matlab reading variables with varying lengths into the workspace - matlab

Question

This is a follow up question to
Reading parameters from a text file into the workspace
I am wondering, how would I read the following:
% ---------------------- details --------------------------
%---------------------------------------------------------------
% location:
lat = 54.35
lon = -2.9833
%
Eitan T suggested using:
fid = fopen(filename);
C = textscan(fid, '%[^= ]%*[= ]%f', 'CommentStyle', '%')
fclose(fid);
to obtain the information from the file and then
lat = C{2}(strcmp(C{1}, 'lat'));
lon = C{2}(strcmp(C{1}, 'lon'));
to obtain the relevant parameters. How could I alter this to read the following:
% ---------------------- details --------------------------
%---------------------------------------------------------------
% location:
lat = 54.35
lon = -2.9833
heights = 0, 2, 4, 7, 10, 13, 16, 19, 22, 25, 30, 35
Volume = 6197333, 5630000, 4958800, 4419400, 3880000, 3340600,
3146800, 2780200, 2413600, 2177000, 1696000, 811000
%
where the variable should contain all of the data points following the equal sign (up until the start of the next variable, Volume in this case)?
Thanks for your help

Answer 1

Here's one method, which uses some filthy string hacking and eval to get the result. This works on your example, but I wouldn't really recommend it:
fid = fopen('data.txt');
contents = textscan(fid, '%s', 'CommentStyle', '%', 'Delimiter', '\n');
contents = contents{1};
for ii = 1:length(contents)
line = contents{ii};
eval( [strrep(line, '=', '=['), '];'] ) # convert to valid Matlab syntax
end
A better method would be to read each of the lines using textscan
for ii = 1:length(contents)
idx = strfind(contents{ii}, ' = ');
vars{ii} = contents{ii}(1:idx-1);
vals(ii) = textscan(contents{ii}(idx+3:end), '%f', 'Delimiter', ',');
end
Now the variables vars and vals have the names of your variables, and their values. To extract the values you could do something like
ix = strmatch('lat', vars, 'exact');
lat = vals{ix};
ix = strmatch('lon', vars, 'exact');
lon = vals{ix};
ix = strmatch('heights', vars, 'exact');
heights = vals{ix};
ix = strmatch('Volume', vars, 'exact');
volume = vals{ix};

Answer 2

This can be accomplished using a 2-step approach:
Read the leading string (first element), equals sign (ignored), and the rest of the line as a string (second element)
Convert these strings-of-the-rest-of-the-lines to floats (second element)
There is however a slight drawback here; your lines seem to follow two formats; one is the one described in step 1), the other is a continuation of the previous line, which contains numbers.
Because of this, an extra step is required:
Read the leading string (first element), equals sign (ignored), and the rest of the line as a string (second element)
This will fail when the "other format" is encountered. Detect this, correct this, and continue
Convert these strings-of-the-rest-of-the-lines to floats (second element)
I think this will do the trick:
fid = fopen('data.txt');
C = [];
try
while ~feof(fid)
% Read next set of data, assuming the "first format"
C_new = textscan(fid, '%[^= ]%*[= ]%s', 'CommentStyle', '%', 'Delimiter', '');
C = [C; C_new]; %#ok
% When we have non-trivial data, check for the "second format"
if ~all(cellfun('isempty', C_new))
% Second format detected!
if ~feof(fid)
% get the line manually
D = fgetl(fid);
% Insert new data from fgetl
C{2}(end) = {[C{2}{end} C{1}{end} D]};
% Correct the cell
C{1}(end) = [];
end
else
% empty means we've reached the end
C = C(1:end-1,:);
end
end
fclose(fid);
catch ME
% Just to make sure to not have any file handles lingering about
fclose(fid);
throw(ME);
end
% convert second elements to floats
C{2} = cellfun(#str2num, C{2}, 'UniformOutput', false);

Answer 3

If you can get rid of the multi-line Volume line, what you have written is valid matlab. So, just invoke the parameter file as a matlab script using the run command.
run(scriptName)
Your only other alternative, as others have shown, is to write what will end up looking like a bastardized Matlab parser. There are definitely better ways to spend your time than doing that!