I have a bash script that runs my Go program. That's all it does, and when I run it from the command line, it works fine.
But when I run it by double clicking on it in Finder, it returns
/Users/colin/go/metgen/metaphorgenerator.sh: line 2: ./binary: No such file or directory
So I made it echo it's working directory, and it just prints /Users/colin, my home directory.
How do I get it to run the code from the directory the file is in? (I want it to work no matter what directory it's in)
You need to point it to the location of the executable, either with a relative path from the working directory as in go/metgen/binary, absolute path like /Users/colin/go/metgen/binary, or absolute path based on the parent executable (unfortunately not reliable).
$0 is the full program name. So you can get the directory with HERE=$(dirname "$0").
Then, line 2 should have ${HERE}/binary.
Related
Lets say I have a folder called xfile, inside of it we have runthis which is executable.
So, in order to execute that, I have to cd xfile and then ./runthis. and it will work just fine.
But, If I have a script call it run.sh and do this ./xfile/runthis, it will says run.sh: line xx: ./xfile/runthis: No such file or directory
So, how to make it happend without doing cd? It's been bugging me alot lately.
The leading dot in your command ./runthis means "relative to the current directory". But when you want to call it from outside the directory, remove the dot and provide the absolute path of your executable file xfile/runthis. The same is true when you want to call it from your run.sh
I am on Windows.
I have a script file named basics.sh and here is what it contains:
cd opt-out-exam/abduvosid_malikov/IT
mkdir made_by_my_script
cd made_by_my_script
echo "Hello World" > hello.txt
so basically, basics.sh script file is responsible to:
go to folder opt-out-exam/abduvosid_malikov/IT
make a directory made_by_my_script
create hello.txt file with content Hello World
Right now. to execute this basics.sh script, I am going to IT folder and writing this command in the terminal:
./basics.sh
In order to execute this basics.sh script, is it compulsory for me to go to IT folder
OR
is it possible to execute this script file even if I am staying in another folder (lets say currently working directory is opt-out-exam)
The first line is a change directory command followed by a relative path, not absolute. In such cases, it is important where you run the script. (An absolute path would start with the filesystem root, i. e. /.)
If you run this script from a directory (I wouldn't call it a folder in this context) where the relative path opt-out-exam/abduvosid_malikov/IT does not exist, it won't cd into it. But it will make a new directory without any problem, it will also create the file and write a line into it.
So only the first line will fail if it's run somewhere else.
UPD: As Gordon Davisson pointed out, this means that you want to check whether the directory change actually took place or not.
Per the title, I'd like to have a .sh file which I can drop into a directory and then:
Run http-server (the simple node server) in the current directory
Open Chrome and point it to that server
The idea is that when I'm developing I can quickly run this bash file and see the current version of whatever html/css/javascript website I am working on. My issue is that if you run a command in a bash file, it isn't run in the current working directory, but rather in the root directory (as far as I can tell). So if I just write http-server in my file, it will run a server not in the current directory, but in ./.
To fix this, I want to cd to the directory first, and then run the script. current_dir=$(pwd) will give me something close to the current directory, but I can't put that directly into a cd command because it (1) doesn't have quotes around it, so spaces in directory names will make it not work, and (2) it starts with /C/ instead of /C:/. Can anyone advise me on how to fix this?
My current code looks as follows.
curr_dir=$(pwd)
cd $curr_dir
http-server
start chrome localhost/XXX
And, as mentioned, results in the http-server command being run in the wrong place:
Starting up http-server, serving ./
Available on:
http://192.168.56.1:8081
http://192.168.1.21:8081
http://127.0.0.1:8081
Hit CTRL-C to stop the server
e: For anyone who happens on this later, this isn't actually a problem -- see the accepted answer. I had another typo.
No; the commands you run always run in the current directory. Otherwise, if your hypothesis were correct, e.g. ls would always show the files in the root directory.
If you want to run a command which is in a different directory, you want
../relative/path/to/command
or
/absolute/path/to/command
Neither of these change the current directory of the shell. The current working directory of the process you create will remain the directory you were in when you run this command. (Even if you run it in the background and subsequently change to a different directory in your interactive shell, for example.)
If you want the command to run in the directory where the script lives on the disk, something like this can occasionally be useful.
cd "$(dirname "$0")"
but again, most of the time, you want and need your commands to run in the current directory.
(There are situations where you want a script to process data files in a fixed location, but these are rare exceptions. Until you have such a situation, consider it a bug to use cd in a shell script.)
If you have a command http-server somewhere on your PATH, just http-server will run that. You should normally not have the current directory on the PATH, but to run the binary in the current directory instead of from anywhere on the PATH, you say so:
./http-server
I got Gnu Utilities that adds the command sed to windows, but to use it I have to type:
C:\ProgramFiles\GnuWin32\bin\sed.exe <args>
How do I shorten this to just sed <args>?
To run an executable without a full path, it needs to be either in the current directory, or in the PATH environment variable. In the CMD prompt, there are several ways to do this.
The first way is to put C:\ProgramFiles\GnuWin32\bin in your PATH variable, which makes every program in that directory available without a full path.
set "PATH=%path%;C:\ProgramFiles\GnuWin32\bin"
This updates PATH in the current command prompt. If you need to set it for other CMD windows, see How to persistently set a variable in Windows 7 from a batch file?
The second method is to have sed.exe in the current directory. The most obvious way to do that is to change directories.
cd C:\ProgramFiles\GnuWin32\bin
sed
Or you can copy it to your current directory.
copy C:\ProgramFiles\GnuWin32\bin\sed.exe .\
sed
(This works with sed.exe because it's a self-contained utility. Don't try this with a Windows application like excel.exe)
Finally, you can create a "redirect" somewhere in the current directory or the path.
>.\sed.bat echo C:\ProgramFiles\GnuWin32\bin\sed.exe %*
This creates a batch file in the current directory called sed.bat that calls the full sed.exe. You can drop this file into any directory in your PATH.
mklink .\sed.exe C:\ProgramFiles\GnuWin32\bin\sed.exe
This creates a symlink to the sed.exe in the current directory, much like a symlink in Unix or a shortcut in Windows.
I have a command-line tool written in Cocoa. Let's call it processFile. So if I am in the terminal and I type in the command ./processFile foo, it looks for a file named foo.html in the same directory as the executable of processFile. If it finds one, it reads it and does some stuff to create fooProcessed.html.
Now I want to modify my tool so that it looks for foo.html in the directory from which it was launched. So if I am in the terminal with current directory ~/documents/html, and processFile executable is in usr/bin, and I type in
processFile foo
it will find and process the file ~/documents/foo.html.
The problem is that I don't know how to get the directory from which the tool was invoked. How can I do that?
That's the current working directory. First of all, any attempt to access the file just using its name and no path will automatically use the working directory. So, if you simply take "foo", append ".html", and attempt to open the file, that will work. If the user specified a relative path, like "subdir/foo", that would also work. It would resolve the relative path starting from the current working directory.
You can also query the working directory using the getcwd() routine.