csv file with a whitespace before comma

Hello, I have a CSV file that has a whitespace before comma and when I use csvread it adds more columns of data than it has. For example, following are the contents of the csv file.

11 ,22 ,33

If use csvread the output looks like this

>> csvread('test.csv')
ans =
  11    22    33     0     0

Is there any way around this other than removing the white-spaces before csvread call?

Thank you in advance, Bhaskar

[per isakson]

I cannot reproduce your result on R2013a

>> csvread('cssm.csv')
ans =
    11    22    33
>> version
ans =
8.1.0.604 (R2013a)

Please, upload your csv-file with the help of the paper-clip-button

[Bhaskar Edara]

This seems to be an issue with 2015a. It works fine in 2014a for me.

Answer 1

[Kirby Fears]

I reproduced the error in 2015a. Indeed there are two trailing 0 columns when the csv file contains white space, but no trailing columns after deleting the whitespace.

I recommend using delimread . Download the function, add it to your path, then read your file using:

num = delimread('test.csv', ',' ,'num');

[per isakson]

csvread (R2015a) returns the correct result for

11, 22, 33

but not for

11 ,22 ,33

Did I get it right?

[Kirby Fears]

Right

[Bhaskar Edara]

Thank you. For now I will use some preprocessing to make sure. Is there no way to get MATLAB to fix these kind of things? I am surprised such an inconsistency between two versions. This works fine on 2014a but not on 2015a.

[Kirby Fears]

You can report this as a bug; it may get fixed eventually. The built-in csvread and dlmread functions both have this bug that you've pointed out. They also have very limited functionality, so it may be in your best interest to switch to a new function now instead of putting it off.

I think Mathworks is pushing users to adopt tables and readtable() and therefore have neglected these old csv reading functions. So switching to readtable seems like the smart approach if you're cool with dealing in tables instead of simple arrays.

I made the delimread function (linked above) when I ran into the same problem you're facing now. It reads much more flexible input types but sticks with simple array (or cell) outputs.

[Walter Roberson]

csvread() has historically called dlmread() to do the work. dlmread() has historically called textscan() to do its work. textscan() is being actively maintained -- it was, for example, enhanced to handle date formats.

Answer 2

[Jae Song]

How about using the offset parameters of the csvread function. For your example:

csvread('test.csv',0,0,[0,0,0,2])

[Bhaskar Edara]

with out the knowledge of the file offset is difficult to use though.

Answer 3

[Brendan Hamm]

You can also use xlsread or readtable (the latter requires version 2013b or later).

xlsread('test.csv')
readtable('test.csv','ReadVariableNames',false)

[Bhaskar Edara]

Thank you. Unfortunately this is an existing code that seems to work fine for some users and it is a pain to go and change all the code but will keep these options in mind for new development.

[Kirby Fears]

It's worth noting that xlsread should not be used on files that are not explicitly excel-formatted files (.xls, .xlsx, .xlsm, etc) since it will fail to read .csv files in non-Windows operating systems. The xlsread function also has way more overhead than simple text reading functions. It is extremely slow for large csv files.