Add new procedure that computes the hash of an array of strings

2 files changed, 17 insertions, 3 deletions

diff --git a/gn3/computations/gemma.py b/gn3/computations/gemma.py
index 790aecb..ea0f86a 100644
--- a/gn3/computations/gemma.py
+++ b/gn3/computations/gemma.py
@@ -2,6 +2,8 @@
 import random
 import string
 
+from base64 import b64encode
+from hashlib import md5
 
 def generate_random_n_string(n_length: int) -> str:
     """Generate a random string that is N chars long"""
@@ -9,6 +11,11 @@ def generate_random_n_string(n_length: int) -> str:
                    for _ in range(n_length))
 
 
+def generate_hash_of_string(unhashed_str: str) -> str:
+    """Given an UNHASHED_STRING, generate it's md5 hash while removing the '==' at
+the end"""
+    hashed_str = md5(unhashed_str.encode("utf-8")).digest()
+    return b64encode(hashed_str).decode("utf-8").replace("==", "")
 def generate_pheno_txt_file(trait_filename: str,
                             values: str,
                             tmpdir: str = "/tmp") -> str:
diff --git a/tests/unit/computations/test_gemma.py b/tests/unit/computations/test_gemma.py
index 89c8d00..23707ef 100644
--- a/tests/unit/computations/test_gemma.py
+++ b/tests/unit/computations/test_gemma.py
@@ -2,12 +2,12 @@
 import unittest
 
 from unittest import mock
+from gn3.computations.gemma import generate_hash_of_string
 from gn3.computations.gemma import generate_pheno_txt_file
 
 
 class TestGemma(unittest.TestCase):
     """Test cases for computations.gemma module"""
-
     def test_generate_pheno_txt_file(self):
         """Test that the pheno text file is generated correctly"""
         open_mock = mock.mock_open()
@@ -16,9 +16,16 @@ class TestGemma(unittest.TestCase):
             _file = generate_pheno_txt_file(tmpdir="/tmp",
                                             trait_filename="phenotype.txt",
                                             values=["x", "x", "BXD07 438.700"])
-            self.assertEqual(_file, "/tmp/gn2/phenotype.txt")
-        open_mock.assert_called_with("/tmp/gn2/phenotype.txt", "w")
+            self.assertEqual(_file, ("/tmp/gn2/phenotype_"
+                                     "P7y6QWnwBPedSZdL0+m/GQ.txt"))
+        open_mock.assert_called_with(("/tmp/gn2/phenotype_"
+                                      "P7y6QWnwBPedSZdL0+m/GQ.txt"), "w")
         open_mock.return_value.write.assert_has_calls([
             mock.call("NA\n"),
             mock.call("NA\n"),
             mock.call("BXD07 438.700\n")])
+
+    def test_generate_hash_of_string(self):
+        """Test that a string is hashed correctly"""
+        self.assertEqual(generate_hash_of_string("I^iQP&TlSR^z"),
+                         "hMVRw8kbEp49rOmoIkhMjA")