add run_length_encode()

Author: cleoold

doc/api/more/types_linq.more.more_enumerable.rst

@@ -459,6 +459,49 @@ Revisions:
 
 ----
 
+instancemethod ``run_length_encode()``
+----------------------------------------
+
+
+Returns
+  - ``MoreEnumerable[Tuple[TSource_co, int]]``
+
+Run-length encodes the sequence into a sequence of tuples where each tuple contains an
+(the first) element and its number of contingent occurrences, where equality is based on
+`==`.
+
+Example
+    >>> MoreEnumerable('abbcaeeeaa').run_length_encode().to_list()
+    [('a', 1), ('b', 2), ('c', 1), ('a', 1), ('e', 3), ('a', 2)]
+
+Revisions:
+    - main: New.
+
+----
+
+instancemethod ``run_length_encode(__comparer)``
+--------------------------------------------------
+
+Parameters
+  - `__comparer` (``Callable[[TSource_co, TSource_co], bool]``)
+
+Returns
+  - ``MoreEnumerable[Tuple[TSource_co, int]]``
+
+Run-length encodes the sequence into a sequence of tuples where each tuple contains an
+(the first) element and its number of contingent occurrences, where equality is determined by
+the comparer.
+
+Example
+    >>> MoreEnumerable('abBBbcaEeeff') \
+    >>>     .run_length_encode(lambda x, y: x.lower() == y.lower()).to_list()
+    [('a', 1), ('b', 4), ('c', 1), ('a', 1), ('E', 3), ('f', 2)]
+
+Revisions:
+    - main: New.
+
+----
+
 staticmethod ``traverse_breath_first[TSource](root, children_selector)``
 --------------------------------------------------------------------------
 

doc/api_spec.py

@@ -205,6 +205,7 @@ class ClassSpec(TypedDict):
                     'pipe',
                     'rank',
                     'rank_by',
+                    'run_length_encode',
                     'traverse_breath_first',
                     'traverse_depth_first',
                 },

tests/test_more_usage.py

@@ -429,6 +429,35 @@ def test_overload2_empty(self):
             .to_list() == []
 
 
+class TestRunLengthEncodeMethod:
+    def test_overload1(self):
+        en = MoreEnumerable('abbcaeeeaa')
+        assert en.run_length_encode().to_list() \
+            == [('a', 1), ('b', 2), ('c', 1), ('a', 1), ('e', 3), ('a', 2)]
+
+    def test_overload1_empty(self):
+        en = MoreEnumerable(())
+        assert en.run_length_encode().to_list() == []
+
+    def test_overload1_one_run(self):
+        en = MoreEnumerable('AAAAA')
+        assert en.run_length_encode().to_list() == [('A', 5)]
+
+    def test_overload1_one_elem(self):
+        en = MoreEnumerable('A')
+        assert en.run_length_encode().to_list() == [('A', 1)]
+
+    def test_overload1_no_run(self):
+        en = MoreEnumerable('abcdefghijklmnopqrstuvwxyz')
+        assert en.run_length_encode().to_list() == \
+            en.select(lambda x: (x, 1)).to_list()
+
+    def test_overload2(self):
+        en = MoreEnumerable('abBBbcaEeeff')
+        assert en.run_length_encode(lambda x, y: x.lower() == y.lower()).to_list() \
+            == [('a', 1), ('b', 4), ('c', 1), ('a', 1), ('E', 3), ('f', 2)]
+
+
 class TestCycleMethod:
     def test_repeat(self):
         en = MoreEnumerable([1, 2, 3])

types_linq/more/more_enumerable.py

@@ -192,6 +192,34 @@ def inner():
                 yield rank_map[key]
         return MoreEnumerable(inner)
 
+    def run_length_encode(self,
+        *args: Callable[[TSource_co, TSource_co], bool],
+    ) -> MoreEnumerable[Tuple[TSource_co, int]]:
+        if len(args) == 0:
+            comparer = lambda x, y: x == y
+        else:
+            comparer = args[0]
+        def inner():
+            iterator = iter(self)
+            try:
+                prev_elem = next(iterator)
+            except StopIteration:
+                return
+            count = 1
+            while True:
+                try:
+                    elem = next(iterator)
+                except StopIteration:
+                    break
+                if comparer(prev_elem, elem):
+                    count += 1
+                else:
+                    yield (prev_elem, count)
+                    prev_elem = elem
+                    count = 1
+            yield (prev_elem, count)
+        return MoreEnumerable(inner)
+
     @staticmethod
     def traverse_breath_first(
         root: TSource,

types_linq/more/more_enumerable.pyi

@@ -317,6 +317,39 @@ class MoreEnumerable(Enumerable[TSource_co]):
             - v1.0.0: New.
         '''
 
+    @overload
+    def run_length_encode(self) -> MoreEnumerable[Tuple[TSource_co, int]]:
+        '''
+        Run-length encodes the sequence into a sequence of tuples where each tuple contains an
+        (the first) element and its number of contingent occurrences, where equality is based on
+        `==`.
+
+        Example
+            >>> MoreEnumerable('abbcaeeeaa').run_length_encode().to_list()
+            [('a', 1), ('b', 2), ('c', 1), ('a', 1), ('e', 3), ('a', 2)]
+
+        Revisions:
+            - main: New.
+        '''
+
+    @overload
+    def run_length_encode(self,
+        __comparer: Callable[[TSource_co, TSource_co], bool],
+    ) -> MoreEnumerable[Tuple[TSource_co, int]]:
+        '''
+        Run-length encodes the sequence into a sequence of tuples where each tuple contains an
+        (the first) element and its number of contingent occurrences, where equality is determined by
+        the comparer.
+
+        Example
+            >>> MoreEnumerable('abBBbcaEeeff') \\
+            >>>     .run_length_encode(lambda x, y: x.lower() == y.lower()).to_list()
+            [('a', 1), ('b', 4), ('c', 1), ('a', 1), ('E', 3), ('f', 2)]
+
+        Revisions:
+            - main: New.
+        '''
+
     @staticmethod
     def traverse_breath_first(
         root: TSource,