what does `{}` mean as a file name in output from egrep? - grep

Question

I am on ubuntu debian 12.04, and I ran a find command to add something to all of my python files:
find . iname "*.py" -exec echo "import os" >> {} \;
The command runs without error and I want to validate the results so I egrep all of the files:
egrep -in "import os" *
And I get results looking like this:
{}:35:import os
{}:36:import os
{}:37:import os
{}:38:import os
{}:39:import os
...and the numbers go until 51 for some reason. What does this mean?
Thank you.

Answer 1

Your first command:
find . iname "*.py" -exec echo "import os" >> {} \;
Is looking for files ending in .py, and for each one is putting the string "import os" in a file called {}. Presumably there are 51 matches.
So egrep, when you run it, the * matches all files, including your file called {}. With {}:35:import os it's telling you that "in the file {}, at line 35, there's the string you're looking for"

Answer 2

This command:
find . iname "*.py" -exec echo "import os" >> {} \;
...creates a file named {} (in bash, and other shells which honor redirections in positions other than head and tail -- this is an extension which the POSIX sh standard does not require). It does not modify the files found by find. (This is because the >> is acting as a command to the shell that's starting find; it's not modifying the behavior of -exec -- and even if it did, -exec directly uses execve() to invoke the command given; it doesn't start that command through a shell, so it doesn't honor shell constructs such as redirections, so you'd be passing literal >> as an argument to echo on any shells not implementing this extension, still not performing a redirection on the individual files found).
Now, if you did want to modify the files found by find, you might do so like this:
find . -iname '*.py' -exec sh -c 'for f; do echo "import os" >>"$f"; done' {} +
Noteworthy differences:
The redirection is invoked inside a shell started with exec sh; thus, there's a shell present to honor it after the individual filenames have been resolved.
-exec ... {} + is used, which is much more efficient than -exec ... {} ; (the former runs as few subcommands as possible; the latter runs one per file found).

Answer 3

{} is a placeholder that is replaced by find with the filename that matches the given condition, in this case {} is replaced with filename that match the pattern "*.py".
However your find command isn't actually doing that, as the >> {} is not actually part of the -exec block, but interpreted by the shell as a redirect for the whole find command, so the {} never gets replaced by find with the proper filename and instead you are redirecting into a file called {}. To make things more clear, the command you are actually executing is this:
find . iname "*.py" -exec echo "import os" \; >> {}
Meaning for every *.py file you add a line containing "import os" into a file called {}. The output of grep is just filename:linenumber:matched_line so you get a {} in there as that is the filename.
If you are wondering how the \; survives and why you are not getting a:
find: missing argument to `-exec'
The shell doesn't actually care where in the command line the redirect occurs:
echo 1 2 3 4 5 6 7 > foo
is the same as:
echo 1 2 > foo 3 4 5 6 7
and gives you this each time:
$ cat foo
1 2 3 4 5 6 7
Also worth to mention >> is an append operator, so even if you fix your command you are adding to the end of the Python files, while import os probably should go to the top of the file.