def recipe_from_url(request):
url = request.POST['url']
headers = {
'User-Agent':
'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/83.0.4103.106 Safari/537.36' # noqa: E501
}
try:
response = requests.get(url, headers=headers)
except requests.exceptions.ConnectionError:
return JsonResponse(
{
'error': True,
'msg': _('The requested page could not be found.')
},
status=400)
if response.status_code == 403:
return JsonResponse(
{
'error':
True,
'msg':
_('The requested page refused to provide any information (Status Code 403).'
) # noqa: E501
},
status=400)
return get_from_html(response.text, url)
def test_ld_json(self):
test_list = [
{
'file': 'cookbook/tests/resources/websites/ld_json_1.html',
'result_length': 3128
},
{
'file': 'cookbook/tests/resources/websites/ld_json_2.html',
'result_length': 1450
},
{
'file': 'cookbook/tests/resources/websites/ld_json_3.html',
'result_length': 1545
},
{
'file': 'cookbook/tests/resources/websites/ld_json_4.html',
'result_length': 1657
},
{
'file':
'cookbook/tests/resources/websites/ld_json_invalid.html',
'result_length': 115
},
{
'file':
'cookbook/tests/resources/websites/ld_json_itemList.html',
'result_length': 3131
},
{
'file':
'cookbook/tests/resources/websites/ld_json_multiple.html',
'result_length': 1546
},
{
'file': 'cookbook/tests/resources/websites/micro_data_1.html',
'result_length': 1022
},
{
'file': 'cookbook/tests/resources/websites/micro_data_2.html',
'result_length': 1384
},
{
'file': 'cookbook/tests/resources/websites/micro_data_3.html',
'result_length': 1100
},
{
'file': 'cookbook/tests/resources/websites/micro_data_4.html',
'result_length': 4231
},
]
for test in test_list:
with open(test['file'], 'rb') as file:
parsed_content = json.loads(
get_from_html(file.read(), 'test_url').content)
self.assertEqual(len(str(parsed_content)),
test['result_length'])
file.close()
def test_ld_json(self):
test_list = [
{
'file': 'cookbook/tests/resources/websites/ld_json_1.html',
'result_length': 3237
},
{
'file': 'cookbook/tests/resources/websites/ld_json_2.html',
'result_length': 1510
},
{
'file': 'cookbook/tests/resources/websites/ld_json_3.html',
'result_length': 1629
},
{
'file': 'cookbook/tests/resources/websites/ld_json_4.html',
'result_length': 1744
},
{
'file':
'cookbook/tests/resources/websites/ld_json_itemList.html',
'result_length': 3206
},
{
'file':
'cookbook/tests/resources/websites/ld_json_multiple.html',
'result_length': 1621
},
{
'file': 'cookbook/tests/resources/websites/micro_data_1.html',
'result_length': 1079
},
{
'file': 'cookbook/tests/resources/websites/micro_data_2.html',
'result_length': 1438
},
{
'file': 'cookbook/tests/resources/websites/micro_data_3.html',
'result_length': 1148
},
{
'file': 'cookbook/tests/resources/websites/micro_data_4.html',
'result_length': 4396
},
]
for test in test_list:
with open(test['file'], 'rb') as file:
print(
f'Testing {test["file"]} expecting length {test["result_length"]}'
)
parsed_content = json.loads(
get_from_html(file.read(), 'test_url').content)
self.assertEqual(len(str(parsed_content)),
test['result_length'])
file.close()