alex/ArchiveBox
1
0
Fork 0
mirror of https://github.com/ArchiveBox/ArchiveBox.git synced 2026-04-06 07:47:53 +10:00
Code Issues Packages Projects Releases Wiki Activity

Fix CLI tests to use subprocess and remove mocks

Browse Source 
- Fix conftest.py: use subprocess for init, remove unused cli_env fixture
- Update all test files to use data_dir parameter instead of env
- Remove mock-based TestJSONLOutput class from tests_piping.py
- Remove unused imports (MagicMock, patch)
- Fix file permissions for cli_utils.py

All tests now use real subprocess calls per CLAUDE.md guidelines:
- NO MOCKS - tests exercise real code paths
- NO SKIPS - every test runs
This commit is contained in:
Claude
2025-12-31 10:53:45 +00:00
parent bb52b5902a
commit b87bbbbecb
6 changed files with 218 additions and 279 deletions
Download Patch File Download Diff File Expand all files Collapse all files

80
archivebox/tests/test_cli_snapshot.py
View File

@@ -22,13 +22,13 @@ from archivebox.tests.conftest import (
class TestSnapshotCreate:
"""Tests for `archivebox snapshot create`."""
def test_create_from_url_args(self, cli_env, initialized_archive):
def test_create_from_url_args(self, initialized_archive):
"""Create snapshot from URL arguments."""
url = create_test_url()
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'create', url],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0, f"Command failed: {stderr}"
@@ -39,19 +39,19 @@ class TestSnapshotCreate:
assert records[0]['type'] == 'Snapshot'
assert records[0]['url'] == url
def test_create_from_crawl_jsonl(self, cli_env, initialized_archive):
def test_create_from_crawl_jsonl(self, initialized_archive):
"""Create snapshots from Crawl JSONL input."""
url = create_test_url()
# First create a crawl
stdout1, _, _ = run_archivebox_cmd(['crawl', 'create', url], env=cli_env)
stdout1, _, _ = run_archivebox_cmd(['crawl', 'create', url], data_dir=initialized_archive)
crawl = parse_jsonl_output(stdout1)[0]
# Pipe crawl to snapshot create
stdout2, stderr, code = run_archivebox_cmd(
['snapshot', 'create'],
stdin=json.dumps(crawl),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0, f"Command failed: {stderr}"
@@ -65,20 +65,20 @@ class TestSnapshotCreate:
snapshot = next(r for r in records if r['type'] == 'Snapshot')
assert snapshot['url'] == url
def test_create_with_tag(self, cli_env, initialized_archive):
def test_create_with_tag(self, initialized_archive):
"""Create snapshot with --tag flag."""
url = create_test_url()
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'create', '--tag=test-tag', url],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
records = parse_jsonl_output(stdout)
assert 'test-tag' in records[0].get('tags_str', '')
def test_create_pass_through_other_types(self, cli_env, initialized_archive):
def test_create_pass_through_other_types(self, initialized_archive):
"""Pass-through records of other types unchanged."""
tag_record = {'type': 'Tag', 'id': 'fake-tag-id', 'name': 'test'}
url = create_test_url()
@@ -87,7 +87,7 @@ class TestSnapshotCreate:
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'create'],
stdin=stdin,
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -97,13 +97,13 @@ class TestSnapshotCreate:
assert 'Tag' in types
assert 'Snapshot' in types
def test_create_multiple_urls(self, cli_env, initialized_archive):
def test_create_multiple_urls(self, initialized_archive):
"""Create snapshots from multiple URLs."""
urls = [create_test_url() for _ in range(3)]
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'create'] + urls,
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -118,24 +118,24 @@ class TestSnapshotCreate:
class TestSnapshotList:
"""Tests for `archivebox snapshot list`."""
def test_list_empty(self, cli_env, initialized_archive):
def test_list_empty(self, initialized_archive):
"""List with no snapshots returns empty."""
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'list'],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
assert 'Listed 0 snapshots' in stderr
def test_list_returns_created(self, cli_env, initialized_archive):
def test_list_returns_created(self, initialized_archive):
"""List returns previously created snapshots."""
url = create_test_url()
run_archivebox_cmd(['snapshot', 'create', url], env=cli_env)
run_archivebox_cmd(['snapshot', 'create', url], data_dir=initialized_archive)
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'list'],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -143,14 +143,14 @@ class TestSnapshotList:
assert len(records) >= 1
assert any(r.get('url') == url for r in records)
def test_list_filter_by_status(self, cli_env, initialized_archive):
def test_list_filter_by_status(self, initialized_archive):
"""Filter snapshots by status."""
url = create_test_url()
run_archivebox_cmd(['snapshot', 'create', url], env=cli_env)
run_archivebox_cmd(['snapshot', 'create', url], data_dir=initialized_archive)
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'list', '--status=queued'],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -158,14 +158,14 @@ class TestSnapshotList:
for r in records:
assert r['status'] == 'queued'
def test_list_filter_by_url_contains(self, cli_env, initialized_archive):
def test_list_filter_by_url_contains(self, initialized_archive):
"""Filter snapshots by URL contains."""
url = create_test_url(domain='unique-domain-12345.com')
run_archivebox_cmd(['snapshot', 'create', url], env=cli_env)
run_archivebox_cmd(['snapshot', 'create', url], data_dir=initialized_archive)
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'list', '--url__icontains=unique-domain-12345'],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -173,14 +173,14 @@ class TestSnapshotList:
assert len(records) == 1
assert 'unique-domain-12345' in records[0]['url']
def test_list_with_limit(self, cli_env, initialized_archive):
def test_list_with_limit(self, initialized_archive):
"""Limit number of results."""
for _ in range(3):
run_archivebox_cmd(['snapshot', 'create', create_test_url()], env=cli_env)
run_archivebox_cmd(['snapshot', 'create', create_test_url()], data_dir=initialized_archive)
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'list', '--limit=2'],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -191,16 +191,16 @@ class TestSnapshotList:
class TestSnapshotUpdate:
"""Tests for `archivebox snapshot update`."""
def test_update_status(self, cli_env, initialized_archive):
def test_update_status(self, initialized_archive):
"""Update snapshot status."""
url = create_test_url()
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], env=cli_env)
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], data_dir=initialized_archive)
snapshot = parse_jsonl_output(stdout1)[0]
stdout2, stderr, code = run_archivebox_cmd(
['snapshot', 'update', '--status=started'],
stdin=json.dumps(snapshot),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -209,16 +209,16 @@ class TestSnapshotUpdate:
records = parse_jsonl_output(stdout2)
assert records[0]['status'] == 'started'
def test_update_add_tag(self, cli_env, initialized_archive):
def test_update_add_tag(self, initialized_archive):
"""Update snapshot by adding tag."""
url = create_test_url()
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], env=cli_env)
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], data_dir=initialized_archive)
snapshot = parse_jsonl_output(stdout1)[0]
stdout2, stderr, code = run_archivebox_cmd(
['snapshot', 'update', '--tag=new-tag'],
stdin=json.dumps(snapshot),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -228,46 +228,46 @@ class TestSnapshotUpdate:
class TestSnapshotDelete:
"""Tests for `archivebox snapshot delete`."""
def test_delete_requires_yes(self, cli_env, initialized_archive):
def test_delete_requires_yes(self, initialized_archive):
"""Delete requires --yes flag."""
url = create_test_url()
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], env=cli_env)
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], data_dir=initialized_archive)
snapshot = parse_jsonl_output(stdout1)[0]
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'delete'],
stdin=json.dumps(snapshot),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 1
assert '--yes' in stderr
def test_delete_with_yes(self, cli_env, initialized_archive):
def test_delete_with_yes(self, initialized_archive):
"""Delete with --yes flag works."""
url = create_test_url()
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], env=cli_env)
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], data_dir=initialized_archive)
snapshot = parse_jsonl_output(stdout1)[0]
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'delete', '--yes'],
stdin=json.dumps(snapshot),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
assert 'Deleted 1 snapshots' in stderr
def test_delete_dry_run(self, cli_env, initialized_archive):
def test_delete_dry_run(self, initialized_archive):
"""Dry run shows what would be deleted."""
url = create_test_url()
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], env=cli_env)
stdout1, _, _ = run_archivebox_cmd(['snapshot', 'create', url], data_dir=initialized_archive)
snapshot = parse_jsonl_output(stdout1)[0]
stdout, stderr, code = run_archivebox_cmd(
['snapshot', 'delete', '--dry-run'],
stdin=json.dumps(snapshot),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0