From: Aristarkh A Zagorodnikov <xm@...3d.ru>
Date: 2001-07-14T23:28:02+09:00
Subject: [ruby-talk:17842] Bug ... probably another one


		Hello!

	Consider the following example:

def expand_path(path, _path_info = '')
    doc_path = 'documents'
    path = File.expand_path doc_path + path
    pwd = Dir.pwd << '/' << doc_path
    document = path.dup
    path.slice! 0, pwd.length
    puts document
    if File.exist? document
        puts _path_info.inspect
        puts((_path_info.scan /.{8}$/).inspect)
    else
        path = path.split '/'
        _path_info.insert 0, ('/' << path.pop)
        expand_path((path.join '/'), _path_info)
    end
end
expand_path '/index.html/' << ARGV[0].to_s

	This is a ripped-down version of function which is intended to locate the real document and it's path info from the uri: i.e if
given uri is "/index.html/test", and /index.html is a plain file, then uri stripped to "/index.html" and path_info is "test". The
algorithm itself is simple (recursing, cutting one path-delimited element at once until document with such name is found). This
function does not perform the task (real one has support for directories and directory index dociments), but I just stripped it down
to demonstrate:

	Make 'documents' directory, and place 'index.html' file inside it, then put above example in the directory where 'documents' reside
in. Go to that dir and just run it, supplying optional path_info argument.

	The bug does not show first - the path_info string is obviously correct, but when I try to get last 8 characters it gets crazy.

empty path_info - right (empty path_info, no match):
xm@ns:~/ruby/test$ ruby tb.rb
/home/xm/ruby/test/documents/index.html
""
[]

short path_info - right (normal path_info, no match):
xm@ns:~/ruby/test$ ruby tb.rb 123456
/home/xm/ruby/test/documents/index.html/123456
/home/xm/ruby/test/documents/index.html
"/123456"
[]

8-char path_info - right (normal path_info, normal match):
xm@ns:~/ruby/test$ ruby tb.rb 12345678
/home/xm/ruby/test/documents/index.html/12345678
/home/xm/ruby/test/documents/index.html
"/12345678"
["12345678"]

9-char path_info - right (normal path_info, normal match):
xm@ns:~/ruby/test$ ruby tb.rb 1234567890
/home/xm/ruby/test/documents/index.html/1234567890
/home/xm/ruby/test/documents/index.html
"/1234567890"
["34567890"]

a bit longer path_info - WRONG (normal path_info, match contains something which looks like garbage):
xm@ns:~/ruby/test$ ruby tb.rb 12345678901
/home/xm/ruby/test/documents/index.html/12345678901
/home/xm/ruby/test/documents/index.html
"/12345678901"
["\320<\"@\020\000\000\000"]

even longer path_info - WRONG (normal path_info, match contains something which looks like garbage):xm@ns:~/ruby/test$ ruby tb.rb
12345678901324324
/home/xm/ruby/test/documents/index.html/12345678901324324
/home/xm/ruby/test/documents/index.html
"/12345678901324324"
["\000\000)\000\000\000\a\000"]

	What the hell is going on here? Maybe regexps don't work in this case for some reason? What's that garbage then?

Aristarkh A Zagorodnikov, Lead Programmer, W3D Group
http://www.w3d.ru /// xm@w3d.ru /// ICQ UIN 36987938