1. RE: Looking for the best "Parse"...
Euphoria allows certain reverse slices (see refman) which lets you
simlify your code -
k = n - m
if v = 1 then
p = r & u[m+1..n]
elsif v = 1 + k then
p = u[1..k] & r
else
p = u[1..v-1] & r & u[v+m..n]
end if
to this -
p = u[1..v-1] & r & u[v+m..n]
Try it. - Colin
Igor Kachan wrote:
> Hello Cassidy Napoli,
>
> Just once more iteration with your program
> and your very interesting *recursive* idea,
> which you can make with the
> *delete doubl* method to get a simple
> and understandable code.
>
> Try please:
>
> --[begin code]
> include wildcard.e
>
> global function parseUserText
> (sequence u, --unParsedText -- u
> sequence e, --escape -- e
> sequence r) --replace -- r
> --
> sequence p --ParsedText -- p
> integer v --varPos -- v
> integer m -- le
> integer n -- lu
> integer k
> --
> v = match(lower(e), lower(u)) -- add explanations in your
> if not v then return u end if -- native language and in English
> --
> m = length(e) -- place for explanation
> n = length(u) -- place for explanation
> k = n - m -- place for explanation
> if v = 1 -- place for explanation
> then p = r & u[m+1..n] -- place for explanation
> elsif v = 1 + k -- place for explanation
> then p = u[1..k] & r -- place for explanation
> else p = u[1..v-1] & r & u[v+m..n] -- place for explanation
> end if -- place for explanation
> return parseUserText(p,e,r) -- place for explanation
> end function
>
> sequence text
>
> text = parseUserText("The $A jumped over the lazy dog.","$A","quick
> brown
> fox")
>
> puts (1, text)
>
> --[end code]
>
> The speed is another question, it requires
> the bench-testings on monotasking OS,
> in plain Dos-32 mode, for example.
>
> Regards,
> Igor Kachan
> kinz at peterlink.ru
>
>
2. RE: Looking for the best "Parse"...
Thanks everyone. I'm still studying the different methods Jiri, Igor,
and Colin posted. But I'll probably uses Igor's slightly improved
version of my method because I understand it the most clearly.
Jiri - Could you explain more about what is happening using your method?
I'm not exactly sure I follow it, what makes it better than recursion?
I'm willing to take it as a matter of faith that it's faster; even
though I'm not concerned with marginal speed differences for this
particular program, I'll store the knowledge away for when I AM
concerned with speed.
Igor, What's "double delete"?
Thanks Again,
Cassidy
3. RE: Looking for the best "Parse"...
Igor and Cassidy,
To summarise reverse slices, if you have a sequence s and a position p
in that sequence, then a slice s[p..p-1] produces an empty sequence {}.
This is handy when doing operations such as inserting a string into
another string, since it works regardless of the insertion point. You
don't have to make a special case for p = 1 or p = length(s)+1.
For example, you can insert t into s at p by the general formula:
s = s[1..p-1] & t & s[p..length(s)]
this will work for any value of p from 1 to length(s)+1. If p = 1 --
ie. prepend t to s -- the formula resolves to:
s = s[1..0] & t & s[1..length(s)] --> s = t & s
if p = length(s)+1 -- ie. append t to s -- then the formula resolves to:
s = s[1..length(s)] & t & s[length(s)+1..length(s)] --> s = s & t
So, Igor's code can be reduced to:
--[begin code]
include wildcard.e
global function parseUserText(sequence u, sequence e, sequence r)
integer v
v = match(lower(e), lower(u))
if not v then
return u
end if
u = u[1..v-1] & r & u[v+length(e)..length(u)]
return parseUserText(u,e,r)
end function
--[end code]
This routine is short, but will be very slow. Jiri's approach should be
used if speed is an issue (which it usually is).
- Colin
Cassidy Napoli wrote:
> Thanks everyone. I'm still studying the different methods Jiri, Igor,
> and Colin posted. But I'll probably uses Igor's slightly improved
> version of my method because I understand it the most clearly.
>
> Jiri - Could you explain more about what is happening using your method?
>
> I'm not exactly sure I follow it, what makes it better than recursion?
> I'm willing to take it as a matter of faith that it's faster; even
> though I'm not concerned with marginal speed differences for this
> particular program, I'll store the knowledge away for when I AM
> concerned with speed.
>
> Igor, What's "double delete"?
>
> Thanks Again,
> Cassidy
>
>
4. RE: Looking for the best "Parse"...
- Posted by bensler at mail.com
Oct 27, 2001
How about just "do/end do" or "loop"
To me "repeat" insinuates I can loop a specified number of times,
similar to a for loop w/o the explicit counter variable(I think that
would be more useful than a simple loop).
For the sake of a single boolean comparison though, I don't think it's
worthwhile to introduce a new structure to euphoria.
How about...
loop x do
--code goes here
end loop
Where x is the number of iterations, or 0 for an infinite loop.
A loop with a value of 0 would be optimized to a simple loop when bound.
Chris
Igor Kachan wrote:
> Yes, Derek, *just* is not good and your
> variant is more similar to Euphoria's
> syntax.
>
> > Yes, a simple loop construct would be handy. The standard method of
> > doing
> > this...
> >
> > while 1 do
> > . . .
> > end while
> >
> > introduces an extra comparison for each iteration. An alternative syntax
> of
> > ...
> >
> > repeat do
> > . . .
> > end repeat
> >
> > might not be such a bad idea.
> > -------
> > Derek
>
>
> Regards,
> Igor Kachan
> kinz at peterlink.ru
>
>
