github/gallery-dl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[blogger] add search extractor (#925)

Browse Source
This commit is contained in:
Mike Fährmann
2020-08-06 19:43:39 +02:00
parent d5067c51c5
commit 2b88c90f6f
2 changed files with 33 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/supportedsites.rst
View File

@@ -20,7 +20,7 @@ Archived.Moe https://archived.moe/ Threads
ArtStation https://www.artstation.com/ |artstation-C|
baraag https://baraag.net/ Images from Statuses, User Profiles Optional (`OAuth <https://github.com/mikf/gallery-dl#oauth>`__)
Behance https://www.behance.net/ Collections, Galleries, User Profiles
Blogger https://www.blogger.com/ Blogs, Posts
Blogger https://www.blogger.com/ Blogs, Posts, Search Results
Danbooru https://danbooru.donmai.us/ Pools, Popular Images, Posts, Tag Searches Optional
Desuarchive https://desuarchive.org/ Threads
DeviantArt https://www.deviantart.com/ |deviantart-C| Optional (`OAuth <https://github.com/mikf/gallery-dl#oauth>`__)

36
gallery_dl/extractor/blogger.py
View File

@@ -147,7 +147,7 @@ class BloggerPostExtractor(BloggerExtractor):
class BloggerBlogExtractor(BloggerExtractor):
"""Extractor for an entire Blogger blog"""
subcategory = "blog"
pattern = BASE_PATTERN + "(?:/search/label/([^/?&#]+))?/?$"
pattern = BASE_PATTERN + r"/?$"
test = (
("https://julianbphotography.blogspot.com/", {
"range": "1-25",
@@ -160,13 +160,36 @@ class BloggerBlogExtractor(BloggerExtractor):
}),
)
def posts(self, blog):
return self.api.blog_posts(blog["id"])
class BloggerSearchExtractor(BloggerExtractor):
"""Extractor for search resuls and labels"""
subcategory = "search"
pattern = BASE_PATTERN + r"/search(?:/?\?q=([^/?&#]+)|/label/([^/?&#]+))"
test = (
("https://julianbphotography.blogspot.com/search?q=400mm", {
"count": "< 10"
}),
("https://dmmagazine.blogspot.com/search/label/D%26D", {
"range": "1-25",
"count": 25,
}),
)
def __init__(self, match):
BloggerExtractor.__init__(self, match)
label = match.group(3)
self.label = text.unquote(label) if label else None
query = match.group(3)
if query:
self.query, self.label = query, None
else:
self.query, self.label = None, match.group(4)
def posts(self, blog):
return self.api.blog_posts(blog["id"], self.label)
if self.query:
return self.api.blog_search(blog["id"], text.unquote(self.query))
return self.api.blog_posts(blog["id"], text.unquote(self.label))
class BloggerAPI():
@@ -188,6 +211,11 @@ class BloggerAPI():
params = {"labels": label}
return self._pagination(endpoint, params)
def blog_search(self, blog_id, query):
endpoint = "blogs/{}/posts/search".format(blog_id)
params = {"q": query}
return self._pagination(endpoint, params)
def post_by_path(self, blog_id, path):
endpoint = "blogs/{}/posts/bypath".format(blog_id)
return self._call(endpoint, {"path": path}, "post")