Add builder for unixfs dags (#12)

* Add builder for unixfs chunked files and directories

data/builder/dir_test.go

@@ -0,0 +1,72 @@
+package builder
+
+import (
+	"bytes"
+	"fmt"
+	"testing"
+
+	"github.com/ipfs/go-unixfsnode"
+	dagpb "github.com/ipld/go-codec-dagpb"
+	"github.com/ipld/go-ipld-prime"
+	cidlink "github.com/ipld/go-ipld-prime/linking/cid"
+)
+
+func mkEntries(cnt int, ls *ipld.LinkSystem) ([]dagpb.PBLink, error) {
+	entries := make([]dagpb.PBLink, 0, cnt)
+	for i := 0; i < cnt; i++ {
+		r := bytes.NewBufferString(fmt.Sprintf("%d", i))
+		f, s, err := BuildUnixFSFile(r, "", ls)
+		if err != nil {
+			return nil, err
+		}
+		e, err := BuildUnixFSDirectoryEntry(fmt.Sprintf("file %d", i), int64(s), f)
+		if err != nil {
+			return nil, err
+		}
+		entries = append(entries, e)
+	}
+	return entries, nil
+}
+
+func TestBuildUnixFSDirectory(t *testing.T) {
+	ls := cidlink.DefaultLinkSystem()
+	storage := cidlink.Memory{}
+	ls.StorageReadOpener = storage.OpenRead
+	ls.StorageWriteOpener = storage.OpenWrite
+
+	testSizes := []int{100, 1000, 50000}
+	for _, cnt := range testSizes {
+		entries, err := mkEntries(cnt, &ls)
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		dl, err := BuildUnixFSDirectory(entries, &ls)
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		pbn, err := ls.Load(ipld.LinkContext{}, dl, dagpb.Type.PBNode)
+		if err != nil {
+			t.Fatal(err)
+		}
+		ufd, err := unixfsnode.Reify(ipld.LinkContext{}, pbn, &ls)
+		if err != nil {
+			t.Fatal(err)
+		}
+		observedCnt := 0
+
+		li := ufd.MapIterator()
+		for !li.Done() {
+			_, _, err := li.Next()
+			if err != nil {
+				t.Fatal(err)
+			}
+			observedCnt++
+		}
+		if observedCnt != cnt {
+			fmt.Printf("%+v\n", ufd)
+			t.Fatalf("unexpected number of dir entries %d vs %d", observedCnt, cnt)
+		}
+	}
+}

data/builder/directory.go

@@ -0,0 +1,130 @@
+package builder
+
+import (
+	"fmt"
+	"io/fs"
+	"os"
+	"path"
+
+	"github.com/ipfs/go-unixfsnode/data"
+	dagpb "github.com/ipld/go-codec-dagpb"
+	"github.com/ipld/go-ipld-prime"
+	cidlink "github.com/ipld/go-ipld-prime/linking/cid"
+	"github.com/multiformats/go-multihash"
+)
+
+// https://github.com/ipfs/go-ipfs/pull/8114/files#diff-eec963b47a6e1080d9d8023b4e438e6e3591b4154f7379a7e728401d2055374aR319
+const shardSplitThreshold = 262144
+
+// https://github.com/ipfs/go-unixfs/blob/ec6bb5a4c5efdc3a5bce99151b294f663ee9c08d/io/directory.go#L29
+const defaultShardWidth = 256
+
+// BuildUnixFSRecursive returns a link pointing to the UnixFS node representing
+// the file or directory tree pointed to by `root`
+func BuildUnixFSRecursive(root string, ls *ipld.LinkSystem) (ipld.Link, uint64, error) {
+	info, err := os.Lstat(root)
+	if err != nil {
+		return nil, 0, err
+	}
+
+	m := info.Mode()
+	switch {
+	case m.IsDir():
+		entries, err := os.ReadDir(root)
+		if err != nil {
+			return nil, 0, err
+		}
+		lnks := make([]dagpb.PBLink, 0, len(entries))
+		for _, e := range entries {
+			lnk, sz, err := BuildUnixFSRecursive(path.Join(root, e.Name()), ls)
+			if err != nil {
+				return nil, 0, err
+			}
+			entry, err := BuildUnixFSDirectoryEntry(e.Name(), int64(sz), lnk)
+			if err != nil {
+				return nil, 0, err
+			}
+			lnks = append(lnks, entry)
+		}
+		outLnk, err := BuildUnixFSDirectory(lnks, ls)
+		return outLnk, 0, err
+	case m.Type() == fs.ModeSymlink:
+		content, err := os.Readlink(root)
+		if err != nil {
+			return nil, 0, err
+		}
+		return BuildUnixFSSymlink(content, ls)
+	case m.IsRegular():
+		fp, err := os.Open(root)
+		if err != nil {
+			return nil, 0, err
+		}
+		defer fp.Close()
+		return BuildUnixFSFile(fp, "", ls)
+	default:
+		return nil, 0, fmt.Errorf("cannot encode non regular file: %s", root)
+	}
+}
+
+// estimateDirSize estimates if a directory is big enough that it warrents sharding.
+// The estimate is the sum over the len(linkName) + bytelen(linkHash)
+// https://github.com/ipfs/go-unixfs/blob/master/io/directory.go#L152-L162
+func estimateDirSize(entries []dagpb.PBLink) int {
+	s := 0
+	for _, e := range entries {
+		s += len(e.Name.Must().String())
+		lnk := e.Hash.Link()
+		cl, ok := lnk.(cidlink.Link)
+		if ok {
+			s += cl.ByteLen()
+		} else {
+			s += len(lnk.Binary())
+		}
+	}
+	return s
+}
+
+// BuildUnixFSDirectory creates a directory link over a collection of entries.
+func BuildUnixFSDirectory(entries []dagpb.PBLink, ls *ipld.LinkSystem) (ipld.Link, error) {
+	if estimateDirSize(entries) > shardSplitThreshold {
+		return BuildUnixFSShardedDirectory(defaultShardWidth, multihash.MURMUR3_128, entries, ls)
+	}
+	ufd, err := BuildUnixFS(func(b *Builder) {
+		DataType(b, data.Data_Directory)
+	})
+	if err != nil {
+		return nil, err
+	}
+	pbb := dagpb.Type.PBNode.NewBuilder()
+	pbm, err := pbb.BeginMap(2)
+	if err != nil {
+		return nil, err
+	}
+	if err = pbm.AssembleKey().AssignString("Data"); err != nil {
+		return nil, err
+	}
+	if err = pbm.AssembleValue().AssignBytes(data.EncodeUnixFSData(ufd)); err != nil {
+		return nil, err
+	}
+	if err = pbm.AssembleKey().AssignString("Links"); err != nil {
+		return nil, err
+	}
+	lnks, err := pbm.AssembleValue().BeginList(int64(len(entries)))
+	if err != nil {
+		return nil, err
+	}
+	// sorting happens in codec-dagpb
+	for _, e := range entries {
+		if err := lnks.AssembleValue().AssignNode(e); err != nil {
+			return nil, err
+		}
+	}
+	if err := lnks.Finish(); err != nil {
+		return nil, err
+	}
+	if err := pbm.Finish(); err != nil {
+		return nil, err
+	}
+	node := pbb.Build()
+	return ls.Store(ipld.LinkContext{}, fileLinkProto, node)
+}