Randomly generated media IDs (#1336)

* Filenames based on hashes but media IDs randomly generated and per-upload metadata

* Lint

* Comment generateMediaID

5 files changed, 85 insertions, 6 deletions

diff --git a/mediaapi/storage/interface.go b/mediaapi/storage/interface.go
index 672e8ef5..84319971 100644
--- a/mediaapi/storage/interface.go
+++ b/mediaapi/storage/interface.go
@@ -24,6 +24,7 @@ import (
 type Database interface {
 	StoreMediaMetadata(ctx context.Context, mediaMetadata *types.MediaMetadata) error
 	GetMediaMetadata(ctx context.Context, mediaID types.MediaID, mediaOrigin gomatrixserverlib.ServerName) (*types.MediaMetadata, error)
+	GetMediaMetadataByHash(ctx context.Context, mediaHash types.Base64Hash, mediaOrigin gomatrixserverlib.ServerName) (*types.MediaMetadata, error)
 	StoreThumbnail(ctx context.Context, thumbnailMetadata *types.ThumbnailMetadata) error
 	GetThumbnail(ctx context.Context, mediaID types.MediaID, mediaOrigin gomatrixserverlib.ServerName, width, height int, resizeMethod string) (*types.ThumbnailMetadata, error)
 	GetThumbnails(ctx context.Context, mediaID types.MediaID, mediaOrigin gomatrixserverlib.ServerName) ([]*types.ThumbnailMetadata, error)
diff --git a/mediaapi/storage/postgres/media_repository_table.go b/mediaapi/storage/postgres/media_repository_table.go
index e975530a..1d3264ca 100644
--- a/mediaapi/storage/postgres/media_repository_table.go
+++ b/mediaapi/storage/postgres/media_repository_table.go
@@ -59,9 +59,14 @@ const selectMediaSQL = `
 SELECT content_type, file_size_bytes, creation_ts, upload_name, base64hash, user_id FROM mediaapi_media_repository WHERE media_id = $1 AND media_origin = $2
 `
 
+const selectMediaByHashSQL = `
+SELECT content_type, file_size_bytes, creation_ts, upload_name, media_id, user_id FROM mediaapi_media_repository WHERE base64hash = $1 AND media_origin = $2
+`
+
 type mediaStatements struct {
-	insertMediaStmt *sql.Stmt
-	selectMediaStmt *sql.Stmt
+	insertMediaStmt       *sql.Stmt
+	selectMediaStmt       *sql.Stmt
+	selectMediaByHashStmt *sql.Stmt
 }
 
 func (s *mediaStatements) prepare(db *sql.DB) (err error) {
@@ -73,6 +78,7 @@ func (s *mediaStatements) prepare(db *sql.DB) (err error) {
 	return statementList{
 		{&s.insertMediaStmt, insertMediaSQL},
 		{&s.selectMediaStmt, selectMediaSQL},
+		{&s.selectMediaByHashStmt, selectMediaByHashSQL},
 	}.prepare(db)
 }
 
@@ -113,3 +119,23 @@ func (s *mediaStatements) selectMedia(
 	)
 	return &mediaMetadata, err
 }
+
+func (s *mediaStatements) selectMediaByHash(
+	ctx context.Context, mediaHash types.Base64Hash, mediaOrigin gomatrixserverlib.ServerName,
+) (*types.MediaMetadata, error) {
+	mediaMetadata := types.MediaMetadata{
+		Base64Hash: mediaHash,
+		Origin:     mediaOrigin,
+	}
+	err := s.selectMediaStmt.QueryRowContext(
+		ctx, mediaMetadata.Base64Hash, mediaMetadata.Origin,
+	).Scan(
+		&mediaMetadata.ContentType,
+		&mediaMetadata.FileSizeBytes,
+		&mediaMetadata.CreationTimestamp,
+		&mediaMetadata.UploadName,
+		&mediaMetadata.MediaID,
+		&mediaMetadata.UserID,
+	)
+	return &mediaMetadata, err
+}
diff --git a/mediaapi/storage/postgres/storage.go b/mediaapi/storage/postgres/storage.go
index 756913d3..f89501de 100644
--- a/mediaapi/storage/postgres/storage.go
+++ b/mediaapi/storage/postgres/storage.go
@@ -67,6 +67,19 @@ func (d *Database) GetMediaMetadata(
 	return mediaMetadata, err
 }
 
+// GetMediaMetadataByHash returns metadata about media stored on this server.
+// The media could have been uploaded to this server or fetched from another server and cached here.
+// Returns nil metadata if there is no metadata associated with this media.
+func (d *Database) GetMediaMetadataByHash(
+	ctx context.Context, mediaHash types.Base64Hash, mediaOrigin gomatrixserverlib.ServerName,
+) (*types.MediaMetadata, error) {
+	mediaMetadata, err := d.statements.media.selectMediaByHash(ctx, mediaHash, mediaOrigin)
+	if err != nil && err == sql.ErrNoRows {
+		return nil, nil
+	}
+	return mediaMetadata, err
+}
+
 // StoreThumbnail inserts the metadata about the thumbnail into the database.
 // Returns an error if the combination of MediaID and Origin are not unique in the table.
 func (d *Database) StoreThumbnail(
diff --git a/mediaapi/storage/sqlite3/media_repository_table.go b/mediaapi/storage/sqlite3/media_repository_table.go
index dcc1b41e..bcef609d 100644
--- a/mediaapi/storage/sqlite3/media_repository_table.go
+++ b/mediaapi/storage/sqlite3/media_repository_table.go
@@ -60,11 +60,16 @@ const selectMediaSQL = `
 SELECT content_type, file_size_bytes, creation_ts, upload_name, base64hash, user_id FROM mediaapi_media_repository WHERE media_id = $1 AND media_origin = $2
 `
 
+const selectMediaByHashSQL = `
+SELECT content_type, file_size_bytes, creation_ts, upload_name, media_id, user_id FROM mediaapi_media_repository WHERE base64hash = $1 AND media_origin = $2
+`
+
 type mediaStatements struct {
-	db              *sql.DB
-	writer          sqlutil.Writer
-	insertMediaStmt *sql.Stmt
-	selectMediaStmt *sql.Stmt
+	db                    *sql.DB
+	writer                sqlutil.Writer
+	insertMediaStmt       *sql.Stmt
+	selectMediaStmt       *sql.Stmt
+	selectMediaByHashStmt *sql.Stmt
 }
 
 func (s *mediaStatements) prepare(db *sql.DB, writer sqlutil.Writer) (err error) {
@@ -79,6 +84,7 @@ func (s *mediaStatements) prepare(db *sql.DB, writer sqlutil.Writer) (err error)
 	return statementList{
 		{&s.insertMediaStmt, insertMediaSQL},
 		{&s.selectMediaStmt, selectMediaSQL},
+		{&s.selectMediaByHashStmt, selectMediaByHashSQL},
 	}.prepare(db)
 }
 
@@ -122,3 +128,23 @@ func (s *mediaStatements) selectMedia(
 	)
 	return &mediaMetadata, err
 }
+
+func (s *mediaStatements) selectMediaByHash(
+	ctx context.Context, mediaHash types.Base64Hash, mediaOrigin gomatrixserverlib.ServerName,
+) (*types.MediaMetadata, error) {
+	mediaMetadata := types.MediaMetadata{
+		Base64Hash: mediaHash,
+		Origin:     mediaOrigin,
+	}
+	err := s.selectMediaStmt.QueryRowContext(
+		ctx, mediaMetadata.Base64Hash, mediaMetadata.Origin,
+	).Scan(
+		&mediaMetadata.ContentType,
+		&mediaMetadata.FileSizeBytes,
+		&mediaMetadata.CreationTimestamp,
+		&mediaMetadata.UploadName,
+		&mediaMetadata.MediaID,
+		&mediaMetadata.UserID,
+	)
+	return &mediaMetadata, err
+}
diff --git a/mediaapi/storage/sqlite3/storage.go b/mediaapi/storage/sqlite3/storage.go
index d5c3031e..9e510fa3 100644
--- a/mediaapi/storage/sqlite3/storage.go
+++ b/mediaapi/storage/sqlite3/storage.go
@@ -70,6 +70,19 @@ func (d *Database) GetMediaMetadata(
 	return mediaMetadata, err
 }
 
+// GetMediaMetadataByHash returns metadata about media stored on this server.
+// The media could have been uploaded to this server or fetched from another server and cached here.
+// Returns nil metadata if there is no metadata associated with this media.
+func (d *Database) GetMediaMetadataByHash(
+	ctx context.Context, mediaHash types.Base64Hash, mediaOrigin gomatrixserverlib.ServerName,
+) (*types.MediaMetadata, error) {
+	mediaMetadata, err := d.statements.media.selectMediaByHash(ctx, mediaHash, mediaOrigin)
+	if err != nil && err == sql.ErrNoRows {
+		return nil, nil
+	}
+	return mediaMetadata, err
+}
+
 // StoreThumbnail inserts the metadata about the thumbnail into the database.
 // Returns an error if the combination of MediaID and Origin are not unique in the table.
 func (d *Database) StoreThumbnail(