1
00:00:00,710 --> 00:00:02,210
- [Instructor] Next up
we're going to show you

2
00:00:02,210 --> 00:00:04,830
how to search through recent tweets,

3
00:00:04,830 --> 00:00:08,990
so again, Twitter makes
available up to the last

4
00:00:08,990 --> 00:00:12,470
seven days worth of tweets
for a given account,

5
00:00:12,470 --> 00:00:14,880
and you can use the API method search

6
00:00:14,880 --> 00:00:18,240
to search for specific query strings

7
00:00:18,240 --> 00:00:20,820
that you supply as an argument.

8
00:00:20,820 --> 00:00:22,890
It's not guaranteed to give you back

9
00:00:22,890 --> 00:00:25,320
all matching tweets, just so you know,

10
00:00:25,320 --> 00:00:28,870
so you will get what is
effectively a sampling of tweets

11
00:00:28,870 --> 00:00:33,520
containing those query
terms that you specify.

12
00:00:33,520 --> 00:00:36,090
You're going to be
using the search method,

13
00:00:36,090 --> 00:00:37,410
which in turn is going to call

14
00:00:37,410 --> 00:00:40,480
the Twitter Search API's tweets method,

15
00:00:40,480 --> 00:00:42,030
and that is going to enable you

16
00:00:42,030 --> 00:00:45,030
to get 15 tweets at a time by default,

17
00:00:45,030 --> 00:00:49,410
but you can get up to
100 of those at a time.

18
00:00:49,410 --> 00:00:51,370
Now, of course if you're going

19
00:00:51,370 --> 00:00:53,970
to be requesting a lot of results,

20
00:00:53,970 --> 00:00:57,570
then you're going to want to
use a cursor object as well,

21
00:00:57,570 --> 00:01:00,870
and in this case we won't
be using a cursor object

22
00:01:00,870 --> 00:01:02,550
because we're only going to request

23
00:01:02,550 --> 00:01:05,450
a few results for demonstration purposes.

24
00:01:05,450 --> 00:01:09,730
So, let me switch back
over to my terminal window,

25
00:01:09,730 --> 00:01:11,450
and you'll notice that I've already

26
00:01:11,450 --> 00:01:14,100
pasted in a snippet of code here.

27
00:01:14,100 --> 00:01:16,830
We have a file that we've provided for you

28
00:01:16,830 --> 00:01:19,930
in this lessons examples folder

29
00:01:19,930 --> 00:01:21,533
that is called tweetutilities.py,

30
00:01:22,990 --> 00:01:25,670
and it has a method, or a function rather,

31
00:01:25,670 --> 00:01:28,710
in it called print_tweets
that we will be using

32
00:01:28,710 --> 00:01:33,310
in this example to display
the content of each tweet.

33
00:01:33,310 --> 00:01:35,660
So, I'm going to import that function,

34
00:01:35,660 --> 00:01:38,600
and I just want to show you
that function quickly here.

35
00:01:38,600 --> 00:01:42,180
So, this is the tweetutilities.py file,

36
00:01:42,180 --> 00:01:44,770
which just has a number of
functions that we predefined

37
00:01:44,770 --> 00:01:47,930
for you that we use in
subsequent examples,

38
00:01:47,930 --> 00:01:50,150
and as I scroll down here this is

39
00:01:50,150 --> 00:01:52,840
the definition of the
print_tweets function.

40
00:01:52,840 --> 00:01:56,340
It's going to receive a
collection of status objects

41
00:01:56,340 --> 00:02:00,830
representing the tweets that
are returned in this example

42
00:02:00,830 --> 00:02:03,230
that will be from the search API,

43
00:02:03,230 --> 00:02:06,390
and for each of the
tweets in that collection

44
00:02:06,390 --> 00:02:09,040
we're going to get the user object

45
00:02:09,040 --> 00:02:11,680
from the tweet and show the screen_name.

46
00:02:11,680 --> 00:02:16,110
And then if the tweet is
in the English language

47
00:02:16,110 --> 00:02:18,920
we're going to display the tweet's text,

48
00:02:18,920 --> 00:02:22,080
and if it's not in
English and the language

49
00:02:22,080 --> 00:02:25,240
is not UND, which means undefined,

50
00:02:25,240 --> 00:02:29,440
then we're going to show you
the original tweet's text

51
00:02:29,440 --> 00:02:32,290
and we're going to translate that tweet

52
00:02:32,290 --> 00:02:35,560
using TextBlob's translate capabilities,

53
00:02:35,560 --> 00:02:40,140
so we did import the
TextBlob class up above,

54
00:02:40,140 --> 00:02:43,500
so we're going to create a
TextBlob from the tweet's text,

55
00:02:43,500 --> 00:02:46,090
then we're going to invoke
the translate method

56
00:02:46,090 --> 00:02:47,890
on that to get back the translated

57
00:02:47,890 --> 00:02:50,490
string and display that string.

58
00:02:50,490 --> 00:02:52,580
So, of course Twitter is used worldwide

59
00:02:52,580 --> 00:02:56,110
and tweets come in in many
different spoken languages.

60
00:02:56,110 --> 00:02:58,500
If Twitter is able to
recognize that language

61
00:02:58,500 --> 00:03:00,630
it will encode the tweet accordingly,

62
00:03:00,630 --> 00:03:02,700
and if it's not able to recognize it

63
00:03:02,700 --> 00:03:06,070
it will list that at UND for undefined.

64
00:03:06,070 --> 00:03:09,080
So, going back over to our session now,

65
00:03:09,080 --> 00:03:12,450
the next thing we want to
do is perform a search.

66
00:03:12,450 --> 00:03:15,260
So, we're going to call
the API method search,

67
00:03:15,260 --> 00:03:17,860
and again, I'll keep stating this

68
00:03:17,860 --> 00:03:19,920
throughout several of the examples,

69
00:03:19,920 --> 00:03:23,000
we really should be using
cursors if we're going

70
00:03:23,000 --> 00:03:27,080
to invoke search and get
a lot of results back.

71
00:03:27,080 --> 00:03:29,240
In this case we're getting
only three results,

72
00:03:29,240 --> 00:03:33,540
so there really is no need to
use a cursor for this example.

73
00:03:33,540 --> 00:03:37,020
The important argument is the Q argument,

74
00:03:37,020 --> 00:03:38,930
which represents the query string.

75
00:03:38,930 --> 00:03:42,450
In this case we're
looking for recent tweets

76
00:03:42,450 --> 00:03:45,900
for the Mars Opportunity Rover.

77
00:03:45,900 --> 00:03:48,420
So, we're going to execute that,

78
00:03:48,420 --> 00:03:50,730
and now we have tweet objects back

79
00:03:50,730 --> 00:03:54,420
and we can call out print_tweets
method, or function rather,

80
00:03:54,420 --> 00:03:57,570
and we can simply hand
it the tweets object

81
00:03:57,570 --> 00:04:00,290
that we just created in
the previous snippet,

82
00:04:00,290 --> 00:04:04,200
and it's going to then
display those tweets to us.

83
00:04:04,200 --> 00:04:05,720
And in this case we did happen

84
00:04:05,720 --> 00:04:09,010
to get a tweet that was translated,

85
00:04:09,010 --> 00:04:12,750
so it looks like this might be Spanish

86
00:04:12,750 --> 00:04:15,450
or possibly Portuguese.

87
00:04:15,450 --> 00:04:17,180
I don't speak either of those languages,

88
00:04:17,180 --> 00:04:19,630
so I'm gonna guess Spanish
based on what I'm seeing here.

89
00:04:19,630 --> 00:04:22,670
But in any case, it was able to go ahead

90
00:04:22,670 --> 00:04:24,800
and translate that into English

91
00:04:24,800 --> 00:04:29,180
as well as part of the display here.

92
00:04:29,180 --> 00:04:31,550
Now, separately we have the ability

93
00:04:31,550 --> 00:04:34,710
to work with Twitter Search operators

94
00:04:34,710 --> 00:04:37,900
to refine the searches
that we're performing.

95
00:04:37,900 --> 00:04:41,530
So, let me jump back over to
the slides for a moment here,

96
00:04:41,530 --> 00:04:45,130
and the search operators allow you

97
00:04:45,130 --> 00:04:48,610
to create basically more complex queries.

98
00:04:48,610 --> 00:04:51,660
I've actually already
opened a browser window

99
00:04:51,660 --> 00:04:54,760
for the Twitter Search homepage,

100
00:04:54,760 --> 00:04:57,190
and the reason I did that
is you'll notice right below

101
00:04:57,190 --> 00:05:00,030
the search field here
there's this operators link,

102
00:05:00,030 --> 00:05:03,520
and if you click that it
will take you to a page

103
00:05:03,520 --> 00:05:06,310
that allows you to see all the different

104
00:05:06,310 --> 00:05:08,440
search operators with examples.

105
00:05:08,440 --> 00:05:11,740
I have some examples that I've
pasted into my slides here,

106
00:05:11,740 --> 00:05:14,500
so these are just some
of the ways in which

107
00:05:14,500 --> 00:05:16,440
you can use Twitter Search operators

108
00:05:16,440 --> 00:05:19,490
as part of your query
strings that you supply

109
00:05:19,490 --> 00:05:22,700
to that API object's search method.

110
00:05:22,700 --> 00:05:25,420
So, if you just specify words,

111
00:05:25,420 --> 00:05:28,880
that's implicitly going to
be a logical and operation.

112
00:05:28,880 --> 00:05:31,120
So, if we search for "python twitter,"

113
00:05:31,120 --> 00:05:32,520
it's going to look for tweets that

114
00:05:32,520 --> 00:05:35,470
contain both "python" and "twitter."

115
00:05:35,470 --> 00:05:38,460
If you include the word OR between words,

116
00:05:38,460 --> 00:05:40,930
that's going to be your
logical OR operator,

117
00:05:40,930 --> 00:05:42,687
so it will find tweets that contain

118
00:05:42,687 --> 00:05:45,830
"python" or "twitter" or both.

119
00:05:45,830 --> 00:05:48,410
If you say "python" and you include

120
00:05:48,410 --> 00:05:50,630
a question mark in the search string

121
00:05:50,630 --> 00:05:52,450
it looks for tweets that are asking

122
00:05:52,450 --> 00:05:55,420
questions about the specified topic.

123
00:05:55,420 --> 00:05:59,170
If you specify a search term followed

124
00:05:59,170 --> 00:06:02,100
by a minus sign and another term

125
00:06:02,100 --> 00:06:05,780
it's going to find tweets
containing the original term

126
00:06:05,780 --> 00:06:07,930
that don't contain the other one,

127
00:06:07,930 --> 00:06:10,610
so this particular one
would look for tweets

128
00:06:10,610 --> 00:06:12,180
that contain the word "planets"

129
00:06:12,180 --> 00:06:15,560
but don't contain the word "mars."

130
00:06:15,560 --> 00:06:19,010
Some other examples, if you
include a smiley face emoji,

131
00:06:19,010 --> 00:06:22,200
that's going to look for
positive sentiment tweets,

132
00:06:22,200 --> 00:06:24,710
so this is a little bit
of sentiment analysis

133
00:06:24,710 --> 00:06:26,770
in the context of Twitter Search.

134
00:06:26,770 --> 00:06:30,100
If you put in a sad
face you'll get negative

135
00:06:30,100 --> 00:06:33,090
sentiment tweets on that specified topic.

136
00:06:33,090 --> 00:06:35,660
You also can search based on timeframes,

137
00:06:35,660 --> 00:06:39,210
so I can use the "since"
keyword followed by

138
00:06:39,210 --> 00:06:41,810
a colon and a date of the form shown here,

139
00:06:41,810 --> 00:06:45,720
which is a four digit
year, a month, and a day,

140
00:06:45,720 --> 00:06:50,140
and it will find tweets on
or after that timeframe,

141
00:06:50,140 --> 00:06:54,400
and again, remember you can
only do for the past seven days.

142
00:06:54,400 --> 00:06:59,400
So, if today were, well,
today is June 12th of 2019,

143
00:07:01,030 --> 00:07:04,960
so I could go back as far as
seven days ago for tweets,

144
00:07:04,960 --> 00:07:08,240
so maybe I only want tweets
as far back as two days ago

145
00:07:08,240 --> 00:07:10,850
and I could specify a date if I would

146
00:07:10,850 --> 00:07:12,960
like with the "since" keyword.

147
00:07:12,960 --> 00:07:16,170
I can also look for tweets
near a particular location,

148
00:07:16,170 --> 00:07:18,150
so I can give it a string like a city,

149
00:07:18,150 --> 00:07:20,690
in this case "New York City"
and it will find tweets

150
00:07:20,690 --> 00:07:24,890
that came from that general
area or that vicinity,

151
00:07:24,890 --> 00:07:27,360
and you can also use the
"from" and "to" keywords

152
00:07:27,360 --> 00:07:29,770
to specify the particular accounts

153
00:07:29,770 --> 00:07:31,390
from which a tweet was sent

154
00:07:31,390 --> 00:07:34,760
or to which a tweet was sent as well.

155
00:07:34,760 --> 00:07:37,880
So, with that in mind we can go ahead

156
00:07:37,880 --> 00:07:41,940
and do a demonstration of some Python,

157
00:07:41,940 --> 00:07:44,640
or not Python, excuse me,
Twitter Search operators.

158
00:07:44,640 --> 00:07:49,570
So, let's say that we would
like to search for tweets

159
00:07:49,570 --> 00:07:53,410
from the NASA account
since a particular date.

160
00:07:53,410 --> 00:07:55,690
Now, I've gone ahead here

161
00:07:55,690 --> 00:08:00,150
and changed this to June 6th of 2019.

162
00:08:00,150 --> 00:08:03,990
Again, this is June 12th of 2019,

163
00:08:03,990 --> 00:08:07,290
and I'm looking for a maximum
of three tweets in this case,

164
00:08:07,290 --> 00:08:09,450
so I'll go ahead and execute that.

165
00:08:09,450 --> 00:08:12,780
And once again, I will
call my print_tweets

166
00:08:12,780 --> 00:08:15,600
function to display those tweets,

167
00:08:15,600 --> 00:08:19,360
so although I'm not showing
dates with each of these tweets,

168
00:08:19,360 --> 00:08:23,220
all of these would've
been within seven days

169
00:08:23,220 --> 00:08:27,560
of the time that I'm doing
this particular recording.

170
00:08:27,560 --> 00:08:30,720
Now, you can also search
for things like hashtags,

171
00:08:30,720 --> 00:08:33,610
so of course if you're
familiar with Twitter

172
00:08:33,610 --> 00:08:37,350
hashtags are a way that people

173
00:08:37,350 --> 00:08:40,690
point out important things in a tweet.

174
00:08:40,690 --> 00:08:44,500
So, for example if I want to do a search

175
00:08:44,500 --> 00:08:46,290
for the hashtag #collegefootball

176
00:08:47,660 --> 00:08:51,410
I can use a pound sign as
part of my query string,

177
00:08:51,410 --> 00:08:54,647
and now it's going to look
for tweets that include

178
00:08:54,647 --> 00:08:58,790
#collegefootball as
part of the tweet text,

179
00:08:58,790 --> 00:09:02,160
and in this case we're just
going to get two tweets,

180
00:09:02,160 --> 00:09:04,383
and we'll call print_tweets once again.

181
00:09:05,640 --> 00:09:08,240
So, in this case it looks like three,

182
00:09:08,240 --> 00:09:11,210
but in reality one of these is a tweet

183
00:09:11,210 --> 00:09:15,320
that has a line break in the middle of it,

184
00:09:15,320 --> 00:09:17,320
so it's not really three
tweets that came back.

185
00:09:17,320 --> 00:09:20,063
It's two tweets in this particular case.