Using Output of Previous Commands in Bash

Using output of previous commands in bash

Since the amount of output is indeterminate, it doesn't make sense for bash to store it for you for re-display. But there's an alternate solution to your problem:

The tee command allows you to duplicate an output stream to a file. So if you're willing to use a file for temporary storage, you can do something like this:

make | tee output.txt
grep "warning" output.txt

This solution avoids running make twice, which could be (a) expensive and (b) inconsistent: the second make may be doing less work than the first because some targets were already made the first time around.

Note: I haven't tried this. You may need to fiddle with joining the error and output streams, or such.

Automatically capture output of last command into a variable using Bash?

This is a really hacky solution, but it seems to mostly work some of the time. During testing, I noted it sometimes didn't work very well when getting a ^C on the command line, though I did tweak it a bit to behave a bit better.

This hack is an interactive mode hack only, and I am pretty confident that I would not recommend it to anyone. Background commands are likely to cause even less defined behavior than normal. The other answers are a better way of programmatically getting at results.

---

That being said, here is the "solution":

PROMPT_COMMAND='LAST="`cat /tmp/x`"; exec >/dev/tty; exec > >(tee /tmp/x)'

Set this bash environmental variable and issues commands as desired. $LAST will usually have the output you are looking for:

startide seth> fortune
Courtship to marriage, as a very witty prologue to a very dull play.
                -- William Congreve
startide seth> echo "$LAST"
Courtship to marriage, as a very witty prologue to a very dull play.
                -- William Congreve

How to read the output of previous command and then proceed to the next command

try this:

#!/bin/bash
Commands=(
  "command1 parameter1 parameter2"
  "command2 parameter1"
  "command3 parameter1 parameter2 parameter3"
)

tmplog=$(mktemp)
for cmd in "${Commands[@]}"; do
   echo "$cmd"
   $cmd >"$tmplog"
   tail -1 "$tmplog" | grep -v "^Successfully" && break
done
rm "$tmplog"

Execute the output of previous command line

This should do it I believe.

my-first-command | bash

In a bash pipe, take the output of the previous command as a variable to the next command (Eg. if statement)

Following your narrow request looks like:

sha256sum abc.txt  |
  awk '{print $1}' |
  if [ "$(cat)" = "8237491082roieuwr0r9812734iur" ]; then echo "match"; fi

...as cat with no arguments reads the command's stdin, and in a pipeline, content generated from prior stages are streamed into their successors.

Alternately:

sha256sum abc.txt  |
  awk '{print $1}' |
  if read -r line && [ "$line" = "8237491082roieuwr0r9812734iur" ]; then echo "match"; fi

...wherein we read only a single line from stdin instead of using cat. (To instead loop over all lines given on stdin, see BashFAQ #1).

---

However, I would strongly suggest writing this instead as:

if [ "$(sha256sum abc.txt | awk '{print $1}')" = "8237491082roieuwr0r9812734iur" ]; then
  echo "match"
fi

...which, among other things, keeps your logic outside the pipeline, so your if statement can set variables that remain set after the pipeline exits. See BashFAQ #24 for more details on the problems inherent in running code in pipelines.

---

Related Topics