MATLAB Central Newsreader - sscanf

sscanf

doe john — Mon, 27 Oct 2008 13:46:01 -0400

HI, I have a chain of characters string + number delimited by semi '; ' as following

line ='name.rrr; ; 78;7;6;;string;;0'

first I replacethe ';' by a tab
line = regexprep(line,';','\t')
Then, I try to use
[A, count, errmsg, nextindex]=sscanf(line, '%s, %d,%d,%d,%d,%s,%s,%s,%d')
but I was not able to retrieve those different data. I get the name.rr in A but in error , I get matching failure in format. I don't understand why it doesnt retrieve other data.

Thanks

Joe

Re: sscanf

Walter Roberson — Mon, 27 Oct 2008 15:00:28 -0400

doe john wrote:
> HI, I have a chain of characters string + number delimited by semi '; ' as following

> line ='name.rrr; ; 78;7;6;;string;;0'

> first I replacethe ';' by a tab
> line = regexprep(line,';','\t')
> Then, I try to use
> [A, count, errmsg, nextindex]=sscanf(line, '%s, %d,%d,%d,%d,%s,%s,%s,%d')
> but I was not able to retrieve those different data. I get the name.rr in A but in error ,
> I get matching failure in format. I don't understand why it doesnt retrieve other data.

When you put a comma in your sscanf format, then sscanf will look for a literal comma in
your input. Do not put any delimiter between the various format elements. A lot of people
put spaces, but it is best not to even put those. For example, you might use

'%s%d%d%d%d%s%s%s%d'

However, you are still going to have problems with the empty fields. Both of the
format elements you have specified, %s and %d, skip -all- leading whitespace, where
"whitespace" is tab or space (or vertical tab, or carriage return, or newline.)
The leading %s format will read the 'name.rrr', but the %d after that will skip over
to the 78, the second %d will get the 7, the third %d will get the 6, then the
fourth %d will skip over and try reading from 'string', which is going to fail.
And if it did somehow work (e.g., the string happened to be all digits), then
the %s that follows in the format string would skip over the whitespace to the 0
leaving you with a %s and %d not satisfied.

I would suggest that for your purposes, it would probably be easier to split
at the semi-colon and parse the individual parts.

Re: sscanf

Andres — Mon, 27 Oct 2008 15:02:02 -0400

"doe john" <johnnydoe1515@gmail.com> wrote in message <ge4gmp$r05$1@fred.mathworks.com>...
> HI, I have a chain of characters string + number delimited by semi '; ' as following
>
> line ='name.rrr; ; 78;7;6;;string;;0'
>
> first I replacethe ';' by a tab
> line = regexprep(line,';','\t')
> Then, I try to use
> [A, count, errmsg, nextindex]=sscanf(line, '%s, %d,%d,%d,%d,%s,%s,%s,%d')
> but I was not able to retrieve those different data. I get the name.rr in A but in error , I get matching failure in format. I don't understand why it doesnt retrieve other data.

sscanf reads the first string, but then expects a comma as it appears after the first %s in your format string.

Besides, reading mixed strings and numbers is not very comfortable with sscanf since it will put everything together into a single numeric matrix (read the "Output Characteristics: Both Numeric and Character Values Read" paragraph in the sscanf doc).

Matlab's textscan is much better here:

line ='name.rrr; ; 78;7;6;;string;;0'
A = textscan(line,'%s %d %d %d %d %s %s %s %d','Delimiter',';');

(it should definitely be listed in the "See also" section of the sscanf help page, too)