sea: support argv in sea config

implement argv config for sea including snapshot

Author: hotpineapple

src/node_sea.cc

@@ -142,6 +142,18 @@ size_t SeaSerializer::Write(const SeaResource& sea) {
       written_total += WriteStringView(arg, StringLogMode::kAddressAndContent);
     }
   }
+
+  if (static_cast<bool>(sea.flags & SeaFlags::kIncludeUserArgv)) {
+    Debug("Write SEA resource user argv size %zu\n", sea.user_argv.size());
+    written_total += WriteArithmetic<size_t>(sea.user_argv.size());
+    for (const auto& arg : sea.user_argv) {
+      Debug("Write SEA resource user arg %s at %p, size=%zu\n",
+            arg.data(),
+            arg.data(),
+            arg.size());
+      written_total += WriteStringView(arg, StringLogMode::kAddressAndContent);
+    }
+  }
   return written_total;
 }
 
@@ -224,13 +236,29 @@ SeaResource SeaDeserializer::Read() {
       exec_argv.emplace_back(arg);
     }
   }
+
+  std::vector<std::string_view> user_argv;
+  if (static_cast<bool>(flags & SeaFlags::kIncludeUserArgv)) {
+    size_t user_argv_size = ReadArithmetic<size_t>();
+    Debug("Read SEA resource user args size %zu\n", user_argv_size);
+    user_argv.reserve(user_argv_size);
+    for (size_t i = 0; i < user_argv_size; ++i) {
+      std::string_view arg = ReadStringView(StringLogMode::kAddressAndContent);
+      Debug("Read SEA resource user arg %s at %p, size=%zu\n",
+            arg.data(),
+            arg.data(),
+            arg.size());
+      user_argv.emplace_back(arg);
+    }
+  }
   return {flags,
-          exec_argv_extension,
-          code_path,
-          code,
-          code_cache,
-          assets,
-          exec_argv};
+        exec_argv_extension,
+        code_path,
+        code,
+        code_cache,
+        assets,
+        exec_argv,
+        user_argv};
 }
 
 std::string_view FindSingleExecutableBlob() {
@@ -316,12 +344,14 @@ std::tuple<int, char**> FixupArgsForSEA(int argc, char** argv) {
     static std::vector<char*> new_argv;
     static std::vector<std::string> exec_argv_storage;
     static std::vector<std::string> cli_extension_args;
+    static std::vector<std::string> user_argv_storage;
 
     SeaResource sea_resource = FindSingleExecutableResource();
 
     new_argv.clear();
     exec_argv_storage.clear();
     cli_extension_args.clear();
+    user_argv_storage.clear();
 
     // Handle CLI extension mode for --node-options
     if (sea_resource.exec_argv_extension == SeaExecArgvExtension::kCli) {
@@ -341,10 +371,11 @@ std::tuple<int, char**> FixupArgsForSEA(int argc, char** argv) {
       }
     }
 
-    // Reserve space for argv[0], exec argv, cli extension args, original argv,
-    // and nullptr
+    // Reserve space for two argv[0] , exec argv, cli extension args, user argv,
+    // original argv and nullptr
     new_argv.reserve(argc + sea_resource.exec_argv.size() +
-                     cli_extension_args.size() + 2);
+                     cli_extension_args.size() + sea_resource.user_argv.size() +
+                     3);
     new_argv.emplace_back(argv[0]);
 
     // Insert exec argv from SEA config
@@ -363,8 +394,18 @@ std::tuple<int, char**> FixupArgsForSEA(int argc, char** argv) {
       new_argv.emplace_back(exec_argv_storage.back().data());
     }
 
-    // Add actual run time arguments
-    new_argv.insert(new_argv.end(), argv, argv + argc);
+    new_argv.emplace_back(argv[0]);
+    // Insert user argv from SEA config
+    if (!sea_resource.user_argv.empty()) {
+      user_argv_storage.reserve(sea_resource.user_argv.size());
+      for (const auto& arg : sea_resource.user_argv) {
+        user_argv_storage.emplace_back(arg);
+        new_argv.emplace_back(user_argv_storage.back().data());
+      }
+    }
+
+    // Add actual run time arguments.
+    new_argv.insert(new_argv.end(), argv + 1, argv + argc);
     new_argv.emplace_back(nullptr);
     argc = new_argv.size() - 1;
     argv = new_argv.data();
@@ -382,6 +423,7 @@ struct SeaConfig {
   SeaExecArgvExtension exec_argv_extension = SeaExecArgvExtension::kEnv;
   std::unordered_map<std::string, std::string> assets;
   std::vector<std::string> exec_argv;
+  std::vector<std::string> user_argv;
 };
 
 std::optional<SeaConfig> ParseSingleExecutableConfig(
@@ -544,6 +586,35 @@ std::optional<SeaConfig> ParseSingleExecutableConfig(
                 config_path);
         return std::nullopt;
       }
+    } else if (key == "argv") {
+      simdjson::ondemand::array argv_array;
+      FPrintF(stderr, "\"argv\" start \n");
+
+      if (field.value().get_array().get(argv_array)) {
+        FPrintF(stderr,
+                "\"argv\" field of %s is not an array of strings\n",
+                config_path);
+        return std::nullopt;
+      }
+      std::vector<std::string> user_argv;
+
+      for (auto argv : argv_array) {
+        FPrintF(stderr, "\"argv\"\n");
+
+        std::string_view argv_str;
+        if (argv.get_string().get(argv_str)) {
+          FPrintF(stderr,
+                  "\"argv\" field of %s is not an array of strings\n",
+                  config_path);
+          return std::nullopt;
+        }
+        FPrintF(stderr, "\"argv\" : %s \n", argv_str);
+        user_argv.emplace_back(argv_str);
+      }
+      if (!user_argv.empty()) {
+        result.flags |= SeaFlags::kIncludeUserArgv;
+        result.user_argv = std::move(user_argv);
+      }
     }
   }
 
@@ -579,9 +650,11 @@ ExitCode GenerateSnapshotForSEA(const SeaConfig& config,
                                 const SnapshotConfig& snapshot_config,
                                 std::vector<char>* snapshot_blob) {
   SnapshotData snapshot;
-  // TODO(joyeecheung): make the arguments configurable through the JSON
-  // config or a programmatic API.
   std::vector<std::string> patched_args = {args[0], config.main_path};
+  if (!config.user_argv.empty()) {
+    patched_args.insert(
+      patched_args.end(), config.user_argv.begin(),config.user_argv.end());
+  }
   ExitCode exit_code = SnapshotBuilder::Generate(&snapshot,
                                                  patched_args,
                                                  exec_args,
@@ -741,6 +814,10 @@ ExitCode GenerateSingleExecutableBlob(
   for (const auto& arg : config.exec_argv) {
     exec_argv_view.emplace_back(arg);
   }
+  std::vector<std::string_view> user_argv_view;
+  for (const auto& arg : config.user_argv) {
+    user_argv_view.emplace_back(arg);
+  }
   SeaResource sea{
       config.flags,
       config.exec_argv_extension,
@@ -750,7 +827,8 @@ ExitCode GenerateSingleExecutableBlob(
           : std::string_view{main_script.data(), main_script.size()},
       optional_sv_code_cache,
       assets_view,
-      exec_argv_view};
+      exec_argv_view,
+      user_argv_view};
 
   SeaSerializer serializer;
   serializer.Write(sea);

src/node_sea.h

@@ -29,6 +29,7 @@ enum class SeaFlags : uint32_t {
   kUseCodeCache = 1 << 2,
   kIncludeAssets = 1 << 3,
   kIncludeExecArgv = 1 << 4,
+  kIncludeUserArgv = 1 << 5,
 };
 
 enum class SeaExecArgvExtension : uint8_t {
@@ -45,6 +46,7 @@ struct SeaResource {
   std::optional<std::string_view> code_cache;
   std::unordered_map<std::string_view, std::string_view> assets;
   std::vector<std::string_view> exec_argv;
+  std::vector<std::string_view> user_argv;
 
   bool use_snapshot() const;
   bool use_code_cache() const;

test/fixtures/sea-argv.js

@@ -0,0 +1,13 @@
+const assert = require('assert');
+
+process.emitWarning('This warning should not be shown in the output', 'TestWarning');
+
+console.log('process.argv:', JSON.stringify(process.argv));
+
+assert.deepStrictEqual(process.argv.slice(2), [
+  'user-arg1',
+  'user-arg2',
+  'user-arg3'
+]);
+
+console.log('multiple argv test passed');

test/sequential/test-single-executable-application-argv.js

@@ -0,0 +1,65 @@
+'use strict';
+
+require('../common');
+
+const {
+  generateSEA,
+  skipIfSingleExecutableIsNotSupported,
+} = require('../common/sea');
+
+skipIfSingleExecutableIsNotSupported();
+
+// This tests the argv functionality with multiple arguments in single executable applications.
+
+const fixtures = require('../common/fixtures');
+const tmpdir = require('../common/tmpdir');
+const { copyFileSync, writeFileSync, existsSync } = require('fs');
+const { spawnSyncAndAssert, spawnSyncAndExitWithoutError } = require('../common/child_process');
+const { join } = require('path');
+const assert = require('assert');
+
+const configFile = tmpdir.resolve('sea-config.json');
+const seaPrepBlob = tmpdir.resolve('sea-prep.blob');
+const outputFile = tmpdir.resolve(process.platform === 'win32' ? 'sea.exe' : 'sea');
+
+tmpdir.refresh();
+
+// Copy test fixture to working directory
+copyFileSync(fixtures.path('sea-argv.js'), tmpdir.resolve('sea.js'));
+
+writeFileSync(configFile, `
+{
+  "main": "sea.js",
+  "output": "sea-prep.blob",
+  "disableExperimentalSEAWarning": true,
+  "argv": ["user-arg1", "user-arg2"]
+}
+`);
+
+spawnSyncAndExitWithoutError(
+  process.execPath,
+  ['--experimental-sea-config', 'sea-config.json'],
+  { cwd: tmpdir.path });
+
+assert(existsSync(seaPrepBlob));
+
+generateSEA(outputFile, process.execPath, seaPrepBlob);
+
+// Test that multiple argv are properly applied
+spawnSyncAndAssert(
+  outputFile,
+  [
+    'user-arg3',
+  ],
+  {
+    env: {
+      ...process.env,
+      NODE_NO_WARNINGS: '0',
+      COMMON_DIRECTORY: join(__dirname, '..', 'common'),
+      NODE_DEBUG_NATIVE: 'SEA',
+    }
+  },
+  {
+    stdout: /multiple argv test passed/,
+    trim: true,
+  });

test/sequential/test-single-executable-application-snapshot-with-argv.js

@@ -0,0 +1,109 @@
+'use strict';
+
+require('../common');
+
+const {
+  generateSEA,
+  skipIfSingleExecutableIsNotSupported,
+} = require('../common/sea');
+
+skipIfSingleExecutableIsNotSupported();
+
+// This tests the snapshot support in single executable applications.
+
+const tmpdir = require('../common/tmpdir');
+const { writeFileSync, existsSync } = require('fs');
+const {
+  spawnSyncAndAssert,
+  spawnSyncAndExit,
+} = require('../common/child_process');
+const assert = require('assert');
+
+const configFile = tmpdir.resolve('sea-config.json');
+const seaPrepBlob = tmpdir.resolve('sea-prep.blob');
+const outputFile = tmpdir.resolve(process.platform === 'win32' ? 'sea.exe' : 'sea');
+
+{
+  tmpdir.refresh();
+
+  writeFileSync(tmpdir.resolve('snapshot.js'), '', 'utf-8');
+  writeFileSync(configFile, `
+  {
+    "main": "snapshot.js",
+    "output": "sea-prep.blob",
+    "useSnapshot": true
+  }
+  `);
+
+  spawnSyncAndExit(
+    process.execPath,
+    ['--experimental-sea-config', 'sea-config.json'],
+    {
+      cwd: tmpdir.path
+    },
+    {
+      status: 1,
+      signal: null,
+      stderr: /snapshot\.js does not invoke v8\.startupSnapshot\.setDeserializeMainFunction\(\)/
+    });
+}
+
+{
+  tmpdir.refresh();
+  const code = `
+  const {
+    setDeserializeMainFunction,
+  } = require('v8').startupSnapshot;
+  
+  setDeserializeMainFunction(() => {
+    console.log('Hello from snapshot with config argv:', process.argv[2]);
+  });
+  `;
+
+  writeFileSync(tmpdir.resolve('snapshot.js'), code, 'utf-8');
+  writeFileSync(configFile, `
+  {
+    "main": "snapshot.js",
+    "output": "sea-prep.blob",
+    "useSnapshot": true,
+    "argv": ["user-arg"]
+  }
+  `);
+
+  spawnSyncAndAssert(
+    process.execPath,
+    ['--experimental-sea-config', 'sea-config.json'],
+    {
+      cwd: tmpdir.path,
+      env: {
+        NODE_DEBUG_NATIVE: 'SEA',
+        ...process.env,
+      },
+    },
+    {
+      stderr: /Single executable application is an experimental feature/
+    });
+
+  assert(existsSync(seaPrepBlob));
+
+  generateSEA(outputFile, process.execPath, seaPrepBlob);
+
+  spawnSyncAndAssert(
+    outputFile,
+    {
+      env: {
+        NODE_DEBUG_NATIVE: 'SEA,MKSNAPSHOT',
+        ...process.env,
+      }
+    },
+    {
+      trim: true,
+      stdout: 'Hello from snapshot with config argv: user-arg',
+      stderr(output) {
+        assert.doesNotMatch(
+          output,
+          /Single executable application is an experimental feature/);
+      }
+    }
+  );
+}