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

72
archivebox/tests/test_cli_crawl.py
View File

@@ -23,13 +23,13 @@ from archivebox.tests.conftest import (
class TestCrawlCreate:
"""Tests for `archivebox crawl create`."""
def test_create_from_url_args(self, cli_env, initialized_archive):
def test_create_from_url_args(self, initialized_archive):
"""Create crawl from URL arguments."""
url = create_test_url()
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'create', url],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0, f"Command failed: {stderr}"
@@ -41,7 +41,7 @@ class TestCrawlCreate:
assert records[0]['type'] == 'Crawl'
assert url in records[0]['urls']
def test_create_from_stdin_urls(self, cli_env, initialized_archive):
def test_create_from_stdin_urls(self, initialized_archive):
"""Create crawl from stdin URLs (one per line)."""
urls = [create_test_url() for _ in range(3)]
stdin = '\n'.join(urls)
@@ -49,7 +49,7 @@ class TestCrawlCreate:
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'create'],
stdin=stdin,
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0, f"Command failed: {stderr}"
@@ -62,33 +62,33 @@ class TestCrawlCreate:
for url in urls:
assert url in crawl['urls']
def test_create_with_depth(self, cli_env, initialized_archive):
def test_create_with_depth(self, initialized_archive):
"""Create crawl with --depth flag."""
url = create_test_url()
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'create', '--depth=2', url],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
records = parse_jsonl_output(stdout)
assert records[0]['max_depth'] == 2
def test_create_with_tag(self, cli_env, initialized_archive):
def test_create_with_tag(self, initialized_archive):
"""Create crawl with --tag flag."""
url = create_test_url()
stdout, stderr, code = run_archivebox_cmd(
['crawl', '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()
@@ -97,7 +97,7 @@ class TestCrawlCreate:
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'create'],
stdin=stdin,
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -108,18 +108,18 @@ class TestCrawlCreate:
assert 'Tag' in types
assert 'Crawl' in types
def test_create_pass_through_existing_crawl(self, cli_env, initialized_archive):
def test_create_pass_through_existing_crawl(self, initialized_archive):
"""Existing Crawl records (with id) are passed through."""
# First create a crawl
url = create_test_url()
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]
# Now pipe it back - should pass through
stdout2, stderr, code = run_archivebox_cmd(
['crawl', 'create'],
stdin=json.dumps(crawl),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -131,24 +131,24 @@ class TestCrawlCreate:
class TestCrawlList:
"""Tests for `archivebox crawl list`."""
def test_list_empty(self, cli_env, initialized_archive):
def test_list_empty(self, initialized_archive):
"""List with no crawls returns empty."""
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'list'],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
assert 'Listed 0 crawls' in stderr
def test_list_returns_created(self, cli_env, initialized_archive):
def test_list_returns_created(self, initialized_archive):
"""List returns previously created crawls."""
url = create_test_url()
run_archivebox_cmd(['crawl', 'create', url], env=cli_env)
run_archivebox_cmd(['crawl', 'create', url], data_dir=initialized_archive)
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'list'],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -156,14 +156,14 @@ class TestCrawlList:
assert len(records) >= 1
assert any(url in r.get('urls', '') 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 crawls by status."""
url = create_test_url()
run_archivebox_cmd(['crawl', 'create', url], env=cli_env)
run_archivebox_cmd(['crawl', 'create', url], data_dir=initialized_archive)
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'list', '--status=queued'],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -171,15 +171,15 @@ class TestCrawlList:
for r in records:
assert r['status'] == 'queued'
def test_list_with_limit(self, cli_env, initialized_archive):
def test_list_with_limit(self, initialized_archive):
"""Limit number of results."""
# Create multiple crawls
for _ in range(3):
run_archivebox_cmd(['crawl', 'create', create_test_url()], env=cli_env)
run_archivebox_cmd(['crawl', 'create', create_test_url()], data_dir=initialized_archive)
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'list', '--limit=2'],
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -190,18 +190,18 @@ class TestCrawlList:
class TestCrawlUpdate:
"""Tests for `archivebox crawl update`."""
def test_update_status(self, cli_env, initialized_archive):
def test_update_status(self, initialized_archive):
"""Update crawl status."""
# Create a crawl
url = create_test_url()
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]
# Update it
stdout2, stderr, code = run_archivebox_cmd(
['crawl', 'update', '--status=started'],
stdin=json.dumps(crawl),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
@@ -214,46 +214,46 @@ class TestCrawlUpdate:
class TestCrawlDelete:
"""Tests for `archivebox crawl 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(['crawl', 'create', url], env=cli_env)
stdout1, _, _ = run_archivebox_cmd(['crawl', 'create', url], data_dir=initialized_archive)
crawl = parse_jsonl_output(stdout1)[0]
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'delete'],
stdin=json.dumps(crawl),
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(['crawl', 'create', url], env=cli_env)
stdout1, _, _ = run_archivebox_cmd(['crawl', 'create', url], data_dir=initialized_archive)
crawl = parse_jsonl_output(stdout1)[0]
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'delete', '--yes'],
stdin=json.dumps(crawl),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0
assert 'Deleted 1 crawls' 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(['crawl', 'create', url], env=cli_env)
stdout1, _, _ = run_archivebox_cmd(['crawl', 'create', url], data_dir=initialized_archive)
crawl = parse_jsonl_output(stdout1)[0]
stdout, stderr, code = run_archivebox_cmd(
['crawl', 'delete', '--dry-run'],
stdin=json.dumps(crawl),
env=cli_env,
data_dir=initialized_archive,
)
assert code == 0