From 7b1cda9ca20d2e03f00d34e8c2fb0f8e6569b115 Mon Sep 17 00:00:00 2001
From: Eelco Dolstra <edolstra@gmail.com>
Date: Fri, 3 Feb 2023 16:04:19 +0100
Subject: [PATCH] GitArchiveInputScheme: Restore the lastModified attribute

---
 src/libfetchers/git-accessor.cc | 11 ++++++++--
 src/libfetchers/git-utils.hh    |  8 ++++++-
 src/libfetchers/github.cc       | 38 +++++++++++++++++----------------
 3 files changed, 36 insertions(+), 21 deletions(-)

diff --git a/src/libfetchers/git-accessor.cc b/src/libfetchers/git-accessor.cc
index 45d387fa9..c50e4355a 100644
--- a/src/libfetchers/git-accessor.cc
+++ b/src/libfetchers/git-accessor.cc
@@ -243,7 +243,7 @@ static Repository openTarballCache()
     }
 }
 
-Hash importTarball(Source & source)
+TarballInfo importTarball(Source & source)
 {
     auto repo = openTarballCache();
 
@@ -308,6 +308,8 @@ Hash importTarball(Source & source)
 
     size_t componentsToStrip = 1;
 
+    time_t lastModified = 0;
+
     for (;;) {
         // FIXME: merge with extract_archive
         struct archive_entry * entry;
@@ -321,6 +323,8 @@ Hash importTarball(Source & source)
         else
             archive.check(r);
 
+        lastModified = std::max(lastModified, archive_entry_mtime(entry));
+
         auto pathComponents = tokenizeString<std::vector<std::string>>(path, "/");
 
         std::span<const std::string> pathComponents2{pathComponents};
@@ -388,7 +392,10 @@ Hash importTarball(Source & source)
 
     auto [oid, _name] = popBuilder();
 
-    return Hash::parseAny(git_oid_tostr_s(&oid), htSHA1);
+    return TarballInfo {
+        .treeHash = Hash::parseAny(git_oid_tostr_s(&oid), htSHA1),
+        .lastModified = lastModified
+    };
 }
 
 ref<InputAccessor> makeTarballCacheAccessor(const Hash & rev)
diff --git a/src/libfetchers/git-utils.hh b/src/libfetchers/git-utils.hh
index 775a0d021..227e31942 100644
--- a/src/libfetchers/git-utils.hh
+++ b/src/libfetchers/git-utils.hh
@@ -6,7 +6,13 @@ namespace nix {
 
 ref<InputAccessor> makeGitInputAccessor(const CanonPath & path, const Hash & rev);
 
-Hash importTarball(Source & source);
+struct TarballInfo
+{
+    Hash treeHash;
+    time_t lastModified;
+};
+
+TarballInfo importTarball(Source & source);
 
 ref<InputAccessor> makeTarballCacheAccessor(const Hash & rev);
 
diff --git a/src/libfetchers/github.cc b/src/libfetchers/github.cc
index 756dcef8b..5e75a0992 100644
--- a/src/libfetchers/github.cc
+++ b/src/libfetchers/github.cc
@@ -168,6 +168,8 @@ struct GitArchiveInputScheme : InputScheme
 
     void checkLocks(const Input & specified, const Input & final) const override
     {
+        InputScheme::checkLocks(specified, final);
+
         if (auto prevTreeHash = getTreeHash(specified)) {
             if (getTreeHash(final) != prevTreeHash)
                 throw Error("Git tree hash mismatch in input '%s', expected '%s'",
@@ -201,7 +203,7 @@ struct GitArchiveInputScheme : InputScheme
 
     virtual DownloadUrl getDownloadUrl(const Input & input) const = 0;
 
-    std::pair<Input, Hash> downloadArchive(ref<Store> store, Input input) const
+    std::pair<Input, TarballInfo> downloadArchive(ref<Store> store, Input input) const
     {
         if (!maybeGetStrAttr(input.attrs, "ref")) input.attrs.insert_or_assign("ref", "HEAD");
 
@@ -214,13 +216,17 @@ struct GitArchiveInputScheme : InputScheme
         auto cache = getCache();
 
         auto treeHashKey = fmt("git-rev-to-tree-hash-%s", rev->gitRev());
+        auto lastModifiedKey = fmt("git-rev-to-last-modified-%s", rev->gitRev());
 
         if (auto treeHashS = cache->queryFact(treeHashKey)) {
-            auto treeHash = Hash::parseAny(*treeHashS, htSHA1);
-            if (tarballCacheContains(treeHash))
-                return {std::move(input), treeHash};
-            else
-                debug("Git tree with hash '%s' has disappeared from the cache, refetching...", treeHash.gitRev());
+            if (auto lastModifiedS = cache->queryFact(lastModifiedKey)) {
+                auto treeHash = Hash::parseAny(*treeHashS, htSHA1);
+                auto lastModified = string2Int<time_t>(*lastModifiedS).value();
+                if (tarballCacheContains(treeHash))
+                    return {std::move(input), TarballInfo { .treeHash = treeHash, .lastModified = lastModified }};
+                else
+                    debug("Git tree with hash '%s' has disappeared from the cache, refetching...", treeHash.gitRev());
+            }
         }
 
         /* Stream the tarball into the tarball cache. */
@@ -232,26 +238,22 @@ struct GitArchiveInputScheme : InputScheme
             getFileTransfer()->download(std::move(req), sink);
         });
 
-        auto treeHash = importTarball(*source);
+        auto tarballInfo = importTarball(*source);
 
-        cache->upsertFact(treeHashKey, treeHash.gitRev());
+        cache->upsertFact(treeHashKey, tarballInfo.treeHash.gitRev());
+        cache->upsertFact(lastModifiedKey, std::to_string(tarballInfo.lastModified));
 
-        return {std::move(input), treeHash};
+        return {std::move(input), tarballInfo};
     }
 
     std::pair<ref<InputAccessor>, Input> getAccessor(ref<Store> store, const Input & _input) const override
     {
-        auto [input, treeHash] = downloadArchive(store, _input);
+        auto [input, tarballInfo] = downloadArchive(store, _input);
 
-        input.attrs.insert_or_assign("treeHash", treeHash.gitRev());
+        input.attrs.insert_or_assign("treeHash", tarballInfo.treeHash.gitRev());
+        input.attrs.insert_or_assign("lastModified", uint64_t(tarballInfo.lastModified));
 
-        auto accessor = makeTarballCacheAccessor(treeHash);
-
-        #if 0
-        auto lastModified = accessor->getLastModified();
-        assert(lastModified);
-        input.attrs.insert_or_assign("lastModified", uint64_t(*lastModified));
-        #endif
+        auto accessor = makeTarballCacheAccessor(tarballInfo.treeHash);
 
         accessor->setPathDisplay("«" + input.to_string() + "»");