Scrapper should return `HitPage`s and `HitCollection`s instead of itself

[ronaldtse]

Implement a class HitCollection < Array (which already includes Enumerable) so that we not only have the each method but also the rest, like length?

An array of HitPages should be the result returned by Scrapper.

This allows something like:

s = Isobib::IsoBibliography.search '19115' # => #<Isobib::HitPages ...>
page1 = s.first  # => #<Isobib::HitCollection ... >
page1.first # => #<Isobib::Hit ...>

Accessing the second page of search results:

page2 = s[1]  # => #<Isobib::HitCollection @fetched=false... >
page2.fetch  # => #<Isobib::HitCollection @fetched=true >
page2.first # => #<Isobib::Hit ...>

[ronaldtse]

Related to https://github.com/riboseinc/isobib/issues/12

[andrew2net]

@ronaldtse do we really need HitPages? It's possible to implement Array's methods for HitCollection. We can lazy load pages automatically when an index is out of a current page. So we don't have to manipulate pages by hand.

[ronaldtse]

@andrew2net yes it's possible to use HitCollection indeed -- my concern was what is the best way to communicate to users when lazy load will happen, and also how to make it not happen (as you imagine it is important to server-side code).

We can have a LazyCollection that automatically calls hit.fetch on access too. Thoughts?

[ronaldtse]

P.S. I agree with you that HitPages is not something the user needs to access, it might be useful internally, though.

[andrew2net]

implemented HitCollection and HitPage