Merge pull request #6 from rameshrvr/master

New Feature (finding occurrence)

README.rst

@@ -8,6 +8,7 @@ A small Python library which enables:
 
 #. key lookups on deeply nested documents.
 #. fetching all keys from a nested dictionary.
+#. get the number of occurrences of a key/value from a nested dictionary
 
 Documents may be built out of dictionaries (dicts) and/or lists.
 
@@ -50,6 +51,14 @@ quick tutorial
  >>> get_all_keys(document)
  ['taco', 'salsa', 'burrito', 'taco']
 
+ >>> from nested_lookup import get_occurrence_of_key, get_occurrence_of_value
+
+ >>> get_occurrence_of_key(document, key='taco')
+ 2
+
+ >>> get_occurrence_of_value(document, value='42')
+ 1
+
 longer tutorial
 ===============
 
@@ -93,6 +102,20 @@ To get a list of every nested key in a document, run this:
   
   ['name', 'email_address', 'other', 'secondary_email', 'EMAIL_RECOVERY', 'email_address']
 
+To get the number of occurrence of the given key/value
+
+.. code-block:: python
+
+  from nested_lookup import get_occurrence_of_key, get_occurrence_of_value
+
+  no_of_key_occurrence = get_occurrence_of_key(my_document, key='email_address')
+
+  print(no_of_key_occurrence)  # result => 2
+
+  no_of_value_occurrence = get_occurrence_of_value(my_document, value='test2@example.com')
+
+  print(no_of_value_occurrence)  # result => 1
+
 Next, we could act `wild` and find all the email addresses like this:
 
 .. code-block:: python

nested_lookup/__init__.py

@@ -1 +1,2 @@
-from .nested_lookup import nested_lookup, get_all_keys
+from .nested_lookup import nested_lookup, get_all_keys, get_occurrence_of_key,\
+    get_occurrence_of_value

nested_lookup/nested_lookup.py

@@ -62,3 +62,58 @@ def get_all_keys(dictionary):
 
     recrusion(dictionary=dictionary)
     return result_list
+
+
+def get_occurrence_of_key(dictionary, key):
+    """
+    Method to get occurrence of a key in a nested dictionary
+
+    Args:
+        dictionary: Nested dictionary
+        key: Key to search for the occurrences
+    Return:
+        Number of occurrence (Integer)
+    """
+    return _get_occurrence(dictionary=dictionary, item='key', keyword=key)
+
+
+def get_occurrence_of_value(dictionary, value):
+    """
+    Method to get occurrence of a value in a nested dictionary
+
+    Args:
+        dictionary: Nested dictionary
+        value: Value to search for the occurrences
+    Return:
+        Number of occurrence (Integer)
+    """
+    return _get_occurrence(dictionary=dictionary, item='value', keyword=value)
+
+
+def _get_occurrence(dictionary, item, keyword):
+    """
+    Method to get occurrence of a key or value in a nested dictionary
+
+    Args:
+        dictionary: Nested dictionary
+        item: Mostly (key or value)
+        keyword: key word to find occurrence
+    Return:
+        Number of occurrence of the given keyword in the dict
+    """
+    occurrence = [0]
+
+    def recrusion(dictionary):
+        if item == 'key':
+            occurrence[0] += 1 if dictionary.get(keyword) else 0
+        elif keyword in dictionary.values():
+            occurrence[0] += dictionary.values().count(keyword)
+        for key, value in iteritems(dictionary):
+            if isinstance(value, dict):
+                recrusion(dictionary=value)
+            elif isinstance(value, list):
+                for list_items in value:
+                    recrusion(dictionary=list_items)
+
+    recrusion(dictionary=dictionary)
+    return occurrence[0]

setup.py

@@ -13,7 +13,7 @@ with open('requirements.txt', 'r') as f:
 
 setup( 
     name = 'nested-lookup',
-    version = '0.1.6',
+    version = '0.1.7',
     description = 'lookup a key in a deeply nested document of dicts and lists',
     keywords = 'nested document dictionary dict list lookup schema json xml yaml',
     long_description = open('README.rst').read(),

test_nested_loopkup.py

@@ -1,6 +1,7 @@
 from unittest import TestCase
 
-from nested_lookup import nested_lookup, get_all_keys
+from nested_lookup import nested_lookup, get_all_keys, get_occurrence_of_key,\
+    get_occurrence_of_value
 
 
 class TestNestedLookup(TestCase):
@@ -160,5 +161,80 @@ class TestGetAllKeys(TestCase):
             self.assertIn(key, result)
 
 
+class TestGetOccurrence(TestCase):
+    def setUp(self):
+        self.sample1 = {
+            "build_version": {
+                "model_name": "MacBook Pro",
+                "build_version": {
+                    "processor_name": "Intel Core i7",
+                    "processor_speed": "2.7 GHz",
+                    "core_details": {
+                        "build_version": "4",
+                        "l2_cache(per_core)": "256 KB"
+                    }
+                },
+                "number_of_cores": "4",
+                "memory": "256 KB",
+            },
+            "os_details": {
+                "product_version": "10.13.6",
+                "build_version": "17G65"
+            },
+            "name": "Test",
+            "date": "YYYY-MM-DD HH:MM:SS"
+        }
+        self.sample2 = {
+            "hardware_details": {
+                "model_name": "MacBook Pro",
+                "processor_details": [{
+                    "processor_name": "4",
+                    "processor_speed": "2.7 GHz",
+                    "core_details": {
+                        "total_numberof_cores": "4",
+                        "l2_cache(per_core)": "256 KB"
+                    }
+                }],
+                "total_number_of_cores": "4",
+                "memory": "16 GB",
+            }
+        }
+        self.sample3 = {
+            "hardware_details": {
+                "model_name": "MacBook Pro",
+                "processor_details": [
+                    {
+                        "total_number_of_cores": "4",
+                        "processor_speed": "2.7 GHz",
+                    },
+                    {
+                        "total_number_of_cores": "4",
+                        "l2_cache(per_core)": "256 KB"
+                    }
+                ],
+                "total_number_of_cores": "4",
+                "memory": "16 GB",
+            }
+        }
+
+    def test_sample_data1(self):
+        result = get_occurrence_of_key(self.sample1, 'build_version')
+        self.assertEqual(4, result)
+        result = get_occurrence_of_value(self.sample1, '256 KB')
+        self.assertEqual(2, result)
+
+    def test_sample_data2(self):
+        result = get_occurrence_of_key(self.sample2, 'core_details')
+        self.assertEqual(1, result)
+        result = get_occurrence_of_value(self.sample2, '4')
+        self.assertEqual(3, result)
+
+    def test_sample_data3(self):
+        result = get_occurrence_of_key(self.sample3, 'total_number_of_cores')
+        self.assertEqual(3, result)
+        result = get_occurrence_of_value(self.sample3, '4')
+        self.assertEqual(3, result)
+
+
 if __name__ == "__main__":
     pass