Maximum file size among files in a folder

[Maxime Roy]

Hi,

I would like to automate the process of determining the maximum file size among files located in a folder. A code that would work if the size of a file was stored somewhere after use of the "ls" command, say in r(size), would be:

clear
local test = "C:/Users/Stata/Test"
cd `test'
local max = 0
local filelist: dir "." files "*.dta"
foreach file in `filelist' {
ls `file'
local size = r(size)
if `size'>`max'{
local max = `size'
}
}
di `max'

Looking forward to your hints and suggestions.

Thank you!

Max

[Aljar Meesters]

-checksum- returns the length of file in bytes in r(filelen).

[Maxime Roy]

Thank you Aljar!

[Sergiy Radyakin]

An easy but inefficient way of determining the file size. Checksum will also read the file which will take time. For a bunch of gb files that can take a lot of time. Sergiy

[Friedrich Huebler]

Running the code below in a directory with 10 files of about 1 GB each took 55 seconds. The same code took 13 seconds for a directory with about 3000 files with a combined size of 1.2 GB.

local max = 0
local filelist: dir . files "*.*"
foreach file in `filelist' {
  quietly checksum "`file'"
  local size = r(filelen)
  if `size' > `max' {
    local max = `size'
  }
}
di `max'

[Robert Picard]

The new version of filelist (from SSC) can determine the file size with no overhead. See this recent announcement.

[Friedrich Huebler]

I compared filelist to the code in post #5. I looked again at 10 files of about 1 GB each. With the code from post #5 it took 58 seconds to identify the size of the largest file.

timer on 1
local max = 0
local filelist: dir . files "*.*"
foreach file in `filelist' {
  quietly checksum "`file'"
  local size = r(filelen)
  if `size' > `max' {
    local max = `size'
  }
}
di `max'
timer off 1
timer list 1

Result of above code:

. di `max'
1.014e+09

. timer list 1
   1:     58.01 /        1 =      58.0120

With Robert's filelist the size of the largest file was returned almost immediately, in less than 1 second. Unfortunately, I couldn't time the task because filelist seems to clear all timers.

timer on 1
filelist
sum fsize
di r(max)
timer off 1
timer list 1

Result of above code (note that timer list returns no result):

. di r(max)
1.014e+09

. timer list 1

[Robert Picard]

Thanks again Friedrich. I had to scratch my head on this one for a few seconds. Turns out that the clear statement in the program is interpreted using version 9 syntax and that apparently included clearing the timers. I switched to drop _all and that should fix the problem. I'll continue holding on to the fixed version for a few days just in case more comes up.

In the mean time, you can use

set rmsg on

to display the execution time of each command.