From fbcb897e21e6c7b866a7aed97129141c0e7caa22 Mon Sep 17 00:00:00 2001
From: Eelco Dolstra <edolstra@gmail.com>
Date: Mon, 16 Mar 2020 13:20:32 +0100
Subject: [PATCH] Add a test for shallow Git clones

Also, don't return a revCount anymore for shallow or dirty Git trees,
since it's incorrect.

Closes #2988.
---
 src/libexpr/primops/fetchGit.cc |  5 +++--
 src/libstore/fetchers/git.cc    | 14 +++++++++++---
 tests/fetchGit.sh               |  8 +++++++-
 3 files changed, 21 insertions(+), 6 deletions(-)

diff --git a/src/libexpr/primops/fetchGit.cc b/src/libexpr/primops/fetchGit.cc
index 18774371b..638c14970 100644
--- a/src/libexpr/primops/fetchGit.cc
+++ b/src/libexpr/primops/fetchGit.cc
@@ -67,8 +67,9 @@ static void prim_fetchGit(EvalState & state, const Pos & pos, Value * * args, Va
     auto rev2 = input2->getRev().value_or(Hash(htSHA1));
     mkString(*state.allocAttr(v, state.symbols.create("rev")), rev2.gitRev());
     mkString(*state.allocAttr(v, state.symbols.create("shortRev")), rev2.gitShortRev());
-    assert(tree.info.revCount);
-    mkInt(*state.allocAttr(v, state.symbols.create("revCount")), *tree.info.revCount);
+    // Backward compatibility: set 'revCount' to 0 for a dirty tree.
+    mkInt(*state.allocAttr(v, state.symbols.create("revCount")),
+        tree.info.revCount.value_or(0));
     v.attrs->sort();
 
     if (state.allowedPaths)
diff --git a/src/libstore/fetchers/git.cc b/src/libstore/fetchers/git.cc
index 9276b0993..46ca187fe 100644
--- a/src/libstore/fetchers/git.cc
+++ b/src/libstore/fetchers/git.cc
@@ -34,6 +34,8 @@ static void cacheGitInfo(
     const Tree & tree,
     const Hash & rev)
 {
+    if (!tree.info.revCount || !tree.info.lastModified) return;
+
     nlohmann::json json;
     json["storePath"] = store.printStorePath(tree.storePath);
     json["name"] = name;
@@ -169,6 +171,7 @@ struct GitInput : Input
             return url.path;
         return {};
     }
+
     void markChangedFile(std::string_view file, std::optional<std::string> commitMsg) const override
     {
         auto sourcePath = getSourcePath();
@@ -271,7 +274,6 @@ struct GitInput : Input
                     .actualPath = store->printStorePath(storePath),
                     .storePath = std::move(storePath),
                     .info = TreeInfo {
-                        .revCount = haveCommits ? std::stoull(runProgram("git", true, { "-C", actualUrl, "rev-list", "--count", "HEAD" })) : 0,
                         // FIXME: maybe we should use the timestamp of the last
                         // modified dirty file?
                         .lastModified = haveCommits ? std::stoull(runProgram("git", true, { "-C", actualUrl, "log", "-1", "--format=%ct", "HEAD" })) : 0,
@@ -357,8 +359,11 @@ struct GitInput : Input
                 input->rev = Hash(chomp(readFile(localRefFile)), htSHA1);
         }
 
+        bool isShallow = chomp(runProgram("git", true, { "-C", repoDir, "rev-parse", "--is-shallow-repository" })) == "true";
+
         if (auto tree = lookupGitInfo(store, name, *input->rev)) {
             assert(*input->rev == tree->first);
+            if (isShallow) tree->second.info.revCount.reset();
             return {std::move(tree->second), input};
         }
 
@@ -380,14 +385,17 @@ struct GitInput : Input
         unpackTarfile(*source, tmpDir);
 
         auto storePath = store->addToStore(name, tmpDir);
-        auto revCount = std::stoull(runProgram("git", true, { "-C", repoDir, "rev-list", "--count", input->rev->gitRev() }));
+
         auto lastModified = std::stoull(runProgram("git", true, { "-C", repoDir, "log", "-1", "--format=%ct", input->rev->gitRev() }));
 
         auto tree = Tree {
             .actualPath = store->toRealPath(storePath),
             .storePath = std::move(storePath),
             .info = TreeInfo {
-                .revCount = revCount,
+                .revCount =
+                    !isShallow
+                    ? std::optional(std::stoull(runProgram("git", true, { "-C", repoDir, "rev-list", "--count", input->rev->gitRev() })))
+                    : std::nullopt,
                 .lastModified = lastModified
             }
         };
diff --git a/tests/fetchGit.sh b/tests/fetchGit.sh
index b64f444a5..6e5a4750e 100644
--- a/tests/fetchGit.sh
+++ b/tests/fetchGit.sh
@@ -11,7 +11,7 @@ repo=$TEST_ROOT/git
 
 export _NIX_FORCE_HTTP=1
 
-rm -rf $repo ${repo}-tmp $TEST_HOME/.cache/nix/gitv* $TEST_ROOT/worktree
+rm -rf $repo ${repo}-tmp $TEST_HOME/.cache/nix/git* $TEST_ROOT/worktree $TEST_ROOT/shallow
 
 git init $repo
 git -C $repo config user.email "foobar@example.com"
@@ -147,3 +147,9 @@ NIX=$(command -v nix)
 # Try again, with 'git' available.  This should work.
 path5=$(nix eval --impure --raw --expr "(builtins.fetchGit { url = $repo; ref = \"dev\"; }).outPath")
 [[ $path3 = $path5 ]]
+
+# Check that shallow clones work and don't return a revcount.
+git clone --depth 1 file://$repo $TEST_ROOT/shallow
+path6=$(nix eval --impure --raw --expr "(builtins.fetchGit { url = $TEST_ROOT/shallow; ref = \"dev\"; }).outPath")
+[[ $path3 = $path6 ]]
+[[ $(nix eval --impure --expr "(builtins.fetchTree { type = \"git\"; url = \"file://$TEST_ROOT/shallow\"; ref = \"dev\"; }).revCount or 123") == 123 ]]