From 1ca4529dc02902830be2a668a4e598f22f154071 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?St=C3=A9phane=20Bidoul?= <stephane.bidoul@gmail.com>
Date: Mon, 29 May 2023 12:34:21 +0200
Subject: [PATCH] Ignore invalid origin.json in wheel cache

---
 news/11985.bugfix.rst      |  1 +
 src/pip/_internal/cache.py | 42 ++++++++++++++++++++++++++++----------
 tests/unit/test_req.py     | 19 +++++++++++++++++
 3 files changed, 51 insertions(+), 11 deletions(-)
 create mode 100644 news/11985.bugfix.rst

diff --git a/news/11985.bugfix.rst b/news/11985.bugfix.rst
new file mode 100644
index 000000000..66c8e8786
--- /dev/null
+++ b/news/11985.bugfix.rst
@@ -0,0 +1 @@
+Ignore invalid or unreadable ``origin.json`` files in the cache of locally built wheels.
diff --git a/src/pip/_internal/cache.py b/src/pip/_internal/cache.py
index 05f0a9acb..8d3a664c7 100644
--- a/src/pip/_internal/cache.py
+++ b/src/pip/_internal/cache.py
@@ -194,7 +194,17 @@ class CacheEntry:
         self.origin: Optional[DirectUrl] = None
         origin_direct_url_path = Path(self.link.file_path).parent / ORIGIN_JSON_NAME
         if origin_direct_url_path.exists():
-            self.origin = DirectUrl.from_json(origin_direct_url_path.read_text())
+            try:
+                self.origin = DirectUrl.from_json(
+                    origin_direct_url_path.read_text(encoding="utf-8")
+                )
+            except Exception as e:
+                logger.warning(
+                    "Ignoring invalid cache entry origin file %s for %s (%s)",
+                    origin_direct_url_path,
+                    link.filename,
+                    e,
+                )
 
 
 class WheelCache(Cache):
@@ -257,16 +267,26 @@ class WheelCache(Cache):
     @staticmethod
     def record_download_origin(cache_dir: str, download_info: DirectUrl) -> None:
         origin_path = Path(cache_dir) / ORIGIN_JSON_NAME
-        if origin_path.is_file():
-            origin = DirectUrl.from_json(origin_path.read_text())
-            # TODO: use DirectUrl.equivalent when https://github.com/pypa/pip/pull/10564
-            # is merged.
-            if origin.url != download_info.url:
+        if origin_path.exists():
+            try:
+                origin = DirectUrl.from_json(origin_path.read_text(encoding="utf-8"))
+            except Exception as e:
                 logger.warning(
-                    "Origin URL %s in cache entry %s does not match download URL %s. "
-                    "This is likely a pip bug or a cache corruption issue.",
-                    origin.url,
-                    cache_dir,
-                    download_info.url,
+                    "Could not read origin file %s in cache entry (%s). "
+                    "Will attempt to overwrite it.",
+                    origin_path,
+                    e,
                 )
+            else:
+                # TODO: use DirectUrl.equivalent when
+                # https://github.com/pypa/pip/pull/10564 is merged.
+                if origin.url != download_info.url:
+                    logger.warning(
+                        "Origin URL %s in cache entry %s does not match download URL "
+                        "%s. This is likely a pip bug or a cache corruption issue. "
+                        "Will overwrite it with the new value.",
+                        origin.url,
+                        cache_dir,
+                        download_info.url,
+                    )
         origin_path.write_text(download_info.to_json(), encoding="utf-8")
diff --git a/tests/unit/test_req.py b/tests/unit/test_req.py
index c9742812b..74b9712dc 100644
--- a/tests/unit/test_req.py
+++ b/tests/unit/test_req.py
@@ -445,6 +445,25 @@ class TestRequirementSet:
             assert isinstance(req.download_info.info, ArchiveInfo)
             assert req.download_info.info.hash == hash
 
+    def test_download_info_archive_cache_with_invalid_origin(
+        self, tmp_path: Path, shared_data: TestData, caplog: pytest.LogCaptureFixture
+    ) -> None:
+        """Test an invalid origin.json is ignored."""
+        url = shared_data.packages.joinpath("simple-1.0.tar.gz").as_uri()
+        finder = make_test_finder()
+        wheel_cache = WheelCache(str(tmp_path / "cache"))
+        cache_entry_dir = wheel_cache.get_path_for_link(Link(url))
+        Path(cache_entry_dir).mkdir(parents=True)
+        Path(cache_entry_dir).joinpath("origin.json").write_text("{")  # invalid json
+        wheel.make_wheel(name="simple", version="1.0").save_to_dir(cache_entry_dir)
+        with self._basic_resolver(finder, wheel_cache=wheel_cache) as resolver:
+            ireq = get_processed_req_from_line(f"simple @ {url}")
+            reqset = resolver.resolve([ireq], True)
+            assert len(reqset.all_requirements) == 1
+            req = reqset.all_requirements[0]
+            assert req.is_wheel_from_cache
+            assert "Ignoring invalid cache entry origin file" in caplog.messages[0]
+
     def test_download_info_local_wheel(self, data: TestData) -> None:
         """Test that download_info is set for requirements from a local wheel."""
         finder = make_test_finder()