Yet another git cheatsheet.
- Git Cheatsheet
git branch # show branches
git branch --no-merged # show unmerged branches
git checkout my-branch # check out branch "my-branch"
git checkout -b my-branch # create a branch and checkout
git checkout -b my-branch 0766c053 # create a branch from commit 0766c053 and checkout
git checkout other-branch /path/to/file # check out file from other branch
git branch -d my-branch # delete local branch
git push -u origin my-branch # push local branch upstream
git for-each-ref --sort=-committerdate refs/heads/ --format='%(committerdate:short) %(authorname) %(refname:short)'
# list branches showing date (I usually alias this command in ~/.gitconfig: git branchlog)git add path/file1 path/file2 # add these files
git add . # add all new untracked files in current path
git add -u . # add all deleted files in current path
git add -A # add all new untracked and deleted files in current path
git reset # unstage all added filesgit commit # Commit staged changes
git commit -a # Stage and commit
git commit --amend # Fix last commit, forgot to add files, typo in message etc.
git commit --dry-run # See what will happen
git cherry -v # List unpushed commits
git push # Push all unpushed commits upstreamgit pull # fetch and merge from upstream branch
git merge other-branch # merge other branch into current branch
git checkout other-branch path/file1 path/file2 # pick specific files from another branch
git merge --strategy-option ours other-branch # Automatically resolve conflicts with our version,
# "theirs" for the other branch
git merge --squash other-branch # squash all commits to one big one before merge.
# Don't use if you intend to merge back... Fix conflicts manually by editing the file (or using some fancy merge tool), or, if the conflict is simple:
# Option 1, the correct file is in the other branch:
git checkout --theirs path/to/file
# Option 2, the correct file is in this branch
git checkout --ours path/to/file
# Option 3
git merge --abort # Abort the merge, then
git merge --strategy-option ours other-branch # Automatically resolve conflicts to "ours" (or "theirs")
# Finish the merge by adding and committingI changed my mind, I don't want to merge yet:
(master|MERGING)$ git merge --abort # abort merge process, revert to pre-merge situationOk, so you want to merge in another branch, but you have some content you want to keep:
First do
git config --global merge.ours.driver true # Adds properties to ~/.gitconfig
The create the file .gitattributes in your repo, adding for example:
#file: .gitattributes
content/** merge=ours
app/app.html merge=ours
.gitignore merge=oursNow you can merge without these files being overwritten.
Show status with timestamps (last modified)
git status -s | while read -n2 mode; read -n1; read file; do echo $mode $(date -r "$file" "+%Y-%m-%d %H:%M") $file; done|sortgit log --pretty=format:"%h%x09%an%x09%ad%x09%s" # Concise with times
git log # List commits in current branch
git log other-branch # List commits from another branch
git log --follow path/to/file # All commits for specific file
git show HEAD~1:path/to/file # Display file contents one commit back
git show 835a0edccd:path/to/file # Display file contents specific commit
git log -S'foobar' # Find string 'foobar' in history
git log -S'foobar' -- path/to/file # Find string 'foobar' in history of specific file
git log -1 # Show only -[number] latest commits
git log --since 2020-12-20 # Show only commits sice this date
for repo in *; do git -C ./$repo/ log -1; done; # Loop through all repos in this directory and display the last commit
git diff # Show changes since last commit
git diff my-branch other-branch # Show diff between branches
git diff other-branch -- path/to/file # Specific file
git diff --name-status other-branch # List files with differences in another branch
git diff --cached # List diff of staged files (double-check what you're about to commit)
git cherry -v # List unpushed commits
git fetch; git diff ..origin/master # check diff before merge or pull
git diff @{1}.. # What did I just merge in? - Post merge diff.
git diff 1ab2d4..901p07 # Show diff between two commitsStashing is useful when you want to temporary clean up modified files in your repo.
git stash # Stash uncommitted files, clean up working directory
git stash apply # Apply the stash to the current branch
git stash branch newbranch # Short for stash, checkout new branch, apply
git stash list # List stashes
git stash list --date=short # List stashes with creation dates
git stash apply stash@{2} # Apply second stash in list
git stash show -p | git apply -R # Un-apply last applied stash
git stash show -p stash@{0} | git apply -R # Un-apply specific stash
git clean -dfn # Delete untracked files (d = also directories, n = dry-run (remove n when sure))
Git is fantastic for undoing things and restoring from its unlimited history, but only if you've told git about the changes you've made. Commit frequently!
Undo accidental commit
When you did git commit -a and meant just to commit a single staged file. Or committed to the wrong branch.
git reset HEAD~ # Undo last commit, unstage changed files
git reset --soft HEAD~ # As above, but leave the files staged,
# if you only want to change the message or something.Getting pre-restore info
git log # List commits and get keys (eg. 835a0edccd) to copy-paste
git log --follow path/to/file # All commits for specific file
git show 835a0edccd:path/to/file # Display file contents specific commitUnstaged files
git reset # unstage added files
git stash # Discard uncommited changes
git clean -dfn # Delete untracked files (d = also directories, n = dry-run (remove n when sure))Restore specific files
git checkout 835a0edccd -- path/to/file # Get file from this commit, then: git commit
git checkout 835a0edccd^ -- path/to/file # Restore file deleted in this commit. The caret (^) means
# "as it was before committing", e.g. before deletion
git checkout other-branch path/file1 path/file2 # pick specific files from another branchRevert to or restore a specific commit
git cherry-pick 2e744aba6c # pick a specific commit from another branch
git revert 835a0edccd # Undo this commit without losing anything else
git revert a867b4af 25eee4ca 0766c053 # Revert a list of commits
git revert a867b4af..0766c053 # Revert a range of commits
git revert --abort # As with merge --abort, this lets you abort the process if it doesn't look good
git checkout -b newbranch 835a0edccd # Check out a new branch created from this commit
git reset --hard 835a0edccd # Reset to this commit. Removes subsequent commits, use revert if unsure.NOTE that revert apllies to the specified commit, while reset --hard keeps the specified commit and removes subsequent sommits.
**Revert to upstream situation **
git fetch origin # fetch everything from upstream
git reset --hard origin/master # Reset to this
Tags are good for release version numbers.
git tag # list tags
git tag -a v1.4 -m "My version 1.4"
git push --tags # push tags, pull --tags to pullFor npm projects, there's a cool auto versioning tool that both creates a git tag and updates the version of package.json. See https://docs.npmjs.com/cli/version
npm install version
npm version patch -m "My new patch" # Bump the patch version e.g. 1.4.1 => 1.4.2 ( patch | minor | major )
npm version v1.5.0 -m "My new version %s" # Bump the version explicitly. %s is a mask for the versionCreate the new repo on Bitbucket/Github, copy-paste the instructions from there or do:
cd /path/to/your/project
git init
git remote add origin [email protected]:me/newrepo.git
nano .gitignore # Do this now if you already have some files. Example below.
git add . # add everything. Too many files? git reset, edit .gitignore, add again
git commit # Initial commit
git push -u origin --all # First pushWhoops, did I set the wrong upstream url?
git remote -v # List upstream repos
git remote set-url origin [email protected]:me/correct-repo.git
You need to upload your public SSH key in order to be able to push upstream. This is quite simple:
cat ~/.ssh/id_rsa.pub # displays your public key. It should look something like this (only longer):
ssh-rsa AAAAB3NzaC1ycz7Xr1/9raqEhJ1fJDc6MQIp6JUswzHANJrG6VIR0qahCege1/zWMJ1
aC1ycz7Xr1/9raqEhJ1fJDc6MQIp6JUswzHANJrG6VIR0
O7bHPH54nnzteLhWti1KIlRbjDnXT+lsE6SCUNOSpLaoVQspL me@mymachine
If you get "No such file or directory" you need to generate a key pair first, do this and the repeat the first step. When you see the key, copy it to the clipboard (the whole thing starting with ssh-rsa).
# SKIP THIS STEP if you already have managed to display and copy down the key.
ssh-keygen # enter, enter, enter until done unless you want to change something. You now have a key pair, repeat the above step.
In Github/Bitbucket, go to settings (either user wide or a certain repo), look for SSH keys, Access keys or some such and paste the key you just copied as a new access key.
Copy the cloning address from Github/Bitbucket, choose ssh for repos you intend to commit to and have uploaded your SSH key to, https for repos you only want to use or don't have write permissions to.
cd /path/to/dir
# ssh:
git clone [email protected]:me/somerepo.git
# clone with submodules:
git clone [email protected]:me/somerepo.git --recurse-submodules
# https:
git clone https://[email protected]/me/somerepo.gitCloning will create a directory for the repo, so do this from the path you want the repo to reside in.
Clone just one remote branch into a new repository
git clone [email protected]:me/somerepo.git -b some-branch --single-branch /path/to/new_repo
Add a submodule to a repo (assuming the submodule already has an upstream repo):
~/myrepo/$ git submodule add [email protected]:me/mysubmod.git mysub # will add subdirectory ~/myrepo/mysubmod
Breakout existing dir into submodule (dir is named app in this example):
~/my_repo$ mv app ~/app_tmp # move Move "app" out the repo
~/my_repo$ git commit -a # Commit the deletion af "app"
# Go to github/bitbucket, greate a new repo "mysubmod"
# Then go to the moved folder and init the repo there:
~/app_tmp$ git init; git add .; git commit # The usual stuff, add .gitignore if needed
~/app_tmp$ git remote add origin [email protected]:me/mysubmod.git
~/app_tmp$ git push -u origin master
# If the submodule repo looks ok upstream, go back to the original repo:
~/my_repo$ git submodule add [email protected]:me/mysubmod.git app # Clones the repo into app
~/my_repo$ git commit -a
~/my_repo$ git push
Clone repo with submodule:
git clone [email protected]:me/myrepo.git --recurse-submodules
# Submodule is in detached head mode, so checkout branch if you want to commit something:
~/myrepo/$ cd mysubmod
~/myrepo/mysubmod$ git checkout master
# if you forgot to clone with --recurse-submodules, you can do:
~/myrepo/$ git submodule update --init --recursive
See also:
https://github.com/blog/2104-working-with-submodules
https://chrisjean.com/git-submodules-adding-using-removing-and-updating/
Make the current branch visible in the prompt (Ubuntu/Debian). In ~/.bashrc (somewhere before the prompt line) do:
# Include the git-prompt (this isn't needed on some systems, but safe to do)
source /etc/bash_completion.d/git-prompt
# Show dirty state with * like this: (master *)$
export GIT_PS1_SHOWDIRTYSTATE=1Then add $(__git_ps1) to the end of the prompt line, something like:
PS1='${debian_chroot:[ ... ] \w\[\033[00m\]$(__git_ps1)\$ 'Uncomment the line force_color_prompt=yes while you're at it.
Remember to secure your git directory in the web server, Apache 2.4 in this case. Use apache2.conf for system wide effect. Might as well add some other stuff while you're at it:
# Block all hidden files (.gitignore for example)
<FilesMatch "^\.">
Require all denied
</FilesMatch>
# Block all hidden directories (.git for example)
<DirectoryMatch "^(.*/)*\..*">
Require all denied
</DirectoryMatch>
# Block requests filename~ formatted backup files
<Files ~ "\~$">
Require all denied
</Files>
# Block requests for misc extensions, (?i) is case insensitive
<Files ~ "\.(?i)(inc|bak|orig|old|0|tmp?)$">
Require all denied
</Files>
Some basic excludes, save as .gitignore in your repo root.
*.orig
*.bak
*.log
*.tmp
*~
# npm projects
node_modules
bower_components
# Visual Studio Code metadata
.vscode
# Railo or Coldfusion metadata
WEB-INF
# For TypeScript projects like Angular2
app/**/*.js
app/**/*.mapSometimes you want to start ignoring files you've earlier tracked with git. Adding a line to gitignore won't be enough in this case, you also have to remove it from git's index with:
git rm --cached file/to/ignore
git rm -r 'app/*.js' # recursive wildcard, note single quotes
The --cached flag means "remove from git index only, don't remove from filesystem".
HOWEVER, the file WILL be deleted in another repository if you push-pull this commit there. So if you want to keep the file in the other repo, do the git rm --cached thing there as well (if too late, first restore the file with git checkout HEAD^ file/to/ignore).
Local exceptions to .gitignore
Local exceptions to .gitignore can be made in .git/info/exclude (same syntax as .gitignore). This is simpler than adding .gitignore to .gitignore, since it works immediately without the git rm --cached stuff.
After adding modified files to .git/info/exclude, do git update-index --assume-unchanged my_modified_file to assume unchanged.
Put aliases for useful commands in ~/.gitconfig
[alias]
branchdiff = diff --name-status
prettylog = log --pretty=format:"%h%x09%an%x09%ad%x09%s"
branchlog = for-each-ref --sort=-committerdate refs/heads/ --format='%(committerdate:short) %(authorname) %(refname:short)'When working on a web server with a devel and a production repo side by side, and you get bored with doing this all the time:
www/site_devel (master)$ git push
# wait...
www/site_devel (master)$ cd ../site
www/site (master)$ git pull
# wait...
www/site (master)$ cd ../site_devel/Instead, you could use my push-pull-script, and simply do:
www/site_devel (master)$ pushpullThe script shows which commits will be pushed and checks that you are in the correct dir and branch. The script is located at: https://github.com/subsite/misc-scripts/blob/master/pushpull
Wouldn't it be nice to have the changes you've made to the database included in the commit? One way is to dump the schema to a file in a pre-commit hook. (PostgreSQL shown here, for MySQL, look into the command mysqldump --no-data)
For a more manageable dump, you could use a script to split the dump into chunks by database object. This one here, for instance: https://github.com/subsite/misc-scripts/blob/master/split_pgdump.php
Save the following in your repo as .git/hooks/pre-commit (or add to it if it exists):
#!/bin/sh
## Dump database schema
db_database="database_name" # The database to dump
db_user="database_user" # The database user (password in ~/.pgpass)
db_schemafile="database/database_schema.sql"; # File to contain the schema, relative to repo root
db_outdir="database/schema_files" # OPTIONAL, for split_pgdump only
#
echo "Updating $db_schemafile"
db_schemafile="$(git rev-parse --show-toplevel)/$db_schemafile"
pg_dump -h localhost -U $db_user -s -f "$db_schemafile" $db_database
git add "$db_schemafile"
# OPTIONAL, for split_pgdump only
split_pgdump.php $db_schemafile $db_outdir
git add $db_outdir/*
## Dump database schema DONE