Allow Python scraper to keep empty spans with ids

[thewheat]

Fixes #1081

Source has 3 IDs

Scraped page has 2

Turns out that

• Empty spans with IDs are removed in core/clean_text

  devdocs/lib/docs/filters/core/clean_text.rb

  Line 10 in 887c879

  while html.gsub!(EMPTY_NODES_RGX, ''); end

• sphinx/clean_html does some ID manipulation and removal of some empty spans

  devdocs/lib/docs/filters/sphinx/clean_html.rb

  Lines 39 to 42 in 887c879

  	css('span[id]:empty').each do |node|
  	(node.next_element || node.previous_element)['id'] ||= node['id'] if node.next_element || node.previous_element
  	node.remove
  	end

Code changes done

• Created new sphinx_keep_empty_ids parameter to bypass sphinx/clean_html processing and set python scraper to options[:sphinx_keep_empty_ids] = true
• options[:clean_text] = false to bypass core/clean_text

Other notes

In testing this solve the issue mentioned and likely other scenarios (could be up to 54 scenarios in total)

➜  python~3.10 git:(allow-python-empty-spans) ✗ pwd
/Users/thewheat/src/devdocs/docs/python~3.10
➜  python~3.10 git:(allow-python-empty-spans) ✗ grep -irn  "<span id=.*</span><span id="  | grep datetime
./library/datetime.html:2660:<span id="strftime-strptime-behavior"></span><span id="index-0"></span><h2><code class="xref py py-meth docutils literal notranslate"><span class="pre">strftime()</span></code> and <code class="xref py py-meth docutils literal notranslate"><span class="pre">strptime()</span></code> Behavior<a class="headerlink" href="#strftime-and-strptime-behavior" title="Permalink to this headline">¶</a></h2>
➜  python~3.10 git:(allow-python-empty-spans) ✗ grep -irn  "<span id=.*</span><span id="  | wc -l        
      54

Some others affected that are fixed in my local setup

• Search option flags https://devdocs.io/python~3.7/library/doctest#doctest-options
• Search doctest directives https://devdocs.io/python~3.7/library/doctest#doctest-directives

[simon04]

Great, thank you!