Add misra-gries streaming algorithm

README.md

@@ -307,6 +307,7 @@ If you want to uninstall algorithms, it is as simple as:
     - [is_sorted](algorithms/stack/is_sorted.py)
 - [streaming](algorithms/streaming)
     - [1-sparse-recovery](algorithms/streaming/one_sparse_recovery.py)
+    - [misra-gries](algorithms/streaming/misra_gries.py)
 - [strings](algorithms/strings)
     - [fizzbuzz](algorithms/strings/fizzbuzz.py)
     - [delete_reoccurring](algorithms/strings/delete_reoccurring.py)

algorithms/streaming/__init__.py

@@ -1 +1,2 @@
-from .one_sparse_recovery import *
+from .one_sparse_recovery import *
+from .misra_gries import *

algorithms/streaming/misra_gries.py

@@ -0,0 +1,49 @@
+
+"""
+Implementation of the Misra-Gries algorithm.
+Given a list of items and a value k, it returns the every item in the list that appears at least n/k times, where n is the length of the array
+By default, k is set to 2, solving the majority problem. 
+For the majority problem, this algorithm only guarantees that if there is an element that appears more than n/2 times, it will be outputed. If there
+is no such element, any arbitrary element is returned by the algorithm. Therefore, we need to iterate through again at the end. But since we have filtred
+out the suspects, the memory complexity is significantly lower than it would be to create counter for every element in the list.
+
+For example:
+Input misras_gries([1,4,4,4,5,4,4])
+Output {'4':5}
+Input misras_gries([0,0,0,1,1,1,1])
+Output {'1':4}
+Input misras_gries([0,0,0,0,1,1,1,2,2],3)
+Output {'0':4,'1':3}
+Input misras_gries([0,0,0,1,1,1]
+Output None
+"""
+def misras_gries(array,k=2):
+  keys = {}
+  for i in range(len(array)):
+    val = str(array[i])
+    if val in keys:
+      keys[val] = keys[val] + 1
+
+    elif len(keys) < k - 1:
+      keys[val] = 1
+
+    else:
+      for key in list(keys):
+        keys[key] = keys[key] - 1
+        if keys[key] == 0:
+          del keys[key]
+
+  suspects =  keys.keys()
+  frequencies = {}
+  for suspect in suspects:
+    freq = _count_frequency(array,int(suspect))
+    if freq >= len(array) / k:
+      frequencies[suspect] = freq
+
+  return frequencies if len(frequencies) > 0 else None
+
+
+def _count_frequency(array,element):
+  return array.count(element)
+
+

tests/test_streaming.py

@@ -1,7 +1,22 @@
+from algorithms.streaming.misra_gries import (
+    misras_gries,
+)
 from algorithms.streaming import (
     one_sparse
 )
-import unittest   
+import unittest
+
+
+class TestMisraGreis(unittest.TestCase):
+    def test_misra_correct(self):
+        self.assertEqual({'4':5},misras_gries([1,4,4,4,5,4,4]))
+        self.assertEqual({'1':4},misras_gries([0,0,0,1,1,1,1]))
+        self.assertEqual({'0':4,'1':3},misras_gries([0,0,0,0,1,1,1,2,2],3))
+
+    def test_misra_incorrect(self):
+        self.assertEqual(None,misras_gries([1,2,5,4,5,4,4,5,4,4,5]))
+        self.assertEqual(None,misras_gries([0,0,0,2,1,1,1]))
+        self.assertEqual(None,misras_gries([0,0,0,1,1,1]))
 
 class TestOneSparse(unittest.TestCase):
     def test_one_sparse_correct(self):
@@ -12,4 +27,4 @@ def test_one_sparse_correct(self):
     def test_one_sparse_incorrect(self):
         self.assertEqual(None,one_sparse([(2,'+'),(2,'+'),(2,'+'),(2,'+'),(2,'+'),(2,'+'),(1,'+')])) #Two values remaining
         self.assertEqual(None,one_sparse([(2,'+'),(2,'+'),(2,'+'),(2,'+'),(2,'-'),(2,'-'),(2,'-'),(2,'-')])) # No values remaining
-        self.assertEqual(None,one_sparse([(2,'+'),(2,'+'),(4,'+'),(4,'+')])) # Bitsum sum of sign is inccorect
+        self.assertEqual(None,one_sparse([(2,'+'),(2,'+'),(4,'+'),(4,'+')])) # Bitsum sum of sign is inccorect