From 2374cf77c20530c3b85eab72784a958c2cf41e5b Mon Sep 17 00:00:00 2001
From: Tamir Duberstein <tamird@gmail.com>
Date: Fri, 29 Mar 2024 15:51:51 +0000
Subject: [PATCH] Initialize cache before writing data

Writing cache data is interruptible; this reduces the probability of a
pathological case where interrupting a cache write can cause the cache
directory to never be properly initialized with its supporting files.

Unify `Cache.mkdir` with `Cache.set` while I'm here so the former also
properly initializes the cache directory.

Closes #12167.
---
 changelog/12167.trivial.rst  |  1 +
 src/_pytest/cacheprovider.py | 44 ++++++++++++++++++++++--------------
 2 files changed, 28 insertions(+), 17 deletions(-)
 create mode 100644 changelog/12167.trivial.rst

diff --git a/changelog/12167.trivial.rst b/changelog/12167.trivial.rst
new file mode 100644
index 00000000000..305c6ca3d95
--- /dev/null
+++ b/changelog/12167.trivial.rst
@@ -0,0 +1 @@
+cache: create supporting files (``CACHEDIR.TAG``, ``.gitignore``, etc.) in the cache directory before writing cache data, reducing the probability of those files being lost.
diff --git a/src/_pytest/cacheprovider.py b/src/_pytest/cacheprovider.py
index 81703ddac44..1f62baaa1eb 100755
--- a/src/_pytest/cacheprovider.py
+++ b/src/_pytest/cacheprovider.py
@@ -14,6 +14,7 @@
 from typing import List
 from typing import Optional
 from typing import Set
+from typing import Tuple
 from typing import Union
 
 from .pathlib import resolve_from_str
@@ -21,6 +22,7 @@
 from .reports import CollectReport
 from _pytest import nodes
 from _pytest._io import TerminalWriter
+from _pytest.compat import assert_never
 from _pytest.config import Config
 from _pytest.config import ExitCode
 from _pytest.config import hookimpl
@@ -123,6 +125,10 @@ def warn(self, fmt: str, *, _ispytest: bool = False, **args: object) -> None:
             stacklevel=3,
         )
 
+    def _mkdir(self, path: Path) -> None:
+        self._ensure_cache_dir_and_supporting_files()
+        path.mkdir(exist_ok=True, parents=True)
+
     def mkdir(self, name: str) -> Path:
         """Return a directory path object with the given name.
 
@@ -141,7 +147,7 @@ def mkdir(self, name: str) -> Path:
         if len(path.parts) > 1:
             raise ValueError("name is not allowed to contain path separators")
         res = self._cachedir.joinpath(self._CACHE_PREFIX_DIRS, path)
-        res.mkdir(exist_ok=True, parents=True)
+        self._mkdir(res)
         return res
 
     def _getvaluepath(self, key: str) -> Path:
@@ -178,19 +184,13 @@ def set(self, key: str, value: object) -> None:
         """
         path = self._getvaluepath(key)
         try:
-            if path.parent.is_dir():
-                cache_dir_exists_already = True
-            else:
-                cache_dir_exists_already = self._cachedir.exists()
-                path.parent.mkdir(exist_ok=True, parents=True)
+            self._mkdir(path.parent)
         except OSError as exc:
             self.warn(
                 f"could not create cache path {path}: {exc}",
                 _ispytest=True,
             )
             return
-        if not cache_dir_exists_already:
-            self._ensure_supporting_files()
         data = json.dumps(value, ensure_ascii=False, indent=2)
         try:
             f = path.open("w", encoding="UTF-8")
@@ -203,17 +203,27 @@ def set(self, key: str, value: object) -> None:
             with f:
                 f.write(data)
 
-    def _ensure_supporting_files(self) -> None:
-        """Create supporting files in the cache dir that are not really part of the cache."""
-        readme_path = self._cachedir / "README.md"
-        readme_path.write_text(README_CONTENT, encoding="UTF-8")
+    def _ensure_cache_dir_and_supporting_files(self) -> None:
+        """Create the cache dir and its supporting files."""
+        if self._cachedir.is_dir():
+            return
 
-        gitignore_path = self._cachedir.joinpath(".gitignore")
-        msg = "# Created by pytest automatically.\n*\n"
-        gitignore_path.write_text(msg, encoding="UTF-8")
+        self._cachedir.mkdir(exist_ok=True, parents=True)
 
-        cachedir_tag_path = self._cachedir.joinpath("CACHEDIR.TAG")
-        cachedir_tag_path.write_bytes(CACHEDIR_TAG_CONTENT)
+        files: Iterable[Tuple[str, Union[str, bytes]]] = (
+            ("README.md", README_CONTENT),
+            (".gitignore", "# Created by pytest automatically.\n*\n"),
+            ("CACHEDIR.TAG", CACHEDIR_TAG_CONTENT),
+        )
+        for file, content in files:
+            if isinstance(content, str):
+                with open(self._cachedir.joinpath(file), "xt", encoding="UTF-8") as f:
+                    f.write(content)
+            elif isinstance(content, bytes):
+                with open(self._cachedir.joinpath(file), "xb") as f:
+                    f.write(content)
+            else:
+                assert_never(content)
 
 
 class LFPluginCollWrapper: