src: add config file support

Author: marco-ippolito

doc/api/cli.md

@@ -911,6 +911,48 @@ added: v23.6.0
 
 Enable experimental import support for `.node` addons.
 
+### `--experimental-config-file`
+
+<!-- YAML
+added: REPLACEME
+-->
+
+> Stability: 1.0 - Early development
+
+Use this flag to specify a configuration file that will be loaded and parsed
+before the application starts.
+Node.js will read the configuration file and apply the settings.
+The configuration file should be a JSON file
+with the following structure:
+
+```json
+{
+  "experimental_transform_types": true,
+  "import": [
+    "amaro/transform"
+  ]
+}
+```
+
+The priority in configuration is as follows:
+
+* NODE\_OPTIONS and command-line options
+* Config file
+* Dotenv NODE\_OPTIONS
+
+If multiple keys are present in the configuration file,
+the last one will override the previous ones.
+Unknown keys will be ignored.
+
+It possible to use the official json schema to validate the configuration file,
+which may vary depending on the Node.js version.
+
+```json
+{
+  "$schema": "https://nodejs.org/dist/REPLACEME/node_config_json_schema.json",
+}
+```
+
 ### `--experimental-eventsource`
 
 <!-- YAML

doc/node.1

@@ -166,6 +166,9 @@ Interpret the entry point as a URL.
 .It Fl -experimental-addon-modules
 Enable experimental addon module support.
 .
+.It Fl -experimental-config-file
+Enable support for experimental config file
+.
 .It Fl -experimental-import-meta-resolve
 Enable experimental ES modules support for import.meta.resolve().
 .

doc/node_config_json_schema.json

@@ -0,0 +1,16 @@
+{
+    "$schema": "https://json-schema.org/draft/2020-12/schema",
+    "type": "object",
+    "properties": {
+        "experimental_transform_types": {
+            "type": "boolean"
+        },
+        "import": {
+            "type": "array",
+            "items": {
+                "type": "string"
+            }
+        }
+    },
+    "additionalProperties": false
+}

lib/internal/process/pre_execution.js

@@ -116,6 +116,8 @@ function prepareExecution(options) {
   initializeSourceMapsHandlers();
   initializeDeprecations();
 
+  setupConfigFile();
+
   require('internal/dns/utils').initializeDns();
 
   if (isMainThread) {
@@ -312,6 +314,12 @@ function setupSQLite() {
   BuiltinModule.allowRequireByUsers('sqlite');
 }
 
+function setupConfigFile() {
+  if (getOptionValue('--experimental-config-file')) {
+    emitExperimentalWarning('--experimental-config-file');
+  }
+}
+
 function setupQuic() {
   if (!getOptionValue('--experimental-quic')) {
     return;

node.gyp

@@ -130,6 +130,7 @@
       'src/node_process_events.cc',
       'src/node_process_methods.cc',
       'src/node_process_object.cc',
+      'src/node_rc.cc',
       'src/node_realm.cc',
       'src/node_report.cc',
       'src/node_report_module.cc',
@@ -262,6 +263,7 @@
       'src/node_platform.h',
       'src/node_process.h',
       'src/node_process-inl.h',
+      'src/node_rc.h',
       'src/node_realm.h',
       'src/node_realm-inl.h',
       'src/node_report.h',

src/node.cc

@@ -21,6 +21,7 @@
 
 #include "node.h"
 #include "node_dotenv.h"
+#include "node_rc.h"
 #include "node_task_runner.h"
 
 // ========== local headers ==========
@@ -150,6 +151,9 @@ namespace per_process {
 // Instance is used to store environment variables including NODE_OPTIONS.
 node::Dotenv dotenv_file = Dotenv();
 
+// node_rc.h
+node::ConfigReader config_reader = ConfigReader();
+
 // node_revert.h
 // Bit flag used to track security reverts.
 unsigned int reverted_cve = 0;
@@ -884,6 +888,23 @@ static ExitCode InitializeNodeWithArgsInternal(
     per_process::dotenv_file.AssignNodeOptionsIfAvailable(&node_options);
   }
 
+  auto result = per_process::config_reader.GetDataFromArgs(*argv);
+  if (result.has_value()) {
+    switch (per_process::config_reader.ParseConfig(result.value())) {
+      case ConfigReader::ParseResult::Valid:
+        break;
+      case ConfigReader::ParseResult::InvalidContent:
+        errors->push_back(result.value() + ": invalid format");
+        break;
+      case ConfigReader::ParseResult::FileError:
+        errors->push_back(result.value() + ": not found");
+        break;
+      default:
+        UNREACHABLE();
+    }
+    per_process::config_reader.AssignNodeOptions(&node_options);
+  }
+
 #if !defined(NODE_WITHOUT_NODE_OPTIONS)
   if (!(flags & ProcessInitializationFlags::kDisableNodeOptionsEnv)) {
     // NODE_OPTIONS environment variable is preferred over the file one.

src/node_options.cc

@@ -671,6 +671,9 @@ EnvironmentOptionsParser::EnvironmentOptionsParser() {
             "set environment variables from supplied file",
             &EnvironmentOptions::optional_env_file);
   Implies("--env-file-if-exists", "[has_env_file_string]");
+  AddOption("--experimental-config-file",
+            "set config file from supplied file",
+            &EnvironmentOptions::experimental_config_file);
   AddOption("--test",
             "launch test runner on startup",
             &EnvironmentOptions::test_runner);

src/node_options.h

@@ -256,6 +256,7 @@ class EnvironmentOptions : public Options {
 
   bool report_exclude_env = false;
   bool report_exclude_network = false;
+  std::string experimental_config_file;
 
   inline DebugOptions* get_debug_options() { return &debug_options_; }
   inline const DebugOptions& debug_options() const { return debug_options_; }

src/node_rc.cc

@@ -0,0 +1,119 @@
+#include "node_rc.h"
+#include "debug_utils-inl.h"
+#include "env-inl.h"
+#include "node_errors.h"
+#include "node_file.h"
+#include "node_internals.h"
+#include "simdjson.h"
+
+#include <functional>
+#include <map>
+#include <string>
+
+namespace node {
+
+std::optional<std::string> ConfigReader::GetDataFromArgs(
+    const std::vector<std::string>& args) {
+  constexpr std::string_view flag = "--experimental-config-file";
+
+  for (auto it = args.begin(); it != args.end(); ++it) {
+    if (*it == flag) {
+      // Case: "--experimental-config-file foo"
+      if (auto next = std::next(it); next != args.end()) {
+        return *next;
+      }
+    } else if (it->starts_with(flag)) {
+      // Case: "--experimental-config-file=foo"
+      if (it->size() > flag.size() && (*it)[flag.size()] == '=') {
+        return it->substr(flag.size() + 1);
+      }
+    }
+  }
+
+  return std::nullopt;
+}
+
+ConfigReader::ParseResult ConfigReader::ParseConfig(
+    const std::string& config_path) {
+  std::string file_content;
+  // Read the configuration file
+  int r = ReadFileSync(&file_content, config_path.c_str());
+  if (r != 0) {
+    const char* err = uv_strerror(r);
+    FPrintF(
+        stderr, "Cannot read configuration from %s: %s\n", config_path, err);
+    return ParseResult::FileError;
+  }
+
+  // Parse the configuration file
+  simdjson::ondemand::parser json_parser;
+  simdjson::ondemand::document document;
+  if (json_parser.iterate(file_content).get(document)) {
+    FPrintF(stderr, "Can't parse %s\n", config_path.c_str());
+    return ParseResult::InvalidContent;
+  }
+
+  simdjson::ondemand::object main_object;
+  // If document is not an object, throw an error.
+  if (auto root_error = document.get_object().get(main_object)) {
+    if (root_error == simdjson::error_code::INCORRECT_TYPE) {
+      FPrintF(stderr,
+              "Root value unexpected not an object for %s\n\n",
+              config_path.c_str());
+    } else {
+      FPrintF(stderr, "Can't parse %s\n", config_path.c_str());
+    }
+    return ParseResult::InvalidContent;
+  }
+
+  ConfigReader::Config config;
+  simdjson::ondemand::value ondemand_value;
+  simdjson::ondemand::raw_json_string key;
+
+  for (auto field : main_object) {
+    if (field.key().get(key) || field.value().get(ondemand_value)) {
+      return ParseResult::InvalidContent;
+    }
+    if (key == "experimental_transform_types") {
+      if (ondemand_value.get_bool().get(config.experimental_transform_types)) {
+        FPrintF(stderr, "Invalid value for experimental_transform_types\n");
+        return ParseResult::InvalidContent;
+      }
+    }
+
+    if (key == "import") {
+      simdjson::ondemand::array raw_imports;
+      if (ondemand_value.get_array().get(raw_imports)) {
+        FPrintF(stderr, "Invalid value for import\n");
+        return ParseResult::InvalidContent;
+      }
+      for (auto raw_import : raw_imports) {
+        std::string_view import;
+        if (raw_import.get_string(import)) {
+          FPrintF(stderr, "Invalid value for import\n");
+          return ParseResult::InvalidContent;
+        }
+        config.import.push_back(std::string(import));
+      }
+    }
+  }
+
+  config_ = config;
+  return ParseResult::Valid;
+}
+
+void ConfigReader::AssignNodeOptions(std::string* node_options) {
+  std::string result = "";
+  if (config_.experimental_transform_types) {
+    result += "--experimental-transform-types";
+  }
+
+  if(config_.import.size() > 0) {
+    for (const auto& import : config_.import) {
+      result += " --import=" + import;
+    }
+  }
+  *node_options = result;
+  return;
+}
+}  // namespace node

src/node_rc.h

@@ -0,0 +1,36 @@
+#ifndef SRC_NODE_RC_H_
+#define SRC_NODE_RC_H_
+
+#if defined(NODE_WANT_INTERNALS) && NODE_WANT_INTERNALS
+
+#include <map>
+#include <string>
+#include <variant>
+#include "simdjson.h"
+#include "util-inl.h"
+
+namespace node {
+
+class ConfigReader {
+ public:
+  enum ParseResult { Valid, FileError, InvalidContent };
+  struct Config {
+    bool experimental_transform_types;
+    std::vector<std::string> import;
+  };
+  ConfigReader::ParseResult ParseConfig(const std::string& config_path);
+
+  std::optional<std::string> GetDataFromArgs(
+      const std::vector<std::string>& args);
+
+  void AssignNodeOptions(std::string* node_options);
+
+ private:
+  ConfigReader::Config config_;
+};
+
+}  // namespace node
+
+#endif  // defined(NODE_WANT_INTERNALS) && NODE_WANT_INTERNALS
+
+#endif  // SRC_NODE_RC_H_