function tex2txt(infile, outfile)
% Clear the most of the TeX markup from the tex file
%
% tex2txt(infile, outfile)
%
% It reads a .tex file, removes all  the formulae $x$ and $$x$$ 
% repleces '~' for ' ', '---' for '-', gathers paragraphs together
%
% Example
% tex2txt('tex2txt.txt')
%
% http://strijov.com

% just to show an example 
switch nargin
    case 1
        outfile = 'testout.txt';
    case 0
        infile  = 'tex2txt.txt'
        outfile = 'testout.txt';
end

[sts] = textread(infile,'%s','delimiter','\n'); % read the text file

flag=0;
disp('1. remove comments and outline formulas');
for i=1:length(sts)
    s = sts{i};
    if length(s) > 0
        idx = strfind(s,'%');
        if length(idx) >0
            s(idx(1):end) = [];
        end
    end
    [s, flag] = strclear(s,'$$',flag);
    sts{i} = s; 
end

disp('2. remove inline formulas');
for i=1:length(sts)
    s = sts{i};
    [s, flag] = strclear(s,'$',flag);
    sts{i} = s; 
end

disp('2a. remove equations');
for i=1:length(sts)
    s = sts{i};
    [s, flag] = strclear(s,'{equation}',flag);
    sts{i} = s; 
end

disp('3. replace ~ for space and --- for - and space,. for,.');
for i=1:length(sts)
    s = sts{i};
    idx =  strfind(s,'~');
    s(idx) = ' ';    
    idx = strfind(s,'---');
    s(idx) = []; 
    idx = strfind(s,'--');
    s(idx) = []; 
    idx = strfind(s,' .');
    s(idx) = [];
    idx = strfind(s,' ,');
    s(idx) = [];
    sts{i} = s; 
end

disp('4. gather the paragraphs');
cnt=0;
out={};
for i=1:length(sts)
    s = sts{i};
    if isempty(s)
       cnt = cnt + 1; 
       out{cnt}='';
    else
        out{cnt}=[out{cnt}, ' ', s];
    end
end

disp('5. remove double spaces');
flag = 1;
while flag
    flag = 0;
    for i=1:length(out)
        s = out{i};
        idx =  strfind(s,'  ');
        if ~isempty(idx), flag = 1; end
        s(idx) = []; 
        out{i} = s;
    end
    disp('pass...');
end

disp('6. write the output file');
fid=fopen(outfile,'w+'); % dicard existing content, if any
for i=1:length(out)
    s = out{i};
   % disp(s);
    fprintf(fid,'%s\n',s);
end
fclose(fid);
disp('done!')
return

function [str,flag] = strclear(str,pattern,flag) 
    n = length(pattern);
    idx =  strfind(str,pattern);
    tmp=zeros(1,length(str));
    
    start = 1; 
    while ~isempty(idx)
        if flag
            tmp(start:idx(1)+n-1) = 1; % 1 to remove
        end
        flag = swch(flag);
        start = idx(1);
        idx(1)=[];
    end
    
    if flag, tmp(start:end) = 1; end % if there were no flag but the idx is not empty
    
    idx = find(tmp==1);
    str(idx) = [];
return
    
function flag = swch(flag)
% a simple switch function
if flag > 0
    flag = 0;
else
    flag = 1;
end
return