	i = 0	
+	for chun in ann[1]:
+		inds = sorted(chun.indices)				
+		i += 1
+		for element in inds:
+			row[element] = i	
+	return row
+
+
+def isOK(comparer, chunks):
+	i = 0	
+	err = 0
+	ok = 0	
+	tab = []
+#	print comparer
+#	print chunks
+	for toks in chunks:
+		if toks != 0 and have_value(toks, tab) and isinstance( toks, int ):	
+			tab.append(toks)
+
+	for nam_id in tab:
+		if check_one_channel(comparer, chunks, nam_id):	
+			ok += 1
+		else:
+			err += 1
+	return [ok, err]
+
+def check_one_channel(comparer, chunks, nam_id):
+	i = 0	
+	tab = []	
+	for toks in chunks:
+		if toks == nam_id:
+			tab.append(comparer[i])
+		i += 1
+	value = tab[0]
+	for v in tab:
+		if v != value:
+			return False
+	return True
+
+def have_value(toks, tab):
+	for v in tab:
+		if v == toks:
+			return False
+	return True
+
+	
+if __name__ == '__main__':
+    main()
+