feat: final update to RAG generation

Cloud-Code-AI · Aug 16, 2024 · ad403d9 · ad403d9
1 parent d5ebb47
commit ad403d9
Show file tree

Hide file tree

Showing 22 changed files with 1,298 additions and 630 deletions.
diff --git a/.gitignore b/.gitignore
@@ -164,4 +164,5 @@ cython_debug/
 node_modules
 .next
 
-.cloudcode
+.cloudcode
+tree_sitter_languages/
diff --git a/.kaizen/unit_test/kaizen/helpers/test_create_folder.py b/.kaizen/unit_test/kaizen/helpers/test_create_folder.py
@@ -6,33 +6,44 @@
 # Mock logger
 logger = mock.Mock()
 
+
 @pytest.fixture
 def mock_os_path_exists():
-    with mock.patch('os.path.exists') as mock_exists:
+    with mock.patch("os.path.exists") as mock_exists:
         yield mock_exists
 
+
 @pytest.fixture
 def mock_os_makedirs():
-    with mock.patch('os.makedirs') as mock_makedirs:
+    with mock.patch("os.makedirs") as mock_makedirs:
         yield mock_makedirs
 
+
 @pytest.fixture
 def mock_logger_debug():
-    with mock.patch('kaizen.helpers.output.logger.debug') as mock_debug:
+    with mock.patch("kaizen.helpers.output.logger.debug") as mock_debug:
         yield mock_debug
 
-def test_create_new_folder_when_not_exists(mock_os_path_exists, mock_os_makedirs, mock_logger_debug):
-    folder_path = 'new_folder'
+
+def test_create_new_folder_when_not_exists(
+    mock_os_path_exists, mock_os_makedirs, mock_logger_debug
+):
+    folder_path = "new_folder"
     mock_os_path_exists.return_value = False
 
     create_folder(folder_path)
 
     mock_os_path_exists.assert_called_once_with(folder_path)
     mock_os_makedirs.assert_called_once_with(folder_path)
-    mock_logger_debug.assert_called_once_with(f"Folder '{folder_path}' created successfully.")
+    mock_logger_debug.assert_called_once_with(
+        f"Folder '{folder_path}' created successfully."
+    )
+
 
-def test_do_nothing_when_folder_already_exists(mock_os_path_exists, mock_os_makedirs, mock_logger_debug):
-    folder_path = 'existing_folder'
+def test_do_nothing_when_folder_already_exists(
+    mock_os_path_exists, mock_os_makedirs, mock_logger_debug
+):
+    folder_path = "existing_folder"
     mock_os_path_exists.return_value = True
 
     create_folder(folder_path)
@@ -41,52 +52,71 @@ def test_do_nothing_when_folder_already_exists(mock_os_path_exists, mock_os_make
     mock_os_makedirs.assert_not_called()
     mock_logger_debug.assert_called_once_with(f"Folder '{folder_path}' already exists.")
 
+
 def test_raise_value_error_when_folder_path_is_empty():
     with pytest.raises(ValueError, match="Folder path cannot be empty"):
-        create_folder('')
+        create_folder("")
+
 
-def test_create_deeply_nested_folder(mock_os_path_exists, mock_os_makedirs, mock_logger_debug):
-    folder_path = 'a/b/c/d/e/f/g'
+def test_create_deeply_nested_folder(
+    mock_os_path_exists, mock_os_makedirs, mock_logger_debug
+):
+    folder_path = "a/b/c/d/e/f/g"
     mock_os_path_exists.return_value = False
 
     create_folder(folder_path)
 
     mock_os_path_exists.assert_called_once_with(folder_path)
     mock_os_makedirs.assert_called_once_with(folder_path)
-    mock_logger_debug.assert_called_once_with(f"Folder '{folder_path}' created successfully.")
+    mock_logger_debug.assert_called_once_with(
+        f"Folder '{folder_path}' created successfully."
+    )
+
 
-def test_create_folder_with_special_characters(mock_os_path_exists, mock_os_makedirs, mock_logger_debug):
-    folder_path = 'folder_with_special_!@#$%^&*()'
+def test_create_folder_with_special_characters(
+    mock_os_path_exists, mock_os_makedirs, mock_logger_debug
+):
+    folder_path = "folder_with_special_!@#$%^&*()"
     mock_os_path_exists.return_value = False
 
     create_folder(folder_path)
 
     mock_os_path_exists.assert_called_once_with(folder_path)
     mock_os_makedirs.assert_called_once_with(folder_path)
-    mock_logger_debug.assert_called_once_with(f"Folder '{folder_path}' created successfully.")
+    mock_logger_debug.assert_called_once_with(
+        f"Folder '{folder_path}' created successfully."
+    )
 
-def test_create_folder_with_max_path_length(mock_os_path_exists, mock_os_makedirs, mock_logger_debug):
+
+def test_create_folder_with_max_path_length(
+    mock_os_path_exists, mock_os_makedirs, mock_logger_debug
+):
     # Adjusting the max path length to a more typical value for modern filesystems
-    max_path_length = os.pathconf('/', 'PC_PATH_MAX')
-    folder_path = 'a' * max_path_length
+    max_path_length = os.pathconf("/", "PC_PATH_MAX")
+    folder_path = "a" * max_path_length
     mock_os_path_exists.return_value = False
 
     create_folder(folder_path)
 
     mock_os_path_exists.assert_called_once_with(folder_path)
     mock_os_makedirs.assert_called_once_with(folder_path)
-    mock_logger_debug.assert_called_once_with(f"Folder '{folder_path}' created successfully.")
+    mock_logger_debug.assert_called_once_with(
+        f"Folder '{folder_path}' created successfully."
+    )
+
 
-def test_create_folder_with_invalid_characters(mock_os_path_exists, mock_os_makedirs, mock_logger_debug):
+def test_create_folder_with_invalid_characters(
+    mock_os_path_exists, mock_os_makedirs, mock_logger_debug
+):
     # Assuming the filesystem does not allow characters like ':', '*', '?', '<', '>', '|'
-    invalid_characters = [':', '*', '?', '<', '>', '|']
+    invalid_characters = [":", "*", "?", "<", ">", "|"]
     for char in invalid_characters:
-        folder_path = f'invalid{char}folder'
+        folder_path = f"invalid{char}folder"
         mock_os_path_exists.return_value = False
 
         with pytest.raises(OSError):
             create_folder(folder_path)
 
         mock_os_path_exists.assert_called_once_with(folder_path)
         mock_os_makedirs.assert_not_called()
-        mock_logger_debug.assert_not_called()
+        mock_logger_debug.assert_not_called()
diff --git a/.kaizen/unit_test/kaizen/helpers/test_create_pr_description.py b/.kaizen/unit_test/kaizen/helpers/test_create_pr_description.py
@@ -2,51 +2,91 @@
 import time
 from kaizen.helpers.output import create_pr_description
 
-DESC_COLLAPSIBLE_TEMPLATE = "<details><summary>Original Description</summary>\n\n{desc}\n\n</details>"
+DESC_COLLAPSIBLE_TEMPLATE = (
+    "<details><summary>Original Description</summary>\n\n{desc}\n\n</details>"
+)
 
-@pytest.mark.parametrize("desc, original_desc, expected", [
-    # Normal Cases
-    ("This is a PR description", "This is the original detailed description",
-     "This is a PR description\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\nThis is the original detailed description\n\n</details>"),
-    ("Fixes a bug", "This fixes a bug in the system",
-     "Fixes a bug\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\nThis fixes a bug in the system\n\n</details>"),
-    # Edge Cases
-    ("", "This is the original detailed description",
-     "\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\nThis is the original detailed description\n\n</details>"),
-    ("This is a PR description", "",
-     "This is a PR description\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n\n\n</details>"),
-    ("", "",
-     "\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n\n\n</details>"),
-    ("# Heading\n* Bullet", "**Bold**\n_Italic_",
-     "# Heading\n* Bullet\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n**Bold**\n_Italic_\n\n</details>"),
-    # Special Characters and HTML Tags
-    ("<h1>Title</h1>", "<p>This is a <strong>bold</strong> statement</p>",
-     "<h1>Title</h1>\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n<p>This is a <strong>bold</strong> statement</p>\n\n</details>"),
-    ("Special characters: !@#$%^&*()", "More special characters: ~`<>?",
-     "Special characters: !@#$%^&*()\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\nMore special characters: ~`<>?\n\n</details>"),
-])
+
+@pytest.mark.parametrize(
+    "desc, original_desc, expected",
+    [
+        # Normal Cases
+        (
+            "This is a PR description",
+            "This is the original detailed description",
+            "This is a PR description\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\nThis is the original detailed description\n\n</details>",
+        ),
+        (
+            "Fixes a bug",
+            "This fixes a bug in the system",
+            "Fixes a bug\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\nThis fixes a bug in the system\n\n</details>",
+        ),
+        # Edge Cases
+        (
+            "",
+            "This is the original detailed description",
+            "\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\nThis is the original detailed description\n\n</details>",
+        ),
+        (
+            "This is a PR description",
+            "",
+            "This is a PR description\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n\n\n</details>",
+        ),
+        (
+            "",
+            "",
+            "\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n\n\n</details>",
+        ),
+        (
+            "# Heading\n* Bullet",
+            "**Bold**\n_Italic_",
+            "# Heading\n* Bullet\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n**Bold**\n_Italic_\n\n</details>",
+        ),
+        # Special Characters and HTML Tags
+        (
+            "<h1>Title</h1>",
+            "<p>This is a <strong>bold</strong> statement</p>",
+            "<h1>Title</h1>\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n<p>This is a <strong>bold</strong> statement</p>\n\n</details>",
+        ),
+        (
+            "Special characters: !@#$%^&*()",
+            "More special characters: ~`<>?",
+            "Special characters: !@#$%^&*()\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\nMore special characters: ~`<>?\n\n</details>",
+        ),
+    ],
+)
 def test_create_pr_description_normal_and_edge_cases(desc, original_desc, expected):
     assert create_pr_description(desc, original_desc) == expected
 
-@pytest.mark.parametrize("desc, original_desc, expected_error_message", [
-    # Error Handling
-    (None, "This is the original detailed description", "desc must be a string"),
-    (123, "This is the original detailed description", "desc must be a string"),
-    ([], "This is the original detailed description", "desc must be a string"),
-    ("This is a PR description", None, "original_desc must be a string"),
-    ("This is a PR description", 123, "original_desc must be a string"),
-    ("This is a PR description", [], "original_desc must be a string"),
-])
-def test_create_pr_description_error_handling(desc, original_desc, expected_error_message):
+
+@pytest.mark.parametrize(
+    "desc, original_desc, expected_error_message",
+    [
+        # Error Handling
+        (None, "This is the original detailed description", "desc must be a string"),
+        (123, "This is the original detailed description", "desc must be a string"),
+        ([], "This is the original detailed description", "desc must be a string"),
+        ("This is a PR description", None, "original_desc must be a string"),
+        ("This is a PR description", 123, "original_desc must be a string"),
+        ("This is a PR description", [], "original_desc must be a string"),
+    ],
+)
+def test_create_pr_description_error_handling(
+    desc, original_desc, expected_error_message
+):
     with pytest.raises(TypeError) as exc_info:
         create_pr_description(desc, original_desc)
     assert str(exc_info.value) == expected_error_message
 
-@pytest.mark.parametrize("desc, original_desc", [
-    # Boundary Conditions
-    ("a" * 10000, "b" * 10000),
-    ("a" * 100000, "b" * 100000),
-])
+
+@pytest.mark.parametrize(
+    "desc, original_desc",
+    [
+        # Boundary Conditions
+        ("a" * 10000, "b" * 10000),
+        ("a" * 100000, "b" * 100000),
+    ],
+)
 def test_create_pr_description_boundary_conditions(desc, original_desc):
     start_time = time.time()
     result = create_pr_description(desc, original_desc)
@@ -56,9 +96,18 @@ def test_create_pr_description_boundary_conditions(desc, original_desc):
     assert result.startswith(desc)
     assert result.endswith(DESC_COLLAPSIBLE_TEMPLATE.format(desc=original_desc))
     assert "> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️" in result
-    assert len(result) == len(desc) + len(original_desc) + len("\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n\n\n</details>") - 2
+    assert (
+        len(result)
+        == len(desc)
+        + len(original_desc)
+        + len(
+            "\n\n> ✨ Generated with love by [Kaizen](https://cloudcode.ai) ❤️\n\n<details><summary>Original Description</summary>\n\n\n\n</details>"
+        )
+        - 2
+    )
     # Removed the arbitrary 1-second boundary condition
     print(f"Execution time: {execution_time} seconds")
 
+
 if __name__ == "__main__":
-    pytest.main()
+    pytest.main()
diff --git a/.kaizen/unit_test/kaizen/helpers/test_create_pr_review_text.py b/.kaizen/unit_test/kaizen/helpers/test_create_pr_review_text.py
@@ -15,6 +15,7 @@
 </details>
 """
 
+
 @pytest.fixture
 def setup_single_topic_single_review():
     return {
@@ -32,6 +33,7 @@ def setup_single_topic_single_review():
         ]
     }
 
+
 @pytest.fixture
 def setup_multiple_topics_multiple_reviews():
     return {
@@ -55,7 +57,7 @@ def setup_multiple_topics_multiple_reviews():
                 "end_line": 40,
                 "file_name": "file2.py",
                 "severity_level": 7,
-            }
+            },
         ],
         "topic2": [
             {
@@ -68,14 +70,18 @@ def setup_multiple_topics_multiple_reviews():
                 "file_name": "file3.py",
                 "severity_level": 5,
             }
-        ]
+        ],
     }
 
+
 def test_empty_topics():
     topics = {}
-    expected_output = "## Code Review\n\n✅ **All Clear:** This PR is ready to merge! 👍\n\n"
+    expected_output = (
+        "## Code Review\n\n✅ **All Clear:** This PR is ready to merge! 👍\n\n"
+    )
     assert create_pr_review_text(topics) == expected_output
 
+
 def test_single_topic_single_review(setup_single_topic_single_review):
     topics = setup_single_topic_single_review
     expected_output = (
@@ -96,6 +102,7 @@ def test_single_topic_single_review(setup_single_topic_single_review):
     )
     assert create_pr_review_text(topics) == expected_output
 
+
 def test_multiple_topics_multiple_reviews(setup_multiple_topics_multiple_reviews):
     topics = setup_multiple_topics_multiple_reviews
     expected_output = (
@@ -139,6 +146,7 @@ def test_multiple_topics_multiple_reviews(setup_multiple_topics_multiple_reviews
     )
     assert create_pr_review_text(topics) == expected_output
 
+
 def test_reviews_with_missing_fields():
     topics = {
         "topic1": [
@@ -181,7 +189,7 @@ def test_reviews_with_missing_fields():
                 "end_line": 80,
                 "file_name": "final_test_file.py",
                 # Missing severity_level
-            }
+            },
         ]
     }
     expected_output = (
@@ -235,6 +243,7 @@ def test_reviews_with_missing_fields():
     )
     assert create_pr_review_text(topics) == expected_output
 
+
 def test_reviews_with_missing_comment():
     topics = {
         "topic1": [
@@ -268,9 +277,10 @@ def test_reviews_with_missing_comment():
     )
     assert create_pr_review_text(topics) == expected_output
 
+
 def test_empty_list_in_topics():
-    topics = {
-        "topic1": []
-    }
-    expected_output = "## Code Review\n\n✅ **All Clear:** This PR is ready to merge! 👍\n\n"
-    assert create_pr_review_text(topics) == expected_output
+    topics = {"topic1": []}
+    expected_output = (
+        "## Code Review\n\n✅ **All Clear:** This PR is ready to merge! 👍\n\n"
+    )
+    assert create_pr_review_text(topics) == expected_output