Reading a hex table with some undefined values

I have a text table with hex values. Some of the values are of the form 0xXXXX. I'd like to read in the entire table as doubles with the 0xXXXX entries read as NaN. How do I do that?

[Walter Roberson]

What is the form of the other entries? Could you show us a short example?

[Robert]

All values are in the form of 5 digit hex values such as 0x1BEEF or 0x4DEAD. Some values are 0xXXXXX indicating an undefined value. The columns which have any 0xXXXXX entries are interpreted as strings. Any column with all legal hex values are interpreted as uint32. Ideally it would be nice if this could work with any size hex value. Here's a partial line:

0x00000 0xXXXXX 0xXXXXX 0xXXXXX 0xXXXXX 0xXXXXX 0x0002D 0x0002D

[Robert]

I should mention I used readtable in 2020a.

Answer 1

[Robert]

Thanks for even looking at this. That example helped. But to complete the answer to my original question (convert everything to double and use NaN to mark 0xXXXXX entries) I did this:

% read in data
tbl = readtable('hex.txt');
data = table2cell(tbl);

% find XXXXX entries 
missingindication = '0xXXXXX';
idx = find(strcmp(missingindication, data));   

% read again 
tbl = readtable('hex.txt', 'treatasmissing', missingindication, 'multi', true);
data = double(table2array(tbl));
data(idx) = NaN; 

I don't know if that's the most efficient but it works. Anyway, thanks for your help.

[Walter Roberson]

If it was the "convert to double" that is the reason not to use the code I posted, then after my code use

varnames = data.Properties.VariableNames;
data = varfun(@double, data, '');   %changes variable names
data.Properties.VariableName = varnames;

Answer 2

[Walter Roberson]

filename = 'hex.txt';

missingindication = {'0xXXXXX'};
data = readtable(filename, 'treatasmissing', missingindication, 'readvariable',false, 'delimiter', ' ', 'multi', true);

This code will work for up to 16 hex digits; any column that has a value 2^32 or larger will be returned as uint64 .

If the missing data indicator itself can have different numbers of X characters, then you can expand the cell array, such as

missingindication = arrayfun(@(c) ['0x', repmat('X', 1, c)], 1:32, 'uniform', 0);