32 lines
760 B
Ruby
32 lines
760 B
Ruby
|
require 'nokogiri'
|
||
|
|
|
||
|
|
def toc_data(page_content)
|
||
|
|
html_doc = Nokogiri::HTML::DocumentFragment.parse(page_content)
|
||
|
|
|
||
|
|
# get a flat list of headers
|
||
|
|
headers = []
|
||
|
|
html_doc.css('h1, h2, h3').each do |header|
|
||
|
|
headers.push({
|
||
|
|
id: header.attribute('id').to_s,
|
||
|
|
content: header.children,
|
||
|
|
title: header.children.to_s.gsub(/<[^>]*>/, ''),
|
||
|
|
level: header.name[1].to_i,
|
||
|
|
children: []
|
||
|
|
})
|
||
|
|
end
|
||
|
|
|
||
|
|
[3,2].each do |header_level|
|
||
|
|
header_to_nest = nil
|
||
|
|
headers = headers.reject do |header|
|
||
|
|
if header[:level] == header_level
|
||
|
|
header_to_nest[:children].push header if header_to_nest
|
||
|
|
true
|
||
|
|
else
|
||
|
|
header_to_nest = header if header[:level] < header_level
|
||
|
|
false
|
||
|
|
end
|
||
|
|
end
|
||
|
|
end
|
||
|
|
headers
|
||
|
|
end
|