[bug#50708] gnu: web: Add jupyter-service

Attached is an updated patch.
From d264da0811480a8d2acd5e73f58d320e15bfa9f3 Mon Sep 17 00:00:00 2001
From: Jesse <gib19014@byui.edu>
Date: Mon, 20 Sep 2021 16:01:22 -0600
Subject: [PATCH] gnu: web: Add jupyter-service

* gnu/services/web.scm (gitile-service-type): New variable.
* doc/guix.texi (Version Control Services): Document it.
---
 doc/guix.texi        |  52 +++++++++++++++++++
 gnu/services/web.scm | 120 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 172 insertions(+)

Hello Jesse,

Jesse <jgibbons2357@gmail.com> writes:

> Attached is an updated patch.
>
> From d264da0811480a8d2acd5e73f58d320e15bfa9f3 Mon Sep 17 00:00:00 2001
> From: Jesse <gib19014@byui.edu>
> Date: Mon, 20 Sep 2021 16:01:22 -0600
> Subject: [PATCH] gnu: web: Add jupyter-service
>
> * gnu/services/web.scm (gitile-service-type): New variable.
> * doc/guix.texi (Version Control Services): Document it.
> ---
>  doc/guix.texi        |  52 +++++++++++++++++++
>  gnu/services/web.scm | 120 +++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 172 insertions(+)
>
> diff --git a/doc/guix.texi b/doc/guix.texi
> index 6436e83a7c..0f8aaca413 100644
> --- a/doc/guix.texi
> +++ b/doc/guix.texi
> @@ -26197,6 +26197,58 @@ The file which should store the logging output of Agate.
>  @end table
>  @end deftp
>  
> +@cindex jupyter
> +The Jupyter Notebook is a web application that allows you to create
> +and share documents that contain live code, equations, visualizations
> +and explanatory text.  Uses include: data cleaning and transformation,
> +numerical simulation, statistical modeling, machine learning and much
> +more.
> +
> +@deffn (Scheme Variable) jupyter-service-type
> +This is the type of the agate service, whose value should be an
                           ^
                           Agate

I'm a bit confused by this service referring to Agate; seems it'd be
clearer for the neophyte to simply refer to Jupyter in its place, even
if it's technically an Agate server powering it?

> +@code{jupyter-service-type} object, as in this example:
> +
> +@lisp
> +(service jupyter-service-type
> + (jupyter-configuration
> +  (log-file "/var/log/jupyter.log")
> +  (server-config
> +   (kernels
> +    (list jupyter-guile-kernel)))))
> +
> +@end lisp

I'd put the (list jupyter-guile-kernel) on the same line as (kernels
...)

> +The example above tells the the Jupyter service to provide the
                           ^ typo
                           
> +@code{jupyter-guile-kernel} as an optional kernel.
> +@end deffn
> +
> +@deftp {Data Type} jupyter-configuration
> +Data type representing the configuration of Agate.
> +
> +@table @asis
> +@item @code{jupyter-package} (default: @code{jupyter})
> +The jupyter package to use.
       ^
       Jupyter

I'd capitalize Jupyter in other places as well where it's not decorated
as a command or code, since it's a proper name.

> +
> +@item @code{kernels} (default: @code{'()}
> +A list of kernel packages to use with the jupyter service.
                                             ^ here
> +
> +@item @code{root-dir} (default: @file{"/var/lib/jupyter/notebooks"})
> +The directory where the jupyter server looks for jupyter notebooks.
                                                    ^ and here
> +@item @code{log-file} (default: @code{#f})
> +The location of the log file. If #f is given, no log is produced.
>
> +@item @code{shell} (default: @code{(file-append bash "/bin/bash")}
> +The location of the shell used when starting a web terminal (if it is configured to allow terminals).
> +
> +@item @code{server-config} (default: @code{#f})
> +A file contianing the jupyter server's config file. If @code{#f} is given, an empty file is produced,
          ^ containing

> and the default settings are used.
> +
> +@end table
> +@end deftp
> +
> +
>  @node Certificate Services
>  @subsection Certificate Services
>  
> diff --git a/gnu/services/web.scm b/gnu/services/web.scm
> index 6a093cf4e4..2d2946f7a7 100644
> --- a/gnu/services/web.scm
> +++ b/gnu/services/web.scm
> @@ -15,6 +15,7 @@
>  ;;; Copyright © 2020 Arun Isaac <arunisaac@systemreboot.net>
>  ;;; Copyright © 2020 Oleg Pykhalov <go.wigust@gmail.com>
>  ;;; Copyright © 2020, 2021 Alexandru-Sergiu Marton <brown121407@posteo.ro>
> +;;; Copyright © 2020, 2021 Jesse Gibbons <jgibbons2357+guix@gmail.com>
>  ;;;
>  ;;; This file is part of GNU Guix.
>  ;;;
> @@ -41,11 +42,13 @@
>    #:use-module (gnu system shadow)
>    #:use-module (gnu packages admin)
>    #:use-module (gnu packages base)
> +  #:use-module (gnu packages bash)
>    #:use-module (gnu packages databases)
>    #:use-module (gnu packages web)
>    #:use-module (gnu packages patchutils)
>    #:use-module (gnu packages php)
>    #:use-module (gnu packages python)
> +  #:use-module (gnu packages python-xyz)
>    #:use-module (gnu packages gnupg)
>    #:use-module (gnu packages guile)
>    #:use-module (gnu packages logging)
> @@ -91,6 +94,8 @@
>  
>              httpd-service-type
>  
> +            jupyter-configuration
> +
>              nginx-configuration
>              nginx-configuration?
>              nginx-configuration-nginx
> @@ -1994,3 +1999,118 @@ root=/srv/gemini
>            (service-extension shepherd-root-service-type
>                               agate-shepherd-service)))
>     (default-value (agate-configuration))))
> +
> +
> +;;; Jupyter configuration
> +;;; Jupyter server configuration
> +;;; -> includes a list of kernels to allow in the jupyter server
> +;;; -> includes a list of configuration options specifically for the jupyter server
> +;;; --> These options can be concatenated to a python file
> +;;; --> Nested alist
> +;;; ---> '((NotebookApp (allow-remote-access #t))) -> "c.NotebookApp.allow_remote_access = True"
> +;;; -> Include some settings specifying how to run the daemon
> +;;; --> location of log file
> +;;; --> If a console should be allowed, the location of the shell to be used.
> +;;; --> The package containing the jupyter server itself, default
> jupyter

What is the meaning of the arrows in the above comment?  It seems to me
that they were meant to be used during development, not in the final
result?  Otherwise, the types of the fields could be self-documented and
controlled by the use of the 'define-configuration' procedure and
friends from the (gnu services configuration) module.

> +(define-public default-jupyter-config "#")
> +
> +(define-record-type* <jupyter-configuration> jupyter-configuration
> +                     this-jupyter-configuration jupyter-configuration?
> +                     (jupyter-package jupyter-configuration-jupyter-package
> +                                      (default jupyter))
> +                     (kernels jupyter-configuration-kernels
> +                              (default '()))
> +                     (root-dir jupyter-configuration-root-dir
> +                               (default "/var/lib/jupyter/notebooks"))
> +                     (log-file jupyter-configuration-log-file
> +                               (default #f))
> +                     (shell jupyter-configuration-shell
> +                            (default (file-append bash "/bin/bash")))
> +                     (server-config jupyter-configuration-server-config
> +                                    (default #f))); TODO: Make configuration DSL.
> +
> +
> +(define (search-path-string search-path-pair)
> +  (string-append (search-path-specification-variable (car search-path-pair))
> +                 "="
> +                 (cdr search-path-pair)))
> +
> +;;;TODO: Add actions to list jupyter servers, change passwords, etc.
> +(define (jupyter-shepherd-service config)
> +  (list (shepherd-service
> +          (provision '(jupyter)) ;TODO: Add magic to allow multiple Jupyter servers
> +          (requirement '(loopback))
> +          (start #~(make-forkexec-constructor
> +                     (list "/run/current-system/profile/bin/jupyter"
> +                           "notebook"
> +                           (string-append "--config="
> +                     #$(or (jupyter-configuration-server-config config)
> +                           (plain-file "jupyter_notebook_config.py"
> +                                       default-jupyter-config))
> +                     #$(jupyter-configuration-root-dir config)))
> +                     #:user "jupyter"
> +                     #:group "jupyter"
> +                     #:environment-variables ; TODO use search-paths magic instead of hard-coding these things.
> +                     (list "GI_TYPELIB_PATH=/run/current-system/profile/lib/girepository-1.0"
> +                           "GUILE_LOAD_COMPILED_PATH=/run/current-system/profile/lib/guile/3.0/site-ccache:/run/current-system/profile/share/guile/site/3.0"
> +                           "GUILE_LOAD_PATH=/run/current-system/profile/share/guile/site/3.0"
> +                           "HOME=/var/lib/jupyter"
> +                           "JUPYTER_PATH=/run/current-system/profile/share/jupyter"
> +                           "PATH=/run/current-system/profile/bin:/run/current-system/profile/sbin"
> +                           "PYTHONPATH=/run/current-system/profile/lib/python3.8/site-packages"
> +                           "R_LIBS_SITE=/run/current-system/profile/site-library/"
> +                           "TEXMF=/run/current-system/profile/share/texmf-dist"
> +                           "TEXMFCNF=/run/current-system/profile//share/texmf-dist/web2c"
                                                                    ^
                                                                    extra slash
                                                                    
> +                           "XDG_DATA_DIRS=/run/current-system/profile/share")
> +

Hmm.  Would it work if search paths were set on the 'jupyter' package
instead?  The above is too fragile, with hard coded versions baked in.

> +                     #:directory #$(jupyter-configuration-root-dir config)
> +                     #:log-file #$(jupyter-configuration-log-file config)))
> +          (stop #~(make-kill-destructor))
> +          (documentation "Runs a Jupyter Notebook server. A Jupyter Notebook is
                                                            ^ Use two
                                                            spaces after period.

>a web application that allows you to create and share documents that contain live code, equations, visualizations, and explanatory text."))))
> +
> +(define (jupyter-account config)
> +  (list
> +    (user-account
> +      (name "jupyter")
> +      (group "jupyter")
> +      (comment "Jupyter Notebook Server")
> +      (home-directory "/var/lib/jupyter")
> +      (shell (jupyter-configuration-shell config))
> +      (system? #t))
> +    (user-group
> +      (name "jupyter")
> +      (system? #t))))
> +
> +(define (jupyter-profile config)
> +  (cons*
> +    (jupyter-configuration-jupyter-package config)
> +    (jupyter-configuration-kernels config)))
> +
> +(define (jupyter-activation config)
> +  #~(begin
> +      (let ((root-dir #$(jupyter-configuration-root-dir config))
> +            (pw (getpwnam "jupyter")))
> +        (unless (file-exists? root-dir)
> +          (mkdir root-dir)
> +          (chown root-dir (passwd:uid pw)
> +                 (passwd:gid pw))
> +          (chmod root-dir #o700)))))
> +
> +;;; Jupyter service type
> +;;; -> Information vital to settinng up the server, like the port and accepted parameters
> +;;; -> list of kernels considered permissible.
> +;;; -> a shepherd service extension that runs the jupyter server
> +;;; --> shepherd service to list the running servers, set the password, etc.
> +;;; --> Make a log file only readable by root?
> +;;; -> an accounts service extension describing the user that runs the accounts
> +;;; --> group "jupyter"

Same comment as earlier.

> +(define-public jupyter-service-type
> +  (service-type
> +    (name "jupyter")
> +    (extensions (list
> +                  (service-extension shepherd-root-service-type jupyter-shepherd-service)
> +                  (service-extension account-service-type jupyter-account)
> +                  (service-extension activation-service-type jupyter-activation)
> +                  (service-extension profile-service-type jupyter-profile)))
> +    (description "Runs a Jupyter Notebook server. A Jupyter Notebook is a web application that allows you to create and share documents that contain live code, equations, visualizations, and explanatory text.")
> +    (default-value (jupyter-configuration))))

I haven't tried it, but it looks promising.  If we can find a better
solution than the hard coded version values in the environment variables
list, I think it'd be in good shape for inclusion, although a system
test would be a nice to have.

Thank you!

Maxim