How can I download only a specific folder or directory from a remote Git repo hosted on GitHub?
Say the example GitHub repo lives here:
[email protected]:foobar/Test.git
Its directory structure:
Test/
├── foo/
│ ├── a.py
│ └── b.py
└── bar/
├── c.py
└── d.py
I want to download only the foo folder and not clone the whole Test project.
Update Apr. 2021: there are a few tools created by the community that can do this for you:
Note: if you're trying to download a large number of files, you may need to provide a token to these tools to avoid rate limiting.
Original (manual) approach: Checking out an individual directory is not supported by git natively, but Github can do this via SVN. If you checkout your code with subversion, Github will essentially convert the repo from git to subversion on the backend, then serve up the requested directory.
Here's how you can use this feature to download a specific folder. I'll use the popular javascript library lodash as an example.
Navigate to the folder you want to download. Let's download /test from master branch.
Modify the URL for subversion. Replace tree/master with trunk.
https://github.com/lodash/lodash/tree/master/test ➜
https://github.com/lodash/lodash/trunk/test
Download the folder. Go to the command line and grab the folder with SVN.
svn checkout https://github.com/lodash/lodash/trunk/test
You might not see any activity immediately because Github takes up to 30 seconds to convert larger repositories, so be patient.
Full URL format explanation:
- If you're interested in
masterbranch, usetrunkinstead. So the full path istrunk/foldername- If you're interested in
foobranch, usebranches/fooinstead. The full path looks likebranches/foo/foldername- Protip: You can use
svn lsto see available tags and branches before downloading if you wish
That's all! Github supports more subversion features as well, including support for committing and pushing changes.
You can DIRECTLY DOWNLOAD or create DOWNLOAD LINK for any GitHub public directory or file from DownGit. Here is a simple demonstration-
You may also configure properties of the downloaded file- detailed usage.
Disclaimer: I fell into the same problem as the question-asker and could not find any proper solution. So, I created this tool for my own use first, then opened it for everyone :)
Usage:
Get Token:
http://kinolien.github.io/gitzip by using GitHub API, and JSZip, FileSaver.js libraries.
Step1: Input github url to the field at the top-right.
Step2: Press enter or click download for download zip directly or click search for view the list of sub-folders and files.
Step3: Click "Download Zip File" or "Get File" button to get files.
In most cases, it works fine, except that the folder contains more than 1,000 files, because of the Github Trees API limitation. (refers to Github API#Contents)
And it also can support private/public repos and upgrade the rate limit, if you have GitHub account and use "get token" link in this site.
If you have svn, you can use svn export to do this:
svn export https://github.com/foobar/Test.git/trunk/foo
Notice the URL format:
https://github.com//trunk appended at the endBefore you run svn export, it's good to first verify the content of the directory with:
svn ls https://github.com/foobar/Test.git/trunk/foo
If you want to download files, not clone the repository with history, you can do this with git-archive.
git-archive makes a compressed zip or tar archive of a git repository. Some things that make it special:
.git/ folder, or any untracked files in the repository it's run on.An example of creating an archive of the docs/usage directory from a remote repo you're connected to with ssh:
# in terminal
$ git archive --format tar --remote ssh://server.org/path/to/git HEAD docs/usage > /tmp/usage_docs.tar
More information in this blog post and the git documentation.
Nothing wrong with other answers but I just thought I'd share step-by-step instructions for those wandering through this process for the first time.
~ To open Terminal just click spotlight and type terminal then hit enter
BTW - If you are on Windows or some other platform you can find a binary download of subversion (svn) at http://subversion.apache.org
~ If you want to checkout the folder rather than simply download it try using the svn help (tldr: replace export with checkout)
Update
Regarding the comment on resuming an interrupted download/checkout. I would try running svn cleanup followed by svn update. Please search SO for additional options.
Whoever is working on specific folder he needs to clone that particular folder itself, to do so please follow below steps by using sparse checkout.
Create a directory.
Initialize a Git repository. (git init)
Enable Sparse Checkouts. (git config core.sparsecheckout true)
Tell Git which directories you want (echo 2015/brand/May( refer to folder you want to work on) >> .git/info/sparse-checkout)
Add the remote (git remote add -f origin https://jafartke.com/mkt-imdev/DVM.git)
Fetch the files (git pull origin master )
You cannot; unlike Subversion, where each subdirectory can be checked out individually, Git operates on a whole-repository basis.
For projects where finer-grained access is necessary, you can use submodules -- each submodule is a separate Git project, and thus can be cloned individually.
It is conceivable that a Git front-end (e.g. GitHub's web interface, or gitweb) could choose to provide an interface for you to extract a given folder, but to my knowledge none of them do that (though they do let you download individual files, so if the folder does not contain too many files, that is an option)
Edit - GitHub actually offers access via SVN, which would allow you to do just this (as per comment). See https://github.com/blog/1438-improved-svn-here-to-stay-old-svn-going-away for latest instructions on how to do this
There are a variety of ways to handle this, depending on whether or not you want to do this manually or programmatically.
There are four options summarized below. And for those that prefer a more hands-on explanation, I've put together a YouTube video: Download Individual Files and Folders from GitHub.
Also, I've posted a similar answer on StackOverflow for those that need to download single files from GitHub (as opposed to folders).
1. GitHub User Interface
2. Third Party Tools
https://github.com/babel/babel-eslint/tree/master/lib) and press the "Download" button.3. Subversion
svn export https://github.com/babel/babel-eslint/trunk/lib, and press enter.4. GitHub API
https://api.github.com/repos/:owner/:repo/contents/:path. After replacing the placeholders, an example endpoint is: https://api.github.com/repos/babel/babel-eslint/contents/lib. This gives you JSON data for all of the content that exists in that folder. The data has everything you need, including whether or not the content is a folder or file, a download URL if it's a file, and an API endpoint if it's a folder (so that you can get the data for that folder). Using this data, the script can recursively go through all content in the target folder, create folders for nested folders, and download all of the files for each folder. Check out DownGit's code for inspiration.If you truly just want to just "download" the folder and not "clone" it (for development), the easiest way to simply get a copy of the most recent version of the repository (and therefore a folder/file within it), without needing to clone the whole repo or even install git in the first place, is to download a zip archive (for any repo, fork, branch, commit, etc.) by going to the desired repository/fork/branch/commit on GitHub (e.g. http(s)://github.com/<user>/<repo>/commit/<Sha1> for a copy of the files as they were after a specific commit) and selecting the Downloads button near the upper-right.
This archive format contains none of the git-repo magic, just the tracked files themselves (and perhaps a few .gitignore files if they were tracked, but you can ignore those :p) - that means that if the code changes and you want to stay on top, you'll have to manually re-download it, and it also means you won't be able to use it as a git repository...
Not sure if that's what you're looking for in this case (again, "download"/view vs "clone"/develop), but it can be useful nonetheless...
There's a Python3 pip package called githubdl that can do this*:
export GIT_TOKEN=1234567890123456789012345678901234567890123
pip install githubdl
githubdl -u http://github.com/foobar/test -d foo
The project page is here
* Disclaimer: I wrote this package.
If you are comfortable with unix commands, you don't need special dependencies or web apps for this. You can download the repo as a tarball and untar only what you need.
Example (woff2 files from a subdirectory in fontawesome):
curl -L https://api.github.com/repos/FortAwesome/Font-Awesome/tarball | tar xz --wildcards "*/web-fonts-with-css/webfonts/*.woff2" --strip-components=3
*/) to match any directory. Github creates a wrapper directory with the commit ref in the name, so it can't be known.--strip-components to be the same as the amount of slashes (/) in the path (previous argument).This will download the whole tarball. Use the SVN method mentioned in the other answers if this has to be avoided or if you want to be nice to the GitHub servers.
Another specific example:
Like I want to download 'iOS Pro Geo' folder from the url
https://github.com/alokc83/APRESS-Books-Source-Code-/tree/master/%20Pro%20iOS%20Geo
and I can do so via
svn checkout https://github.com/alokc83/APRESS-Books-Source-Code-/trunk/%20Pro%20iOS%20Geo
Note trunk in the path
Edited: (as per Tommie C's comment)
Yes, using export instead of checkout would give a clean copy without extra git repository files.
svn export https://github.com/alokc83/APRESS-Books-Source-Code-/trunk/%20Pro%20iOS%20Geo
Edited: If tree/master is not there in url then Fork it and it will be there in Forked url.
You can do a simple download of the directory tree:
git archive --remote [email protected]:foobar/Test.git HEAD:foo | tar xf -
But if you mean to check it out, and be able to do commits and push them back, no you can't do that.
git clone --filter from git 2.19 now works on GitHubTested 2020-09-18, git 2.25.1.
This option was added together with an update to the remote protocol, and it truly prevents objects from being downloaded from the server.
E.g., to clone only objects required for d1 of this repository: https://github.com/cirosantilli/test-git-partial-clone I can do:
git clone \
--depth 1 \
--filter=blob:none \
--no-checkout \
https://github.com/cirosantilli/test-git-partial-clone \
;
cd test-git-partial-clone
git checkout master -- d1
I have covered this in more detail at: Git: How do I clone a subdirectory only of a Git repository?
None of the answers helped in my situation. If you are developing for Windows, you likely don't have svn. In many situations one can't count on users to have Git installed either, or don't want to download entire repositories for other reasons. Some of the people that answered this question, such as Willem van Ketwich and aztack, made tools to accomplish this task. However, if the tool isn't written for the language you are using, or you don't want to install a third party library, these don't work.
However, there is a much easier way. GitHub has an API that allows you to download a single file or an entire directory's contents using GET requests. You can access a directory using https://api.github.com/repos/:owner/:repo_name/contents/:path that returns a JSON object enumerating all the files in the directory. Included in the enumeration is a link to the raw content of the file, the download_url parameter. The file can then be downloaded using that URL.
It's a two step process that requires the ability to make GET requests, but this can be implemented in pretty much any language, on any platform. It can be used to get files or directories.
git sparse-checkoutGit 2.25.0 includes a new experimental
git sparse-checkoutcommand that makes the existing feature easier to use, along with some important performance benefits for large repositories. (The GitHub Blog)
Example with current version:
git clone --filter=blob:none --sparse https://github.com/git/git.git
cd git
git sparse-checkout init --cone
git sparse-checkout add t
--sparse checks out only top-level directory files of git repository into working copygit sparse-checkout add t incrementally adds/checks out t subfolder of git git sparse-checkout init does some preparations to enable partial checkouts--filter=blob:none optimizes data fetching by downloading only necessary git objects (take a look at partial clone feature for further infos)--cone also speeds up performance by applying more restricted file inclusion patternsGitHub is still evaluating this feature internally while it’s enabled on a select few repositories [...]. As the feature stabilizes and matures, we’ll keep you updated with its progress. (docs)
This is how I do it with git v2.25.0, also tested with v2.26.2. This trick doesn't work with v2.30.1
git clone --no-checkout --filter=tree:0 https://github.com/opencv/opencv
cd opencv
# requires git 2.25.x to 2.26.2
git sparse-checkout set data/haarcascades
You can use Docker to avoid installing a specific version of git
git clone --no-checkout --filter=tree:0 https://github.com/opencv/opencv
cd opencv
# requires git 2.25.x to 2.26.2
docker run --rm -it -v $PWD/:/code/ --workdir=/code/ alpine/git:v2.26.2 sparse-checkout set data/haarcascades
# bare minimum clone of opencv
$ git clone --no-checkout --filter=tree:0 https://github.com/opencv/opencv
...
Resolving deltas: 100% (529/529), done.
# Downloaded only ~7.3MB , takes ~3 seconds
# du = disk usage, -s = summary, -h = human-readable
$ du -sh opencv
7.3M opencv/
# Set target dir
$ cd opencv
$ git sparse-checkout set data/haarcascades
...
Updating files: 100% (17/17), done.
# Takes ~10 seconds, depending on your specs
# View downloaded files
$ du -sh data/haarcascades/
9.4M data/haarcascades/
$ ls data/haarcascades/
haarcascade_eye.xml haarcascade_frontalface_alt2.xml haarcascade_licence_plate_rus_16stages.xml haarcascade_smile.xml
haarcascade_eye_tree_eyeglasses.xml haarcascade_frontalface_alt_tree.xml haarcascade_lowerbody.xml haarcascade_upperbody.xml
haarcascade_frontalcatface.xml haarcascade_frontalface_default.xml haarcascade_profileface.xml
haarcascade_frontalcatface_extended.xml haarcascade_fullbody.xml haarcascade_righteye_2splits.xml
haarcascade_frontalface_alt.xml haarcascade_lefteye_2splits.xml haarcascade_russian_plate_number.xml
Just to amplify the answers above, a real example from a real GitHub repository to a local directory would be:
svn ls https://github.com/rdcarp/playing-cards/trunk/PumpkinSoup.PlayingCards.Interfaces
svn export https://github.com/rdcarp/playing-cards/trunk/PumpkinSoup.PlayingCards.Interfaces /temp/SvnExport/Washburn
Sometimes a concrete example helps clarify the substitutions proposed.
It's one of the few places where SVN is better than Git.
In the end we've gravitated towards three options:
For whatever reason, the svn solution does not work for me, and since I have no need of svn for anything else, it did not make sense to spend time trying to make it, so I looked for a simple solution using tools I already had. This script uses only curl and awk to download all files in a GitHub directory described as "/:user:repo/contents/:path".
The returned body of a call to the GitHub REST API
"GET /repos/:user:repo/contents/:path" command returns an object that includes a "download_url" link for each file in a directory.
This command-line script calls that REST API using curl and sends the result through AWK, which filters out all but the "download_url" lines, erases quote marks and commas from the links, and then downloads the links using another call to curl.
curl -s https://api.github.com/repos/:user/:repo/contents/:path | awk \
'/download_url/ { gsub("\"|,", "", $2); system("curl -O "$2"); }'
cd %ProgramFiles%\SlikSvn\binhttps://github.com/ZeBobo5/Vlc.DotNet/tree/develop/src/Samplestree/develop
or tree/master with trunksvn export https://github.com/ZeBobo5/Vlc.DotNet/trunk/src/Samples
Our team wrote a bash script to do this because we didn't want to have to install SVN on our bare bones server.
https://github.com/ojbc/docker/blob/master/java8-karaf3/files/git-download.sh
It uses the github API and can be run from the command line like this:
git-download.sh https://api.github.com/repos/ojbc/main/contents/shared/ojb-certs
I work with CentOS 7 servers on which I don't have root access, nor git, svn, etc (nor want to!) so made a python script to download any github folder: https://github.com/andrrrl/github-folder-downloader
Usage is simple, just copy the relevant part from a github project, let's say the project is https://github.com/MaxCDN/php-maxcdn/, and you want a folder where some source files are only, then you need to do something like:
$ python gdownload.py "/MaxCDN/php-maxcdn/tree/master/src" /my/target/dir/
(will create target folder if doesn't exist)
It requires lxml library, can be installed with easy_install lxml
If you don't have root access (like me) you can create a .pydistutils.py file into your $HOME dir with these contents:
[install]
user=1
And easy_install lxml will just work (ref: https://stackoverflow.com/a/33464597/591257).
Open repo to codesandbox by replacing github to githubbox in url and on codesandbox go to file menu and Export it as a zip.
For following repo: https://github.com/geist-org/react/tree/master/examples/custom-themes
Enter following url: https://githubbox.com/geist-org/react/tree/master/examples/custom-themes
In codesandbox go to file menu and Export it as a Zip.
You can use ghget with any URL copied from the address bar:
ghget https://github.com/fivethirtyeight/data/tree/master/airline-safety
It's a self-contained portable shell script that doesn't use SVN (which didn't work for me on a big repo). It also doesn't use the API so it doesn't require a token and isn't rate-limited.
Disclaimer: I made it.
To export a directory from GitHub, replace "/tree/master/" in the directory's url with "/trunk/".
For example, to export the directory from the following URL:
https://github.com/liferay/liferay-plugins/tree/master/portlets/sample-hibernate-portlet
run the following command:
svn export https://github.com/liferay/liferay-plugins/trunk/portlets/sample-hibernate-portlet
A straightforward answer to this is to first tortoise svn from following link.
while installation turn on CLI option, so that it can be used from command line interface.
copy the git hub sub directory link.
Example
https://github.com/tensorflow/models/tree/master/research/deeplab
replace tree/master with trunk
and do
svn checkout https://github.com/tensorflow/models/trunk/research/deeplab
files will be downloaded to the deeplab folder in the current directory.
