From e9ea9e2316052a3d10df7ba7e59bdd630237f9c8 Mon Sep 17 00:00:00 2001
From: Harmen Stoppels <harmenstoppels@gmail.com>
Date: Thu, 22 Dec 2022 09:48:05 +0100
Subject: index.json.hash, no fatal error if key cannot be fetched (#34643)

---
 lib/spack/spack/binary_distribution.py | 36 ++++++++++++++++++----------------
 lib/spack/spack/test/bindist.py        |  3 +--
 2 files changed, 20 insertions(+), 19 deletions(-)

(limited to 'lib')

diff --git a/lib/spack/spack/binary_distribution.py b/lib/spack/spack/binary_distribution.py
index 6ab71e3965..8ae7207768 100644
--- a/lib/spack/spack/binary_distribution.py
+++ b/lib/spack/spack/binary_distribution.py
@@ -2389,33 +2389,35 @@ class DefaultIndexFetcher:
         self.url = url
         self.local_hash = local_hash
         self.urlopen = urlopen
+        self.headers = {"User-Agent": web_util.SPACK_USER_AGENT}
+
+    def get_remote_hash(self):
+        # Failure to fetch index.json.hash is not fatal
+        url_index_hash = url_util.join(self.url, _build_cache_relative_path, "index.json.hash")
+        try:
+            response = self.urlopen(urllib.request.Request(url_index_hash, headers=self.headers))
+        except urllib.error.URLError:
+            return None
+
+        # Validate the hash
+        remote_hash = response.read(64)
+        if not re.match(rb"[a-f\d]{64}$", remote_hash):
+            return None
+        return remote_hash.decode("utf-8")
 
     def conditional_fetch(self):
         # Do an intermediate fetch for the hash
         # and a conditional fetch for the contents
-        if self.local_hash:
-            url_index_hash = url_util.join(self.url, _build_cache_relative_path, "index.json.hash")
 
-            try:
-                response = self.urlopen(urllib.request.Request(url_index_hash))
-            except urllib.error.URLError as e:
-                raise FetchIndexError("Could not fetch {}".format(url_index_hash), e) from e
-
-            # Validate the hash
-            remote_hash = response.read(64)
-            if not re.match(rb"[a-f\d]{64}$", remote_hash):
-                raise FetchIndexError("Invalid hash format in {}".format(url_index_hash))
-            remote_hash = remote_hash.decode("utf-8")
-
-            # No need to update further
-            if remote_hash == self.local_hash:
-                return FetchIndexResult(etag=None, hash=None, data=None, fresh=True)
+        # Early exit if our cache is up to date.
+        if self.local_hash and self.local_hash == self.get_remote_hash():
+            return FetchIndexResult(etag=None, hash=None, data=None, fresh=True)
 
         # Otherwise, download index.json
         url_index = url_util.join(self.url, _build_cache_relative_path, "index.json")
 
         try:
-            response = self.urlopen(urllib.request.Request(url_index))
+            response = self.urlopen(urllib.request.Request(url_index, headers=self.headers))
         except urllib.error.URLError as e:
             raise FetchIndexError("Could not fetch index from {}".format(url_index), e)
 
diff --git a/lib/spack/spack/test/bindist.py b/lib/spack/spack/test/bindist.py
index 653d783969..dc6cb35177 100644
--- a/lib/spack/spack/test/bindist.py
+++ b/lib/spack/spack/test/bindist.py
@@ -854,8 +854,7 @@ def test_default_index_invalid_hash_file(index_json):
         url="https://www.example.com", local_hash=index_json_hash, urlopen=urlopen
     )
 
-    with pytest.raises(bindist.FetchIndexError, match="Invalid hash format"):
-        fetcher.conditional_fetch()
+    assert fetcher.get_remote_hash() is None
 
 
 def test_default_index_json_404():
-- 
cgit v1.2.3-60-g2f50