URL prediction verification with file name pattern

Introduced file name as a parameter to allow to also verify url predictions in link_generator.py
dev
Peery 3 years ago
parent 52030c3409
commit 6fab754612

@ -100,13 +100,13 @@ class ArtNetManager:
if url is None: if url is None:
return None return None
#tags = ArtNet.web.Scrap_Tags.scrap_tags(file_name, url, ArtNet.web.link_generator.predict_domain(file_name)) tags = LinkGenerator.get_instance().scrape_tags(url=url, file_name=file_name,
tags = LinkGenerator.get_instance().scrape_tags(url, LinkGenerator.get_instance().predict_domain(file_name)) domain=LinkGenerator.get_instance().predict_domain(file_name))
if tags is None: if tags is None:
return None return None
already_applied_tags = self.db_connection.get_art_tags_by_ID(art_ID) already_applied_tags = self.db_connection.get_art_tags_by_ID(art_ID)
for i in range(len(already_applied_tags)): for i in range(len(already_applied_tags)): # converting the list to List[str]
already_applied_tags[i] = self.db_connection.get_tag_by_ID(already_applied_tags[i])[0][1].strip() already_applied_tags[i] = self.db_connection.get_tag_by_ID(already_applied_tags[i])[0][1].strip()
importable_tags = [] importable_tags = []

@ -73,11 +73,14 @@ class DomainLinkGenerator:
""" """
raise NotImplementedError raise NotImplementedError
def scrape_tags(self, url: str, headers: dict) -> list: def scrape_tags(self, url: str, headers: dict, file_name: str) -> list:
""" """
Scrape the tags from the given url for all tags associated with the work. Scrape the tags from the given url for all tags associated with the work.
The file_name can also be used to check the given url against prediction results.
:param url: :param url:
:param headers: :param headers:
:param file_name:
:return: :return:
""" """
raise NotImplementedError raise NotImplementedError
@ -149,11 +152,12 @@ class LinkGenerator:
return None return None
def scrape_tags(self, url: str, domain: int) -> dict: def scrape_tags(self, url: str, domain: int, file_name: str) -> dict:
""" """
Scrapes the tags from the given url Scrapes the tags from the given url
:param url: :param url:
:param domain: :param domain:
:param file_name:
:return: :return:
""" """
headers = { headers = {
@ -163,7 +167,7 @@ class LinkGenerator:
for g in self.__link_generators: for g in self.__link_generators:
if g.get_identifier() == domain or g.get_domain_name() == url_domain: if g.get_identifier() == domain or g.get_domain_name() == url_domain:
try: try:
return g.scrape_tags(url, headers) return g.scrape_tags(url=url, headers=headers, file_name=file_name)
except NotImplementedError: except NotImplementedError:
pass pass
return None return None

Loading…
Cancel
Save