git-deploy-php is a simple php-based tool that deploys your Git repositories to FTP servers, and keeps them updated automatically.
Deploys a local git repository quickly and easily onto a remote server via FTP.
Git-deploy only requires FTP access, no SSH, so it will work even on the most restrictive of shared hosting plans. It supports both pushing and pulling, so if you already have something deployed it's simple to pull it down, put it under version control, and deploy it again! Even better, when pushing up to the server only changed files are uploaded. You save both bandwidth and time.
And best of all, it's quick, simple, and user-friendly. You just have to know how to open up the command line.
--dry-run flag to test out a command before doing it for real.--update-self command, to update to the latest version of the script from GitHub.git-deploy file.development or production.Using git-deploy for the first time doesn't take a lot of work.
git-deploy into your project.php git-deploy --setup
deploy-config.ini. When you're done, execute php git-deploy to deploy your changes.Super simple.
And later, pushing changes is as easy as running php git-deploy once more. You're done!
The configuration progess is pretty simple and the default config file walks you through it. The only thing to note is that if you have your project deployed already and want to upload a new revision, you'll either have to clean the remote location first and upload from scratch, or upload a file named REVISION containing the SHA-1 hash of the currently uploaded revision. There's a convenience command to do this for you: First your server in deploy-config.ini, then run php git-deploy --make-revfile=<identifier>. You'll be prompted for the revision to set, and it'll be automatically uploaded for you.
If you don't know what revision the server is on (but you know how many commits back it is from HEAD), running git rev-parse --short HEAD~<number of commits back> will return the SHA-1 hash for that commit, on the current branch. More on that here. In particular, check out this section on ancestry references. The basic guideline is below:
HEAD~n returns the ancestor of the current branch n number of steps back from HEAD on the current branch.branch~n is like HEAD~n, but it works relative the tip of branch.After running --setup you'll notice a file called .deployignore. It works similarly to a .gitignore file, except that it excludes files from being uploaded (and it supports optional regex patters!). You'll find more information in the default file.
Git-deploy has a lot of user-friendly functionality built in. Command line arguments don't take long, complicated inputs, and git-deploy strives to make sure you don't break anything unintentionally. That means that if you pass a dangerous argument, or two that don't make sense together, git-deploy will stop and ask you what to do. The last thing you want is your hard work accidentally deleted!
If you're afraid of wrecking something on your server, run git-deploy with the --dry-run flag turned on. Dry run mode won't run any commands that change files, but will still log output as if it were operating normally. This lets you know exactly what would change if you were to run that command without --dry-run turned on.
Git-deploy doesn't have to be called via php git-deploy either. You can safely rename the script and it will still operate correctly.
Git-deploy only has two requirements.
Simply place git-deploy in the root directory of your git repository and run:
php git-deploy --setup
This will create all needed files under the .deploy/ directory. Once that is complete, you can edit the deploy-config.ini file found inside with the appropriate values. The file is well-commented, so it shouldn't be too hard.
If the configuration file is ever mangled or broken, removing it and re-running --setup will restore it to its default state. If git-deploy is run and no configuration file is found, the script will ask you to either create it yourself or to run --setup again.
The syntax isn't complicated.
php git-deploy [<tree-ish>] [flags]
Make sure to specify the [<tree-ish>' command first, so it doesn't get parsed as an argument to one of the flags.
php git-deploy HEAD~1 -csn --only testserver
...which will clean the upload directory of the server identified by 'testserver' (as configured in deploy-config.ini) without logging or printing output.
[<tree-ish>]
Specify the revision to deploy. You can use any format recognizable by git-diff (hash, tag, etc). Defaults to HEAD of the current branch if not specified. Note that files matching the following patterns will not be deployed for security reasons: git-deploy*, .deploy/, and a match for the script's curent name. (In case of rename.)
--setup
Run setup.
-c--clean
Completely clear out the upload directory.
-l--list
List files to upload or delete.
-p=<identifier>--pull=<identifier>
Pull contents down from the server specified with the given identifier into the directory of the git-deploy script. Useful for putting an existing deployment under version control for the first time. (Just run --pull=<identifier>, git add -A, git commit and you're done!) Certain files or directories are excluded from being pulled. These include anything that match the patterns git-deploy*, .deploy/, or REVISION. If the file is named the same as this script (in case you renamed it from git-deploy), it will not be pulled down either. (Pattern: basename(__FILE__)*)
-u--update-self
Fetch a newer copy of this script if it exists on GitHub and update to it. This script will keep its name, so even if it isn't named git-deploy this command (like all the others) will still work.
-m=<identifier>
--make-revfile=<identifier>
Uploads a REVISION file to the server with the configured identifier. This is basically only useful for first-time setup, where you might already have code uploaded but without a REVISION file to identify the currently deployed revision to git-deploy. In that case you can run --make-revfile and pass the identifier of the server you want to upload the REVISION file to. You'll be prompted during upload for the revision hash/tag/etc. to set.
-o=<identifier-list>--only=<identifier-list>
Only run git-deploy on the servers with the given identifiers in the config file. If more than one is specified, they should be enclosed in quote characters and space delineated.
-t=<taglist>
--tagged-with=<taglist>
Deploys only to servers tagged with the given identifiers in deploy-config.ini. If more than one tag is specified, the list should be enclosed in quotes and space-delineated. A server only has to have one tag in the list in order to be deployed, but you can specifically exlude certain tags by prefixing the tag with an exclamation mark. Note that --tagged-with will not override --only. If both --only and --tagged-with are specified, a server must match both to be included.
-s--silent
Silence command line output.
-n--nolog
Do not create a log file.
-d--dry-run
Run a command without changing local or remote files, but log results as if the command were actually being run.
Overrides --silent, but not --nolog.
-r--resume-only
If a RESUME file exists for a given server, only resume the upload and do not continue with a new one.
-f--show-filtered
Show which files are excluded from upload, and by which exclude pattern. Useful for debugging the .deployignore file.
Git-deploy stores file called REVISION on your server inside the root path to your application. This file stores the current revision of your application residing on your server.
When you run a git deploy, git-deploy downloads the REVISION file, and checks to see what files are different between revisions and either upload the changed files or deletes them from the server.
Feel free to contribute! The source code is pretty clean and easy to go through, even if your PHP skills aren't that great. I've documented it more than strictly necessary for exactly that reason -- so get cracking! No pull request is too small.