1 00:00:05,120 --> 00:00:09,360 To demonstrate the problem with something that JSON doesn't recognise, I'll change our list to 2 00:00:09,360 --> 00:00:15,280 be a list of tuples, instead of a list of lists. IntelliJ has a search and replace feature, 3 00:00:15,280 --> 00:00:19,040 and you can restrict its scope by selecting the text that you want to change. 4 00:00:19,040 --> 00:00:27,268 I'll select all our inner lists, on lines 4 to 11. From the Edit menu, choose Find > Replace. 5 00:00:27,360 --> 00:00:32,479 In the top box, which is the text to find, enter an opening square bracket 6 00:00:32,479 --> 00:00:36,673 and in the bottom box, we'll replace it with an opening parenthesis. 7 00:00:36,800 --> 00:00:40,240 Then click Replace All. We get a lot of errors, 8 00:00:40,240 --> 00:00:43,280 but they'll go when we replace the closing brackets. 9 00:00:43,280 --> 00:00:47,680 Select the same lines again, and enter a closing square bracket for the text to find, 10 00:00:47,680 --> 00:00:52,720 and replace it with a closing parenthesis. Then click Replace All again. 11 00:00:52,720 --> 00:00:57,200 That's changed the lists to tuples. We can close the search and replace bar, 12 00:00:57,200 --> 00:01:02,960 with the X over on the right. Ok, uncomment lines 14 and 15, 13 00:01:02,960 --> 00:01:07,760 and let's see what happens this time: We're going to write to the test.json file again, 14 00:01:07,760 --> 00:01:12,570 this time we'll be serializing a list of tuples. Run the program. 15 00:01:17,920 --> 00:01:23,360 And we get back a list of lists. The JSON format doesn't support tuples. 16 00:01:23,360 --> 00:01:27,360 That's not surprising; many languages don't even have tuples, so they'd be 17 00:01:27,360 --> 00:01:33,710 unable to parse them out of the JSON text. Remember that JSON is a format for exchanging data. 18 00:01:33,710 --> 00:01:37,498 As long as you can get the data in a form that you can work with, in your code, 19 00:01:37,498 --> 00:01:41,520 then it's serving its purpose. JSON has to be language independent, 20 00:01:41,520 --> 00:01:45,360 otherwise it wouldn't be an open standard that can be used with any language. 21 00:01:45,360 --> 00:01:48,960 That means it can only support objects that are common in all languages. 22 00:01:48,960 --> 00:01:54,000 Let's see what types of data JSON supports. There's a table showing the JSON data types 23 00:01:54,000 --> 00:01:59,003 in the documentation for the json module. I'll go to that page in my browser. 24 00:02:02,480 --> 00:02:06,220 On the left are the eight different types of data that you can get from JSON. 25 00:02:06,220 --> 00:02:09,840 The right hand column shows the equivalent Python data types. 26 00:02:09,840 --> 00:02:14,080 JSON only defines seven data types, as shown in that table. 27 00:02:14,080 --> 00:02:18,800 The Python decoder is clever enough to detect if a number is an int or a float, and can return the 28 00:02:18,800 --> 00:02:25,360 appropriate Python type. But as far as JSON is concerned, there's only a single type, number. 29 00:02:25,360 --> 00:02:31,293 That table shows the JSON types, and the corresponding Python type, when a JSON string is decoded – 30 00:02:31,293 --> 00:02:35,840 such as when we read it from a file. Scroll down to the next table. 31 00:02:37,200 --> 00:02:42,880 It's almost the same. This time, it shows the Python types on the left, and the corresponding 32 00:02:42,880 --> 00:02:48,800 JSON type that will be produced, when encoding (or serializing) the objects to JSON. 33 00:02:48,800 --> 00:02:54,161 As you can see, both ints and floats result in the same JSON type, which is number. 34 00:02:54,240 --> 00:02:57,770 We haven't talked about enums yet, but they're just numbers. 35 00:02:57,840 --> 00:03:01,976 So, going back to our code, and checking the output in the Run pane, 36 00:03:02,080 --> 00:03:06,080 the Python decoder has produced lists, rather than tuples, because there's 37 00:03:06,080 --> 00:03:12,678 no specific tuple representation in JSON. Ok, I'll summarise all this with some slides. 38 00:03:13,920 --> 00:03:19,680 Use JSON when you want to store data, or transmit it, in a format that other systems can use. 39 00:03:19,680 --> 00:03:24,880 JSON isn't suitable for storing your program's data, if you need to preserve the exact type. 40 00:03:24,880 --> 00:03:29,600 As we've seen, it can only be used to represent a limited number of data types. 41 00:03:29,600 --> 00:03:33,871 Those types are usually sufficient, when you're only interested in the actual data. 42 00:03:34,000 --> 00:03:37,520 They're not suitable when the exact type of the object is important. 43 00:03:37,520 --> 00:03:41,520 That's rarely the case, when transferring JSON, but is important if you want to 44 00:03:41,520 --> 00:03:46,331 save data in your program, and get back exactly what you saved, later. 45 00:03:48,720 --> 00:03:52,320 The main limitation of JSON is related to the problem we've already seen: 46 00:03:52,320 --> 00:03:56,880 JSON only supports seven data types. In particular, there's no support 47 00:03:56,880 --> 00:04:00,160 for things like dates. The decoding program needs to 48 00:04:00,160 --> 00:04:04,720 know how you've stored your dates in strings. There is an international standard for storing 49 00:04:04,720 --> 00:04:10,960 dates (google ISO 8601 for details). When including dates in your JSON data, 50 00:04:10,960 --> 00:04:15,111 we definitely recommend sticking to one of the formats in that standard. 51 00:04:17,360 --> 00:04:20,531 Related to that is the lack of typing, in JSON. 52 00:04:20,640 --> 00:04:25,360 The JSON format doesn't provide any indication about the type of data it contains. 53 00:04:25,360 --> 00:04:30,960 Because it has a limited number of data types, a JSON parser can get the data back quite easily. 54 00:04:30,960 --> 00:04:36,000 But it's then up to the program consuming the data to work out what each piece of data represents. 55 00:04:36,000 --> 00:04:41,520 Dates, once again, are a good example of this. After parsing the data to get a string, 56 00:04:41,520 --> 00:04:47,010 the program will need to know the exact format that was used, to represent things like dates. 57 00:04:49,040 --> 00:04:55,635 Another example is representing complex numbers. The usual way, in JSON, is to use a list. 58 00:04:55,635 --> 00:05:02,160 For example 12.5 + 3i would be stored as [12.5, 3] and it's left to the decoding 59 00:05:02,160 --> 00:05:06,000 program to know that a complex number is present at that position in the data. 60 00:05:06,000 --> 00:05:12,240 A complex number can be specified in XML, to give one example, with code such as the XML below: 61 00:05:12,240 --> 00:05:15,680 JSON is far less verbose than some other serialization formats, 62 00:05:15,680 --> 00:05:21,340 such as XML, but as a consequence, it lacks the ability to specify the data type. 63 00:05:23,760 --> 00:05:28,480 Despite those limitations, JSON has become a popular format on the internet. 64 00:05:28,480 --> 00:05:33,392 It's size is often many times smaller than formats such as XML, for example. 65 00:05:33,392 --> 00:05:36,240 That can be important when you want a web page to load quickly, 66 00:05:36,240 --> 00:05:39,828 without having to wait while a large amount of data is downloaded. 67 00:05:39,828 --> 00:05:44,000 JSON is also easy to parse – in part because of its simplicity. 68 00:05:44,000 --> 00:05:48,480 Encoders and decoders are available for most programming languages in use today. 69 00:05:48,480 --> 00:05:52,640 It's also very easy to read and edit. Whitespace is unimportant 70 00:05:52,640 --> 00:05:58,228 (except inside strings of course) and it's quite hard to mess up a JSON document by accident. 71 00:05:59,440 --> 00:06:03,200 In the next couple of videos, we'll fetch some JSON data from the internet, 72 00:06:03,200 --> 00:06:08,033 and use it in our Python code. I'll see you, in the next video.