gophernotes - Use Go in Jupyter notebooks and nteract

gophernotes is a Go kernel for Jupyter notebooks and nteract. It lets you use Go interactively in a browser-based notebook or desktop app. Use gophernotes to create and share documents that contain live Go code, equations, visualizations and explanatory text. These notebooks, with the live Go code, can then be shared with others via email, Dropbox, GitHub and the Jupyter Notebook Viewer. Go forth and do data science, or anything else interesting, with Go notebooks!

Acknowledgements - This project utilizes a Go interpreter called gomacro under the hood to evaluate Go code interactively. The gophernotes logo was designed by the brilliant Marcus Olsson and was inspired by Renee French's original Go Gopher design.

• Examples
• Install gophernotes:
  • Prerequisites
  • FreeBSD
  • Linux
  • Mac
  • Windows
  • Docker
• Getting Started
• Limitations
• Troubleshooting

Examples

Jupyter Notebook:

nteract:

Example Notebooks (download and run them locally, follow the links to view in Github, or use the Jupyter Notebook Viewer):

• Worker Pools
• Matrix Operations
• Facial Recognition
• Display Images, HTML, LaTeX...

Installation

Prerequisites

• Go 1.13+ - including GOPATH/bin added to your PATH (i.e., you can run Go binaries that you go install).
• Jupyter Notebook or nteract
• git - usually already present on Linux and Mac OS X. If not present, follow the instructions at https://git-scm.com/download

Linux or FreeBSD

The instructions below should work both on Linux and on FreeBSD.

Method 1: quick installation as module

  go install github.com/gopherdata/gophernotes@v0.7.5
  mkdir -p ~/.local/share/jupyter/kernels/gophernotes
  cd ~/.local/share/jupyter/kernels/gophernotes
  cp "$(go env GOPATH)"/pkg/mod/github.com/gopherdata/gophernotes@v0.7.5/kernel/*  "."
  chmod +w ./kernel.json # in case copied kernel.json has no write permission
  sed "s|gophernotes|$(go env GOPATH)/bin/gophernotes|" < kernel.json.in > kernel.json

Method 2: manual installation from GOPATH

  mkdir -p "$(go env GOPATH)"/src/github.com/gopherdata
  cd "$(go env GOPATH)"/src/github.com/gopherdata
  git clone https://github.com/gopherdata/gophernotes
  cd gophernotes
  git checkout -f v0.7.5
  go install
  mkdir -p ~/.local/share/jupyter/kernels/gophernotes
  cp kernel/* ~/.local/share/jupyter/kernels/gophernotes
  cd ~/.local/share/jupyter/kernels/gophernotes
  chmod +w ./kernel.json # in case copied kernel.json has no write permission
  sed "s|gophernotes|$(go env GOPATH)/bin/gophernotes|" < kernel.json.in > kernel.json

To confirm that the gophernotes binary is installed in GOPATH, execute it directly:

  "$(go env GOPATH)"/bin/gophernotes

and you should see the following:

2017/09/20 10:33:12 Need a command line argument specifying the connection file.

Note - if you have the JUPYTER_PATH environmental variable set or if you are using an older version of Jupyter, you may need to copy this kernel config to another directory. You can check which directories will be searched by executing:

  jupyter --data-dir

Mac

Important Note - gomacro relies on the plugin package when importing third party libraries. This package works reliably on Mac OS X with Go 1.10.2+ as long as you never execute the command strip gophernotes.

Method 1: quick installation as module

  go install github.com/gopherdata/gophernotes@v0.7.5
  mkdir -p ~/Library/Jupyter/kernels/gophernotes
  cd ~/Library/Jupyter/kernels/gophernotes
  cp "$(go env GOPATH)"/pkg/mod/github.com/gopherdata/gophernotes@v0.7.5/kernel/*  "."
  chmod +w ./kernel.json # in case copied kernel.json has no write permission
  sed "s|gophernotes|$(go env GOPATH)/bin/gophernotes|" < kernel.json.in > kernel.json

Method 2: manual installation from GOPATH

  mkdir -p "$(go env GOPATH)"/src/github.com/gopherdata
  cd "$(go env GOPATH)"/src/github.com/gopherdata
  git clone https://github.com/gopherdata/gophernotes
  cd gophernotes
  git checkout -f v0.7.5
  go install
  mkdir -p ~/Library/Jupyter/kernels/gophernotes
  cp kernel/* ~/Library/Jupyter/kernels/gophernotes
  cd ~/Library/Jupyter/kernels/gophernotes
  chmod +w ./kernel.json # in case copied kernel.json has no write permission
  sed "s|gophernotes|$(go env GOPATH)/bin/gophernotes|" < kernel.json.in > kernel.json

To confirm that the gophernotes binary is installed in GOPATH, execute it directly:

  "$(go env GOPATH)"/bin/gophernotes

and you should see the following:

2017/09/20 10:33:12 Need a command line argument specifying the connection file.

Note - if you have the JUPYTER_PATH environmental variable set or if you are using an older version of Jupyter, you may need to copy this kernel config to another directory. You can check which directories will be searched by executing:

  jupyter --data-dir

Windows

Important Note - gomacro relies on the plugin package when importing third party libraries. This package is only supported on Linux and Mac OS X currently. Thus, if you need to utilize third party packages in your Go notebooks and you are running on Windows, you should use the Docker install and run gophernotes/Jupyter in Docker.

1. Download gophernotes inside GOPATH, compile and install it

   go env GOPATH > temp.txt
   set /p GOPATH=<temp.txt
   mkdir %GOPATH%\src\github.com\gopherdata
   cd /d %GOPATH%\src\github.com\gopherdata
   git clone https://github.com/gopherdata/gophernotes
   cd gophernotes
   git checkout -f v0.7.5
   go install
   

2. Copy the kernel config:

   mkdir %APPDATA%\jupyter\kernels\gophernotes
   xcopy %GOPATH%\src\github.com\gopherdata\gophernotes\kernel %APPDATA%\jupyter\kernels\gophernotes /s
   

   Note, if you have the JUPYTER_PATH environmental variable set or if you are using an older version of Jupyter, you may need to copy this kernel config to another directory. You can check which directories will be searched by executing:

   jupyter --data-dir
   

3. Update %APPDATA%\jupyter\kernels\gophernotes\kernel.json with the FULL PATH to your gophernotes.exe (usually in %GOPATH%\bin). For example:

   {
       "argv": [
         "C:\\gopath\\bin\\gophernotes.exe",
         "{connection_file}"
         ],
       "display_name": "Go",
       "language": "go",
       "name": "go"
   }
   

Docker

You can try out or run Jupyter + gophernotes without installing anything using Docker. To run a Go notebook that only needs things from the standard library, run:

  docker run -it -p 8888:8888 gopherdata/gophernotes

Or to run a Go notebook with access to common Go data science packages (gonum, gota, golearn, etc.), run:

  docker run -it -p 8888:8888 gopherdata/gophernotes:latest-ds

In either case, running this command should output a link that you can follow to access Jupyter in a browser. Also, to save notebooks to and/or load notebooks from a location outside of the Docker image, you should utilize a volume mount. For example:

  docker run -it -p 8888:8888 -v /path/to/local/notebooks:/path/to/notebooks/in/docker gopherdata/gophernotes

Getting Started

Jupyter

• If you completed one of the local installs above (i.e., not the Docker install), start the jupyter notebook server:

  jupyter notebook
  

• Select Go from the New drop down menu.

• Have fun!

nteract

• Launch nteract.

• From the nteract menu select Language -> Go.

• Have fun!

Special commands

In addition to Go code, the following special commands are also supported - they must be on a line by their own:

• %cd [path]
• %go111module {on|off}
• %help
• $ shell_command [args...]

Limitations

gophernotes uses gomacro under the hood to evaluate Go code interactively. You can evaluate most any Go code with gomacro, but there are some limitations, which are discussed in further detail here. Most notably, gophernotes does NOT support:

• importing third party packages when running natively on Windows - This is a current limitation of the Go plugin package.
• some corner cases on interpreted interfaces, as interface -> interface type switch and type assertion, are not implemented yet.
• some corner cases on recursive types may not work correctly.
• conversion from typed constant to interpreted interface is not implemented. Workaround: assign the constant to a variable, then convert the variable to the interpreted interface type.
• conversions from/to unsafe.Pointer are not supported.
• goto is only partially implemented.
• out-of-order code in the same cell is supported, but not heavily tested. It has some known limitations for composite literals.

Also, creation of new named types is emulated, and their methods are visible only to interpreted code.

Troubleshooting

gophernotes not found

Depending on your environment, you may need to manually change the path to the gophernotes executable in kernel/kernel.json before copying it to ~/.local/share/jupyter/kernels/gophernotes. You can put the full path to the gophernotes executable here, and you shouldn't have any further issues.

"Kernel error" in a running notebook

Traceback (most recent call last):
  File "/usr/local/lib/python2.7/site-packages/notebook/base/handlers.py", line 458, in wrapper
    result = yield gen.maybe_future(method(self, *args, **kwargs))
  File "/usr/local/lib/python2.7/site-packages/tornado/gen.py", line 1008, in run
    value = future.result()
  ...
  File "/usr/local/Cellar/python/2.7.11/Frameworks/Python.framework/Versions/2.7/lib/python2.7/subprocess.py", line 1335, in _execute_child
    raise child_exception
OSError: [Errno 2] No such file or directory

Stop jupyter, if it's already running.

Add a symlink to /go/bin/gophernotes from your path to the gophernotes executable. If you followed the instructions above, this will be:

sudo ln -s $HOME/go/bin/gophernotes /go/bin/gophernotes

Restart jupyter, and you should now be up and running.

error "could not import C (no metadata for C)" when importing a package

At a first analysis, it seems to be a limitation of the new import mechanism that supports Go modules. You can switch the old (non module-aware) mechanism with the command %go111module off

To re-enable modules support, execute %go111module on

Look at Jupyter notebook's logs for debugging

In order to see the logs for your Jupyter notebook, use the --log-level option

jupyter notebook --log-level DEBUG