Check tarball cache validity

src/libfetchers/git-accessor.cc

@@ -1,3 +1,4 @@
+#include "git-utils.hh"
 #include "input-accessor.hh"
 
 #include <span>
@@ -43,6 +44,14 @@ static Repository openRepo(const CanonPath & path)
     return Repository(_repo);
 }
 
+git_oid hashToOID(const Hash & hash)
+{
+    git_oid oid;
+    if (git_oid_fromstr(&oid, hash.gitRev().c_str()))
+        throw Error("cannot convert '%s' to a Git OID", hash.gitRev());
+    return oid;
+}
+
 struct GitInputAccessor : InputAccessor
 {
     Repository repo;
@@ -51,9 +60,7 @@ struct GitInputAccessor : InputAccessor
     GitInputAccessor(Repository && repo_, const Hash & rev)
         : repo(std::move(repo_))
     {
-        git_oid oid;
-        if (git_oid_fromstr(&oid, rev.gitRev().c_str()))
-            throw Error("cannot convert '%s' to a Git OID", rev.gitRev());
+        auto oid = hashToOID(rev);
 
         git_object * obj = nullptr;
         if (git_object_lookup(&obj, repo.get(), &oid, GIT_OBJECT_ANY)) {
@@ -389,4 +396,20 @@ ref<InputAccessor> makeTarballCacheAccessor(const Hash & rev)
     return make_ref<GitInputAccessor>(openTarballCache(), rev);
 }
 
+bool tarballCacheContains(const Hash & treeHash)
+{
+    auto repo = openTarballCache();
+
+    auto oid = hashToOID(treeHash);
+
+    git_object * obj = nullptr;
+    if (auto errCode = git_object_lookup(&obj, repo.get(), &oid, GIT_OBJECT_TREE)) {
+        if (errCode == GIT_ENOTFOUND) return false;
+        auto err = git_error_last();
+        throw Error("getting Git object '%s': %s", treeHash.gitRev(), err->message);
+    }
+
+    return true;
+}
+
 }

src/libfetchers/git-utils.hh

@@ -0,0 +1,15 @@
+#pragma once
+
+#include "input-accessor.hh"
+
+namespace nix {
+
+ref<InputAccessor> makeGitInputAccessor(const CanonPath & path, const Hash & rev);
+
+Hash importTarball(Source & source);
+
+ref<InputAccessor> makeTarballCacheAccessor(const Hash & rev);
+
+bool tarballCacheContains(const Hash & treeHash);
+
+}

src/libfetchers/git.cc

@@ -8,6 +8,7 @@
 #include "util.hh"
 #include "git.hh"
 #include "fs-input-accessor.hh"
+#include "git-utils.hh"
 
 #include "fetch-settings.hh"
 

src/libfetchers/github.cc

@@ -9,6 +9,7 @@
 #include "fetch-settings.hh"
 #include "input-accessor.hh"
 #include "tarball.hh"
+#include "git-utils.hh"
 
 #include <optional>
 #include <nlohmann/json.hpp>
@@ -196,8 +197,10 @@ struct GitArchiveInputScheme : InputScheme
 
         if (auto treeHashS = cache->queryFact(treeHashKey)) {
             auto treeHash = Hash::parseAny(*treeHashS, htSHA1);
-            // FIXME: verify that treeHash exists in the tarball cache.
-            return {std::move(input), treeHash};
+            if (tarballCacheContains(treeHash))
+                return {std::move(input), treeHash};
+            else
+                debug("Git tree with hash '%s' has disappeared from the cache, refetching...", treeHash.gitRev());
         }
 
         /* Stream the tarball into the tarball cache. */
@@ -272,6 +275,7 @@ struct GitHubInputScheme : GitArchiveInputScheme
         return getStrAttr(input.attrs, "repo");
     }
 
+    /* .commit.tree.sha, .commit.committer.date */
     Hash getRevFromRef(nix::ref<Store> store, const Input & input) const override
     {
         auto host = getHost(input);

src/libfetchers/input-accessor.hh

@@ -117,12 +117,6 @@ ref<InputAccessor> makePatchingInputAccessor(
     ref<InputAccessor> next,
     const std::vector<std::string> & patches);
 
-ref<InputAccessor> makeGitInputAccessor(const CanonPath & path, const Hash & rev);
-
-Hash importTarball(Source & source);
-
-ref<InputAccessor> makeTarballCacheAccessor(const Hash & rev);
-
 struct SourcePath
 {
     ref<InputAccessor> accessor;