Trying to sort on two fields first is second criteria

Question

I'm trying to sort on multiple columns a file with GNU sort.

The format of the source file tmp/file.txt is :

03A12345679   2014-09-30-00.00.01.00

• "03" => first column
• "A" => field separator
• "12345679 " => second column
• "2014-09-30-00.00.01.00" => other column (not sorted)

I've tried this command :

sort --field-separator="A" -k2,11 -k1,2 tmp/file.txt > tmp/file_out.txt

With :

• --field-separator="A" => field separator is "A"
• -k2,11 => the second column is the first sort ("12345679 ")
• -k1,2 => the first column is the second sort ("03")

But in the result the sort is not correct (the "03" is before the "01" and "02" lines) :

03A12345679   2014-09-30-00.00.01.000000-0000
01A12345679   12345679  132132155VDVLDLV
02A12345679   FONDSEURO   000012664,120000000M
02A12345679   FR09999956570000009999,999990000F
01A12345680   472958519598601200000000047295824
02A12345680   122121EUR     0000099999,370000000M

I want to have this result :

01A12345679   12345679  132132155VDVLDLV
02A12345679   FONDSEURO   000012664,120000000M
02A12345679   FR09999956570000009999,999990000F
03A12345679   2014-09-30-00.00.01.000000-0000
01A12345680   472958519598601200000000047295824
02A12345680   122121EUR     0000099999,370000000M

How can i do that ?

Answer 1

You seem to be mixing up field positions and character positions. There is no need to split on A; with GNU sort, you can sort on a range of characters within the first space-delimited field:

$ sort -k1.4,1.11 -k1.1,1.2 file
01A12345679   12345679  132132155VDVLDLV
02A12345679   FONDSEURO   000012664,120000000M
02A12345679   FR09999956570000009999,999990000F
03A12345679   2014-09-30-00.00.01.000000-0000
01A12345680   472958519598601200000000047295824
02A12345680   122121EUR     0000099999,370000000M

Answer 2

Use -g (compare according to general numerical value) to sort:

sort --field-separator="A" -g -k2,11 -k1,2 tmp/file.txt > tmp/file_out.txt