def __init__(self, number: int):
self.approval_certificate = ApprovalCertificate.objects(
number=number).first()
if self.approval_certificate is None:
self.approval_certificate = ApprovalCertificate(number=number)
soup = self.fetch_website(f"https://consultaca.com/{number}")
self.parse_website(soup)
self.approval_certificate.save()
class ApprovalCertificateExtractor(BaseExtractor):
def __init__(self, number: int):
self.approval_certificate = ApprovalCertificate.objects(
number=number).first()
if self.approval_certificate is None:
self.approval_certificate = ApprovalCertificate(number=number)
soup = self.fetch_website(f"https://consultaca.com/{number}")
self.parse_website(soup)
self.approval_certificate.save()
def parse_website(self, soup: BeautifulSoup):
try:
title = soup.title.get_text().lower()
except (IndexError, AttributeError) as exception:
raise ErrorParsingWebsite from exception
text = nlp(title)
detected_labels = {entity.label_: entity.text for entity in text.ents}
ac_class = _find_first(["PFF3", "PFF2", "PFF1"], detected_labels)
if ac_class is None:
return
self.approval_certificate.ac_class = ac_class
try:
self.approval_certificate.valid = (soup.find(
"div", {
"id": "box_result"
}).find_all("p")[7].span.get_text().lower() == "válido")
self.approval_certificate.manufacturer = title.split(
" - ")[2].split(" ")[0].lower()
except (IndexError, AttributeError) as exception:
raise ErrorParsingWebsite from exception
self.approval_certificate.good_ac = True
def mocked_extract_approval_certificate(self: RespiratorExtractor,
ac_real: int,
ac_candidate: str,
_: str = None):
try:
number = int(ac_candidate.replace(".", ""))
except (ValueError, AttributeError):
return False
if ac_real == number:
self.respirator.approval_certificate = ApprovalCertificate(
number=number, good_ac=True)
return True
return False
def setup_class(cls):
ApprovalCertificate.drop_collection()
Respirator.drop_collection()
def invalid_approval_certificate(self):
approval_certificate = ApprovalCertificate(number=777,
manufacturer="test",
good_ac=False)
return approval_certificate.save()
def approval_certificate(self):
approval_certificate = ApprovalCertificate(number=666,
manufacturer="test",
good_ac=True)
return approval_certificate.save()