|
|
|
@ -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")
|
|
|
|
|