I have begun to combine different commands in the linux terminal. I am wondering why the backslash and semicolon are required for a command such as:
find ./ -name 'blabla' -exec cp {} ./test \;
when a simple cp command is simply:
cp randomfile ./test
without the \;
Are they to clearly indicate the end of a command, or is it simply required in the documentation? What is the underlying principle?
The backslash before the semicolon is used, because ; is one of list operators (or &&, ||) for separating shell commands. In example:
command1; command2
The find utility is using ; or + to terminate the shell commands invoked by -exec.
So to avoid special shell characters from interpretation, they need to be escaped with a backslash to remove any special meaning for the next character read and for line continuation.
Therefore the following example syntax is allowed for find command:
find . -exec echo {} \;
find . -exec echo {} ';'
find . -exec echo {} ";"
find . -exec echo {} \+
find . -exec echo {} +
See also:
Using semicolon (;) vs plus (+) with exec in find
Simple unix command, what is the {} and \; for
from "man find":
All following
arguments to find are taken to be arguments to the command until
an argument consisting of ';' is encountered.
find needs to know when the arguments of exec are terminated. It is natural to terminate a shell command with ; because also the shell uses this character. For the very same reason such a character must be escaped when inserted through the shell.
Related
This question already has answers here:
Why are the backslash and semicolon required with the find command's -exec option?
(2 answers)
Closed 9 years ago.
This is a relatively simple command, so if a duplicate exists and someone could refer me, I'm sorry and I'll delete/close this question.
Man page for find
find . -type f -exec file '{}' \;
Runs 'file' on every file in or below the current directory. Notice that the braces are enclosed in single quote marks to protect them from interpretation
as shell script punctuation. The semicolon is similarly protected by the use of a backslash, though ';' could have been used in that case also.
I do not understand the notation \;. What in the world is that?
In the find command, the action -exec is followed by a command and that command's arguments. Because there can be any number of arguments, find needs some way of knowing when it ends. The semicolon is what tells find that it has reached the end of the command's arguments.
Left to their own devices, most shells would eat the semicolon. We want that semicolon to be passed to the find command. Therefore, we escape it with a backslash. This tells the shell to treat the semicolon as just one of the arguments to the find command.
MORE: Why not, one may ask, just assume that the exec command's argument just go to the end of the line? Why do we need to signal an end to the exec command's arguments at all? The reason is that find has advanced features. Just for example, consider:
find . -name '*.pdf' -exec echo Yes, we have a pdf: {} \; -o -exec echo No, not a pdf: {} \;
I am trying to manipulate the filename from the find command:
find . -name "*.xib" -exec echo '{}' ';'
For example, this might print:
./Views/Help/VCHelp.xib
I would like to make it:
./Views/Help/VCHelp.strings
What I tried:
find . -name "*.xib" -exec echo ${'{}'%.*} ';'
But, the '{}' is not being recognized as a string or something...
I also tried the following:
find . -name "*.xib" -exec filename='{}' ";" -exec echo ${filename%.*} ";"
But it is trying to execute a command called "filename" instead of assigning the variable:
find: filename: No such file or directory
You can't use Parameter Expansion with literal string. Try to store it in a variable first:
find . -name '*.xib' -exec bash -c "f='{}' ; echo \${f%.xib}.strings" \;
-exec sees first argument after it as the command, therefore you can't simply give it filename='{}' because find doesn't use sh to execute what you give it. If you want to run some shell stuff, you need to use sh or bash to wrap up.
Or use sed:
find . -name '*.xib' | sed 's/.xlib$/.strings/'
For such a simple search, you can use a pure bash solution:
#!/bin/bash
shopt -s globstar
shopt -s nullglob
found=( **.xib )
for f in "${found[#]}"; do
echo "${f%xib}strings"
done
Turning the globstar shell option on enables the ** to "match all files and zero or more directories and subdirectories" (as quoted from the bash reference manual). The nullglob option helps if there's no match: in this case, the glob will be expanded to nothing instead of the ugly **.xib. This solution is safe regarding filenames containing funny characters.
find . -name "*.xib" | sed -e 's/\.xib$/.strings/'
I would like to pass the output list of elements of a command as a parameter of another command. I have found some other pages:
How to display the output of a Linux command on stdout and also pipe it to another command?
Use output of bash command (with pipe) as a parameter for another command
but they seem to be more complex.
I just would like to copy a file to every result of a call to the Linux find command.
What is wrong here?:
find . -name myFile 2>&1 | cp /home/myuser/myFile $1
Thanks
This is what you want:
find . -name myFile -exec cp /home/myuser/myFile {} ';'
A breakdown / explanation of this:
find: invoking the find command
.: start search from current working directory.
Since no depth flags are specified, this will search recursively for all subfolders
-name myFile: find files with the explicit name myFile
-exec: for the search results, perform additional commands with them
cp /home/myuser/myFile {}: copies /home/myuser/myFile to overwrite each result returned by find to ; think of {} as where each search result goes.
';': used to separate different commands to be run after find
There are a couple of ways to solve this, depending on whether you need to worry about files with spaces or other special characters in their names.
If none of the filenames have spaces or special characters (they consist only of letters, numbers, dashes, and underscores), then the following is a simple solution that will work. You can use $(command) to execute a command, and substitute the results into the arguments of another command. The shell will split the result on spaces, tabs, or newlines, and for assign each value to $f in turn, and run the command on each value.
for f in $(find . -name myFile)
do
cp something $f
done
If you do have spaces or tabs, you could use find's -exec option. You pass -exec command args, putting {} where you want the filename to be substituted, and ending the arguments with a ;. You need to quote the {} and ; so that the shell doesn't interpret them.
find . -name myFile -exec cp something "{}" \;
Sometimes -exec is not sufficient. For example, in this question, they wanted to use Bash parameter expansion to compute the filename. In order to do that, you need to pass -exec bash -c 'your command', but then you will run into quoting problems with the {} substitution. To solve this, you can use -print0 from find to print the results delimited with null characters (which are invalid in filenames), and pipe it to a while read loop that splits parameters on nulls:
find . -name myFile -print0 | (while read -d $'\0' f; do
cp something "$f"
done)
The pipe will send the output of one program to the input of another. cp does not read from its input stream at the terminal, it merely uses the arguments on the command line.
You want to either use xargs with the pipe or find's exec argument instead of pipes.
find . -name myFile 2>&1 | xargs -I {} cp /home/myuser/myFile {}
Note: option -I {} defines {} as the place holder you could alternatively use someother placeholder if it conflicts with command to be executed.
I use command of find, for example:
find . -name "*.log" -exec grep "running" {} \;
why the command of find need {} , a blank and \?
This is because of the -exec parameter: the {} is a placeholder for the file that will be passed to the command.
the semicolon (;) tells find that the -exec argument list is over, but since ; is also a shell operator, you need to escape it so that it reaches find: \;
-exec works like this: for every file that is found, the first argument after -exec (the command) is executed and all parameters up until the ; are passed as arguments to the command. The {} is then replaced by the current filename that is found by find.
{} is a placeholder for path, which find replaces with the actual found path.
\; terminates find's exec arguments. Without \ shell would treat it as shell statement terminator, hence it needs to be quoted with \ for the shell to pass ; it to find.
I'd say that ; was an unfortunate choice for find exec command terminator.
Note that {} \; sequence can be replaced with {} +:
-exec command {} +
This variant of the -exec action runs the specified command on
the selected files, but the command line is built by appending
each selected file name at the end; the total number of invoca‐
tions of the command will be much less than the number of
matched files. The command line is built in much the same way
that xargs builds its command lines. Only one instance of `{}'
is allowed within the command. The command is executed in the
starting directory.
Just read the manpages
-exec command ;
Execute command; true if 0 status is returned. All following
arguments to find are taken to be arguments to the command until
an argument consisting of `;' is encountered. The string `{}'
is replaced by the current file name being processed everywhere
it occurs in the arguments to the command, not just in arguments
where it is alone, as in some versions of find. Both of these
constructions might need to be escaped (with a `\') or quoted to
protect them from expansion by the shell. See the EXAMPLES sec‐
tion for examples of the use of the -exec option. The specified
command is run once for each matched file. The command is exe‐
cuted in the starting directory. There are unavoidable secu‐
rity problems surrounding use of the -exec action; you should
use the -execdir option instead.
The backslash is an escape to protect the semicolon from being misinterpreted.
The braces are placeholders for the full path outputted by find.
I want to know exactly what {} \; and {} \+ and | xargs ... do. Please clarify these with explanations.
Below 3 commands run and output same result but the first command takes a little time and the format is also little different.
find . -type f -exec file {} \;
find . -type f -exec file {} \+
find . -type f | xargs file
It's because 1st one runs the file command for every file coming from the find command. So, basically it runs as:
file file1.txt
file file2.txt
But latter 2 find with -exec commands run file command once for all files like below:
file file1.txt file2.txt
Then I run the following commands on which first one runs without problem but second one gives error message.
find . -type f -iname '*.cpp' -exec mv {} ./test/ \;
find . -type f -iname '*.cpp' -exec mv {} ./test/ \+ #gives error:find: missing argument to `-exec'
For command with {} \+, it gives me the error message
find: missing argument to `-exec'
why is that? can anyone please explain what am I doing wrong?
The manual page (or the online GNU manual) pretty much explains everything.
find -exec command {} \;
For each result, command {} is executed. All occurences of {} are replaced by the filename. ; is prefixed with a slash to prevent the shell from interpreting it.
find -exec command {} +
Each result is appended to command and executed afterwards. Taking the command length limitations into account, I guess that this command may be executed more times, with the manual page supporting me:
the total number of invocations of the command will be much less than the number of matched files.
Note this quote from the manual page:
The command line is built in much the same way that xargs builds its command lines
That's why no characters are allowed between {} and + except for whitespace. + makes find detect that the arguments should be appended to the command just like xargs.
The solution
Luckily, the GNU implementation of mv can accept the target directory as an argument, with either -t or the longer parameter --target. It's usage will be:
mv -t target file1 file2 ...
Your find command becomes:
find . -type f -iname '*.cpp' -exec mv -t ./test/ {} \+
From the manual page:
-exec command ;
Execute command; true if 0 status is returned. All following arguments to find are taken to be arguments to the command until an argument consisting of `;' is encountered. The string `{}' is replaced by the current file name being processed everywhere it occurs in the arguments to the command, not just in arguments where it is alone, as in some versions of find. Both of these constructions might need to be escaped (with a `\') or quoted to protect them from expansion by the shell. See the EXAMPLES section for examples of the use of the -exec option. The specified command is run once for each matched file. The command is executed in the starting directory. There are unavoidable security problems surrounding use of the -exec action; you should use the -execdir option instead.
-exec command {} +
This variant of the -exec action runs the specified command on the selected files, but the command line is built by appending each selected file name at the end; the total number of invocations of the command will be much less than the number of matched files. The command line is built in much the same way that xargs builds its command lines. Only one instance of `{}' is allowed within the command. The command is executed in the starting directory.
I encountered the same issue on Mac OSX, using a ZSH shell: in this case there is no -t option for mv, so I had to find another solution.
However the following command succeeded:
find .* * -maxdepth 0 -not -path '.git' -not -path '.backup' -exec mv '{}' .backup \;
The secret was to quote the braces. No need for the braces to be at the end of the exec command.
I tested under Ubuntu 14.04 (with BASH and ZSH shells), it works the same.
However, when using the + sign, it seems indeed that it has to be at the end of the exec command.
The standard equivalent of find -iname ... -exec mv -t dest {} + for find implementations that don't support -iname or mv implementations that don't support -t is to use a shell to re-order the arguments:
find . -name '*.[cC][pP][pP]' -type f -exec sh -c '
exec mv "$#" /dest/dir/' sh {} +
By using -name '*.[cC][pP][pP]', we also avoid the reliance on the current locale to decide what's the uppercase version of c or p.
Note that +, contrary to ; is not special in any shell so doesn't need to be quoted (though quoting won't harm, except of course with shells like rc that don't support \ as a quoting operator).
The trailing / in /dest/dir/ is so that mv fails with an error instead of renaming foo.cpp to /dest/dir in the case where only one cpp file was found and /dest/dir didn't exist or wasn't a directory (or symlink to directory).
find . -name "*.mp3" -exec mv --target-directory=/home/d0k/Музика/ {} \+
no, the difference between + and \; should be reversed. + appends the files to the end of the exec command then runs the exec command and \; runs the command for each file.
The problem is find . -type f -iname '*.cpp' -exec mv {} ./test/ \+ should be find . -type f -iname '*.cpp' -exec mv {} ./test/ + no need to escape it or terminate the +
xargs I haven't used in a long time but I think works like +.