header_hierarchy_problem.rb

# The goal of this problem is to extract headers from a block of text,
# and arrange them hierarchically.
#
# See the specs for more detail on the output

require 'net/http'
require 'nokogiri'

def header_hierarchy(html)
  # https://www.w3.org/MarkUp/html3/headings.html
  Nokogiri::HTML(html).css("h1, h2, h3, h4, h5, h6").map do |h|
    indentation = "  " * h.name[/\d/].to_i.pred
    "#{indentation}[#{h.name}] #{h.text}"
  end
end



describe '#header_hierarchy' do
  context 'EASY' do
    it 'can extract a single header' do
      expect(header_hierarchy("<h1>Foo</h1>")).to eq(['[h1] Foo'])
    end

    it 'can extract one nested level of header' do
      expect(
        header_hierarchy("<h1>Foo</h1><h2>Bar</h2>")
      ).to eq([
        '[h1] Foo',
        '  [h2] Bar'
      ])
    end
  end

  context 'MEDIUM' do
    it 'can extract multiple levels of nested headers' do
      expect(
        header_hierarchy("<h1>Foo</h1><h2>Bar</h2><h3>Baz</h3><h4>Bam</h4>")
      ).to eq([
        '[h1] Foo',
        '  [h2] Bar',
        '    [h3] Baz',
        '      [h4] Bam'
      ])
    end
  end

  context 'HARD' do
    it 'can extract multiple nested headers in multiple branches' do
      expect(
        header_hierarchy("<h1>Foo</h1><h2>Bar</h2><h3>Baz</h3><h2>Bam</h2><h3>Ba</h3>")
      ).to eq([
        '[h1] Foo',
        '  [h2] Bar',
        '    [h3] Baz',
        '  [h2] Bam',
        '    [h3] Ba'
      ])
    end
  end

  describe 'True parsing' do
    it 'can parse an entire document' do
      html = Net::HTTP.get(URI("https://jquery.com/"))
      expect(header_hierarchy(html)).to eq([
        "  [h2] jQuery",
        "    [h3] Lightweight Footprint",
        "    [h3] CSS3 Compliant",
        "    [h3] Cross-Browser",
        "  [h2] What is jQuery?",
        "  [h2] Other Related Projects",
        "    [h3] Resources",
        "  [h2] A Brief Look",
        "    [h3] DOM Traversal and Manipulation",
        "    [h3] Event Handling",
        "    [h3] Ajax",
        "    [h3] Books"
      ])
    end
  end
end