Merge pull request #3851 from aduffeck/keep-index-in-sync

Fix keeping the index in sync when directories contain special chars
owncloud · May 23, 2022 · 8130d4e · 8130d4e
2 parents cd87cae + 644386b
commit 8130d4e
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 14 deletions.
diff --git a/changelog/unreleased/fix-index-integrity.md b/changelog/unreleased/fix-index-integrity.md
@@ -0,0 +1,5 @@
+Bugfix: Fix search index getting out of sync
+
+We fixed a problem where the search index got out of sync with child elements of a parent containing special characters.
+
+https://github.com/owncloud/ocis/pull/3851
diff --git a/extensions/search/pkg/search/index/index.go b/extensions/search/pkg/search/index/index.go
@@ -23,6 +23,7 @@ import (
 	"errors"
 	"math"
 	"path"
+	"regexp"
 	"strings"
 	"time"
 
@@ -111,7 +112,7 @@ func (i *Index) markAsDeleted(id string, deleted bool) error {
 	if doc.Type == uint64(sprovider.ResourceType_RESOURCE_TYPE_CONTAINER) {
 		query := bleve.NewConjunctionQuery(
 			bleve.NewQueryStringQuery("RootID:"+doc.RootID),
-			bleve.NewQueryStringQuery("Path:"+doc.Path+"/*"),
+			bleve.NewQueryStringQuery("Path:"+queryEscape(doc.Path+"/*")),
 		)
 		bleveReq := bleve.NewSearchRequest(query)
 		bleveReq.Size = math.MaxInt
@@ -187,7 +188,7 @@ func (i *Index) Move(id *sprovider.ResourceId, fullPath string) error {
 	if doc.Type == uint64(sprovider.ResourceType_RESOURCE_TYPE_CONTAINER) {
 		query := bleve.NewConjunctionQuery(
 			bleve.NewQueryStringQuery("RootID:"+doc.RootID),
-			bleve.NewQueryStringQuery("Path:"+oldName+"/*"),
+			bleve.NewQueryStringQuery("Path:"+queryEscape(oldName+"/*")),
 		)
 		bleveReq := bleve.NewSearchRequest(query)
 		bleveReq.Size = math.MaxInt
@@ -217,8 +218,8 @@ func (i *Index) Search(ctx context.Context, req *searchsvc.SearchIndexRequest) (
 	query := bleve.NewConjunctionQuery(
 		bleve.NewQueryStringQuery(req.Query),
 		deletedQuery, // Skip documents that have been marked as deleted
-		bleve.NewQueryStringQuery("RootID:"+req.Ref.ResourceId.StorageId+"!"+req.Ref.ResourceId.OpaqueId), // Limit search to the space
-		bleve.NewQueryStringQuery("Path:"+utils.MakeRelativePath(path.Join(req.Ref.Path, "/"))+"*"),       // Limit search to this directory in the space
+		bleve.NewQueryStringQuery("RootID:"+req.Ref.ResourceId.StorageId+"!"+req.Ref.ResourceId.OpaqueId),        // Limit search to the space
+		bleve.NewQueryStringQuery("Path:"+queryEscape(utils.MakeRelativePath(path.Join(req.Ref.Path, "/"))+"*")), // Limit search to this directory in the space
 	)
 	bleveReq := bleve.NewSearchRequest(query)
 	bleveReq.Size = 200
@@ -340,3 +341,8 @@ func idToBleveId(id *sprovider.ResourceId) string {
 	}
 	return id.StorageId + "!" + id.OpaqueId
 }
+
+func queryEscape(s string) string {
+	re := regexp.MustCompile(`([` + regexp.QuoteMeta(`+=&|><!(){}[]^\"~*?:\/`) + `\-\s])`)
+	return re.ReplaceAllString(s, "\\$1")
+}
diff --git a/extensions/search/pkg/search/index/index_test.go b/extensions/search/pkg/search/index/index_test.go
@@ -29,21 +29,21 @@ var _ = Describe("Index", func() {
 		ri        *sprovider.ResourceInfo
 		parentRef = &sprovider.Reference{
 			ResourceId: rootId,
-			Path:       "./my/sudbir",
+			Path:       "./my/sub d!r",
 		}
 		parentRi = &sprovider.ResourceInfo{
 			Id: &sprovider.ResourceId{
 				StorageId: "storageid",
 				OpaqueId:  "parentopaqueid",
 			},
-			Path:  "subdir",
+			Path:  "sub d!r",
 			Size:  12345,
 			Type:  sprovider.ResourceType_RESOURCE_TYPE_CONTAINER,
 			Mtime: &typesv1beta1.Timestamp{Seconds: 4000},
 		}
 		childRef = &sprovider.Reference{
 			ResourceId: rootId,
-			Path:       "./my/sudbir/child.pdf",
+			Path:       "./my/sub d!r/child.pdf",
 		}
 		childRi = &sprovider.ResourceInfo{
 			Id: &sprovider.ResourceId{
@@ -298,12 +298,12 @@ var _ = Describe("Index", func() {
 		It("marks a resource as deleted", func() {
 			err := i.Add(parentRef, parentRi)
 			Expect(err).ToNot(HaveOccurred())
-			assertDocCount(rootId, "subdir", 1)
+			assertDocCount(rootId, `sub\ d!r`, 1)
 
 			err = i.Delete(parentRi.Id)
 			Expect(err).ToNot(HaveOccurred())
 
-			assertDocCount(rootId, "subdir", 0)
+			assertDocCount(rootId, `sub\ d!r`, 0)
 		})
 
 		It("also marks child resources as deleted", func() {
@@ -312,13 +312,13 @@ var _ = Describe("Index", func() {
 			err = i.Add(childRef, childRi)
 			Expect(err).ToNot(HaveOccurred())
 
-			assertDocCount(rootId, "subdir", 1)
+			assertDocCount(rootId, `sub\ d\!r`, 1)
 			assertDocCount(rootId, "child.pdf", 1)
 
 			err = i.Delete(parentRi.Id)
 			Expect(err).ToNot(HaveOccurred())
 
-			assertDocCount(rootId, "subdir", 0)
+			assertDocCount(rootId, `sub\ d\!r`, 0)
 			assertDocCount(rootId, "child.pdf", 0)
 		})
 	})
@@ -332,13 +332,13 @@ var _ = Describe("Index", func() {
 			err = i.Delete(parentRi.Id)
 			Expect(err).ToNot(HaveOccurred())
 
-			assertDocCount(rootId, "subdir", 0)
+			assertDocCount(rootId, `sub\ d!r`, 0)
 			assertDocCount(rootId, "child.pdf", 0)
 
 			err = i.Restore(parentRi.Id)
 			Expect(err).ToNot(HaveOccurred())
 
-			assertDocCount(rootId, "subdir", 1)
+			assertDocCount(rootId, `sub\ d!r`, 1)
 			assertDocCount(rootId, "child.pdf", 1)
 		})
 	})
@@ -354,7 +354,7 @@ var _ = Describe("Index", func() {
 			err = i.Move(parentRi.Id, "./somewhere/else/newname")
 			Expect(err).ToNot(HaveOccurred())
 
-			assertDocCount(rootId, "subdir", 0)
+			assertDocCount(rootId, `sub\ d!r`, 0)
 
 			matches := assertDocCount(rootId, "Name:child.pdf", 1)
 			Expect(matches[0].Entity.Ref.Path).To(Equal("./somewhere/else/newname/child.pdf"))