# HTML Outliner 0.5
#
# See the tests at the bottom for usage examples
#
# Copyright Christoffer Sawicki 2006-2008
# Licensed under the same terms as Ruby
#
# Please send bug reports and improvements to
# christoffer.sawicki@gmail.com
#
# Canonical source
# http://code.vemod.net/svn/misc/hpricot_goodies/html_outliner.rb
#
# Noteworthy changes
# * 0.5: Should now handle malformed HTML gracefully.
# * 0.4: An element can only have one id,
# so pass the current id to the slugifier.
require "hpricot"
class SimpleTree
attr_accessor :root, :subtrees
def initialize(root, subtrees = [])
@root = root
@subtrees = subtrees
end
end
class HTMLOutliner
def initialize(hpricot_doc)
@doc = hpricot_doc
end
def add_header_anchors!(slugifier = method(:default_slugifier))
headers.each_with_index do |header, index|
header["id"] = slugifier.call(header.inner_html, index, header["id"])
end
end
def default_slugifier(string, index, current_id)
# Just overwrite the current id
"section-" + string.downcase.gsub(/\s+/, "_")
end
def headers
find_headers(@doc.children)
end
def outline
tree_stack = []
headers.inject([]) do |result, header|
new_tree = SimpleTree.new(header)
until tree_stack.empty? || tree_stack.last.root.name < header.name
tree_stack.pop
end
if tree_stack.empty?
tree_stack.push(new_tree)
next(result + [ new_tree ])
else
tree_stack.last.subtrees.push(new_tree)
tree_stack.push(new_tree)
next(result)
end
end
end
private
def find_headers(nodes)
nodes.inject([]) do |sum, node|
if node.is_a?(Hpricot::Text) || node.is_a?(Hpricot::BogusETag)
sum
elsif node.name.match(/^h[1-6]$/i)
sum + [ node ] + find_headers(node.children)
else
sum + find_headers(node.children)
end
end
end
end
if __FILE__ == $0
require "test/unit"
class HTMLOutlinerTest < Test::Unit::TestCase
def test_construction_with_root
assert_nothing_raised do
HTMLOutliner.new(Hpricot("")).headers
end
end
def test_construction_without_root
assert_nothing_raised do
HTMLOutliner.new(Hpricot("")).headers
end
end
def test_headers
input = <<-END
Hello
World
END
doc = Hpricot(input)
outliner = HTMLOutliner.new(doc)
assert_equal 2, outliner.headers.size
assert_equal "Hello", outliner.headers[0].inner_html
assert_equal "World", outliner.headers[1].inner_html
end
def test_add_anchors
input = <<-END
Hello
World
END
expected_output = <<-END
Hello
World
END
doc = Hpricot(input)
outliner = HTMLOutliner.new(doc)
outliner.add_header_anchors!
assert_equal(expected_output, doc.to_s)
end
def test_add_anchors_with_existing_id
input = <<-END
Hello
END
expected_output = <<-END
Hello
END
doc = Hpricot(input)
outliner = HTMLOutliner.new(doc)
outliner.add_header_anchors!
assert_equal(expected_output, doc.to_s)
end
def test_add_anchors_with_custom_slugifier
input = <<-END
Hello
END
expected_output = <<-END
Hello
END
doc = Hpricot(input)
outliner = HTMLOutliner.new(doc)
slugifier = lambda { |x, i, current_id| "chapter#{i + 1}" }
outliner.add_header_anchors!(slugifier)
assert_equal(expected_output, doc.to_s)
end
def test_outline
doc = Hpricot(<<-END)
A
B
C
D
E
END
outline = HTMLOutliner.new(doc).outline
assert_equal 2, outline.size
assert_equal 0, outline[0].subtrees.size
assert_equal 2, outline[1].subtrees.size
assert_equal 1, outline[1].subtrees[0].subtrees.size
assert_equal 0, outline[1].subtrees[0].subtrees[0].subtrees.size
assert_equal 0, outline[1].subtrees[1].subtrees.size
assert_equal "A", outline[0].root.inner_html
assert_equal "B", outline[1].root.inner_html
assert_equal "C", outline[1].subtrees[0].root.inner_html
assert_equal "D", outline[1].subtrees[0].subtrees[0].root.inner_html
assert_equal "E", outline[1].subtrees[1].root.inner_html
end
def test_malformed_html
doc = Hpricot(<<-END)
Uh!
B
C
END
assert_nothing_raised do
HTMLOutliner.new(doc).headers
end
assert_nothing_raised do
HTMLOutliner.new(doc).outline
end
end
end
end