Write an application that searches for a provided term on google, then clicks on the first link and lints the HTML on the page using W3C. The application should not assume that it will always use google or W3C.

I've assumed that "clicks on the first link" from the brief is paraphrasing (rather than typing out "makes a request to" each time) and used simple Net::HTTP requests to implement the behaviour specified in most cases, other than sending multipart form data to W3C, for which I've used RestClient.

However, a downside of this is a dependency on W3C accepting cross-origin POST requests, and a dependency on the HTML structure of Google search results.

I could use a library like mechanize or similar to drive a web browser (even a pseudo one) instead to mitigate this, at the expense of much higher overhead when running the code.

Normally I would use feature branches, bugfix branches, and pull requests. However, as I'm the only person working on this project there's no need; I'm going to work directly on master.

git clone [email protected]:kalleth/first-result-linter.git
cd first-result-linter/
# install ruby 2.2 if required; your ruby version manager should handle this
# rvm/rbenv install 2.2

$ ./first_result_linter "cheesy wotsits"

Linting the first result on google for 'cheesy wotsits'
RESULTS:

Category: Info
  The Content-Type was text/html. Using the HTML parser.

Category: Info
  Using the schema for HTML with SVG 1.1, MathML 3.0, RDFa 1.1, and ITS 2.0 support.

Category: Error
  Element link is missing required attribute property.
  From line 451, column 1; to line 451, column 298

bundle exec rspec

• Ensure a Search::DuckDuckGo (or other) matches this API:
  • #call(term) returns the URL of the first result
• Switch it out in the initializer of FirstResultLinter

• Ensure a Linter::HTMLTidy (or other) matches this API:
  • #call(html) returns an array of Linter::Response objects.

• Output the URL returned / that's being linted. This will involve restructuring the linter to perform linting and fetching in a two stage process, or returning a two-element return value.
• Extract the fetching of the HTML of the first result from the Orchestrating class (FirstResultLinter) to a class with the sole responsibility of fetching the HTML.
• Add an integration test which runs the script directly using a VCR cassette.
• Refactor Linter::W3C#parse so it doesn't violate demeter so much.