mbrukman / scruffy-server Goto Github PK

scruffy-server is a micro web server front-end for Scruffy: A UML class/sequence diagram generator from human text, like yUML, with a very clean interface.

Features:

• UML Class diagram
• UML Sequence diagram
• PNG / SVG export
• Simple and fast interface
• Works also without JavaScript
• Mobile friendly

If you have Docker installed:

$ docker run -d -p 8080:8080 wernight/scruffy-server

And browse to http://localhost:8080/

That Dockerize container is:

• Small: Using [Debian image][debian] is below 100 MB (while Ubuntu is about 230 MB), and removing build packages.
• Simple: Exposes default port, easy to extend.
• Secure: Runs as non-root UID/GID 35726 (selected randomly to avoid mapping to an existing user).

You can deploy on Kubernetes (for example on Google Container Engine), once you've kubectl and a running cluster:

$ kubectl run-container scruffy --image=wernight/scruffy-server:latest --port=80
$ kubectl expose rc scruffy --port=80 --target-port=8080 --create-external-load-balancer

Wait may be a minute or so and you should have a public IP (the second one here):

$ kubectl get svc scruffy
NAME      LABELS        SELECTOR      IP(S)           PORT(S)
scruffy   run=scruffy   run=scruffy   10.255.255.10   80/TCP
                                      104.123.45.67
$ xdg-open http://104.123.45.67

Be aware that there is no ACL, no spam check, and no caching (i.e. anyone can access by default and evil doers might use all your CPU). So you probably would want it only on a private network or proxied behind a password protected URL. For example you could put it behind Nginx as reverse proxy, with a basic authentication.

1. Install Scruffy pre-requisites:

   • On Ubuntu Linux you'd do: $ sudo apt-get install python-dev python-setuptools graphviz plotutils librsvg2-bin
   • On Arch Linux you'd do: $ pacman -S graphviz python2 plotutils librsvg
   • In general: You'll need Python, dot, libRSVG binaries, pic2plot, and Python Imaging Library (PIL) or Python Pillow.

2. Run Scruffy-Server:

   $ git clone https://github.com/wernight/scruffy-server.git
   $ cd scruffy-server
   $ virtualenv ENV --system-site-packages && source ENV/bin/activate    # (optional)
   $ pip install -r requirements.txt
   $ python server.py
   

3. Browse http://localhost:8080/

Edit the end of server.py to change the port or IP binding.

See also Bottle Deployment.

• js-sequence-diagrams using JavaScript only.
• yUML Which is very similar commercial alternative, with Use Case, Activity, and Class diagram support (and not a so great mobile support).
• PlantUML An OpenSource Java solution, also text based which supports most UML diagrams with a more classic look (and not a so great mobile support).
• nomnoml A UML class diagram using JavaScript (only).
• draw.io and Lucidchart are also online solutions but not auto-generated from text.

scruffy-server is good for lean short UML diagrams even from a mobile phone or tablet.

The only real risk is that your server isn't well sandboxed and a user sends malicious UML code.

For users however there is no real need for password protection because the URL shared is the entire UML diagram source code. This means that no one can try to guess your UML by trying random URLs (as it would only generate all possible UMLs), and any change to an existing UML will generate a new URL. So you can generate a diagram, share it, without ever needing a password. Just remember: URL = UML.

Check that the user running server.py can execute Scruffy suml command.

Chech Arch Fonts - Pango Warnings. You may want to install ttf-tlwg to have Purisa and a more scruffy look (a bit like Comic Sans).

You can also add ..., '--font-family', 'Purisa', ... to suml parameters, see "How to change the UML font and style" question below.

If you just want to use another font, you can set environement variable SCRUFFY_FONT to the font family you'd like to use.

For a full list of possible settings, execute $ suml --help to find what is allowed and change the check_output(...) parameters in server.py.

Improvement ideas and pull requests are welcome via Github Issue Tracker.