Skip to content
Snippets Groups Projects
README.md 6.74 KiB
Newer Older
Filip Kučerák's avatar
Filip Kučerák committed
# KDTesting 

Python library for testing applications using IO.

# Test 

Test is container that holds name, arbitrary and value provided by said arbitrary (if you use constructor with None, it will generate value using arbitrary). The test also contains functions `shrink` and `to_str` (both just use the functions from arbitrary on the value held by the test). 
Filip Kučerák's avatar
Filip Kučerák committed

# Tester

Tester is abstract class, the most important function `run_test(test: Test[G])` is not defined. Tester defines how shrinking works etc.
Filip Kučerák's avatar
Filip Kučerák committed

# Arbitraries 

`Arbitrary[T]` defines an object capable of providing arbitrary value with type `T`, shrinking value of type `T` to a simpler instance and converting value of type `T` to `str`.

To define your custom Arbitrary, either extend one of the predefined structures (below) or extend Arbitrary. Note that you have to define functions `shrink`, you don't have to define `to_str` but it will probably create undesirable output. The function `get` uses the value generator from the constructor.

The constructor of `Arbitrary[T]` takes ValueGenerator, function with type `Callable[[], T]`, this function is used in get(). There are some predefined value generatos, see ValueGenerator section.
Filip Kučerák's avatar
Filip Kučerák committed

### Get
`get() -> T`

returns arbitrary value
Filip Kučerák's avatar
Filip Kučerák committed


### Shrink
`shrink(value: T) -> List[T]`

shrinks any value (created by the Arbitrary) to simpler value
Filip Kučerák's avatar
Filip Kučerák committed

### To string
`to_str(value: T) -> str`

converts the value to string 

## The arbitraries currently present are: 

### NumberA(min_value, max_value)
get returns: number in [min_value, max_value)

shrinks value to: val // 2, val moved towards 0

### ConstantA(constant)
get returns: constant 

shrinks value to: nothing

### TupleA(arbs)
get returns: tuple where i-th value uses i-th arbitrary from arbs

shrinks value to: list of different tuples with shrunk values 

### ListA(arb, min_size, max_size)
get returns: list with length [min_size, max_size)

shrinks value to: shorter lists

### ChoiceA(choices)
get returns: tuple (index of chosen arb, value generated by that arb), choice is random

shrinks value to: tuples with shrunk values (index of the chosen arb, value shrunk using said arb)

# ValueGenerator

There are some predefined factories. Note that you have to CALL the factory, to recieve value generator.

*- = returns generator that creates 

## predefined value generators creators 

### Atomic 
Filip Kučerák's avatar
Filip Kučerák committed
- `integer_gen(min_value: int, max_value: int)` *- random integer in range [min_value, max_value)
- `pos_integer_gen(max_value: int)` *- random integer in range [1, max_valaue)
- `lower_case_let_gen()` *- lower case letter
- `upper_case_let_gen()` *- upper case letter

### Generic  
- `constant_gen(const: T)` *- the constant 
- `oneof_gen(constants: List[T])` *- one of the constants
- `vector_gen(gen: Callable[[], T], size: int)` *- list with constant size with elements generated by the generator
- `listof_gen(gen: Callable[[], T], min_size: int, max_size: int)` *- list with random size [min_value, max_value) with elements generated by the generator
- `tupleof_gen(generators: List[ValueGenerator])` *- tuple created from the values of the different generators
- `choices_gen(choices: List[Tuple[int, Callable[[], T]]])` *- choices are list of tuples in form (weight, generator), generates value created by one of the generators, chooses using weighted random

# Testing with reference implementation 

## RefImlTester

Tester that runs tests on your implementation and the reference implementation, then runs your 'process output tests' on the outputs. More below.

### ProcessOutputTest

Type is `Callable[[ProcessOutput, ProcessOutput], bool]`, it takes the outputs (in format provided by the `subprocess.run`) and returns bool.

There are some presets: 
- `stdout_equality()`: returns test, that only compares the raw stdout
- `stdout_processed_equality(test_preprocessor, exp_preprocessor)`: returns test that compares processed stdouts. `test_preprocessor` and `exp_preprocessor` are `Callable[[str], str]` that "process" the stdout (you have to define these, there is `str_id` that returns the input unchaged `x -*> x`)

### Example use

```python
def ignore_numbers(x: str):
    return re.sub(r'\d+', '', x)

# `program_tst` and `program_exp` are absolute paths pointing to the executables
# 3 - verbosity (default is 5, too much) 
tester = RefImplTester(program_tst, program_exp, 3)
tester.add_output_test(stdout_processed_eqality(ignore_numbers, ignore_numbers))

for i in range(5):
    # `command_sequence` is a custom arbitrary that returns sequence of 
    # `\n` delimitered commands
    tester.add_test(Test(f"all_{i}", command_sequence)) 

# runs the test, returns (reduced test, number of shrinks) tuple 
reduced_tests = tester.run()

# prints the failed tests, first the name, the number of shrinks, the test, 
# then saves it as .tst file 
for test, shrinks in reduced_tests:
    print(test.name, shrinks)
    print(test.to_str())
    tester.save_shrunk(reduced_tests)

# also saves the original tests
tester.save_tests()
```

## Command 

Command is very similar to TupleA, it takes list of arbitraries, and provies 'tupled' values. The difference is, that Command's `to_str` returns by space separated values, instead of the standard (). 

Command can be provided with symbolic table `Dict[str, List[int]]`, table `symbolic name -> list of indeces`. This table can be later used in `get_symbol(self, symbol: str, value: Tuple[Generatable, ...])` to get value from the tuple by name (this is used in CommandSequence)

## CommandSequence

CommandSequence takes in commands in form of choice-style list of tuples. It is very similar to ListA, but it renders the values separated by '\n'. It also contains the `get_symbol(self, symbol: str, values: List[Tuple[int, Generatable]])` function, for retrieving symbols. This time it returns all the values present in all the commands.

## Example use

```python 
# This creates command in the form:
# git commit -m <string[5,10)>
git_commit = Command([
    ConstantA("git commit -m"),
    StringA(5, 10)
], {
    "message": [1]
})

# This creates command in the form:
# git pull
git_pull = Command([
    ConstantA("git pull")
])

# This creates sequence with pull being twice as likely as happen as commit
command_sequence = CommandSequence([
    (1, git_commit),
    (2, git_pull)
], 10, 15)

# Let't get random value from the command sequence
random_command_sequence = command_sequence.get()

print(command_sequence.to_str(random_command_sequence))

# Let's get the symbol message from commits
messages = command_sequence.get_symbol("message", random_command_sequence)
messages = [''.join(m) for m in messages]
print(messages)
```

Output 

```
git commit -m nYWMY
git pull
git pull
git pull
git commit -m lpejbIJr
git pull
git pull
git pull
git commit -m ZZNfJHHF
git pull
git commit -m UxkaTibR

['nYWMY', 'lpejbIJr', 'ZZNfJHHF', 'UxkaTibR']
```