1
00:00:00,03 --> 00:00:02,01
- [Instructor] Now that we've seen the location tool,

2
00:00:02,01 --> 00:00:04,01
let's take a look at the weather tool.

3
00:00:04,01 --> 00:00:05,08
The weather tool is pretty straightforward.

4
00:00:05,08 --> 00:00:07,05
The only file that we need to import,

5
00:00:07,05 --> 00:00:09,07
the only library we need is requests

6
00:00:09,07 --> 00:00:13,02
because we're going to be using the Open-Meteo APIs.

7
00:00:13,02 --> 00:00:15,03
And this Open-Meteo API, as you can see here,

8
00:00:15,03 --> 00:00:16,09
just gives us a URL,

9
00:00:16,09 --> 00:00:18,08
that we pass it the latitude

10
00:00:18,08 --> 00:00:21,06
and the longitude of where we are,

11
00:00:21,06 --> 00:00:23,07
and then some other parameters, for example,

12
00:00:23,07 --> 00:00:27,02
like we want hourly weather, precipitation probability,

13
00:00:27,02 --> 00:00:30,07
weather code, stuff like that to come back from it.

14
00:00:30,07 --> 00:00:32,04
We're going to use requests

15
00:00:32,04 --> 00:00:34,08
to get that URL to get a response.

16
00:00:34,08 --> 00:00:36,07
If there's an error or a bad response,

17
00:00:36,07 --> 00:00:39,02
we understand that we're just going to exit out of here

18
00:00:39,02 --> 00:00:41,01
and say there's an error fetching the data.

19
00:00:41,01 --> 00:00:43,07
And again, our agent using this tool,

20
00:00:43,07 --> 00:00:47,02
if it sees an error, could come up with an alternate plan.

21
00:00:47,02 --> 00:00:49,01
But once we've done that and we get it back,

22
00:00:49,01 --> 00:00:53,02
then we want to get the weather for the day-to-day.

23
00:00:53,02 --> 00:00:55,08
And we're going to get that as a summary

24
00:00:55,08 --> 00:00:58,05
with that weather data that we've returned.

25
00:00:58,05 --> 00:01:00,00
So, to summarize the weather,

26
00:01:00,00 --> 00:01:01,05
I've made a couple of decisions here

27
00:01:01,05 --> 00:01:03,07
that one of them in particular would be...

28
00:01:03,07 --> 00:01:05,01
We want to hike,

29
00:01:05,01 --> 00:01:08,05
and we're generally going to be hiking in daytime hours.

30
00:01:08,05 --> 00:01:13,07
So, this code is going to take a look at today

31
00:01:13,07 --> 00:01:18,06
between 0800 and 1700, so between 8:00 AM and 5:00 PM,

32
00:01:18,06 --> 00:01:21,02
and use that to be able to get a summary

33
00:01:21,02 --> 00:01:24,00
of the daytime weather hours.

34
00:01:24,00 --> 00:01:26,01
So, once we've done that,

35
00:01:26,01 --> 00:01:28,09
this is how we're going to be able to generate,

36
00:01:28,09 --> 00:01:32,03
remember that English language forecast for the day

37
00:01:32,03 --> 00:01:35,01
that the engine would be able to reason across,

38
00:01:35,01 --> 00:01:39,04
because everything that the service is returning to me

39
00:01:39,04 --> 00:01:42,02
would be, for example, a table containing,

40
00:01:42,02 --> 00:01:43,08
at this time, it's this temperature,

41
00:01:43,08 --> 00:01:45,09
this precipitation probability.

42
00:01:45,09 --> 00:01:47,07
And that would be very useful to us.

43
00:01:47,07 --> 00:01:50,04
But I wanted to process this weather data

44
00:01:50,04 --> 00:01:53,08
to get that concise summary for today's hours.

45
00:01:53,08 --> 00:01:55,06
So, that's all this code is doing here,

46
00:01:55,06 --> 00:01:57,02
is that it's looking at the data

47
00:01:57,02 --> 00:02:00,00
that it was getting back in this response

48
00:02:00,00 --> 00:02:01,08
from the weather API

49
00:02:01,08 --> 00:02:06,02
and turning that into a decent English language summary

50
00:02:06,02 --> 00:02:07,09
of the weather for today

51
00:02:07,09 --> 00:02:10,07
between the hours of 8:00 AM and 5:00 PM.

52
00:02:10,07 --> 00:02:13,02
All this code is very specific to that API.

53
00:02:13,02 --> 00:02:15,09
You could go into more detail by looking at that API.

54
00:02:15,09 --> 00:02:18,08
I'm not going to waste your time by going into that detail now,

55
00:02:18,08 --> 00:02:20,05
but that's just the important part here,

56
00:02:20,05 --> 00:02:23,01
is that the idea was to get that summary

57
00:02:23,01 --> 00:02:24,07
for the daylight hours

58
00:02:24,07 --> 00:02:28,02
and then return that here as this weather description.

59
00:02:28,02 --> 00:02:30,07
So, the weather description comes

60
00:02:30,07 --> 00:02:32,06
from a number of these codes.

61
00:02:32,06 --> 00:02:34,05
This service will return these codes.

62
00:02:34,05 --> 00:02:38,06
So, for example, if it was to return code number 73,

63
00:02:38,06 --> 00:02:41,00
the text for that is moderate snowfall.

64
00:02:41,00 --> 00:02:44,09
And I want the AI to be able to reason across the text,

65
00:02:44,09 --> 00:02:47,01
rather than having to deal with a code.

66
00:02:47,01 --> 00:02:50,09
So, this is where the weather tool is taking those codes

67
00:02:50,09 --> 00:02:53,02
that are from the Open-Meteo

68
00:02:53,02 --> 00:02:55,04
and turning that into that summary

69
00:02:55,04 --> 00:02:57,01
that we saw an example of earlier on

70
00:02:57,01 --> 00:02:59,03
that's saying today is 20 degrees,

71
00:02:59,03 --> 00:03:02,02
average temperature of 20 degrees, it's going to be sunny

72
00:03:02,02 --> 00:03:05,01
and a maximum precipitation probability of whatever,

73
00:03:05,01 --> 00:03:08,08
so that the service has something to reason across.

74
00:03:08,08 --> 00:03:10,08
And I've also made this to be...

75
00:03:10,08 --> 00:03:13,01
You could use it as a standalone if you want to test it,

76
00:03:13,01 --> 00:03:16,00
where I've just hard coded some latitude and longitude

77
00:03:16,00 --> 00:03:22,02
so we can run it and see how it looks.

78
00:03:22,02 --> 00:03:25,02
I do that here by running python weather.py.

79
00:03:25,02 --> 00:03:27,04
And for that latitude and longitude,

80
00:03:27,04 --> 00:03:29,06
today's forecast, clear sky,

81
00:03:29,06 --> 00:03:32,03
if you remember clear sky here

82
00:03:32,03 --> 00:03:34,02
should correlate to...

83
00:03:34,02 --> 00:03:37,01
Yep, the WMO code zero here.

84
00:03:37,01 --> 00:03:40,06
The API returns that WMO code.

85
00:03:40,06 --> 00:03:44,04
I chose that one and turned that into English language.

86
00:03:44,04 --> 00:03:45,08
And then the average temperature

87
00:03:45,08 --> 00:03:47,07
was the temperature if we looked at

88
00:03:47,07 --> 00:03:49,08
the data that came back from the service

89
00:03:49,08 --> 00:03:52,00
between the hours of 8:00 AM and 5:00 PM

90
00:03:52,00 --> 00:03:53,04
and averaged that out.

91
00:03:53,04 --> 00:03:54,09
And then we took a look at

92
00:03:54,09 --> 00:03:57,03
the maximum precipitation probability again,

93
00:03:57,03 --> 00:04:00,03
looking through the table of data that the service returned,

94
00:04:00,03 --> 00:04:03,04
and in this case, found the maximum probability of 0%,

95
00:04:03,04 --> 00:04:07,02
so we get that really nice English language forecast

96
00:04:07,02 --> 00:04:10,03
that the AI model could reason across.

97
00:04:10,03 --> 00:04:12,02
So, that's it for the weather tool.

98
00:04:12,02 --> 00:04:14,02
Pretty straightforward, as you can see.

99
00:04:14,02 --> 00:04:15,08
But the next video, we're going to then

100
00:04:15,08 --> 00:04:18,00
take a look at our final and most important tool,

101
00:04:18,00 --> 00:04:21,00
and that's the one that's for the park's APIs.