class Raven::Processor::UTF8Conversion
Constants
- REPLACE
Slightly misnamed - actually just removes any bytes with invalid encoding Previously, our JSON backend required UTF-8. Since we now use the built-in JSON, we can use any encoding, but it must be valid anyway so we can do things like call match and slice on strings
Public Instance Methods
process(value)
click to toggle source
# File lib/raven/processor/utf8conversion.rb, line 9 def process(value) case value when Hash !value.frozen? ? value.merge!(value) { |_, v| process v } : value.merge(value) { |_, v| process v } when Array !value.frozen? ? value.map! { |v| process v } : value.map { |v| process v } when Exception return value if value.message.valid_encoding? clean_exc = value.class.new(remove_invalid_bytes(value.message)) clean_exc.set_backtrace(value.backtrace) clean_exc when String # Encoding::BINARY / Encoding::ASCII_8BIT is a special binary encoding. # valid_encoding? will always return true because it contains all codepoints, # so instead we check if it only contains actual ASCII codepoints, and if # not we assume it's actually just UTF8 and scrub accordingly. if value.encoding == Encoding::BINARY && !value.ascii_only? value = value.dup value.force_encoding(Encoding::UTF_8) end return value if value.valid_encoding? remove_invalid_bytes(value) else value end end
Private Instance Methods
remove_invalid_bytes(string)
click to toggle source
# File lib/raven/processor/utf8conversion.rb, line 43 def remove_invalid_bytes(string) string.scrub(REPLACE) end