Add list and filter commands

v2/cmd/car/car.go

@@ -28,10 +28,28 @@ func main() {
 				},
 			},
 			{
-				Name:    "split",
-				Aliases: []string{"s"},
-				Usage:   "Split an index to a detached file",
-				Action:  SplitCar,
+				Name:   "detach-index",
+				Usage:  "Detach an index to a detached file",
+				Action: DetachCar,
+			},
+			{
+				Name:    "list",
+				Aliases: []string{"l"},
+				Usage:   "List the CIDs in a car",
+				Action:  ListCar,
+			},
+			{
+				Name:    "filter",
+				Aliases: []string{"f"},
+				Usage:   "Filter the CIDs in a car",
+				Action:  FilterCar,
+				Flags: []cli.Flag{
+					&cli.StringFlag{
+						Name:      "cid-file",
+						Usage:     "A file to read CIDs from",
+						TakesFile: true,
+					},
+				},
 			},
 		},
 	}

v2/cmd/car/split.go → v2/cmd/car/detach.go

@@ -9,8 +9,8 @@ import (
 	"github.com/urfave/cli/v2"
 )
 
-// SplitCar is a command to output the index part of a car.
-func SplitCar(c *cli.Context) error {
+// DetachCar is a command to output the index part of a car.
+func DetachCar(c *cli.Context) error {
 	r, err := carv2.OpenReader(c.Args().Get(0))
 	if err != nil {
 		return err

v2/cmd/car/filter.go

@@ -0,0 +1,100 @@
+package main
+
+import (
+	"bufio"
+	"fmt"
+	"io"
+	"os"
+	"strings"
+
+	"github.com/ipfs/go-cid"
+	carv2 "github.com/ipld/go-car/v2"
+	"github.com/ipld/go-car/v2/blockstore"
+	icarv1 "github.com/ipld/go-car/v2/internal/carv1"
+	"github.com/urfave/cli/v2"
+)
+
+// FilterCar is a command to select a subset of a car by CID.
+func FilterCar(c *cli.Context) error {
+	r, err := carv2.OpenReader(c.Args().Get(0))
+	if err != nil {
+		return err
+	}
+	defer r.Close()
+
+	if c.Args().Len() < 2 {
+		return fmt.Errorf("an output filename must be provided")
+	}
+	roots, err := r.Roots()
+	if err != nil {
+		return err
+	}
+	bs, err := blockstore.OpenReadWrite(c.Args().Get(1), roots)
+	if err != nil {
+		return err
+	}
+
+	// Get the set of CIDs from stdin.
+	inStream := os.Stdin
+	if c.IsSet("cidFile") {
+		inStream, err = os.Open(c.String("cidFile"))
+		if err != nil {
+			return err
+		}
+		defer inStream.Close()
+	}
+	cidList, err := parseCIDS(inStream)
+	if err != nil {
+		return err
+	}
+	fmt.Printf("filtering to %d cids\n", len(cidList))
+
+	cidMap := make(map[cid.Cid]struct{})
+	for _, e := range cidList {
+		cidMap[e] = struct{}{}
+	}
+
+	rd, err := icarv1.NewCarReader(r.DataReader())
+	if err != nil {
+		return err
+	}
+
+	for {
+		blk, err := rd.Next()
+		if err != nil {
+			if err == io.EOF {
+				break
+			}
+			return err
+		}
+		if _, ok := cidMap[blk.Cid()]; ok {
+			if err := bs.Put(blk); err != nil {
+				return err
+			}
+		}
+	}
+	return bs.Finalize()
+}
+
+func parseCIDS(r io.Reader) ([]cid.Cid, error) {
+	cb := make([]cid.Cid, 0)
+	br := bufio.NewReader(r)
+	for {
+		line, _, err := br.ReadLine()
+		if err != nil {
+			if err == io.EOF {
+				return cb, nil
+			}
+			return nil, err
+		}
+		trimLine := strings.TrimSpace(string(line))
+		if len(trimLine) == 0 {
+			continue
+		}
+		c, err := cid.Parse(trimLine)
+		if err != nil {
+			return nil, err
+		}
+		cb = append(cb, c)
+	}
+}

v2/cmd/car/list.go

@@ -0,0 +1,47 @@
+package main
+
+import (
+	"fmt"
+	"io"
+	"os"
+
+	carv2 "github.com/ipld/go-car/v2"
+	icarv1 "github.com/ipld/go-car/v2/internal/carv1"
+	"github.com/urfave/cli/v2"
+)
+
+// ListCar is a command to output the cids in a car.
+func ListCar(c *cli.Context) error {
+	r, err := carv2.OpenReader(c.Args().Get(0))
+	if err != nil {
+		return err
+	}
+	defer r.Close()
+
+	outStream := os.Stdout
+	if c.Args().Len() >= 2 {
+		outStream, err = os.Create(c.Args().Get(1))
+		if err != nil {
+			return err
+		}
+	}
+	defer outStream.Close()
+
+	rd, err := icarv1.NewCarReader(r.DataReader())
+	if err != nil {
+		return err
+	}
+
+	for {
+		blk, err := rd.Next()
+		if err != nil {
+			if err == io.EOF {
+				break
+			}
+			return err
+		}
+		outStream.WriteString(fmt.Sprintf("%s\n", blk.Cid()))
+	}
+
+	return err
+}