Support hard links in tarballs

Fixes #10395.

src/libfetchers/git-utils.cc

@@ -115,10 +115,10 @@ git_oid hashToOID(const Hash & hash)
     return oid;
 }
 
-Object lookupObject(git_repository * repo, const git_oid & oid)
+Object lookupObject(git_repository * repo, const git_oid & oid, git_object_t type = GIT_OBJECT_ANY)
 {
     Object obj;
-    if (git_object_lookup(Setter(obj), repo, &oid, GIT_OBJECT_ANY)) {
+    if (git_object_lookup(Setter(obj), repo, &oid, type)) {
         auto err = git_error_last();
         throw Error("getting Git object '%s': %s", oid, err->message);
     }
@@ -909,6 +909,50 @@ struct GitFileSystemObjectSinkImpl : GitFileSystemObjectSink
         addToTree(*pathComponents.rbegin(), oid, GIT_FILEMODE_LINK);
     }
 
+    void createHardlink(const Path & path, const CanonPath & target) override
+    {
+        auto pathComponents = tokenizeString<std::vector<std::string>>(path, "/");
+        if (!prepareDirs(pathComponents, false)) return;
+
+        auto relTarget = CanonPath(path).parent()->makeRelative(target);
+
+        auto dir = pendingDirs.rbegin();
+
+        // For each ../ component at the start, go up one directory.
+        std::string_view relTargetLeft(relTarget);
+        while (hasPrefix(relTargetLeft, "../")) {
+            if (dir == pendingDirs.rend())
+                throw Error("invalid hard link target '%s'", target);
+            ++dir;
+            relTargetLeft = relTargetLeft.substr(3);
+        }
+
+        // Look up the remainder of the target, starting at the
+        // top-most `git_treebuilder`.
+        std::variant<git_treebuilder *, git_oid> curDir{dir->builder.get()};
+        Object tree; // needed to keep `entry` alive
+        const git_tree_entry * entry = nullptr;
+
+        for (auto & c : CanonPath(relTargetLeft)) {
+            if (auto builder = std::get_if<git_treebuilder *>(&curDir)) {
+                if (!(entry = git_treebuilder_get(*builder, std::string(c).c_str())))
+                    throw Error("cannot find hard link target '%s'", target);
+                curDir = *git_tree_entry_id(entry);
+            } else if (auto oid = std::get_if<git_oid>(&curDir)) {
+                tree = lookupObject(*repo, *oid, GIT_OBJECT_TREE);
+                if (!(entry = git_tree_entry_byname((const git_tree *) &*tree, std::string(c).c_str())))
+                    throw Error("cannot find hard link target '%s'", target);
+                curDir = *git_tree_entry_id(entry);
+            }
+        }
+
+        assert(entry);
+
+        addToTree(*pathComponents.rbegin(),
+            *git_tree_entry_id(entry),
+            git_tree_entry_filemode(entry));
+    }
+
     Hash sync() override {
         updateBuilders({});
 

src/libfetchers/git-utils.hh

@@ -7,7 +7,7 @@ namespace nix {
 
 namespace fetchers { struct PublicKey; }
 
-struct GitFileSystemObjectSink : FileSystemObjectSink
+struct GitFileSystemObjectSink : ExtendedFileSystemObjectSink
 {
     /**
      * Flush builder and return a final Git hash.

src/libutil/fs-sink.hh

@@ -41,6 +41,19 @@ struct FileSystemObjectSink
     virtual void createSymlink(const Path & path, const std::string & target) = 0;
 };
 
+/**
+ * An extension of `FileSystemObjectSink` that supports file types
+ * that are not supported by Nix's FSO model.
+ */
+struct ExtendedFileSystemObjectSink : FileSystemObjectSink
+{
+    /**
+     * Create a hard link. The target must be the path of a previously
+     * encountered file relative to the root of the FSO.
+     */
+    virtual void createHardlink(const Path & path, const CanonPath & target) = 0;
+};
+
 /**
  * Recursively copy file system objects from the source into the sink.
  */

src/libutil/tarfile.cc

@@ -163,7 +163,7 @@ void unpackTarfile(const Path & tarFile, const Path & destDir)
     extract_archive(archive, destDir);
 }
 
-time_t unpackTarfileToSink(TarArchive & archive, FileSystemObjectSink & parseSink)
+time_t unpackTarfileToSink(TarArchive & archive, ExtendedFileSystemObjectSink & parseSink)
 {
     time_t lastModified = 0;
 
@@ -183,7 +183,12 @@ time_t unpackTarfileToSink(TarArchive & archive, FileSystemObjectSink & parseSin
 
         lastModified = std::max(lastModified, archive_entry_mtime(entry));
 
-        switch (archive_entry_filetype(entry)) {
+        if (auto target = archive_entry_hardlink(entry)) {
+            parseSink.createHardlink(path, CanonPath(target));
+            continue;
+        }
+
+        switch (auto type = archive_entry_filetype(entry)) {
 
         case AE_IFDIR:
             parseSink.createDirectory(path);
@@ -220,7 +225,7 @@ time_t unpackTarfileToSink(TarArchive & archive, FileSystemObjectSink & parseSin
         }
 
         default:
-            throw Error("file '%s' in tarball has unsupported file type", path);
+            throw Error("file '%s' in tarball has unsupported file type %d", path, type);
         }
     }
 

src/libutil/tarfile.hh

@@ -41,6 +41,6 @@ void unpackTarfile(Source & source, const Path & destDir);
 
 void unpackTarfile(const Path & tarFile, const Path & destDir);
 
-time_t unpackTarfileToSink(TarArchive & archive, FileSystemObjectSink & parseSink);
+time_t unpackTarfileToSink(TarArchive & archive, ExtendedFileSystemObjectSink & parseSink);
 
 }

tests/functional/tarball.sh

@@ -59,3 +59,12 @@ test_tarball() {
 test_tarball '' cat
 test_tarball .xz xz
 test_tarball .gz gzip
+
+# Test hard links.
+path="$(nix flake prefetch --json "tarball+file://$(pwd)/tree.tar.gz" | jq -r .storePath)"
+[[ $(cat "$path/a/b/foo") = bar ]]
+[[ $(cat "$path/a/b/xyzzy") = bar ]]
+[[ $(cat "$path/a/yyy") = bar ]]
+[[ $(cat "$path/a/zzz") = bar ]]
+[[ $(cat "$path/c/aap") = bar ]]
+[[ $(cat "$path/fnord") = bar ]]