feat: complete blob-extractor.

src/tools/blob-extractor/main.cpp

@@ -1,107 +1,65 @@
+#include "data_format/magic_blob.h"
 
-#define XXH_INLINE_ALL
-#include "xxhash.h"
+#include <argparse/argparse.hpp>
+#include <nlohmann/json.hpp>
 
-class File : public std::ifstream {
-public:
-    using std::ifstream::basic_ifstream;
+using namespace di;
 
-    template <typename T>
-    inline T read() {
-        T value;
-        std::ifstream::read((char*)&value, sizeof(T));
-        return value;
-    }
+auto load_args(int argc, char* argv[]) {
+    argparse::ArgumentParser program("blob-extractor");
 
-    template <std::unsigned_integral T>
-    inline T read_varint() {
-        T   res   = 0;
-        int shift = 0;
-        while (true) {
-            auto byte  = std::ifstream::get();
-            res       |= static_cast<T>(byte & 0x7F) << shift;
-            if ((byte & 0x80) == 0) break;
-            shift += 7;
-        }
-        return res;
-    }
-};
+    struct {
+        std::string m_magic_blob_path;
+        std::string m_output_path;
+    } args;
 
-#define HIDWORD(x) (*((int32_t*)&(x) + 1))
+    // clang-format off
 
-uint64_t unk_hash(uint64_t a1) {
-    unsigned int v1; // eax
-    int          v2; // edx
-    int64_t      v3; // rdx
+    program.add_argument("magicblob")
+        .help("Path to magic blob.")
+        .store_into(args.m_magic_blob_path)
+        .required();
 
-    v1 = ((33
-               * ((4097 * HIDWORD(a1) + 2127912214)
-                  ^ ((unsigned int)(4097 * HIDWORD(a1) + 2127912214) >> 19)
-                  ^ 0xC761C23C)
-           + 374761393)
-          << 9)
-       ^ (33
-              * ((4097 * HIDWORD(a1) + 2127912214)
-                 ^ ((unsigned int)(4097 * HIDWORD(a1) + 2127912214) >> 19)
-                 ^ 0xC761C23C)
-          - 369570787);
-    v2 = 33
-       * ((4097 * a1 + 2127912214)
-          ^ ((unsigned int)(4097 * a1 + 2127912214) >> 19) ^ 0xC761C23C);
-    v3 = (((v2 + 374761393) << 9) ^ (v2 - 369570787))
-       + 8 * (((v2 + 374761393) << 9) ^ (unsigned int)(v2 - 369570787))
-       - 42973499;
-    return (v3 ^ (((unsigned int)v3 ^ 0xB55A4F090000uLL) >> 16))
-         | ((((v1 + 8 * v1 - 42973499) & 0xFFFF0000)
-             ^ (((v1 + 8 * v1 - 42973499) ^ 0xFFFFFFFFB55A4F09uLL) << 16))
-            << 16);
+    program.add_argument("--output", "-o")
+        .help("Path to output symlist.")
+        .store_into(args.m_output_path)
+        .required();
+
+    // clang-format on
+
+    program.parse_args(argc, argv);
+
+    return args;
 }
 
-struct Entry {
-    std::bitset<64> flags;
-    uint32_t        rva;
-    uint64_t        hash;
+int main(int argc, char* argv[]) try {
 
-    constexpr bool is_function() { return flags[0]; }
-    constexpr bool _unk2() { return flags[1]; }
-    constexpr bool is_verbose() { return flags[2]; }
-    constexpr bool _unk4() { return flags[3]; }
-};
+    auto args = load_args(argc, argv);
 
-int main(int argc, char** argv) {
-    File data("bedrock_runtime_data", std::ios::binary);
+    data_format::MagicBlob blob;
+    blob.read(args.m_magic_blob_path);
 
-    uint32_t t_rva{};
-    auto     record_seed = data.read<uint64_t>();
-    auto     twin_seed   = unk_hash(record_seed);
+    nlohmann::json data;
+    blob.for_each([&data](hash_t hash, const MagicEntry& entry) {
+        data.emplace_back(nlohmann::json{
+            {"hash",        hash               },
+            {"rva",         entry.rva          },
+            {"is_function", entry.is_function()},
+            {"_unk2",       entry._unk2()      },
+            {"is_verbose",  entry.is_verbose() },
+            {"_unk4",       entry._unk4()      }
+        });
+    });
 
-    std::println("Record seed: {:#x}", record_seed);
-    std::println("Twin seed: {:#x}", twin_seed);
-
-    std::unordered_map<uint64_t, Entry> map;
-
-    while (data.peek() != EOF) {
-        Entry entry;
-        entry.flags = data.read_varint<uint64_t>();
-        entry.rva   = data.read_varint<uint32_t>();
-        entry.hash  = data.read<uint64_t>();
-
-        t_rva     += entry.rva;
-        entry.rva  = t_rva;
-
-        map.emplace(entry.hash, entry);
-        // entry.print_debug_string();
+    std::ofstream ofs(args.m_output_path);
+    if (!ofs) {
+        throw std::runtime_error("Failed to open file!");
     }
 
-    std::string_view test_query_name = "main";
-    auto             test_query_hash =
-        XXH64(test_query_name.data(), test_query_name.size(), twin_seed);
-
-    if (map.contains(test_query_hash)) {
-        std::println("RVA of main(): {:#x}", map.at(test_query_hash).rva);
-    } else {
-        std::println("RVA of main(): INVALID.");
-    }
+    ofs << data.dump(4);
 
     return 0;
+} catch (const std::exception& e) {
+    std::println("E: {}", e.what());
+    return -1;
 }

xmake.lua

@@ -48,14 +48,13 @@ target('libdi')
     add_files('src/**.cpp')
     set_pcxxheader('src/pch.h')
     
+    remove_files('src/tools/**')
     set_basename('di')
 
     add_packages(
         'nlohmann_json'
     )
 
-    remove_files('src/tools/**')
-
 target('askrva')
     set_kind('binary')
     add_deps('libdi')
@@ -78,6 +77,11 @@ target('blob-extractor')
     add_files('src/tools/blob-extractor/**.cpp')
     set_pcxxheader('src/pch.h')
 
+    add_packages(
+        'nlohmann_json',
+        'argparse'
+    )
+
 target('dumpsym')
     set_kind('shared')
     add_deps('libdi')