reading data from txt file in fortran

Slight modification to the @Andrés Argüello Guillén answer.

Unlike most other solutions, my code does not force you to specify in advance the number of rows and columns.

CHARACTER(128) :: buffer

integer strlen, rows, cols
real, dimension(:,:), allocatable :: x

OPEN (1, file = 'matrix.txt', status='old', action='read')

!Count the number of columns

read(1,'(a)') buffer !read first line WITH SPACES INCLUDED
REWIND(1) !Get back to the file beginning

strlen = len(buffer) !Find the REAL length of a string read
do while (buffer(strlen:strlen) == ' ') 
  strlen = strlen - 1 
enddo

cols=0 !Count the number of spaces in the first line
do i=0,strlen
  if (buffer(i:i) == ' ') then
    cols=cols+1
  endif
enddo

cols = cols+1

!Count the number of rows

rows = 0 !Count the number of lines in a file
DO
  READ(1,*,iostat=io)
  IF (io/=0) EXIT
  rows = rows + 1
END DO

REWIND(1)

print*, 'Number of rows:', rows
print*, 'Number of columns:', cols

allocate(x(rows,cols))

do I=1,rows,1
  read(1,*) x(I,:)
  write(*,*) x(I,:)
enddo

CLOSE (1)

matrix.txt

0.0 1.0 2.0
3.0 4.0 5.0
6.0 7.0 8.0

I had the hardest time ever trying to use read, but finally... If you want to read a matrix stored in a .txt file use this:

program FILEREADER

   real, dimension(:,:), allocatable :: x
   integer :: n,m

   open (unit=99, file='array.txt', status='old', action='read')
   read(99, *), n
   read(99, *), m
   allocate(x(n,m))

   do I=1,n,1
      read(99,*) x(I,:)
      write(*,*) x(I,:)
   enddo

end

And the "array.txt" file must be like this for instance (And placed in the same folder of the main):

4
3
0.0 1.0 2.0
3.0 4.0 5.0
6.0 7.0 8.0
9.0 10.0 11.0

Hope it works for everyone out there

List-directed IO (i.e., *) is easier, especially on input. Nevertheless, there are times to use full IO control so that is worth understanding. On input, the data items and descriptors must line up by column. For input, in Fw.d, the d doesn't matter if you have a decimal point in the data item. The fields must be wide enough on both input and output. There need to be enough descriptors, of types which match the variables and the data items. Compare to this example program:

program test_read

   implicit none
   integer, parameter :: VLI_K = selected_int_kind (18)
   integer, parameter :: DR_K = selected_real_kind (14)

   integer (VLI_K) :: i
   real (DR_K) :: a, b, c, d

   open (unit=15, file="data.txt", status='old',    &
             access='sequential', form='formatted', action='read' )

   read (15, 110)  i, a, b, c, d
   110 format (I16, 4(1X, F10.0) )
   write (*, 120) i, a, b, c, d
   120 format ( I18, 4 (2X, F12.3) )

   read (15, *) i, a, b, c, d
   write (*, 120) i, a, b, c, d

end program test_read

I used fixed format because the editing and inspecting of input files having fixed column structure is easier than that of zigzag data. My problem was how the Fortran run-time reader procedures interpret the presence and absence of decimal dots. I am not sure that my solution was the best but I read the data lines as character arrays, split them to fields having length 12 characters then I read the fields by read(*) statements.