[RubyConf] Create scrubber for replacing double breakpoints into paragraph nodes #284
Conversation
josecolella
changed the title
|
@torihuang @josecolella Thanks again for this PR! Looking at the failing tests, it seems like the HTML4 and the HTML5 parser handle newlines differently, and that's causing a failure one way or the other depending on whether there are newlines in the expected result or not. Demonstration: #! /usr/bin/env ruby
require "bundler/inline"
gemfile do
source "https://rubygems.org"
gem "loofah"
end
input = "<html><body><h1>Hello</h1><div>World</div></body></html>"
doc = Nokogiri::HTML4::Document.parse(input)
doc.at_css("h1").add_next_sibling("<b>there</b>")
doc.at_css("body").inner_html
# => "<h1>Hello</h1>\n" + "<b>there</b><div>World</div>"
doc = Nokogiri::HTML4::Document.parse(input)
doc.at_css("h1").add_next_sibling("<p>there</p>")
doc.at_css("body").inner_html
# => "<h1>Hello</h1>\n" + "<p>there</p>\n" + "<div>World</div>"
doc = Nokogiri::HTML5::Document.parse(input)
doc.at_css("h1").add_next_sibling("<b>there</b>")
doc.at_css("body").inner_html
# => "<h1>Hello</h1><b>there</b><div>World</div>"
doc = Nokogiri::HTML5::Document.parse(input)
doc.at_css("h1").add_next_sibling("<p>there</p>")
doc.at_css("body").inner_html
# => "<h1>Hello</h1><p>there</p><div>World</div>"In any case, now that I understand what's going on, I'll wrap this up by tomorrow! |
|
Sounds good. Thanks @flavorjones |
|
@flavorjones were you still interested in getting this merged? |
|
@josecolella Totally! I've been really distracted the last few weeks, but I will absolutely circle back on this. |
|
@flavorjones Any update here? |
|
@josecolella Really sorry for the delay. This was harder than I expected to wrap up (at least in a way that I didn't think was gross). I will be spending a few weeks (at my new job!) on the sanitizer stack starting in late October and will do my best to get this merged then. |
flavorjones
force-pushed
the
jc-th-add-breakpoint-scrubber
branch
from
59aac52 to
5ac043d
Compare
because the html4 and html5 parsers just handle tags and insert newlines differently, and their presence/absence is orthogonal to wrapping in a `p` tag.
flavorjones
force-pushed
the
jc-th-add-breakpoint-scrubber
branch
from
5ac043d to
4d94183
Compare
|
@torihuang and @josecolella -- Sorry this took so long to get back to! Thank you so much for the work, and for your patience. Merging, will be in v2.24.0 shortly. |
|
Thank you @flavorjones for your time in getting this merged! |
Why?
What?
:double_breakpoint) that replaces double breakpoints into paragraph nodes (thank you @flavorjones )How did we test?
Important
There is a failing test right now where the expectation and actual result match except for newline characters. In discussing with @flavorjones, this might be related to minitest and how it formats html
References #279