Merge branch '16265-security-updates' into dependabot/bundler/services/api/nokogiri...

[arvados.git] / sdk / go / arvados / collection.go

diff --git a/sdk/go/arvados/collection.go b/sdk/go/arvados/collection.go
index 35fd3fd7406fcd24b0f167d38fc9dd009f0e4d69..030665d77f7a075f289c92d4a715f1d414de9a9b 100644 (file)
--- 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"
 
@@ -90,3 +92,28 @@ 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)
+}