Fix escaping of names in IR

src/ir/names.cpp

@@ -70,59 +70,4 @@ std::string MinifiedNameGenerator::getName() {
   return name;
 }
 
-static bool isIdChar(char ch) {
-  return (ch >= '0' && ch <= '9') || (ch >= 'A' && ch <= 'Z') ||
-         (ch >= 'a' && ch <= 'z') || ch == '!' || ch == '#' || ch == '$' ||
-         ch == '%' || ch == '&' || ch == '\'' || ch == '*' || ch == '+' ||
-         ch == '-' || ch == '.' || ch == '/' || ch == ':' || ch == '<' ||
-         ch == '=' || ch == '>' || ch == '?' || ch == '@' || ch == '^' ||
-         ch == '_' || ch == '`' || ch == '|' || ch == '~';
-}
-
-static char formatNibble(int nibble) {
-  return nibble < 10 ? '0' + nibble : 'a' - 10 + nibble;
-}
-
-Name escape(Name name) {
-  bool allIdChars = true;
-  for (char c : name.view()) {
-    if (!(allIdChars = isIdChar(c))) {
-      break;
-    }
-  }
-  if (allIdChars) {
-    return name;
-  }
-  // encode name, if at least one non-idchar (per WebAssembly spec) was found
-  std::string escaped;
-  for (char c : name.view()) {
-    if (isIdChar(c)) {
-      escaped.push_back(c);
-      continue;
-    }
-    // replace non-idchar with `\xx` escape
-    escaped.push_back('\\');
-    escaped.push_back(formatNibble((unsigned char)c >> 4));
-    escaped.push_back(formatNibble((unsigned char)c & 15));
-  }
-  return escaped;
-}
-
-std::string unescape(Name name) {
-  std::string output;
-  std::string_view input = name.view();
-  for (size_t i = 0; i < input.length(); i++) {
-    if ((input[i] == '\\') && (i + 2 < input.length()) &&
-        isxdigit(input[i + 1]) && isxdigit(input[i + 2])) {
-      std::string byte = std::string(input.substr(i + 1, 2));
-      i += 2;
-      char chr = (char)(int)strtol(byte.c_str(), nullptr, 16);
-      output.push_back(chr);
-    } else {
-      output.push_back(input[i]);
-    }
-  }
-  return output;
-}
-
 } // namespace wasm::Names

src/ir/names.h

@@ -138,13 +138,6 @@ class MinifiedNameGenerator {
   std::string getName();
 };
 
-// Escapes a string into a valid WebAssembly identifier by converting invalid
-// characters to \xx hex sequences.
-Name escape(Name name);
-// Unescapes a WebAssembly identifier back into its original human-readable
-// string.
-std::string unescape(Name name);
-
 } // namespace wasm::Names
 
 #endif // wasm_ir_names_h

src/passes/Asyncify.cpp

@@ -475,32 +475,29 @@ class PatternMatcher {
   std::set<Name> names;
   std::set<std::string> patterns;
   std::set<std::string> patternsMatched;
-  std::map<std::string, std::string> unescaped;
 
   PatternMatcher(std::string designation,
                  Module& module,
                  const String::Split& list)
     : designation(designation) {
     // The lists contain human-readable strings. Turn them into the
-    // internal escaped names for later comparisons
+    // internal names for later comparisons.
     for (auto& name : list) {
-      auto escaped = Names::escape(name);
-      unescaped[escaped.toString()] = name;
       if (name.find('*') != std::string::npos) {
-        patterns.insert(escaped.toString());
+        patterns.insert(name);
       } else {
-        auto* func = module.getFunctionOrNull(escaped);
+        auto* func = module.getFunctionOrNull(name);
         if (!func) {
           std::cerr << "warning: Asyncify " << designation
-                    << "list contained a non-existing function name: " << name
-                    << " (" << escaped << ")\n";
+                    << "list contained a non-existing function name: '" << name
+                    << "'\n";
         } else if (func->imported()) {
           Fatal() << "Asyncify " << designation
                   << "list contained an imported function name (use the import "
                      "list for imports): "
                   << name << '\n';
         }
-        names.insert(escaped.str);
+        names.insert(name);
       }
     }
   }
@@ -523,8 +520,8 @@ class PatternMatcher {
     for (auto& pattern : patterns) {
       if (!patternsMatched.contains(pattern)) {
         std::cerr << "warning: Asyncify " << designation
-                  << "list contained a non-matching pattern: "
-                  << unescaped[pattern] << " (" << pattern << ")\n";
+                  << "list contained a non-matching pattern: '" << pattern
+                  << "'\n";
       }
     }
   }

src/tools/wasm-split/split-options.cpp

@@ -365,12 +365,6 @@ WasmSplitOptions::WasmSplitOptions()
          {Mode::Instrument, Mode::MergeProfiles, Mode::MultiSplit},
          Options::Arguments::One,
          [&](Options* o, const std::string& argument) { output = argument; })
-    .add("--unescape",
-         "-u",
-         "Un-escape function names (in print-profile output)",
-         WasmSplitOption,
-         Options::Arguments::Zero,
-         [&](Options* o, const std::string& argument) { unescape = true; })
     .add("--verbose",
          "-v",
          "Verbose output mode. Prints the functions that will be kept "

src/tools/wasm-split/split-options.h

@@ -45,7 +45,6 @@ struct WasmSplitOptions : ToolOptions {
   StorageKind storageKind = StorageKind::InGlobals;
 
   bool usePlaceholders = true;
-  bool unescape = false;
   bool verbose = false;
   bool emitBinary = true;
   bool symbolMap = false;

src/tools/wasm-split/wasm-split.cpp

@@ -431,7 +431,7 @@ void multiSplitModule(const WasmSplitOptions& options) {
       }
       continue;
     }
-    Name name = Names::escape(line);
+    Name name = line;
     if (newSection) {
       if (name.endsWith(":")) {
         name = name.substr(0, name.size() - 1);
@@ -586,9 +586,7 @@ void printReadableProfile(const WasmSplitOptions& options) {
 
   auto printFnSet = [&](auto funcs, std::string prefix) {
     for (auto it = funcs.begin(); it != funcs.end(); ++it) {
-      std::cout << prefix << " "
-                << (options.unescape ? Names::unescape(*it) : it->toString())
-                << std::endl;
+      std::cout << prefix << " " << it->toString() << std::endl;
     }
   };
 

src/wasm-binary.h

@@ -1489,7 +1489,6 @@ class WasmBinaryWriter {
 
   // helpers
   void writeInlineString(std::string_view name);
-  void writeEscapedName(std::string_view name);
   void writeInlineBuffer(const char* data, size_t size);
   void writeData(const char* data, size_t size);
 

src/wasm/wasm-binary.cpp

@@ -917,7 +917,7 @@ void WasmBinaryWriter::writeNames() {
   if (emitModuleName && wasm->name.is()) {
     auto substart =
       startSubsection(BinaryConsts::CustomSections::Subsection::NameModule);
-    writeEscapedName(wasm->name.view());
+    writeInlineString(wasm->name.view());
     finishSubsection(substart);
   }
 
@@ -946,7 +946,7 @@ void WasmBinaryWriter::writeNames() {
       o << U32LEB(functionsWithNames.size());
       for (auto& [index, global] : functionsWithNames) {
         o << U32LEB(index);
-        writeEscapedName(global->name.view());
+        writeInlineString(global->name.view());
       }
       finishSubsection(substart);
     }
@@ -1006,7 +1006,7 @@ void WasmBinaryWriter::writeNames() {
         o << U32LEB(localsWithNames.size());
         for (auto& [indexInBinary, name] : localsWithNames) {
           o << U32LEB(indexInBinary);
-          writeEscapedName(name.view());
+          writeInlineString(name.view());
         }
         emitted++;
       }
@@ -1029,7 +1029,7 @@ void WasmBinaryWriter::writeNames() {
       o << U32LEB(namedTypes.size());
       for (auto type : namedTypes) {
         o << U32LEB(indexedTypes.indices[type]);
-        writeEscapedName(wasm->typeNames[type].name.view());
+        writeInlineString(wasm->typeNames[type].name.view());
       }
       finishSubsection(substart);
     }
@@ -1056,7 +1056,7 @@ void WasmBinaryWriter::writeNames() {
 
       for (auto& [index, table] : tablesWithNames) {
         o << U32LEB(index);
-        writeEscapedName(table->name.view());
+        writeInlineString(table->name.view());
       }
 
       finishSubsection(substart);
@@ -1082,7 +1082,7 @@ void WasmBinaryWriter::writeNames() {
       o << U32LEB(memoriesWithNames.size());
       for (auto& [index, memory] : memoriesWithNames) {
         o << U32LEB(index);
-        writeEscapedName(memory->name.view());
+        writeInlineString(memory->name.view());
       }
       finishSubsection(substart);
     }
@@ -1107,7 +1107,7 @@ void WasmBinaryWriter::writeNames() {
       o << U32LEB(globalsWithNames.size());
       for (auto& [index, global] : globalsWithNames) {
         o << U32LEB(index);
-        writeEscapedName(global->name.view());
+        writeInlineString(global->name.view());
       }
       finishSubsection(substart);
     }
@@ -1132,7 +1132,7 @@ void WasmBinaryWriter::writeNames() {
 
       for (auto& [index, elem] : elemsWithNames) {
         o << U32LEB(index);
-        writeEscapedName(elem->name.view());
+        writeInlineString(elem->name.view());
       }
 
       finishSubsection(substart);
@@ -1156,7 +1156,7 @@ void WasmBinaryWriter::writeNames() {
         auto& seg = wasm->dataSegments[i];
         if (seg->hasExplicitName) {
           o << U32LEB(i);
-          writeEscapedName(seg->name.view());
+          writeInlineString(seg->name.view());
         }
       }
       finishSubsection(substart);
@@ -1187,7 +1187,7 @@ void WasmBinaryWriter::writeNames() {
         o << U32LEB(fieldNames.size());
         for (auto& [index, name] : fieldNames) {
           o << U32LEB(index);
-          writeEscapedName(name.view());
+          writeInlineString(name.view());
         }
       }
       finishSubsection(substart);
@@ -1213,7 +1213,7 @@ void WasmBinaryWriter::writeNames() {
       o << U32LEB(tagsWithNames.size());
       for (auto& [index, tag] : tagsWithNames) {
         o << U32LEB(index);
-        writeEscapedName(tag->name.view());
+        writeInlineString(tag->name.view());
       }
       finishSubsection(substart);
     }
@@ -1846,37 +1846,6 @@ void WasmBinaryWriter::writeInlineString(std::string_view name) {
   o.writeInlineString(name);
 }
 
-static bool isHexDigit(char ch) {
-  return (ch >= '0' && ch <= '9') || (ch >= 'a' && ch <= 'f') ||
-         (ch >= 'A' && ch <= 'F');
-}
-
-static int decodeHexNibble(char ch) {
-  return ch <= '9' ? ch & 15 : (ch & 15) + 9;
-}
-
-void WasmBinaryWriter::writeEscapedName(std::string_view name) {
-  if (name.find('\\') == std::string_view::npos) {
-    writeInlineString(name);
-    return;
-  }
-  // decode escaped by escapeName (see below) function names
-  std::string unescaped;
-  for (size_t i = 0; i < name.size();) {
-    char ch = name[i++];
-    // support only `\xx` escapes; ignore invalid or unsupported escapes
-    if (ch != '\\' || i + 1 >= name.size() || !isHexDigit(name[i]) ||
-        !isHexDigit(name[i + 1])) {
-      unescaped.push_back(ch);
-      continue;
-    }
-    unescaped.push_back(
-      char((decodeHexNibble(name[i]) << 4) | decodeHexNibble(name[i + 1])));
-    i += 2;
-  }
-  writeInlineString({unescaped.data(), unescaped.size()});
-}
-
 void WasmBinaryWriter::writeInlineBuffer(const char* data, size_t size) {
   o << U32LEB(size);
   writeData(data, size);
@@ -5310,21 +5279,19 @@ void WasmBinaryReader::readTags() {
 namespace {
 
 // Performs necessary processing of names from the name section before using
-// them. Specifically it escapes and deduplicates them.
+// them. Specifically it deduplicates them.
 class NameProcessor {
 public:
-  // Returns a unique, escaped name. Notes that name for the items to follow to
+  // Returns a unique name. Notes that name for the items to follow to
   // keep them unique as well.
-  Name process(Name name) { return deduplicate(Names::escape(name)); }
-
-private:
-  std::unordered_set<Name> usedNames;
-
-  Name deduplicate(Name base) {
-    auto name = Names::getValidNameGivenExisting(base, usedNames);
+  Name process(Name name) {
+    name = Names::getValidNameGivenExisting(name, usedNames);
     usedNames.insert(name);
     return name;
   }
+
+private:
+  std::unordered_set<Name> usedNames;
 };
 
 } // anonymous namespace

test/binaryen.js/debug-names.js.txt

@@ -26,7 +26,7 @@
  (memory $lo 0 0)
  (table $wor 0 0 funcref)
  (func $of (param $js i32)
-  (local $!#$%&'*+-./:<=>?@\5c^_`|~ f64)
+  (local $!#$%&'*+-./:<=>?@\^_`|~ f64)
  )
 )
 
@@ -37,7 +37,7 @@
  (memory $lo 0 0)
  (table $wor 0 0 funcref)
  (func $of (param $js i32)
-  (local $!#$%&'*+-./:<=>?@\5c^_`|~ f64)
+  (local $!#$%&'*+-./:<=>?@\^_`|~ f64)
  )
 )
 

test/complexBinaryNames.wasm.fromBinary

@@ -1,11 +1,11 @@
 (module
  (type $0 (func))
  (export "$zoo (.bar)" (func $1))
- (func $foo\20\28.bar\29
+ (func $"foo (.bar)"
   (nop)
  )
  (func $1
-  (call $foo\20\28.bar\29)
+  (call $"foo (.bar)")
  )
 )
 

test/finalize/shared_add_to_table.wasm.out

@@ -12,13 +12,13 @@
  (import "GOT.func" "_Z14waka_func_minei" (global $gimport$4 (mut i32)))
  (import "GOT.mem" "waka_mine" (global $gimport$5 (mut i32)))
  (import "GOT.mem" "waka_others" (global $gimport$6 (mut i32)))
- (import "env" "_Z16waka_func_theirsi" (func $waka_func_theirs\28int\29 (param i32) (result i32)))
+ (import "env" "_Z16waka_func_theirsi" (func $"waka_func_theirs(int)" (param i32) (result i32)))
  (global $global$0 i32 (i32.const 0))
  (global $global$1 i32 (i32.const 0))
  (data $0 (global.get $gimport$1) "*\00\00\00")
  (export "__wasm_call_ctors" (func $__wasm_call_ctors))
  (export "__wasm_apply_relocs" (func $__wasm_apply_relocs))
- (export "_Z14waka_func_minei" (func $waka_func_mine\28int\29))
+ (export "_Z14waka_func_minei" (func $"waka_func_mine(int)"))
  (export "__original_main" (func $__original_main))
  (export "waka_mine" (global $global$0))
  (export "main" (func $main))
@@ -28,7 +28,7 @@
  )
  (func $__wasm_apply_relocs
  )
- (func $waka_func_mine\28int\29 (param $0 i32) (result i32)
+ (func $"waka_func_mine(int)" (param $0 i32) (result i32)
   (i32.add
    (local.get $0)
    (i32.const 1)