do not cache metadata lookups for git or file links

Author: cosmicexplorer

src/pip/_internal/operations/prepare.py

@@ -14,7 +14,6 @@
 from typing import Any, Dict, Iterable, List, Optional
 
 from pip._vendor.packaging.utils import canonicalize_name
-from pip._vendor.requests.exceptions import InvalidSchema
 
 from pip._internal.cache import LinkMetadataCache
 from pip._internal.distributions import make_distribution_for_install_requirement
@@ -428,11 +427,22 @@ def _fetch_metadata_only(
             self._cache_metadata(req.link, computed_metadata)
         return computed_metadata
 
-    def _fetch_cached_metadata(
-        self,
-        link: Link,
-    ) -> Optional[BaseDistribution]:
+    def _should_cache_metadata(self, link: Link) -> bool:
         if self._metadata_cache is None:
+            return False
+
+        # Some types of links may host changing content we don't want to cache.
+        if link.scheme.startswith("git"):
+            logger.debug("not attempting to cache metadata from git url %s", link)
+            return False
+        if link.scheme == "file":
+            logger.debug("not attempting to cache metadata from local file %s", link)
+            return False
+
+        return True
+
+    def _fetch_cached_metadata(self, link: Link) -> Optional[BaseDistribution]:
+        if not self._should_cache_metadata(link):
             return None
         try:
             cached_path = self._metadata_cache.cache_path(link)
@@ -459,7 +469,7 @@ def _cache_metadata(
         link: Link,
         metadata_dist: BaseDistribution,
     ) -> None:
-        if self._metadata_cache is None:
+        if not self._should_cache_metadata(link):
             return
         try:
             cached_path = self._metadata_cache.cache_path(link)
@@ -564,17 +574,7 @@ def _complete_partial_requirements(
         links_to_fully_download: Dict[Link, InstallRequirement] = {}
         for req in partially_downloaded_reqs:
             assert req.link
-            # If this is e.g. a git url, we don't know how to handle that in the
-            # BatchDownloader, so leave it for self._prepare_linked_requirement() at the
-            # end of this method, which knows how to handle any URL.
-            can_simply_download = True
-            try:
-                # This will raise InvalidSchema if our Session can't download it.
-                self._session.get_adapter(req.link.url)
-            except InvalidSchema:
-                can_simply_download = False
-            if can_simply_download:
-                links_to_fully_download[req.link] = req
+            links_to_fully_download[req.link] = req
 
         batch_download = self._batch_download(
             links_to_fully_download.keys(),