search: allow searching by file wholeref

pkg/index/corpus.go

@@ -1156,6 +1156,11 @@ func (c *Corpus) GetMediaTagsLocked(fileRef blob.Ref) (map[string]string, error)
 	return tags, nil
 }
 
+func (c *Corpus) GetWholeRefLocked(fileRef blob.Ref) (wholeRef blob.Ref, ok bool) {
+	wholeRef, ok = c.fileWholeRef[fileRef]
+	return
+}
+
 func (c *Corpus) FileLatLongLocked(fileRef blob.Ref) (lat, long float64, ok bool) {
 	wholeRef, ok := c.fileWholeRef[fileRef]
 	if !ok {

pkg/search/query.go

@@ -321,11 +321,17 @@ func (c *Constraint) onlyMatchesPermanode() bool {
 type FileConstraint struct {
 	// (All non-zero fields must match)
 
-	FileSize *IntConstraint `json:"fileSize,omitempty"`
-	FileName *StringConstraint
-	MIMEType *StringConstraint
-	Time     *TimeConstraint
-	ModTime  *TimeConstraint
+	FileSize *IntConstraint    `json:"fileSize,omitempty"`
+	FileName *StringConstraint `json:"fileName,omitempty"`
+	MIMEType *StringConstraint `json:"mimeType,omitempty"`
+	Time     *TimeConstraint   `json:"time,omitempty"`
+	ModTime  *TimeConstraint   `json:"modTime,omitempty"`
+
+	// WholeRef if non-zero only matches if the entire checksum of the
+	// file (the concatenation of all its blobs) is equal to the
+	// provided blobref. The index may not have every file's digest for
+	// every known hash algorithm.
+	WholeRef blob.Ref `json:"wholeRef,omitempty"`
 
 	// For images:
 	IsImage  bool                `json:"isImage,omitempty"`
@@ -1403,6 +1409,15 @@ func (c *FileConstraint) blobMatches(s *search, br blob.Ref, bm camtypes.BlobMet
 		}
 	}
 	corpus := s.h.corpus
+	if c.WholeRef.Valid() {
+		if corpus == nil {
+			return false, nil
+		}
+		wholeRef, ok := corpus.GetWholeRefLocked(br)
+		if !ok || wholeRef != c.WholeRef {
+			return false, nil
+		}
+	}
 	var width, height int64
 	if c.Width != nil || c.Height != nil || c.WHRatio != nil {
 		if corpus == nil {

pkg/search/query_test.go

@@ -569,6 +569,35 @@ func TestQueryFileConstraint(t *testing.T) {
 	})
 }
 
+func TestQueryFileConstraint_WholeRef(t *testing.T) {
+	testQueryTypes(t, memIndexTypes, func(qt *queryTest) {
+		id := qt.id
+		fileRef, _ := id.UploadFile("some-stuff.txt", "hello", time.Unix(123, 0))
+		qt.t.Logf("fileRef = %q", fileRef)
+		p1 := id.NewPlannedPermanode("1")
+		id.SetAttribute(p1, "camliContent", fileRef.String())
+
+		fileRef2, _ := id.UploadFile("other-file", "hellooooo", time.Unix(456, 0))
+		qt.t.Logf("fileRef2 = %q", fileRef2)
+		p2 := id.NewPlannedPermanode("2")
+		id.SetAttribute(p2, "camliContent", fileRef2.String())
+
+		sq := &SearchQuery{
+			Constraint: &Constraint{
+				Permanode: &PermanodeConstraint{
+					Attr: "camliContent",
+					ValueInSet: &Constraint{
+						File: &FileConstraint{
+							WholeRef: blob.SHA1FromString("hello"),
+						},
+					},
+				},
+			},
+		}
+		qt.wantRes(sq, p1)
+	})
+}
+
 func TestQueryPermanodeModtime(t *testing.T) {
 	testQuery(t, func(qt *queryTest) {
 		id := qt.id