Refactor spec so it fails consistently and properly retrieve filename.

[jmacdonald]

This corrects the previous deficient spec, and provides a proper implementation.

[kevinsawicki]

Mind rebasing this on master?

[jmacdonald]

Of course not. :wink:

[kevinsawicki]

My main concern here is speed, this current implementation to creates a couple arrays and also splits the entire string when only an end of string comparison is needed.

The benchmarks appear to suggest a 30-40% slower filtering time.

Current Master

> npm run benchmark

> fuzzaldrin@2.0.0 benchmark /Users/kevin/github/fuzzaldrin
> coffee benchmark/benchmark.coffee

Filtering 66672 entries for 'index' took 150ms for 6168 results
Matching 66672 entries for 'index' took 186ms for 6168 results

This PR

> npm run benchmark

> fuzzaldrin@2.0.0 benchmark /Users/kevin/github/fuzzaldrin
> coffee benchmark/benchmark.coffee

Filtering 66672 entries for 'index' took 217ms for 6168 results
Matching 66672 entries for 'index' took 179ms for 6168 results

Would you be cool tweaking it to not use split, but instead something like:

queryIsLastPathSegment = (string, query) ->
  index = string.lastIndexOf(query)
  index is string.length - query.length and string[index - 1] is '/'

[jmacdonald]

Good catch! Considering this is something that'll be run on a regular basis where performance is key, those gains are well worth it. :thumbsup:

[kevinsawicki]

Great, thanks for this.