def write_to_code_file(mode, data, path_to_load, path_to_export,
path_raw_code):
g = data.grammar
nt = {v: reverse_typename(k) for k, v in g.node_type_to_id.items()}
#print(nt,g.node_type_to_id)
v = data.terminal_vocab
raw = []
with open(path_raw_code, 'r') as f:
for line in f:
raw.append(line[:-1])
with open(path_to_load, 'r') as f:
l = json.load(f, encoding='utf8')
l_code = []
for i in range(len(l)):
# print(raw[i])
try:
t = ASTNode.from_dict(l[i], nt, v)
ast_tree = parse.decode_tree_to_python_ast(t)
code = astor.to_source(ast_tree)[:-1]
real_code = parse.de_canonicalize_code(code, raw[i])
if (mode == "hs"):
real_code = " ".join(parse.tokenize_code_adv(
real_code, True)).replace("\n", "#NEWLINE#").replace(
"#NEWLINE# ", "").replace("#INDENT# ", "")
real_code = " ".join(parse.tokenize_code_adv(real_code, False))
#print(real_code,raw[i])
l_code.append(real_code)
except:
print "Tree %d impossible to parse" % (i)
l_code.append("")
with open(path_to_export, 'w') as f:
for c in l_code:
f.write(c + "\n")
return reverse_typename(t[:-1])
else:
return vars(ast)[t]
if __name__ == '__main__':
flag = "hs"
path_to_load = "../data/exp/results/test_hs_10_iter.json"
if flag == "django":
train_data, dev_data, test_data = deserialize_from_file("../../django.cleaned.dataset.freq5.par_info.refact.space_only.bin")
elif flag == "hs":
train_data, dev_data, test_data = deserialize_from_file("../../hs.freq3.pre_suf.unary_closure.bin")
data = test_data
g = data.grammar
nt = {v:reverse_typename(k) for k,v in g.node_type_to_id.items()}
#print(nt,g.node_type_to_id)
v = data.terminal_vocab
results = []
with open(path_to_load,'r') as f:
l = json.load(f, encoding='utf8')
for i in range(len(l)):
t = ASTNode.from_dict(l[i], nt,v)
ast_tree = parse.decode_tree_to_python_ast(t)
results.append(ast_tree)
evaluate_decode_results(flag, test_data, results, verbose=True)
