Add split by total size

SabreTools.Core/README.1ST

@@ -721,6 +721,15 @@ Features and Options:
 			  first DAT, and everything greater than or equal goes in the
 			  second.
 
+	  -tis, --total-size		Split DAT(s) or folder by total game sizes
+		  For a DAT, or set of DATs, allow for splitting based on the combined
+		  sizes of the games, splitting into individual chunks.
+
+		  -cs=, --chunk-size=		Set a chunk size to output
+			  Set the total game size to cut off at for each chunked DAT. It is
+			  recommended to use a sufficiently large size such as 1GB or else
+			  you may run into issues.
+
 	  -ts, --type				 Split DAT(s) or folder by file types (rom/disk)
 		  For a DAT, or set of DATs, allow for splitting based on the types of
 		  the files, specifically if the type is a rom or a disk.

SabreTools.DatTools/Splitter.cs

@@ -353,6 +353,79 @@ namespace SabreTools.DatTools
             return (lessThan, greaterThan);
         }
 
+        /// <summary>
+        /// Split a DAT by size of Rom
+        /// </summary>
+        /// <param name="datFile">Current DatFile object to split</param>
+        /// <param name="chunkSize">Long value representing the total size to split at</param>
+        /// <returns>Less Than and Greater Than DatFiles</returns>
+        public static List<DatFile> SplitByTotalSize(DatFile datFile, long chunkSize)
+        {
+            // If the size is invalid, just return
+            if (chunkSize <= 0)
+                return new List<DatFile>();
+
+            // Create each of the respective output DATs
+            InternalStopwatch watch = new InternalStopwatch($"Splitting DAT by total size");
+
+            // Sort the DatFile by machine name
+            datFile.Items.BucketBy(ItemKey.Machine, DedupeType.None);
+
+            // Get the keys in a known order for easier sorting
+            var keys = datFile.Items.SortedKeys;
+
+            // Get the output list
+            List<DatFile> datFiles = new List<DatFile>();
+
+            // Initialize everything
+            long currentSize = 0;
+            long currentIndex = 0;
+            DatFile currentDat = DatFile.Create(datFile.Header.CloneStandard());
+            currentDat.Header.FileName += $"_{currentIndex}";
+            currentDat.Header.Name += $"_{currentIndex}";
+            currentDat.Header.Description += $"_{currentIndex}";
+
+            // Loop through each machine 
+            foreach (string machine in keys)
+            {
+                // Get the current machine
+                var items = datFile.Items[machine];
+                if (items == null || !items.Any())
+                    continue;
+
+                // Get the total size of the current machine
+                long machineSize = 0;
+                foreach (var item in items)
+                {
+                    if (item is Rom rom)
+                        machineSize += rom.Size ?? 0;
+                }
+
+                // If the current machine size makes the current DatFile too big, split
+                if (currentSize + machineSize > chunkSize)
+                {
+                    datFiles.Add(currentDat);
+                    currentSize = 0;
+                    currentIndex++;
+                    currentDat = DatFile.Create(datFile.Header.CloneStandard());
+                    currentDat.Header.FileName += $"_{currentIndex}";
+                    currentDat.Header.Name += $"_{currentIndex}";
+                    currentDat.Header.Description += $"_{currentIndex}";
+                }
+
+                // Add the current machine to the current DatFile
+                currentDat.Items[machine] = items;
+                currentSize += machineSize;
+            }
+
+            // Add the final DatFile to the list
+            datFiles.Add(currentDat);
+
+            // Then return the list
+            watch.Stop();
+            return datFiles;
+        }
+
         /// <summary>
         /// Split a DAT by type of DatItem
         /// </summary>

SabreTools/Features/BaseFeature.cs

@@ -38,10 +38,11 @@ namespace SabreTools.Features
             None = 0x00,
 
             Extension = 1 << 0,
-            Hash = 1 << 2,
+            Hash = 1 << 1,
-            Level = 1 << 3,
+            Level = 1 << 2,
-            Type = 1 << 4,
+            Type = 1 << 3,
-            Size = 1 << 5,
+            Size = 1 << 4,
+            TotalSize = 1 << 5,
         }
 
         /// <summary>
@@ -1080,6 +1081,20 @@ namespace SabreTools.Features
             }
         }
 
+        internal const string TotalSizeValue = "total-size";
+        internal static Feature TotalSizeFlag
+        {
+            get
+            {
+                return new Feature(
+                    TotalSizeValue,
+                    new List<string>() { "-tis", "--total-size" },
+                    "Split DAT(s) or folder by total game sizes",
+                    ParameterType.Flag,
+                    longDescription: "For a DAT, or set of DATs, allow for splitting based on the combined sizes of the games, splitting into individual chunks.");
+            }
+        }
+
         internal const string TrimValue = "trim";
         internal static Feature TrimFlag
         {
@@ -1172,6 +1187,20 @@ namespace SabreTools.Features
 
         #region Int64 features
 
+        internal const string ChunkSizeInt64Value = "chunk-size";
+        internal static Feature ChunkSizeInt64Input
+        {
+            get
+            {
+                return new Feature(
+                    ChunkSizeInt64Value,
+                    new List<string>() { "-cs", "--chunk-size" },
+                    "Set a chunk size to output",
+                    ParameterType.Int64,
+                    longDescription: "Set the total game size to cut off at for each chunked DAT. It is recommended to use a sufficiently large size such as 1GB or else you may run into issues.");
+            }
+        }
+
         internal const string RadixInt64Value = "radix";
         internal static Feature RadixInt64Input
         {
@@ -1958,6 +1987,8 @@ Some special strings that can be used:
                 splittingMode |= SplittingMode.Level;
             if (GetBoolean(features, SizeValue))
                 splittingMode |= SplittingMode.Size;
+            if (GetBoolean(features, TotalSizeValue))
+                splittingMode |= SplittingMode.TotalSize;
             if (GetBoolean(features, TypeValue))
                 splittingMode |= SplittingMode.Type;
 

SabreTools/Features/Split.cs

@@ -39,6 +39,8 @@ namespace SabreTools.Features
             this[LevelFlag].AddFeature(BaseFlag);
             AddFeature(SizeFlag);
             this[SizeFlag].AddFeature(RadixInt64Input);
+            AddFeature(TotalSizeFlag);
+            this[TotalSizeFlag].AddFeature(ChunkSizeInt64Input);
             AddFeature(TypeFlag);
         }
 
@@ -120,6 +122,23 @@ namespace SabreTools.Features
                     watch.Stop();
                 }
 
+                // Total Size splitting
+                if (splittingMode.HasFlag(SplittingMode.TotalSize))
+                {
+                    logger.Warning("This feature is not implemented: level-split");
+                    List<DatFile> sizedDats = DatTools.Splitter.SplitByTotalSize(internalDat, GetInt64(features, ChunkSizeInt64Value));
+
+                    InternalStopwatch watch = new InternalStopwatch("Outputting total-size-split DATs");
+
+                    // Loop through each type DatFile
+                    Parallel.ForEach(sizedDats, Globals.ParallelOptions, sizedDat =>
+                    {
+                        Writer.Write(sizedDat, OutputDir);
+                    });
+
+                    watch.Stop();
+                }
+
                 // Type splitting
                 if (splittingMode.HasFlag(SplittingMode.Type))
                 {