From 4db5717854c05f9fdd4cfbb81e8f3ce7efc691d4 Mon Sep 17 00:00:00 2001
From: Jakub Kuczys <me@jacken.men>
Date: Tue, 28 Mar 2023 04:54:55 +0200
Subject: [PATCH] Stop treating files with only line endings changed as
 doc-only

---
 src/label_doconly_changes/base_hooks.py   |  5 +--
 src/label_doconly_changes/hooks/python.py |  5 ++-
 tests/hooks/python_test.py                | 38 ++++++++++++++++++++++-
 3 files changed, 44 insertions(+), 4 deletions(-)

diff --git a/src/label_doconly_changes/base_hooks.py b/src/label_doconly_changes/base_hooks.py
index cbe699e..4b61639 100644
--- a/src/label_doconly_changes/base_hooks.py
+++ b/src/label_doconly_changes/base_hooks.py
@@ -49,11 +49,12 @@ class HookOutputDict(TypedDict):
 
 def _get_file_from_ref(*, ref: str, filename: str) -> str | None:
     try:
+        # this can't use `check_output()`'s encoding or text kwarg
+        # instead of `.decode("utf-8")` because that forces universal newline behavior
         return subprocess.check_output(
             ("git", "cat-file", "blob", f"{ref}:{filename}"),
             stderr=subprocess.PIPE,
-            encoding="utf-8",
-        )
+        ).decode("utf-8")
     except subprocess.CalledProcessError as e:
         prefixes = tuple(
             f"fatal: path '{filename}' {error_msg} '"
diff --git a/src/label_doconly_changes/hooks/python.py b/src/label_doconly_changes/hooks/python.py
index e96f00d..4d8a3ac 100644
--- a/src/label_doconly_changes/hooks/python.py
+++ b/src/label_doconly_changes/hooks/python.py
@@ -19,6 +19,9 @@
 _NodeT = TypeVar("_NodeT", bound=cst.CSTNode)
 _ExprParentT = TypeVar("_ExprParentT")
 _ExprT = TypeVar("_ExprT")
+# ensure consistent default to avoid false positives
+# caused by parser's detection mechanism for this value
+PARSER_CONFIG = cst.PartialParserConfig(default_newline="\n")
 
 
 class DocstringLocation(NamedTuple, Generic[_ExprParentT, _ExprT]):
@@ -69,7 +72,7 @@ def on_visit(self, node: cst.CSTNode) -> bool | None:
 
     @classmethod
     def from_contents(cls, contents: str) -> Self:
-        self = cls(cst.parse_module(contents))
+        self = cls(cst.parse_module(contents, PARSER_CONFIG))
         self.base_node.visit(self)
         return self
 
diff --git a/tests/hooks/python_test.py b/tests/hooks/python_test.py
index 649a942..cef1275 100644
--- a/tests/hooks/python_test.py
+++ b/tests/hooks/python_test.py
@@ -29,7 +29,7 @@ def leave_List(
 
 def simplify_list(nodes: list[cst.CSTNode], *, max_depth: int = 1) -> str:
     visitor = _ListSimplifier(max_depth=max_depth)
-    return cst.parse_module(repr(nodes)).visit(visitor).code
+    return cst.parse_module(repr(nodes), python.PARSER_CONFIG).visit(visitor).code
 
 
 def print_extractor_info(tracker: python.ModuleTracker) -> None:
@@ -72,3 +72,39 @@ def test_is_doc_only_false(contents_before: str, contents_after: str) -> None:
     except Exception:
         print_analyzer_info(analyzer)
         raise
+
+
+@pytest.mark.parametrize(
+    "contents_before,contents_after",
+    (
+        (
+            "x = 1\ny = 2\n",
+            "x = 1\r\ny = 2\r\n",
+        ),
+        # LibCST's native parser panics with this one :D
+        #
+        # (
+        #     "x = 1\ny = 2\n",
+        #     "x = 1\ry = 2\r\n",
+        # ),
+        (
+            "x = 1\ny = 2\n",
+            "x = 1\ry = 2\r",
+        ),
+        (
+            "x = 1\ny = 2\n",
+            "x = 1\ny = 2\r\n",
+        ),
+        (
+            "x = 1\r\ny = 2\n",
+            "x = 1\ny = 2\r\n",
+        ),
+    ),
+)
+def test_different_line_endings(contents_before: str, contents_after: str) -> None:
+    analyzer = python.PythonAnalyzer(contents_before, contents_after)
+    try:
+        assert not analyzer.is_docstring_only()
+    except Exception:
+        print_analyzer_info(analyzer)
+        raise