From fa9463e1d8506b0789bd4242536a059ed3210638 Mon Sep 17 00:00:00 2001
From: Yagiz Nizipli <yagiz@nizipli.com>
Date: Sat, 18 Nov 2023 15:37:57 -0500
Subject: [PATCH] src: add `--config-file=path` support to Node.js

---
 doc/api/cli.md              | 12 ++++++
 doc/node-config-schema.json | 14 ++++++
 node.gyp                    |  2 +
 src/node.cc                 | 19 ++++++--
 src/node_config_file.cc     | 86 +++++++++++++++++++++++++++++++++++++
 src/node_config_file.h      | 34 +++++++++++++++
 src/node_options.cc         |  4 ++
 src/node_options.h          |  2 +
 8 files changed, 169 insertions(+), 4 deletions(-)
 create mode 100644 doc/node-config-schema.json
 create mode 100644 src/node_config_file.cc
 create mode 100644 src/node_config_file.h

diff --git a/doc/api/cli.md b/doc/api/cli.md
index 6b1860ecc81980..19b3a4f36f89ab 100644
--- a/doc/api/cli.md
+++ b/doc/api/cli.md
@@ -374,6 +374,18 @@ For example, to run a module with "development" resolutions:
 node -C development app.js
 ```
 
+### `--config-file=path`
+
+> Stability: 1.1 - Active development
+
+<!-- YAML
+added: REPLACEME
+-->
+
+Write a documentation for here.
+- Mention the order of preferrence for environment variables, config file, and command line arguments.
+- Mention the JSON schema.
+
 ### `--cpu-prof`
 
 <!-- YAML
diff --git a/doc/node-config-schema.json b/doc/node-config-schema.json
new file mode 100644
index 00000000000000..b554f5b0e4e52d
--- /dev/null
+++ b/doc/node-config-schema.json
@@ -0,0 +1,14 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "object",
+  "properties": {
+    "options": {
+      "type": "object",
+      "properties": {
+        "inspect": {
+          "type": "string"
+        }
+      }
+    }
+  }
+}
diff --git a/node.gyp b/node.gyp
index 811d15b0df9ad3..210ff3182aa677 100644
--- a/node.gyp
+++ b/node.gyp
@@ -97,6 +97,7 @@
       'src/node_buffer.cc',
       'src/node_builtins.cc',
       'src/node_config.cc',
+      'src/node_config_file.cc',
       'src/node_constants.cc',
       'src/node_contextify.cc',
       'src/node_credentials.cc',
@@ -213,6 +214,7 @@
       'src/node_blob.h',
       'src/node_buffer.h',
       'src/node_builtins.h',
+      'src/node_config_file.h',
       'src/node_constants.h',
       'src/node_context_data.h',
       'src/node_contextify.h',
diff --git a/src/node.cc b/src/node.cc
index 524f80ee69ee52..d93156d313379e 100644
--- a/src/node.cc
+++ b/src/node.cc
@@ -20,6 +20,7 @@
 // USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "node.h"
+#include "node_config_file.h"
 #include "node_dotenv.h"
 
 // ========== local headers ==========
@@ -30,6 +31,7 @@
 #include "memory_tracker-inl.h"
 #include "node_binding.h"
 #include "node_builtins.h"
+#include "node_config_file.h"
 #include "node_errors.h"
 #include "node_internals.h"
 #include "node_main_instance.h"
@@ -145,6 +147,10 @@ namespace per_process {
 // Instance is used to store environment variables including NODE_OPTIONS.
 node::Dotenv dotenv_file = Dotenv();
 
+// node_config_file.h
+// Instance is used to store configuration variables.
+node::ConfigFile config_file = ConfigFile();
+
 // node_revert.h
 // Bit flag used to track security reverts.
 unsigned int reverted_cve = 0;
@@ -858,13 +864,18 @@ static ExitCode InitializeNodeWithArgsInternal(
   HandleEnvOptions(per_process::cli_options->per_isolate->per_env);
 
   std::string node_options;
-  auto file_paths = node::Dotenv::GetPathFromArgs(*argv);
-
-  if (!file_paths.empty()) {
+  // TODO(@anonrig): Prefer what to choose.
+  //   - Environment variable
+  //   - Config file
+  //   - Dotenv file
+  auto config_file_path = node::ConfigFile::GetPathFromArgs(*argv);
+  auto dotenv_file_paths = node::Dotenv::GetPathFromArgs(*argv);
+
+  if (!dotenv_file_paths.empty()) {
     CHECK(!per_process::v8_initialized);
     auto cwd = Environment::GetCwd(Environment::GetExecPath(*argv));
 
-    for (const auto& file_path : file_paths) {
+    for (const auto& file_path : dotenv_file_paths) {
       std::string path = cwd + kPathSeparator + file_path;
       auto path_exists = per_process::dotenv_file.ParsePath(path);
 
diff --git a/src/node_config_file.cc b/src/node_config_file.cc
new file mode 100644
index 00000000000000..9be3a75182a0e8
--- /dev/null
+++ b/src/node_config_file.cc
@@ -0,0 +1,86 @@
+#include "node_config_file.h"
+#include "env-inl.h"
+#include "simdjson.h"
+#include "util.h"
+#include "uv.h"
+
+namespace node {
+
+using v8::NewStringType;
+using v8::String;
+
+std::optional<std::string> ConfigFile::GetPathFromArgs(
+     const std::vector<std::string>& args) {
+   std::string_view flag = "--config-file";
+   // Match the last `--config-file`
+   // This is required to imitate the default behavior of Node.js CLI argument
+   // matching.
+   auto path =
+       std::find_if(args.rbegin(), args.rend(), [&flag](const std::string& arg) {
+         return strncmp(arg.c_str(), flag.data(), flag.size()) == 0;
+       });
+
+   if (path == args.rend()) {
+     return std::nullopt;
+   }
+
+   auto equal_char = path->find('=');
+
+   if (equal_char != std::string::npos) {
+     return path->substr(equal_char + 1);
+   }
+
+   auto next_arg = std::prev(path);
+
+   if (next_arg == args.rend()) {
+     return std::nullopt;
+   }
+
+   return *next_arg;
+ }
+
+bool ConfigFile::ParsePath(const std::string_view path) {
+  std::string raw_file;
+  if (ReadFileSync(&raw_file, path.data()) < 0) {
+    // Loading file failed.
+    // TODO(@anonrig): Throw appropiate error.
+    return false;
+  }
+
+  simdjson::ondemand::parser parser;
+  simdjson::ondemand::document document;
+  simdjson::ondemand::object main_object;
+  auto error = parser.iterate(raw_file).get(main_object);
+
+  if (error || document.get_object().get(main_object)) {
+    return false;
+  }
+
+  simdjson::ondemand::value ondemand_value;
+  simdjson::ondemand::raw_json_string raw_json_string;
+  std::string_view key;
+  std::string_view value;
+
+  // Config file schema is kept up to date on doc/node-config-schema.json
+  for (auto field : main_object) {
+    if (field.key().get(raw_json_string) || field.value().get(ondemand_value)) {
+      return false;
+    }
+
+    if (key == "inspect") {
+      // Implement the rest.
+    }
+  }
+
+  return true;
+}
+
+void ConfigFile::AssignNodeOptionsIfAvailable(std::string* node_options) {
+  // TODO(@anonrig): Implement this.
+}
+
+void ConfigFile::SetEnvironment(Environment* env) {
+  // TODO(@anonrig): Implement this.
+}
+
+}  // namespace node
diff --git a/src/node_config_file.h b/src/node_config_file.h
new file mode 100644
index 00000000000000..52ac5a99042c1a
--- /dev/null
+++ b/src/node_config_file.h
@@ -0,0 +1,34 @@
+#ifndef SRC_NODE_CONFIG_FILE_H_
+#define SRC_NODE_CONFIG_FILE_H_
+
+#if defined(NODE_WANT_INTERNALS) && NODE_WANT_INTERNALS
+
+#include "util-inl.h"
+
+#include <optional>
+#include <map>
+
+namespace node {
+
+class ConfigFile {
+ public:
+  ConfigFile() = default;
+  ConfigFile(const ConfigFile& d) = default;
+  ConfigFile(ConfigFile&& d) noexcept = default;
+  ConfigFile& operator=(ConfigFile&& d) noexcept = default;
+  ConfigFile& operator=(const ConfigFile& d) = default;
+  ~ConfigFile() = default;
+
+  bool ParsePath(const std::string_view path);
+  void AssignNodeOptionsIfAvailable(std::string* node_options);
+  void SetEnvironment(Environment* env);
+
+  static std::optional<std::string> GetPathFromArgs(
+       const std::vector<std::string>& args);
+};
+
+}  // namespace node
+
+#endif  // defined(NODE_WANT_INTERNALS) && NODE_WANT_INTERNALS
+
+#endif  // SRC_NODE_CONFIG_FILE_H_
diff --git a/src/node_options.cc b/src/node_options.cc
index 417d4679e84ec3..b81e0532b75782 100644
--- a/src/node_options.cc
+++ b/src/node_options.cc
@@ -342,6 +342,10 @@ DebugOptionsParser::DebugOptionsParser() {
 }
 
 EnvironmentOptionsParser::EnvironmentOptionsParser() {
+  AddOption("--config-file",
+            "set node.js configuration from supplied file",
+            &EnvironmentOptions::env_file);
+  Implies("--config-file", "[has_config_file_string]");
   AddOption("--conditions",
             "additional user conditions for conditional exports and imports",
             &EnvironmentOptions::conditions,
diff --git a/src/node_options.h b/src/node_options.h
index 04c8a9e438dcfb..92ad12135e6ddb 100644
--- a/src/node_options.h
+++ b/src/node_options.h
@@ -125,6 +125,8 @@ class EnvironmentOptions : public Options {
   std::vector<std::string> allow_fs_write;
   bool allow_child_process = false;
   bool allow_worker_threads = false;
+  std::string config_file;
+  bool has_config_file_string = false;
   bool experimental_repl_await = true;
   bool experimental_vm_modules = false;
   bool expose_internals = false;