[Surfraw-devel] OS X sed and basic regular expressions

[Donavan-Ross Costaras]

Hi all,

Can I suggest...

--- surfraw-original    2014-11-21 23:40:16.000000000 +0200
+++ surfraw-fix         2014-11-21 23:40:59.000000000 +0200
@@ -324,7 +324,7 @@

 list_elvi() {
     local dir="$1"
-    sed -n 's/^.*elvis:[       ]\+\(.*\)$/\1/p' $(find "$dir" ! -type d ! -type l ! -name '*~' | sort)
+    sed -n 's/^.*elvis:[       ]\{1,\}\(.*\)$/\1/p' $(find "$dir" ! -type d ! -type l ! -name '*~' | sort)
     for script in $(find "$dir" ! -type d -type l | sort) ; do
        printf '%-16s--> %s\n' $(basename "$script") $(basename $(readlink "${script}"))
     done
 

...as OS X basic regular expressions do not support the `+'
metacharacter. I believe this is POSIX.2 and the `+' metacharacter is
non standard, though I may be wrong.

On the dev list archive there is an OS X user who seems to have it
working but he may have forgotten his GNU sed symlink
http://lists.alioth.debian.org/pipermail/surfraw-devel/2011-October/000968.html

Here is an extract of the OS X RE_FORMAT man page with the relevent section.

     Obsolete (``basic'') regular expressions differ in several respects.  `|'
     is an ordinary character and there is no equivalent for its functional-
     ity.  `+' and `?' are ordinary characters, and their functionality can be
     expressed using bounds (`{1,}' or `{0,1}' respectively).

I'd be happy to push the change myself though I've never done so for a
public project and would need guidance. A lot of tests seem to fail with
503 or 403 but I suspect this change would not effect those on my OS X
system but would need to be tested in other OSes.

Thanks for your time.

-- 
Regards,
Donavan-Ross Costaras