Reading and Writing text to a NEW file - Matlab - file

Question

I have a file that contains a full set of values for some sentences which have transcribed for a speech recognition program. Ive been trying to write some matlab code to go through this file and extract the values for each sentence and write them to a new individual file. So instead of having them all in one 'mlf' file i want them in separate files for each sentence.
For example by 'mlf' file (contains all values for all sentences) looks like this:
#!MLF!#
"/N001.lab"
AH
SEE
I
GOT
THEM
MONTHS
AGO
.
"/N002.lab"
WELL
WORK
FOR
LIVE
WIRE
BUT
ERM
.
"/N003.lab"
IM
GOING
TO
SEE
JAMES
VINCENT
MCMORROW
.
etc
So each sentences is separated by the 'Nxxx.lab' and the '.'. I need to create a new file for every Nxxx.lab, for example the file for N001 would just contain:
AH
SEE
I
GOT
THEM
MONTHS
AGO
I've been trying to use fgetline to specify the 'Nxxx.lab' and '.' boundaries, but it doesn't work as i don't know how to write the content into a new file separate from the 'mlf'.
If anyone can give me any guidance of what sort of approach to use would be greatly appreciated!
Cheers!

Answer 1

Try this code (input file test.mlf has to be in the working directory):
%# read the file
filename = 'test.mlf';
fid = fopen(filename,'r');
lines = textscan(fid,'%s','Delimiter','\n','HeaderLines',1);
lines = lines{1};
fclose(fid);
%# find start and stop indices
istart = find(cellfun(#(x) strcmp(x(1),'"'), lines));
istop = find(strcmp(lines, '.'));
assert(numel(istop)==numel(istop) && all(istop>istart),'Check the input file format.')
%# write lines to new files
for k = 1:numel(istart)
filenew = lines{istart(k)}(2:end-1);
fout = fopen(filenew,'wt');
for l = (istart(k)+1):(istop(k)-1)
fprintf(fout,'%s\n',lines{l});
end
fclose(fout);
end
The code assume that the file names are in double-quotes as in your example. If not, you can find istart indices base on a pattern. Or just assuming that entries for new file start from the 2nd line and follows the dot: istart = [1; istop(1:end-1)+1];

Answer 2

You could use a growing cell array to gather the information.
Read one line at a time from the file.
Grab the file name and put it into the first column if its the first read for the sentence.
If the line read is a period, add it to the string and move the index to a row in the array. Write the new file with the content.
This bit of code should help you in building the cell array while appending a string within it. I assume reading line by line is not a problem. You can also retain the carriage returns/new lines within the string ('\n').
%% Declare A
A = {}
%% Fill row 1
A(1,1) = {'file1'}
A(1,2) = {'Sentence 1'}
A(1,2) = { strcat(A{1,2}, ', has been appended')}
%% Fill row 2
A(2,1) = {'file2'}
A(2,2) = {'Sentence 2'}

Answer 3

While I'm sure you can do this with MATLAB, I would suggest you use Perl to split the original file and then process the individual files using MATLAB.
The following Perl script reads the entire file ("xxx.txt") and writes out the individual files according the the "NAME.lab" lines:
open(my $fh, "<", "xxx.txt");
# read the entire file into $contents
# This may not be a good idea if the file is huge.
my $contents = do { local $/; <$fh> };
# iterate over the $contents string and extract the individual
# files
while($contents =~ /"(.*)"\n((.*\n)*?)\./mg) {
# We arrive here with $1 holding the filename
# and $2 the content up to the "." ending the section/sentence.
open(my $fout, ">", $1);
print $fout $2;
close($fout);
}
close($fh);
The multiline regular expression is a bit difficult but it does the job.
For these sort of text manipulation, perl is much faster and useful. A good tool to learn if you process a lot of text.