Bash: Separating a file by blank lines and assigning to a list - bash

So i have a file for example
a
b
c
d
I'd like to make the list of the lines with data out of this. The empty line would be the seperator. So above file's list would be
First element = a
Second element = b
c
Third element = d


Replace blank lines with ,, then remove newline characters:
cat <file> | sed 's/^$/, /' | tr -d '\n'


The following awk would do:
awk 'BEGIN{RS="";ORS=",";FS="\n";OFS=""}($1=$1)' file
This adds an extra , at the end. You can get rid of that in the following way:
awk 'BEGIN{RS="";ORS=",";FS="\n";OFS=""}
{$1=$1;s=s $0 ORS}END{sub(ORS"$","",s); print s}' file
But what happened now, by making this slight modification to eliminate the last ORS (i.e. comma), you have to store the full thing in memory. So you could then just do it more boring and less elegant by storing the full file in memory:
awk '{s=s $0}END{gsub(/\n\n/,",",s);gsub(/\n/,"",s); print s}' file
The following sed does exactly the same. Store the full file in memory and process it.
sed ':a;N;$!ba;s/\n\n/,/g;s/\n//g' <file>
There is, however, a way to play it a bit more clever with awk.
awk 'BEGIN{RS=OFS="";FS="\n"}{$1=$1; print (NR>1?",":"")$0}' file


It depends on what you need to do with that data.
With perl, you have a one-liner:
$ perl -00 -lnE 'say "element $. = $_"' file.txt
element 1 = a
element 2 = b
c
element 3 = d
But clearly you need to process the elements in some way, and I suspect Perl is not your cup of tea.
With bash you could do:
elements=()
n=0
while IFS= read -r line; do
[[ $line ]] && elements[n]+="$line"$'\n' || ((n++))
done < file.txt
# strip the trailing newline from each element
elements=("${elements[#]/%$'\n'/}")
# and show what's in the array
declare -p elements
declare -a elements='([0]="a" [1]="b
c" [2]="d")'


$ awk -v RS= '{print "Element " NR " = " $0}' file
Element 1 = a
Element 2 = b
c
Element 3 = d
If you really want to say First Element instead of Element 1 then enjoy the exercise :-).

Related

Grep a line from a file and replace a substring and append the line to the original file in bash?

This is what I want to do.
for example my file contains many lines say :
ABC,2,4
DEF,5,6
GHI,8,9
I want to copy the second line and replace a substring EF(all occurrences) and make it XY and add this line back to the file so the file looks like this:
ABC,2,4
DEF,5,6
GHI,8,9
DXY,5,6
how can I achieve this in bash?
EDIT : I want to do this in general and not necessarily for the second line. I want to grep EF, and do the substition in whatever line is returned.

Here's a simple Awk script.
awk -F, -v pat="EF" -v rep="XY" 'BEGIN { OFS=FS }
$1 ~ pat { x = $1; sub(pat, rep, x); y = $0; sub($1, x, y); a[++n] = y }
1
END { for(i=1; i<=n; i++) print a[i] }' file
The -F , says to use comma as the input field separator (internal variable FS) and in the BEGIN block, we also set that as the output field separator (OFS).
If the first field matches the pattern, we copy the first field into x, substitute pat with rep, and then substitute the first field of the whole line $0 with the new result, and append it to the array a.
1 is a shorthand to say "print the current input line".
Finally, in the END block, we output the values we have collected into a.
This could be somewhat simplified by hardcoding the pattern and the replacement, but I figured it's more useful to make it modular so that you can plug in whatever values you need.
While this all could be done in native Bash, it tends to get a bit tortured; spending the 30 minutes or so that it takes to get a basic understanding of Awk will be well worth your time. Perhaps tangentially see also while read loop extremely slow compared to cat, why? which explains part of the rationale for preferring to use an external tool like Awk over a pure Bash solution.

You can use the sed command:
sed '
/EF/H # copy all matching lines
${ # on the last line
p # print it
g # paste the copied lines
s/EF/XY/g # replace all occurences
s/^\n// # get rid of the extra newline
}'
As a one-liner:
sed '/EF/H;${p;g;s/EF/XY/g;s/^\n//}' file.csv

If ed is available/acceptable, something like:
#!/bin/sh
ed -s file.txt <<-'EOF'
$kx
g/^.*EF.*,.*/t'x
'x+;$s/EF/XY/
,p
Q
EOF
Or in one-line.
printf '%s\n' '$kx' "g/^.*EF.*,.*/t'x" "'x+;\$s/EF/XY/" ,p Q | ed -s file.txt
Change Q to w if in-place editing is needed.
Remove the ,p to silence the output.

Using BASH:
#!/bin/bash
src="${1:-f.dat}"
rep="${2:-XY}"
declare -a new_lines
while read -r line ; do
if [[ "$line" == *EF* ]] ; then
new_lines+=("${line/EF/${rep}}")
fi
done <"$src"
printf "%s\n" "${new_lines[#]}" >> "$src"
Contents of f.dat before:
ABC,2,4
DEF,5,6
GHI,8,9
Contents of f.dat after:
ABC,2,4
DEF,5,6
GHI,8,9
DXY,5,6

Following on from the great answer by #tripleee, you can create a variation that uses a single call to sub() by outputting all records before the substitution is made, then add the updated record to the array to be output with the END rule, e.g.
awk -F, '1; /EF/ {sub(/EF/,"XY"); a[++n]=$0} END {for(i=1;i<=n;i++) print a[i]}' file
Example Use/Output
An expanded input based on your answer to my comment below the question that all occurrences of EF will be replaced with XY in all records, e.g.
$ cat file
ABC,2,4
DEF,5,6
GHI,8,9
EFZ,3,7
Use and output:
$ awk -F, '1; /EF/ {sub(/EF/,"XY"); a[++n]=$0} END {for(i=1;i<=n;i++) print a[i]}' file
ABC,2,4
DEF,5,6
GHI,8,9
EFZ,3,7
DXY,5,6
XYZ,3,7
Let me know if you have questions.

Error in bash script: arithmetic error

I am wrote a simple script to extract text from a bunch of files (*.out) and add two lines at the beginning and a line at the end. Then I add the extracted text with another file to create a new file. The script is here.
#!/usr/bin/env bash
#A simple bash script to extract text from *.out and create another file
for f in *.out; do
#In the following line, n is a number which is extracted from the file name
n=$(echo $f | cut -d_ -f6)
t=$((2 * $n ))
#To extract the necessary text/data
grep " B " $f | tail -${t} | awk 'BEGIN {OFS=" ";} {print $1, $4, $5, $6}' | rev | column -t | rev > xyz.xyz
#To add some text as the first, second and last lines.
sed -i '1i -1 2' xyz.xyz
sed -i '1i $molecule' xyz.xyz
echo '$end' >> xyz.xyz
#To combine the extracted info with another file (ea_input.in)
cat xyz.xyz ./input_ea.in > "${f/abc.out/pqr.in}"
done
./script.sh: line 4: (ls file*.out | cut -d_ -f6: syntax error: invalid arithmetic operator (error token is ".out) | cut -d_ -f6")
How I can correct this error?

In bash, when you use:
$(( ... ))
it treats the contents of the brackets as an arithmetic expression, returning the result of the calculation, and when you use:
$( ... )
it executed the contents of the brackets and returns the output.
So, to fix your issue, it should be as simple as to replace line 4 with:
n=$(ls $f | cut -d_ -f6)
This replaces the outer double brackets with single, and removes the additional brackets around ls $f which should be unnecessary.

The arithmetic error can be avoided by adding spaces between parentheses. You are already using var=$((arithmetic expression)) correctly elsewhere in your script, so it should be easy to see why $( ((ls "$f") | cut -d_ -f6)) needs a space. But the subshells are completely superfluous too; you want $(ls "$f" | cut -d_ -f6). Except ls isn't doing anything useful here, either; use $(echo "$f" | cut -d_ -f6). Except the shell can easily, albeit somewhat clumsily, extract a substring with parameter substitution; "${f#*_*_*_*_*_}". Except if you're using Awk in your script anyway, it makes more sense to do this - and much more - in Awk as well.
Here is an at empt at refactoring most of the processing into Awk.
for f in *.out; do
awk 'BEGIN {OFS=" " }
# Extract 6th _-separated field from input filename
FNR==1 { split(FILENAME, f, "_"); t=2*f[6] }
# If input matches regex, add to array b
/ B / { b[++i] = $1 OFS $4 OFS $5 OFS $6 }
# If array size reaches t, start overwriting old values
i==t { i=0; m=t }
END {
# Print two prefix lines
print "$molecule"; print -1, 2;
# Handle array smaller than t
if (!m) m=i
# Print starting from oldest values (index i + 1)
for(j=1; j<=m; j++) {
# Wrap to beginning of array at end
if(i+j > t) i-=t
print b[i+j]; }
print "$end" }' "$f" |
rev | column -t | rev |
cat - ./input_ea.in > "${f/foo.out/bar.in}"
done
Notice also how we avoid using a temporary file (this would certainly have been avoidable without the Awk refactoring, too) and how we take care to quote all filename variables in double quotes.
The array b contains (up to) the latest t values from matching lines; we collect these into an array which is constrained to never contain more than t values by wrapping the index i back to the beginning of the array when we reach index t. This "circular array" avoids keeping too many values in memory, which would make the script slow if the input file contains many matches.

Search file A for a list of strings located in file B and append the value associated with that string to the end of the line in file A

This is a bit complicated, well I think it is..
I have two files, File A and file B
File A contains delay information for a pin and is in the following format
AD22 15484
AB22 9485
AD23 10945
File B contains a component declaration that needs this information added to it and is in the format:
'DXN_0':
PIN_NUMBER='(AD22,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)';
'DXP_0':
PIN_NUMBER='(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,AD23,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)';
'VREFN_0':
PIN_NUMBER='(AB22,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)';
So what I am trying to achieve is the following output
'DXN_0':
PIN_NUMBER='(AD22,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)';
PIN_DELAY='15484';
'DXP_0':
PIN_NUMBER='(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,AD23,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)';
PIN_DELAY='10945';
'VREFN_0':
PIN_NUMBER='(AB22,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)';
PIN_DELAY='9485';
There is no order to the pin numbers in file A or B
So I'm assuming the following needs to happen
open file A, read first line
search file B for first string field in the line just read
once found in file B at the end of the line add the text "\nPIN_DELAY='"
add the second string filed of the line read from file A
add the following text at the end "';"
repeat by opening file A, read the second line
I'm assuming it will be a combination of sed and awk commands and I'm currently trying to work it out but think this is beyond my knowledge. Many thanks in advance as I know it's complicated..

FILE2=`cat file2`
FILE1=`cat file1`
TMPFILE=`mktemp XXXXXXXX.tmp`
FLAG=0
for line in $FILE1;do
echo $line >> $TMPFILE
for line2 in $FILE2;do
if [ $FLAG == 1 ];then
echo -e "PIN_DELAY='$(echo $line2 | awk -F " " '{print $1}')'" >> $TMPFILE
FLAG=0
elif [ "`echo $line | grep $(echo $line2 | awk -F " " '{print $1}')`" != "" ];then
FLAG=1
fi
done
done
mv $TMPFILE file1
Works for me, you can also add a trap for remove tmp file if user send sigint.

awk to the rescue...
$ awk -vq="'" 'NR==FNR{a[$1]=$2;next} {print; for(k in a) if(match($0,k)) {print "PIN_DELAY=" q a[k] q ";"; next}}' keys data
'DXN_0':
PIN_NUMBER='(AD22,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)';
PIN_DELAY='15484';
'DXP_0':
PIN_NUMBER='(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,AD23,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)';
PIN_DELAY='10945';
'VREFN_0':
PIN_NUMBER='(AB22,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)';
PIN_DELAY='9485';
Explanation: scan the first file for key/value pairs. For each line in the second data file print the line, for any matching key print value of the key in the requested format. Single quotes in awk is little tricky, setting a q variable is one way of handling it.

FINAL Script for my application, A big thank you to all that helped..
# ! /usr/bin/sh
# script created by Adam with a LOT of help from users on stackoverflow
# must pass $1 file (package file from Xilinx)
# must pass $2 file (chips.prt file from the PCB design office)
# remove these temp files, throws error if not present tho, whoops!!
rm DELAYS.txt CHIP.txt OUTPUT.txt
# BELOW::create temp files for the code thanks to Glastis#stackoverflow https://stackoverflow.com/users/5101968/glastis I now know how to do this
DELAYS=`mktemp DELAYS.txt`
CHIP=`mktemp CHIP.txt`
OUTPUT=`mktemp OUTPUT.txt`
# BELOW::grep input file 1 (pkg file from Xilinx) for lines containing a delay in the form of n.n and use TAIL to remove something (can't remember), sed to remove blanks and replace with single space, sed to remove space before \n, use awk to print columns 3,9,10 and feed into awk again to calculate delay provided by fedorqui#stackoverflow https://stackoverflow.com/users/1983854/fedorqui
# In awk, NF refers to the number of fields on the current line. Since $n refers to the field number n, with $(NF-1) we refer to the penultimate field.
# {...}1 do stuff and then print the resulting line. 1 evaluates as True and anything True triggers awk to perform its default action, which is to print the current line.
# $(NF-1) + $NF)/2 * 141 perform the calculation: `(penultimate + last) / 2 * 141
# {$(NF-1)=sprintf( ... ) assign the result of the previous calculation to the penultimate field. Using sprintf with %.0f we make sure the rounding is performed, as described above.
# {...; NF--} once the calculation is done, we have its result in the penultimate field. To remove the last column, we just say "hey, decrease the number of fields" so that the last one gets "removed".
grep -E -0 '[0-9]\.[0-9]' $1 | tail -n +2 | sed -e 's/[[:blank:]]\+/ /g' -e 's/\s\n/\n/g' | awk '{print ","$3",",$9,$10}' | awk '{$(NF-1)=sprintf("%.0f", ($(NF-1) + $NF)/2 * 169); NF--}1' >> $DELAYS
# remove blanks in part file and add additional commas (,) so that the following awk command works properly
cat $2 | sed -e "s/[[:blank:]]\+//" -e "s/(/(,/g" -e 's/)/,)/g' >> $CHIP
# this awk command is provided by karakfa#stackoverflow https://stackoverflow.com/users/1435869/karakfa Explanation: scan the first file for key/value pairs. For each line in the second data file print the line, for any matching key print value of the key in the requested format. Single quotes in awk is little tricky, setting a q variable is one way of handling it. https://stackoverflow.com/questions/32458680/search-file-a-for-a-list-of-strings-located-in-file-b-and-append-the-value-assoc
awk -vq="'" 'NR==FNR{a[$1]=$2;next} {print; for(k in a) if(match($0,k)) {print "PIN_DELAY=" q a[k] q ";"; next}}' $DELAYS $CHIP >> $OUTPUT
# remove the additional commas (,) added in earlier before ) and after ( and you are done..
cat $OUTPUT | sed -e 's/(,/(/g' -e 's/,)/)/g' >> chipsd.prt

How can I find unique characters per line of input?

Is there any way to extract the unique characters of each line?
I know I can find the unique lines of a file using
sort -u file
I would like to determine the unique characters of each line (something like sort -u for each line).
To clarify: given this input:
111223234213
111111111111
123123123213
121212122212
I would like to get this output:
1234
1
123
12

Using sed
sed ':;s/\(.\)\(.*\)\1/\1\2/;t' file
Basically what it does is capture a character and check if it appears anywhere else on the line. It also captures all the characters between these.
Then it replaces all of that including the second occurence with just first occurence and then what was inbetween.
t is test and jumps to the : label if the previous command was successful. Then this repeats until the s/// command fails meaning only unique characters remain.
; just separates commands.
1234
1
123
12
Keeps order as well.

It doesn't get things in the original order, but this awk one-liner seems to work:
awk '{for(i=1;i<=length($0);i++){a[substr($0,i,1)]=1} for(i in a){printf("%s",i)} print "";delete a}' input.txt
Split apart for easier reading, it could be stand-alone like this:
#!/usr/bin/awk -f
{
# Step through the line, assigning each character as a key.
# Repeated keys overwrite each other.
for(i=1;i<=length($0);i++) {
a[substr($0,i,1)]=1;
}
# Print items in the array.
for(i in a) {
printf("%s",i);
}
# Print a newline after we've gone through our items.
print "";
# Get ready for the next line.
delete a;
}
Of course, the same concept can be implemented pretty easily in pure bash as well:
#!/usr/bin/env bash
while read s; do
declare -A a
while [ -n "$s" ]; do
a[${s:0:1}]=1
s=${s:1}
done
printf "%s" "${!a[#]}"
echo ""
unset a
done < input.txt
Note that this depends on bash 4, due to the associative array. And this one does get things in the original order, because bash does a better job of keeping array keys in order than awk.
And I think you've got a solution using sed from Jose, though it has a bunch of extra pipe-fitting involved. :)
The last tool you mentioned was grep. I'm pretty sure you can't do this in traditional grep, but perhaps some brave soul might be able to construct a perl-regexp variant (i.e. grep -P) using -o and lookarounds. They'd need more coffee than is in me right now though.

One way using perl:
perl -F -lane 'print do { my %seen; grep { !$seen{$_}++ } #F }' file
Results:
1234
1
123
12

Another solution,
while read line; do
grep -o . <<< $line | sort -u | paste -s -d '\0' -;
done < file
grep -o . convert 'row line' to 'column line'
sort -u sort letters and remove repetead letters
paste -s -d '\0' - convert 'column line' to 'row line'
- as a filename argument to paste to tell it to use standard input.

This awk should work:
awk -F '' '{delete a; for(i=1; i<=NF; i++) a[$i]; for (j in a) printf "%s", j; print ""}' file
1234
1
123
12
Here:
-F '' will break the record char by char giving us single character in $1, $2 etc.
Note: For non-gnu awk use:
awk 'BEGIN{FS=""} {delete a; for(i=1; i<=NF; i++) a[$i];
for (j in a) printf "%s", j; print ""}' file

This might work for you (GNU sed):
sed 's/\B/\n/g;s/.*/echo "&"|sort -u/e;s/\n//g' file
Split each line into a series of lines. Unique sort those lines. Combine the result back into a single line.

Unique and sorted alternative to the others, using sed and gnu tools:
sed 's/\(.\)/\1\n/g' file | sort | uniq
which produces one character per line; If you want those on one line, just do:
sed 's/\(.\)/\1\n/g' file | sort | uniq | sed ':a;N;$!ba;s/\n//g;'
This has the advantage of showing the characters in sorted order, rather than order of appearance.

How to grep the last occurrence of a line pattern

I have a file with contents
x
a
x
b
x
c
I want to grep the last occurrence,
x
c
when I try
sed -n "/x/,/b/p" file
it lists all the lines, beginning x to c.

I'm not sure if I got your question right, so here are some shots in the dark:
Print last occurence of x (regex):
grep x file | tail -1
Alternatively:
tac file | grep -m1 x
Print file from first matching line to end:
awk '/x/{flag = 1}; flag' file
Print file from last matching line to end (prints all lines in case of no match):
tac file | awk '!flag; /x/{flag = 1};' | tac

grep -A 1 x file | tail -n 2
-A 1 tells grep to print one line after a match line
with tail you get the last two lines.
or in a reversed way:
tac fail | grep -B 1 x -m1 | tac
Note: You should make sure your pattern is "strong" enough so it gets you the right lines. i.e. by enclosing it with ^ at the start and $ at the end.

This might work for you (GNU sed):
sed 'H;/x/h;$!d;x' file
Saves the last x and what follows in the hold space and prints it out at end-of-file.

not sure how to do it using sed, but you can try awk
awk '{a=a"\n"$0; if ($0 == "x"){ a=$0}} END{print a}' file

POSIX vi (or ex or ed), in case it is useful to someone
Done in Command mode, of course
:set wrapscan
Go to the first line and just search Backwards!
1G?pattern
Slower way, without :set wrapscan
G$?pattern
Explanation:
G go to the last line
Move to the end of that line $
? search Backwards for pattern
The first backwards match will be the same as the last forward match
Either way, you may now delete all lines above current (match)
:1,.-1d
or
kd1G
You could also delete to the beginning of the matched line prior to the line deletions with d0 in case there were multiple matches on the same line.
POSIX awk, as suggested at
get last line from grep search on multiple files
awk '(FNR==1)&&s{print s; s=""}/PATTERN/{s=$0}END{if(s) print s}'

if you wanna do awk in truly hideous one-liner fashion but getting awk to resemble closer to functional programming paradigm syntax without having to keep track when the last occurrence is
mawk/mawk2/gawk 'BEGIN { FS = "=7713[0-9]+="; RS = "^$";
} END { print ar0[split($(0 * sub(/\n.+$/,"",$NF)), ar0, ORS)] }'
Here i'm employing multiple awk short-hands :
sub(/[\n.+$/, "", $NF) # trimming all extra rows after pattern
g/sub() returns # of substitutions made, so multiplying that by 0 forces the split() to be splitting $0, the full file, instead.
split() returns # of items in the array (which is another way of saying the position of last element), so even though I've already trimmed out the trailing \n, i still can directly print ar0[split()], knowing that ORS will fill in the missing trailing \n.
That's why this code looks like i'm trying to extract array items before the array itself is defined, but due to flow of logic needed, the array will become defined by the time it reaches print.
Now if you want something simpler, these 2 also work
mawk/gawk 'BEGIN { FS="=7713[0-9]+="; RS = "^$"
} END { $NF = substr($NF, 1, index($NF, ORS));
FS = ORS; $0 = $0; print $(NF-1) }'
or
mawk/gawk '/=7713[0-9]+=/ { lst = $0 } END { print lst }'
I didn't use the same x|c requirements as OP just to showcase these work regardless of whether you need fixed-strings or regex based matches.

The above solutions only work for one single file, to print the last occurrence for many files (say with suffix .txt), use the following bash script
#!/bin/bash
for fn in `ls *.txt`
do
result=`grep 'pattern' $fn | tail -n 1`
echo $result
done
where 'pattern' is what you would like to grep.

Resources