build: speed up compilation of mksnapshot output

Incremental compilation of Node.js is slow. Currently on a powerful
Linux machine, it takes about 5.8 seconds to compile
`gen/node_snapshot.cc` with g++.

As in the previous PR which dealt with `node_js2c`, we add a new build
define `NODE_MKSNAPSHOT_USE_STRING_LITERALS` which is used by
`node_mksnapshot`. When this flag is set, we emit string literals
instead of array literals for the snapshot blob and for the code cache,
i.e.:

```c++
// old: static const uint8_t X[] = { ... };
static const uint8_t *X = "...";
```

I only enabled the new flag on Linux/macOS, since those are systems that
I have available for testing. On my Linux system with gcc, it speeds up
compilation of this file by 3.7s (5.8s -> 2.1s). On my Mac system with
clang, it speeds up compilation by 1.7s (3.4s -> 1.7s).

Again, the right thing here is probably to generate separate files for
the snapshot blob and for each code cache output, but this is a nice
intermediate speedup.

Refs: #47984
Refs: #48160
PR-URL: #48162
Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com>
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>

Author: kvakil

node.gyp

@@ -872,6 +872,9 @@
           'node_target_type=="executable"', {
           'defines': [ 'NODE_ENABLE_LARGE_CODE_PAGES=1' ],
         }],
+        ['OS in "linux mac"', {
+          'defines': [ 'NODE_MKSNAPSHOT_USE_STRING_LITERALS' ],
+        }],
         [ 'use_openssl_def==1', {
           # TODO(bnoordhuis) Make all platforms export the same list of symbols.
           # Teach mkssldef.py to generate linker maps that UNIX linkers understand.

src/node_snapshotable.cc

@@ -742,18 +742,56 @@ static std::string FormatSize(size_t size) {
   return buf;
 }
 
-static void WriteStaticCodeCacheData(std::ostream* ss,
-                                     const builtins::CodeCacheInfo& info) {
+#ifdef NODE_MKSNAPSHOT_USE_STRING_LITERALS
+static void WriteDataAsCharString(std::ostream* ss,
+                                  const uint8_t* data,
+                                  size_t length) {
+  for (size_t i = 0; i < length; i++) {
+    const uint8_t ch = data[i];
+    // We can print most printable characters directly. The exceptions are '\'
+    // (escape characters), " (would end the string), and ? (trigraphs). The
+    // latter may be overly conservative: we compile with C++17 which doesn't
+    // support trigraphs.
+    if (ch >= ' ' && ch <= '~' && ch != '\\' && ch != '"' && ch != '?') {
+      *ss << ch;
+    } else {
+      // All other characters are blindly output as octal.
+      const char c0 = '0' + ((ch >> 6) & 7);
+      const char c1 = '0' + ((ch >> 3) & 7);
+      const char c2 = '0' + (ch & 7);
+      *ss << "\\" << c0 << c1 << c2;
+    }
+    if (i % 64 == 63) {
+      // Go to a newline every 64 bytes since many text editors have
+      // problems with very long lines.
+      *ss << "\"\n\"";
+    }
+  }
+}
+
+static void WriteStaticCodeCacheDataAsStringLiteral(
+    std::ostream* ss, const builtins::CodeCacheInfo& info) {
+  *ss << "static const uint8_t *" << GetCodeCacheDefName(info.id)
+      << "= reinterpret_cast<const uint8_t *>(\"";
+  WriteDataAsCharString(ss, info.data.data, info.data.length);
+  *ss << "\");\n";
+}
+#else
+static void WriteStaticCodeCacheDataAsArray(
+    std::ostream* ss, const builtins::CodeCacheInfo& info) {
   *ss << "static const uint8_t " << GetCodeCacheDefName(info.id) << "[] = {\n";
   WriteVector(ss, info.data.data, info.data.length);
-  *ss << "};";
+  *ss << "};\n";
 }
+#endif
 
-static void WriteCodeCacheInitializer(std::ostream* ss, const std::string& id) {
+static void WriteCodeCacheInitializer(std::ostream* ss,
+                                      const std::string& id,
+                                      size_t size) {
   std::string def_name = GetCodeCacheDefName(id);
   *ss << "    { \"" << id << "\",\n";
   *ss << "      {" << def_name << ",\n";
-  *ss << "       arraysize(" << def_name << "),\n";
+  *ss << "       " << size << ",\n";
   *ss << "      }\n";
   *ss << "    },\n";
 }
@@ -767,21 +805,34 @@ void FormatBlob(std::ostream& ss, const SnapshotData* data) {
 // This file is generated by tools/snapshot. Do not edit.
 
 namespace node {
-
-static const char v8_snapshot_blob_data[] = {
 )";
+
+#ifdef NODE_MKSNAPSHOT_USE_STRING_LITERALS
+  ss << R"(static const char *v8_snapshot_blob_data = ")";
+  WriteDataAsCharString(
+      &ss,
+      reinterpret_cast<const uint8_t*>(data->v8_snapshot_blob_data.data),
+      data->v8_snapshot_blob_data.raw_size);
+  ss << R"(";)";
+#else
+  ss << R"(static const char v8_snapshot_blob_data[] = {)";
   WriteVector(&ss,
               data->v8_snapshot_blob_data.data,
               data->v8_snapshot_blob_data.raw_size);
-  ss << R"(};
+  ss << R"(};)";
+#endif
 
-static const int v8_snapshot_blob_size = )"
+  ss << R"(static const int v8_snapshot_blob_size = )"
      << data->v8_snapshot_blob_data.raw_size << ";";
 
-  // Windows can't deal with too many large vector initializers.
-  // Store the data into static arrays first.
   for (const auto& item : data->code_cache) {
-    WriteStaticCodeCacheData(&ss, item);
+#ifdef NODE_MKSNAPSHOT_USE_STRING_LITERALS
+    WriteStaticCodeCacheDataAsStringLiteral(&ss, item);
+#else
+    // Windows can't deal with too many large vector initializers.
+    // Store the data into static arrays first.
+    WriteStaticCodeCacheDataAsArray(&ss, item);
+#endif
   }
 
   ss << R"(const SnapshotData snapshot_data {
@@ -808,7 +859,7 @@ static const int v8_snapshot_blob_size = )"
   // -- code_cache begins --
   {)";
   for (const auto& item : data->code_cache) {
-    WriteCodeCacheInitializer(&ss, item.id);
+    WriteCodeCacheInitializer(&ss, item.id, item.data.length);
   }
   ss << R"(
   }