47 lines
1.5 KiB
Lua
47 lines
1.5 KiB
Lua
local WEBSITE_NAME = "9to5mac"
|
|
local WEBSITE_HOME = "https://9to5mac.com"
|
|
|
|
add_route("Nineto5mac", "/9to5mac")
|
|
Nineto5mac = {}
|
|
function Nineto5mac.route(args)
|
|
local xml = get("https://9to5mac.com/feed/")
|
|
local rss_parser = Feed()
|
|
local feed = rss_parser:new(xml)
|
|
|
|
local articles = feed.channel.articles
|
|
local existing_articles = db:check_if_articles_in_feed_exist(feed)
|
|
|
|
log:debug("Fetching missing articles from the database")
|
|
for _, article in ipairs(articles) do
|
|
if existing_articles[article.guid.value] then
|
|
log:debug("Article already exists in the database: " .. article.title)
|
|
article.description = existing_articles[article.guid.value]
|
|
goto continue
|
|
end
|
|
|
|
log:debug("Getting article: " .. article.title .. " from " .. article.link)
|
|
|
|
local article_content = get(article.link)
|
|
local html_parser = HtmlParser()
|
|
html_parser:parse(article_content)
|
|
|
|
-- Remove sections
|
|
html_parser:delete_element("header")
|
|
html_parser:delete_element("script")
|
|
html_parser:delete_element(".ad-disclaimer-container")
|
|
html_parser:delete_element("#after_disclaimer_placement")
|
|
html_parser:delete_element(".adsbygoogle")
|
|
html_parser:delete_element(".google-news-link")
|
|
|
|
local elements = html_parser:select_element('.post-content')
|
|
local element = elements
|
|
[1]
|
|
article.description =
|
|
element
|
|
sleep(500)
|
|
::continue::
|
|
end
|
|
|
|
return feed:render(), feed
|
|
end
|