sed command failing in script with arguments

Question

I am trying to perform a two step string editing:

1. Match for the string "payload". If found, then add some "#if #endif" code as shown in output file.
2. Then copy the line containing "payload", replace it with "upper_payload" and then paste it within the "if endif" conditions.

Input File:

Let's see if this works or not payload.
alskdjaslkdjn.
aslkdjbnalkgfjbaglbjbjdgfsbkgjbsg;
Also here is another_payload.
alskdjbnalgkjb.
kjhsdfjlkgfjlknbsgdfljkbgsbljk((*&)(&;

Expected Output File:

Let's see if this works or not payload.
#if defined(NV_C2C_UPPER_PD)
Let's see if this works or not upper_payload.
#endif
alskdjaslkdjn.
aslkdjbnalkgfjbaglbjbjdgfsbkgjbsg;
Also here is another_payload.
#if defined(NV_C2C_UPPER_PD)
Also here is another_upper_payload.
#endif
alskdjbnalgkjb.
kjhsdfjlkgfjlknbsgdfljkbgsbljk((*&)(&;

I finally came up with the following code in a shell file so that I can run this on multiple text files:

#!/bin/csh -f
sed -i '/payload/a #if defined(NV_C2C_UPPER_PD)\n#endif' $1
sed -i "/NV_C2C_UPPER_PD/a `sed -n '/payload/s/payload/upper_payload/p' $1`" $1

With the above shell script when I pass a command such as:

> ./my_shell.sh test.txt

the text file is modified to the following output:

Let's see if this works or not payload.
#if defined(NV_C2C_UPPER_PD)
Let's see if this works or not upper_payload.
#endif
alskdjaslkdjn.
aslkdjbnalkgfjbaglbjbjdgfsbkgjbsg;
Also here is another_payload.
#if defined(NV_C2C_UPPER_PD)
Let's see if this works or not upper_payload.
#endif
alskdjbnalgkjb.
kjhsdfjlkgfjlknbsgdfljkbgsbljk((*&)(&;

And it also gives this error/warning on the terminal:

sed: can't read Also here is another_upper_payload.: No such file or directory

In the output we can observe that the second "upper_payload" substitution was incorrect because the line copying is incorrect. It copied the first line instead of the newer line containing "another_payload".

I have tried different options but I am failing in solving this problem. Would appreciate some help.

Answer 1

Instead of trying to invoke another sed process to create input for one of your sed expressions, consider using the s/// command to reorganize the current line as soon as you find the trigger substring (payload).

/payload/!b

s/.*/&\
#if defined(NV_C2C_UPPER_PD)\
&\
#endif/

s/payload/upper_&/2

The above sed editing script starts by skipping any line containing our trigger string. The b ("branch") command skips to the very end of the script when used without a label.

On the remaining lines, we replace the entire contents with itself, followed by the #if line, the original line a second time, and the #endif line. One may insert newlines using the s/// command if they are escaped as shown above (GNU sed allows for inserting literal newlines with \n too).

The second occurrence of the string payload is then prepended by the string upper_ in a separate substitution.

This solution assumes that the trigger string does not occur multiple times in a single line, or the last substitution would not replace the correct substring.

On the command line,

sed -e '/payload/!b' \
    -e 's/.*/&\
#if defined(NV_C2C_UPPER_PD)\
&\
#endif/' \
    -e 's/payload/upper_&/2' file

or, using a separate script file for the script,

sed -f script file

or, using a here-document for the script,

sed -f /dev/stdin file <<'SED_SCRIPT'
/payload/!b

s/.*/&\
#if defined(NV_C2C_UPPER_PD)\
&\
#endif/

s/payload/upper_&/2
SED_SCRIPT

or, using GNU sed,

sed -e '/payload/!b' \
    -e 's/.*/&\n#if defined(NV_C2C_UPPER_PD)\n&\n#endif/' \
    -e 's/payload/upper_&/2' file

Result:

Let's see if this works or not payload.
#if defined(NV_C2C_UPPER_PD)
Let's see if this works or not upper_payload.
#endif
alskdjaslkdjn.
aslkdjbnalkgfjbaglbjbjdgfsbkgjbsg;
Also here is another_payload.
#if defined(NV_C2C_UPPER_PD)
Also here is another_upper_payload.
#endif
alskdjbnalgkjb.
kjhsdfjlkgfjlknbsgdfljkbgsbljk((*&)(&;

Answer 2

You can try this sed

$ sed -i 's/\(.*\)\(payload\)\(.*\)/&\n#if defined(NV_C2C_UPPER_PD)\n\1upper_\2\3\n#endif/ ' input_file

\(.*\)\(payload\) Here, the match is split into 2 groups so we can easily back reference them on the return

& will return the full match unaltered

\1upper_\2 This is where we return the grouping in the order we require, the first group matches all the way to payload which allows upper_ to be inserted before payload

Output

$ sed 's/\(.*\)\(payload\)\(.*\)/&\n#if defined(NV_C2C_UPPER_PD)\n\1upper_\2\3\n#endif/' input_file
Let's see if this works or not payload
#if defined(NV_C2C_UPPER_PD)
Let's see if this works or not upper_payload
#endif.
alskdjaslkdjn.
aslkdjbnalkgfjbaglbjbjdgfsbkgjbsg;
Also here is another_payload
#if defined(NV_C2C_UPPER_PD)
Also here is another_upper_payload
#endif.
alskdjbnalgkjb.
kjhsdfjlkgfjlknbsgdfljkbgsbljk((*&)(&;

Answer 3

Great sed solutions here already, but I'd personally prefer awk for this:

awk \
-v if='#if defined(NV_C2C_UPPER_PD)' \
-v endif='#endif' \
'
/payload/{
  line=$0
  gsub("payload","upper_payload",$0);
  printf "%s\n%s\n%s\n%s\n",$line,if,$0,endif
  next
}
1' file

Answer 4

One way using POSIXly only constructs:

sed -e 'p
  /payload/!d
  s//upper_&/
  i\
#if defined(NV_C2C_UPPER_PD)
  a\
#endif
' file

---

If you don't want escaped newlines, you can do as follows while still being POSIXly compliant:

sed -e 'p
  /payload/!d
  s//upper_&/
  H;s/.*//;x;G
  s/^/#if defined(NV_C2C_UPPER_PD)/
  s/$/#endif/
' file