├── README.md
├── assets
└── Plotly-on-Zeppelin-screenie.png
├── offline
├── __init__.py
└── offline.py
└── sample
└── Plot.ly on Zeppelin.json
/README.md:
--------------------------------------------------------------------------------
1 | # Overview
2 |
3 | [Apache Zeppelin](https://zeppelin.incubator.apache.org/) comes with basic charts out of the box and even offers the ability to call into the underlying NVD3 library. Sometimes, though, you just want some [Plot.ly](https://plot.ly/) goodness. While this awesome visualization library can be imported directly within Jupyter, having it work on Zeppelin needs a bit more of work. This repository is one way of making that happen.
4 |
5 | 
6 |
7 | It does two things:
8 | * The Zeppelin web app sources `plotly-latest.min.js` from the CDN.
9 | * It overrides Plot.ly's offline module with the code contained here such that calls to it will render via Zeppelin's display system instead of Jupyter's.
10 |
11 | # Setup
12 |
13 | These steps apply to the node that runs Zeppelin.
14 |
15 | 1. Install the following Python libraries:
16 | * [Plot.ly](https://plot.ly/python/getting-started/)
17 | * [Cufflinks](https://plot.ly/ipython-notebooks/cufflinks/)
18 | * [Pandas](http://pandas.pydata.org/)
19 | * [IPython](https://ipython.org/ipython-doc/2/install/install.html)
20 |
21 | 1. Clone this repository in a commonly accessible path.
22 |
23 | 1. Update `index.html` in zeppelin-web such that it sources `plotly-latest.min.js` during startup. The cleaner way would be to modify Zeppelin's code and then compile. But I assume you have already have Zeppelin set up and you just want to get Plot.ly running on it as fast as possible. In this case, we just extract `index.html` from `zeppelin-web-*.war`, insert the script tag for plotly.js, and then add `index.html` back into the same WAR file.
24 |
25 | 1. Find the `zeppelin-web-*.war` file inside the root folder of your Zeppelin install (i.e. `zeppelin-web-0.6.0-incubating-SNAPSHOT.war`). Back it up somewhere so you can restore the original copy if you need to.
26 | 1. Extract `index.html`
27 |
28 | ```
29 | jar xvf zeppelin-web-0.6.0-incubating-SNAPSHOT.war index.html
30 | ```
31 |
32 | 1. Edit `index.html` to insert `plotly-latest.min.js` script tag just before the `` tag.
33 |
34 | ```
35 |
36 | ```
37 | 1. Insert this modified `index.html` back into the WAR file.
38 |
39 | ```
40 | jar uvf zeppelin-web-0.6.0-incubating-SNAPSHOT.war index.html
41 | ```
42 |
43 | Your modifications to the setup are now done and you can just start or restart your Zeppelin instance. In case it fails to start, you probably mistyped something when editing `index.html`. You can always restore the original WAR file you backed up if all else fails.
44 |
45 | # Usage
46 |
47 | 1. At the top of a Zeppelin notebook, copy and execute the following code. Replace with the actual path where you cloned the repository.
48 |
49 | ```
50 | %pyspark
51 | # Load this module from a non-standard path (non-installed package)
52 | import sys
53 | sys.path.insert(0, "PATH_TO_YOUR_REPO/zeppelin-plotly")
54 |
55 | import offline
56 |
57 | # Override Plotly's module loading to use our patch instead.
58 | sys.modules["plotly"].offline = offline
59 | sys.modules["plotly.offline"] = offline
60 |
61 | # Import cufflinks (Pandas, etc) like usual.
62 | import cufflinks as cf
63 |
64 | # We need to use Cufflinks in offline mode.
65 | cf.go_offline()
66 | ```
67 |
68 | 1. You can then use Plotly and Cufflinks (offline mode) as usual in the suceeding notebook paragraphs.
69 |
70 | ```
71 | %pyspark
72 | import pandas as pd
73 | import numpy as np
74 |
75 | df = pd.DataFrame(np.random.randn(1000, 4), columns=['a', 'b', 'c', 'd'])
76 | df.scatter_matrix()
77 | ```
78 |
79 | Check out the showcase notebook included inside the `sample` folder if you want to see the different visualizations available on Plot.ly. Just import the `json` file like how you would any other Zeppelin notebookand then point the `sys.path.insert` bit to the actual location of your cloned repo.
80 |
81 | # TODO
82 |
83 | * Maybe contribute this back to the Plot.ly and/or Zeppelin projects so that integration becomes even easier.
84 | * Write tests
85 |
86 |
--------------------------------------------------------------------------------
/assets/Plotly-on-Zeppelin-screenie.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/beljun/zeppelin-plotly/8ed116738a1aa56d138378c20fc1318dae83d325/assets/Plotly-on-Zeppelin-screenie.png
--------------------------------------------------------------------------------
/offline/__init__.py:
--------------------------------------------------------------------------------
1 | """
2 | offline
3 | ======
4 | This module provides offline functionality.
5 | """
6 | from . offline import (
7 | init_notebook_mode,
8 | iplot
9 | )
10 |
11 |
--------------------------------------------------------------------------------
/offline/offline.py:
--------------------------------------------------------------------------------
1 | """ Plotly Offline
2 | A module to use Plotly's graphing library with Python
3 | without connecting to a public or private plotly enterprise
4 | server.
5 | """
6 | from __future__ import absolute_import
7 |
8 | import json
9 | import os
10 | import uuid
11 | import warnings
12 | from pkg_resources import resource_string
13 |
14 | import plotly
15 | from plotly import tools, utils
16 | from plotly.exceptions import PlotlyError
17 |
18 |
19 | __PLOTLY_OFFLINE_INITIALIZED = True
20 |
21 | def get_plotlyjs():
22 | path = os.path.join('offline', 'plotly.min.js')
23 | plotlyjs = resource_string('plotly', path).decode('utf-8')
24 | return plotlyjs
25 |
26 | def init_notebook_mode():
27 | pass
28 |
29 | def iplot(figure_or_data, show_link=False, link_text='',
30 | validate=True):
31 | """
32 | Draw plotly graphs inside a Zeppelin notebook without
33 | connecting to an external server.
34 | To save the chart to Plotly Cloud or Plotly Enterprise, use
35 | `plotly.plotly.iplot`.
36 | To embed an image of the chart, use `plotly.image.ishow`.
37 |
38 | figure_or_data -- a plotly.graph_objs.Figure or plotly.graph_objs.Data or
39 | dict or list that describes a Plotly graph.
40 | See https://plot.ly/python/ for examples of
41 | graph descriptions.
42 |
43 | Keyword arguments:
44 | show_link (default=True) -- display a link in the bottom-right corner of
45 | of the chart that will export the chart to
46 | Plotly Cloud or Plotly Enterprise
47 | link_text (default='Export to plot.ly') -- the text of export link
48 | validate (default=True) -- validate that all of the keys in the figure
49 | are valid? omit if your version of plotly.js
50 | has become outdated with your version of
51 | graph_reference.json or if you need to include
52 | extra, unnecessary keys in your figure.
53 |
54 | Example:
55 | ```
56 | from plotly.offline import init_notebook_mode, iplot
57 | init_notebook_mode()
58 |
59 | iplot([{'x': [1, 2, 3], 'y': [5, 2, 7]}])
60 | ```
61 | """
62 |
63 | figure = tools.return_figure_from_figure_or_data(figure_or_data, validate)
64 |
65 | width = figure.get('layout', {}).get('width', '100%')
66 | height = figure.get('layout', {}).get('height', 525)
67 | try:
68 | float(width)
69 | except (ValueError, TypeError):
70 | pass
71 | else:
72 | width = str(width) + 'px'
73 |
74 | try:
75 | float(width)
76 | except (ValueError, TypeError):
77 | pass
78 | else:
79 | width = str(width) + 'px'
80 |
81 | plotdivid = uuid.uuid4()
82 | jdata = json.dumps(figure.get('data', []), cls=utils.PlotlyJSONEncoder)
83 | jlayout = json.dumps(figure.get('layout', {}), cls=utils.PlotlyJSONEncoder)
84 |
85 | config = {}
86 | config['showLink'] = show_link
87 | config['linkText'] = link_text
88 | config['displayModeBar'] = False
89 | jconfig = json.dumps(config)
90 |
91 | script = '\n'.join([
92 | 'Plotly.plot("{id}", {data}, {layout}, {config}).then(function() {{',
93 | ' $(".{id}.loading").remove();',
94 | '}})'
95 | ]).format(id=plotdivid,
96 | data=jdata,
97 | layout=jlayout,
98 | config=jconfig)
99 |
100 | print("""%html
101 |