Fix the stage diff algorithm (#1031)

RELEASE-NOTES.md

@@ -9,6 +9,7 @@
 
 ## Fixes and improvements
 * More human-friendly errors in case of corrupted `config.toml` file.
+* Fixed a bug in `snow app` that caused files to be re-uploaded unnecessarily.
 
 # v2.2.0
 

src/snowflake/cli/plugins/stage/diff.py

@@ -7,7 +7,7 @@
 
 from snowflake.cli.api.exceptions import SnowflakeSQLExecutionError
 from snowflake.cli.api.secure_path import UNLIMITED, SecurePath
-from snowflake.connector.cursor import SnowflakeCursor
+from snowflake.connector.cursor import DictCursor
 
 from .manager import StageManager
 
@@ -144,13 +144,13 @@ def strip_stage_name(path: str) -> str:
     return "/".join(path.split("/")[1:])
 
 
-def build_md5_map(list_stage_cursor: SnowflakeCursor) -> Dict[str, str]:
+def build_md5_map(list_stage_cursor: DictCursor) -> Dict[str, str]:
     """
     Returns a mapping of relative stage paths to their md5sums.
     """
     return {
-        strip_stage_name(name): md5
-        for (name, size, md5, modified) in list_stage_cursor.fetchall()
+        strip_stage_name(file["name"]): file["md5"]
+        for file in list_stage_cursor.fetchall()
     }
 
 

tests/stage/test_diff.py

@@ -1,12 +1,13 @@
 import hashlib
 from pathlib import Path
-from typing import Dict, List, Tuple, Union
+from typing import Dict, List, Union
 from unittest import mock
 
 import pytest
 from snowflake.cli.api.exceptions import SnowflakeSQLExecutionError
 from snowflake.cli.plugins.stage.diff import (
     DiffResult,
+    build_md5_map,
     delete_only_on_stage_files,
     enumerate_files,
     get_stage_path_from_file,
@@ -40,13 +41,18 @@ def md5_of(contents: Union[str, bytes]) -> str:
 
 def stage_contents(
     files: Dict[str, Union[str, bytes]], last_modified: str = DEFAULT_LAST_MODIFIED
-) -> List[Tuple[str, int, str, str]]:
+) -> List[Dict[str, Union[str, int]]]:
     """
     Return file contents as they would be listed by a SNOWFLAKE_SSE stage
     if they were uploaded with the given structure and contents.
     """
     return [
-        (f"stage/{relpath}", len(contents), md5_of(contents), last_modified)
+        {
+            "name": f"stage/{relpath}",
+            "size": len(contents),
+            "md5": md5_of(contents),
+            "last_modified": last_modified,
+        }
         for (relpath, contents) in files.items()
     ]
 
@@ -199,6 +205,23 @@ def test_put_files_on_stage(mock_put, overwrite_param):
         assert mock_put.mock_calls == expected
 
 
+def test_build_md5_map(mock_cursor):
+    actual = build_md5_map(
+        mock_cursor(
+            rows=stage_contents(FILE_CONTENTS),
+            columns=STAGE_LS_COLUMNS,
+        )
+    )
+
+    expected = {
+        "README.md": "9b650974f65cc49be96a5ed34ac6d1fd",
+        "my.jar": "fc605d0e2e50cf3e71873d57f4c598b0",
+        "ui/streamlit.py": "a7dfdfaf892ecfc5f164914123c7f2cc",
+    }
+
+    assert actual == expected
+
+
 @mock.patch(f"{STAGE_MANAGER}.remove")
 def test_sync_local_diff_with_stage(mock_remove, other_directory):
     temp_dir = Path(other_directory)

tests_integration/test_nativeapp.py

@@ -550,7 +550,15 @@ def test_nativeapp_init_deploy(
                 dict(name=app_name),
             )
 
-            # make sure we always delete the app
+            # re-deploying should be a no-op; make sure we don't issue any PUT commands
+            result = runner.invoke_with_connection_json(
+                ["app", "deploy", "--debug"],
+                env=TEST_ENV,
+            )
+            assert result.exit_code == 0
+            assert "Successfully uploaded chunk 0 of file" not in result.output
+
+            # make sure we always delete the package
             result = runner.invoke_with_connection_json(
                 ["app", "teardown"],
                 env=TEST_ENV,