Command output with empty values to csv - bash

Question

> lsblk -o NAME,LABEL,FSTYPE,MOUNTPOINT,SIZE,TYPE -x NAME
NAME LABEL FSTYPE MOUNTPOINT SIZE TYPE
nvme0n1 894.3G disk
nvme0n1p1 [SWAP] 4G part
nvme0n1p2 1G part
nvme0n1p3 root /home/cg/root 889.3G part
I need the output of this command in csv format, but all the methods I've tried so far don't handle the empty values correctly, thus generating bad rows like these I got with sed:
> lsblk -o NAME,LABEL,FSTYPE,MOUNTPOINT,SIZE,TYPE -x NAME | sed -E 's/ +/,/g'
NAME,LABEL,FSTYPE,MOUNTPOINT,SIZE,TYPE
nvme0n1,894.3G,disk
nvme0n1p1,[SWAP],4G,part
nvme0n1p2,1G,part
nvme0n1p3,root,/home/cg/root,889.3G,part
Any idea how to add the extra commas for the empty fields?
NAME,LABEL,FSTYPE,MOUNTPOINT,SIZE,TYPE
nvme0n1,,,,894.3G,disk

Answer 1

Make sure that the fields that are possibly empty are at the end of the line. And then re-arrange them in the required sequence.
lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINT,LABEL -x NAME | awk '{ print $1,";",$6,";",$4,";",$5,";",$2,";",$3 }'

Answer 2

Just:
lsblk -o NAME,LABEL,FSTYPE,MOUNTPOINT,SIZE,TYPE -x NAME -r | tr ' ' ','

Answer 3

Not really bash, but a quick and dirty Perl would be something like:
my $state=0;
my #input=<>;
my $maxlength=0;
for my $line ( 0 .. $#input){
my $curlength= length($input[$line]);
if ($curlength>$maxlength){$maxlength=$curlength;}
}
my $fill=' ' x $maxlength;
for my $line ( 0 .. $#input){
chomp $input[$line];
$input[$line]="$input[$line] $fill";
}
for (my $pos=0; $pos<$maxlength; $pos++){
my $spacecol=1;
for my $line ( 0 .. $#input){
if (substr($input[$line],$pos,1) ne ' '){
$spacecol=0;
}
}
if ($spacecol==1){
for my $line ( 0 .. $#input){
substr($input[$line],$pos,1)=';';
}
}
}
for my $line ( 0 .. $#input){
print "$input[$line]\n";
}

Answer 4

Assumptions:
output format is fixed-width
header record does not contain any blank fields
no fields contain white space (ie, only white space occurs between fields)
Design overview:
parse header to get initial index for each field; if all columns are left-justified then this would be all we need to do however, with the existence of right-justified columns (eg, SIZE) we need to look for right-justified values that are longer than the associated header field (ie, the value has a lower index than the associated header)
for non-header rows we loop through our set of potential fields, using substr()/match() to find the non-space fields in the line and ...
if said field starts and ends before the next field's index then add the field's value to our output variable but ...
if said field starts before next field's index but ends after next field's index then we're looking at a right-justified value of the next field which happens to have an earlier index than the associated header's index; in this case update the index for the next field and add a blank value (for the current field) to our output variable
if said field starts after the index of the next field then the current field is empty; again, add the empty/blank value to our output variable
once we've completed processing a line of input print the output to stdout
One awk idea:
awk '
BEGIN { OFS="," }
# use header record to determine initial set of indexes
FNR==1 { maxNF=NF
header=$0
out=sep=""
for (i=1;i<=maxNF;i++) {
match(header,/[^[:space:]]+/) # find first non-space string
ndx[i]=ndx[i-1] + prevlen + RSTART - (i==1 ? 0 : 1) # make note of index
out=out sep substr(header,RSTART,RLENGTH) # add value to our output variable
sep=OFS
prevlen=RLENGTH # need for next pass through loop
header=substr(header,RSTART+RLENGTH) # strip off matched string and repeat loop
}
print out # print header to stdout
ndx[1]=1 # in case 1st field is right-justified, override index and set to 1
next
}
# for rest of records need to determine which fields are empty and/or which fields need the associated index updated
{ out=sep=""
for (i=1;i<maxNF;i++) { # loop through all but last field
restofline=substr($0,ndx[i]) # work with current field thru to end of line
if ( match(restofline,/[^[:space:]]+/) ) # if we find a non-space match ...
if ( ndx[i]-1+RSTART < ndx[i+1] ) # if match starts before index of next field and ...
if ( ndx[i]-1+RSTART+RLENGTH < ndx[i+1] ) # ends before index of next field then ...
out=out sep substr(restofline,RSTART,RLENGTH) # append value to our output variable
else { # else if match finished beyond index of next field then ...
out=out sep "" # this field is empty and ...
diff=ndx[i+1]-(ndx[i]+RSTART-1) # figure the difference and ...
ndx[i+1]-=diff # update the index for the next field
}
else # current field is empty
out=out sep ""
sep=OFS
}
field=substr($0,ndx[maxNF]) # process last field
gsub(/[[:space:]]/,"",field) # remove all remaining spaces
print out, field # print new line to stdout
}
' lsblk.out
This generates:
NAME,LABEL,FSTYPE,MOUNTPOINT,SIZE,TYPE
nvme0n1,,,,894.3G,disk
nvme0n1p1,,,[SWAP],4G,part
nvme0n1p2,,,,1G,part
nvme0n1p3,root,,/home/cg/root,889.3G,part