Sed/awk: Aligning words in a file - linux

Question

I have a file with the following structure:
# #################################################################
# TEXT: MORE TEXT
# TEXT: MORE TEXT
# #################################################################
___________________________________________________________________
ITEM 1
___________________________________________________________________
PROPERTY1: VALUE1_1
PROPERTY222: VALUE2_1
PROPERTY33: VALUE3_1
PROPERTY4444: VALUE4_1
PROPERTY55: VALUE5_1
Description1: Some text goes here
Description2: Some text goes here
___________________________________________________________________
ITEM 2
___________________________________________________________________
PROPERTY1: VALUE1_2
PROPERTY222: VALUE2_2
PROPERTY33: VALUE3_2
PROPERTY4444: VALUE4_2
PROPERTY55: VALUE5_2
Description1: Some text goes here
Description2: Some text goes here
I want to add another item to the file, using sed or awk:
sed -i -r "\$a$PROPERTY1: VALUE1_3" file.txt
sed -i -r "\$a$PROPERTY2222: VALUE2_3" file.txt
etc. So my next item looks like this:
___________________________________________________________________
ITEM 3
___________________________________________________________________
PROPERTY1: VALUE1_3
PROPERTY222: VALUE2_3
PROPERTY33: VALUE3_3
PROPERTY4444: VALUE4_3
PROPERTY55: VALUE5_3
Description1: Some text goes here
Description2: Some text goes here
The column values is jagged. How do I align my values to the left like for previous items? I can see 2 solutions here:
To align the values while inserting them into the file.
To insert the values into the file the way I did it and align them next.
The command
sed -i -r "s|.*:.*|&|g" file.txt
catches the properties and values I want to align, but I haven't been able to align them properly, i.e.
awk '/^.*:.*$/{ printf "%-40s %-70s\n", $1, $2 }' file.txt
It prints out the file, but it includes the description values and tags, cuts the values if they include spaces or dashes. It just a big mess.
I've tried more commands based on what I've found on Stack Overflow and some blogs, but nothing does what I need.
Note: Values of the description tags are not jagged- this is because I write them to the file in a separate way.
What is wrong with my commands? How do I achieve what I need?

Answer 1

When your file is without tabs, try this:
sed -r 's/: +/:\t/' file.txt | expand -20
When this works, redirect the output to a tmpfile and move the tmpfile to file.txt.

Answer 2

You can use gensub and thoughtful field seperators to take care of this:
for i in {1..5}; do
echo $(( 10 ** i )): $i;
done | awk -F ':::' '/^[^:]+:.+/{
$0 = gensub(/: +/, ":::", $0 );
key=( $1 ":" );
printf "%-40s %s\n", key, $2;
}'
The relevant part being where we swap out ": +" for just ":::" and then do a printf to bring it back together.

Answer 3

You could use \t to insert tabs (rather than spaces which is why you get 'jagged' values)
instead of
sed -i -r "\$a$PROPERTY1: VALUE1_3" file.txt
use
sed -i -r "\$a$PROPERTY1:\t\tVALUE1_3" file.txt

Answer 4

All you need to do is remember the existing indentation when inserting the new line, e.g.:
echo 'PROPERTY732: VALUE9_8_7' |
awk -v prop="PROPERTY1" -v val="VALUE1_3" '
match($0,/^PROPERTY[^[:space:]]+[[:space:]]+/) { wid=RLENGTH }
{ print }
END { printf "%-*s%s\n", wid, prop":", val }
'
PROPERTY732: VALUE9_8_7
PROPERTY1: VALUE1_3
but it's not clear that adding 1 line at a time makes sense or where all of the other text you're adding is coming from.
The above will work with any awk on any UNIX system.
If your "properties" don't actually start with the word PROPERTY then you just need to edit your question to show more realistic sample input/output and tell/show us how to distinguish a PROPERTY line from a Description line and, again, the solution will be trivial with awk.