"TypedDict" In Python Is Actually AWESOME!

[@tfr]

i love the way you pronounce “category”, very unique haha

[@kamurashev]

I read it as …you pronounce category and unique. I listened the whole thing one more time to find out how does he pronounces “unique” and then I get back to this comment and actually read it correctly, so now I am laughing.

[@man0utoftime]

"ka-'TEG-o-ree" is a 1st for me, as well. 😂

[@EverythingTechTime]

Cet egery

[@tribaltalker1608]

I had a student who pronounced "integer" as "in tee grr". He had never heard anyone say the word, so he guessed the pronunciation from reading it. English doesn't have much consistency in how we spell words as we have (mostly) spurned the accent marks that other languages use. It's a linguistic minefield.

[@herrpez]

@@tribaltalker1608 It's been quite a while since English spelling and pronunciation drifted apart. But once upon a time they were actually fairly close. Add the fact that English is basically three languages in a trench coat... 😂 It's no wonder people get confused.

[@kvelez]

from typing import TypedDict, NotRequired class Coordinate(TypedDict): # inheritance x: float y: float label: str category: NotRequired[str] # could be excluded / type needed # IDE helps with type: Coordinate coordinate: Coordinate = {"x":10, "y":10, 'label':"Profit", 'category':"Finance"} coordinated: Coordinate = {"x":15, "y":25, 'label':"Expenses"} # category not used. 6:40 from typing import TypedDict, NotRequired, Required Vote = TypedDict('Vote', {"for":int, "against":int}, total=False) # override for Votes = TypedDict('Votes', {"for":int, "against":Required[int]}, total=False) # type needed vote : Vote = {"for":100, "against":250} votes : Votes = {"against": 450}

[@kolterdyx]

Hell yeah! I had no idea this was possible! I've been missing this feature for ever since I started using typescript. Having an interface for a simple data type like a dictionary is the best!

[@AlmondAxis987]

Ooh this is quite similar to c++'s 'typedef struct' declaration. It's super useful in data management for vectors and matrices.

[@ajdziaj]

Nice feature. I’m definitely going to try it out in next project. And I like your style of these videos, keep up the good work 🙂

[@erickemayot9552]

I really enjoy learning from your video compilations. Thanks so much for the good work. Keep them coming through.

[@maswinkels]

Great video. Useful info and clearly explained. Thanks! (One small thing: the word 'category' has emphasis on the first syllable: CATegory).

[@berkaybakacak]

This is not Python that I know 😂 This channel teached me a lot. Thank you so much

[@SkyyySi]

Please note that this is a bad use of TypedDict. The reason to use TypedDict is to adapt code that's already depending on untyped dictionaries to use types. Perhaps to add a type structure to the result of json.load(). However, designing your structured code from the ground up around dictionaries is a bad idea, because the purpose of a dict is to be unstructured. For designing strucutred types, Python has a dedicated mechanism: Classes. In particular, the dataclass from the dataclasses module. These are designed to have a structure, and are generally more convenient to use that way (for example, because they support looking up attributes with object.field instead of object["field"] out of the box).

[@keithkaranu4258]

Not: I'm not a python guy so idk if this is a good practice in Python, but I can imagine this being useful for implementing algebraic data types sort of like interfaces in typescript.

[@SkyyySi]

@@keithkaranu4258 I don't see how using a dictionary with a fixed type structure would change the things you can model over a class

[@youtubeenjoyer1743]

Please note that this comment is wrong. There is not such a thing as ten commandments discouraging heterogenous dictionaries. The purpose of the dictionary data structure is to map keys to values. The attribute lookup syntax is irrelevant, except it is more limited in comparison to the key lookup syntax.

[@matteolacki4533]

Hello! Is it not against the idea of a dictionary to specify in advance the keys and their types? Python also has namedtuples that should be way more efficient in use, especially given that one needs to declare types in advance, so obviously one needs to know them in advance. Is there a version of namedtuples where one can specify types?

[@paljain01]

I luv TypedDict, I had a long json body and keys were long I have to go back and forth to copy and paste that key. once I created the TypedDict the auto complete feels like magic ✨. unfortunately i had to use python 3.7 in prod so I had to install typing-extensions module to use TypedDict.

[@rogumann838]

I dont see why I would use this over creating a dataclass of some sort

[@alansnyder8448]

What version of python3 is this? The "typing" package seems to change between sub-versions. For example, I would have thought you would use Optional and didn't even know NotRequired existed.

[@alidev425]

please make a video about class methods and object methods(all kind of it ) and the differences between them just for clarification (and also usage of them in real situations) because the way you teach is amazing and thank you very much for your consideration

[@Indently]

I already did

[@flames9310]

Thanks for the info! I feel like this will make OOP easier to a certain extent.

[@thisoldproperty]

I'm now wondering when to use TypedDict vs @dataclass

[@walker-gx1lx]

same here...

[@Skull0]

If you want your data structure to behave like a dict but with a typed schema to prevent error this is the solution, you'll have all the expected methods and behavior or a normal dict which differ from a object, even if the typing is only statically checked this is a good solution to that need. Otherwise you'll have to create a class that derives from dict and then add validation to ensure all parameters are of the right type, it will execute at runtime so add overload which may not be necessary compare to just let the developer know that the a parameter type is wrong at development time with static analysis.

[@anasssofti9271]

Emm This will definitely help to organise more the codes, thanks for the tip

[@ada-victoria]

Love your content! not sure if it's intentional but your channel logo is so similar to LinkedIn I keep thinking it's a video posted by them fyi...

[@smyvensestime6995]

What code editor are you using? it looks very nice

[@SalticHash]

Seams to be pycharm

[@kunalsoni7681]

it seems very interesting to do 😀💯💙✨

[@IndraKurniawan]

Whoa. Bracket in Python!

[@timelessjc]

thanks for sharing. I tried this code: ====== from typing import TypedDict class Person(TypedDict): name: str age: int salary: float person: Person = { 'weight':125, 'height': '6ft', } print(person) ======= ... {'weight': 125, 'height': '6ft'} it worked fine and nothing stops me from giving anything I want to the 'person' dictionary. am I missing something here or I just don't get it?

[@youtubeenjoyer1743]

The TypedDict annotation is not enforced by the interpreter at runtime. If you want runtime type checking, you have to write it yourself by inspecting the Person class and comparing the dictionary with the declared type annotations.

[@EW-mb1ih]

What's the advantage of TypedDict vs dataclass?

[@Alister222222]

I am also wondering this. I guess one advantage might be that a Dict is (?) smaller and more lightweight than a class, since classes are kind of extended Dicts with a lot of extra functionality.

[@blackdereker4023]

@@Alister222222 I would put that under premature optimization. Unless you are creating hundreds of thousands of data or running Python on a microship you really shouldn't worry about that.

[@spaghettiking653]

@@Alister222222 What would be the difference with a slotted class, for example, memory-wise? If you make use of slots, I believe that dramatically shrinks the possible values, because the class no longer wraps a dictionary, but rather has distinct fields for the properties, exactly like the intended use of TypedDict?

[@mntraders8258]

Thank you

[@viktortodosijevic3270]

Why use this instead of a @dataclass?

[@YatharthMathur]

Because dataclasses aren't an exact replacement of python native dictionaries. TypedDict is just a type that helps with IDE auto completion. It's not adding any additional functionality like dataclasses.

[@Jason-b9t]

If you can implement your resource initialization yourself, then dataclass will indeed be better. But in most cases you will get data from other libraries, and these data structures are usually very "dirty". For example, processing from files (toml, json, xlsx), http/database responses, etc. are almost all nested dictionaries/lists. Converting these things into dataclasses will be cumbersome, creating a lot of dependencies and prone to errors .

[@lordcasper3357]

how and where to use this dictionaries?

[@Indently]

They're just normal dictionaries, you use them exactly the same way you'd use other dictionaries.

[@grippnault]

I seem to not have Required and NotRequired in typing module. I'm using PyCharm

[@grippnault]

For me, I do not have Required and NotRequired. My version of Python (3.10.5) would only except Optional

[@Indently]

You’re right, it’s new as of Python 3.11

[@grippnault]

@@Indently Thank you for replying.

[@alex_brutforce]

I receive - ImportError: cannot import name 'NotRequired' from 'typing'. Python 3.9.5

[@DrDeuteron]

ikr, a version minimum would be nice. I can't get the latest thanks to the security ppl.

[@rishiraj2548]

Good day greetings

[@xzex2609]

you realLY love typing :D , why would you even consider that the first coordinate = {...} has anything to do with that class Coordinate?

[@albertashabaaheebwa9111]

Awesome. Although you didn't demostrate the use of reserved keywords since for != 'for', in != 'in' Nonetheless, Awesome tutorial

[@darrenlefcoe]

great video. It seems that python is chasing statically typed languages in this regard. Better to keep to core, simple, readable python...

[@呀咧呀咧]

Thanks I learnt sth new

[@papetoast]

One really annoying thing about TypedDict is that you can't inherit from a TypedDict and then change the type of the values already defined For example, this does not work: class A(TypedDict): x: int Class B(A): x: str

[@codershorts]

for was not acceptable in class but "for" is acceptable in class - for is reserved, not "for"

[@falklumo]

So, a feature was introduced in 3.8, deprecated in 3.11 and to be removed in 3.13? Seriously? Is Python only for kids playing around?

[@youtubeenjoyer1743]

Python is a toy scripting language.

[@tourdesource]

> Invent a dynamically typed language for quick-and-dirty scripting and glueing > Decide you need static types and end up with a language as messy as Ruby and as cumbersome as Java, except 90x slower > mUh eCoSyStEm

[@DrDeuteron]

quick and not dirty. This fixation on typing is anti-pythonic.

[@K9Megahertz]

This is why I just stick with C/C++. If you're going to write code, why use an interpreted language that's orders of magnitude slower than something else. In either case you're going to be writing a lot of code, so why gas up and drive a Pinto when you can do it with a Ferrari... Every program boils down to a few core concepts: loops, lists, arrays, variables, conditional statements, etc.. The part I don't understand is all these languages come out that try to be better/safer than C++ and then they just slowly all regress back towards C++.

[@DrDeuteron]

@@K9Megahertz You are bastardizing Greenspuns 10th Law

[@spaghettiking653]

@@DrDeuteron I disagree, why not use types when your functions are required to take a specific type of data? Obviously some functions don't do this, e.g. print can print anything, but what about a function that iterates the lines of a file? Obviously it needs to take a file(-like) object, so why not type it as such? Then it's more obvious what the code does. And why not annotate what the function returns, because then you can tell without having to read the code? It's just easier, not to mention that the code editor can help you when it knows the possible types.

[@DrDeuteron]

Sentinel pattern: iter(fp.readline, “”) Is sufficient

[@rongarza9488]

So what language does the world pick for a lingua franca? English! hahahaha Want to know why? Because of the Industrial Revolution, and also the airline industry, and movies, and TV, and on and on. Moral of the story: you don't have to always be "right", you just have to be persistent and gain the leading edge. Don't ever give up.

[@SirSX3]

kategari? 😂

[@TomLeg]

English is a messed-up language ... "caaaaah-tegory'

[@vishalprajapati1004]

Please Hindi subtitle

[@spaghettiking653]

Sir I do not speak Hindi but you can try automatic subtitles and automatically translate to Hindi

[@БогданЗинченко-к5п]

pydantic with less abilities

[@kutilkol]

What a crap