Issue #14591 has been updated by jeremyevans0 (Jeremy Evans).


This issue started in Ruby 2.5 and is still present in 2.6.  It appears to be related to the code page.  If I have a directory with Japanese characters (ƥ), the error occurs with my default code page (437), but does not occur if I switch to a UTF-8 code page (65001), or if I use the `-Ku` flag when starting Ruby.  I'm not sure whether this behavior is expected with the non-UTF8 code page.

----------------------------------------
Bug #14591: Files with invalid multi-byte characters will cause Find::find() to raise EINVAL exception
https://bugs.ruby-lang.org/issues/14591#change-81050

* Author: jeffgrover (Jeff Grover)
* Status: Open
* Priority: Normal
* Assignee: 
* Target version: 
* ruby -v: ruby 2.5.0p0 (2017-12-25 revision 61468) [x64-mingw32]
* Backport: 2.3: UNKNOWN, 2.4: UNKNOWN, 2.5: UNKNOWN
----------------------------------------
This can be easily duplicated by the following simple program.  I believe this is mostly going to be a problem for users on Windows, where Unicode filenames are common.  The example below was the name of a real file in my Recycle Bin:

First, create the problematic file:

~~~
c:\Users\me>copy con .000100000003f582f1e810a56094d18e
File contents
^Z
        1 file(s) copied.

c:\Users\me>dir
 Volume in drive C is Windows
 Volume Serial Number is 64A1-A9E3

 Directory of c:\Users\grove\Documents\Ruby

03/08/2018  07:18 AM    <DIR>          .
03/08/2018  07:18 AM    <DIR>          ..
03/08/2018  07:18 AM                15 .000100000003f582f1e810a56094d18e
~~~

Then, run the following simple Ruby program:

require 'find'

~~~
Find.find('.') { |path_name|
	puts path_name
}
~~~

You will see the exception raised:

~~~
Traceback (most recent call last):
        6: from dirhog.rb:22:in `<main>'
        5: from C:/Ruby25-x64/lib/ruby/2.5.0/find.rb:43:in `find'
        4: from C:/Ruby25-x64/lib/ruby/2.5.0/find.rb:43:in `each'
        3: from C:/Ruby25-x64/lib/ruby/2.5.0/find.rb:48:in `block in find'
        2: from C:/Ruby25-x64/lib/ruby/2.5.0/find.rb:48:in `catch'
        1: from C:/Ruby25-x64/lib/ruby/2.5.0/find.rb:51:in `block (2 levels) in find'
C:/Ruby25-x64/lib/ruby/2.5.0/find.rb:51:in `lstat': Invalid argument @ rb_file_s_lstat - c:\$Recycle.Bin/S-1-5-21-2582874610-2078213686-3622711573-1001/.????000100000003f582f1e810a56094d18e (Errno::EINVAL)
~~~

The (work-around) solution I came up with was to add to the list of exceptions already handled by "ignore_errors" in lib/find.rb line 52:

~~~
          begin
            s = File.lstat(file)
          rescue Errno::ENOENT, Errno::EACCES, Errno::ENOTDIR, Errno::ELOOP, Errno::ENAMETOOLONG, Errno::EINVAL
            raise unless ignore_error
            next
          end
~~~

This seems a reasonable compromise for now, although there is probably a better solution which involves dealing with the invalid characters and translating them to something that can be handled.  The reason I am submitting this bug instead of monkey-patching a solution or something is that it is almost impossible to do this externally, you can't catch the exception while in the block of find().

If you want, I can do a GitHub pull request with the change.

As a side note, I'd also like to consider making "ignore_errors=false" the default, instead of true... as it hides problems the programmer might want to know about.  If the programmer doesn't care about errors (as in my case) the documentation should clearly emphasize this option.




-- 
https://bugs.ruby-lang.org/

Unsubscribe: <mailto:ruby-core-request / ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>