Ruby: collect a target key's value into array from nested hash - ruby

Question

I have a file like this:
$urls = [
{name:'Venture Capitals',
sites: [
'http://blog.ycombinator.com/posts.atom',
'http://themacro.com/feed.xml',
'http://a16z.com/feed/',
'http://firstround.com/review/feed.xml',
'http://www.kpcb.com/blog.rss',
'https://library.gv.com/feed',
'http://theaccelblog.squarespace.com/blog?format=RSS',
'https://medium.com/feed/accel-insights',
'http://500.co/blog/posts/feed/',
'http://feeds.feedburner.com/upfrontinsights?format=xml',
'http://versionone.vc/feed/',
'http://nextviewventures.com/blog/feed/',
]},
{name:'Companies and Groups',
sites: [
{name:'Product Companies',
sites: [
'https://m.signalvnoise.com/feed',
'http://feeds.feedburner.com/insideintercom',
'http://www.kickstarter.com/blog.atom',
'http://blog.invisionapp.com/feed/',
'http://feeds.feedburner.com/bufferapp',
'https://open.buffer.com/feed/',
'https://blog.asana.com/feed/',
'http://blog.drift.com/rss.xml',
'https://www.groovehq.com/blog/feed',]},
{name:'Consulting Groups, Studios',
sites: [
'http://svpg.com/articles/rss',
'http://www.thoughtworks.com/rss/insights.xml',
'http://zurb.com/blog/rss',]},
{name:'Communities',
sites: [
'http://alistapart.com/main/feed',
'https://www.mindtheproduct.com/feed/',]},
]},
]
I have organized the $url into different groups. Now I want to extract all the urls out (the link in the sites), how should I do?
The main problem is that, there are sites within sites, as the file showed above.
My problems are:
Am I using a proper file structure to save these links? (array within array). If not, what would be good way to save and group them?
How can I extract all the urls out into a flattened array? so I can later iterate through the list.
I can do this pretty manually, like the code shown below.
sites = []
$urls.each do |item|
item[:sites].each do |sub_item|
if sub_item.is_a?(Hash)
sites.concat sub_item[:sites]
else
sites.append sub_item
end
end
end
File.open('lib/flatten_sites.yaml', 'w') { |fo| fo.puts sites.to_yaml }
But I just feel this is bad code.
An alternative in this specific case, is to collect all the sites attribute, but I feel this is also very constrained, and may not help in some other cases.

Answer 1

If you have Hash, you can use this recursive method
Input
urls = [
{
:name => 'Venture Capitals',
:sites => [
'http://blog.ycombinator.com/posts.atom',
'http://themacro.com/feed.xml',
'http://a16z.com/feed/',
'http://firstround.com/review/feed.xml',
'http://www.kpcb.com/blog.rss',
'https://library.gv.com/feed',
'http://theaccelblog.squarespace.com/blog?format=RSS',
'https://medium.com/feed/accel-insights',
'http://500.co/blog/posts/feed/',
'http://feeds.feedburner.com/upfrontinsights?format=xml',
'http://versionone.vc/feed/',
'http://nextviewventures.com/blog/feed/',
]
},
{
:name => 'Companies and Groups',
:sites => [
{
:name => 'Product Companies',
:sites => [
'https://m.signalvnoise.com/feed',
'http://feeds.feedburner.com/insideintercom',
'http://www.kickstarter.com/blog.atom',
'http://blog.invisionapp.com/feed/',
'http://feeds.feedburner.com/bufferapp',
'https://open.buffer.com/feed/',
'https://blog.asana.com/feed/',
'http://blog.drift.com/rss.xml',
'https://www.groovehq.com/blog/feed',]
},
{
:name => 'Consulting Groups, Studios',
:sites => [
'http://svpg.com/articles/rss',
'http://www.thoughtworks.com/rss/insights.xml',
'http://zurb.com/blog/rss',]
},
{
:name => 'Communities',
:sites => [
'http://alistapart.com/main/feed',
'https://www.mindtheproduct.com/feed/',]
}
]
}
]
Method
def get_all_sites(data)
data[:sites].map { |r| Hash === r ? get_all_sites(r) : r }
end
urls.map { |r| get_all_sites(r) }.flatten
Output
[
"http://blog.ycombinator.com/posts.atom",
"http://themacro.com/feed.xml",
"http://a16z.com/feed/",
"http://firstround.com/review/feed.xml",
"http://www.kpcb.com/blog.rss",
"https://library.gv.com/feed",
"http://theaccelblog.squarespace.com/blog?format=RSS",
"https://medium.com/feed/accel-insights",
"http://500.co/blog/posts/feed/",
"http://feeds.feedburner.com/upfrontinsights?format=xml",
"http://versionone.vc/feed/",
"http://nextviewventures.com/blog/feed/",
"https://m.signalvnoise.com/feed",
"http://feeds.feedburner.com/insideintercom",
"http://www.kickstarter.com/blog.atom",
"http://blog.invisionapp.com/feed/",
"http://feeds.feedburner.com/bufferapp",
"https://open.buffer.com/feed/",
"https://blog.asana.com/feed/",
"http://blog.drift.com/rss.xml",
"https://www.groovehq.com/blog/feed",
"http://svpg.com/articles/rss",
"http://www.thoughtworks.com/rss/insights.xml",
"http://zurb.com/blog/rss",
"http://alistapart.com/main/feed",
"https://www.mindtheproduct.com/feed/"
]
I hope this helps

Answer 2

The solution similar to what Lukas Baliak proposed, but using more suitable Proc instead of redundant method (works for any amount of level’s nesting):
deep_map = ->(data) do
data[:sites].flat_map { |r| r.is_a?(String) ? r : deep_map.(r) }
end
urls.flat_map(&deep_map)