fix: read file size after streams are closed in converter

Tools/CacheConverter/Program.cs

@@ -102,47 +102,49 @@ await Parallel.ForEachAsync(jsonFiles, options, async (jsonFile, cancellationTok
             dataTable.Columns.Add(colName, colType);
         }
 
-        // Stream JSON and write to protobuf in batches
-        await using var inputFs = new FileStream(jsonFile, FileMode.Open, FileAccess.Read, FileShare.Read, 256 * 1024, FileOptions.SequentialScan | FileOptions.Asynchronous);
-        await using var decompressStream = new DecompressionStream(inputFs);
-        await using var outputFs = new FileStream(outputFile, FileMode.Create, FileAccess.Write, FileShare.None, 256 * 1024, FileOptions.Asynchronous);
-        await using var compressStream = new CompressionStream(outputFs, level: 10);
-
         int rowCount = 0;
         int batchCount = 0;
 
-        // True streaming: DeserializeAsyncEnumerable streams each array element without loading entire JSON
+        // Stream JSON and write to protobuf in batches
-        var jsonOptions = new JsonSerializerOptions { PropertyNameCaseInsensitive = true };
+        // Use explicit dispose to ensure file is closed before reading size
-        await foreach (var element in JsonSerializer.DeserializeAsyncEnumerable<JsonElement>(
-            decompressStream,
-            jsonOptions,
-            cancellationToken))
         {
-            var row = dataTable.NewRow();
+            await using var inputFs = new FileStream(jsonFile, FileMode.Open, FileAccess.Read, FileShare.Read, 256 * 1024, FileOptions.SequentialScan | FileOptions.Asynchronous);
-            ReadJsonElement(element, row, dataTable);
+            await using var decompressStream = new DecompressionStream(inputFs);
-            dataTable.Rows.Add(row);
+            await using var outputFs = new FileStream(outputFile, FileMode.Create, FileAccess.Write, FileShare.None, 256 * 1024, FileOptions.Asynchronous);
-            rowCount++;
+            await using var compressStream = new CompressionStream(outputFs, level: 10);
 
-            // Write batch when we hit the batch size
+            // True streaming: DeserializeAsyncEnumerable streams each array element without loading entire JSON
-            if (dataTable.Rows.Count >= BatchSize)
+            var jsonOptions = new JsonSerializerOptions { PropertyNameCaseInsensitive = true };
+            await foreach (var element in JsonSerializer.DeserializeAsyncEnumerable<JsonElement>(
+                decompressStream,
+                jsonOptions,
+                cancellationToken))
+            {
+                var row = dataTable.NewRow();
+                ReadJsonElement(element, row, dataTable);
+                dataTable.Rows.Add(row);
+                rowCount++;
+
+                // Write batch when we hit the batch size
+                if (dataTable.Rows.Count >= BatchSize)
+                {
+                    using var reader = dataTable.CreateDataReader();
+                    DataSerializer.Serialize(compressStream, reader);
+                    dataTable.Clear();
+                    batchCount++;
+                }
+            }
+
+            // Write remaining rows
+            if (dataTable.Rows.Count > 0)
             {
                 using var reader = dataTable.CreateDataReader();
                 DataSerializer.Serialize(compressStream, reader);
-                dataTable.Clear();
                 batchCount++;
             }
-        }
+        } // Streams closed here
-
-        // Write remaining rows
-        if (dataTable.Rows.Count > 0)
-        {
-            using var reader = dataTable.CreateDataReader();
-            DataSerializer.Serialize(compressStream, reader);
-            batchCount++;
-        }
-
-        await compressStream.FlushAsync(cancellationToken);
 
+        // Read file size after streams are fully closed
         var newSize = new FileInfo(outputFile).Length;
         Interlocked.Add(ref totalNewSize, newSize);