dev-cpp-users

Re: [Dev-C++] atof 2...

[Per W. <pw...@ia...>]

Hello!

Don't use atof() for converting from text string to a double.

Use strtod() instead, since it will inform you how many characters it
consumed.

char mystring[] =3D "1.2E-3*3";
char *p =3D mystring;
char *end;
num =3D strtod(p,&end);

Now end will point to the '*' character, since '1.2E-3' is consumed and
returned as the double 0.0012.

Note also when parsing infix expressions, that you normally have to take
care of unary signs, i.e. 1+-2 which should be evaluated as 1 + (-2).

Easiest is to perform the parsing with a recursive-descent parser, i.e.
something like:

eval_expr()
  return eval_additive()

eval_additive()
  while (*p =3D=3D '+' || *p =3D=3D '-')
    if (*p =3D=3D '+') {
      p++;
      n +=3D eval_multiplicative()
    } else {
      p++;
      n -=3D eval_multiplicative()
    }
  return n;

eval_multiplicative()
  n =3D eval_factor()
  while (*p =3D=3D '*' || *p =3D=3D '/')
    if (*p =3D=3D '*') {
      p++;
      n *=3D eval_factor();
    } else {
      p++;
      n /=3D eval_factor();
    }
  return n;

eval_factor()
  if (*p =3D=3D =B4+') {p++; return eval_factor(); }
  if (*p =3D=3D '-') {p++; return -eval_factor(); }
  if (*p =3D=3D '(') {p++; n =3D eval_expr(); if (p !=3D ')') error(); retu=
rn n; }
  return number();

Good luck!

/Per W

On Sun, 19 Jan 2003, j b wrote:

>
>
>
> ---------------------------------
> Gesendet von http://mail.yahoo.de.
> Bis zu 100 MB Speicher bei http://premiummail.yahoo.de.

Re: [Dev-C++] atof 2...

[Ioannis V. <no...@ho...>]

------Original Message-------=0D
 =0D
From: Per Westermark=0D
Date: Sunday, 19 January 2003 10:48:26 PM=0D
To: j b=0D
Cc: dev...@li...=0D
Subject: Re: [Dev-C++] atof 2...=0D
 =0D
Hello!=0D
=0D
Don't use atof() for converting from text string to a double.=0D
=0D
Use strtod() instead, since it will inform you how many characters it=0D
consumed.=0D
=0D
=0D
=0D
=3D=3D> What about sscanf()?=0D
=0D
=0D
=0D
Ioannis Vranos=0D
=0D
* Programming pages: http://www.noicys.freeurl.com=0D
* Alternative URL: http://run.to/noicys=20

Re: [Dev-C++] atof 2...

[Per W. <pw...@ia...>]

sscanf can convert one or multiplie parameters from a text string into
corresponding variables. The problem is that you have very little control
of what sscanf considers white characters etc. And in case of a conversion
problem, it isn't easy to point exactly at the failing character.

With strtod, you get a pointer to the first un-processed character of your
input string. For a parser, the normal action is to process characters one
at a time - that means that strotod is never called unless the parser
knows that the first character is the start of a number, i.e. the next
character to process is 0-9 or a decimal separator (. or , depending
on locale). Uppon return of atrtod, the parser may then immediately
continue parsing at the endptr position.

/Per W

On Sun, 19 Jan 2003, Ioannis Vranos wrote:

> ------Original Message-------
>
> From: Per Westermark
> Date: Sunday, 19 January 2003 10:48:26 PM
> To: j b
> Cc: dev...@li...
> Subject: Re: [Dev-C++] atof 2...
>
> Hello!
>
> Don't use atof() for converting from text string to a double.
>
> Use strtod() instead, since it will inform you how many characters it
> consumed.
>
>
>
> ==> What about sscanf()?
>
>
>
> Ioannis Vranos
>
> * Programming pages: http://www.noicys.freeurl.com
> * Alternative URL: http://run.to/noicys

Re: [Dev-C++] atof 2...

[Ioannis V. <no...@ho...>]

-------Original Message-------=0D
 =0D
From: Per Westermark=0D
Date: =D4=F1=DF=F4=E7, 28 =C9=E1=ED=EF=F5=DC=F1=E9=EF=F2 2003 10:43:25 =F0=
=EC=0D
To: Ioannis Vranos=0D
Cc: dev...@li...=0D
Subject: Re: [Dev-C++] atof 2...=0D
 =0D
sscanf can convert one or multiplie parameters from a text string into=0D
corresponding variables. The problem is that you have very little control=
=0D
of what sscanf considers white characters etc. And in case of a conversio=
n=0D
problem, it isn't easy to point exactly at the failing character.=0D
=0D
With strtod, you get a pointer to the first un-processed character of you=
r=0D
input string. For a parser, the normal action is to process characters on=
e=0D
at a time - that means that strotod is never called unless the parser=0D
knows that the first character is the start of a number, i.e. the next=0D
character to process is 0-9 or a decimal separator (. or , depending=0D
on locale). Uppon return of atrtod, the parser may then immediately=0D
continue parsing at the endptr position.=0D
=0D
/Per W=0D
=0D
=0D
=3D=3D> Yes it's a nice for-double-only valid way too. But i prefer sscan=
f() for
all this kind of stuff. :)=0D
=0D
=0D
Ioannis Vranos=0D
=0D
* Programming pages: http://www.noicys.freeurl.com=0D
* Alternative URL: http://run.to/noicys=20

Re: [Dev-C++] atof 2...

[Per W. <pw...@ia...>]

Just as long as you are aware that strtod, strtof, strtol, ... can perform
tricks that sscanf can not. sscanf works reasonable for processing a
configuration file with known contents. It doesn't work when parsing a
language, or an expression, since you don't know the contents of the line
until you have parsed it.

/Per W

On Tue, 28 Jan 2003, Ioannis Vranos wrote:

> -------Original Message-------
>
> From: Per Westermark
> Date: =D4=F1=DF=F4=E7, 28 =C9=E1=ED=EF=F5=DC=F1=E9=EF=F2 2003 10:43:25 =
=F0=EC
> To: Ioannis Vranos
> Cc: dev...@li...
> Subject: Re: [Dev-C++] atof 2...
>
> sscanf can convert one or multiplie parameters from a text string into
> corresponding variables. The problem is that you have very little control
> of what sscanf considers white characters etc. And in case of a conversio=
n
> problem, it isn't easy to point exactly at the failing character.
>
> With strtod, you get a pointer to the first un-processed character of you=
r
> input string. For a parser, the normal action is to process characters on=
e
> at a time - that means that strotod is never called unless the parser
> knows that the first character is the start of a number, i.e. the next
> character to process is 0-9 or a decimal separator (. or , depending
> on locale). Uppon return of atrtod, the parser may then immediately
> continue parsing at the endptr position.
>
> /Per W
>
>
> =3D=3D> Yes it's a nice for-double-only valid way too. But i prefer sscan=
f() for
> all this kind of stuff. :)
>
>
> Ioannis Vranos
>
> * Programming pages: http://www.noicys.freeurl.com
> * Alternative URL: http://run.to/noicys