# encoding: utf-8
require 'html/pipeline'
require 'task_list'

class TaskList
  # Returns a `Nokogiri::DocumentFragment` object.
  def self.filter(*args)
    Filter.call(*args)
  end

  # TaskList filter replaces task list item markers (`[ ]` and `[x]`) with
  # checkboxes, marked up with metadata and behavior.
  #
  # This should be run on the HTML generated by the Markdown filter, after the
  # SanitizationFilter.
  #
  # Syntax
  # ------
  #
  # Task list items must be in a list format:
  #
  # ```
  # - [ ] incomplete
  # - [x] complete
  # ```
  #
  # Results
  # -------
  #
  # The following keys are written to the result hash:
  #   :task_list_items - An array of TaskList::Item objects.
  class Filter < HTML::Pipeline::Filter

    Incomplete = "[ ]".freeze
    Complete   = "[x]".freeze

    IncompletePattern = /\[[[:space:]]\]/.freeze # matches all whitespace
    CompletePattern   = /\[[xX]\]/.freeze        # matches any capitalization

    # All valid checkbox patterns.
    # Useful for overriding ItemPattern wih custom formats.
    CheckboxPatterns  = /#{CompletePattern}|#{IncompletePattern}/

    # Pattern used to identify all task list items.
    # Useful when you need iterate over all items.
    ItemPattern = /
      ^
      (?:\s*[-+*]|(?:\d+\.))? # optional list prefix
      \s*                     # optional whitespace prefix
      (#{CheckboxPatterns})   # checkbox
      (?=\s)                  # followed by whitespace
    /x

    # List of `TaskList::Item` objects that were recognized in the document.
    # This is available in the result hash as `:task_list_items`.
    #
    # Returns an Array of TaskList::Item objects.
    def task_list_items
      result[:task_list_items] ||= []
    end

    # Computes attributes for the item input.
    #
    # Returns an String of HTML attributes.
    def checkbox_attributes(item)
      'checked="checked"' if item.complete?
    end

    # Renders the item checkbox in a span including the item state.
    #
    # Returns an HTML-safe String.
    def render_item_checkbox(item)
      %(<input type="checkbox"
        class="task-list-item-checkbox"
        #{checkbox_attributes(item)}
        disabled="disabled"
      />)
    end

    # Public: Marks up the task list item checkbox with metadata and behavior.
    #
    # NOTE: produces a string that, when assigned to a Node's `inner_html`,
    # will corrupt the string contents' encodings. Instead, we parse the
    # rendered HTML and explicitly set its encoding so that assignment will
    # not change the encodings.
    #
    # See [this pull](https://github.com/github/github/pull/8505) for details.
    #
    # Returns the marked up task list item Nokogiri::XML::NodeSet object.
    def render_task_list_item(item)
      Nokogiri::HTML.fragment \
        item.source.sub(ItemPattern, render_item_checkbox(item)), 'utf-8'
    end

    # Filters the source for task list items.
    #
    # Each item is wrapped in HTML to identify, style, and layer
    # useful behavior on top of.
    #
    # Modifications apply to the parsed document directly.
    #
    # Returns nothing.
    def filter!
      doc.xpath(".//li").reverse.each do |li|
        container = li.xpath("./p[1]")[0] || li
        next if not container.inner_html =~ ItemPattern

        item = TaskList::Item.new($1, container.inner_html)
        task_list_items.unshift item
        container.inner_html = render_task_list_item(item)
        li.parent.add_class('task-list')
        li.add_class('task-list-item')
      end
    end

    def call
      filter!
      doc
    end
  end
end