decompiler: Cleanup duplication in extractor/decompiler and make it easier to enable streamed audio ripping from CLI (#3560)

This centralizes the code that both `extractor` and the decompiler
executes. In the past this code was partially-duplicated, meaning that
the `extractor` could only do _some_ operations and not others (ie.
could not extract the audio files).

I also simplified the process to enable audio streaming in the
configuration. This is to support a new feature in the launcher that
allows you to enable these options for the decompiler:


![image](https://github.com/open-goal/jak-project/assets/13153231/8e6c20a1-8b5b-46f0-bceb-7644f713989f)

common/util/os.cpp

@@ -3,20 +3,29 @@
 #include "common/common_types.h"
 #include "common/log/log.h"
 
-#ifdef __linux__
-
+#ifdef _WIN32
+// clang-format off
+#define NOMINMAX
+#define WIN32_LEAN_AND_MEAN
+#include <Windows.h>
+#include <psapi.h>
+// clang-format on
+size_t get_peak_rss() {
+  HANDLE hProcess = GetCurrentProcess();
+  PROCESS_MEMORY_COUNTERS pmc;
+  if (GetProcessMemoryInfo(hProcess, &pmc, sizeof(pmc))) {
+    return pmc.PeakWorkingSetSize;
+  } else {
+    return 0;
+  }
+}
+#else
 #include <sys/resource.h>
-
 size_t get_peak_rss() {
   rusage x;
   getrusage(RUSAGE_SELF, &x);
   return x.ru_maxrss * 1024;
 }
-
-#else
-size_t get_peak_rss() {
-  return 0;
-}
 #endif
 
 #ifdef _WIN32

common/util/os.h

@@ -3,7 +3,6 @@
 #include <cstddef>
 #include <string>
 
-// Note: these are not implemented on windows and will return zero.
 size_t get_peak_rss();
 void setup_cpu_info();
 

decompiler/CMakeLists.txt

@@ -89,7 +89,8 @@ add_library(
         VuDisasm/VuDisassembler.cpp
         VuDisasm/VuInstruction.cpp
 
-        config.cpp)
+        config.cpp
+        decompilation_process.cpp)
 
 target_link_libraries(decomp
         lzokay

decompiler/config.cpp

@@ -64,7 +64,6 @@ Config make_config_via_json(nlohmann::json& json) {
         inputs_json.at("str_art_file_names").get<std::vector<std::string>>();
   }
 
-  config.audio_dir_file_name = inputs_json.at("audio_dir_file_name").get<std::string>();
   config.streamed_audio_file_names =
       inputs_json.at("streamed_audio_file_names").get<std::vector<std::string>>();
 
@@ -321,6 +320,9 @@ Config make_config_via_json(nlohmann::json& json) {
   if (json.contains("save_texture_pngs")) {
     config.save_texture_pngs = json.at("save_texture_pngs").get<bool>();
   }
+  if (json.contains("rip_streamed_audio")) {
+    config.rip_streamed_audio = json.at("rip_streamed_audio").get<bool>();
+  }
 
   if (inputs_json.contains("animated_textures")) {
     config.animated_textures =

decompiler/config.h

@@ -102,8 +102,6 @@ struct Config {
   std::vector<std::string> str_file_names;
   std::vector<std::string> str_texture_file_names;
   std::vector<std::string> str_art_file_names;
-
-  std::string audio_dir_file_name;
   std::vector<std::string> streamed_audio_file_names;
 
   std::string obj_file_name_map_file;
@@ -175,6 +173,7 @@ struct Config {
   std::vector<std::string> levels_to_extract;
   bool levels_extract;
   bool save_texture_pngs = false;
+  bool rip_streamed_audio = false;
 
   DecompileHacks hacks;
 

decompiler/config/jak1/jak1_config.jsonc

@@ -122,6 +122,9 @@
   // save game textures as .png files to decompiler_out/<game>/textures
   "save_texture_pngs": false,
 
+  // whether or not to dump out streamed audio files to decompiler_out/<game>/audio
+  "rip_streamed_audio": false,
+
   ////////////////////////////
   // PATCHING OPTIONS
   ////////////////////////////

decompiler/config/jak1/ntsc_v1/inputs.jsonc

@@ -258,10 +258,6 @@
     "TEXT/6COMMON.TXT"
   ],
 
-  // uncomment the next line to extract audio to wave files.
-  //"audio_dir_file_name": "jak1/VAG",
-  "audio_dir_file_name": "",
-
   "streamed_audio_file_names": [
     "VAGWAD.ENG",
     "VAGWAD.FRE",

decompiler/config/jak1_demo/default/inputs.jsonc

@@ -228,9 +228,5 @@
     "TEXT/6COMMON.TXT"
   ],
 
-  // uncomment the next line to extract audio to wave files.
-  //"audio_dir_file_name": "jak1/VAG",
-  "audio_dir_file_name": "",
-
   "streamed_audio_file_names": ["VAGWAD.ENG", "VAGWAD.JAP"]
 }

decompiler/config/jak2/jak2_config.jsonc

@@ -127,6 +127,9 @@
   "rip_collision": false,
   // save game textures as .png files to decompiler_out/<game>/textures
   "save_texture_pngs": false,
+  
+  // whether or not to dump out streamed audio files to decompiler_out/<game>/audio
+  "rip_streamed_audio": false,
 
   ////////////////////////////
   // PATCHING OPTIONS

decompiler/config/jak2/ntsc_v1/inputs.jsonc

@@ -441,10 +441,6 @@
     "TEXT/7COMMON.TXT"
   ],
 
-  // uncomment the next line to extract audio to wave files.
-  // "audio_dir_file_name": "jak2/VAG",
-  "audio_dir_file_name": "",
-
   "streamed_audio_file_names": [
     "VAGWAD.ENG",
     "VAGWAD.FRE",

decompiler/config/jak3/jak3_config.jsonc

@@ -127,6 +127,9 @@
   // save game textures as .png files to decompiler_out/<game>/textures
   "save_texture_pngs": false,
 
+  // whether or not to dump out streamed audio files to decompiler_out/<game>/audio
+  "rip_streamed_audio": false,
+
   ////////////////////////////
   // PATCHING OPTIONS
   ////////////////////////////

decompiler/config/jak3/ntsc_v1/inputs.jsonc

@@ -333,10 +333,6 @@
     "TEXT/11COMMON.TXT"
   ],
 
-  // uncomment the next line to extract audio to wave files.
-  // "audio_dir_file_name": "jak3/VAG",
-  "audio_dir_file_name": "",
-
   "streamed_audio_file_names": [
     "VAGWAD.ENG",
     "VAGWAD.FRE",

decompiler/decompilation_process.cpp

@@ -0,0 +1,284 @@
+#include "decompilation_process.h"
+
+#include <string>
+#include <vector>
+
+#include "config.h"
+
+#include "common/log/log.h"
+#include "common/util/Timer.h"
+#include "common/util/os.h"
+
+#include "Disasm/OpcodeInfo.h"
+#include "ObjectFile/ObjectFileDB.h"
+#include "data/streamed_audio.h"
+#include "level_extractor/extract_level.h"
+
+int run_decompilation_process(decompiler::Config config,
+                              const fs::path& in_folder,
+                              const fs::path& out_folder,
+                              const bool minimal_for_extractor) {
+  using namespace decompiler;
+  Timer decomp_timer;
+
+  lg::info("[Mem] Top of main: {} MB\n", get_peak_rss() / (1024 * 1024));
+
+  init_opcode_info();
+
+  lg::info("[Mem] After init: {} MB\n", get_peak_rss() / (1024 * 1024));
+
+  std::vector<fs::path> dgos, objs, strs, tex_strs, art_strs;
+  if (minimal_for_extractor) {
+    // TODO - does this even matter, or can we just make the DGOs lazily loaded (does it already
+    // happen?)
+    for (const auto& dgo_name : config.dgo_names) {
+      std::string common_name = "GAME.CGO";
+      if (dgo_name.length() > 3 && dgo_name.substr(dgo_name.length() - 3) == "DGO") {
+        // ends in DGO, it's a level
+        dgos.push_back(in_folder / dgo_name);
+      } else if (dgo_name.length() >= common_name.length() &&
+                 dgo_name.substr(dgo_name.length() - common_name.length()) == common_name) {
+        // it's COMMON.CGO, we need that too.
+        dgos.push_back(in_folder / dgo_name);
+      }
+    }
+  } else {
+    for (const auto& dgo_name : config.dgo_names) {
+      dgos.push_back(in_folder / dgo_name);
+    }
+  }
+
+  if (minimal_for_extractor) {
+    // TODO - does this even matter, or can we just make the DGOs lazily loaded (does it already
+    // happen?)
+    for (const auto& obj_name : config.object_file_names) {
+      if (obj_name.length() > 3 && obj_name.substr(obj_name.length() - 3) == "TXT") {
+        // ends in TXT
+        objs.push_back(in_folder / obj_name);
+      }
+    }
+  } else {
+    for (const auto& obj_name : config.object_file_names) {
+      objs.push_back(in_folder / obj_name);
+    }
+  }
+
+  if (!minimal_for_extractor) {
+    for (const auto& str_name : config.str_file_names) {
+      strs.push_back(in_folder / str_name);
+    }
+  }
+
+  for (const auto& str_name : config.str_texture_file_names) {
+    tex_strs.push_back(in_folder / str_name);
+  }
+
+  for (const auto& str_name : config.str_art_file_names) {
+    art_strs.push_back(in_folder / str_name);
+  }
+
+  lg::info("[Mem] After config read: {} MB", get_peak_rss() / (1024 * 1024));
+
+  // build file database
+  lg::info("Setting up object file DB...");
+  ObjectFileDB db(dgos, fs::path(config.obj_file_name_map_file), objs, strs, tex_strs, art_strs,
+                  config);
+
+  // Explicitly fail if a file in the 'allowed_objects' list wasn't found in the DB
+  // as this is another silent error that can be confusing
+  if (!config.allowed_objects.empty()) {
+    for (const auto& expected_obj : config.allowed_objects) {
+      if (db.obj_files_by_name.count(expected_obj) == 0) {
+        // TODO - this is wrong for jak1, fix eventually as this is now done in 3 places
+        lg::error(
+            "Expected to find '{}' in the ObjectFileDB but did not. Check "
+            "./decompiler/config/{}/inputs.jsonc",
+            expected_obj, config.game_name);
+        return 1;
+      }
+    }
+  }
+
+  lg::info("[Mem] After DB setup: {} MB", get_peak_rss() / (1024 * 1024));
+
+  // write out DGO file info
+  file_util::create_dir_if_needed(out_folder);
+  file_util::write_text_file(out_folder / "dgo.txt", db.generate_dgo_listing());
+  // write out object file map (used for future decompilations, if desired)
+  file_util::write_text_file(out_folder / "obj.txt",
+                             db.generate_obj_listing(config.merged_objects));
+
+  // dump raw objs
+  if (config.dump_objs) {
+    auto path = out_folder / "raw_obj";
+    file_util::create_dir_if_needed(path);
+    db.dump_raw_objects(path);
+  }
+
+  // process files (required for all analysis)
+  db.process_link_data(config);
+  lg::info("[Mem] After link data: {} MB", get_peak_rss() / (1024 * 1024));
+  db.find_code(config);
+  db.process_labels();
+  lg::info("[Mem] After code: {} MB", get_peak_rss() / (1024 * 1024));
+
+  // top level decompile (do this before printing asm so we get function names)
+  if (config.find_functions) {
+    db.ir2_top_level_pass(config);
+  }
+
+  // print disassembly
+  if (config.disassemble_code || config.disassemble_data) {
+    db.write_disassembly(out_folder, config.disassemble_data, config.disassemble_code,
+                         config.write_hex_near_instructions);
+  }
+
+  if (config.process_art_groups) {
+    db.extract_art_info();
+    // dump art info to json if requested
+    if (config.dump_art_group_info) {
+      auto ag_file_name = out_folder / "dump" / "art-group-info.min.json";
+      nlohmann::json ag_json = db.dts.art_group_info;
+      file_util::create_dir_if_needed_for_file(ag_file_name);
+      file_util::write_text_file(ag_file_name, ag_json.dump(-1));
+      lg::info("[DUMP] Dumped art group info to {}", ag_file_name.string());
+    }
+    if (config.dump_joint_geo_info) {
+      auto jg_file_name = out_folder / "dump" / "joint-node-info.min.json";
+      nlohmann::json jg_json = db.dts.jg_info;
+      file_util::create_dir_if_needed_for_file(jg_file_name);
+      file_util::write_text_file(jg_file_name, jg_json.dump(-1));
+      lg::info("[DUMP] Dumped joint node info to {}", jg_file_name.string());
+    }
+  } else if (!config.art_group_info_dump.empty() || !config.jg_info_dump.empty()) {
+    // process art groups (used in decompilation)
+    // - if the config has a path to the art info dump, just use that
+    // - otherwise (or if we want to dump it fresh) extract it
+    if (!config.art_group_info_dump.empty()) {
+      db.dts.art_group_info = config.art_group_info_dump;
+    }
+    if (!config.jg_info_dump.empty()) {
+      db.dts.jg_info = config.jg_info_dump;
+    }
+  } else {
+    lg::error("`process_art_groups` was false and no art-group-info dump was provided!");
+    return 1;
+  }
+
+  if (config.process_tpages && !config.texture_info_dump.empty()) {
+    db.dts.textures = config.texture_info_dump;
+  }
+
+  // main decompile.
+  if (config.decompile_code) {
+    db.analyze_functions_ir2(out_folder, config, {}, {}, {});
+  }
+
+  if (config.generate_all_types) {
+    ASSERT_MSG(config.decompile_code, "Must decompile code to generate all-types");
+    db.ir2_analyze_all_types(out_folder / "new-all-types.gc", config.old_all_types_file,
+                             config.hacks.types_with_bad_inspect_methods);
+  }
+
+  lg::info("[Mem] After decomp: {} MB", get_peak_rss() / (1024 * 1024));
+
+  // write out all symbols
+  file_util::write_text_file(out_folder / "all-syms.gc", db.dts.dump_symbol_types());
+
+  // write art groups
+  if (config.process_art_groups) {
+    db.dump_art_info(out_folder);
+  }
+
+  if (config.hexdump_code || config.hexdump_data) {
+    db.write_object_file_words(out_folder, config.hexdump_data, config.hexdump_code);
+  }
+
+  // data stuff
+  if (config.write_scripts) {
+    db.find_and_write_scripts(out_folder);
+  }
+
+  // ensure asset dir exists
+  file_util::create_dir_if_needed(out_folder / "assets");
+
+  if (config.process_game_text) {
+    auto result = db.process_game_text_files(config);
+    if (!result.empty()) {
+      file_util::write_text_file(out_folder / "assets" / "game_text.txt", result);
+    }
+  }
+
+  lg::info("[Mem] After text: {} MB", get_peak_rss() / (1024 * 1024));
+
+  if (config.process_subtitle_text || config.process_subtitle_images) {
+    auto result = db.process_all_spool_subtitles(
+        config, config.process_subtitle_images ? out_folder / "assets" / "subtitle-images" : "");
+    if (!result.empty()) {
+      file_util::write_text_file(out_folder / "assets" / "game_subs.txt", result);
+    }
+  }
+
+  lg::info("[Mem] After spool handling: {} MB", get_peak_rss() / (1024 * 1024));
+
+  TextureDB tex_db;
+  if (config.process_tpages || config.levels_extract) {
+    auto textures_out = out_folder / "textures";
+    auto dump_out = out_folder / "import";
+    file_util::create_dir_if_needed(textures_out);
+    auto result = db.process_tpages(tex_db, textures_out, config, dump_out);
+    if (!result.empty() && config.process_tpages) {
+      file_util::write_text_file(textures_out / "tpage-dir.txt", result);
+      file_util::write_text_file(textures_out / "tex-remap.txt",
+                                 tex_db.generate_texture_dest_adjustment_table());
+    }
+    if (config.dump_tex_info) {
+      auto texture_file_name = out_folder / "dump" / "tex-info.min.json";
+      nlohmann::json texture_json = db.dts.textures;
+      file_util::create_dir_if_needed_for_file(texture_file_name);
+      file_util::write_text_file(texture_file_name, texture_json.dump(-1));
+      lg::info("[DUMP] Dumped texture info to {}", texture_file_name.string());
+    }
+  }
+
+  lg::info("[Mem] After textures: {} MB", get_peak_rss() / (1024 * 1024));
+
+  // Merge textures before replacing them, in other words, replacements take priority
+  auto texture_merge_path = file_util::get_jak_project_dir() / "game" / "assets" /
+                            game_version_names[config.game_version] / "texture_merges";
+  if (fs::exists(texture_merge_path)) {
+    tex_db.merge_textures(texture_merge_path);
+  }
+
+  auto replacements_path = file_util::get_jak_project_dir() / "custom_assets" /
+                           game_version_names[config.game_version] / "texture_replacements";
+  if (fs::exists(replacements_path)) {
+    tex_db.replace_textures(replacements_path);
+  }
+
+  if (config.process_game_count) {
+    auto result = db.process_game_count_file();
+    if (!result.empty()) {
+      file_util::write_text_file(out_folder / "assets" / "game_count.txt", result);
+    }
+  }
+
+  if (config.levels_extract) {
+    auto level_out_path =
+        file_util::get_jak_project_dir() / "out" / game_version_names[config.game_version] / "fr3";
+    file_util::create_dir_if_needed(level_out_path);
+    extract_all_levels(db, tex_db, config.levels_to_extract, "GAME.CGO", config, level_out_path);
+  }
+
+  lg::info("[Mem] After extraction: {} MB", get_peak_rss() / (1024 * 1024));
+
+  if (config.rip_streamed_audio) {
+    auto streaming_audio_out = out_folder / "audio";
+    file_util::create_dir_if_needed(streaming_audio_out);
+    process_streamed_audio(config, streaming_audio_out, in_folder,
+                           config.streamed_audio_file_names);
+  }
+
+  lg::info("Decompiler has finished successfully in {:.2f} seconds.", decomp_timer.getSeconds());
+  return 0;
+}

decompiler/decompilation_process.h

@@ -0,0 +1,8 @@
+#pragma once
+
+#include "config.h"
+
+int run_decompilation_process(decompiler::Config config,
+                              const fs::path& in_folder,
+                              const fs::path& out_folder,
+                              const bool minimal_for_extractor);

decompiler/extractor/main.cpp

@@ -5,10 +5,9 @@
 #include "common/util/term_util.h"
 #include "common/util/unicode_util.h"
 
-#include "decompiler/ObjectFile/ObjectFileDB.h"
 #include "decompiler/config.h"
+#include "decompiler/decompilation_process.h"
 #include "decompiler/extractor/extractor_util.h"
-#include "decompiler/level_extractor/extract_level.h"
 #include "goalc/compiler/Compiler.h"
 
 #include "third-party/CLI11.hpp"
@@ -102,116 +101,24 @@ std::tuple<std::optional<ISOMetadata>, ExtractorErrorCode> validate(
   };
 }
 
-// TODO - remove code duplication, most of this is copying what happens in decompiler's `main.cpp`
-void decompile(const fs::path& iso_data_path,
-               const std::string& data_subfolder,
-               const std::string& config_override) {
-  using namespace decompiler;
-
+ExtractorErrorCode decompile(const fs::path& in_folder,
+                             const std::string& data_subfolder,
+                             const std::string& config_override) {
   // Determine which config to use from the database
-  const auto version_info = get_version_info_or_default(iso_data_path);
+  const auto version_info = get_version_info_or_default(in_folder);
 
-  Config config = read_config_file(file_util::get_jak_project_dir() / "decompiler" / "config" /
-                                       version_info.game_name /
-                                       fmt::format("{}_config.jsonc", version_info.game_name),
-                                   version_info.decomp_config_version, config_override);
+  decompiler::Config config = decompiler::read_config_file(
+      file_util::get_jak_project_dir() / "decompiler" / "config" / version_info.game_name /
+          fmt::format("{}_config.jsonc", version_info.game_name),
+      version_info.decomp_config_version, config_override);
 
-  std::vector<fs::path> dgos, objs, tex_strs, art_strs;
-
-  // grab all DGOS we need (level + common)
-  // TODO - Jak 2 - jak 1 specific code?
-  for (const auto& dgo_name : config.dgo_names) {
-    std::string common_name = "GAME.CGO";
-    if (dgo_name.length() > 3 && dgo_name.substr(dgo_name.length() - 3) == "DGO") {
-      // ends in DGO, it's a level
-      dgos.push_back(iso_data_path / dgo_name);
-    } else if (dgo_name.length() >= common_name.length() &&
-               dgo_name.substr(dgo_name.length() - common_name.length()) == common_name) {
-      // it's COMMON.CGO, we need that too.
-      dgos.push_back(iso_data_path / dgo_name);
-    }
-  }
-
-  // grab all the object files we need (just text)
-  for (const auto& obj_name : config.object_file_names) {
-    if (obj_name.length() > 3 && obj_name.substr(obj_name.length() - 3) == "TXT") {
-      // ends in TXT
-      objs.push_back(iso_data_path / obj_name);
-    }
-  }
-
-  for (const auto& str_name : config.str_texture_file_names) {
-    tex_strs.push_back(iso_data_path / str_name);
-  }
-
-  for (const auto& str_name : config.str_art_file_names) {
-    art_strs.push_back(iso_data_path / str_name);
-  }
-
-  // set up objects
-  ObjectFileDB db(dgos, fs::path(config.obj_file_name_map_file), objs, {}, tex_strs, art_strs,
-                  config);
-
-  // save object files
   auto out_folder = file_util::get_jak_project_dir() / "decompiler_out" / data_subfolder;
-  auto raw_obj_folder = out_folder / "raw_obj";
-  file_util::create_dir_if_needed(raw_obj_folder);
-  db.dump_raw_objects(raw_obj_folder);
 
-  // analyze object file link data
-  db.process_link_data(config);
-  db.find_code(config);
-  db.process_labels();
-
-  // ensure asset dir exists
-  file_util::create_dir_if_needed(out_folder / "assets");
-
-  // text files
-  {
-    auto result = db.process_game_text_files(config);
-    if (!result.empty()) {
-      file_util::write_text_file(out_folder / "assets" / "game_text.txt", result);
-    }
-  }
-
-  // textures
-  decompiler::TextureDB tex_db;
-  auto textures_out = out_folder / "textures";
-  auto dump_out = out_folder / "import";
-  file_util::create_dir_if_needed(textures_out);
-  file_util::write_text_file(textures_out / "tpage-dir.txt",
-                             db.process_tpages(tex_db, textures_out, config, dump_out));
-
-  // texture merges
-  // TODO - put all this stuff in somewhere common
-  auto texture_merge_path = file_util::get_jak_project_dir() / "game" / "assets" /
-                            game_version_names[config.game_version] / "texture_merges";
-  if (fs::exists(texture_merge_path)) {
-    tex_db.merge_textures(texture_merge_path);
-  }
-
-  // texture replacements
-  auto replacements_path = file_util::get_jak_project_dir() / "custom_assets" /
-                           game_version_names[config.game_version] / "texture_replacements";
-  if (fs::exists(replacements_path)) {
-    tex_db.replace_textures(replacements_path);
-  }
-
-  // game count
-  {
-    auto result = db.process_game_count_file();
-    if (!result.empty()) {
-      file_util::write_text_file(out_folder / "assets" / "game_count.txt", result);
-    }
-  }
-
-  // levels
-  {
-    auto level_out_path =
-        file_util::get_jak_project_dir() / "out" / game_version_names[config.game_version] / "fr3";
-    file_util::create_dir_if_needed(level_out_path);
-    extract_all_levels(db, tex_db, config.levels_to_extract, "GAME.CGO", config, level_out_path);
+  const auto result = run_decompilation_process(config, in_folder, out_folder, true);
+  if (result != 0) {
+    return ExtractorErrorCode::DECOMPILATION_GENERIC_ERROR;
   }
+  return ExtractorErrorCode::SUCCESS;
 }
 
 const std::unordered_map<std::string, GameIsoFlags> game_iso_flag_names = {
@@ -315,7 +222,6 @@ int main(int argc, char** argv) {
   }
 
   // - SETUP
-  decompiler::init_opcode_info();
   if (!project_path_override.empty()) {
     if (!fs::exists(project_path_override)) {
       lg::error("Error: project path override '{}' does not exist", project_path_override.string());
@@ -426,7 +332,11 @@ int main(int argc, char** argv) {
       // Get hash and file count
       const auto [hash, file_count] = calculate_extraction_hash(iso_data_path);
       // Validate
-      auto [version_info, code] = validate(iso_data_path, hash, file_count);
+      auto [version_info, validate_code] = validate(iso_data_path, hash, file_count);
+      if (validate_code == ExtractorErrorCode::VALIDATION_BAD_EXTRACTION ||
+          (flag_fail_on_validation && validate_code != ExtractorErrorCode::SUCCESS)) {
+        return static_cast<int>(validate_code);
+      }
     }
 
     // write out a json file with some metadata for the game
@@ -455,7 +365,10 @@ int main(int argc, char** argv) {
 
   if (flag_decompile) {
     try {
-      decompile(iso_data_path, data_subfolder, decomp_config_override);
+      const auto status_code = decompile(iso_data_path, data_subfolder, decomp_config_override);
+      if (status_code != ExtractorErrorCode::SUCCESS) {
+        return static_cast<int>(status_code);
+      }
     } catch (std::exception& e) {
       lg::error("Error during decompile: {}", e.what());
       return static_cast<int>(ExtractorErrorCode::DECOMPILATION_GENERIC_ERROR);
@@ -463,7 +376,10 @@ int main(int argc, char** argv) {
   }
 
   if (flag_compile) {
-    compile(iso_data_path, data_subfolder);
+    const auto status_code = compile(iso_data_path, data_subfolder);
+    if (status_code != ExtractorErrorCode::SUCCESS) {
+      return static_cast<int>(status_code);
+    }
   }
 
   if (flag_play) {

decompiler/main.cpp

@@ -1,33 +1,15 @@
-#include <cstdio>
 #include <string>
-#include <vector>
 
 #include "config.h"
+#include "decompilation_process.h"
 
 #include "common/log/log.h"
 #include "common/util/FileUtil.h"
-#include "common/util/Timer.h"
-#include "common/util/diff.h"
-#include "common/util/os.h"
 #include "common/util/set_util.h"
 #include "common/util/term_util.h"
 #include "common/util/unicode_util.h"
-#include "common/versions/versions.h"
-
-#include "ObjectFile/ObjectFileDB.h"
-#include "decompiler/data/TextureDB.h"
-#include "decompiler/data/streamed_audio.h"
-#include "decompiler/level_extractor/extract_level.h"
 
 #include "third-party/CLI11.hpp"
-#include "third-party/json.hpp"
-
-template <typename... Args>
-static void mem_log(const std::string& format, Args&&... args) {
-#ifndef _WIN32
-  lg::info("[Mem] " + format, std::forward<Args>(args)...);
-#endif
-}
 
 int main(int argc, char** argv) {
   ArgumentGuard u8_guard(argc, argv);
@@ -81,11 +63,9 @@ int main(int argc, char** argv) {
     return 1;
   }
 
-  using namespace decompiler;
-
-  Config config;
+  decompiler::Config config;
   try {
-    config = read_config_file(config_path, config_game_version, config_override);
+    config = decompiler::read_config_file(config_path, config_game_version, config_override);
   } catch (const std::exception& e) {
     lg::error("Failed to parse config: {}", e.what());
     return 1;
@@ -115,7 +95,7 @@ int main(int argc, char** argv) {
 
   // Warning message if expected ELF isn't found, user could be using bad assets / didn't extract
   // the ISO properly
-  if (!config.expected_elf_name.empty() && !exists(in_folder / config.expected_elf_name)) {
+  if (!config.expected_elf_name.empty() && !fs::exists(in_folder / config.expected_elf_name)) {
     lg::error(
         "WARNING - '{}' does not contain the expected ELF file '{}'.  Was the ISO extracted "
         "properly or is there a version mismatch?",
@@ -123,235 +103,5 @@ int main(int argc, char** argv) {
   }
 
   // -- Begin the Decompilation!
-
-  Timer decomp_timer;
-
-  mem_log("Top of main: {} MB\n", get_peak_rss() / (1024 * 1024));
-
-  init_opcode_info();
-
-  mem_log("After init: {} MB\n", get_peak_rss() / (1024 * 1024));
-
-  std::vector<fs::path> dgos, objs, strs, tex_strs, art_strs;
-  for (const auto& dgo_name : config.dgo_names) {
-    dgos.push_back(in_folder / dgo_name);
-  }
-
-  for (const auto& obj_name : config.object_file_names) {
-    objs.push_back(in_folder / obj_name);
-  }
-
-  for (const auto& str_name : config.str_file_names) {
-    strs.push_back(in_folder / str_name);
-  }
-
-  for (const auto& str_name : config.str_texture_file_names) {
-    tex_strs.push_back(in_folder / str_name);
-  }
-
-  for (const auto& str_name : config.str_art_file_names) {
-    art_strs.push_back(in_folder / str_name);
-  }
-
-  mem_log("After config read: {} MB", get_peak_rss() / (1024 * 1024));
-
-  // build file database
-  lg::info("Setting up object file DB...");
-  ObjectFileDB db(dgos, fs::path(config.obj_file_name_map_file), objs, strs, tex_strs, art_strs,
-                  config);
-
-  // Explicitly fail if a file in the 'allowed_objects' list wasn't found in the DB
-  // as this is another silent error that can be confusing
-  if (!config.allowed_objects.empty()) {
-    for (const auto& expected_obj : config.allowed_objects) {
-      if (db.obj_files_by_name.count(expected_obj) == 0) {
-        // TODO - this is wrong for jak1, fix eventually as this is now done in 3 places
-        lg::error(
-            "Expected to find '{}' in the ObjectFileDB but did not. Check "
-            "./decompiler/config/{}/inputs.jsonc",
-            expected_obj, config.game_name);
-        return 1;
-      }
-    }
-  }
-
-  mem_log("After DB setup: {} MB", get_peak_rss() / (1024 * 1024));
-
-  // write out DGO file info
-  file_util::write_text_file(out_folder / "dgo.txt", db.generate_dgo_listing());
-  // write out object file map (used for future decompilations, if desired)
-  file_util::write_text_file(out_folder / "obj.txt",
-                             db.generate_obj_listing(config.merged_objects));
-
-  // dump raw objs
-  if (config.dump_objs) {
-    auto path = out_folder / "raw_obj";
-    file_util::create_dir_if_needed(path);
-    db.dump_raw_objects(path);
-  }
-
-  // process files (required for all analysis)
-  db.process_link_data(config);
-  mem_log("After link data: {} MB", get_peak_rss() / (1024 * 1024));
-  db.find_code(config);
-  db.process_labels();
-  mem_log("After code: {} MB", get_peak_rss() / (1024 * 1024));
-
-  // top level decompile (do this before printing asm so we get function names)
-  if (config.find_functions) {
-    db.ir2_top_level_pass(config);
-  }
-
-  // print disassembly
-  if (config.disassemble_code || config.disassemble_data) {
-    db.write_disassembly(out_folder, config.disassemble_data, config.disassemble_code,
-                         config.write_hex_near_instructions);
-  }
-
-  if (config.process_art_groups) {
-    db.extract_art_info();
-    // dump art info to json if requested
-    if (config.dump_art_group_info) {
-      auto ag_file_name = out_folder / "dump" / "art-group-info.min.json";
-      nlohmann::json ag_json = db.dts.art_group_info;
-      file_util::create_dir_if_needed_for_file(ag_file_name);
-      file_util::write_text_file(ag_file_name, ag_json.dump(-1));
-      lg::info("[DUMP] Dumped art group info to {}", ag_file_name.string());
-    }
-    if (config.dump_joint_geo_info) {
-      auto jg_file_name = out_folder / "dump" / "joint-node-info.min.json";
-      nlohmann::json jg_json = db.dts.jg_info;
-      file_util::create_dir_if_needed_for_file(jg_file_name);
-      file_util::write_text_file(jg_file_name, jg_json.dump(-1));
-      lg::info("[DUMP] Dumped joint node info to {}", jg_file_name.string());
-    }
-  } else if (!config.art_group_info_dump.empty() || !config.jg_info_dump.empty()) {
-    // process art groups (used in decompilation)
-    // - if the config has a path to the art info dump, just use that
-    // - otherwise (or if we want to dump it fresh) extract it
-    if (!config.art_group_info_dump.empty()) {
-      db.dts.art_group_info = config.art_group_info_dump;
-    }
-    if (!config.jg_info_dump.empty()) {
-      db.dts.jg_info = config.jg_info_dump;
-    }
-  } else {
-    lg::error("`process_art_groups` was false and no art-group-info dump was provided!");
-    return 1;
-  }
-
-  if (config.process_tpages && !config.texture_info_dump.empty()) {
-    db.dts.textures = config.texture_info_dump;
-  }
-
-  // main decompile.
-  if (config.decompile_code) {
-    db.analyze_functions_ir2(out_folder, config, {}, {}, {});
-  }
-
-  if (config.generate_all_types) {
-    ASSERT_MSG(config.decompile_code, "Must decompile code to generate all-types");
-    db.ir2_analyze_all_types(out_folder / "new-all-types.gc", config.old_all_types_file,
-                             config.hacks.types_with_bad_inspect_methods);
-  }
-
-  mem_log("After decomp: {} MB", get_peak_rss() / (1024 * 1024));
-
-  // write out all symbols
-  file_util::write_text_file(out_folder / "all-syms.gc", db.dts.dump_symbol_types());
-
-  // write art groups
-  if (config.process_art_groups) {
-    db.dump_art_info(out_folder);
-  }
-
-  if (config.hexdump_code || config.hexdump_data) {
-    db.write_object_file_words(out_folder, config.hexdump_data, config.hexdump_code);
-  }
-
-  // data stuff
-  if (config.write_scripts) {
-    db.find_and_write_scripts(out_folder);
-  }
-
-  if (config.process_game_text) {
-    auto result = db.process_game_text_files(config);
-    if (!result.empty()) {
-      file_util::write_text_file(out_folder / "assets" / "game_text.txt", result);
-    }
-  }
-
-  mem_log("After text: {} MB", get_peak_rss() / (1024 * 1024));
-
-  if (config.process_subtitle_text || config.process_subtitle_images) {
-    auto result = db.process_all_spool_subtitles(
-        config, config.process_subtitle_images ? out_folder / "assets" / "subtitle-images" : "");
-    if (!result.empty()) {
-      file_util::write_text_file(out_folder / "assets" / "game_subs.txt", result);
-    }
-  }
-
-  mem_log("After spool handling: {} MB", get_peak_rss() / (1024 * 1024));
-
-  TextureDB tex_db;
-  if (config.process_tpages || config.levels_extract) {
-    auto textures_out = out_folder / "textures";
-    auto dump_out = out_folder / "import";
-    file_util::create_dir_if_needed(textures_out);
-    auto result = db.process_tpages(tex_db, textures_out, config, dump_out);
-    if (!result.empty() && config.process_tpages) {
-      file_util::write_text_file(textures_out / "tpage-dir.txt", result);
-      file_util::write_text_file(textures_out / "tex-remap.txt",
-                                 tex_db.generate_texture_dest_adjustment_table());
-    }
-    if (config.dump_tex_info) {
-      auto texture_file_name = out_folder / "dump" / "tex-info.min.json";
-      nlohmann::json texture_json = db.dts.textures;
-      file_util::create_dir_if_needed_for_file(texture_file_name);
-      file_util::write_text_file(texture_file_name, texture_json.dump(-1));
-      lg::info("[DUMP] Dumped texture info to {}", texture_file_name.string());
-    }
-  }
-
-  mem_log("After textures: {} MB", get_peak_rss() / (1024 * 1024));
-
-  // Merge textures before replacing them, in other words, replacements take priority
-  auto texture_merge_path = file_util::get_jak_project_dir() / "game" / "assets" /
-                            game_version_names[config.game_version] / "texture_merges";
-  if (fs::exists(texture_merge_path)) {
-    tex_db.merge_textures(texture_merge_path);
-  }
-
-  auto replacements_path = file_util::get_jak_project_dir() / "custom_assets" /
-                           game_version_names[config.game_version] / "texture_replacements";
-  if (fs::exists(replacements_path)) {
-    tex_db.replace_textures(replacements_path);
-  }
-
-  if (config.process_game_count) {
-    auto result = db.process_game_count_file();
-    if (!result.empty()) {
-      file_util::write_text_file(out_folder / "assets" / "game_count.txt", result);
-    }
-  }
-
-  if (config.levels_extract) {
-    auto level_out_path =
-        file_util::get_jak_project_dir() / "out" / game_version_names[config.game_version] / "fr3";
-    file_util::create_dir_if_needed(level_out_path);
-    extract_all_levels(db, tex_db, config.levels_to_extract, "GAME.CGO", config, level_out_path);
-  }
-
-  mem_log("After extraction: {} MB", get_peak_rss() / (1024 * 1024));
-
-  if (!config.audio_dir_file_name.empty()) {
-    auto streaming_audio_in = in_folder / "VAG";
-    auto streaming_audio_out = out_folder / "assets" / "streaming_audio";
-    file_util::create_dir_if_needed(streaming_audio_out);
-    process_streamed_audio(config, streaming_audio_out, in_folder,
-                           config.streamed_audio_file_names);
-  }
-
-  lg::info("Decompiler has finished successfully in {:.2f} seconds.", decomp_timer.getSeconds());
-  return 0;
+  return run_decompilation_process(config, in_folder, out_folder, false);
 }