Fix issue #93 - uses re to parse links in tweet

[pravj]

Before this, the open {id} command was using only prefix-matching to parse links from tweets.
In old way, it used to split the tweet text and collected all the words with http/https as prefix.

As explained in a sample tweet in issue #93:

Super excited to have David DeSandro (@desandro) helping us revamp forecast.io (soon to be http://darksky.net).

The parsed links here, will be 'http://darksky.net).' instead of 'http://darksky.net', which is wrong.

Now it uses regex module to parse the same.

As shortened URLs in tweets have 22/23 string length(based on protocol). So, all the short URLs have unique-id of length 10 as suffix.

So, a simple re.findall with https?://t.co/[a-zA-Z0-9]{10} as search pattern, will return a list of all the links in the tweet text.

That's it.

[orakaro]

Hi @pravj ! Thanks for suggest this. I will test for a while before merge 🍺

[Tenzer]

Wouldn't it be simpler and more reliable to just use the entities list Twitter includes with every tweet? https://dev.twitter.com/overview/api/entities-in-twitter-objects#urls

[orakaro]

Thanks @Tenzer .I will take a further look for the entities list

[Tenzer]

@dtvd You can take a look at the change to repall I made, that was switched to use entities: #90.

[orakaro]

@Tenzer thanks. Just update this function using entities