Decoding removes £ entity when given a SafeBuffer

[nathany]

Use Case

The Rails number_to_currency helper returns currency symbols as HTML entities. When exporting an HTML report to CSV, we would like to use the UTF-8 symbol instead.

Versions

• ruby 2.0.0p247 (2013-06-27 revision 41674) [x86_64-darwin12.5.0]
• Rails 4.0.0 (same issue with 3.x)
• htmlentities (4.3.1)

Steps to Reproduce

New rails app:

rails new testapp
cd testapp
echo "gem 'htmlentities'" >> Gemfile
bundle
rails c

In console:

HTMLEntities.new.decode("£12.34") # => "£12.34"
ActiveSupport::SafeBuffer.new("£12.34") # => "£12.34"
HTMLEntities.new.decode(ActiveSupport::SafeBuffer.new("£12.34")) # => "12.34"

It decodes the way I would want when not given a SafeBuffer, but removes the entity entirely when given a SafeBuffer.

Work around

Instead of using a nice generic solution like HTMLEntities, I end up using a few gsubs to get the job done:

str.to_s.gsub("£", "\u00a3").gsub("€", "\u20ac").gsub("č", "\u010d")

(which works, but then this list needs to be maintained).

[threedaymonk]

I've looked into this. In short, ActiveRecord::SafeBuffer breaks gsub.

"hello".gsub(/(.)/) { $1 } # => "hello"
ActiveSupport::SafeBuffer.new("hello").gsub(/(.)/) { $1 } # => "ooooo"

I can't find a generic way to fix this, but I can suggest a workaround: call to_str on the buffer:

HTMLEntities.new.decode(ActiveSupport::SafeBuffer.new("£12.34").to_str) # => "£12.34"

[nathany]

Thanks. I was looking for a work-around and never tried to_str (to_s didn't help).