refactor(tvix/nar-bridge): rename writer package to exporter r/6696

Change-Id: I7cf6a3ed0dca4c450fadd9228dc0154ad6254f51
Reviewed-on: https://cl.tvl.fyi/c/depot/+/9530
Tested-by: BuildkiteCI
Reviewed-by: Connor Brewster <cbrewster@hey.com>
Autosubmit: flokli <flokli@flokli.de>

3 files changed, 550 insertions, 0 deletions

diff --git a/tvix/nar-bridge/pkg/exporter/export.go b/tvix/nar-bridge/pkg/exporter/export.go
new file mode 100644
index 000000000000..40ccc4b6380c
--- /dev/null
+++ b/tvix/nar-bridge/pkg/exporter/export.go
@@ -0,0 +1,279 @@
+package exporter
+
+import (
+	"fmt"
+	"io"
+	"path"
+
+	castorev1pb "code.tvl.fyi/tvix/castore/protos"
+	storev1pb "code.tvl.fyi/tvix/store/protos"
+	"github.com/nix-community/go-nix/pkg/nar"
+)
+
+type DirectoryLookupFn func([]byte) (*castorev1pb.Directory, error)
+type BlobLookupFn func([]byte) (io.ReadCloser, error)
+
+// Export will traverse a given PathInfo structure, and write the contents
+// in NAR format to the passed Writer.
+// It uses directoryLookupFn and blobLookupFn to resolve references.
+func Export(
+	w io.Writer,
+	pathInfo *storev1pb.PathInfo,
+	directoryLookupFn DirectoryLookupFn,
+	blobLookupFn BlobLookupFn,
+) error {
+	// initialize a NAR writer
+	narWriter, err := nar.NewWriter(w)
+	if err != nil {
+		return fmt.Errorf("unable to initialize nar writer: %w", err)
+	}
+	defer narWriter.Close()
+
+	// populate rootHeader
+	rootHeader := &nar.Header{
+		Path: "/",
+	}
+
+	// populate a stack
+	// we will push paths and directories to it when entering a directory,
+	// and emit individual elements to the NAR writer, draining the Directory object.
+	// once it's empty, we can pop it off the stack.
+	var stackPaths = []string{}
+	var stackDirectories = []*castorev1pb.Directory{}
+
+	// peek at the pathInfo root and assemble the root node and write to writer
+	// in the case of a regular file, we retrieve and write the contents, close and exit
+	// in the case of a symlink, we write the symlink, close and exit
+	switch v := (pathInfo.GetNode().GetNode()).(type) {
+	case *castorev1pb.Node_File:
+		rootHeader.Type = nar.TypeRegular
+		rootHeader.Size = int64(v.File.GetSize())
+		rootHeader.Executable = v.File.GetExecutable()
+		err := narWriter.WriteHeader(rootHeader)
+		if err != nil {
+			return fmt.Errorf("unable to write root header: %w", err)
+		}
+
+		// if it's a regular file, retrieve and write the contents
+		blobReader, err := blobLookupFn(v.File.GetDigest())
+		if err != nil {
+			return fmt.Errorf("unable to lookup blob: %w", err)
+		}
+		defer blobReader.Close()
+
+		_, err = io.Copy(narWriter, blobReader)
+		if err != nil {
+			return fmt.Errorf("unable to read from blobReader: %w", err)
+		}
+
+		err = blobReader.Close()
+		if err != nil {
+			return fmt.Errorf("unable to close content reader: %w", err)
+		}
+
+		err = narWriter.Close()
+		if err != nil {
+			return fmt.Errorf("unable to close nar reader: %w", err)
+		}
+
+		return nil
+
+	case *castorev1pb.Node_Symlink:
+		rootHeader.Type = nar.TypeSymlink
+		rootHeader.LinkTarget = string(v.Symlink.GetTarget())
+		err := narWriter.WriteHeader(rootHeader)
+		if err != nil {
+			return fmt.Errorf("unable to write root header: %w", err)
+		}
+
+		err = narWriter.Close()
+		if err != nil {
+			return fmt.Errorf("unable to close nar reader: %w", err)
+		}
+
+		return nil
+	case *castorev1pb.Node_Directory:
+		// We have a directory at the root, look it up and put in on the stack.
+		directory, err := directoryLookupFn(v.Directory.Digest)
+		if err != nil {
+			return fmt.Errorf("unable to lookup directory: %w", err)
+		}
+		stackDirectories = append(stackDirectories, directory)
+		stackPaths = append(stackPaths, "/")
+
+		err = narWriter.WriteHeader(&nar.Header{
+			Path: "/",
+			Type: nar.TypeDirectory,
+		})
+
+		if err != nil {
+			return fmt.Errorf("error writing header: %w", err)
+		}
+	}
+
+	// as long as the stack is not empty, we keep running.
+	for {
+		if len(stackDirectories) == 0 {
+			return nil
+		}
+
+		// Peek at the current top of the stack.
+		topOfStack := stackDirectories[len(stackDirectories)-1]
+		topOfStackPath := stackPaths[len(stackPaths)-1]
+
+		// get the next element that's lexicographically smallest, and drain it from
+		// the current directory on top of the stack.
+		nextNode := drainNextNode(topOfStack)
+
+		// If nextNode returns nil, there's nothing left in the directory node, so we
+		// can emit it from the stack.
+		// Contrary to the import case, we don't emit the node popping from the stack, but when pushing.
+		if nextNode == nil {
+			// pop off stack
+			stackDirectories = stackDirectories[:len(stackDirectories)-1]
+			stackPaths = stackPaths[:len(stackPaths)-1]
+
+			continue
+		}
+
+		switch n := (nextNode).(type) {
+		case *castorev1pb.DirectoryNode:
+			err := narWriter.WriteHeader(&nar.Header{
+				Path: path.Join(topOfStackPath, string(n.GetName())),
+				Type: nar.TypeDirectory,
+			})
+			if err != nil {
+				return fmt.Errorf("unable to write nar header: %w", err)
+			}
+
+			d, err := directoryLookupFn(n.GetDigest())
+			if err != nil {
+				return fmt.Errorf("unable to lookup directory: %w", err)
+			}
+
+			// add to stack
+			stackDirectories = append(stackDirectories, d)
+			stackPaths = append(stackPaths, path.Join(topOfStackPath, string(n.GetName())))
+		case *castorev1pb.FileNode:
+			err := narWriter.WriteHeader(&nar.Header{
+				Path:       path.Join(topOfStackPath, string(n.GetName())),
+				Type:       nar.TypeRegular,
+				Size:       int64(n.GetSize()),
+				Executable: n.GetExecutable(),
+			})
+			if err != nil {
+				return fmt.Errorf("unable to write nar header: %w", err)
+			}
+
+			// copy file contents
+			contentReader, err := blobLookupFn(n.GetDigest())
+			if err != nil {
+				return fmt.Errorf("unable to get blob: %w", err)
+			}
+			defer contentReader.Close()
+
+			_, err = io.Copy(narWriter, contentReader)
+			if err != nil {
+				return fmt.Errorf("unable to copy contents from contentReader: %w", err)
+			}
+
+			err = contentReader.Close()
+			if err != nil {
+				return fmt.Errorf("unable to close content reader: %w", err)
+			}
+		case *castorev1pb.SymlinkNode:
+			err := narWriter.WriteHeader(&nar.Header{
+				Path:       path.Join(topOfStackPath, string(n.GetName())),
+				Type:       nar.TypeSymlink,
+				LinkTarget: string(n.GetTarget()),
+			})
+			if err != nil {
+				return fmt.Errorf("unable to write nar header: %w", err)
+			}
+		}
+	}
+}
+
+// TODO: add validation functions to Directory in both rust and golang, to
+// validate the keys in directories, files and symlinks are sorted.
+
+// drainNextNode will drain a directory message with one of its child nodes,
+// whichever comes first alphabetically.
+func drainNextNode(d *castorev1pb.Directory) interface{} {
+	switch v := (smallestNode(d)).(type) {
+	case *castorev1pb.DirectoryNode:
+		d.Directories = d.Directories[1:]
+		return v
+	case *castorev1pb.FileNode:
+		d.Files = d.Files[1:]
+		return v
+	case *castorev1pb.SymlinkNode:
+		d.Symlinks = d.Symlinks[1:]
+		return v
+	case nil:
+		return nil
+	default:
+		panic("invalid type encountered")
+	}
+}
+
+// smallestNode will return the node from a directory message,
+// whichever comes first alphabetically.
+func smallestNode(d *castorev1pb.Directory) interface{} {
+	childDirectories := d.GetDirectories()
+	childFiles := d.GetFiles()
+	childSymlinks := d.GetSymlinks()
+
+	if len(childDirectories) > 0 {
+		if len(childFiles) > 0 {
+			if len(childSymlinks) > 0 {
+				// directories,files,symlinks
+				return smallerNode(smallerNode(childDirectories[0], childFiles[0]), childSymlinks[0])
+			} else {
+				// directories,files,!symlinks
+				return smallerNode(childDirectories[0], childFiles[0])
+			}
+		} else {
+			// directories,!files
+			if len(childSymlinks) > 0 {
+				// directories,!files,symlinks
+				return smallerNode(childDirectories[0], childSymlinks[0])
+			} else {
+				// directories,!files,!symlinks
+				return childDirectories[0]
+			}
+		}
+	} else {
+		// !directories
+		if len(childFiles) > 0 {
+			// !directories,files
+			if len(childSymlinks) > 0 {
+				// !directories,files,symlinks
+				return smallerNode(childFiles[0], childSymlinks[0])
+			} else {
+				// !directories,files,!symlinks
+				return childFiles[0]
+			}
+		} else {
+			//!directories,!files
+			if len(childSymlinks) > 0 {
+				//!directories,!files,symlinks
+				return childSymlinks[0]
+			} else {
+				//!directories,!files,!symlinks
+				return nil
+			}
+		}
+	}
+}
+
+// smallerNode compares two nodes by their name,
+// and returns the one with the smaller name.
+// both nodes may not be nil, we do check for these cases in smallestNode.
+func smallerNode(a interface{ GetName() []byte }, b interface{ GetName() []byte }) interface{ GetName() []byte } {
+	if string(a.GetName()) < string(b.GetName()) {
+		return a
+	} else {
+		return b
+	}
+}
diff --git a/tvix/nar-bridge/pkg/exporter/export_test.go b/tvix/nar-bridge/pkg/exporter/export_test.go
new file mode 100644
index 000000000000..ffee62f29183
--- /dev/null
+++ b/tvix/nar-bridge/pkg/exporter/export_test.go
@@ -0,0 +1,220 @@
+package exporter_test
+
+import (
+	"bytes"
+	"context"
+	"encoding/base64"
+	"fmt"
+	"io"
+	"os"
+	"testing"
+
+	castorev1pb "code.tvl.fyi/tvix/castore/protos"
+	"code.tvl.fyi/tvix/nar-bridge/pkg/exporter"
+	"code.tvl.fyi/tvix/nar-bridge/pkg/importer"
+	storev1pb "code.tvl.fyi/tvix/store/protos"
+	"github.com/stretchr/testify/require"
+	"lukechampine.com/blake3"
+)
+
+func mustDirectoryDigest(d *castorev1pb.Directory) []byte {
+	dgst, err := d.Digest()
+	if err != nil {
+		panic(err)
+	}
+	return dgst
+}
+
+func mustBlobDigest(r io.Reader) []byte {
+	hasher := blake3.New(32, nil)
+	_, err := io.Copy(hasher, r)
+	if err != nil {
+		panic(err)
+	}
+	return hasher.Sum([]byte{})
+}
+
+func TestSymlink(t *testing.T) {
+	pathInfo := &storev1pb.PathInfo{
+
+		Node: &castorev1pb.Node{
+			Node: &castorev1pb.Node_Symlink{
+				Symlink: &castorev1pb.SymlinkNode{
+					Name:   []byte("doesntmatter"),
+					Target: []byte("/nix/store/somewhereelse"),
+				},
+			},
+		},
+	}
+
+	var buf bytes.Buffer
+
+	err := exporter.Export(&buf, pathInfo, func([]byte) (*castorev1pb.Directory, error) {
+		panic("no directories expected")
+	}, func([]byte) (io.ReadCloser, error) {
+		panic("no files expected")
+	})
+	require.NoError(t, err, "exporter shouldn't fail")
+
+	f, err := os.Open("../../testdata/symlink.nar")
+	require.NoError(t, err)
+
+	bytesExpected, err := io.ReadAll(f)
+	if err != nil {
+		panic(err)
+	}
+
+	require.Equal(t, bytesExpected, buf.Bytes(), "expected nar contents to match")
+}
+
+func TestRegular(t *testing.T) {
+	// The blake3 digest of the 0x01 byte.
+	BLAKE3_DIGEST_0X01 := []byte{
+		0x48, 0xfc, 0x72, 0x1f, 0xbb, 0xc1, 0x72, 0xe0, 0x92, 0x5f, 0xa2, 0x7a, 0xf1, 0x67, 0x1d,
+		0xe2, 0x25, 0xba, 0x92, 0x71, 0x34, 0x80, 0x29, 0x98, 0xb1, 0x0a, 0x15, 0x68, 0xa1, 0x88,
+		0x65, 0x2b,
+	}
+
+	pathInfo := &storev1pb.PathInfo{
+		Node: &castorev1pb.Node{
+			Node: &castorev1pb.Node_File{
+				File: &castorev1pb.FileNode{
+					Name:       []byte("doesntmatter"),
+					Digest:     BLAKE3_DIGEST_0X01,
+					Size:       1,
+					Executable: false,
+				},
+			},
+		},
+	}
+
+	var buf bytes.Buffer
+
+	err := exporter.Export(&buf, pathInfo, func([]byte) (*castorev1pb.Directory, error) {
+		panic("no directories expected")
+	}, func(blobRef []byte) (io.ReadCloser, error) {
+		if !bytes.Equal(blobRef, BLAKE3_DIGEST_0X01) {
+			panic("unexpected blobref")
+		}
+		return io.NopCloser(bytes.NewBuffer([]byte{0x01})), nil
+	})
+	require.NoError(t, err, "exporter shouldn't fail")
+
+	f, err := os.Open("../../testdata/onebyteregular.nar")
+	require.NoError(t, err)
+
+	bytesExpected, err := io.ReadAll(f)
+	if err != nil {
+		panic(err)
+	}
+
+	require.Equal(t, bytesExpected, buf.Bytes(), "expected nar contents to match")
+}
+
+func TestEmptyDirectory(t *testing.T) {
+	// construct empty directory node this refers to
+	emptyDirectory := &castorev1pb.Directory{
+		Directories: []*castorev1pb.DirectoryNode{},
+		Files:       []*castorev1pb.FileNode{},
+		Symlinks:    []*castorev1pb.SymlinkNode{},
+	}
+	emptyDirectoryDigest := mustDirectoryDigest(emptyDirectory)
+
+	pathInfo := &storev1pb.PathInfo{
+		Node: &castorev1pb.Node{
+			Node: &castorev1pb.Node_Directory{
+				Directory: &castorev1pb.DirectoryNode{
+					Name:   []byte("doesntmatter"),
+					Digest: emptyDirectoryDigest,
+					Size:   0,
+				},
+			},
+		},
+	}
+
+	var buf bytes.Buffer
+
+	err := exporter.Export(&buf, pathInfo, func(directoryRef []byte) (*castorev1pb.Directory, error) {
+		if !bytes.Equal(directoryRef, emptyDirectoryDigest) {
+			panic("unexpected directoryRef")
+		}
+		return emptyDirectory, nil
+	}, func([]byte) (io.ReadCloser, error) {
+		panic("no files expected")
+	})
+	require.NoError(t, err, "exporter shouldn't fail")
+
+	f, err := os.Open("../../testdata/emptydirectory.nar")
+	require.NoError(t, err)
+
+	bytesExpected, err := io.ReadAll(f)
+	if err != nil {
+		panic(err)
+	}
+
+	require.Equal(t, bytesExpected, buf.Bytes(), "expected nar contents to match")
+}
+
+func TestFull(t *testing.T) {
+	// We pipe nar_1094wph9z4nwlgvsd53abfz8i117ykiv5dwnq9nnhz846s7xqd7d.nar to the exporter,
+	// and store all the file contents and directory objects received in two hashmaps.
+	// We then feed it to the writer, and test we come up with the same NAR file.
+
+	f, err := os.Open("../../testdata/nar_1094wph9z4nwlgvsd53abfz8i117ykiv5dwnq9nnhz846s7xqd7d.nar")
+	require.NoError(t, err)
+
+	narContents, err := io.ReadAll(f)
+	require.NoError(t, err)
+
+	blobsMap := make(map[string][]byte, 0)
+	directoriesMap := make(map[string]*castorev1pb.Directory)
+
+	pathInfo, err := importer.Import(
+		context.Background(),
+		bytes.NewBuffer(narContents),
+		func(blobReader io.Reader) ([]byte, error) {
+			// read in contents, we need to put it into filesMap later.
+			contents, err := io.ReadAll(blobReader)
+			require.NoError(t, err)
+
+			dgst := mustBlobDigest(bytes.NewReader(contents))
+
+			// put it in filesMap
+			blobsMap[base64.StdEncoding.EncodeToString(dgst)] = contents
+
+			return dgst, nil
+		},
+		func(directory *castorev1pb.Directory) ([]byte, error) {
+			dgst := mustDirectoryDigest(directory)
+
+			directoriesMap[base64.StdEncoding.EncodeToString(dgst)] = directory
+			return dgst, nil
+		},
+	)
+
+	require.NoError(t, err)
+
+	// done populating everything, now actually test the export :-)
+	var buf bytes.Buffer
+	err = exporter.Export(
+		&buf,
+		pathInfo,
+		func(directoryDgst []byte) (*castorev1pb.Directory, error) {
+			d, found := directoriesMap[base64.StdEncoding.EncodeToString(directoryDgst)]
+			if !found {
+				panic(fmt.Sprintf("directory %v not found", base64.StdEncoding.EncodeToString(directoryDgst)))
+			}
+			return d, nil
+		},
+		func(blobDgst []byte) (io.ReadCloser, error) {
+			blobContents, found := blobsMap[base64.StdEncoding.EncodeToString(blobDgst)]
+			if !found {
+				panic(fmt.Sprintf("blob      %v not found", base64.StdEncoding.EncodeToString(blobDgst)))
+			}
+			return io.NopCloser(bytes.NewReader(blobContents)), nil
+		},
+	)
+
+	require.NoError(t, err, "exporter shouldn't fail")
+	require.Equal(t, narContents, buf.Bytes())
+}
diff --git a/tvix/nar-bridge/pkg/exporter/pick_next_node_test.go b/tvix/nar-bridge/pkg/exporter/pick_next_node_test.go
new file mode 100644
index 000000000000..c80261ae6b6a
--- /dev/null
+++ b/tvix/nar-bridge/pkg/exporter/pick_next_node_test.go
@@ -0,0 +1,51 @@
+package exporter
+
+import (
+	"testing"
+
+	castorev1pb "code.tvl.fyi/tvix/castore/protos"
+	"github.com/google/go-cmp/cmp"
+	"github.com/stretchr/testify/require"
+	"google.golang.org/protobuf/testing/protocmp"
+)
+
+func requireProtoEq(t *testing.T, expected interface{}, actual interface{}) {
+	if diff := cmp.Diff(expected, actual, protocmp.Transform()); diff != "" {
+		t.Errorf("unexpected difference:\n%v", diff)
+	}
+}
+
+func TestPopNextNode(t *testing.T) {
+	t.Run("empty directory", func(t *testing.T) {
+		d := &castorev1pb.Directory{
+			Directories: []*castorev1pb.DirectoryNode{},
+			Files:       []*castorev1pb.FileNode{},
+			Symlinks:    []*castorev1pb.SymlinkNode{},
+		}
+
+		n := drainNextNode(d)
+		require.Equal(t, nil, n)
+	})
+	t.Run("only directories", func(t *testing.T) {
+		ds := &castorev1pb.Directory{
+			Directories: []*castorev1pb.DirectoryNode{{
+				Name:   []byte("a"),
+				Digest: []byte{},
+				Size:   0,
+			}, {
+				Name:   []byte("b"),
+				Digest: []byte{},
+				Size:   0,
+			}},
+			Files:    []*castorev1pb.FileNode{},
+			Symlinks: []*castorev1pb.SymlinkNode{},
+		}
+
+		n := drainNextNode(ds)
+		requireProtoEq(t, &castorev1pb.DirectoryNode{
+			Name:   []byte("a"),
+			Digest: []byte{},
+			Size:   0,
+		}, n)
+	})
+}