We are using Oracle Apps where we also develop programs using UNIX. In our Oracle Apps UNIX program, while running a program, we will pass parameters and the parameter values will be stored in $1 value. And this $1 will also contain other information like instance, what is program id, program ran user name and user id etc., as below
XX_VENDOR_ACH_HOST FCP_REQID=9946271 FCP_LOGIN="APPS/trn12fnd" FCP_USERID=127 FCP_USERNAME="USERNAME" FCP_PRINTER="noprint" FCP_SAVE_OUT=Y FCP_NUM_COPIES=0 "TEST Pay Batch 021120-c"
here, TEST Pay Batch 021120-c is the only parameter value. Now, I need to fetch this parameter value. So, I wrote the command as below
echo $1 | cut -d" " -f9 | sed 's/"//g'
But, it is just printing the word TEST not the complete value. Can anyone please help me with the complete command to fetch TEST Pay Batch 021120-c
Regards,
Srivathsava
Assuming;
The parameter value is enclosed with double-quotes
The parameter value is located at the end of $1
then how about;
echo "$1" | sed 's/.*"\([^"]\+\)"$/\1/'
which outputs:
TEST Pay Batch 021120-c
If my assumptions are incorrect and if the parameter value string has
some distinct characteristics (e.g. it starts with TEST Pay Batch),
please let me know. Then I'll tweak the regex to extract the parameter value.
Assuming the parameter value is located at the end of $1 something like can do the work:
echo $1|awk -F\" '{print $(NF-1)}'
Related
I have a rename script below rename.sh. I want to introduce a variable such that I can pass a date argument when executing the script like./rename.sh 20151103 such that 20151103 replaces 20140306 in the script.
for f in *.CDR*; do
echo mv "$f" "${f/-20140306/-0-20140306}"
done
Thinking of automating this as I don't want to manually edit the script each time i'm doing a rename. Any other method will be highly welcomed.
#!/bin/bash
pattern="$1"
for f in *.CDR*; do
echo mv "$f" "${f/-${pattern}/-0-${pattern}}"
done
Explanation:
The #!-line says we're running this as a bash script.
The script will populate the variables $1, $2 etc. with the arguments handed to it on the command line. These are called the positional parameters ($0 usually holds the name of the script).
We take $1, because we know that should contain the pattern we're replacing, and assign it to the variable $pattern. In much more complex scripts, here is where we would handle command line switches (with getopts, but that's an answer for another day).
We quote $1, just because. (It's good practice to quote user input, just to be sure no shell-globbing characters, such as * gets expanded).
The rest is the script like you had from before, but with the string 20140306 replaced by ${pattern}. I'm using ${pattern} rather than $pattern here for readability only. In general, you need to use ${a} rather than $a if you, for example, interpolate a string like "${a}nospaceafter".
Then it should just be a matter of making the script executable before testing it:
$ chmod +x rename.sh
This is the one of the method you can consider:
#!/bin/bash
input=$1
for f in *.CDR*; do
echo mv "$f" "${f/-$input/-0-$input}"
done
What is the right way to assign commands into variable . a.k.a how should i use quote commas and how assign it to variable , so the command is up to date? For example
i am at home/desktop
i assign variable
b=`pwd`
echo "$b" // home/desktop
but when i change the directory
cd games
pwd // home/desktop/games
echo "$b" // home/desktop
it does not update. Also i want to do something like
g="-l"
ls $g
is it possible?
Variables never update automatically, you have to re-assign them, i.e.,
b=`pwd`
echo "$b" // home/desktop
cd games
pwd // home/desktop/games
b=`pwd`
echo "$b" // home/desktop/games
as for your second question, I think that's only possible using eval:
g=" -la"
eval "ls"$g
BE VERY CAREFUL WITH THIS It's very easy to write something that can compromise security and indeed most people (including myself!) would strongly advise you never to use eval unless all else has failed.
Consider the code above for demonstration purposes only, under no circumstances to be used in a production system.
I think you are confusing assigning variables with aliases.
assigning a variable means you store the result of the command in a variable
what this command does
b=`pwd`
is running pwd and stores the answer in a variable b.
and alias means giving some command a different name. so running alias b pwd will make it so whenever you run b, you will actually run pwd
b is only set when you run the first command (assigning the output of pwd to b, it doesn't link those items together (so that b gets set on every invocation). There are ways to do that also (do man bash on your machine) (for example, the shell variable $PWD contains this information.
to your second question, yes, you can do that, but a nice way to so it is to use alias. So, alias ll='ls -l' creates a command on your system that will run ls -l when you type ll
I have a string in different ranges :
WATSON_AJAY_AB04_DOTHING.data
WATSON_NAVNEET_CK4_DOTHING.data
WATSON_PRASHANTH_KJ56_DOTHING.data
WATSON_ABHINAV_KD323_DOTHING.data
On these above string how can I extract
AB04,CK4,KJ56,KD323
in Unix?
echo "$string" | cut -d'_' -f3
You could use sed or grep for this task. But since the string is so simple, I dont think you will need to.
One method is to use the bash 'cut' command. Below is an example directly on the BASH shell/command line:
jimm#pi$ string='WATSON_AJAY_AB04_DOTHING.data'
jimm#pi$ cut -d '_' -f 3 <<< "$string"
AB04 <-- outputs the result directly
(edit: of course Lucas' answer above is also a quick 'one-liner' that does the same thing as above - he beat me to it) :)
The cut will take an _ character as the delimiter (the -d '_' part), then display the 3rd slice of the string (the -f 3 part).
Or, if you want to output that 3rd slice from a list of content (using your list above), you can write a simple BASH script.
First, save the lines above ('WATSON...etc') into something like text.txt. Then open up your favorite text editor and type:
#!/bin/sh
cut -d '_' -f 3 < $1
Save that script to some useful name like slice.sh, and make sure it is executable with something like chmod 775 slice.sh.
Then at the command line you can execute the script against your text file, and immediately get an output of those parts of the file you want (in this case the third set of text, separated by the _ character):
$ ./slice.sh text.txt
AB04
CK4
KJ56
KD323
Hope that helps! Bear in mind that the commands above may vary a bit, depending on the flavor of *nix you are using, but it should at least point you in the right direction.
How can I retrieve control-m JOBNAME in unix shell (ksh) script
From what I read %%JOBNAME should give me the JOBNAME but unix does not support %%JOBNAME. I tried it but not successful
Also I tried using $JOBNAME but it did not work either
First you will have to get your local %%JOBNAME variable into a global variable, you can do that pretty easily from within the job form.
Once you have a global variable, there is a ctmvar utility included with the CTM Server/Agents.
You can use that to read in Control-M global auto edit variables into script-able shell environments like bash as follows:
UNIXVAR=$(ctmvar -action list | grep %%CTMGLOBALVARIABLENAME | awk '{print $2}')
To do the reverse and set a CTM global variable from a unix shell variable do this:
ctmvar -action set -var "%%\CTMGLOBALVARIABLE" -varexpr "$UNIXVAR"
I do this all the time in shell scripts and it works great. The one caveat is that once you set a global variable is it visible to all agents and all servers by the same variable name, so be careful you use unique variable names so you don't step on your own toes, and also that you clean up after yourself or you will have a ton of global variables left lying about.
the easiest way is to define a parameter in the job that submits the unix task. Name the parameter PARM1 (or PARM2) and give it the value %%JOBNAME. in the unix script, you'll access the value with $1 (or $2).
For example,
I executed "pwd" and it shows the current working directory. Then if I want to reuse that result in my another command, it would convenient to get it via a Unix command or built-in variable. Does it exist?
You can get the result, as in return code, using $?. In order to get the output you'll need to explicitly keep it around - e.g. with:
MYVAR=`pwd`
echo $MYVAR
Use $? inorder to get the status of the last executed command. Its value will be zero if the last executed command was successful else non zero.
The internal variable $? holds the return value of the last executed command or program. Example: http://tldp.org/LDP/abs/html/complexfunct.html#MAX.
If you don't need to run one command first, you can also try using pipes | to connect commands. I am constantly piping long directory listings over to more, so I can page through the results, with
ls -al | more
so if you want to use the results of running pwd as input to another program, you can try something like piping the results of pwd over to more with
pwd|more