Question

I am trying to read a dataset which looks like this

DATE,TIME,val
1/1/2001,1:00:00,0

with the program

program main
implicit none

real :: val
character(len=8) :: date
character(len=7) :: time
open(1,file='data.csv',status='old')
read(1,*) ! header
read(1,fmt=100)date,time,val
100 FORMAT (A,1x,A,1x,F3.1)
end program

This is all fine if the date and time always have 8 or 7 characters but they don’t e.g.

 4/21/2001,19:00:00,0

How should I declare the format for fortran to read both the date,time,val lines in the example?

Thank you

Answer 1

You could read all the entire data, time, value line into a single string and then process that to extract the individual elements, something along the lines of

program main
implicit none

real :: val
character(len=10) :: date
character(len=8) :: time

character(len=100) :: line
integer :: n1, n2, end

open(1, file='data.csv', status='old')

! Read entire second line into `line`, ignoring the header
read(1,*) ! header
read(1,'(A100)') line

! Determine locations of the first and last comma in `line` and the
! end of the line:
n1  = index(line, ',')
n2  = index(line, ',', back=.True.)
end = len_trim(data)

! Split line up according to position of commas and assign to 
! appropriate variables
date = data(1:n1-1)
time = data(n1+1:n2-1)
read(data(n2+1:end), *) val ! Internal read, for converting from string to float

end program main

Note that this is highly specialised to the example data in the question, but it shouldn’t be too hard to generalise this code.