1. RE: match() in depth!
- Posted by bensler at mail.com
Feb 24, 2002
What's faster?
find("is",test)
or match({"is"},test)
Chris
Derek Parnell wrote:
> Hi Kat,
> I'm not quite sure what you are driving at either.
>
> When you run this code of yours, what do you get displayed?
>
> When I run it, it only reports that 'is' is found. And this is exactly
> what I would have expected.
>
> My assumption is that 'parse()' converts :
>
> "this is a test of match()"
>
> to:
>
> {"this","is","a","test","of","match()"}
>
> In which case the match({"is"},test) call will return 2. Because "is" is
> the second element of the
> test sequence.
>
> In which case the match({"s"},test) call will return 0. Because "s" is
> not any of the elements in
> the test sequence.
>
> What did you find surprising with your test?
> ------
> Derek.
>
>
> 25/02/2002 1:43:42 PM, Kat <gertie at PELL.NET> wrote:
>
> >
> >On 24 Feb 2002, at 22:52, rforno at tutopia.com wrote:
> >
> >>
> >> OK. But what is the strange thing about it? I don't understand.
> >
> >The {bracketed} and "quoted" parts in the examples below.
> >
> >Kat
> >
> >
> >> ----- Original Message -----
> >> From: "Kat" <gertie at PELL.NET>
> >> To: "EUforum" <EUforum at topica.com>
> >> Sent: Sunday, February 24, 2002 12:49 AM
> >> Subject: match() in depth!
> >>
> >>
> >> > I didn't know we could do this!
> >> >
> >> > include strtok.e
> >> > with trace
> >> > sequence test
> >> >
> >> > test = "this is a test of match()"
> >> > test = parse(test,32)
> >> >
> >> >
> >> > if match({"is"},test) then
> >> > puts(1,"found is\n")
> >> > end if
> >> >
> >> > if match({"s"},test) then
> >> > puts(1,"found s\n")
> >> > end if
> >> >
> >> > trace(1)
> >> > abort(0)
> >> >
> >> >
> ---------
> Cheers,
> Derek Parnell
>
>
2. Re: RE: match() in depth!
Chris,
it probably doesn't matter too much which is faster as I imagine that what Kat
is trying to do is
find out which *word* is the "is" word, rather than what its character position
is in the original
string.
However, I find that the find() is about 10% faster than match().
25/02/2002 2:48:26 PM, bensler at mail.com wrote:
>
>What's faster?
>
>find("is",test)
>or match({"is"},test)
>
>Chris
>
>Derek Parnell wrote:
>> Hi Kat,
>> I'm not quite sure what you are driving at either.
>>
>> When you run this code of yours, what do you get displayed?
>>
>> When I run it, it only reports that 'is' is found. And this is exactly
>> what I would have expected.
>>
>> My assumption is that 'parse()' converts :
>>
>> "this is a test of match()"
>>
>> to:
>>
>> {"this","is","a","test","of","match()"}
>>
>> In which case the match({"is"},test) call will return 2. Because "is" is
>> the second element of the
>> test sequence.
>>
>> In which case the match({"s"},test) call will return 0. Because "s" is
>> not any of the elements in
>> the test sequence.
>>
>> What did you find surprising with your test?
>> ------
>> Derek.
>>
>>
>> 25/02/2002 1:43:42 PM, Kat <gertie at PELL.NET> wrote:
>>
>> >
>> >On 24 Feb 2002, at 22:52, rforno at tutopia.com wrote:
>> >
>> >>
>> >> OK. But what is the strange thing about it? I don't understand.
>> >
>> >The {bracketed} and "quoted" parts in the examples below.
>> >
>> >Kat
>> >
>> >
>> >> ----- Original Message -----
>> >> From: "Kat" <gertie at PELL.NET>
>> >> To: "EUforum" <EUforum at topica.com>
>> >> Sent: Sunday, February 24, 2002 12:49 AM
>> >> Subject: match() in depth!
>> >>
>> >>
>> >> > I didn't know we could do this!
>> >> >
>> >> > include strtok.e
>> >> > with trace
>> >> > sequence test
>> >> >
>> >> > test = "this is a test of match()"
>> >> > test = parse(test,32)
>> >> >
>> >> >
>> >> > if match({"is"},test) then
>> >> > puts(1,"found is\n")
>> >> > end if
>> >> >
>> >> > if match({"s"},test) then
>> >> > puts(1,"found s\n")
>> >> > end if
>> >> >
>> >> > trace(1)
>> >> > abort(0)
>> >> >
>> >> >
>> ---------
>> Cheers,
>> Derek Parnell
>>
>>
>
>
>
---------
Cheers,
Derek Parnell
3. Re: RE: match() in depth!
On 25 Feb 2002, at 15:16, Derek Parnell wrote:
>
> Chris,
> it probably doesn't matter too much which is faster as I imagine that what Kat
> is trying to do is find out which *word* is the "is" word, rather than what
> its
> character position is in the original string.
What i was trying to do is find a faster match than the find() was doing in
find_all() in strtok.e. I was surpised that match accepted the {""} in place of
a
"" or a {}. I was hoping there was a way to nest the {} or "" to specify a
nesting level to search in match(). For instance (non-tested code follows):
test = {{"one","two"},{"three","four"}}
match({"three"},test) = 2
match("three",test) = 0
But it doesn't work like that.
Kat
> However, I find that the find() is about 10% faster than match().
> 25/02/2002 2:48:26 PM, bensler at mail.com wrote:
>
> >
> >What's faster?
> >
> >find("is",test)
> >or match({"is"},test)
> >
> >Chris
> >
> >Derek Parnell wrote:
> >> Hi Kat,
> >> I'm not quite sure what you are driving at either.
> >>
> >> When you run this code of yours, what do you get displayed?
> >>
> >> When I run it, it only reports that 'is' is found. And this is exactly
> >> what I would have expected.
> >>
> >> My assumption is that 'parse()' converts :
> >>
> >> "this is a test of match()"
> >>
> >> to:
> >>
> >> {"this","is","a","test","of","match()"}
> >>
> >> In which case the match({"is"},test) call will return 2. Because "is" is
> >> the
> >> second element of the test sequence.
> >>
> >> In which case the match({"s"},test) call will return 0. Because "s" is
> >> not any of the elements in
> >> the test sequence.
> >>
> >> What did you find surprising with your test?
> >> ------
> >> Derek.
> >>
> >>
> >> 25/02/2002 1:43:42 PM, Kat <gertie at PELL.NET> wrote:
> >>
> >> >
> >> >On 24 Feb 2002, at 22:52, rforno at tutopia.com wrote:
> >> >
> >> >>
> >> >> OK. But what is the strange thing about it? I don't understand.
> >> >
> >> >The {bracketed} and "quoted" parts in the examples below.
> >> >
> >> >Kat
> >> >
> >> >
> >> >> ----- Original Message -----
> >> >> From: "Kat" <gertie at PELL.NET>
> >> >> To: "EUforum" <EUforum at topica.com>
> >> >> Sent: Sunday, February 24, 2002 12:49 AM
> >> >> Subject: match() in depth!
> >> >>
> >> >>
> >> >> > I didn't know we could do this!
> >> >> >
> >> >> > include strtok.e
> >> >> > with trace
> >> >> > sequence test
> >> >> >
> >> >> > test = "this is a test of match()"
> >> >> > test = parse(test,32)
> >> >> >
> >> >> >
> >> >> > if match({"is"},test) then
> >> >> > puts(1,"found is\n")
> >> >> > end if
> >> >> >
> >> >> > if match({"s"},test) then
> >> >> > puts(1,"found s\n")
> >> >> > end if
> >> >> >
> >> >> > trace(1)
> >> >> > abort(0)
> >> >> >
> >> >> >
> >> ---------
> >> Cheers,
> >> Derek Parnell
> >>
> >>
> ---------
> Cheers,
> Derek Parnell
>
>
>
>
4. RE: match() in depth!
- Posted by bensler at mail.com
Feb 24, 2002
> > If we use some symbolic representation of the test sequence above it
> > could be
> > coded as :
> >
> > {TIAtOM}
> >
> > where I've simple replaced each word with a letter representing that
> > word.
> >
> > Thus the find for "is" similarly coded is match({I}, test). This will
> > return 2.
{TIAtOM} = parse("this is a test of match()",32}
^ this should really be {T,I,A,T,O,M}
I = "is"
I[1] = 'i'
> > The scan for the "s" word is coded match({S},test) and returns 0 because
> > there
> > is no "s" word in the test sequence.
>
This is a bit confusing. match({S},test) will not return 0, it will fail
with "variable not defined", because there is no S variable.
> You lost me there. You said match({I}, test), which i understand, but
> that is
> {I}, same as "I", which meets the specs for match() in the help files. I
> used
> {"I"}.
{I} != "I"
{I} = {"is"}
{"I"} = {{'I'}} -- (confused again? :P)
> > May I recommend trying to locate "is" word in a standard Euphoria
> > sequence
> > string by doing :
> >
> > find (" is ", test) -- Note the spaces around the word. This might save
> > you
> > parsing it too often.
>
> I know that works too, same thing about the {" "} though. But about
> parsing
> it too often, i parse the lines once, then each item is a slice, and i
> deparse
> before writing each out again. There is at least one change on each of
> 200,000+ lines, and some have dozens of changes.
The reason I asked if find() was faster than match() is because that is
what I thought, but you confused me with match({"is"},test) :)
find(x,s) will compare x against any ELEMENT of s
match(x,s) will compare x against any SLICE of s
EXAMPLE:
find("the","look for the matching element")
returns 0, because there is no element that matches "the"
find("the",parse("look for the matching element",32))
returns 3
match("the","look for the matching slice")
returns 10
If you're parsing every line, find() should be quite a bit faster.
find("is",parse("this is a test",32))
If you only need to find one word in the string, match() without parsing
is faster.
match(" is "," " & "this is a test" & " ") -- you'll need the
concatenated spaces so you can compare the first and last slices
Chris
