fix: ignore zero or empty CRC and check file size (#537)

libretro · Mar 12, 2024 · 2140a68 · 2140a68
1 parent f83948d
commit 2140a68
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 8 deletions.
diff --git a/dat/dat.go b/dat/dat.go
@@ -36,6 +36,7 @@ type CRC uint32
 type ROM struct {
 	XMLName xml.Name `xml:"rom"`
 	Name    string   `xml:"name,attr"`
+	Size    int64    `xml:"size,attr"`
 	CRC     CRC      `xml:"crc,attr"`
 }
 
@@ -63,7 +64,12 @@ func Parse(dat []byte) Dat {
 }
 
 // FindByCRC loops over the Dats in the DB and concurrently matches CRC checksums.
-func (db *DB) FindByCRC(romPath string, romName string, crc uint32, games chan (Game)) {
+func (db *DB) FindByCRC(romPath string, romName string, crc uint32, size int64, games chan (Game)) {
+	// skip empty file
+	if size == 0 || crc == 0 {
+		return
+	}
+
 	var wg sync.WaitGroup
 	wg.Add(len(*db))
 	// For every Dat in the DB
@@ -74,8 +80,8 @@ func (db *DB) FindByCRC(romPath string, romName string, crc uint32, games chan (
 				if len(game.ROMs) == 0 {
 					continue
 				}
-				// If the checksums match
-				if crc == uint32(game.ROMs[0].CRC) {
+				// If the checksums and sizes match
+				if crc == uint32(game.ROMs[0].CRC) && size == game.ROMs[0].Size {
 					game.Path = romPath
 					game.System = system
 					games <- game

diff --git a/scanner/scanner.go b/scanner/scanner.go
@@ -11,6 +11,7 @@ import (
 	"path/filepath"
 	"strconv"
 	"strings"
+
 	"github.com/libretro/ludo/dat"
 	ntf "github.com/libretro/ludo/notifications"
 	"github.com/libretro/ludo/playlists"
@@ -113,19 +114,20 @@ func Scan(dir string, roms []string, games chan (dat.Game), n *ntf.Notification)
 			}
 			for _, rom := range z.File {
 				romExt := filepath.Ext(rom.Name)
+				size := int64(rom.UncompressedSize64)
 				// these 4 systems might have headered or headerless roms and need special logic
 				if headerSize, ok := headerSizes[romExt]; ok {
 					crc, crcHeaderless, err := checksumHeaderless(rom, headerSize)
 					if err != nil {
 						n.Update(ntf.Error, err.Error())
 						continue
 					}
-					state.DB.FindByCRC(f, rom.Name, crc, games)
-					state.DB.FindByCRC(f, rom.Name, crcHeaderless, games)
+					state.DB.FindByCRC(f, rom.Name, crc, size, games)
+					state.DB.FindByCRC(f, rom.Name, crcHeaderless, size-int64(headerSize), games)
 					n.Update(ntf.Info, strconv.Itoa(i)+"/"+strconv.Itoa(len(roms))+" "+f)
 				} else if rom.CRC32 > 0 {
 					// Look for a matching game entry in the database
-					state.DB.FindByCRC(f, rom.Name, rom.CRC32, games)
+					state.DB.FindByCRC(f, rom.Name, rom.CRC32, size, games)
 					n.Update(ntf.Info, strconv.Itoa(i)+"/"+strconv.Itoa(len(roms))+" "+f)
 				}
 			}
@@ -140,11 +142,16 @@ func Scan(dir string, roms []string, games chan (dat.Game), n *ntf.Notification)
 				n.Update(ntf.Error, err.Error())
 				continue
 			}
+			s, err := os.Stat(f)
+			if err != nil {
+				n.Update(ntf.Error, err.Error())
+				continue
+			}
 			crc := crc32.ChecksumIEEE(bytes)
-			state.DB.FindByCRC(f, utils.FileName(f), crc, games)
+			state.DB.FindByCRC(f, utils.FileName(f), crc, s.Size(), games)
 			if headerSize, ok := headerSizes[ext]; ok {
 				crcHeaderless := crc32.ChecksumIEEE(bytes[headerSize:])
-				state.DB.FindByCRC(f, utils.FileName(f), crcHeaderless, games)
+				state.DB.FindByCRC(f, utils.FileName(f), crcHeaderless, s.Size()-int64(headerSize), games)
 			}
 			n.Update(ntf.Info, strconv.Itoa(i)+"/"+strconv.Itoa(len(roms))+" "+f)
 		}