Implemented interface ot csv.DictReader for #92

Author: EntilZha

functional/streams.py

@@ -136,6 +136,20 @@ def csv(self, csv_file, dialect='excel', **fmt_params):
         csv_input = csvapi.reader(input_file, dialect=dialect, **fmt_params)
         return self(csv_input).cache(delete_lineage=True)
 
+    def csv_dict_reader(self, csv_file, fieldnames=None, restkey=None, restval=None,
+                        dialect='excel', **kwds):
+        if isinstance(csv_file, str):
+            file_open = get_read_function(csv_file, self.disable_compression)
+            input_file = file_open(csv_file)
+        elif hasattr(csv_file, 'next') or hasattr(csv_file, '__next__'):
+            input_file = csv_file
+        else:
+            raise ValueError('csv_file must be a file path or implement the iterator interface')
+
+        csv_input = csvapi.DictReader(input_file, fieldnames=fieldnames, restkey=restkey,
+                                      restval=restval, dialect=dialect, **kwds)
+        return self(csv_input).cache(delete_lineage=True)
+
     def jsonl(self, jsonl_file):
         """
         Reads and parses the input of a jsonl file stream or file.

functional/test/data/test_header.csv

@@ -0,0 +1,3 @@
+a,b,c
+1,2,3
+4,5,6

functional/test/test_streams.py

@@ -76,6 +76,27 @@ def test_csv(self):
         with self.assertRaises(ValueError):
             self.seq.csv(1)
 
+    def test_csv_dict_reader(self):
+        result = self.seq.csv_dict_reader('functional/test/data/test_header.csv').to_list()
+        self.assertEqual(result[0]['a'], '1')
+        self.assertEqual(result[0]['b'], '2')
+        self.assertEqual(result[0]['c'], '3')
+        self.assertEqual(result[1]['a'], '4')
+        self.assertEqual(result[1]['b'], '5')
+        self.assertEqual(result[1]['c'], '6')
+
+        with open('functional/test/data/test_header.csv', 'r') as f:
+            result = self.seq.csv_dict_reader(f).to_list()
+        self.assertEqual(result[0]['a'], '1')
+        self.assertEqual(result[0]['b'], '2')
+        self.assertEqual(result[0]['c'], '3')
+        self.assertEqual(result[1]['a'], '4')
+        self.assertEqual(result[1]['b'], '5')
+        self.assertEqual(result[1]['c'], '6')
+
+        with self.assertRaises(ValueError):
+            self.seq.csv_dict_reader(1)
+
     def test_gzip_csv(self):
         result = self.seq.csv('functional/test/data/test.csv.gz').to_list()
         expect = [['1', '2', '3', '4'], ['a', 'b', 'c', 'd']]