Support data symbols in WASM binary format

4 files changed, 45 insertions, 2 deletions

diff --git a/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/linking/WasmDataSegmentIndex.java b/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/linking/WasmDataSegmentIndex.java
new file mode 100644
index 00000000..80124ae9
--- /dev/null
+++ b/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/linking/WasmDataSegmentIndex.java
@@ -0,0 +1,4 @@
+package org.shedlang.compiler.backends.wasm.structure.ast.linking;
+
+public record WasmDataSegmentIndex(int value) {
+}
diff --git a/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/linking/WasmDataSymbol.java b/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/linking/WasmDataSymbol.java
new file mode 100644
index 00000000..864d46e9
--- /dev/null
+++ b/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/linking/WasmDataSymbol.java
@@ -0,0 +1,9 @@
+package org.shedlang.compiler.backends.wasm.structure.ast.linking;
+
+public record WasmDataSymbol(
+    String name,
+    WasmDataSegmentIndex dataSegmentIndex,
+    int offset,
+    int size
+) {
+}
diff --git a/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/modules/WasmDataSegment.java b/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/modules/WasmDataSegment.java
index 7227f54b..911a6a01 100644
--- a/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/modules/WasmDataSegment.java
+++ b/src/main/java/org/shedlang/compiler/backends/wasm/structure/ast/modules/WasmDataSegment.java
@@ -1,11 +1,15 @@
 package org.shedlang.compiler.backends.wasm.structure.ast.modules;
 
 import org.shedlang.compiler.backends.wasm.structure.ast.instructions.WasmExpression;
+import org.shedlang.compiler.backends.wasm.structure.ast.linking.WasmDataSymbol;
+
+import java.util.List;
 
 public record WasmDataSegment(
     String name,
     byte[] init,
     WasmExpression offset,
-    WasmAlignment alignment
+    WasmAlignment alignment,
+    List<WasmDataSymbol> symbols
 ) {
 }
diff --git a/src/main/java/org/shedlang/compiler/backends/wasm/structure/binary/WasmBinaryFormatLinking.java b/src/main/java/org/shedlang/compiler/backends/wasm/structure/binary/WasmBinaryFormatLinking.java
index 3cb75953..ecc06f71 100644
--- a/src/main/java/org/shedlang/compiler/backends/wasm/structure/binary/WasmBinaryFormatLinking.java
+++ b/src/main/java/org/shedlang/compiler/backends/wasm/structure/binary/WasmBinaryFormatLinking.java
@@ -1,6 +1,8 @@
 package org.shedlang.compiler.backends.wasm.structure.binary;
 
 import org.shedlang.compiler.backends.wasm.structure.BinaryWriter;
+import org.shedlang.compiler.backends.wasm.structure.ast.linking.WasmDataSegmentIndex;
+import org.shedlang.compiler.backends.wasm.structure.ast.linking.WasmDataSymbol;
 import org.shedlang.compiler.backends.wasm.structure.ast.linking.WasmInitFunc;
 import org.shedlang.compiler.backends.wasm.structure.ast.linking.WasmSymbolIndex;
 import org.shedlang.compiler.backends.wasm.structure.ast.modules.WasmAlignment;
@@ -136,6 +138,14 @@ public class WasmBinaryFormatLinking {
                 writeFuncIndex(function.funcIndex, writer);
                 writeString(function.name(), writer);
             }
+            case SymbolInfo.Data data -> {
+                writer.writeByte(SymbolType.DATA.id);
+                writeFlags(data.flags, writer);
+                writeString(data.dataSymbol.name(), writer);
+                writeDataSegmentIndex(data.dataSymbol.dataSegmentIndex(), writer);
+                Leb128Encoding.encodeUnsignedInt32(data.dataSymbol.offset(), writer);
+                Leb128Encoding.encodeUnsignedInt32(data.dataSymbol.size(), writer);
+            }
             case SymbolInfo.Global global -> {
                 writer.writeByte(SymbolType.GLOBAL.id);
                 writeFlags(global.flags, writer);
@@ -192,6 +202,12 @@ public class WasmBinaryFormatLinking {
                 );
             });
 
+        var dataSymbols = module.dataSegments().stream()
+            .flatMap(dataSegment ->
+                dataSegment.symbols().stream()
+                    .map(dataSymbol -> new SymbolInfo.Data(List.of(), dataSymbol))
+            );
+
         var globalSymbols = module.indexedGlobals().stream()
             .map(global -> {
                 var flags = switch (global) {
@@ -208,7 +224,10 @@ public class WasmBinaryFormatLinking {
                 );
             });
 
-        var symbolInfos = Stream.<SymbolInfo>concat(functionSymbols, globalSymbols).toList();
+        var symbolInfos = Stream.<Stream<? extends SymbolInfo>>of(functionSymbols, dataSymbols, globalSymbols)
+            .<SymbolInfo>flatMap(x -> x)
+            .toList();
+
         return new SymbolTable(symbolInfos, functionSymbolIndexOffset);
     }
 
@@ -218,6 +237,9 @@ public class WasmBinaryFormatLinking {
 
         record Global(List<SymbolFlag> flags, WasmGlobalIndex globalIndex, String name) implements SymbolInfo {
         }
+
+        record Data(List<SymbolFlag> flags, WasmDataSymbol dataSymbol) implements SymbolInfo {
+        }
     }
 
     private enum SymbolFlag {
@@ -250,6 +272,10 @@ public class WasmBinaryFormatLinking {
         Leb128Encoding.encodeUnsignedInt32(flagsToInt(flags), writer);
     }
 
+    private static void writeDataSegmentIndex(WasmDataSegmentIndex dataSegmentIndex, BinaryWriter writer) {
+        Leb128Encoding.encodeUnsignedInt32(dataSegmentIndex.value(), writer);
+    }
+
     private static void writeSymbolIndex(WasmSymbolIndex symbolIndex, BinaryWriter writer) {
         Leb128Encoding.encodeUnsignedInt32(symbolIndex.value(), writer);
     }