How do you normalize a file path in Bash?

2019-01-06 08:36发布

站内文章 / Linux

13 0

我命由我不由天

女 | 书童

私信

可以将文章内容翻译成中文,广告屏蔽插件可能会导致该功能失效(如失效，请关闭广告屏蔽插件后再试):

问题:

I want to transform /foo/bar/.. to /foo

Is there a bash command which does this?

Edit: in my practical case, the directory does exist.

回答1:

if you're wanting to chomp part of a filename from the path, "dirname" and "basename" are your friends, and "realpath" is handy too.

dirname /foo/bar/baz 
# /foo/bar 
basename /foo/bar/baz
# baz
dirname $( dirname  /foo/bar/baz  ) 
# /foo 
realpath ../foo
# ../foo: No such file or directory
realpath /tmp/../tmp/../tmp
# /tmp

realpath alternatives

If realpath is not supported by your shell, you can try

readlink -f /path/here/..

Also

readlink -m /path/there/../../

Works the same as

realpath -s /path/here/../../

in that the path doesn't need to exist to be normalized.

回答2:

I don't know if there is a direct bash command to do this, but I usually do

normalDir="`cd "${dirToNormalize}";pwd`"
echo "${normalDir}"

and it works well.

回答3:

Try realpath. Below is the source in its entirety, hereby donated to the public domain.

// realpath.c: display the absolute path to a file or directory.
// Adam Liss, August, 2007
// This program is provided "as-is" to the public domain, without express or
// implied warranty, for any non-profit use, provided this notice is maintained.

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <libgen.h>   
#include <limits.h>

static char *s_pMyName;
void usage(void);

int main(int argc, char *argv[])
{
    char
        sPath[PATH_MAX];


    s_pMyName = strdup(basename(argv[0]));

    if (argc < 2)
        usage();

    printf("%s\n", realpath(argv[1], sPath));
    return 0;
}    

void usage(void)
{
    fprintf(stderr, "usage: %s PATH\n", s_pMyName);
    exit(1);
}

回答4:

Use the readlink utility from the coreutils package.

MY_PATH=$(readlink -f "$0")

回答5:

A portable and reliable solution is to use python, which is preinstalled pretty much everywhere (including Darwin). You have two options:

abspath returns an absolute path but does not resolve symlinks:

python -c "import os,sys; print os.path.abspath(sys.argv[1])" path/to/file
realpath returns an absolute path and in doing so resolves symlinks, generating a canonical path:

python -c "import os,sys; print os.path.realpath(sys.argv[1])" path/to/file

In each case, path/to/file can be either a relative or absolute path.

回答6:

readlink is the bash standard for obtaining the absolute path. It also has the advantage of returning empty strings if paths or a path doesn't exist (given the flags to do so).

To get the absolute path to a directory that may or may not exist, but who's parents do exist, use:

abspath=$(readlink -f $path)

To get the absolute path to a directory that must exist along with all parents:

abspath=$(readlink -e $path)

To canonicalise the given path and follow symlinks if they happen to exist, but otherwise ignore missing directories and just return the path anyway, it's:

abspath=$(readlink -m $path)

The only downside is that readlink will follow links. If you do not want to follow links, you can use this alternative convention:

abspath=$(cd ${path%/*} && echo $PWD/${path##*/})

That will chdir to the directory part of $path and print the current directory along with the file part of $path. If it fails to chdir, you get an empty string and an error on stderr.

回答7:

My recent solution was:

pushd foo/bar/..
dir=`pwd`
popd

Based on the answer of Tim Whitcomb.

回答8:

As Adam Liss noted realpath is not bundled with every distribution. Which is a shame, because it is the best solution. The provided source code is great, and I will probably start using it now. Here is what I have been using until now, which I share here just for completeness:

get_abs_path() {
     local PARENT_DIR=$(dirname "$1")
     cd "$PARENT_DIR"
     local ABS_PATH="$(pwd)"/"$(basename "$1")"
     cd - >/dev/null
     echo "$ABS_PATH"
}

If you want it to resolve symlinks, just replace pwd with pwd -P.

回答9:

Old question, but there is much simpler way if you are dealing with full path names at the shell level:

   abspath="$( cd "$path" && pwd )"

As the cd happens in a subshell it does not impact the main script.

Two variations, supposing your shell built-in commands accept -L and -P, are:

   abspath="$( cd -P "$path" && pwd -P )"    #physical path with resolved symlinks
   abspath="$( cd -L "$path" && pwd -L )"    #logical path preserving symlinks

Personally, I rarely need this later approach unless I'm fascinated with symbolic links for some reason.

FYI: variation on obtaining the starting directory of a script which works even if the script changes it's current directory later on.

name0="$(basename "$0")";                  #base name of script
dir0="$( cd "$( dirname "$0" )" && pwd )"; #absolute starting dir

The use of CD assures you always have the absolute directory, even if the script is run by commands such as ./script.sh which, without the cd/pwd, often gives just .. Useless if the script does a cd later on.

回答10:

Talkative, and a bit late answer. I need to write one since I'm stuck on older RHEL4/5. I handles absolute and relative links, and simplifies //, /./ and somedir/../ entries.

test -x /usr/bin/readlink || readlink () {
        echo $(/bin/ls -l $1 | /bin/cut -d'>' -f 2)
    }


test -x /usr/bin/realpath || realpath () {
    local PATH=/bin:/usr/bin
    local inputpath=$1
    local changemade=1
    while [ $changemade -ne 0 ]
    do
        changemade=0
        local realpath=""
        local token=
        for token in ${inputpath//\// }
        do 
            case $token in
            ""|".") # noop
                ;;
            "..") # up one directory
                changemade=1
                realpath=$(dirname $realpath)
                ;;
            *)
                if [ -h $realpath/$token ] 
                then
                    changemade=1
                    target=`readlink $realpath/$token`
                    if [ "${target:0:1}" = '/' ]
                    then
                        realpath=$target
                    else
                        realpath="$realpath/$target"
                    fi
                else
                    realpath="$realpath/$token"
                fi
                ;;
            esac
        done
        inputpath=$realpath
    done
    echo $realpath
}

mkdir -p /tmp/bar
(cd /tmp ; ln -s /tmp/bar foo; ln -s ../.././usr /tmp/bar/link2usr)
echo `realpath /tmp/foo`

回答11:

Not exactly an answer but perhaps a follow-up question (original question was not explicit):

readlink is fine if you actually want to follow symlinks. But there is also a use case for merely normalizing ./ and ../ and // sequences, which can be done purely syntactically, without canonicalizing symlinks. readlink is no good for this, and neither is realpath.

for f in $paths; do (cd $f; pwd); done

works for existing paths, but breaks for others.

A sed script would seem to be a good bet, except that you cannot iteratively replace sequences (/foo/bar/baz/../.. -> /foo/bar/.. -> /foo) without using something like Perl, which is not safe to assume on all systems, or using some ugly loop to compare the output of sed to its input.

FWIW, a one-liner using Java (JDK 6+):

jrunscript -e 'for (var i = 0; i < arguments.length; i++) {println(new java.io.File(new java.io.File(arguments[i]).toURI().normalize()))}' $paths

回答12:

I'm late to the party, but this is the solution I've crafted after reading a bunch of threads like this:

resolve_dir() {
        (builtin cd `dirname "${1/#~/$HOME}"`'/'`basename "${1/#~/$HOME}"` 2>/dev/null; if [ $? -eq 0 ]; then pwd; fi)
}

This will resolve the absolute path of $1, play nice with ~, keep symlinks in the path where they are, and it won't mess with your directory stack. It returns the full path or nothing if it doesn't exist. It expects $1 to be a directory and will probably fail if it's not, but that's an easy check to do yourself.

回答13:

Try our new Bash library product realpath-lib that we have placed on GitHub for free and unencumbered use. It's thoroughly documented and makes a great learning tool.

It resolves local, relative and absolute paths and doesn't have any dependencies except Bash 4+; so it should work just about anywhere. It's free, clean, simple and instructive.

You can do:

get_realpath <absolute|relative|symlink|local file path>

This function is the core of the library:

function get_realpath() {

if [[ -f "$1" ]]
then 
    # file *must* exist
    if cd "$(echo "${1%/*}")" &>/dev/null
    then 
        # file *may* not be local
        # exception is ./file.ext
        # try 'cd .; cd -;' *works!*
        local tmppwd="$PWD"
        cd - &>/dev/null
    else 
        # file *must* be local
        local tmppwd="$PWD"
    fi
else 
    # file *cannot* exist
    return 1 # failure
fi

# reassemble realpath
echo "$tmppwd"/"${1##*/}"
return 0 # success

}

It also contains functions to get_dirname, get_filename, get_ stemname and validate_path. Try it across platforms, and help to improve it.

回答14:

Based on @Andre's answer, I might have a slightly better version, in case someone is after a loop-free, completely string-manipulation based solution. It is also useful for those who don't want to dereference any symlinks, which is the downside of using realpath or readlink -f.

It works on bash versions 3.2.25 and higher.

shopt -s extglob

normalise_path() {
    local path="$1"
    # get rid of /../ example: /one/../two to /two
    path="${path//\/*([!\/])\/\.\./}"
    # get rid of /./ and //* example: /one/.///two to /one/two
    path="${path//@(\/\.\/|\/+(\/))//}"
    # remove the last '/.'
    echo "${path%%/.}"
}

$ normalise_path /home/codemedic/../codemedic////.config
/home/codemedic/.config

回答15:

The problem with realpath is that it is not available on BSD (or OSX for that matter). Here is a simple recipe extracted from a rather old (2009) article from Linux Journal, that is quite portable:

function normpath() {
  # Remove all /./ sequences.
  local path=${1//\/.\//\/}

  # Remove dir/.. sequences.
  while [[ $path =~ ([^/][^/]*/\.\./) ]]; do
    path=${path/${BASH_REMATCH[0]}/}
  done
  echo $path
}

Notice this variant also does not require the path to exist.

回答16:

Based on loveborg's excellent python snippet, I wrote this:

#!/bin/sh

# Version of readlink that follows links to the end; good for Mac OS X

for file in "$@"; do
  while [ -h "$file" ]; do
    l=`readlink $file`
    case "$l" in
      /*) file="$l";;
      *) file=`dirname "$file"`/"$l"
    esac
  done
  #echo $file
  python -c "import os,sys; print os.path.abspath(sys.argv[1])" "$file"
done

回答17:

FILEPATH="file.txt"
echo $(realpath $(dirname $FILEPATH))/$(basename $FILEPATH)

This works even if the file doesn't exist. It does require the directory containing the file to exist.

回答18:

I needed a solution that would do all three:

Work on a stock Mac. realpath and readlink -f are addons
Resolve symlinks
Have error handling

None of the answers had both #1 and #2. I added #3 to save others any further yak-shaving.

#!/bin/bash

P="${1?Specify a file path}"

[ -e "$P" ] || { echo "File does not exist: $P"; exit 1; }

while [ -h "$P" ] ; do
    ls="$(ls -ld "$P")"
    link="$(expr "$ls" : '.*-> \(.*\)$')"
    expr "$link" : '/.*' > /dev/null &&
        P="$link" ||
        P="$(dirname "$P")/$link"
done
echo "$(cd "$(dirname "$P")"; pwd)/$(basename "$P")"

Here is a short test case with some twisted spaces in the paths to fully exercise the quoting

mkdir -p "/tmp/test/ first path "
mkdir -p "/tmp/test/ second path "
echo "hello" > "/tmp/test/ first path / red .txt "
ln -s "/tmp/test/ first path / red .txt " "/tmp/test/ second path / green .txt "

cd  "/tmp/test/ second path "
fullpath " green .txt "
cat " green .txt "

回答19:

I know this is an ancient question. I'm still offering an alternative. Recently I met the same issue and found no existing and portable command to do that. So I wrote the following shell script which includes a function that can do the trick.

#! /bin/sh                                                                                                                                                

function normalize {
  local rc=0
  local ret

  if [ $# -gt 0 ] ; then
    # invalid
    if [ "x`echo $1 | grep -E '^/\.\.'`" != "x" ] ; then
      echo $1
      return -1
    fi

    # convert to absolute path
    if [ "x`echo $1 | grep -E '^\/'`" == "x" ] ; then
      normalize "`pwd`/$1"
      return $?
    fi

    ret=`echo $1 | sed 's;/\.\($\|/\);/;g' | sed 's;/[^/]*[^/.]\+[^/]*/\.\.\($\|/\);/;g'`
  else
    read line
    normalize "$line"
    return $?
  fi

  if [ "x`echo $ret | grep -E '/\.\.?(/|$)'`" != "x" ] ; then
    ret=`normalize "$ret"`
    rc=$?
  fi

  echo "$ret"
  return $rc
}

https://gist.github.com/bestofsong/8830bdf3e5eb9461d27313c3c282868c

回答20:

I discovered today that you can use the stat command to resolve paths.

So for a directory like "~/Documents":

You can run this:

stat -f %N ~/Documents

To get the full path:

/Users/me/Documents

For symlinks, you can use the %Y format option:

stat -f %Y example_symlink

Which might return a result like:

/usr/local/sbin/example_symlink

The formatting options might be different on other versions of *NIX but these worked for me on OSX.

回答21:

A simple solution using node.js:

#!/usr/bin/env node
process.stdout.write(require('path').resolve(process.argv[2]));

标签：