[Strings] string.encode (#4776)

4 files changed, 74 insertions, 0 deletions

diff --git a/src/wasm/wasm-binary.cpp b/src/wasm/wasm-binary.cpp
index d81aa0ca4..429ff7bae 100644
--- a/src/wasm/wasm-binary.cpp
+++ b/src/wasm/wasm-binary.cpp
@@ -3927,6 +3927,9 @@ BinaryConsts::ASTNodes WasmBinaryBuilder::readExpression(Expression*& curr) {
       if (maybeVisitStringMeasure(curr, opcode)) {
         break;
       }
+      if (maybeVisitStringEncode(curr, opcode)) {
+        break;
+      }
       if (opcode == BinaryConsts::RefIsFunc ||
           opcode == BinaryConsts::RefIsData ||
           opcode == BinaryConsts::RefIsI31) {
@@ -7190,6 +7193,33 @@ bool WasmBinaryBuilder::maybeVisitStringMeasure(Expression*& out,
   return true;
 }
 
+bool WasmBinaryBuilder::maybeVisitStringEncode(Expression*& out,
+                                               uint32_t code) {
+  StringEncodeOp op;
+  // TODO: share this code with string.measure?
+  if (code == BinaryConsts::StringEncodeWTF8) {
+    auto policy = getU32LEB();
+    switch (policy) {
+      case BinaryConsts::StringPolicy::UTF8:
+        op = StringEncodeUTF8;
+        break;
+      case BinaryConsts::StringPolicy::WTF8:
+        op = StringEncodeWTF8;
+        break;
+      default:
+        throwError("bad policy for string.encode");
+    }
+  } else if (code == BinaryConsts::StringEncodeWTF16) {
+    op = StringEncodeWTF16;
+  } else {
+    return false;
+  }
+  auto* ptr = popNonVoidExpression();
+  auto* ref = popNonVoidExpression();
+  out = Builder(wasm).makeStringEncode(op, ref, ptr);
+  return true;
+}
+
 void WasmBinaryBuilder::visitRefAs(RefAs* curr, uint8_t code) {
   BYN_TRACE("zz node: RefAs\n");
   switch (code) {
diff --git a/src/wasm/wasm-s-parser.cpp b/src/wasm/wasm-s-parser.cpp
index 98ed64c2d..68bd69104 100644
--- a/src/wasm/wasm-s-parser.cpp
+++ b/src/wasm/wasm-s-parser.cpp
@@ -2973,6 +2973,23 @@ Expression* SExpressionWasmBuilder::makeStringMeasure(Element& s,
   return Builder(wasm).makeStringMeasure(op, parseExpression(s[i]));
 }
 
+Expression* SExpressionWasmBuilder::makeStringEncode(Element& s,
+                                                     StringEncodeOp op) {
+  size_t i = 1;
+  if (op == StringEncodeWTF8) {
+    const char* str = s[i++]->c_str();
+    if (strncmp(str, "utf8", 4) == 0) {
+      op = StringEncodeUTF8;
+    } else if (strncmp(str, "wtf8", 4) == 0) {
+      op = StringEncodeWTF8;
+    } else {
+      throw ParseException("bad string.new op", s.line, s.col);
+    }
+  }
+  return Builder(wasm).makeStringEncode(
+    op, parseExpression(s[i]), parseExpression(s[i + 1]));
+}
+
 // converts an s-expression string representing binary data into an output
 // sequence of raw bytes this appends to data, which may already contain
 // content.
diff --git a/src/wasm/wasm-stack.cpp b/src/wasm/wasm-stack.cpp
index 9c086bb47..38c8f1661 100644
--- a/src/wasm/wasm-stack.cpp
+++ b/src/wasm/wasm-stack.cpp
@@ -2281,6 +2281,25 @@ void BinaryInstWriter::visitStringMeasure(StringMeasure* curr) {
   }
 }
 
+void BinaryInstWriter::visitStringEncode(StringEncode* curr) {
+  o << int8_t(BinaryConsts::GCPrefix);
+  switch (curr->op) {
+    case StringEncodeUTF8:
+      o << U32LEB(BinaryConsts::StringEncodeWTF8)
+        << U32LEB(BinaryConsts::StringPolicy::UTF8);
+      break;
+    case StringEncodeWTF8:
+      o << U32LEB(BinaryConsts::StringEncodeWTF8)
+        << U32LEB(BinaryConsts::StringPolicy::WTF8);
+      break;
+    case StringEncodeWTF16:
+      o << U32LEB(BinaryConsts::StringEncodeWTF16);
+      break;
+    default:
+      WASM_UNREACHABLE("invalid string.new*");
+  }
+}
+
 void BinaryInstWriter::emitScopeEnd(Expression* curr) {
   assert(!breakStack.empty());
   breakStack.pop_back();
diff --git a/src/wasm/wasm.cpp b/src/wasm/wasm.cpp
index 749a44339..d4337a87d 100644
--- a/src/wasm/wasm.cpp
+++ b/src/wasm/wasm.cpp
@@ -1192,6 +1192,14 @@ void StringMeasure::finalize() {
   }
 }
 
+void StringEncode::finalize() {
+  if (ref->type == Type::unreachable || ptr->type == Type::unreachable) {
+    type = Type::unreachable;
+  } else {
+    type = Type::i32;
+  }
+}
+
 size_t Function::getNumParams() { return getParams().size(); }
 
 size_t Function::getNumVars() { return vars.size(); }