Question

I’m trying to create a simple script that will select specific columns from the unix df - h command. I can use awk to do this but how can we do this in python?

Here is df -h output:

Filesystem                    Size  Used  Avail  Use%  Mounted on
/dev/mapper/vg_base-lv_root   28G   4.8G    22G   19%  /
tmpfs                        814M   176K   814M    1%  /dev/shm
/dev/sda1                    485M   120M   340M   27%  /boot

I want something like:

Column 1:

Filesystem
/dev/mapper/vg_base-lv_root           
tmpfs                 
/dev/sda1

Column 2:

Size
28G
814M 
485M   

Answer 1

You can use op.popen to run the command and retrieve its output, then splitlines and split to split the lines and fields. Run df -Ph rather than df -h so that lines are not split if a column is too long.

df_output_lines = [s.split() for s in os.popen("df -Ph").read().splitlines()]

The result is a list of lines. To extract the first column, you can use [line[0] for line in df_output_lines] (note that columns are numbered from 0) and so on. You may want to use df_output_lines[1:] instead of df_output_lines to strip the title line.

If you already have the output of df -h stored in a file somewhere, you’ll need to join the lines first.

fixed_df_output = re.sub('\n\s+', ' ', raw_df_output.read())
df_output_lines = [s.split() for s in fixed_df_output.splitlines()]

Note that this assumes that neither the filesystem name nor the mount point contain whitespace. If they do (which is possible with some setups on some unix variants), it’s practically impossible to parse the output of df, even df -P. You can use os.statvfs to obtain information on a given filesystem (this is the Python interface to the C function that df calls internally for each filesystem), but there’s no portable way of enumerating the filesystems.