[cohost] add 'avatar' and 'background' options (#6656)

This commit is contained in:
Mike Fährmann
2024-12-14 20:16:28 +01:00
parent 94d7df186f
commit 7f6a53c347
4 changed files with 63 additions and 1 deletions

View File

@@ -1755,6 +1755,26 @@ Description
Extract ``ask`` posts.
extractor.cohost.avatar
-----------------------
Type
``bool``
Default
``false``
Description
Download ``avatar`` images.
extractor.cohost.background
---------------------------
Type
``bool``
Default
``false``
Description
Download ``background``/``banner``/``header`` images.
extractor.cohost.pinned
-----------------------
Type

View File

@@ -176,6 +176,8 @@
"cohost":
{
"asks" : true,
"avatar" : false,
"background": false,
"pinned" : false,
"replies": true,
"shares" : true

View File

@@ -19,7 +19,7 @@ class CohostExtractor(Extractor):
category = "cohost"
root = "https://cohost.org"
directory_fmt = ("{category}", "{postingProject[handle]}")
filename_fmt = ("{postId}_{headline:?/_/[b:200]}{num}.{extension}")
filename_fmt = ("{postId}{headline:?_//[b:200]}{num:?_//}.{extension}")
archive_fmt = "{postId}_{num}"
def _init(self):
@@ -28,6 +28,14 @@ class CohostExtractor(Extractor):
self.shares = self.config("shares", False)
self.asks = self.config("asks", True)
self.avatar = self.config("avatar", False)
if self.avatar:
self._urls_avatar = {None, ""}
self.background = self.config("background", False)
if self.background:
self._urls_background = {None, ""}
def items(self):
for post in self.posts():
reason = post.get("limitedVisibilityReason")
@@ -43,6 +51,26 @@ class CohostExtractor(Extractor):
post["publishedAt"], "%Y-%m-%dT%H:%M:%S.%fZ")
yield Message.Directory, post
project = post["postingProject"]
if self.avatar:
url = project.get("avatarURL")
if url not in self._urls_avatar:
self._urls_avatar.add(url)
p = post.copy()
p["postId"] = p["kind"] = "avatar"
p["headline"] = p["num"] = ""
yield Message.Url, url, text.nameext_from_url(url, p)
if self.background:
url = project.get("headerURL")
if url not in self._urls_background:
self._urls_background.add(url)
p = post.copy()
p["postId"] = p["kind"] = "background"
p["headline"] = p["num"] = ""
yield Message.Url, url, text.nameext_from_url(url, p)
for post["num"], file in enumerate(files, 1):
url = file["fileURL"]
post.update(file)

View File

@@ -16,6 +16,18 @@ __tests__ = (
"#count" : 20,
},
{
"#url" : "https://cohost.org/infinitebrians",
"#category": ("", "cohost", "user"),
"#class" : cohost.CohostUserExtractor,
"#options" : {"avatar": True, "background": True},
"#range" : "1-2",
"#urls" : (
"https://staging.cohostcdn.org/avatar/3281-abb43502-4c48-407d-9778-2bed7722d3d7-profile.gif",
"https://staging.cohostcdn.org/header/3281-b29dbf4d-45b2-417b-b03b-0f7f07595e66-profile.png",
),
},
{
"#url" : "https://cohost.org/infinitebrians/post/4957017-thank-you-akira-tori",
"#category": ("", "cohost", "post"),