py-pattern-matching 0.1.1

Data-structure pattern matching and unpacking for Python

Usage

Call the patternmatching.match() function with two arguments:

• Argument 1: the pattern to match against

• Argument 2: the data structure to match

• Optional argument flatten: True to have the matched values flattened, i.e. returned as a flat tuple regardless of their position in the structure of the matched data.

Return value:

if flatten is true:
    match_ok, [matched_value, matched_value, matched_value, ...]
if flatten is not true:
    match_ok, matched_data_structure

Caveat: if all matched values in the pattern are ignored, match() does not return a tuple but a boolean.

YES:

match_ok = match(PATTERN, DATA)

NO:

match_ok, = match(PATTERN, DATA)

Built-in match objects

• ANY

• IS_INSTANCE

Any match object can be wrapped with IGNORE() to ignore the value matched by the match object in the returned values.

IGNORE itself can be used as a shortcut for IGNORE(ANY).

assert True == match(IGNORE, 123)

Usage in (unit) tests (but also anywhere else)

Comparing the return values of functions is tedious if you can’t simply use the == operator. For example when a function returns a tuple and you only care about some parts of the tuple and not the entire tuple:

retval = some_function_under_test()
assert retval and isinstance(retval, tuple) \
    and len(retval) == 3 and retval[0] == 'foo' \
    and isinstance(retval[1], tuple) and len(retval[1]) == 2 \
    and isinstance(retval[1][1], SomeException)

With pattern matching:

retval = some_function_under_test()
assert retval == ('foo', (ANY, IS_INSTANCE(SomeException)), ANY)

Examples

MATCH OK

match_ok, value = match(ANY, 'foobar')
assert match_ok and value == 'foobar'

match_ok = match(IGNORE(ANY), 'foobar')
assert match_ok

match_ok, value1, value2 match((ANY, ANY), ('foo', 'bar'))
assert match_ok and value1 == 'foo' and value2 == 'bar'

match_ok, (value1, value2) = match((ANY, ANY), ('foo', 'bar'), flatten=False)
assert match_ok and value1 == 'foo' and value2 == 'bar'

match_ok, (value1, ) = match((ANY, IGNORE(ANY)), ('foo', 'whatev'))
assert match_ok and value1 == 'foo'

match_ok, value1, value2 = match(('foo', ANY, ANY), ('foo', 1, 2))
assert match_ok and value1 == 1 and value2 == 2

match_ok, value1, value2, value3 = match(('foo', ANY, (ANY, ANY)), ('foo', 1, (2, 3)))
assert match_ok and value1 == 1 and value2 == 2 and value3 = 3

match_ok, (value1, (value2, value3)) = match(('foo', ANY, (ANY, ANY)), ('foo', 1, (2, 3)), flatten=False)
assert match_ok and value1 == 1 and value2 == 2 and value3 = 3

NO MATCH

match_ok = match(3, 4)
assert not match_ok

# notice how you can still successfully do unpacking of return values and just ignore `value` if the match failed
match_ok, value = match(IS_INSTANCE(unicode), '123132')
assert not match_ok

# notice how you can still successfully do unpacking of return values and just ignore `value` if the match failed
match_ok, (value1, (value2, value3)) = match(('foo', (ANY, (ANY, ANY))), ('not-foo', (1, (2, 3))))
assert not match_ok

# ...even when the structure of the data completely mismatches
match_ok, (value1, (value2, value3)) = match(('foo', (ANY, (ANY, ANY))), ('foo', 'blabla'))
assert not match_ok

# ...don't rely on `value1`, `value2` and `value3` being `None` though--the matcher can still return whatever it wants there; you have to check `match_ok` yourself.