def outputGraph(self, outf, compute_all_edges=False):
"""
For debugging purposes:
outputs the similarity/relevance graph to a file.
by default, only outputs the spanning graph for each connected component
if compute_all_edges is true, we recompute all the similarities and output the full graph
"""
print('nodedef>name VARCHAR,label VARCHAR,pid VARCHAR,visibility VARCHAR,relevance FLOAT,relevant VARCHAR', file=outf)
for (x,v) in self.authors.items():
visibility = v.paper.visibility
if v.paper.year <= 2012:
visibility = 'NOT_LABELLED'
print(nocomma([x,unidecode(v.paper.title), v.paper.id,
visibility, self.relevance.get(x,None),self.relevance.get(x,-1.)>0.]), file=outf)
print('edgedef>node1 VARCHAR,node2 VARCHAR', file=outf)
if not compute_all_edges:
for (x,y) in self.edge_set:
if y != None:
print(nocomma([x,y]), file=outf)
else:
author_pks = list(self.authors.keys())
for i in range(len(author_pks)):
for j in range(i):
x = author_pks[i]
y = author_pks[j]
if self.classify(x, y):
print(nocomma([x,y]), file=outf)
def outputGraph(self, author_set, outfile, logf):
print('nodedef>name VARCHAR,label VARCHAR,pid VARCHAR,visibility VARCHAR', file=outfile)
author_data = []
for author in author_set:
paper = author.paper
visibility = paper.visibility
if paper.year <= 2012:
visibility = 'NOT_LABELLED'
author_data.append(self.lstData(author))
print(nocomma([author.pk, unidecode(paper.title), paper.id, visibility]), file=outfile)
print('edgedef>node1 VARCHAR,node2 VARCHAR', file=outfile)
authors = list(author_set)
for i in range(len(authors)):
for j in range(i):
output = self.classifyData(author_data[i],author_data[j])
print(str(authors[i].pk)+"-"+str(authors[j].pk)+"\t"+str(output), file=logf)
if output:
print(nocomma([authors[i].pk,authors[j].pk]), file=outfile)
def test_nocomma(self):
self.assertEqual(nocomma(['a','b','cd']), 'a,b,cd')
self.assertEqual(nocomma(['a,','b']), 'a,b')
self.assertEqual(nocomma(['abc','','\n','def']), 'abc, , ,def')
