Handle non-ASCII characters in titles

[calleluks]

Hey, thanks for a great gem!

I tried using an ö in the title of an inlined svg and noticed it got stripped from the output.

I think this is happening because to_html outputs HTML entity codes for these characters by default but when these are then parsed with Nokogiri::XML::Document.parse they get stripped out since they aren't supported in XML:

irb(main):006:0> Nokogiri::XML::Document.parse("<svg>ö</svg>\n")
=> #<Nokogiri::XML::Document:0x3ff16b585394 name="document" children=[#<Nokogiri::XML::Element:0x3ff16b585060 name="svg" children=[#<Nokogiri::XML::Text:0x3ff16b584e80 "ö">]>]>
irb(main):007:0> Nokogiri::XML::Document.parse("<svg>ö</svg>\n").to_html
=> "<svg>&ouml;</svg>\n"
irb(main):008:0> Nokogiri::XML::Document.parse("<svg>&ouml;</svg>\n")
=> #<Nokogiri::XML::Document:0x3ff16af2862c name="document" children=[#<Nokogiri::XML::Element:0x3ff16af282f8 name="svg">]>

Explicitly setting the encoding to "UTF-8" when calling to_html disables generating the entities:

irb(main):009:0> Nokogiri::XML::Document.parse("<svg>ö</svg>\n").to_html(encoding: "UTF-8")
=> "<svg>ö</svg>\n"

Setting the encoding to "UTF-8" when parsing the document makes sure it's implicitly set to "UTF-8" wherever to_html is called later.

irb(main):010:0> Nokogiri::XML::Document.parse("<svg>ö</svg>\n", nil, "UTF-8").to_html
=> "<svg>ö</svg>\n"

[jamesmartin]

@calleerlandsson thank you for the detailed explanation of the problem and your patch, I really appreciate the contribution.

[jamesmartin]

Released in v1.2.3