how to read a data file with textscan when the the data type is inconsistant

Hi all,

I am trying to read a data file containg 100000 lines look as follow:

2.40412 0.45270 3.22916 0.00000 0.00000 2276.6 -1.000 0.54 0 0 Ni

1.92015 2.45198 5.47533 0.02500 0.00000 2847.2 0.543 1.25 1 1 Ni

so my script is: MTRX=textscan(fid, '%f %f %f %f %f %f %f %f %f %f %s','delimiter','\n')

it is working fine but the data file sometimes contains line like:

1.#QNAN 1.57080 1.57080 0.05000 0.00000 0.0 -1.000 180.00 0 0 Ni

So textscan stops reading #QNAN and afterwards. I want to read 1.#QNAN as 1.

How I can get rid of #QNAN using textscan.

Cheers

Hossein

Answer 1

[Walter Roberson]

textscan with CommentStyle {'#', 'AN'}

[Matt Tearle]

Nice. I wondered about CommentStyle, but didn't think of that trick.

Answer 2

[Matt Tearle]

A simple hack workaround would be to read the first column as a string, then remove anything non-numeric from that string and convert to numeric:

MTRX=textscan(fid, '%s %f %f %f %f %f %f %f %f %f %s')
MTRX{1} = str2double(regexprep(MTRX{1},'[^.\d]',''))

Answer 3

[Oleg Komarov]

Couldn't find a way, then 2 solutions:

1) substitute in you file .#QNAN with empty '' and import afterwards, replaceinfile could be useful.

2) Import as sting and manipulate content of first column:

fid = fopen('C:\Users\Oleg\Desktop\data.txt');
MTRX = textscan(fid, '%s %f %f %f %f %f %f %f %f %f %s');
fid = fclose(fid);

temp = regexprep(MTRX{1},'.#QNAN','');
[str2double(temp) MTRX{2:end-1}]

[Matt Tearle]

As an alternative version of (1) (without needing replaceinfile):

txt = fileread('filename.txt');

txt = regexprep(txt,'#QNAN','00000');

MTRX = textscan(txt, '%f %f %f %f %f %f %f %f %f %f %s');

[Hossein Alimadadi]

Thanks for your kind help. It works smoothly.