Re: how does fscanf with "C" work?

[Thomas D. Dean]

On 09/08/14 02:56, Francesco Potortì wrote:

> I'll try to be more specific.  If I get no answer from this list, I'll
> submit a bug report.
>
> I create a text file made of four lines, like this:
>
> ===File ~/math/workarea/test.txt============================
> field1, field2, field3, field4
> A,        a,        1,       1.0,
> B,        b,        2,        2.0,
> C,        c,        3,        3.0,
> ============================================================
>
> Then this is what i get:
>
> octave> version
> ans = 3.8.2
> octave> fid=fopen("test.txt");
> octave> fgetl(fid);
> octave> [v1, v2, v3, v4, count, errmsg] = fscanf(fid, "%s%s%d,%f,", "C")
> v1 = A,
> v2 = a,
> v3 =  1
> v4 =  1
> count =  4
> errmsg =
> octave> fclose(fid);

I have not been following this, but, fscanf is doing exactly what you 
asked it to.

Read 'help fscanf' carefully.  You are using the second form.

fd=fopen("xx.dat")
fgetl(fd)
[v1, v2, v3, v4, count, errmsg] = fscanf(fd, "%s%s%d,%f", "C")

This is iterative and so, slow.

You can read the entire file using the first form of fscanf.  However, 
since the strings are not quoted, octave will translate the ascii 
characters to numerical byte values.
A==65, B==66, etc.

fd=fopen("xx.dat")
fgetl(fd)
[A,count,msg]=fscanf(fd,"%s%s%d,%f%s")

The contents of A show this conversion and also, the extra ',' on each 
line that must be accounted for in the format.

If you only need the numerical values, you can use csvread()
If you do that, you will notice there are 5 columns, not 4.  And, the 
strings are converted to zeros.

A= csvread("xx.dat")

By the way, your first paragraph sounds threatening and makes many 
people not answer.  And, a bug report on this will most likely be ignored.

Tom Dean