added second test for file_tokenizer

backend/code/file_tokenizer.py

@@ -2,7 +2,6 @@
 This file helps in creating and interpreting a .hackthehill file
 """
 
-import hashlib
 import os
 import json
 
@@ -45,10 +44,10 @@ def hash_file_blocks(file_path: str, block_size: int = 512):
 
     block_hashes = {}
 
-    with open(file_path, 'rb') as file:
+    with open(file_path, "r", encoding="utf-8") as file:
         for index in range(num_blocks):
             block = file.read(block_size)
-            block_hash = hashlib.sha256(block).hexdigest()
+            block_hash = custom_hash(block)
             block_hashes[index] = block_hash
 
     header["blocks"] = block_hashes

backend/test/test_file_tokenizer.py

@@ -22,14 +22,9 @@ def test_hash_file_blocks_with_empty_file_returns_correct_value(self):
         We should only see the header.
         """
 
-        testing_file = os.path.join(
-            UPLOADS_FOLDER,
-            "test_hash_file_blocks_with_empty_file_returns_correct_value.txt"
-        )
-        hackthehill_file = os.path.join(
-            SOURCES_FOLDER,
-            "test_hash_file_blocks_with_empty_file_returns_correct_value" + HASH_EXTENSION
-        )
+        function_name = "test_hash_file_blocks_with_empty_file_returns_correct_value"
+        testing_file = os.path.join(UPLOADS_FOLDER, f"{function_name}.txt")
+        hackthehill_file = os.path.join(SOURCES_FOLDER, function_name + HASH_EXTENSION)
         test_header = {
             "header": {
                 "file_name": os.path.basename(testing_file),
@@ -52,3 +47,49 @@ def test_hash_file_blocks_with_empty_file_returns_correct_value(self):
 
         os.remove(testing_file)
         os.remove(hackthehill_file)
+
+    def test_hash_file_blocks_with_non_empty_file_returns_correct_value(self):
+        """
+        We should see the correct header plus the correct hashed blocks
+        """
+
+        function_name = "test_hash_file_blocks_with_non_empty_file_returns_correct_value"
+        testing_file = os.path.join(UPLOADS_FOLDER, f"{function_name}.txt")
+        hackthehill_file = os.path.join(SOURCES_FOLDER, function_name + HASH_EXTENSION)
+        block_size = 1024
+        test_header = {
+            "header": {
+                "file_name": os.path.basename(testing_file),
+                "file_size": 0,
+                "number_of_blocks": 0,
+                "block_size": block_size,
+            },
+            "blocks": {}
+        }
+
+        with open(testing_file, "x", encoding="utf-8") as f:
+            f.write(function_name)
+
+        file_size = os.path.getsize(testing_file)
+        num_blocks = (file_size + block_size - 1) // block_size
+
+        test_header["header"]["file_size"] = file_size
+
+        with open(testing_file, "r", encoding="utf-8") as f:
+            hackthehill_file_hashed_content = hash_file_blocks(testing_file, block_size)
+
+            for index in range(num_blocks):
+                block = f.read(block_size)
+                block_hash = custom_hash(block)
+                test_header["blocks"][str(index)] = str(block_hash)
+                test_header["header"]["number_of_blocks"] += 1
+
+            with open(hackthehill_file, "r", encoding="utf-8") as g:
+                hackthehill_file_content = json.loads(g.read())
+
+                self.assertEqual(hackthehill_file_content, test_header)
+                self.assertEqual(hackthehill_file_hashed_content,
+                                 custom_hash(json.dumps(hackthehill_file_content)))
+
+        os.remove(testing_file)
+        os.remove(hackthehill_file)

backend/test/test_utils.py

@@ -56,13 +56,11 @@ def test_get_filename_by_file_id_with_no_matching_id_returns_none(self):
         If none of the hashed ids match with the file id, we should return None
         """
 
-        testing_file = os.path.join(
-            UPLOADS_FOLDER,
-            "test_get_filename_by_file_id_with_no_matching_id_returns_none.txt")
-        message = "test_get_filename_by_file_id_with_no_matching_id_returns_none"
+        function_name = "test_get_filename_by_file_id_with_no_matching_id_returns_none"
+        testing_file = os.path.join(UPLOADS_FOLDER, f"{function_name}.txt")
 
         with open(testing_file, "x", encoding="utf-8") as f:
-            f.write(message)
+            f.write(function_name)
             file_id = "random-file-id"
 
             self.assertEqual(get_filename_by_file_id(file_id), None)
@@ -75,17 +73,12 @@ def test_get_filename_by_file_id_with_matching_id_returns_tuple(self):
         tuple of filenames
         """
 
-        testing_file = os.path.join(
-            UPLOADS_FOLDER,
-            "test_get_filename_by_file_id_with_matching_id_returns_tuple.txt")
-        hackthehill_file = os.path.join(
-            SOURCES_FOLDER,
-            "test_get_filename_by_file_id_with_matching_id_returns_tuple" + HASH_EXTENSION)
-
-        message = "test_get_filename_by_file_id_with_matching_id_returns_tuple"
+        function_name = "test_get_filename_by_file_id_with_matching_id_returns_tuple"
+        testing_file = os.path.join(UPLOADS_FOLDER, f"{function_name}.txt")
+        hackthehill_file = os.path.join(SOURCES_FOLDER, function_name + HASH_EXTENSION)
 
         with open(testing_file, "x", encoding="utf-8") as f:
-            f.write(message)
+            f.write(function_name)
             hash_file_blocks(testing_file)
 
             with open(hackthehill_file, "r", encoding="utf-8") as g: