def test_get_indent(self):
from utils.line_features import get_indent
self.assertEqual(get_indent("int a = 3; "), 0)
self.assertEqual(get_indent(" int a = 3; "), 1)
self.assertEqual(get_indent(" int a = 3; "), 4)
self.assertEqual(get_indent("\t\tint a = 3; "), 2)
def compute(self, code, annotations):
indentations = {} # how far indented levels should be
lines = code.lines
levels = get_indent_levels(lines)
# find the distribution of indentations for each indent level
for line_no, line in enumerate(lines):
_levels = levels[line_no]
if _levels == None: # lines ignored because they are comments
continue
indent = get_indent(line)
for level in _levels:
if level not in indentations:
indentations[level] = [indent]
else:
indentations[level].append(indent)
# find the most common indentations for each indent level
# ensuring that each successive level is deeper
try:
common_indentation = {}
for level, indents in indentations.iteritems():
s = set(indents)
if level == 0:
common_indentation[level] = max(s, key=indents.count)
else:
last_indentation = common_indentation[level - 1]
most_common = max(s, key=indents.count)
# if indent is not deeper
if most_common <= last_indentation:
# TODO(fil): pass error when indent is not deeper
while most_common <= last_indentation:
s -= set([most_common])
if len(s) == 0:
most_common = max(set(indents), key=indents.count)
break
most_common = max(s, key=indents.count)
common_indentation[level] = most_common
except KeyError as e:
print "Parsing error in feature_indentation (%s)" % e
# find which lines don't match the expected indentation
for line_no, line in enumerate(lines):
_levels = levels[line_no]
if _levels == None:
continue
expected_indents = []
for level in _levels:
try:
expected_indents.append(common_indentation[level])
except KeyError:
continue
indent = get_indent(line)
if expected_indents and indent not in expected_indents:
error = "expected indent %i, got %i" % (expected_indents[0], indent)
self.add_to_annotations(line_no, error, annotations)