In a shell script, I want to download a file from some URL and save it to a specific folder. What is the specific CLI flag I should use to download files to a specific folder with the curl
command, or how else do I get that result?
I don't think you can give a path to curl, but you can CD to the location, download and CD back.
cd target/path && { curl -O URL ; cd -; }
Or using subshell.
(cd target/path && curl -O URL)
Both ways will only download if path exists. -O
keeps remote file name. After download it will return to original location.
If you need to set filename explicitly, you can use small -o
option:
curl -o target/path/filename URL
This option comes in curl 7.73.0:
curl --create-dirs -O --output-dir /tmp/receipes https://example.com/pancakes.jpg
curl
7.68.0 exists --create-dirs
but does not exist --output-dir
curl
doesn't have an option to that (without also specifying the filename), but wget
does. The directory can be relative or absolute. Also, the directory will automatically be created if it doesn't exist.
wget -P relative/dir "$url"
wget -P /absolute/dir "$url"
it works for me:
curl http://centos.mirror.constant.com/8-stream/isos/aarch64/CentOS-Stream-8-aarch64-20210916-boot.iso --output ~/Downloads/centos.iso
where:
--output
allows me to set up the path and the naming of the file and extension file that I want to place.
For powershell in Windows, you can add relative path + filename to --output
flag:
curl -L http://github.com/GorvGoyl/Notion-Boost-browser-extension/archive/master.zip --output build_firefox/master-repo.zip
here build_firefox is relative folder.
Use redirection:
This works to drop a curl
downloaded file into a specified path:
curl https://download.test.com/test.zip > /tmp/test.zip
Obviously "test.zip" is whatever arbitrary name you want to label the redirected file- could be the same name or a different name.
I actually prefer @oderibas solution, but this will get you around the issue until your distro supports curl version 7.73.0 or later-
Use wget
wget -P /your/absolut/path "https://jdbc.postgresql.org/download/postgresql-42.3.3.jar"
Here is an example using Batch to create a safe filename from a URL and save it to a folder named tmp/. I do think it's strange that this isn't an option on the Windows or Linux Curl versions.
@echo off
set url=%1%
for /r %%f in (%url%) do (
set url=%%~nxf.txt
curl --create-dirs -L -v -o tmp/%%~nxf.txt %url%
)
The above Batch file will take a single input, a URL, and convert the filename from the url. If no filename is specified it will be saved as tmp/.txt. So it's not all done for you but it gets the job done in Windows.
Success story sharing
basename /packages
"github.com/ziyaddin/xampp/archive/master.zip". But it says that wrong filename -->basename /packages
(cd target/path; curl -O URL)
cd
is the best option currently, though it seems slightly inelegant.