def lm_file(self):
output = "# Parameters 1\n"
output += "MainLM\tMainLM\n"
corpora = Set()
for key, data in self.lmdata[0].items():
output += "# LMDefinition " + str(len(data)) + "\n"
output += "Name\t" + key + "\n"
for field, value in data.items():
# The coordinates and default LM
# are saved so we can reconstruct the LM in the browser,
# but don't output them in the LM file
if (field=="coords" or field=="default_lm"):
continue
if isinstance(value, list):
if(value[0] == 'lm'):
output += field + "\t" + str(value[1]) + "\n"
elif(value[0] == 'corpus'):
n = str(data["M"])
value.append("2")
output += "Tree\tTree" + value[1] + "_" + n + "\n"
corpora.add((value[1], n))
else:
output += field + "\t" + str(value) + "\n"
for corpus in corpora:
c = Corpus.objects.get(id=corpus[0])
output += "# TreeDefinition 2\n"
output += "Name\tTree" + corpus[0] + "_" + str(corpus[1]) + "\n"
tree_file = files.file_name(c.name) + ".M" + str(corpus[1]) + ".cnt"
output += "File " + os.path.join(files.CLUSTER_CORPORA_DIR, files.TREE_PATH, str(c.id), tree_file) + "\n"
return output
def file_name(self):
return files.file_name(self.corpus.name) + ".M" + str(self.m) + ".cnt"