From: William James <w_a_x_man@...>
Date: 2007-10-03T00:15:09+09:00
Subject: Re: help in my own gsub design..

On Oct 2, 8:24 am, Josselin <josse...@wanadoo.fr> wrote:
> On 2007-10-02 14:46:48 +0200, William James <w_a_x_...@yahoo.com> said:
>
> > On Oct 2, 6:35 am, Josselin <josse...@wanadoo.fr> wrote:
> >> I don't use it very often, rather copying from example, but it's so
> >> useful that I should learn how to write it
> >> I'd appreciate any guru tip on how to proceed once the problem is on
> >> the paper :
>
> >> example :
>
> >> INPUT STRINGS
>
> >> 1-  input can contains spaces (should be eliminated)
>
> >> 2- good patterns :
>
> >> 2a- an Integer  like 99  (no limit)
> >> a   >   >=  <  <=   <>  !=   followed by an Integer      like   >99
>
> >>> =99  <99  <=99  <>99   !=99
>
> >> 2b- a String like:    abcd  abc99  99abc  a99bc
>
> >> 2c- a  % sign  followed by a String
> >> like  %abcd  %abc99  %99abc  %a99bc %abcd
>
> >> 2d- a    String  followed by a  % sign
> >> like  abcd%  abc99%  99abc%  a99bc%
>
> >> 2e- a String leading containing or ending with, one or many * sign(s)
> >> like  a*bcd  a**bc99  99*a*bc  *a99bc a99bc*  *a99bc*     *a9**9b*c*
>
> >> 3-any other pattern will be considered as wrong
>
> >> OUTPUTS
>
> >> 2a ->   an array [characters, Integer]    like   [ "<>", 99]
> >> 2b, 2c, 2d,2e  ->  an array [ nil, String]  like  [nil, abcd]  or [nil,
> >> %abcd] or [nil, a99bc%] or [nil, *a9**9b*c]
>
> >> I am interesting in understanding the design process... : how should I
> >> start when designing such gsub routine ?
> >> should I state it differently on paper first ?  well, any tip that can
> >> help me to build the next ones....
>
> >> thanks for your support
>
> >> joss
>
> > def foo s
> >   case s
> >     when /^(>|>=|<|<=|<>|!=)?(\d+)$/
> >       [ $1 || "", $2.to_i ]
> >     when /^(%)?[a-z\d]+(%)?$/
> >       return nil  if 1 != $~.captures.compact.size
> >       [ nil, $& ]
> >     when /^(?=.*\*)[*a-z\d]+$/
> >       [ nil, $& ]
> >   end
> > end
>
> > p foo( "88" )
> > p foo( "<>88" )
> > p foo( "%33" )
> > p foo( "33%" )
> > p foo( "a*9" )
> > p foo( 'bat' )
>
> thanks a lot , William,  I can see how you built it..  I just have 2
> points to clarify if you don't mind
>
> 1-  return nil  if 1 != $~.captures.compact.size
> I don't catch up

I wrote the function so that it will return nil if the string
doesn't match your patterns.  The line above is equivalent to

if 1 != $~.captures.compact.size
  return nil
end

The number of succesful captures (captured by /^(%)?[a-z\d]+(%)?$/ )
must
be exactly 1; i.e., the string must begin or end with %, but not both.

>
> 2- (?=.*\*)

(?=...) is a zero-width look-ahead; it looks ahead but it doesn't
move ahead in the string.  I put this here to make sure that the
string contains at least one *.

>
> otherwise I commented it .. is my understanding correct ?
>
> def foo s
>   case s

I forgot about spaces.  If they can only be at the start or end of the
string, use
  case s.strip

>     when /^(>|>=|<|<=|<>|!=)?(\d+)$/
>         # start with > OR >= OR < OR <= OR <> OR != (0 or 1 time) then
> digit(0-9) n times

1 or more times.

>       [ $1 || "", $2.to_i ]
>         # first element or nil , second element to integer

If 1st capture is nil, use ""; else use 1st capture.

>
>     when /^(%)?[a-z\d]+(%)?$/
>         # start with % ( 0 or 1 time) then characters a-zA-Z 1 or n times then
> end with  % ( 0 or 1 time)
>       return nil  if 1 != $~.captures.compact.size
>         # return nil if .....
>       [ nil, $& ]
>         # first element nil, second element = parameter

This is a good point.  Instead of $& (what was matched by the reg.ex.)
I should
have used the parameter of the function, s.

>
>     when /^(?=.*\*)[*a-z\d]+$/
>         # start with *( 0 or n times)...... then characters *a-zA-Z 1 or n tiems
>       [ nil, $& ]
>         # first element nil, second element = parameter
>   end
> end