sed/awk conditionally delete lines from the start and end of a file - perl

Question

I have several thousand text files which might start with
"
Start of text
but not all of them have the same number of line breaks and not all of them have "
I would like to remove " (if it exists) and any line breaks, if any.
(and the ending too but I'll probably figure it out if you show me how to remove it from the start)
End of file...
"
perl is also ok
my attempt would be something like this with fish shell. awk is probably more performant though
if head -1 | grep \"
sed -i 1d $file
if head -1 | grep '^\r\n$'
sed -i 1d $file
if head -1 | grep '^\r\n$'
sed -i 1d $file
if head -1 | grep '^\r\n$'
sed -i 1d $file
that might actually work I'm going to try it

Answer 1

The simplest way to do this is a 2-pass approach where on the first pass you figure out the beginning and ending line numbers for the "good" lines and on the second you print the lines between those numbers:
awk '
NR==FNR { if (NF && !/^"$/) { if (!beg) beg=NR; end=NR } next }
(beg <= FNR) && (FNR <= end)
' file file
For example given this input:
$ cat file
"
Start of text
but not all of them have the same number of line breaks and not all of them have "
I would like to remove " (if it exists) and any line breaks, if any.
(and the ending too but I'll probably figure it out if you show me how to remove it from the start)
End of file...
"
We can do the following using any awk in any shell on every UNIX box:
$ awk 'NR==FNR{if (NF && !/^"$/) {if (!beg) beg=NR; end=NR} next} (beg <= FNR) && (FNR <= end)' file file
Start of text
but not all of them have the same number of line breaks and not all of them have "
I would like to remove " (if it exists) and any line breaks, if any.
(and the ending too but I'll probably figure it out if you show me how to remove it from the start)
End of file...

Answer 2

You can use ed to do it in a single pass, too:
Something like
printf '%s\n' '1g/^"$/.,/^./-1d' '$g/^"$/?^.?+1,$d' w | ed -s "$file"
Translated: If the first line is nothing but a quote, delete it and any following empty lines. If the last line is nothing but a quote, delete all preceding empty lines and it. Finally write the file back to disk.

Answer 3

This might work for you (GNU sed):
sed '1{/^"$/d};/\S/!d;:a;${/^"$/Md};/\S/{n;ba};$d;N;ba' file
Delete the first line if contains a single ".
Delete all empty lines from the start of the file.
Form a loop for the remainder of the file.
Delete the last line(s) if it/they contains a single ".
If the current line(s) is/are not empty, print it/them, fetch the next and repeat.
If the current line(s) is/are the last and empty, delete it/them.
The current line(s) is/are empty so append the next line and repeat.
N.B. This is a single pass solution and allows for empty lines within the body of the file.
Alternative, memory intensive:
sed -Ez 's/^"?\n+//;s/\n+("\n)?$/\n/' file

Answer 4

In addition to the two-pass processing, here's a one-pass:
awk '!/^"*$/{print b $0;f=1;b=""} f&&/^"*$/{b=b $0 ORS}' file
The program consists of two small parts:
Whenever there's content (lines that contain more than "), print possibly buffered lines and the current input line, set a flag that content has started, and clear the buffer.
If content had started (f), but the current line doesn't contain any content, we may have reached the end, so we buffer these empty lines. Later, (1) will print them or they will be discarded on EOF.