If I'm reading your question right, you want to take sequences of 3 lines from the file, and swap the order of the first and last in each? With GNU sed, you could do something like this I think:

sed -e :a -e '$!N' -e 's/\(.*\)\n\(.*\)\n\(.*\)$/\2\n\3\n\1/;Ta' file 

which continually slurps up lines until it is able to make the swap, and then starts over.

With awk:

awk '{l=$0;getline;print;getline;print $0"\n"l;}' < file_in > file_out 

Explanation:

1. Save current record in variable l

2. Get next record using getline

3. Print that record (print, invoked without arguments will always print the current record)

4. Get the next record again using getline

5. Print that record, a newline and the first record stored in l

6. For the next record(s) start over at 1.

References

• awk String Functions

For some one-liner transformations, I like to use pyp (distributed as "pypyp").

Unlabeled lines

If your lines are not labelled, you can do an unconditional transformation like so (%3: 0->2 1->0 2->1):

• $ seq 9 | pyp '[elem for group in zip(lines[1::3],lines[2::3],lines[0::3]) for elem in group]'

Shorter example output looks like this:

$ seq 6 | pyp '[elem for group in zip(lines[0::3],lines[2::3],lines[1::3]) for elem in group]' 1 3 2 4 6 5 

Labeled lines

Data setup:

$ pyp -b 'faker.Faker.seed(202418);f=faker.Faker()' \ 'json.dumps( [dict(name=f.first_name(), address=f.address().replace("\n",", "), phone=f.phone_number()) for i in range(2)])' \ | jq '.[]|to_entries[]|(.key + ": " + .value)' -r \ | tee q202418.data 

Parsing by prefix, not as robust as parsing into records:

$ <q202418.data \ pyp -b 'f=lambda categ: filter(lambda l: l.startswith(categ), lines)' \ 'names=f("name:"); addrs=f("address:"); ps=f("phone:"); [line for group in zip(addrs, ps, names) for line in group]' 

For a discussion of more robust multi-line record parsing, see https://stackoverflow.com/questions/13295083/parsing-a-multi-line-data-file-with-python.

Other pyp notes

pyp's --explain flag can help show what's happening, or allow you to save it as a standalone script:

$ seq 9 | pyp --explain '[elem for group in zip(lines[1::3],lines[2::3],lines[0::3]) for elem in group]' #!/usr/bin/env python3 import sys from pyp import pypprint lines = [x.rstrip('\n') for x in sys.stdin] output = [elem for group in zip(lines[1::3], lines[2::3], lines[0::3]) for elem in group] if output is not None: pypprint(output) 

Pyp supports customization through a PYP_CONFIG_FILE.

pyp performs static analysis of the input code, some AST transformations and pruning, then compiles and executes the result, or with --explain, unparses the AST back to source. It imports packages from its environment, and errors early on unknown or undefined symbols.

See other usage examples at https://pypi.org/project/pypyp/.