jason/qotnews
1
0
Fork 0
forked from tanner/qotnews
Code Issues Pull Requests Releases Wiki Activity

stuff comments.

Browse Source
This commit is contained in:
Jason Schwarzenberger
2020-11-19 14:23:01 +13:00
parent dc3d17b171
commit 1fe524207e
4 changed files with 85 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
apiserver/feeds/category.py
View File

@@ -53,7 +53,7 @@ class Category(Base):
# scratchpad so I can quickly develop the parser
if __name__ == '__main__':
print("Category: RadioNZ")
site = Category("https://www.rnz.co.nz/news/")
site = Category({ 'url': "https://www.rnz.co.nz/news/" })
excludes = [
'rnz.co.nz/news/sport',
'rnz.co.nz/weather',
@@ -61,12 +61,12 @@ if __name__ == '__main__':
]
posts = site.feed(excludes)
print(posts[:5])
print(site.story(posts[0]))
print(site.story(posts[0][0], posts[0][1]))
print("Category: Newsroom")
site = Category("https://www.newsroom.co.nz/news/", tz='Pacific/Auckland')
site = Category({ 'url': "https://www.newsroom.co.nz/news/", 'tz': 'Pacific/Auckland'})
posts = site.feed()
print(posts[:5])
print(site.story(posts[0]))
print(site.story(posts[0][0], posts[0][1]))

22
apiserver/feeds/sitemap.py
View File

@@ -76,7 +76,7 @@ class Sitemap(Base):
# scratchpad so I can quickly develop the parser
if __name__ == '__main__':
print("Sitemap: The Spinoff")
site = Sitemap("https://thespinoff.co.nz/sitemap.xml")
site = Sitemap({ 'url': "https://thespinoff.co.nz/sitemap.xml" })
excludes = [
'thespinoff.co.nz/sitemap-misc.xml',
'thespinoff.co.nz/sitemap-authors.xml',
@@ -84,16 +84,18 @@ if __name__ == '__main__':
]
posts = site.feed(excludes)
print(posts[:5])
print(site.story(posts[0]))
print(site.story(posts[0][0], posts[0][1]))
print("Sitemap: Newshub")
site = Sitemap([
'https://www.newshub.co.nz/home/politics.gnewssitemap.xml',
'https://www.newshub.co.nz/home/new-zealand.gnewssitemap.xml',
'https://www.newshub.co.nz/home/world.gnewssitemap.xml',
'https://www.newshub.co.nz/home/money.gnewssitemap.xml',
])
site = Sitemap({
'url': [
'https://www.newshub.co.nz/home/politics.gnewssitemap.xml',
'https://www.newshub.co.nz/home/new-zealand.gnewssitemap.xml',
'https://www.newshub.co.nz/home/world.gnewssitemap.xml',
'https://www.newshub.co.nz/home/money.gnewssitemap.xml',
],
})
posts = site.feed()
print(posts[:5])
print(site.story(posts[0]))
print(site.story(posts[:-1]))
print(site.story(posts[0][0], posts[0][1]))