GraphDash 0.11.1

A web-based dashboard built from graphs and their metadata

GraphDash is a web-based dashboard built on graphs and their metadata. For example, if you have two graphs in a directory:

$ cd default_graph_dir
$ ls
graph.svg graph2.svg

Then you can create two metadata files using YAML format, where you can configure how the graphs will be displayed:

$ cat graph.yaml
name: graph.svg
family: 'Category 1'
title: '*Real serious* graph'
text: |
    The description

$ cat graph2.yaml
name: graph2.svg
family: 'Category 2'
title: 'Another important graph'

You may then start the graph dashboard. You will get a nice web interface displaying your graphs, and a search box with autocompletion. You can easily navigate and share your graphs.

$ GraphDash --root .
* Running on http://0.0.0.0:5555/ (Press CTRL+C to quit)

Installation

Clone and install (in user space):

git clone https://github.com/AmadeusITGroup/graphdash.git
cd graphdash
pip install --user .

Or use the Python package:

pip install --user graphdash

Launch the webapp

For user-space installation, make sure your $PATH includes ~/.local/bin.

$ GraphDash -r default_graph_dir
* Running on http://0.0.0.0:5555/ (Press CTRL+C to quit)

The dashboard can be configured with a YAML config file and the -c/--conf option:

$ cat docs/graphdash.yaml
root: ../default_graph_dir
title: "Example of title ;)"
subtitle: "Example of subtitle"

$ GraphDash -c docs/graphdash.yaml
* Running on http://0.0.0.0:5555/ (Press CTRL+C to quit)

You can generate a template of configuration file:

$ GraphDash -C template.yaml

Serve with Gunicorn

If not already installed on your machine, install Gunicorn:

pip install --user gunicorn setproctitle  # on Fedora you may need to install libffi-devel before

Since you can import the webapp through graphdash:app, you can serve it with Gunicorn:

gunicorn -b 0.0.0.0:8888 --pid server.pid graphdash:app

The configuration file of the webapp can be set with the CONF environment variable. With Gunicorn, you can pass environment variables to the workers with --env:

gunicorn -b 0.0.0.0:8888 --pid server.pid --env CONF=docs/graphdash.yaml graphdash:app

But you should not use these commands yourself, that is what GraphDashManage is for!

GraphDashManage

GraphDashManage is used to start, stop, restart the instances of Gunicorn serving graphdash:app. It needs a configuration file in the current directory:

$ cat settings.sh
ALL_MODES=(
   ['prod']="docs/graphdash.yaml"
   ['test']="docs/graphdash.yaml"
)
ALL_PORTS=(
   ['prod']=1234
   ['test']=5678
)
WORKERS=3

Then you can manage multiple instances of GraphDash using Gunicorn with:

$ GraphDashManage start prod
[INFO] Listening at: http://0.0.0.0:1234
[INFO] Booting worker with pid: 30403
[INFO] Booting worker with pid: 30404
[INFO] Booting worker with pid: 30405

$ GraphDashManage start test
[INFO] Listening at: http://0.0.0.0:5678
...

You can generate a template of settings:

$ GraphDashManage template > template.sh # to be moved to settings.sh

Webapp configuration file

Possible entries (everything is optional):

• root: the root directory of the graphs

• families: path to the families metadata file (optional)

• title: the title of the webapp

• subtitle: the subtitle of the webapp

• placeholder: the default text in the search field

• header: an optional message at the top (markdown syntax)

• footer: an optional message at the bottom (markdown syntax)

• showfamilynumbers: a boolean to toggle family numbering (default is true)

• showgraphnumbers: a boolean to toggle graph numbering (default is true)

• theme: change css theme (default is dark)

• keep: the proportion of common words kept for autocompletion

• logfile: change default log file of the webapp

• raw: when loading, look for all graphs and ignore metadata

• verbose: a boolean indicating verbosity when loading application

• debug: debug mode (enable Grunt livereload, enable Flask debug mode)

• headless: headless mode (only search is available, no page is rendered)

• port: when launched with Flask development server only, port

Graph metadata

Several attributes are supported:

• name: the path to the graph

• title: title of the graph, recommended for display purposes (markdown syntax)

• family: the subsection in which the graph is

• index: an optional list of keywords describing the graph (useful for search feature)

• text: an optional description of the graph (markdown syntax)

• pretext: an optional message appearing before the graph (markdown syntax)

• file: optional path to the raw data

• export: optional path to the exportable graph (for example, a PNG file)

• rank: integer, optional value used to change graphs order (default uses titles)

• showtitle: a boolean to toggle title display for the graph (default is false)

• labels: a list of labels (like 'new') which will be rendered in the UI as colored circles

• other: other metadata not used by GraphDash, but may be needed by other things reading the metadata

Note that if the name attribute is missing, the graph will not be shown and the text will be displayed anyway, like a blog entry.

Family metadata

You may put a .FAMILIES.yaml file at the root of the graph directory. This file may contain metadata for families. It should be a YAML list:

- family: chairs
  rank : 0
- family: tables
  rank : 1
  text: This is a description
  alias: This text will appear instead of "tables"
  labels: new

Each element of the list should be a dict containing:

• family: the family considered

• rank: integer, optional value used to change families order (default uses family name)

• text: an optional description of the family (markdown syntax)

• alias: an optional name who may be longer than the one in the url (useful to build nice urls)

• labels: a list of labels (like new) which will be rendered in the UI as colored circles

Available labels are new, update, bugfix, warning, error, ongoing, obsolete. You may give other labels which will be rendered with defaults colors. For customization, you may specify your own labels with a dict syntax:

labels:
- name: newlabel
  color: white
  text_color: black
  text: "NEW LABEL"
  tooltip: null

Development

If you wish to contribute, you need Grunt to generate new css/js files from sass/coffee source files.

npm install --no-bin-links # may need to repeat
grunt

Debugging can be made with source map files for browser supporting them in their debugging tools. If not, the Gruntfile.js enables an option to generate non-minified assets.

grunt --dev

With the debug mode enabled, Grunt will use the livereload mechanism to reload the browser if any file has changed (and Flask debug mode will reload the server as well).

GraphDash --debug & # or python -m graphdash
grunt watch

If you used Gunicorn with a PID file, Grunt will automatically reload it if any Python files change.

gunicorn -b 0.0.0.0:8888 --pid server.pid graphdash:app &
grunt watch

You can use tox build packages and run tests.

tox