intercal/omnibot
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Format and add some comments to the linkbot module

Browse Source 
Signed-off-by: Alek Ratzloff <alekratz@gmail.com>
This commit is contained in:
Alek Ratzloff
2020-07-02 18:14:38 -07:00
parent e3f30d30c1
commit fcfdb17daa
1 changed files with 15 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
lib/contrib/linkbot.ex
View File

@@ -19,24 +19,27 @@ defmodule Omnibot.Contrib.Linkbot do
plug Tesla.Middleware.FollowRedirects, max_redirects: 10 plug Tesla.Middleware.FollowRedirects, max_redirects: 10
plug Tesla.Middleware.Compression, format: "gzip" plug Tesla.Middleware.Compression, format: "gzip"
# TODO instead of checking for <title> exclusively, do this:
# 1. check for "meta" tag (in the header) with a "property" attribute of "og:title", and fetch the "content" attribute of that tag
# 2. check for meta tag with attribute "name" == "title", and fetch "content" attribute
# 3. Fall back to the <title>
@title_regex ~r"<title>(?<title>.+)</title>"i @title_regex ~r"<title>(?<title>.+)</title>"i
def get_title(url) do def get_title(url) do
# 1. check for "meta" tag (in the header) with a "property" attribute of "og:title", and fetch the "content" attribute of that tag
# 2. check for meta tag with attribute "name" == "title", and fetch "content" attribute
# 3. Fall back to the <title>
html = get_url(url) html = get_url(url)
document = Meeseeks.parse(html) document = Meeseeks.parse(html)
[title | _] = (Meeseeks.all(document, css("meta")) ++ [Meeseeks.one(document, css("title"))]) [title | _] =
|> Enum.map(&( # Get all meta tags and the title tag in a single list
Meeseeks.attr(&1, "property") == "og:title" && Meeseeks.attr(&1, "content") (Meeseeks.all(document, css("meta")) ++ [Meeseeks.one(document, css("title"))])
|| Meeseeks.attr(&1, "name") == "title" && Meeseeks.attr(&1, "content") # Filter/map appropriately
|| Meeseeks.tag(&1) == "title" && Meeseeks.text(&1) |> Enum.map(
)) &((Meeseeks.attr(&1, "property") == "og:title" && Meeseeks.attr(&1, "content")) ||
|> Enum.filter(& &1) (Meeseeks.attr(&1, "name") == "title" && Meeseeks.attr(&1, "content")) ||
(Meeseeks.tag(&1) == "title" && Meeseeks.text(&1)))
)
|> Enum.filter(& &1)
# Return the title
title title
end end