I have this command
cp $(ldd MyApp.out | awk '{print $3}' | sed -E '/^$/d') lib/
and at some point, I want to echo it into a file but a naive approach echo command_above doesn't work.
If I put the command into single quotes, then $3 expands to whitespace.
Is it possible to print that command char-by-char as it is after echo command without any expansion and substitution?
The common approach is to use the << operator to read until some delimiter:
# "cat" just prints what it reads
cat << 'EOF' > output_file
cp $(ldd MyApp.out | awk '{print $3}' | sed -E '/^$/d') lib/
EOF
Use xargs to pass file names list to cp
ldd MyApp.out | awk '$3!=""{print $3}' | xargs -d'\n' -I{} cp {} lib/
For debugging and logging purposes you can use set -x or set -v:
set -v # dump commands below
cp $(ldd MyApp.out | awk '{print $3}' | sed -E '/^$/d') lib/
set +v # stop dumping
Related
I was trying to filter all the files from the URLs and get only paths.
echo -e "http://sub.domain.tld/secured/database_connect.php\nhttp://sub.domain.tld/section/files/image.jpg\nhttp://sub.domain.tld/.git/audio-files/top-secret/audio.mp3" | grep -Ei "(http|https)://[^/\"]+" | sort -u
http://sub.domain.tld
But I want the result like this
http://sub.domain.tld/secured/
http://sub.domain.tld/section/files/
http://sub.domain.tld/.git/audio-files/top-secret/
Is there any way to do it with sed or grep
Using grep
$ echo ... | grep -o '.*/'
http://sub.domain.tld/secured/
http://sub.domain.tld/section/files/
http://sub.domain.tld/.git/audio-files/top-secret/
with grep
If your grep has the -o option:
... | grep -Eio 'https?://.*/'
If there could be multiple URLs per line:
... | grep -Eio 'https?://[^[:space:]]+/'
with sed
If the input is always precisely one URL per line and nothing else, you can just delete the filename part:
... | sed 's/[^/]*$//'
You could use match function of awk, will work in any version of awk. Simple explanation would be, passing echo command's output to awk program. Using match matching everything till last occurrence of / and then printing the sub-string to print just before /(with -1 to RLENGTH).
your_echo_command | awk 'match($0,/.*\//){print substr($0,RSTART,RLENGTH-1)}'
GNU Awk
$ echo ... | awk 'match($0,/.*\//,a){print a[0]}'
$ echo ... | awk '{print gensub(/(.*\/).*/,"\\1",1)}'
$ echo ... | awk 'sub(/[^/]*$/,"")'
http://sub.domain.tld/secured/
http://sub.domain.tld/section/files/
http://sub.domain.tld/.git/audio-files/top-secret/
xargs
$ echo ... | xargs -i sh -c 'echo $(dirname "{}")/'
http://sub.domain.tld/secured/
http://sub.domain.tld/section/files/
http://sub.domain.tld/.git/audio-files/top-secret/
I have tried the below
history -d $(history | grep "echo.*" |awk '{print $1}')
But it is not deleting all the commands from the history with echo
I want to delete any commands start with echo
like
echo "mamam"
echoaaa
echo "hello"
echooooo
You can use this to remove echo entries :
for d in $(history | grep -Po '^\s*\K(\d+)(?= +echo)' | sort -nr); do history -d $d; done
I would do a
history -d $(history | grep -E "^ *[0-9]+ *echo" | awk '{print $1})
The history command produces one column of event number, followed by the command. We need to match an echo, which is following such a event number. The awk then prints just the event number.
An alternative without reverting to awk would be:
history -d $(history | grep -E "^ *[0-9]+ *echo" | grep -Eow '[0-9]+)
history -w
sed -i '/^echo.*/d' ~/.bash_history
history -c
history -r
I have an application (myapp) that gives me a multiline output
result:
abc|myparam1|def
ghi|myparam2|jkl
mno|myparam3|pqr
stu|myparam4|vwx
With grep and sed I can get my parameters as below
myapp | grep '|' | sed -e 's/^[^|]*//' | sed -e 's/|.*//'
But then want these myparamx values as paramaters of a script to be executed for each parameter.
myscript.sh myparam1
myscript.sh myparam2
etc.
Any help greatly appreciated
Please see xargs. For example:
myapp | grep '|' | sed -e 's/^[^|]*//' | sed -e 's/|.*//' | xargs -n 1 myscript.sh
May be this can help -
myapp | awk -F"|" '{ print $2 }' | while read -r line; do /path/to/script/ "$line"; done
I like the xargs -n 1 solution from Dark Falcon, and while read is a classical tool for such kind of things, but just for completeness:
myapp | awk -F'|' '{print "myscript.sh", $2}' | bash
As a side note, speaking about extraction of 2nd field, you could use cut:
myapp | cut -d'|' -f 1 # -f 1 => second field, starting from 0
I have a variable which has the directory path, along with the file name. I want to extract the filename alone from the Unix directory path and store it in a variable.
fspec="/exp/home1/abc.txt"
Use the basename command to extract the filename from the path:
[/tmp]$ export fspec=/exp/home1/abc.txt
[/tmp]$ fname=`basename $fspec`
[/tmp]$ echo $fname
abc.txt
bash to get file name
fspec="/exp/home1/abc.txt"
filename="${fspec##*/}" # get filename
dirname="${fspec%/*}" # get directory/path name
other ways
awk
$ echo $fspec | awk -F"/" '{print $NF}'
abc.txt
sed
$ echo $fspec | sed 's/.*\///'
abc.txt
using IFS
$ IFS="/"
$ set -- $fspec
$ eval echo \${${##}}
abc.txt
You can simply do:
base=$(basename "$fspec")
dirname "/usr/home/theconjuring/music/song.mp3"
will yield
/usr/home/theconjuring/music.
bash:
fspec="/exp/home1/abc.txt"
fname="${fspec##*/}"
echo $fspec | tr "/" "\n"|tail -1
Using bash "here string":
$ fspec="/exp/home1/abc.txt"
$ tr "/" "\n" <<< $fspec | tail -1
abc.txt
$ filename=$(tr "/" "\n" <<< $fspec | tail -1)
$ echo $filename
abc.txt
The benefit of the "here string" is that it avoids the need/overhead of running an echo command. In other words, the "here string" is internal to the shell. That is:
$ tr <<< $fspec
as opposed to:
$ echo $fspec | tr
I run this bash command to display contents of somefile.cf in a Weblogic domain directory.
find $(/usr/ucb/ps auwwx | grep weblogic | tr ' ' '\n' | grep security.policy | grep domain | awk -F'=' '{print $2}' | sed -e 's/weblogic.policy//' -e 's/security\///' -e 's/dep\///' | awk -F'/' '{print "/"$2"/"$3"/"$4"/somefile.cf"}' | sort | uniq) 2> /dev/null -exec ls {} \; -exec cat {} \;
I tried incorporating this in an expect script and also escaped some special characters and double quotes too but it throws an error "extra characters after close-quote"
send "echo ; echo 'Weblogic somefile.cf:' ; find \$(/usr/ucb/ps auwwx | grep weblogic | tr ' ' '\n' | grep security.policy | grep domain | awk -F'=' '{print \$2}' | sed -e 's/weblogic.policy//' -e 's/security\\///' -e 's/dep\\///' | awk -F'/' '{print \"/\"\$2\"/\"\$3\"/\"\$4\"/somefile.cf\"}' | sort | uniq) 2> /dev/null -exec ls {} \\; -exec cat {} \\;
I guess it needs some more escaping of special characters or probably I dint escape the existing ones correctly.
Any help would be appreciated.
give us the syntax error find or bash threw on the other side.
and try adding an extra \ or 2 before the semicolons at the end.
The problem with expect is the number of layers of escapes you need when it get's ugly.
In the awk statement, go escape all the doublequotes ( " -> \" )
and get me an error message :)
If you have a command line with complex quoting that you know works in bash then it's often easier to just go ahead and use bash. Like this:
set cmd {find $(/usr/ucb/ps auwwx | grep weblogic | tr ' ' '\n' | grep security.policy | grep domain | awk -F'=' '{print $2}' | sed -e 's/weblogic.policy//' -e 's/security\///' -e 's/dep\///' | awk -F'/' '{print "/"$2"/"$3"/"$4"/somefile.cf"}' | sort | uniq) 2> /dev/null -exec ls {} \; -exec cat {} \;}
spawn /bin/bash -c $cmd
expect ... whatever is appropriate ...
Notice that I used the Tcl {} operator instead of "" around the command string. This operator is like single quote in bash, it means "literal string, do not interpret the contents in any way" and is appropriate here because I want to pass this string verbatim to the spawned bash subprocess.
There is a " missing at the end of your send line.