freedak/Scrapling
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

chore: migrating to ruff and updating pre-commit hooks

Browse Source
This commit is contained in:
Karim shoair
2025-04-13 17:32:00 +02:00
parent f34b42ea33
commit 0c8dd63f87
35 changed files with 2324 additions and 1182 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/fetchers/sync/test_camoufox.py
+33 -13
View File
@@ -16,12 +16,12 @@ class TestStealthyFetcher:
@pytest.fixture(autouse=True)
def setup_urls(self, httpbin):
"""Fixture to set up URLs for testing"""
self.status_200 = f'{httpbin.url}/status/200'
self.status_404 = f'{httpbin.url}/status/404'
self.status_501 = f'{httpbin.url}/status/501'
self.basic_url = f'{httpbin.url}/get'
self.html_url = f'{httpbin.url}/html'
self.delayed_url = f'{httpbin.url}/delay/10' # 10 Seconds delay response
self.status_200 = f"{httpbin.url}/status/200"
self.status_404 = f"{httpbin.url}/status/404"
self.status_501 = f"{httpbin.url}/status/501"
self.basic_url = f"{httpbin.url}/get"
self.html_url = f"{httpbin.url}/html"
self.delayed_url = f"{httpbin.url}/delay/10" # 10 Seconds delay response
self.cookies_url = f"{httpbin.url}/cookies/set/test/value"
def test_basic_fetch(self, fetcher):
@@ -41,15 +41,21 @@ class TestStealthyFetcher:
def test_waiting_selector(self, fetcher):
"""Test if waiting for a selector make page does not finish loading or not"""
assert fetcher.fetch(self.html_url, wait_selector='h1').status == 200
assert fetcher.fetch(self.html_url, wait_selector='h1', wait_selector_state='visible').status == 200
assert fetcher.fetch(self.html_url, wait_selector="h1").status == 200
assert (
fetcher.fetch(
self.html_url, wait_selector="h1", wait_selector_state="visible"
).status
== 200
)
def test_cookies_loading(self, fetcher):
"""Test if cookies are set after the request"""
assert fetcher.fetch(self.cookies_url).cookies == {'test': 'value'}
assert fetcher.fetch(self.cookies_url).cookies == {"test": "value"}
def test_automation(self, fetcher):
"""Test if automation break the code or not"""
def scroll_page(page):
page.mouse.wheel(10, 0)
page.mouse.move(100, 400)
@@ -60,10 +66,24 @@ class TestStealthyFetcher:
def test_properties(self, fetcher):
"""Test if different arguments breaks the code or not"""
assert fetcher.fetch(self.html_url, block_webrtc=True, allow_webgl=True).status == 200
assert fetcher.fetch(self.html_url, block_webrtc=False, allow_webgl=True).status == 200
assert fetcher.fetch(self.html_url, block_webrtc=True, allow_webgl=False).status == 200
assert fetcher.fetch(self.html_url, extra_headers={'ayo': ''}, os_randomize=True).status == 200
assert (
fetcher.fetch(self.html_url, block_webrtc=True, allow_webgl=True).status
== 200
)
assert (
fetcher.fetch(self.html_url, block_webrtc=False, allow_webgl=True).status
== 200
)
assert (
fetcher.fetch(self.html_url, block_webrtc=True, allow_webgl=False).status
== 200
)
assert (
fetcher.fetch(
self.html_url, extra_headers={"ayo": ""}, os_randomize=True
).status
== 200
)
def test_infinite_timeout(self, fetcher):
"""Test if infinite timeout breaks the code or not"""
tests/fetchers/sync/test_httpx.py
+79 -42
View File
@@ -16,14 +16,14 @@ class TestFetcher:
@pytest.fixture(autouse=True)
def setup_urls(self, httpbin):
"""Fixture to set up URLs for testing"""
self.status_200 = f'{httpbin.url}/status/200'
self.status_404 = f'{httpbin.url}/status/404'
self.status_501 = f'{httpbin.url}/status/501'
self.basic_url = f'{httpbin.url}/get'
self.post_url = f'{httpbin.url}/post'
self.put_url = f'{httpbin.url}/put'
self.delete_url = f'{httpbin.url}/delete'
self.html_url = f'{httpbin.url}/html'
self.status_200 = f"{httpbin.url}/status/200"
self.status_404 = f"{httpbin.url}/status/404"
self.status_501 = f"{httpbin.url}/status/501"
self.basic_url = f"{httpbin.url}/get"
self.post_url = f"{httpbin.url}/post"
self.put_url = f"{httpbin.url}/put"
self.delete_url = f"{httpbin.url}/delete"
self.html_url = f"{httpbin.url}/html"
def test_basic_get(self, fetcher):
"""Test doing basic get request with multiple statuses"""
@@ -36,49 +36,86 @@ class TestFetcher:
assert fetcher.get(self.status_200, stealthy_headers=True).status == 200
assert fetcher.get(self.status_200, follow_redirects=True).status == 200
assert fetcher.get(self.status_200, timeout=None).status == 200
assert fetcher.get(
self.status_200,
stealthy_headers=True,
follow_redirects=True,
timeout=None
).status == 200
assert (
fetcher.get(
self.status_200,
stealthy_headers=True,
follow_redirects=True,
timeout=None,
).status
== 200
)
def test_post_properties(self, fetcher):
"""Test if different arguments with POST request breaks the code or not"""
assert fetcher.post(self.post_url, data={'key': 'value'}).status == 200
assert fetcher.post(self.post_url, data={'key': 'value'}, stealthy_headers=True).status == 200
assert fetcher.post(self.post_url, data={'key': 'value'}, follow_redirects=True).status == 200
assert fetcher.post(self.post_url, data={'key': 'value'}, timeout=None).status == 200
assert fetcher.post(
self.post_url,
data={'key': 'value'},
stealthy_headers=True,
follow_redirects=True,
timeout=None
).status == 200
assert fetcher.post(self.post_url, data={"key": "value"}).status == 200
assert (
fetcher.post(
self.post_url, data={"key": "value"}, stealthy_headers=True
).status
== 200
)
assert (
fetcher.post(
self.post_url, data={"key": "value"}, follow_redirects=True
).status
== 200
)
assert (
fetcher.post(self.post_url, data={"key": "value"}, timeout=None).status
== 200
)
assert (
fetcher.post(
self.post_url,
data={"key": "value"},
stealthy_headers=True,
follow_redirects=True,
timeout=None,
).status
== 200
)
def test_put_properties(self, fetcher):
"""Test if different arguments with PUT request breaks the code or not"""
assert fetcher.put(self.put_url, data={'key': 'value'}).status == 200
assert fetcher.put(self.put_url, data={'key': 'value'}, stealthy_headers=True).status == 200
assert fetcher.put(self.put_url, data={'key': 'value'}, follow_redirects=True).status == 200
assert fetcher.put(self.put_url, data={'key': 'value'}, timeout=None).status == 200
assert fetcher.put(
self.put_url,
data={'key': 'value'},
stealthy_headers=True,
follow_redirects=True,
timeout=None
).status == 200
assert fetcher.put(self.put_url, data={"key": "value"}).status == 200
assert (
fetcher.put(
self.put_url, data={"key": "value"}, stealthy_headers=True
).status
== 200
)
assert (
fetcher.put(
self.put_url, data={"key": "value"}, follow_redirects=True
).status
== 200
)
assert (
fetcher.put(self.put_url, data={"key": "value"}, timeout=None).status == 200
)
assert (
fetcher.put(
self.put_url,
data={"key": "value"},
stealthy_headers=True,
follow_redirects=True,
timeout=None,
).status
== 200
)
def test_delete_properties(self, fetcher):
"""Test if different arguments with DELETE request breaks the code or not"""
assert fetcher.delete(self.delete_url, stealthy_headers=True).status == 200
assert fetcher.delete(self.delete_url, follow_redirects=True).status == 200
assert fetcher.delete(self.delete_url, timeout=None).status == 200
assert fetcher.delete(
self.delete_url,
stealthy_headers=True,
follow_redirects=True,
timeout=None
).status == 200
assert (
fetcher.delete(
self.delete_url,
stealthy_headers=True,
follow_redirects=True,
timeout=None,
).status
== 200
)
tests/fetchers/sync/test_playwright.py
+33 -21
View File
@@ -8,7 +8,6 @@ PlayWrightFetcher.auto_match = True
@pytest_httpbin.use_class_based_httpbin
class TestPlayWrightFetcher:
@pytest.fixture(scope="class")
def fetcher(self):
"""Fixture to create a StealthyFetcher instance for the entire test class"""
@@ -17,12 +16,12 @@ class TestPlayWrightFetcher:
@pytest.fixture(autouse=True)
def setup_urls(self, httpbin):
"""Fixture to set up URLs for testing"""
self.status_200 = f'{httpbin.url}/status/200'
self.status_404 = f'{httpbin.url}/status/404'
self.status_501 = f'{httpbin.url}/status/501'
self.basic_url = f'{httpbin.url}/get'
self.html_url = f'{httpbin.url}/html'
self.delayed_url = f'{httpbin.url}/delay/10' # 10 Seconds delay response
self.status_200 = f"{httpbin.url}/status/200"
self.status_404 = f"{httpbin.url}/status/404"
self.status_501 = f"{httpbin.url}/status/501"
self.basic_url = f"{httpbin.url}/get"
self.html_url = f"{httpbin.url}/html"
self.delayed_url = f"{httpbin.url}/delay/10" # 10 Seconds delay response
self.cookies_url = f"{httpbin.url}/cookies/set/test/value"
def test_basic_fetch(self, fetcher):
@@ -42,12 +41,17 @@ class TestPlayWrightFetcher:
def test_waiting_selector(self, fetcher):
"""Test if waiting for a selector make page does not finish loading or not"""
assert fetcher.fetch(self.html_url, wait_selector='h1').status == 200
assert fetcher.fetch(self.html_url, wait_selector='h1', wait_selector_state='visible').status == 200
assert fetcher.fetch(self.html_url, wait_selector="h1").status == 200
assert (
fetcher.fetch(
self.html_url, wait_selector="h1", wait_selector_state="visible"
).status
== 200
)
def test_cookies_loading(self, fetcher):
"""Test if cookies are set after the request"""
assert fetcher.fetch(self.cookies_url).cookies == {'test': 'value'}
assert fetcher.fetch(self.cookies_url).cookies == {"test": "value"}
def test_automation(self, fetcher):
"""Test if automation break the code or not"""
@@ -60,13 +64,18 @@ class TestPlayWrightFetcher:
assert fetcher.fetch(self.html_url, page_action=scroll_page).status == 200
@pytest.mark.parametrize("kwargs", [
{"disable_webgl": True, "hide_canvas": False},
{"disable_webgl": False, "hide_canvas": True},
# {"stealth": True}, # causes issues with Github Actions
{"useragent": 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:131.0) Gecko/20100101 Firefox/131.0'},
{"extra_headers": {'ayo': ''}}
])
@pytest.mark.parametrize(
"kwargs",
[
{"disable_webgl": True, "hide_canvas": False},
{"disable_webgl": False, "hide_canvas": True},
# {"stealth": True}, # causes issues with Github Actions
{
"useragent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:131.0) Gecko/20100101 Firefox/131.0"
},
{"extra_headers": {"ayo": ""}},
],
)
def test_properties(self, fetcher, kwargs):
"""Test if different arguments breaks the code or not"""
response = fetcher.fetch(self.html_url, **kwargs)
@@ -75,15 +84,18 @@ class TestPlayWrightFetcher:
def test_cdp_url_invalid(self, fetcher):
"""Test if invalid CDP URLs raise appropriate exceptions"""
with pytest.raises(ValueError):
fetcher.fetch(self.html_url, cdp_url='blahblah')
fetcher.fetch(self.html_url, cdp_url="blahblah")
with pytest.raises(ValueError):
fetcher.fetch(self.html_url, cdp_url='blahblah', nstbrowser_mode=True)
fetcher.fetch(self.html_url, cdp_url="blahblah", nstbrowser_mode=True)
with pytest.raises(Exception):
fetcher.fetch(self.html_url, cdp_url='ws://blahblah')
fetcher.fetch(self.html_url, cdp_url="ws://blahblah")
def test_infinite_timeout(self, fetcher, ):
def test_infinite_timeout(
self,
fetcher,
):
"""Test if infinite timeout breaks the code or not"""
response = fetcher.fetch(self.delayed_url, timeout=None)
assert response.status == 200