add test for overwrite flag

This commit is contained in:
Nick Sweeting 2020-08-18 04:52:56 -04:00
parent a3155b88a1
commit a218ceb4e8

View file

@ -8,26 +8,58 @@ def test_depth_flag_is_accepted(process, disable_extractors_dict):
capture_output=True, env=disable_extractors_dict)
assert 'unrecognized arguments: --depth' not in arg_process.stderr.decode("utf-8")
def test_depth_flag_fails_if_it_is_not_0_or_1(process, disable_extractors_dict):
arg_process = subprocess.run(["archivebox", "add", "http://127.0.0.1:8080/static/example.com.html", "--depth=5"],
capture_output=True, env=disable_extractors_dict)
arg_process = subprocess.run(
["archivebox", "add", "--depth=5", "http://127.0.0.1:8080/static/example.com.html"],
capture_output=True,
env=disable_extractors_dict,
)
assert 'invalid choice' in arg_process.stderr.decode("utf-8")
arg_process = subprocess.run(["archivebox", "add", "http://127.0.0.1:8080/static/example.com.html", "--depth=-1"],
capture_output=True, env=disable_extractors_dict)
arg_process = subprocess.run(
["archivebox", "add", "--depth=-1", "http://127.0.0.1:8080/static/example.com.html"],
capture_output=True,
env=disable_extractors_dict,
)
assert 'invalid choice' in arg_process.stderr.decode("utf-8")
def test_depth_flag_0_crawls_only_the_arg_page(tmp_path, process, disable_extractors_dict):
arg_process = subprocess.run(["archivebox", "add", "http://127.0.0.1:8080/static/example.com.html", "--depth=0"],
capture_output=True, env=disable_extractors_dict)
arg_process = subprocess.run(
["archivebox", "add", "--depth=0", "http://127.0.0.1:8080/static/example.com.html"],
capture_output=True,
env=disable_extractors_dict,
)
archived_item_path = list(tmp_path.glob('archive/**/*'))[0]
with open(archived_item_path / "index.json", "r") as f:
output_json = json.load(f)
assert output_json["base_url"] == "127.0.0.1:8080/static/example.com.html"
def test_depth_flag_1_crawls_the_page_AND_links(tmp_path, process, disable_extractors_dict):
arg_process = subprocess.run(["archivebox", "add", "http://127.0.0.1:8080/static/example.com.html", "--depth=1"],
capture_output=True, env=disable_extractors_dict)
arg_process = subprocess.run(
["archivebox", "add", "--depth=1", "http://127.0.0.1:8080/static/example.com.html"],
capture_output=True,
env=disable_extractors_dict,
)
with open(tmp_path / "index.json", "r") as f:
archive_file = f.read()
assert "http://127.0.0.1:8080/static/example.com.html" in archive_file
assert "http://127.0.0.1:8080/static/iana.org.html" in archive_file
def test_overwrite_flag_is_accepted(process, disable_extractors_dict):
subprocess.run(
["archivebox", "add", "--depth=0", "http://127.0.0.1:8080/static/example.com.html"],
capture_output=True,
env=disable_extractors_dict,
)
arg_process = subprocess.run(
["archivebox", "add", "--overwrite", "http://127.0.0.1:8080/static/example.com.html"],
capture_output=True,
env=disable_extractors_dict,
)
assert 'unrecognized arguments: --overwrite' not in arg_process.stderr.decode("utf-8")
assert 'favicon' in arg_process.stdout.decode('utf-8'), 'archive methods probably didnt run, did overwrite work?'