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/async/test_camoufox.py
+55 -43
View File
@@ -17,43 +17,51 @@ class TestStealthyFetcher:
def urls(self, httpbin):
url = httpbin.url
return {
'status_200': f'{url}/status/200',
'status_404': f'{url}/status/404',
'status_501': f'{url}/status/501',
'basic_url': f'{url}/get',
'html_url': f'{url}/html',
'delayed_url': f'{url}/delay/10', # 10 Seconds delay response
'cookies_url': f"{url}/cookies/set/test/value"
"status_200": f"{url}/status/200",
"status_404": f"{url}/status/404",
"status_501": f"{url}/status/501",
"basic_url": f"{url}/get",
"html_url": f"{url}/html",
"delayed_url": f"{url}/delay/10", # 10 Seconds delay response
"cookies_url": f"{url}/cookies/set/test/value",
}
async def test_basic_fetch(self, fetcher, urls):
"""Test doing basic fetch request with multiple statuses"""
assert (await fetcher.async_fetch(urls['status_200'])).status == 200
assert (await fetcher.async_fetch(urls['status_404'])).status == 404
assert (await fetcher.async_fetch(urls['status_501'])).status == 501
assert (await fetcher.async_fetch(urls["status_200"])).status == 200
assert (await fetcher.async_fetch(urls["status_404"])).status == 404
assert (await fetcher.async_fetch(urls["status_501"])).status == 501
async def test_networkidle(self, fetcher, urls):
"""Test if waiting for `networkidle` make page does not finish loading or not"""
assert (await fetcher.async_fetch(urls['basic_url'], network_idle=True)).status == 200
assert (
await fetcher.async_fetch(urls["basic_url"], network_idle=True)
).status == 200
async def test_blocking_resources(self, fetcher, urls):
"""Test if blocking resources make page does not finish loading or not"""
assert (await fetcher.async_fetch(urls['basic_url'], block_images=True)).status == 200
assert (await fetcher.async_fetch(urls['basic_url'], disable_resources=True)).status == 200
assert (
await fetcher.async_fetch(urls["basic_url"], block_images=True)
).status == 200
assert (
await fetcher.async_fetch(urls["basic_url"], disable_resources=True)
).status == 200
async def test_waiting_selector(self, fetcher, urls):
"""Test if waiting for a selector make page does not finish loading or not"""
assert (await fetcher.async_fetch(urls['html_url'], wait_selector='h1')).status == 200
assert (await fetcher.async_fetch(
urls['html_url'],
wait_selector='h1',
wait_selector_state='visible'
)).status == 200
assert (
await fetcher.async_fetch(urls["html_url"], wait_selector="h1")
).status == 200
assert (
await fetcher.async_fetch(
urls["html_url"], wait_selector="h1", wait_selector_state="visible"
)
).status == 200
async def test_cookies_loading(self, fetcher, urls):
"""Test if cookies are set after the request"""
response = await fetcher.async_fetch(urls['cookies_url'])
assert response.cookies == {'test': 'value'}
response = await fetcher.async_fetch(urls["cookies_url"])
assert response.cookies == {"test": "value"}
async def test_automation(self, fetcher, urls):
"""Test if automation break the code or not"""
@@ -64,34 +72,38 @@ class TestStealthyFetcher:
await page.mouse.up()
return page
assert (await fetcher.async_fetch(urls['html_url'], page_action=scroll_page)).status == 200
assert (
await fetcher.async_fetch(urls["html_url"], page_action=scroll_page)
).status == 200
async def test_properties(self, fetcher, urls):
"""Test if different arguments breaks the code or not"""
assert (await fetcher.async_fetch(
urls['html_url'],
block_webrtc=True,
allow_webgl=True
)).status == 200
assert (
await fetcher.async_fetch(
urls["html_url"], block_webrtc=True, allow_webgl=True
)
).status == 200
assert (await fetcher.async_fetch(
urls['html_url'],
block_webrtc=False,
allow_webgl=True
)).status == 200
assert (
await fetcher.async_fetch(
urls["html_url"], block_webrtc=False, allow_webgl=True
)
).status == 200
assert (await fetcher.async_fetch(
urls['html_url'],
block_webrtc=True,
allow_webgl=False
)).status == 200
assert (
await fetcher.async_fetch(
urls["html_url"], block_webrtc=True, allow_webgl=False
)
).status == 200
assert (await fetcher.async_fetch(
urls['html_url'],
extra_headers={'ayo': ''},
os_randomize=True
)).status == 200
assert (
await fetcher.async_fetch(
urls["html_url"], extra_headers={"ayo": ""}, os_randomize=True
)
).status == 200
async def test_infinite_timeout(self, fetcher, urls):
"""Test if infinite timeout breaks the code or not"""
assert (await fetcher.async_fetch(urls['delayed_url'], timeout=None)).status == 200
assert (
await fetcher.async_fetch(urls["delayed_url"], timeout=None)
).status == 200
tests/fetchers/async/test_httpx.py
+92 -51
View File
@@ -16,70 +16,111 @@ class TestAsyncFetcher:
@pytest.fixture(scope="class")
def urls(self, httpbin):
return {
'status_200': f'{httpbin.url}/status/200',
'status_404': f'{httpbin.url}/status/404',
'status_501': f'{httpbin.url}/status/501',
'basic_url': f'{httpbin.url}/get',
'post_url': f'{httpbin.url}/post',
'put_url': f'{httpbin.url}/put',
'delete_url': f'{httpbin.url}/delete',
'html_url': f'{httpbin.url}/html'
"status_200": f"{httpbin.url}/status/200",
"status_404": f"{httpbin.url}/status/404",
"status_501": f"{httpbin.url}/status/501",
"basic_url": f"{httpbin.url}/get",
"post_url": f"{httpbin.url}/post",
"put_url": f"{httpbin.url}/put",
"delete_url": f"{httpbin.url}/delete",
"html_url": f"{httpbin.url}/html",
}
async def test_basic_get(self, fetcher, urls):
"""Test doing basic get request with multiple statuses"""
assert (await fetcher.get(urls['status_200'])).status == 200
assert (await fetcher.get(urls['status_404'])).status == 404
assert (await fetcher.get(urls['status_501'])).status == 501
assert (await fetcher.get(urls["status_200"])).status == 200
assert (await fetcher.get(urls["status_404"])).status == 404
assert (await fetcher.get(urls["status_501"])).status == 501
async def test_get_properties(self, fetcher, urls):
"""Test if different arguments with GET request breaks the code or not"""
assert (await fetcher.get(urls['status_200'], stealthy_headers=True)).status == 200
assert (await fetcher.get(urls['status_200'], follow_redirects=True)).status == 200
assert (await fetcher.get(urls['status_200'], timeout=None)).status == 200
assert (await fetcher.get(
urls['status_200'],
stealthy_headers=True,
follow_redirects=True,
timeout=None
)).status == 200
assert (
await fetcher.get(urls["status_200"], stealthy_headers=True)
).status == 200
assert (
await fetcher.get(urls["status_200"], follow_redirects=True)
).status == 200
assert (await fetcher.get(urls["status_200"], timeout=None)).status == 200
assert (
await fetcher.get(
urls["status_200"],
stealthy_headers=True,
follow_redirects=True,
timeout=None,
)
).status == 200
async def test_post_properties(self, fetcher, urls):
"""Test if different arguments with POST request breaks the code or not"""
assert (await fetcher.post(urls['post_url'], data={'key': 'value'})).status == 200
assert (await fetcher.post(urls['post_url'], data={'key': 'value'}, stealthy_headers=True)).status == 200
assert (await fetcher.post(urls['post_url'], data={'key': 'value'}, follow_redirects=True)).status == 200
assert (await fetcher.post(urls['post_url'], data={'key': 'value'}, timeout=None)).status == 200
assert (await fetcher.post(
urls['post_url'],
data={'key': 'value'},
stealthy_headers=True,
follow_redirects=True,
timeout=None
)).status == 200
assert (
await fetcher.post(urls["post_url"], data={"key": "value"})
).status == 200
assert (
await fetcher.post(
urls["post_url"], data={"key": "value"}, stealthy_headers=True
)
).status == 200
assert (
await fetcher.post(
urls["post_url"], data={"key": "value"}, follow_redirects=True
)
).status == 200
assert (
await fetcher.post(urls["post_url"], data={"key": "value"}, timeout=None)
).status == 200
assert (
await fetcher.post(
urls["post_url"],
data={"key": "value"},
stealthy_headers=True,
follow_redirects=True,
timeout=None,
)
).status == 200
async def test_put_properties(self, fetcher, urls):
"""Test if different arguments with PUT request breaks the code or not"""
assert (await fetcher.put(urls['put_url'], data={'key': 'value'})).status in [200, 405]
assert (await fetcher.put(urls['put_url'], data={'key': 'value'}, stealthy_headers=True)).status in [200, 405]
assert (await fetcher.put(urls['put_url'], data={'key': 'value'}, follow_redirects=True)).status in [200, 405]
assert (await fetcher.put(urls['put_url'], data={'key': 'value'}, timeout=None)).status in [200, 405]
assert (await fetcher.put(
urls['put_url'],
data={'key': 'value'},
stealthy_headers=True,
follow_redirects=True,
timeout=None
)).status in [200, 405]
assert (await fetcher.put(urls["put_url"], data={"key": "value"})).status in [
200,
405,
]
assert (
await fetcher.put(
urls["put_url"], data={"key": "value"}, stealthy_headers=True
)
).status in [200, 405]
assert (
await fetcher.put(
urls["put_url"], data={"key": "value"}, follow_redirects=True
)
).status in [200, 405]
assert (
await fetcher.put(urls["put_url"], data={"key": "value"}, timeout=None)
).status in [200, 405]
assert (
await fetcher.put(
urls["put_url"],
data={"key": "value"},
stealthy_headers=True,
follow_redirects=True,
timeout=None,
)
).status in [200, 405]
async def test_delete_properties(self, fetcher, urls):
"""Test if different arguments with DELETE request breaks the code or not"""
assert (await fetcher.delete(urls['delete_url'], stealthy_headers=True)).status == 200
assert (await fetcher.delete(urls['delete_url'], follow_redirects=True)).status == 200
assert (await fetcher.delete(urls['delete_url'], timeout=None)).status == 200
assert (await fetcher.delete(
urls['delete_url'],
stealthy_headers=True,
follow_redirects=True,
timeout=None
)).status == 200
assert (
await fetcher.delete(urls["delete_url"], stealthy_headers=True)
).status == 200
assert (
await fetcher.delete(urls["delete_url"], follow_redirects=True)
).status == 200
assert (await fetcher.delete(urls["delete_url"], timeout=None)).status == 200
assert (
await fetcher.delete(
urls["delete_url"],
stealthy_headers=True,
follow_redirects=True,
timeout=None,
)
).status == 200
tests/fetchers/async/test_playwright.py
+37 -27
View File
@@ -15,87 +15,97 @@ class TestPlayWrightFetcherAsync:
@pytest.fixture
def urls(self, httpbin):
return {
'status_200': f'{httpbin.url}/status/200',
'status_404': f'{httpbin.url}/status/404',
'status_501': f'{httpbin.url}/status/501',
'basic_url': f'{httpbin.url}/get',
'html_url': f'{httpbin.url}/html',
'delayed_url': f'{httpbin.url}/delay/10',
'cookies_url': f"{httpbin.url}/cookies/set/test/value"
"status_200": f"{httpbin.url}/status/200",
"status_404": f"{httpbin.url}/status/404",
"status_501": f"{httpbin.url}/status/501",
"basic_url": f"{httpbin.url}/get",
"html_url": f"{httpbin.url}/html",
"delayed_url": f"{httpbin.url}/delay/10",
"cookies_url": f"{httpbin.url}/cookies/set/test/value",
}
@pytest.mark.asyncio
async def test_basic_fetch(self, fetcher, urls):
"""Test doing basic fetch request with multiple statuses"""
response = await fetcher.async_fetch(urls['status_200'])
response = await fetcher.async_fetch(urls["status_200"])
assert response.status == 200
@pytest.mark.asyncio
async def test_networkidle(self, fetcher, urls):
"""Test if waiting for `networkidle` make page does not finish loading or not"""
response = await fetcher.async_fetch(urls['basic_url'], network_idle=True)
response = await fetcher.async_fetch(urls["basic_url"], network_idle=True)
assert response.status == 200
@pytest.mark.asyncio
async def test_blocking_resources(self, fetcher, urls):
"""Test if blocking resources make page does not finish loading or not"""
response = await fetcher.async_fetch(urls['basic_url'], disable_resources=True)
response = await fetcher.async_fetch(urls["basic_url"], disable_resources=True)
assert response.status == 200
@pytest.mark.asyncio
async def test_waiting_selector(self, fetcher, urls):
"""Test if waiting for a selector make page does not finish loading or not"""
response1 = await fetcher.async_fetch(urls['html_url'], wait_selector='h1')
response1 = await fetcher.async_fetch(urls["html_url"], wait_selector="h1")
assert response1.status == 200
response2 = await fetcher.async_fetch(urls['html_url'], wait_selector='h1', wait_selector_state='visible')
response2 = await fetcher.async_fetch(
urls["html_url"], wait_selector="h1", wait_selector_state="visible"
)
assert response2.status == 200
@pytest.mark.asyncio
async def test_cookies_loading(self, fetcher, urls):
"""Test if cookies are set after the request"""
response = await fetcher.async_fetch(urls['cookies_url'])
assert response.cookies == {'test': 'value'}
response = await fetcher.async_fetch(urls["cookies_url"])
assert response.cookies == {"test": "value"}
@pytest.mark.asyncio
async def test_automation(self, fetcher, urls):
"""Test if automation break the code or not"""
async def scroll_page(page):
await page.mouse.wheel(10, 0)
await page.mouse.move(100, 400)
await page.mouse.up()
return page
response = await fetcher.async_fetch(urls['html_url'], page_action=scroll_page)
response = await fetcher.async_fetch(urls["html_url"], page_action=scroll_page)
assert response.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": ""}},
],
)
@pytest.mark.asyncio
async def test_properties(self, fetcher, urls, kwargs):
"""Test if different arguments breaks the code or not"""
response = await fetcher.async_fetch(urls['html_url'], **kwargs)
response = await fetcher.async_fetch(urls["html_url"], **kwargs)
assert response.status == 200
@pytest.mark.asyncio
async def test_cdp_url_invalid(self, fetcher, urls):
"""Test if invalid CDP URLs raise appropriate exceptions"""
with pytest.raises(ValueError):
await fetcher.async_fetch(urls['html_url'], cdp_url='blahblah')
await fetcher.async_fetch(urls["html_url"], cdp_url="blahblah")
with pytest.raises(ValueError):
await fetcher.async_fetch(urls['html_url'], cdp_url='blahblah', nstbrowser_mode=True)
await fetcher.async_fetch(
urls["html_url"], cdp_url="blahblah", nstbrowser_mode=True
)
with pytest.raises(Exception):
await fetcher.async_fetch(urls['html_url'], cdp_url='ws://blahblah')
await fetcher.async_fetch(urls["html_url"], cdp_url="ws://blahblah")
@pytest.mark.asyncio
async def test_infinite_timeout(self, fetcher, urls):
"""Test if infinite timeout breaks the code or not"""
response = await fetcher.async_fetch(urls['delayed_url'], timeout=None)
response = await fetcher.async_fetch(urls["delayed_url"], timeout=None)
assert response.status == 200