Merge pull request #650 from weshatheleopard/main

Add support for Brotli compression

CHANGELOG.md

@@ -1,5 +1,10 @@
 # Mechanize CHANGELOG
 
+## next / unreleased
+
+* Introduce experimental support for handling Brotli-compressed responses (CRuby only). (#650) @weshatheleopard
+
+
 ## 2.11.0 / 2024-07-18
 
 * The `accept-charset` header is no longer sent. In early versions of Mechanize, circa 2007, this was a common header but now no modern browser sends it, and servers are instructed to ignore it. See #646 for an example of a server that is confused by its presence. (#647) @flavorjones

Gemfile

@@ -1,3 +1,9 @@
 source "https://rubygems.org"
 
 gemspec
+
+gem "minitest", "~> 5.14"
+gem "rake", "~> 13.0"
+gem "rdoc", "~> 6.3"
+gem "rubocop", "~> 1.12"
+gem "brotli", ">= 0.5" unless RUBY_PLATFORM == "java"

lib/mechanize/http/agent.rb

@@ -495,6 +495,34 @@ def content_encoding_inflate body_io
     body_io.close
   end
 
+  ##
+  # Decodes a Brotli-encoded +body_io+
+  #
+  # (Experimental, CRuby only) Although Mechanize will never request a Brotli-encoded response via
+  # `accept-encoding`, buggy servers may return brotli-encoded responses anyway. Let's try to handle
+  # that case if the Brotli gem is loaded.
+  #
+  # If you need to handle Brotli-encoded responses, install the 'brotli' gem and require it in your
+  # application. If the `Brotli` constant is defined, Mechanize will attempt to use it to inflate
+  # the response.
+  #
+  def content_encoding_brotli(body_io)
+    log.debug('deflate brotli body') if log
+
+    unless defined?(::Brotli)
+      raise Mechanize::Error, "cannot deflate brotli-encoded response. Please install and require the 'brotli' gem."
+    end
+
+    begin
+      return StringIO.new(Brotli.inflate(body_io.read))
+    rescue Brotli::Error
+      log.error("unable to brotli-inflate response") if log
+      raise Mechanize::Error, "error inflating brotli-encoded response."
+    end
+  ensure
+    body_io.close
+  end
+
   def disable_keep_alive request
     request['connection'] = 'close' unless @keep_alive
   end
@@ -831,6 +859,8 @@ def response_content_encoding response, body_io
                content_encoding_inflate body_io
              when 'gzip', 'x-gzip' then
                content_encoding_gunzip body_io
+             when 'br' then
+               content_encoding_brotli body_io
              else
                raise Mechanize::Error,
                  "unsupported content-encoding: #{response['Content-Encoding']}"

mechanize.gemspec

@@ -71,9 +71,4 @@ Gem::Specification.new do |spec|
   spec.add_runtime_dependency("rubyntlm", ">= 0.6.3", "~> 0.6")
   spec.add_runtime_dependency("base64") # removed from bundled gems in 3.4, and needed by rubyntlm (which doesn't declare this dependency)
   spec.add_runtime_dependency("nkf") # removed from bundled gems in 3.4
-
-  spec.add_development_dependency("minitest", "~> 5.14")
-  spec.add_development_dependency("rake", "~> 13.0")
-  spec.add_development_dependency("rdoc", "~> 6.3")
-  spec.add_development_dependency("rubocop", "~> 1.12")
 end

test/test_mechanize_http_agent.rb

@@ -2,6 +2,7 @@
 # frozen_string_literal: true
 
 require 'mechanize/test_case'
+require "brotli" unless RUBY_PLATFORM == "java"
 
 class TestMechanizeHttpAgent < Mechanize::TestCase
 
@@ -924,6 +925,46 @@ def @res.content_length() nil end
     assert_equal 'part', body.read
   end
 
+  def test_response_content_encoding_brotli_when_brotli_not_loaded
+    skip("only test this on jruby which doesn't have brotli support") unless RUBY_ENGINE == 'jruby'
+
+    @res.instance_variable_set :@header, 'content-encoding' => %w[br]
+    body_io = StringIO.new("content doesn't matter for this test")
+
+    e = assert_raises(Mechanize::Error) do
+      @agent.response_content_encoding(@res, body_io)
+    end
+    assert_includes(e.message, "cannot deflate brotli-encoded response")
+
+    assert(body_io.closed?)
+  end
+
+  def test_response_content_encoding_brotli
+    skip("jruby does not have brotli support") if RUBY_ENGINE == 'jruby'
+
+    @res.instance_variable_set :@header, 'content-encoding' => %w[br]
+    body_io = StringIO.new(Brotli.deflate("this is compressed by brotli"))
+
+    body = @agent.response_content_encoding(@res, body_io)
+
+    assert_equal("this is compressed by brotli", body.read)
+    assert(body_io.closed?)
+  end
+
+  def test_response_content_encoding_brotli_corrupt
+    skip("jruby does not have brotli support") if RUBY_ENGINE == 'jruby'
+
+    @res.instance_variable_set :@header, 'content-encoding' => %w[br]
+    body_io = StringIO.new("not a brotli payload")
+
+    e = assert_raises(Mechanize::Error) do
+      @agent.response_content_encoding(@res, body_io)
+    end
+    assert_includes(e.message, "error inflating brotli-encoded response")
+    assert_kind_of(Brotli::Error, e.cause)
+    assert(body_io.closed?)
+  end
+
   def test_response_content_encoding_gzip_corrupt
     log = StringIO.new
     logger = Logger.new log