[arcalive] add 'user' extractor (#5657)

This commit is contained in:
Mike Fährmann
2025-03-14 18:58:10 +01:00
parent d900e868e4
commit 31e57bafab
4 changed files with 82 additions and 13 deletions

View File

@@ -22,6 +22,13 @@ class ArcaliveExtractor(Extractor):
def _init(self):
self.api = ArcaliveAPI(self)
def items(self):
for article in self.articles():
article["_extractor"] = ArcalivePostExtractor
board = self.board or article.get("boardSlug") or "breaking"
url = "{}/b/{}/{}".format(self.root, board, article["id"])
yield Message.Queue, url, article
class ArcalivePostExtractor(ArcaliveExtractor):
"""Extractor for an arca.live post"""
@@ -99,18 +106,26 @@ class ArcalivePostExtractor(ArcaliveExtractor):
class ArcaliveBoardExtractor(ArcaliveExtractor):
"""Extractor for an arca.live board's posts"""
subcategory = "board"
pattern = BASE_PATTERN + r"/b/(\w+)(?:/?\?([^#]+))?$"
pattern = BASE_PATTERN + r"/b/([^/?#]+)/?(?:\?([^#]+))?$"
example = "https://arca.live/b/breaking"
def items(self):
board, query = self.groups
def articles(self):
self.board, query = self.groups
params = text.parse_query(query)
articles = self.api.board(board, params)
return self.api.board(self.board, params)
for article in articles:
article["_extractor"] = ArcalivePostExtractor
url = "{}/b/{}/{}".format(self.root, board, article["id"])
yield Message.Queue, url, article
class ArcaliveUserExtractor(ArcaliveExtractor):
"""Extractor for an arca.live users's posts"""
subcategory = "user"
pattern = BASE_PATTERN + r"/u/@([^/?#]+)/?(?:\?([^#]+))?$"
example = "https://arca.live/u/@USER"
def articles(self):
self.board = None
user, query = self.groups
params = text.parse_query(query)
return self.api.user_posts(text.unquote(user), params)
class ArcaliveAPI():
@@ -132,6 +147,12 @@ class ArcaliveAPI():
endpoint = "/view/article/breaking/" + str(post_id)
return self._call(endpoint)
def user_posts(self, username, params):
endpoint = "/list/channel/breaking"
params["target"] = "nickname"
params["keyword"] = username
return self._pagination(endpoint, params, "articles")
def _call(self, endpoint, params=None):
url = self.root + endpoint
response = self.extractor.request(url, params=params)