Florian Pflug
5/19/2005 8:58:00 PM
Jacob Fugal wrote:
>>>Of course, you want to be careful when using eval, but it should cut
>>>it for the simple case...
>>
>>Is there some specialized version which does not evaluate its
>>(arbitrary) input, but just unformat a formatted string (and no more,
>>in particular refuse non-string lexemes such as arbitrary [dangerous]
>>ruby expressions)?
>
> Unfortunately no. A possibility would be to write an encapsulating
> method using a regex to verify the input:
>
> def unescapeString( str )
> raise TypeError, "Argument must be 'just a string'" unless
> str =~ /some nasty regex/
> eval( str )
> end
>
> I don't really want to deal with writing that regex, though, so I'll
> leave it as an exercise for the reader. :)
Hm, what you have to check is if the string to be eval'd represents
_exactly_ one ruby string-literal. So, native regex could be
/^"[^"]*"$/
This will work as long as your string doesn't contain (escaped)
doublequotes - e.g. '"'.inspect =~ /^"[^"]*"$/ fails.
The general rule is to ignore one character after a backslash, or
to ignore 1-3 digits after a backslash, if the first character after a
backslash is a digit (from 0-7)
That leads to
/^"(([^"\\])|(\\[^0-7])|(\\[0-7][0-7]?[0-7]?))*"$/
It basically says: First a quite, then either something harmless (no
quote or backslash), or a backslash any a abitrary char, or a backslash
and one to three digits, and at the one a quote. The middle part may
occur zeor or more times.
Using that, you can even omit the eval completly. Try this:
class String
def uninspect
return nil unless self =~
/^"(([^"\\])|(\\[^0-7])|(\\[0-7][0-7]?[0-7]?))*"$/
(gsub(/\\([^0-7])/) { $1 }).gsub(/\\([0-7][0-7]?[0-7]?)/) {
$1.to_i(8).chr }
end
end
It implement the trhe rules above, and for every possible
string str == str.inspect.uninspect should hold...
greetings, Florian Pflug