1
00:00:00,580 --> 00:00:02,450
- [Instructor] For our last application

2
00:00:02,450 --> 00:00:06,240
demonstrating IoT pub/sub capabilities,

3
00:00:06,240 --> 00:00:10,150
we're going to go back to
working with the PubNub service.

4
00:00:10,150 --> 00:00:13,060
And in this case we're
going to take advantage

5
00:00:13,060 --> 00:00:16,320
of their pubnub module,
which is a Python module

6
00:00:16,320 --> 00:00:19,610
for interacting with the PubNub service.

7
00:00:19,610 --> 00:00:21,030
And we're going to be using that

8
00:00:21,030 --> 00:00:25,450
to visualize data from their
simulated Market Orders stream.

9
00:00:25,450 --> 00:00:29,170
And that Market Orders stream
has fake stock information

10
00:00:29,170 --> 00:00:32,110
for several different companies.

11
00:00:32,110 --> 00:00:37,110
Now, you can also use the pubnub
module to publish messages

12
00:00:38,270 --> 00:00:40,020
through the PubNub service.

13
00:00:40,020 --> 00:00:41,310
We don't cover that here,

14
00:00:41,310 --> 00:00:44,410
but this link shows you the documentation

15
00:00:44,410 --> 00:00:46,690
for the pubnub Python SDK.

16
00:00:46,690 --> 00:00:50,180
And it describes both,
subscribing to existing streams,

17
00:00:50,180 --> 00:00:54,750
and creating your owns streams
to publish messages as well.

18
00:00:54,750 --> 00:00:56,620
In order to use the pubnub module

19
00:00:56,620 --> 00:00:58,160
you will need to install it.

20
00:00:58,160 --> 00:01:00,200
So, this is the installation command.

21
00:01:00,200 --> 00:01:05,050
And we've provided a
stocklistener.py file, or script,

22
00:01:05,050 --> 00:01:06,680
that is going to subscribe

23
00:01:06,680 --> 00:01:09,230
to that simulated Market Orders stream,

24
00:01:09,230 --> 00:01:14,230
and visualize the stock prices
as they change in a bar plot.

25
00:01:14,330 --> 00:01:17,140
Now, the PubNub client that
we're going to be working with

26
00:01:17,140 --> 00:01:19,100
is going to give us back JSON objects

27
00:01:19,100 --> 00:01:21,770
in the form of Python dictionaries,

28
00:01:21,770 --> 00:01:24,900
and we'll extract the
information we need from that,

29
00:01:24,900 --> 00:01:28,760
and then go ahead and visualize that data.

30
00:01:28,760 --> 00:01:30,720
Now, before we actually get into

31
00:01:30,720 --> 00:01:33,110
the source code for this application,

32
00:01:33,110 --> 00:01:37,180
let's run our script and
demonstrate it in action.

33
00:01:37,180 --> 00:01:39,210
So, we set up this script

34
00:01:39,210 --> 00:01:41,350
with the command-line
argument that specifies

35
00:01:41,350 --> 00:01:44,660
the total number of
messages you wish to receive

36
00:01:44,660 --> 00:01:46,010
from the PubNub service.

37
00:01:46,010 --> 00:01:49,640
So we'll just do 1,000 for
demonstration purposes here.

38
00:01:49,640 --> 00:01:51,970
And as you'll see in a moment,

39
00:01:51,970 --> 00:01:54,750
it will pop open a matplotlib window

40
00:01:54,750 --> 00:01:56,840
with a seaborn bar plot.

41
00:01:56,840 --> 00:02:00,740
And as the stock prices are coming in,

42
00:02:00,740 --> 00:02:02,690
we're displaying the name of the company

43
00:02:02,690 --> 00:02:05,370
and the current stock
price at the command line,

44
00:02:05,370 --> 00:02:08,820
but we're also updating
a dictionary in memory,

45
00:02:08,820 --> 00:02:11,680
and that dictionary's data is being used

46
00:02:11,680 --> 00:02:14,500
to visualize the stock prices here

47
00:02:14,500 --> 00:02:16,110
in the context of the bar plot.

48
00:02:16,110 --> 00:02:18,360
So, the coloring of the bar plot

49
00:02:18,360 --> 00:02:20,800
is not important in this case,

50
00:02:20,800 --> 00:02:22,640
it's just for visual effect.

51
00:02:22,640 --> 00:02:24,500
But the nice thing that we're getting here

52
00:02:24,500 --> 00:02:28,990
is the dynamic, animated
presentation of how those values

53
00:02:28,990 --> 00:02:31,870
are changing with respect to one another.

54
00:02:31,870 --> 00:02:35,990
And notice, by the way, that
the y-axis is auto-scaling

55
00:02:35,990 --> 00:02:39,740
based on the current values
of all the different stocks.

56
00:02:39,740 --> 00:02:44,090
So, the top end of the y-axis
is always going to have to do

57
00:02:44,090 --> 00:02:46,270
with the value of the stock

58
00:02:46,270 --> 00:02:48,703
that has the current highest price.

59
00:02:49,990 --> 00:02:52,330
This is the PubNub webpage

60
00:02:52,330 --> 00:02:55,017
for their simulated Market Orders stream.

61
00:02:55,017 --> 00:02:57,870
And as you can see when
you go to this webpage,

62
00:02:57,870 --> 00:02:59,270
it is going to show you

63
00:02:59,270 --> 00:03:02,890
simulated prices changing over time here.

64
00:03:02,890 --> 00:03:06,060
And if we go and scroll
down further on the page,

65
00:03:06,060 --> 00:03:07,960
you'll see that it tells you the name

66
00:03:07,960 --> 00:03:09,950
of the channel that you need to use,

67
00:03:09,950 --> 00:03:12,220
and the subscription key
that you need to use.

68
00:03:12,220 --> 00:03:15,320
And we'll need both of those
in our source code in a moment.

69
00:03:15,320 --> 00:03:17,550
And it also shows you the format

70
00:03:17,550 --> 00:03:21,380
of the JavaScript Object Notation objects

71
00:03:21,380 --> 00:03:23,360
that represent each message.

72
00:03:23,360 --> 00:03:26,640
So, every message we receive
has these five keys in it,

73
00:03:26,640 --> 00:03:29,220
but we're actually only
going to take advantage,

74
00:03:29,220 --> 00:03:32,290
in this app, of the
symbol, and the bid price

75
00:03:32,290 --> 00:03:35,920
so that we can visualize
the stock price changes

76
00:03:35,920 --> 00:03:37,680
as they occur.

77
00:03:37,680 --> 00:03:39,660
So now that we've seen the animation,

78
00:03:39,660 --> 00:03:41,180
let's take a look at the script

79
00:03:41,180 --> 00:03:43,360
that implements the animation.

80
00:03:43,360 --> 00:03:44,910
Now, we've seen these imports

81
00:03:44,910 --> 00:03:47,060
in lines three through seven previously.

82
00:03:47,060 --> 00:03:49,890
Lines nine through 12
import the various types

83
00:03:49,890 --> 00:03:53,280
that we're going to need
from the pubnub module.

84
00:03:53,280 --> 00:03:55,120
The first type here is a class

85
00:03:55,120 --> 00:03:57,320
that we'll be creating a subclass of

86
00:03:57,320 --> 00:04:00,050
to listen for messages from PubNub.

87
00:04:00,050 --> 00:04:02,540
We'll get two different
kinds of messages from PubNub

88
00:04:02,540 --> 00:04:03,740
in this example.

89
00:04:03,740 --> 00:04:06,360
Administrative messages
that have a constant

90
00:04:06,360 --> 00:04:09,390
in this PNStatusCategory type,

91
00:04:09,390 --> 00:04:12,570
and also messages from
the live stream itself.

92
00:04:12,570 --> 00:04:14,830
We'll use the PNConfiguration object

93
00:04:14,830 --> 00:04:18,060
to specify the subscription key

94
00:04:18,060 --> 00:04:20,900
that's required for connecting to a stream

95
00:04:20,900 --> 00:04:22,670
in the context of PubNub.

96
00:04:22,670 --> 00:04:25,750
And the PubNub client
will not only be used

97
00:04:25,750 --> 00:04:27,320
to register a listener

98
00:04:27,320 --> 00:04:29,900
for all the different
messages that we'll receive,

99
00:04:29,900 --> 00:04:32,600
but also to subscribe to the stream

100
00:04:32,600 --> 00:04:36,500
to start the flow of
messages into our app.

101
00:04:36,500 --> 00:04:37,930
Now, this list right here

102
00:04:37,930 --> 00:04:40,320
contains the names of the
five different companies

103
00:04:40,320 --> 00:04:44,800
we'll be receiving from the
simulated Market Orders stream.

104
00:04:44,800 --> 00:04:48,410
And we use that list to help
create a data-frame here,

105
00:04:48,410 --> 00:04:50,520
where we'll store the
names of the companies,

106
00:04:50,520 --> 00:04:54,120
and their most recent
stock price in each case.

107
00:04:54,120 --> 00:04:58,220
So, we're using a dictionary
to initialize this data frame.

108
00:04:58,220 --> 00:05:00,460
And remember, when you use a dictionary

109
00:05:00,460 --> 00:05:02,850
the keys become the column names,

110
00:05:02,850 --> 00:05:05,350
and the values associated with those keys

111
00:05:05,350 --> 00:05:08,660
are used to populate the
rows within a given column.

112
00:05:08,660 --> 00:05:11,050
So, the company column will have the names

113
00:05:11,050 --> 00:05:13,390
of all the companies
that we defined up above.

114
00:05:13,390 --> 00:05:16,090
And the price column
will initially have zeros

115
00:05:16,090 --> 00:05:18,630
for each of those companies.

116
00:05:18,630 --> 00:05:21,180
Now, the most important
piece in this example

117
00:05:21,180 --> 00:05:24,490
is our subclass of SubscribeCallback.

118
00:05:24,490 --> 00:05:27,440
In that subclass, we've
defined an __init__ method,

119
00:05:27,440 --> 00:05:29,210
which is going to receive the data frame

120
00:05:29,210 --> 00:05:31,810
containing the names of the companies

121
00:05:31,810 --> 00:05:33,370
and their current stock prices.

122
00:05:33,370 --> 00:05:36,580
And we'll keep updating
that as we receive messages.

123
00:05:36,580 --> 00:05:40,670
And limit=1000, this is the
total number of messages

124
00:05:40,670 --> 00:05:44,630
we wish to receive before we
unsubscribe from the stream

125
00:05:44,630 --> 00:05:47,430
in this demonstration example.

126
00:05:47,430 --> 00:05:50,350
We store the data frame and
keep track of the order count,

127
00:05:50,350 --> 00:05:51,870
because we need to be able to check

128
00:05:51,870 --> 00:05:56,190
whether we've reached the
maximum number of orders as well.

129
00:05:56,190 --> 00:05:58,250
Now, we've overwritten
two different methods

130
00:05:58,250 --> 00:05:59,790
in our subclass.

131
00:05:59,790 --> 00:06:01,570
One is the status method,

132
00:06:01,570 --> 00:06:03,820
which will receive administrative messages

133
00:06:03,820 --> 00:06:05,840
from the PubNub service.

134
00:06:05,840 --> 00:06:07,470
And we're looking in this example

135
00:06:07,470 --> 00:06:10,210
for two different status messages.

136
00:06:10,210 --> 00:06:12,050
You'll notice that the status object

137
00:06:12,050 --> 00:06:16,830
we receive as a parameter
has a category attribute.

138
00:06:16,830 --> 00:06:20,250
And that category attribute
will be one of the constants

139
00:06:20,250 --> 00:06:23,050
in the PNStatusCategory type.

140
00:06:23,050 --> 00:06:25,590
So if we get this type of category,

141
00:06:25,590 --> 00:06:28,190
that means we've subscribed to a stream.

142
00:06:28,190 --> 00:06:31,360
And if we get this one it
means that we've unsubscribed.

143
00:06:31,360 --> 00:06:33,820
And in each case, we'll display a message

144
00:06:33,820 --> 00:06:35,340
at the command line.

145
00:06:35,340 --> 00:06:38,050
This is the key method
for receiving messages

146
00:06:38,050 --> 00:06:39,820
from the live stream itself.

147
00:06:39,820 --> 00:06:43,280
It receives the client object,
the pubnub client object,

148
00:06:43,280 --> 00:06:45,980
and the the particular
message from the stream.

149
00:06:45,980 --> 00:06:47,810
Within that message parameter

150
00:06:47,810 --> 00:06:49,780
there is a dictionary called message

151
00:06:49,780 --> 00:06:52,680
that has the dictionary representation

152
00:06:52,680 --> 00:06:55,920
of the JSON object that was sent to us.

153
00:06:55,920 --> 00:06:59,140
And within that JSON
object there's a symbol key

154
00:06:59,140 --> 00:07:01,910
and a bid_price key, as
well as several other keys.

155
00:07:01,910 --> 00:07:04,970
We only need these two in this example.

156
00:07:04,970 --> 00:07:07,290
So, the symbol and bid_price are captured

157
00:07:07,290 --> 00:07:09,170
and then displayed at the command line.

158
00:07:09,170 --> 00:07:12,870
Then we update at the
particular row and column

159
00:07:12,870 --> 00:07:15,430
within our data frame, the bid_price.

160
00:07:15,430 --> 00:07:19,080
So, you'll notice we
figure out the symbol's row

161
00:07:19,080 --> 00:07:22,210
by passing it to the data
frame's index method,

162
00:07:22,210 --> 00:07:25,520
which looks for that symbol in the index,

163
00:07:25,520 --> 00:07:28,070
and then figures out the current row,

164
00:07:28,070 --> 00:07:29,660
or the appropriate row.

165
00:07:29,660 --> 00:07:32,720
And based on that, we update
the corresponding column

166
00:07:32,720 --> 00:07:34,340
with that bid_price.

167
00:07:34,340 --> 00:07:36,580
We also add one to the order count.

168
00:07:36,580 --> 00:07:39,330
And if we've reached the
maximum number of orders,

169
00:07:39,330 --> 00:07:42,320
we tell the PubNub client to unsubscribe

170
00:07:42,320 --> 00:07:45,830
from all the different
streams that we subscribed to.

171
00:07:45,830 --> 00:07:48,590
In this example, there's only one channel

172
00:07:48,590 --> 00:07:52,100
that we will subscribe to,
the Market Orders live stream.

173
00:07:52,100 --> 00:07:54,130
So, if we had multiples,

174
00:07:54,130 --> 00:07:56,270
this would unsubscribe from all of them.

175
00:07:56,270 --> 00:07:58,610
And of course, they also
give you the ability

176
00:07:58,610 --> 00:08:01,323
to unsubscribe from one as well.

177
00:08:02,330 --> 00:08:03,970
Our update method just defines

178
00:08:03,970 --> 00:08:05,220
what's going to be displayed

179
00:08:05,220 --> 00:08:07,090
in every frame of the animation.

180
00:08:07,090 --> 00:08:10,380
And we've done similar things
to this multiple times now.

181
00:08:10,380 --> 00:08:14,060
But just as a reminder, this
clears the current bar plot

182
00:08:14,060 --> 00:08:16,810
from the figure object, from the axes.

183
00:08:16,810 --> 00:08:18,960
cla stands for clear axes.

184
00:08:18,960 --> 00:08:21,240
This creates a new bar plot

185
00:08:21,240 --> 00:08:24,040
with the current data from
the companies data frame,

186
00:08:24,040 --> 00:08:26,240
using the x-axis to plot the companies,

187
00:08:26,240 --> 00:08:29,010
and the y-axis to plot the prices.

188
00:08:29,010 --> 00:08:31,010
Then we set the x and y labels.

189
00:08:31,010 --> 00:08:32,550
And we use a tight_layout

190
00:08:32,550 --> 00:08:34,920
to make sure that everything fits nicely

191
00:08:34,920 --> 00:08:37,910
within the figure window.

192
00:08:37,910 --> 00:08:41,123
And then finally, we have the
main part of our application,

193
00:08:42,090 --> 00:08:46,470
in which we set the background
of our seaborn plot.

194
00:08:46,470 --> 00:08:47,760
We create the window

195
00:08:47,760 --> 00:08:50,140
in which the animation will be displayed.

196
00:08:50,140 --> 00:08:51,980
We configure the animation,

197
00:08:51,980 --> 00:08:53,940
which we've done a number
of times previously,

198
00:08:53,940 --> 00:08:57,030
but again, remember, it
requires the figure object

199
00:08:57,030 --> 00:09:00,200
that specifies where to
display the animation frames,

200
00:09:00,200 --> 00:09:01,610
and the method to call

201
00:09:01,610 --> 00:09:04,580
to create those animation frames as well.

202
00:09:04,580 --> 00:09:08,560
This version of plt.show,
with the block=False argument,

203
00:09:08,560 --> 00:09:10,460
says to display the window,

204
00:09:10,460 --> 00:09:14,580
but don't block and wait for
that window to be closed.

205
00:09:14,580 --> 00:09:18,880
Move on to the next statement
in the example, in this case.

206
00:09:18,880 --> 00:09:21,750
And we'll call plt.show again later

207
00:09:21,750 --> 00:09:24,700
to keep the window on the screen.

208
00:09:24,700 --> 00:09:28,210
config, we create the
PNConfiguration object.

209
00:09:28,210 --> 00:09:30,550
That's where we store
the subscription key.

210
00:09:30,550 --> 00:09:31,780
And in this case,

211
00:09:31,780 --> 00:09:35,060
we did put the subscription
key directly into our code,

212
00:09:35,060 --> 00:09:37,180
but that's not normally what we would do.

213
00:09:37,180 --> 00:09:39,740
The only reason we chose to do that here

214
00:09:39,740 --> 00:09:41,860
is because this is a public key

215
00:09:41,860 --> 00:09:44,000
that they publish on their website

216
00:09:44,000 --> 00:09:45,360
for you to play around with

217
00:09:45,360 --> 00:09:48,430
and learn about streaming with PubNub.

218
00:09:48,430 --> 00:09:49,263
We,

219
00:09:49,263 --> 00:09:52,130
can create, excuse me,
the PubNub client object,

220
00:09:52,130 --> 00:09:54,150
by passing the configuration

221
00:09:54,150 --> 00:09:57,010
into the constructor
for the PubNub client.

222
00:09:57,010 --> 00:09:58,810
Once we have the pubnub object,

223
00:09:58,810 --> 00:10:02,640
we register a listener of
our StockListener type,

224
00:10:02,640 --> 00:10:04,900
giving it the data frame that specifies

225
00:10:04,900 --> 00:10:09,140
the location in which we'll
update the stock prices.

226
00:10:09,140 --> 00:10:13,320
And separately, the total
number of orders to listen to,

227
00:10:13,320 --> 00:10:16,326
and we get that as a
command line argument.

228
00:10:16,326 --> 00:10:20,020
And if it's not a correct
command line argument,

229
00:10:20,020 --> 00:10:23,120
we'll use the number 1,000 by default.

230
00:10:23,120 --> 00:10:26,630
So, the last step is to actually
subscribe to the channel.

231
00:10:26,630 --> 00:10:28,720
And as you can see here,

232
00:10:28,720 --> 00:10:31,930
we call the pubnub
client's subscribe method,

233
00:10:31,930 --> 00:10:35,750
and then we access the resulting
object's channels method

234
00:10:35,750 --> 00:10:39,610
to specify the name of the
channel we wish to subscribe to.

235
00:10:39,610 --> 00:10:42,580
And this, again, was
published on the PubNub page.

236
00:10:42,580 --> 00:10:44,170
And then we call execute,

237
00:10:44,170 --> 00:10:46,880
at which point we now
subscribe to the channel

238
00:10:46,880 --> 00:10:49,040
and begin receiving messages.

239
00:10:49,040 --> 00:10:53,600
Now, if we did not have
this last call to plt.show,

240
00:10:53,600 --> 00:10:55,310
the program would actually terminate

241
00:10:55,310 --> 00:10:57,500
and the window would simply go away.

242
00:10:57,500 --> 00:10:59,760
In this case, calling
show with no arguments

243
00:10:59,760 --> 00:11:01,650
keeps that window on the screen,

244
00:11:01,650 --> 00:11:04,753
and that's what enables
you to see the animation.