Ruby 1.9/Rails method return anomality?

rubyonrails-talk
1

Hello

I have a different return value from a method when the last line is either

  str.encode 'UTF-8'   return str.encode 'UTF-8' # or:   str.encode! 'UTF-8'

In the first case the string is not in UTF-8! Both other return a correct string. I changed this several times back and forth and tried it again because I could not believe it. It really behaves like this.

This strange problem occurs with the following routine:

def convert2utf8 str, charset   str.force_encoding charset rescue   str.force_encoding('BINARY').gsub!(Regexp.new('[\x00-\x08\x0B\x0C\x0E-\x1F0\x80-\xFF]',nil,'n'), '.') ensure   str.force_encoding('BINARY').gsub!(Regexp.new('[\x00-\x08\x0B\x0C\x0E-\x1F\x80-\xFF]',nil,'n'), '.') unless str.valid_encoding?   str.encode 'UTF-8' end

When 'str' (comes from an email parsed with Mail 2.2.15 so str.encoding returns #<Encoding:US-ASCII>) consists of ISO-8859-1 encoded string and 'charset' is 'ISO-8859-1' this happens in a Rails controller. I cannot reproduce this in plain irb (will try in Rails console tomorrow). So is this a Rails problem? Kinda 'lazy encode' problem or something?

Regards, T.

2

Hello

I have a different return value from a method when the last line is either

str.encode 'UTF-8' return str.encode 'UTF-8' # or: str.encode! 'UTF-8'

In the first case the string is not in UTF-8! Both other return a correct string. I changed this several times back and forth and tried it again because I could not believe it. It really behaves like this.

This strange problem occurs with the following routine:

def convert2utf8 str, charset str.force_encoding charset rescue str.force_encoding('BINARY').gsub!(Regexp.new('[\x00-\x08\x0B\x0C\x0E-\x1F0 \x80-\xFF]',nil,'n'), '.') ensure str.force_encoding('BINARY').gsub!(Regexp.new('[\x00-\x08\x0B\x0C\x0E-\x1F\ x80-\xFF]',nil,'n'), '.') unless str.valid_encoding? str.encode 'UTF-8' end

When 'str' (comes from an email parsed with Mail 2.2.15 so str.encoding returns #<Encoding:US-ASCII>) consists of ISO-8859-1 encoded string and 'charset' is 'ISO-8859-1' this happens in a Rails controller. I cannot reproduce this in plain irb (will try in Rails console tomorrow). So is this a Rails problem? Kinda 'lazy encode' problem or something?

The return value of a method isn't the result of the ensure block - it's whatever happens before (ie either str.force_encoding charset or your rescue clause). The latter two alternatives work, either because the change what is being returned to be what you want, or because they modify the string in place

For example, if def foo   1 ensure   2 end

then foo() evaluates to 1, not 2

Fred

3

Frederick Cheung wrote in post #980654:

The return value of a method isn't the result of the ensure block - it's whatever happens before (ie either str.force_encoding charset or your rescue clause). The latter two alternatives work, either because the change what is being returned to be what you want, or because they modify the string in place

For example, if def foo   1 ensure   2 end

Oh, so simple! I didn't test it with an ensure clause. So I feel the floor under my feet is still solid. Thank you!