Using fscanf to read in data

Hello. I am supposed to read in a .txt file that has the following format:

string 12 78 (just another column of numbers)

string 35 63 ...

string 78 47 ...

string 45 12 ...

I need to read those in and then work with the numbers separately

I have to use the command fscanf as told by the instructor. my code is:

fid = fopen(abc.txt, 'r')

data = fscanf(fid, %s %d')

However the result turns out to be weird. What am I doing wrong with '%s %d' format?

Is there another way to read in this .txt file using fscanf?

[dpb]

Are the number of numeric columns consistent? If so, it's no major difference excepting you need to repeat the %d format that many times in the format string so it (the format string, that is) matches the record. You say "just another column" so I'll make the presumption of three total and the format would be

fmt=['%*s repmat('%d',1,3)];
data=fscanf(fid,fmt);

NB: the '*' in the string format to skip it-- fscanf can't read a mixed file of string and numeric data; you'll have to use one of the higher level functions such as textscan for that if you need the nonnumeric data as well.

[Star Strider]

What do you mean by ‘weird’? What does your data variable look like?

It would help if you posted a few lines of abc.txt.

[John]

the assignment told me to read the data in row by row. So for exammple,

row1 = string 35 24 78 19 45 ...

(there is actually 12 columns of numbers)

So how do I do this with fscanf? I have tried this:

row1 = fscanf(fid,'%*s %d')

However, this only gives me 6 numbers in that row (while there are 12 numbers here). So how do I make all 12 numbers show up?

Thank you so much.

[John]

Oh and my abc.txt looks like this: (company name followed by money made each month in millions)

CompanyA 15 24 36 78 45 12 89 78 52 45 11 65

CompanyB ... (same, 12 columns of numbers)

CompanyC ... (same)

[dpb]

Well, as I said before, you've got to make the format match the record -- if there are 12 values/record besides the company name then

fmt=['%*s repmat('%d',1,3)];

becomes

fmt=['%*s repmat('%d',1,12)];

for the 12 values instead of the presumed 3 before.

Since it seems you'd want to keep the company ID as well, you'd be well served to forego the instructor's suggestion and use textscan instead.

Then you'll get a cell array that has the string company name as the first column and the remaining numeric data in the rest.

Or, you could use importdata and let it figure out the text in the left column automagically.

ADDENDUM:

Oh, on the "row-by-row" requirement...that entails using a loop and fgetl would be the simplest tool for that route. Not terribly efficient, of course. But, you could then use sscanf on the line and parse the company name off the left by either finding the first blank and using the length or if it's known to be a fixed length as your sample just hardcoding it for the purpose of a HW assignment.

Answer 1

[Alberto]

You are right dpb, i think textscan is better in this case; the next code extract 3 columns according to the format described:

fid= fopen(filename);

a=textscan(fid, '%s %d %d')

fclose(fid)

Will extract a cell containing the three columns, first containing strings, and two others containing 32 bit signed integers.