Add NodeReaderAt type (read a node from a subpath)

reader.go

@@ -1,5 +1,12 @@
 package ipld
 
+import(
+	"fmt"
+	"errors"
+	"strings"
+	"reflect"
+)
+
 type nodeReadErrors struct { error }
 var NodeReadAbort error = &nodeReadErrors{errors.New("abort")}
 var NodeReadSkip error = &nodeReadErrors{errors.New("skip")}
@@ -73,3 +80,53 @@ type NodeReader interface {
 	Read(f ReadFun) error;
 }
 
+// Convert a []interface{} path to a []string path
+// This should only convert array indices to strings, and such are not ambiguous
+// because array indices and object keys cannot be mixed for the same prefix.
+func ToStringPath(anyPath []interface{}) []string {
+	var res []string
+	for _, e := range anyPath {
+		if str, ok := e.(string); ok {
+			res = append(res, str)
+		} else if i, ok := e.(int); ok {
+			res = append(res, fmt.Sprintf("%d", i))
+		} else {
+			res = append(res, fmt.Sprintf("%v", e))
+		}
+	}
+	return res
+}
+
+// A NodeReader that only read elements from a path prefix
+type NodeReaderAt struct {
+	parent NodeReader
+	prefix []string
+}
+
+func (n *NodeReaderAt) Read(f ReadFun) error {
+	found := false
+	return n.parent.Read(func(path []interface{}, tokenType ReaderToken, value interface{}) error {
+		if isPathPrefix(n.prefix, ToStringPath(path)) {
+			found = true
+			return f(path[len(n.prefix):], tokenType, value)
+		} else if found {
+			// Finished, stop early
+			return NodeReadAbort
+		} else {
+			// Not yet here, continue
+			return nil
+		}
+	})
+}
+
+// Return true if path has prefix
+func isPathPrefix(prefix, path []string) bool {
+	return len(prefix) <= len(path) && reflect.DeepEqual(prefix, path[0:len(prefix)])
+}
+
+// Return an instance of NodeReaderAt that read root only starting at the given
+// prefix.
+func GetReaderAt(root NodeReader, path string) *NodeReaderAt {
+	return &NodeReaderAt{root, strings.Split(path, "/")}
+}
+