Merge pull request #10111 from obsidiansystems/git-objects

Support symlinks properly with `git-hashing` experimental feature

src/libutil/git.cc

@@ -56,17 +56,18 @@ void parseBlob(
     FileSystemObjectSink & sink,
     const Path & sinkPath,
     Source & source,
-    bool executable,
+    BlobMode blobMode,
     const ExperimentalFeatureSettings & xpSettings)
 {
     xpSettings.require(Xp::GitHashing);
 
+    unsigned long long size = std::stoi(getStringUntil(source, 0));
+
+    auto doRegularFile = [&](bool executable) {
         sink.createRegularFile(sinkPath, [&](auto & crf) {
             if (executable)
                 crf.isExecutable();
 
-        unsigned long long size = std::stoi(getStringUntil(source, 0));
-
             crf.preallocateContents(size);
 
             unsigned long long left = size;
@@ -81,6 +82,37 @@ void parseBlob(
                 left -= buf.size();
             }
         });
+    };
+
+    switch (blobMode) {
+
+    case BlobMode::Regular:
+        doRegularFile(false);
+        break;
+
+    case BlobMode::Executable:
+        doRegularFile(true);
+        break;
+
+    case BlobMode::Symlink:
+    {
+        std::string target;
+        target.resize(size, '0');
+        target.reserve(size);
+        for (size_t n = 0; n < target.size();) {
+            checkInterrupt();
+            n += source.read(
+                const_cast<char *>(target.c_str()) + n,
+                target.size() - n);
+        }
+
+        sink.createSymlink(sinkPath, target);
+        break;
+    }
+
+    default:
+        assert(false);
+    }
 }
 
 void parseTree(
@@ -142,7 +174,7 @@ void parse(
     FileSystemObjectSink & sink,
     const Path & sinkPath,
     Source & source,
-    bool executable,
+    BlobMode rootModeIfBlob,
     std::function<SinkHook> hook,
     const ExperimentalFeatureSettings & xpSettings)
 {
@@ -152,7 +184,7 @@ void parse(
 
     switch (type) {
     case ObjectType::Blob:
-        parseBlob(sink, sinkPath, source, executable, xpSettings);
+        parseBlob(sink, sinkPath, source, rootModeIfBlob, xpSettings);
         break;
     case ObjectType::Tree:
         parseTree(sink, sinkPath, source, hook, xpSettings);
@@ -177,7 +209,7 @@ std::optional<Mode> convertMode(SourceAccessor::Type type)
 
 void restore(FileSystemObjectSink & sink, Source & source, std::function<RestoreHook> hook)
 {
-    parse(sink, "", source, false, [&](Path name, TreeEntry entry) {
+    parse(sink, "", source, BlobMode::Regular, [&](Path name, TreeEntry entry) {
         auto [accessor, from] = hook(entry.hash);
         auto stat = accessor->lstat(from);
         auto gotOpt = convertMode(stat.type);
@@ -275,6 +307,13 @@ Mode dump(
     }
 
     case SourceAccessor::tSymlink:
+    {
+        auto target = accessor.readLink(path);
+        dumpBlobPrefix(target.size(), sink, xpSettings);
+        sink(target);
+        return Mode::Symlink;
+    }
+
     case SourceAccessor::tMisc:
     default:
         throw Error("file '%1%' has an unsupported type", path);

src/libutil/git.hh

@@ -75,10 +75,23 @@ ObjectType parseObjectType(
     Source & source,
     const ExperimentalFeatureSettings & xpSettings = experimentalFeatureSettings);
 
+/**
+ * These 3 modes are represented by blob objects.
+ *
+ * Sometimes we need this information to disambiguate how a blob is
+ * being used to better match our own "file system object" data model.
+ */
+enum struct BlobMode : RawMode
+{
+    Regular = static_cast<RawMode>(Mode::Regular),
+    Executable = static_cast<RawMode>(Mode::Executable),
+    Symlink = static_cast<RawMode>(Mode::Symlink),
+};
+
 void parseBlob(
     FileSystemObjectSink & sink, const Path & sinkPath,
     Source & source,
-    bool executable,
+    BlobMode blobMode,
     const ExperimentalFeatureSettings & xpSettings = experimentalFeatureSettings);
 
 void parseTree(
@@ -89,11 +102,15 @@ void parseTree(
 
 /**
  * Helper putting the previous three `parse*` functions together.
+ *
+ * @rootModeIfBlob How to interpret a root blob, for which there is no
+ * disambiguating dir entry to answer that questino. If the root it not
+ * a blob, this is ignored.
  */
 void parse(
     FileSystemObjectSink & sink, const Path & sinkPath,
     Source & source,
-    bool executable,
+    BlobMode rootModeIfBlob,
     std::function<SinkHook> hook,
     const ExperimentalFeatureSettings & xpSettings = experimentalFeatureSettings);
 

tests/functional/git-hashing/simple.sh

@@ -56,3 +56,12 @@ echo Run Hello World! > $TEST_ROOT/dummy3/dir/executable
 path3=$(nix store add --mode git --hash-algo sha1 $TEST_ROOT/dummy3)
 hash3=$(nix-store -q --hash $path3)
 test "$hash3" = "sha256:08y3nm3mvn9qvskqnf13lfgax5lh73krxz4fcjd5cp202ggpw9nv"
+
+rm -rf $TEST_ROOT/dummy3
+mkdir -p $TEST_ROOT/dummy3
+mkdir -p $TEST_ROOT/dummy3/dir
+touch $TEST_ROOT/dummy3/dir/file
+ln -s './hello/world.txt' $TEST_ROOT/dummy3/dir/symlink
+path3=$(nix store add --mode git --hash-algo sha1 $TEST_ROOT/dummy3)
+hash3=$(nix-store -q --hash $path3)
+test "$hash3" = "sha256:1dwazas8irzpar89s8k2bnp72imfw7kgg4aflhhsfnicg8h428f3"

tests/unit/libutil/data/git/tree.txt

@@ -1,3 +1,4 @@
 100644 blob 63ddb340119baf8492d2da53af47e8c7cfcd5eb2	Foo
 100755 blob 63ddb340119baf8492d2da53af47e8c7cfcd5eb2	bAr
 040000 tree 4b825dc642cb6eb9a060e54bf8d69288fbee4904	baZ
+120000 blob 63ddb340119baf8492d2da53af47e8c7cfcd5eb2	quuX

tests/unit/libutil/git.cc

@@ -67,7 +67,7 @@ TEST_F(GitTest, blob_read) {
         StringSink out;
         RegularFileSink out2 { out };
         ASSERT_EQ(parseObjectType(in, mockXpSettings), ObjectType::Blob);
-        parseBlob(out2, "", in, false, mockXpSettings);
+        parseBlob(out2, "", in, BlobMode::Regular, mockXpSettings);
 
         auto expected = readFile(goldenMaster("hello-world.bin"));
 
@@ -115,6 +115,15 @@ const static Tree tree = {
             .hash = Hash::parseAny("4b825dc642cb6eb9a060e54bf8d69288fbee4904", HashAlgorithm::SHA1),
         },
     },
+    {
+        "quuX",
+        {
+            .mode = Mode::Symlink,
+            // hello world with special chars from above (symlink target
+            // can be anything)
+            .hash = Hash::parseAny("63ddb340119baf8492d2da53af47e8c7cfcd5eb2", HashAlgorithm::SHA1),
+        },
+    },
 };
 
 TEST_F(GitTest, tree_read) {
@@ -165,6 +174,12 @@ TEST_F(GitTest, both_roundrip) {
                                 .contents = "good day,\n\0\n\tworld!",
                             },
                         },
+                        {
+                            "quux",
+                            File::Symlink {
+                                .target = "/over/there",
+                            },
+                        },
                     },
                 },
             },
@@ -195,21 +210,24 @@ TEST_F(GitTest, both_roundrip) {
 
     MemorySink sinkFiles2 { files2 };
 
-    std::function<void(const Path, const Hash &, bool)> mkSinkHook;
-    mkSinkHook = [&](auto prefix, auto & hash, auto executable) {
+    std::function<void(const Path, const Hash &, BlobMode)> mkSinkHook;
+    mkSinkHook = [&](auto prefix, auto & hash, auto blobMode) {
         StringSource in { cas[hash] };
         parse(
-            sinkFiles2, prefix, in, executable,
+            sinkFiles2, prefix, in, blobMode,
             [&](const Path & name, const auto & entry) {
                 mkSinkHook(
                     prefix + "/" + name,
                     entry.hash,
-                    entry.mode == Mode::Executable);
+                    // N.B. this cast would not be acceptable in real
+                    // code, because it would make an assert reachable,
+                    // but it should harmless in this test.
+                    static_cast<BlobMode>(entry.mode));
             },
             mockXpSettings);
     };
 
-    mkSinkHook("", root.hash, false);
+    mkSinkHook("", root.hash, BlobMode::Regular);
 
     ASSERT_EQ(files, files2);
 }