I have a file which contains one line of text with tabs
echo -e "foo\tbar\tfoo2\nx\ty\tz" > file.txt
I'd like to get the first column with cut. It works if I do
$ cut -f 1 file.txt
foo
x
But if I read it in a bash script
while read line
do
new_name=`echo -e $line | cut -f 1`
echo -e "$new_name"
done < file.txt
Then I get instead
foo bar foo2
x y z
What am I doing wrong?
/edit: My script looks like that right now
while IFS=$'\t' read word definition
do
clean_word=`echo -e $word | external-command'`
echo -e "$clean_word\t<b>$word</b><br>$definition" >> $2
done < $1
External command removes diacritics from a Greek word. Can the script be optimized any further without changing external-command?
What is happening is that you did not quote $line when reading the file. Then, the original tab-delimited format was lost and instead of tabs, spaces show in between words. And since cut's default delimiter is a TAB, it does not find any and it prints the whole line.
So quoting works:
while read line
do
new_name=`echo -e "$line" | cut -f 1`
#----------------^^^^^^^
echo -e "$new_name"
done < file.txt
Note, however, that you could have used IFS to set the tab as field separator and read more than one parameter at a time:
while IFS=$'\t' read name rest;
do
echo "$name"
done < file.txt
returning:
foo
x
And, again, note that awk is even faster for this purpose:
$ awk -F"\t" '{print $1}' file.txt
foo
x
So, unless you want to call some external command while looping the file, awk (or sed) is better.
Related
I have file with a few lines:
x 1
y 2
z 3 t
I need to pass each line as paramater to some program:
$ program "x 1" "y 2" "z 3 t"
I know how to do it with two commands:
$ readarray -t a < file
$ program "${a[#]}"
How can i do it with one command? Something like that:
$ program ??? file ???
The (default) options of your readarray command indicate that your file items are separated by newlines.
So in order to achieve what you want in one command, you can take advantage of the special IFS variable to use word splitting w.r.t. newlines (see e.g. this doc) and call your program with a non-quoted command substitution:
IFS=$'\n'; program $(cat file)
As suggested by #CharlesDuffy:
you may want to disable globbing by running beforehand set -f, and if you want to keep these modifications local, you can enclose the whole in a subshell:
( set -f; IFS=$'\n'; program $(cat file) )
to avoid the performance penalty of the parens and of the /bin/cat process, you can write instead:
( set -f; IFS=$'\n'; exec program $(<file) )
where $(<file) is a Bash equivalent to to $(cat file) (faster as it doesn't require forking /bin/cat), and exec consumes the subshell created by the parens.
However, note that the exec trick won't work and should be removed if program is not a real program in the PATH (that is, you'll get exec: program: not found if program is just a function defined in your script).
Passing a set of params should be more organized :
In this example case I'm looking for a file containing chk_disk_issue=something etc.. so I set the values by reading a config file which I pass in as a param.
# -- read specific variables from the config file (if found) --
if [ -f "${file}" ] ;then
while IFS= read -r line ;do
if ! [[ $line = *"#"* ]]; then
var="$(echo $line | cut -d'=' -f1)"
case "$var" in
chk_disk_issue)
chk_disk_issue="$(echo $line | tr -d '[:space:]' | cut -d'=' -f2 | sed 's/[^0-9]*//g')"
;;
chk_mem_issue)
chk_mem_issue="$(echo $line | tr -d '[:space:]' | cut -d'=' -f2 | sed 's/[^0-9]*//g')"
;;
chk_cpu_issue)
chk_cpu_issue="$(echo $line | tr -d '[:space:]' | cut -d'=' -f2 | sed 's/[^0-9]*//g')"
;;
esac
fi
done < "${file}"
fi
if these are not params then find a way for your script to read them as data inside of the script and pass in the file name.
What I have is a file (let's call it 'xfile'), containing lines such as
file1 <- this line goes to file1
file2 <- this goes to file2
and what I want to do is run a script that does the work of actually taking the lines and writing them into the file.
The way I would do that manually could be like the following (for the first line)
(echo "this line goes to file1"; echo) >> file1
So, to automate it, this is what I tried to do
IFS=$'\n'
for l in $(grep '[a-z]* <- .*' xfile); do
$(echo $l | sed -e 's/\([a-z]*\) <- \(.*\)/(echo "\2"; echo)\>\>\1/g')
done
unset IFS
But what I get is
-bash: file1(echo "this content goes to file1"; echo)>>: command not found
-bash: file2(echo "this goes to file2"; echo)>>: command not found
(on OS X)
What's wrong?
This solves your problem on Linux
awk -F ' <- ' '{print $2 >> $1}' xfile
Take care in choosing field-separator in such a way that new files does not have leading or trailing spaces.
Give this a try on OSX
You can use the regex capabilities of bash directly. When you use the =~ operator to compare a variable to a regular expression, bash populates the BASH_REMATCH array with matches from the groups in the regex.
re='(.*) <- (.*)'
while read -r; do
if [[ $REPLY =~ $re ]]; then
file=${BASH_REMATCH[1]}
line=${BASH_REMATCH[2]}
printf '%s\n' "$line" >> "$file"
fi
done < xfile
I have a file with contents:
abc|r=1,f=2,c=2
abc|r=1,f=2,c=2;r=3,f=4,c=8
I want a result like below:
abc|r=1,f=2,c=2|1
abc|r=1,f=2,c=2;r=3,f=4,c=8|1
abc|r=1,f=2,c=2;r=3,f=4,c=8|3
The third column value is r value. A new line would be inserted for each occurrence.
I have tried with:
for i in `cat $xxxx.txt`
do
#echo $i
live=$(echo $i | awk -F " " '{print $1}')
home=$(echo $i | awk -F " " '{print $2}')
echo $live
done
but is not working properly. I am a beginner to sed/awk and not sure how can I use them. Can someone please help on this?
awk to the rescue!
$ awk -F'[,;|]' '{c=0;
for(i=2;i<=NF;i++)
if(match($i,/^r=/)) a[c++]=substr($i,RSTART+2);
delim=substr($0,length($0))=="|"?"":"|";
for(i=0;i<c;i++) print $0 delim a[i]}' file
abc|r=1,f=2,c=2|1
abc|r=1,f=2,c=2;r=3,f=4,c=8|1
abc|r=1,f=2,c=2;r=3,f=4,c=8|3
Use an inner routine (made up of GNU grep, sed, and tr) to compile a second more elaborate sed command, the output of which needs further cleanup with more sed. Call the input file "foo".
sed -n $(grep -no 'r=[0-9]*' foo | \
sed 's/^[0-9]*/&s#.*#\&/;s/:r=/|/;s/.*/&#p;/' | \
tr -d '\n') foo | \
sed 's/|[0-9|]*|/|/'
Output:
abc|r=1,f=2,c=2|1
abc|r=1,f=2,c=2;r=3,f=4,c=8|1
abc|r=1,f=2,c=2;r=3,f=4,c=8|3
Looking at the inner sed code:
grep -no 'r=[0-9]*' foo | \
sed 's/^[0-9]*/&s#.*#\&/;s/:r=/|/;s/.*/&#p;/' | \
tr -d '\n'
It's purpose is to parse foo on-the-fly (when foo changes, so will the output), and in this instance come up with:
1s#.*#&|1#p;2s#.*#&|1#p;2s#.*#&|3#p;
Which is almost perfect, but it leaves in old data on the last line:
sed -n '1s#.*#&|1#p;2s#.*#&|1#p;2s#.*#&|3#p;' foo
abc|r=1,f=2,c=2|1
abc|r=1,f=2,c=2;r=3,f=4,c=8|1
abc|r=1,f=2,c=2;r=3,f=4,c=8|1|3
...which old data |1 is what the final sed 's/|[0-9|]*|/|/' removes.
Here is a pure bash solution. I wouldn't recommend actually using this, but it might help you understand better how to work with files in bash.
# Iterate over each line, splitting into three fields
# using | as the delimiter. (f3 is only there to make
# sure a trailing | is not included in the value of f2)
while IFS="|" read -r f1 f2 f3; do
# Create an array of variable groups from $f2, using ;
# as the delimiter
IFS=";" read -a groups <<< "$f2"
for group in "${groups[#]}"; do
# Get each variable from the group separately
# by splitting on ,
IFS=, read -a vars <<< "$group"
for var in "${vars[#]}"; do
# Split each assignment on =, create
# the variable for real, and quit once we
# have found r
IFS== read name value <<< "$var"
declare "$name=$value"
[[ $name == r ]] && break
done
# Output the desired line for the current value of r
printf '%s|%s|%s\n' "$f1" "$f2" "$r"
done
done < $xxxx.txt
Changes for ksh:
read -A instead of read -a.
typeset instead of declare.
If <<< is a problem, you can use a here document instead. For example:
IFS=";" read -A groups <<EOF
$f2
EOF
I would like to remove a file name only from the following configuration file.
Configuration File -- test.conf
knowledgebase/arun/test.rf
knowledgebase/arunraj/tester/test.drl
knowledgebase/arunraj2/arun/test/tester.drl
The above file should be read. And removed contents should went to another file called output.txt
Following are my try. It is not working to me at all. I am getting empty files only.
#!/bin/bash
file=test.conf
while IFS= read -r line
do
# grep --exclude=*.drl line
# awk 'BEGIN {getline line ; gsub("*.drl","", line) ; print line}'
# awk '{ gsub("/",".drl",$NF); print line }' arun.conf
# awk 'NF{NF--};1' line arun.conf
echo $line | rev | cut -d'/' -f 1 | rev >> output.txt
done < "$file"
Expected Output :
knowledgebase/arun
knowledgebase/arunraj/tester
knowledgebase/arunraj2/arun/test
There's the dirname command to make it easy and reliable:
#!/bin/bash
file=test.conf
while IFS= read -r line
do
dirname "$line"
done < "$file" > output.txt
There are Bash shell parameter expansions that will work OK with the list of names given but won't work reliably for some names:
file=test.conf
while IFS= read -r line
do
echo "${line%/*}"
done < "$file" > output.txt
There's sed to do the job — easily with the given set of names:
sed 's%/[^/]*$%%' test.conf > output.txt
It's harder if you have to deal with names like /plain.file (or plain.file — the same sorts of edge cases that trip up the shell expansion).
You could add Perl, Python, Awk variants to the list of ways of doing the job.
You can get the path like this:
path=${fullpath%/*}
It cuts away the string after the last /
Using awk one liner you can do this:
awk 'BEGIN{FS=OFS="/"} {NF--} 1' test.conf
Output:
knowledgebase/arun
knowledgebase/arunraj/tester
knowledgebase/arunraj2/arun/test
The examplary code below writes hi in a new line at every iteration. Is there a way to prevent this?
#!/bin/bash
while read line; do
var=$(echo $line | cut -d \, -f 2)
echo -n " $var"
done < file.csv > output.txt
Desired output is a concatenation of '$var's at each iteration. The code is run in OS X.
[Resolved]
In most cases of similar problems, klashww's answer would be what you want to try so that I would accept it as the answer. Yet, in my case, such options all failed in fixing the bug. The behavior was due to non-displayed character '^M' at the end of each line, since the file was coming from windows. I relearned that we should make sure to get rid of '^M' before processing it in bash via the line below. After that, the original code works fine.
tr -d '\015' < file > newfile
You might like to try using pure bash:
while IFS=',' read nu1 var nu2; do
echo -n " $var"
done < file.csv > output.txt
nu: "not used"
Use echo "hi\c" instead of echo -n "hi" or printf if avaliable , example printf "hi".
In your example, this should work:
while read line; do
var=$(echo $line | cut -d \, -f 2)
printf " $var"
done < file.csv > output.txt
Or you can use a better tool:
awk -F\, '{printf " "$2}' file.csv > output.txt
If everything fails tr brute force:
echo " $var"| tr -d '\n'