def getDependencies(input_file, output_file, git_hub_key, hop, partitions):
dependency_data = {}
no_dependency_data = []
gitLink = []
gh = xkcd2347.GitHub(key=git_hub_key)
result = []
def process_line(link):
link = link.rstrip()
owner, repo = link.split(".com/")[1].split("/")
result = [
dep['packageName'] for dep in gh.get_dependencies(
repo_owner=owner, repo_name=repo, depth=hop)
]
if (len(result) > 0):
return (link, result)
else:
return (link, ["No dependency"])
try:
sc = SparkContext("local[*]", "PySpark Tutorial")
result = sc.textFile(input_file,
partitions).map(process_line).collect()
except:
print("Error running spark job")
with open(output_file, 'w') as fp:
json.dump(dict(result), fp, indent=4)
sc.stop()
def getDependencies(input_file, output_file, git_hub_key, hop):
dependency_data = {}
no_dependency_data = []
gitLink = []
gh = xkcd2347.GitHub(key=git_hub_key)
with open(input_file, 'r') as fp:
gitLink = fp.readlines()
with open(output_file, 'a') as fp:
for link in tqdm(gitLink):
link = link.rstrip()
owner, repo = link.split(".com/")[1].split("/")
result = [
dep['packageName'] for dep in gh.get_dependencies(
repo_owner=owner, repo_name=repo, depth=hop)
]
dependency_data = {}
if (len(result) > 0):
dependency_data[link] = result
json.dump(dependency_data, fp)
fp.write("\n")
else:
no_dependency_data.append(link)
with open('no_result_data_' + output_file, 'w') as fp:
json.dump(no_dependency_data, fp)
def test_cache():
cache_dir = pathlib.Path('test-cache')
if cache_dir.exists():
shutil.rmtree(cache_dir)
cache_dir.mkdir()
cache = diskcache.Cache(cache_dir)
gh = xkcd2347.GitHub(key=key, cache=cache)
deps = list(gh.get_dependencies('docnow', 'twarc'))
assert len(deps) > 0
assert deps[0]['repository']['owner']['login']
def getDependencies(input_file, output_file, git_hub_key, hop):
dependency_data = {}
no_dependency_data = []
gitLink = []
with open(input_file, 'r') as fp:
gitLink = fp.readlines()
link = gitLink[0]
#for link in tqdm(gitLink):
gh = xkcd2347.GitHub(key=git_hub_key)
link = link.rstrip()
owner, repo = link.split(".com/")[1].split("/")
result = [
dep['packageName'] for dep in gh.get_dependencies(
repo_owner=owner, repo_name=repo, depth=hop)
]
if (len(result) > 0):
dependency_data[link] = result
with open(output_file, 'w') as fp:
json.dump(dependency_data, fp, indent=4)
def test_get_dependencies():
gh = xkcd2347.GitHub(key=key)
deps = list(gh.get_dependencies('docnow', 'twarc'))
assert len(deps) > 0
assert deps[0]['repository']['owner']['login']
