X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/f4f0614d06c4ac9abeb34dc6df7c88ecb5feebda..97d0e12a78e9245c8ed29c070ffe0399e5cd6cb4:/sdk/go/arvados/collection.go

diff --git a/sdk/go/arvados/collection.go b/sdk/go/arvados/collection.go
index 623a8d985f..cec20279d1 100644
--- a/sdk/go/arvados/collection.go
+++ b/sdk/go/arvados/collection.go
@@ -6,7 +6,9 @@ package arvados
 
 import (
 	"bufio"
+	"crypto/md5"
 	"fmt"
+	"regexp"
 	"strings"
 	"time"
 
@@ -24,8 +26,8 @@ type Collection struct {
 	Name                      string                 `json:"name"`
 	CreatedAt                 time.Time              `json:"created_at"`
 	ModifiedAt                time.Time              `json:"modified_at"`
-	ModifiedByClientUUID      *string                `json:"modified_by_client_uuid"`
-	ModifiedByUserUUID        *string                `json:"modified_by_user_uuid"`
+	ModifiedByClientUUID      string                 `json:"modified_by_client_uuid"`
+	ModifiedByUserUUID        string                 `json:"modified_by_user_uuid"`
 	PortableDataHash          string                 `json:"portable_data_hash"`
 	ReplicationConfirmed      *int                   `json:"replication_confirmed"`
 	ReplicationConfirmedAt    *time.Time             `json:"replication_confirmed_at"`
@@ -39,10 +41,10 @@ type Collection struct {
 	WritableBy                []string               `json:"writable_by,omitempty"`
 	FileCount                 int                    `json:"file_count"`
 	FileSizeTotal             int64                  `json:"file_size_total"`
-	Version                   *int                   `json:"version"`
+	Version                   int                    `json:"version"`
 	PreserveVersion           bool                   `json:"preserve_version"`
 	CurrentVersionUUID        string                 `json:"current_version_uuid"`
-	Description               *string                `json:"description"`
+	Description               string                 `json:"description"`
 }
 
 func (c Collection) resourceName() string {
@@ -51,6 +53,8 @@ func (c Collection) resourceName() string {
 
 // SizedDigests returns the hash+size part of each data block
 // referenced by the collection.
+//
+// Zero-length blocks are not included.
 func (c *Collection) SizedDigests() ([]SizedDigest, error) {
 	manifestText := c.ManifestText
 	if manifestText == "" {
@@ -74,6 +78,10 @@ func (c *Collection) SizedDigests() ([]SizedDigest, error) {
 				// FIXME: ensure it's a file token
 				break
 			}
+			if strings.HasPrefix(token, "d41d8cd98f00b204e9800998ecf8427e+0") {
+				// Exclude "empty block" placeholder
+				continue
+			}
 			// FIXME: shouldn't assume 32 char hash
 			if i := strings.IndexRune(token[33:], '+'); i >= 0 {
 				token = token[:33+i]
@@ -90,3 +98,28 @@ type CollectionList struct {
 	Offset         int          `json:"offset"`
 	Limit          int          `json:"limit"`
 }
+
+var (
+	blkRe = regexp.MustCompile(`^ [0-9a-f]{32}\+\d+`)
+	tokRe = regexp.MustCompile(` ?[^ ]*`)
+)
+
+// PortableDataHash computes the portable data hash of the given
+// manifest.
+func PortableDataHash(mt string) string {
+	h := md5.New()
+	size := 0
+	_ = tokRe.ReplaceAllFunc([]byte(mt), func(tok []byte) []byte {
+		if m := blkRe.Find(tok); m != nil {
+			// write hash+size, ignore remaining block hints
+			tok = m
+		}
+		n, err := h.Write(tok)
+		if err != nil {
+			panic(err)
+		}
+		size += n
+		return nil
+	})
+	return fmt.Sprintf("%x+%d", h.Sum(nil), size)
+}