Scan from string probs

[CraigGraham]

I've just run into difficulty using the "Scan from string" function to parse
a long-format date, and on investigating discovered the examples aren't too
reliable either. If I use the first example, which is the string:

abc,xyz
12.3+56i 7200
..
and the format string %s,%s%f%2d

then the function returns an error 85: Scan From String was unable to
convert the input string into the datatype indicated by the format
specifier.

NI knowledgebase has nothing that looks relevant.

Although I've gotten around it by using the "Scan string for tokens"
function, this would have been a more elegant way, so I'm curious to know
what could have gone wrong- or if it could be some bizarre bug recently
introduced. The function works fine with a mix o
f strings and numbers, and
only seems to fail where you try and scan two strings seperated by a comma.

[Kevin_work]

Craig Graham wrote:

> I've just run into difficulty using the "Scan from string" function to parse
> a long-format date, and on investigating discovered the examples aren't too
> reliable either. If I use the first example, which is the string:
>
> abc,xyz
> 12.3+56i 7200
> .
> and the format string %s,%s%f%2d
>
> then the function returns an error 85: Scan From String was unable to
> convert the input string into the datatype indicated by the format
> specifier.
>
> NI knowledgebase has nothing that looks relevant.
>
> Although I've gotten around it by using the "Scan string for tokens"
> function, this would have been a more elegant way, so I'm curious to know
> what could have gone wrong- or if it could be some bizarre bug recently
> introduced. The function
works fine with a mix of strings and numbers, and
> only seems to fail where you try and scan two strings seperated by a comma.

Craig,
Indeed the example given is incorrect and misleading in more than one way.
In order for that example to work the format string must be "%3s,%s%f%2d"
Also in the example the 3rd output is actually 12.3 +56.0i IF you set the
indicator to be CBD (complex double) the last output should be 72 (not 7).

Here is the reason for this.
The function kinda pre evaluates the input string. The abc,zyx is a string
so the function pops the whole thing off (Meaning the first output is abc,zyx).

There is not a string left for either the ',' or the zyx. The next thing in
line is
the 12.3+56i. This is known to be a number so the string conversion fails and
you get the error 85.
Putting the 3 in front of the first "s" tells the function to take only 3
characters
so that the rest of the conversion works.

The scan from string is a very powerful function but it does take so
me time
to get used to it's operation.
If you want you can send me the string and I will send you the format string to

parse it out.
Kevin Kent

[Michel Farhi]

Hi Craig,

The help file of the scan from string is effectively wrong and misleading.
As I use LV to manipulate a lot of strings, I have discovered the use of
[^character] to scan for a string until I find a specific character.
Look at the attached example VI.

Also, don't forget the "Spredsheet string to Array" function which is very
useful to parse a string containing data separated by the same character
(like a coma or a tab).

Regards

-Michel Farhi



"Kevin B. Kent" wrote in message
news:394009E3.95625B0A@usa.alcatel.com...
> Craig Graham wrote:
>
> > I've just run into difficulty using the "Scan from string" function to
parse
> > a long-format date, and on investigating discovered the examples aren't
too
> > reliable either. If I use the first example, which is the string:
> >
> > abc,xyz
> > 12.3+56i 7200
> > .
> > and the format string %s,%s%f%2d
> >
> > then the function returns an error 85: Scan From String was unable to
> > convert the input string into the datatype indicated by the format
> > specifier.
> >
> > NI knowledgebase has nothing that looks relevant.
> >
> > Although I've gotten around it by using the "Scan string for tokens"
> > function, this would have been a more elegant way, so I'm curious to
know
> > what could have gone wrong- or if it could be some bizarre bug recently
> > introduced. The function works fine with a mix of strings and numbers,
and
> > only seems to fail where you try and scan two strings seperated by a
comma.
>
> Craig,
> Indeed the example given is incorrect and misleading in more than one way.
> In order for that example to work the format string must be "%3s,%s%f%2d"
> Also in the example the 3rd output is actually 12.3 +56.0i IF you set the
> indicator to be CBD (complex double) the last output should be 72 (not 7).
>
> Here is the reason for this.
> The function kinda pre evaluates the input string. The abc,zyx is a string
> so the function pops the whole thing off (Meaning the first output is
abc,zyx).
>
> There is not a string left for either the ',' or the zyx. The next thing
in
> line is
> the 12.3+56i. This is known to be a number so the string conversion fails
and
> you get the error 85.
> Putting the 3 in front of the first "s" tells the function to take only 3
> characters
> so that the rest of the conversion works.
>
> The scan from string is a very powerful function but it does take some
time
> to get used to it's operation.
> If you want you can send me the string and I will send you the format
string to
>
> parse it out.
> Kevin Kent
>



[Attachment scan example.vi, see below]

[KellyH]

Thank you for commenting on this example. We will look into the issue and try to remedy it in the Scan From String online help topic.

We're constantly striving to improve the documentation, so any specific suggestions you have are welcome. In the future, please send any comments on National Instruments documentation to techpubs@ni.com so we can address your concerns in a timely matter.

Kelly Holmes
LabVIEW Documentation Team