Create pytest compatible tests for existing tests.

[pcdeadeasy]

• Create test_* files that invoke the classes from other modules and run the python to typescript schema generator.
• TBD, add a way to do snapshot testing
• Invoke the tests as part of the CI/CD pipeline

You can run the tests a couple of ways:

cd python/tests
pytest

or

cd python
hatch run test

[DanielRosenwasser]

So I'm not entirely familiar with it, but if we add syrupy, it looks like you can add a parameter called snapshot and you can just write

Suggested change

	def test_python_to_typescript_schema():
	result = python_type_to_typescript_schema(Cart)
	assert result is not None
	assert result.typescript_type_reference == "Cart"
	assert result.typescript_schema_str is not None
	assert not result.errors
	def test_coffeeshop(snapshot):
	result = python_type_to_typescript_schema(Cart)
	assert result == snapshot

Let me know how that works

[pcdeadeasy]

@DanielRosenwasser Worked out really well. Except it looks like the function python_type_to_typescript_schema produced the python to typescript conversions in an order that could change. Look at the outputs:

assert [+ received] == [- snapshot]
E - TypeScriptSchemaConversionResult(typescript_schema_str='interface Cart {\n type: "Cart";\n items: Array<LineItem | UnknownText>;\n}\n\ninterface LineItem {\n type: "LineItem";\n product: BakeryProduct | LatteDrink | CoffeeDrink | EspressoDrink | UnknownText;\n quantity: number;\n}\n\ninterface LatteDrink {\n type: "LatteDrink";\n name: "cappuccino" | "flat white" | "latte" | "latte macchiato" | "mocha" | "chai latte";\n temperature?: "hot" | "extra hot" | "warm" | "iced";\n // The default is \'grande\'\n size?: "short" | "tall" | "grande" | "venti";\n options?: Array<Creamer | Sweetener | Syrup | Topping | Caffeine | LattePreparation>;\n}\n\ninterface CoffeeDrink {\n type: "CoffeeDrink";\n name: "americano" | "coffee";\n temperature?: "hot" | "extra hot" | "warm" | "iced";\n // The default is \'grande\'\n size?: "short" | "tall" | "grande" | "venti";\n options?: Array<Creamer | Sweetener | Syrup | Topping | Caffeine | LattePreparation>;\n}\n\ninterface BakeryProduct {\n type: "BakeryProduct";\n name: "apple bran muffin" | "blueberry muffin" | "lemon poppyseed muffin" | "bagel";\n options?: Array<BakeryOption | BakeryPreparation>;\n}\n\ninterface Topping {\n type: "Topping";\n name: "cinnamon" | "foam" | "ice" | "nutmeg" | "w...

E + TypeScriptSchemaConversionResult(typescript_schema_str='interface Cart {\n type: "Cart";\n items: Array<LineItem | UnknownText>;\n}\n\ninterface LineItem {\n type: "LineItem";\n product: BakeryProduct | LatteDrink | CoffeeDrink | EspressoDrink | UnknownText;\n quantity: number;\n}\n\ninterface BakeryProduct {\n type: "BakeryProduct";\n name: "apple bran muffin" | "blueberry muffin" | "lemon poppyseed muffin" | "bagel";\n options?: Array<BakeryOption | BakeryPreparation>;\n}\n\ninterface BakeryPreparation {\n type: "BakeryPreparation";\n name: "warmed" | "cut in half";\n}\n\n// Represents any text that could not be understood.\ninterface UnknownText {\n type: "UnknownText";\n // The text that wasn\'t understood\n text: string;\n}\n\ninterface CoffeeDrink {\n type: "CoffeeDrink";\n name: "americano" | "coffee";\n temperature?: "hot" | "extra hot" | "warm" | "iced";\n // The default is \'grande\'\n size?: "short" | "tall" | "grande" | "venti";\n options?: Array<Creamer | Sweetener | Syrup | Topping | Caffeine | LattePreparation>;\n}\n\ninterface LattePreparation {\n type: "LattePreparation";\n name: "for here cup" | "lid" | "with room" | "to go" | "dry" | "wet";\n}\n\ninterface LatteDrink {\n type: "LatteDrink";\n name: "cap...

[pcdeadeasy]

As a result the snapshot tests fail. Once we can make the output more deterministic this will work out great.

[DanielRosenwasser]

I'm extremely surprised - you're saying across runs the output changes?

[DanielRosenwasser]

Duh, sets and dicts don't specify an iteration order in Python - they are guaranteed to have insertion order in JavaScript.

[DanielRosenwasser]

I'll fix that by Monday.

[pcdeadeasy]

No worries. The other is test failing was the hello world one but I guess all of them might have the same issue. We need to maintain the order to be able to run the snapshot tests.

[DanielRosenwasser]

#192 should fix the instances in the Py->TS translation where order matters on sets and dicts

[pcdeadeasy]

@DanielRosenwasser and @hillary-mutisya, could you review. With Daniel's last PR, we now have snapshot testing enabled. If tests change, the snapshots will need to be regenerated. (pytest --snapshot-update)

[DanielRosenwasser]

Could you inline the schemas into each test file?

Also, it's a little unfortunate that the baseline output is so unreadable, but we can fix this later with some customization in syrupy.

[DanielRosenwasser]

Could you inline the schemas into each test file?

Also, it's a little unfortunate that the baseline output is so unreadable, but we can fix this later with some customization in syrupy.