Nokolexbor is a drop-in replacement for Nokogiri. It's 5.2x faster at parsing HTML and up to 997x faster at CSS selectors.
It's a performance-focused HTML5 parser for Ruby based on Lexbor. It supports both CSS selectors and XPath. Nokolexbor's API is designed to be 1:1 compatible as much as possible with Nokogiri's API.
Nokolexbor is shipped with pre-compiled gems on most common platforms:
- Linux:
x86_64, with glibc >= 2.17 - macOS:
x86_64andarm64 - Windows:
ucrt64,mingw32andmingw64
If you are on a supported platform, just jump to the Installation section. Otherwise, you need to install CMake to compile C extensions:
brew install cmake
sudo apt-get install cmake
Add to your Gemfile:
gem 'nokolexbor'Then, run bundle install.
Or, install the gem directly:
gem install nokolexbor
require 'nokolexbor'
require 'open-uri'
# Parse HTML document
doc = Nokolexbor::HTML(URI.open('https://github.com/serpapi/nokolexbor'))
# Search for nodes by css
doc.css('#readme h1', 'article h2', 'p[dir=auto]').each do |node|
puts node.content
end
# Search for text nodes by css
doc.css('#readme p > ::text').each do |text|
puts text.content
end
# Search for nodes by xpath
doc.xpath('//div[@id="readme"]//h1', '//article//h2').each do |node|
puts node.content
end- Nokogiri-compatible APIs.
- High performance HTML parsing, DOM manipulation and CSS selectors engine.
- XPath search engine (ported from libxml2).
- Text nodes CSS selector support:
::text.
cssandat_css- Based on Lexbor.
- Only accepts CSS selectors, doesn't support mixed syntax like
div#abc /text(). - To select text nodes, use pseudo element
::text. e.g.div#abc > ::text. - Performance is much higher than libxml2 based methods.
xpathandat_xpath- Based on libxml2.
- Only accepts XPath syntax.
- Works in the same way as Nokogiri's
xpathandat_xpath.
nokogiri_cssandnokogiri_at_css(requires Nokogiri installed)- Based on libxml2.
- Accept mixed syntax like
div#abc /text(). - Works in the same way as Nokogiri's
cssandat_css.
-
For selector
:nth-of-type(n),nis not affected by prior filter. For example, if we want to select the 3rddivexcluding classaand classb, which will be the lastdivin the following HTML:<body> <div></div> <div class="a"></div> <div class="b"></div> <div></div> <div></div> </body>In Nokogiri, the selector should be
div:not(.a):not(.b):nth-of-type(3)In Nokolexbor,
:notdoes affect the place of the lastdiv(same in browsers), the selector should bediv:not(.a):not(.b):nth-of-type(5), but this losts the purpose of filtering though.
Benchmark parsing google result page (368 KB) and selecting nodes using CSS and XPath. Run on MacBook Pro (2019) 2.3 GHz 8-Core Intel Core i9.
Run with: ruby bench/bench.rb
| Nokolexbor (iters/s) | Nokogiri (iters/s) | Diff | |
|---|---|---|---|
| parsing | 487.6 | 93.5 | 5.22x faster |
| at_css | 50798.8 | 50.9 | 997.87x faster |
| css | 7437.6 | 52.3 | 142.11x faster |
| at_xpath | 57.077 | 53.176 | same-ish |
| xpath | 51.523 | 58.438 | same-ish |
Raw data
Warming up --------------------------------------
Nokolexbor parse 56.000 i/100ms
Nokogiri parse 8.000 i/100ms
Calculating -------------------------------------
Nokolexbor parse 487.564 (±10.9%) i/s - 9.688k in 20.117173s
Nokogiri parse 93.470 (±21.4%) i/s - 1.736k in 20.024163s
Comparison:
Nokolexbor parse: 487.6 i/s
Nokogiri parse: 93.5 i/s - 5.22x (± 0.00) slower
Warming up --------------------------------------
Nokolexbor at_css 5.548k i/100ms
Nokogiri at_css 6.000 i/100ms
Calculating -------------------------------------
Nokolexbor at_css 50.799k (±13.8%) i/s - 987.544k in 20.018481s
Nokogiri at_css 50.907 (±35.4%) i/s - 828.000 in 20.666258s
Comparison:
Nokolexbor at_css: 50798.8 i/s
Nokogiri at_css: 50.9 i/s - 997.87x (± 0.00) slower
Warming up --------------------------------------
Nokolexbor css 709.000 i/100ms
Nokogiri css 4.000 i/100ms
Calculating -------------------------------------
Nokolexbor css 7.438k (±14.7%) i/s - 145.345k in 20.083833s
Nokogiri css 52.338 (±36.3%) i/s - 816.000 in 20.042053s
Comparison:
Nokolexbor css: 7437.6 i/s
Nokogiri css: 52.3 i/s - 142.11x (± 0.00) slower
Warming up --------------------------------------
Nokolexbor at_xpath 2.000 i/100ms
Nokogiri at_xpath 4.000 i/100ms
Calculating -------------------------------------
Nokolexbor at_xpath 57.077 (±31.5%) i/s - 920.000 in 20.156393s
Nokogiri at_xpath 53.176 (±35.7%) i/s - 876.000 in 20.036717s
Comparison:
Nokolexbor at_xpath: 57.1 i/s
Nokogiri at_xpath: 53.2 i/s - same-ish: difference falls within error
Warming up --------------------------------------
Nokolexbor xpath 3.000 i/100ms
Nokogiri xpath 3.000 i/100ms
Calculating -------------------------------------
Nokolexbor xpath 51.523 (±31.1%) i/s - 903.000 in 20.102568s
Nokogiri xpath 58.438 (±35.9%) i/s - 852.000 in 20.001408s
Comparison:
Nokogiri xpath: 58.4 i/s
Nokolexbor xpath: 51.5 i/s - same-ish: difference falls within error