class RepositoryAnalyzer:
"""Class to extract code language distribution from a software repository
:params details: if enable, it returns fine-grained results
:param kind: the analyzer kind (e.g., LINGUIST, CLOC)
"""
def __init__(self, details=False, kind=LINGUIST):
self.details = details
self.kind = kind
if kind == LINGUIST:
self.analyzer = Linguist()
else:
self.analyzer = Cloc()
def analyze(self, repository_path):
"""Analyze the content of a repository using Linguist
:param repository_path: repository path
:returns a dict containing the results of the analysis, like the one below
(for instance, repository is based on Python programming language entirely)
{
'Python': 100.0
}
"""
kwargs = {'repository_path': repository_path, 'details': self.details}
if self.kind == CLOC:
kwargs['file_path'] = repository_path
kwargs['repository_level'] = True
analysis = self.analyzer.analyze(**kwargs)
return analysis
def test_analyze_error(self, check_output_mock):
"""Test whether an exception is thrown in case of errors"""
check_output_mock.side_effect = subprocess.CalledProcessError(-1, "command", output=b'output')
cloc = Cloc()
kwargs = {'file_path': os.path.join(self.tmp_data_path, ANALYZER_TEST_FILE)}
with self.assertRaises(GraalError):
_ = cloc.analyze(**kwargs)
class FileAnalyzer:
"""Class to analyse the content of files"""
ALLOWED_EXTENSIONS = ['java', 'py', 'php', 'scala', 'js', 'rb', 'cs', 'cpp', 'c', 'lua', 'go', 'swift']
FORBIDDEN_EXTENSIONS = ['tar', 'bz2', "gz", "lz", "apk", "tbz2",
"lzma", "tlz", "war", "xar", "zip", "zipx"]
def __init__(self, details=False, kind=LIZARD_FILE):
self.details = details
self.kind = kind
if self.kind == LIZARD_FILE:
self.cloc = Cloc()
self.lizard = Lizard()
else:
self.scc = SCC()
def analyze(self, file_path):
"""Analyze the content of a file using CLOC, Lizard and SCC
:param file_path: file path
:returns a dict containing the results of the analysis, like the one below
{
'blanks': ..,
'comments': ..,
'loc': ..,
'ccn': ..,
'avg_ccn': ..,
'avg_loc': ..,
'avg_tokens': ..,
'num_funs': ..,
'tokens': ..,
'funs': [..]
}
"""
kwargs = {'file_path': file_path}
if self.kind == LIZARD_FILE:
cloc_analysis = self.cloc.analyze(**kwargs)
if GraalRepository.extension(file_path) not in self.ALLOWED_EXTENSIONS:
return cloc_analysis
kwargs['details'] = self.details
file_analysis = self.lizard.analyze(**kwargs)
# the LOC returned by CLOC is replaced by the one obtained with Lizard
# for consistency purposes
file_analysis['blanks'] = cloc_analysis['blanks']
file_analysis['comments'] = cloc_analysis['comments']
else:
file_analysis = self.scc.analyze(**kwargs)
return file_analysis
def test_analyze(self):
"""Test whether cloc returns the expected fields data"""
cloc = Cloc()
kwargs = {'file_path': os.path.join(self.tmp_data_path, ANALYZER_TEST_FILE)}
result = cloc.analyze(**kwargs)
self.assertIn('blanks', result)
self.assertTrue(type(result['blanks']), int)
self.assertIn('comments', result)
self.assertTrue(type(result['comments']), int)
self.assertIn('loc', result)
self.assertTrue(type(result['loc']), int)
def test_analyze_repository_level(self):
"""Test whether cloc returns the expected fields data for repository level"""
cloc = Cloc()
kwargs = {'file_path': self.origin_path, 'repository_level': True}
results = cloc.analyze(**kwargs)
result = results[next(iter(results))]
self.assertIn('blanks', result)
self.assertTrue(type(result['blanks']), int)
self.assertIn('comments', result)
self.assertTrue(type(result['comments']), int)
self.assertIn('loc', result)
self.assertTrue(type(result['loc']), int)
self.assertIn('total_files', result)
self.assertTrue(type(result['total_files']), int)
def __analyze_repository(self, repository_path, files_affected, details):
"""Add code complexity information for a given repository
using Lizard and CLOC.
Current information includes cyclomatic complexity (ccn),
lines of code, number of functions, tokens, blanks and comments.
:param repository_path: repository path
:param details: if True, it returns fine-grained results
:returns result: list of the results of the analysis
"""
analysis_result = []
repository_analysis = lizard.analyze(
paths=[repository_path],
threads=1,
exts=lizard.get_extensions([]),
)
cloc = Cloc()
for analysis in repository_analysis:
cloc_analysis = cloc.analyze(file_path=analysis.filename)
file_path = analysis.filename.replace(repository_path + "/", '')
in_commit = True if file_path in files_affected else False
result = {
'loc': analysis.nloc,
'ccn': analysis.CCN,
'tokens': analysis.token_count,
'num_funs': len(analysis.function_list),
'file_path': file_path,
'in_commit': in_commit,
'blanks': cloc_analysis['blanks'],
'comments': cloc_analysis['comments']
}
analysis_result.append(result)
# TODO: implement details option
return analysis_result