A server for executing workbench notebook code fragments

python2.7 server.py --port 8000

This is a temporary interface with some crazy simplifying assumptions:

• There is no authentication or security (so we only listen on localhost!)
• Only one notebook is active at a time
• Only one HTTP request is active at a time

Requests must include a Content-Length header.

Request body: {}

Response body: {}

Request body:

{
  "inputs": [
    "stuff = {}",
    "stuff['k']"
  ]
}

Response body:

{
  "execution_results": [
    {
      "microseconds": 39,
      "result": {"type": "NullValue"}
    },
    {
      "microseconds": 47,
      "result": {
        "type": "ErrorValue",
        "name": "KeyError",
        "message": "k"
      }
    }
  ]
}

server.py
  main

context.py
  Context:        .run_code(), .reset(), maintain locals and globals

stdlib.py
  Video, Image:   these get added to Context's globals

value_repr.py
  value_repr:     Python value -> {"type": "FooValue", ...}

NullValue
StringValue       value
NumberValue       value
BooleanValue      value
ArrayValue        value: [...value representations...]
DictionaryValue   value: {unicode -> value representations}

ReprValue         repr

ImageValue        image_id, width, height
VideoValue        video_id, width, height, first_frame_index, num_frames, fps

InlineImageValue  width, height, data64, ext: "png"

matplotlib is not required.

But if the final line of an evaluated code fragment evaluates to any of the following forms, the figure will be rendered and returned as a high-resolution InlineImageValue:

<matplotlib.figure.Figure object ...>     e.g. from "fig"
<matplotlib.axes.AxesSubplot object ...>  e.g. from "ax"
<matplotlib.text.Text object ...>         e.g. from "ax.set_title('foo')"
[<matplotlib.lines.Line2D object...>]     e.g. from "ax.plot(x, y, 'r')"

Pull requests welcome. This ought to support both, but that's far from the top of my TODO list.

Copyright (c) 2014 Andrew Schaaf and https://github.com/ReclaimSoftware/workbench-kernel contributors.

All contributions from Andrew Schaaf are released under the AGPLv3 (see the license file).

All contributions from other project contributors are released under both the AGPLv3 and the Apache 2 license.

This project might become 100% Apache-licensed in the future, but it hasn't yet.