refactor: polymorphic di::data_format::MagicBlob.

src/data_format/_pl/v1_12_0/blob_impl.cpp

@@ -0,0 +1,74 @@
+#include "blob_impl.h"
+#include "entry_impl.h"
+
+#define XXH_INLINE_ALL
+#include "xxhash.h"
+
+namespace {
+
+constexpr uint64_t hash_qseed(uint64_t stored_seed) {
+    constexpr auto v1 = 0x7ED55D16u;
+    constexpr auto v2 = 0xC761C23Cu;
+    constexpr auto v3 = 0x165667B1u;
+    constexpr auto v4 = 0x160733E3u;
+    constexpr auto v5 = 0x028FB93Bu;
+    constexpr auto v6 = 0xB55A4F09uLL;
+
+    constexpr auto al = [](uint32_t a) {
+        auto c1 = (0x1000 + 1) * a + v1;
+        auto d1 = (0x20 + 1) * (c1 ^ (c1 >> 19) ^ v2);
+        auto e1 = ((d1 + v3) << 9) ^ (d1 - v4);
+        return e1 + 8 * e1 - v5;
+    };
+
+    auto a1 = al(stored_seed >> 32);
+    auto a2 = al(stored_seed);
+
+    auto b1 = v6 | 0xFFFFFFFF00000000uLL;
+
+    auto c1 = ((a1 & 0xFFFF0000) ^ ((a1 ^ b1) << 16)) << 16;
+    auto c2 = a2 ^ ((a2 ^ v6 << 16) >> 16);
+
+    return c1 | c2;
+}
+
+} // namespace
+
+namespace di::data_format::_pl::v1_12_0 {
+
+void MagicBlobImpl::read(const fs::path& path) {
+    StreamedIO::read(path);
+
+    m_stored_seed = eat<uint64_t>();
+    m_query_seed  = hash_qseed(m_stored_seed);
+
+    rva_t n_rva{};
+
+    while (next() != EOF) {
+        auto flags = eat_varint<uint64_t>();
+        auto rva   = eat_varint<rva_t>();
+        auto hash  = eat<hash_t>();
+
+        // What is stored in the original format is not the RVA itself, but the
+        // difference with the previous entry (in MagicBlob, RVA is sorted from
+        // small to large)
+        // But here, we still store the "real" RVA.
+        n_rva += rva;
+        rva    = n_rva;
+
+        m_entries.emplace(
+            hash,
+            std::make_shared<MagicEntryImpl>(hash, rva, flags)
+        );
+    }
+}
+
+MagicBlob::shared_entry_t MagicBlobImpl::query(std::string_view symbol) const {
+    auto query_hash = XXH64(symbol.data(), symbol.size(), m_query_seed);
+    if (m_entries.contains(query_hash)) {
+        return m_entries.at(query_hash);
+    }
+    return nullptr;
+}
+
+} // namespace di::data_format::_pl::v1_12_0

src/data_format/_pl/v1_12_0/blob_impl.h

@@ -0,0 +1,31 @@
+#pragma once
+
+#include "data_format/magic_blob.h"
+
+namespace di::data_format::_pl::v1_12_0 {
+
+class MagicBlobImpl : public MagicBlob {
+public:
+    using for_each_callback_t =
+        std::function<void(hash_t, shared_entry_t const&)>;
+
+    void read(const fs::path& path) override;
+
+    shared_entry_t query(std::string_view symbol) const override;
+
+    void for_each(const for_each_callback_t& callback) const override {
+        for (const auto& [hash, entry] : m_entries) callback(hash, entry);
+    }
+
+    size_t count() const override { return m_entries.size(); }
+
+private:
+    std::unordered_map<hash_t, shared_entry_t> m_entries;
+
+    // MagicBlob uses a custom algorithm to transform the stored seed. When
+    // querying, you should use m_query_seed.
+    uint64_t m_stored_seed{};
+    uint64_t m_query_seed{};
+};
+
+} // namespace di::data_format::_pl::v1_12_0

src/data_format/_pl/v1_12_0/entry_impl.cpp

@@ -0,0 +1,16 @@
+#include "entry_impl.h"
+
+#include <nlohmann/json.hpp>
+
+namespace di::data_format::_pl::v1_12_0 {
+
+void MagicEntryImpl::to_json(nlohmann::json& json) const {
+    MagicEntry::to_json(json);
+
+    json["is_function"] = is_function();
+    json["_unk2"]       = _unk2();
+    json["is_verbose"]  = is_verbose();
+    json["_unk4"]       = _unk4();
+}
+
+} // namespace di::data_format::_pl::v1_12_0

src/data_format/_pl/v1_12_0/entry_impl.h

@@ -0,0 +1,29 @@
+#pragma once
+
+#include "data_format/type/magic_entry.h"
+
+#include <nlohmann/json_fwd.hpp>
+
+namespace di::data_format::_pl::v1_12_0 {
+
+struct MagicEntryImpl : public MagicEntry {
+    using flags_t = std::bitset<64>;
+
+    flags_t flags;
+
+    MagicEntryImpl(hash_t hash, rva_t rva, flags_t flags)
+    : MagicEntry(hash, rva),
+      flags(flags) {}
+
+    void to_json(nlohmann::json& json) const override;
+
+    constexpr bool is_function() const { return flags[0]; }
+    constexpr bool _unk2() const { return flags[1]; }
+    constexpr bool is_verbose() const { return flags[2]; }
+    constexpr bool _unk4() const { return flags[3]; }
+};
+
+// TODO
+// void from_json(const nlohmann::json& json, MagicEntry& entry);
+
+} // namespace di::data_format::_pl::v1_12_0

src/data_format/magic_blob.cpp

@@ -1,67 +1,15 @@
-#include "magic_blob.h"
+#include "data_format/magic_blob.h"
 
-#define XXH_INLINE_ALL
-#include "xxhash.h"
-
-namespace _preloader_v_1_12_0 {
-
-constexpr uint64_t hash_qseed(uint64_t stored_seed) {
-    constexpr auto v1 = 0x7ED55D16u;
-    constexpr auto v2 = 0xC761C23Cu;
-    constexpr auto v3 = 0x165667B1u;
-    constexpr auto v4 = 0x160733E3u;
-    constexpr auto v5 = 0x028FB93Bu;
-    constexpr auto v6 = 0xB55A4F09uLL;
-
-    constexpr auto al = [](uint32_t a) {
-        auto c1 = (0x1000 + 1) * a + v1;
-        auto d1 = (0x20 + 1) * (c1 ^ (c1 >> 19) ^ v2);
-        auto e1 = ((d1 + v3) << 9) ^ (d1 - v4);
-        return e1 + 8 * e1 - v5;
-    };
-
-    auto a1 = al(stored_seed >> 32);
-    auto a2 = al(stored_seed);
-
-    auto b1 = v6 | 0xFFFFFFFF00000000uLL;
-
-    auto c1 = ((a1 & 0xFFFF0000) ^ ((a1 ^ b1) << 16)) << 16;
-    auto c2 = a2 ^ ((a2 ^ v6 << 16) >> 16);
-
-    return c1 | c2;
-}
-
-} // namespace _preloader_v_1_12_0
+#include "data_format/_pl/v1_12_0/blob_impl.h"
 
 namespace di::data_format {
 
-void MagicBlob::read(const fs::path& path) {
-    StreamedIO::read(path);
-
-    m_stored_seed = eat<uint64_t>();
-    m_query_seed  = _preloader_v_1_12_0::hash_qseed(m_stored_seed);
-
-    rva_t n_rva{};
-
-    while (next() != EOF) {
-        auto flags = eat_varint<uint64_t>();
-        auto rva   = eat_varint<rva_t>();
-        auto hash  = eat<hash_t>();
-
-        // see comments in magic_entry.h
-        n_rva += rva;
-        rva    = n_rva;
-
-        m_entries.emplace(hash, std::make_unique<MagicEntry>(flags, rva));
-    }
-}
-
-MagicEntry const* MagicBlob::query(std::string_view symbol) const {
-    auto query_hash = XXH64(symbol.data(), symbol.size(), m_query_seed);
-    if (m_entries.contains(query_hash)) {
-        return m_entries.at(query_hash).get();
+MagicBlob::blob_t MagicBlob::create(FormatVersion version) {
+    switch (version) {
+    case V1_12_0:
+        return std::make_unique<_pl::v1_12_0::MagicBlobImpl>();
     }
     return nullptr;
 }
 
-} // namespace di::data_format
+} // namespace di::data_format

src/data_format/magic_blob.h

@@ -7,25 +7,29 @@ namespace di::data_format {
 
 class MagicBlob : public io::StreamedIO {
 public:
-    using for_each_callback_t = std::function<void(hash_t, MagicEntry const&)>;
+    enum FormatVersion {
+        V1_12_0,
+    };
 
-    void read(const fs::path& path) override;
+    using blob_t  = std::unique_ptr<MagicBlob>;
+    using entry_t = std::unique_ptr<MagicEntry>;
 
-    DI_CONSTEXPR void for_each(const for_each_callback_t& callback) const {
-        for (const auto& [hash, entry] : m_entries) callback(hash, *entry);
-    }
+    using shared_blob_t  = std::shared_ptr<const MagicBlob>;
+    using shared_entry_t = std::shared_ptr<const MagicEntry>;
 
-    constexpr size_t count() const { return m_entries.size(); }
+    using for_each_callback_t =
+        std::function<void(hash_t, shared_entry_t const&)>;
 
-    MagicEntry const* query(std::string_view symbol) const;
+    virtual void read(const fs::path& path) = 0;
 
-private:
-    std::unordered_map<hash_t, std::unique_ptr<MagicEntry>> m_entries;
+    virtual shared_entry_t query(std::string_view symbol) const        = 0;
+    virtual void   for_each(const for_each_callback_t& callback) const = 0;
+    virtual size_t count() const                                       = 0;
 
-    // MagicBlob uses a custom algorithm to transform the stored seed. When
-    // querying, you should use m_query_seed.
-    uint64_t m_stored_seed{};
-    uint64_t m_query_seed{};
+    static blob_t create(FormatVersion version);
+
+protected:
+    MagicBlob() = default;
 };
 
-} // namespace di::data_format
+} // namespace di::data_format

src/data_format/type/magic_entry.cpp

@@ -4,12 +4,9 @@
 
 namespace di {
 
-void to_json(nlohmann::json& json, const MagicEntry& entry) {
-    json["rva"]         = entry.rva;
-    json["is_function"] = entry.is_function();
-    json["_unk2"]       = entry._unk2();
-    json["is_verbose"]  = entry.is_verbose();
-    json["_unk4"]       = entry._unk4();
+void MagicEntry::to_json(nlohmann::json& json) const {
+    json["hash"] = hash;
+    json["rva"]  = rva;
 }
 
 } // namespace di

src/data_format/type/magic_entry.h

@@ -5,28 +5,12 @@
 namespace di {
 
 struct MagicEntry {
-#if __cpp_aggregate_paren_init < 201902L
-    MagicEntry(std::bitset<64> flags, rva_t rva) : flags(flags), rva(rva) {}
-#endif
+    hash_t hash; // key
+    rva_t  rva;  // value
 
-    std::bitset<64> flags;
-    // What is stored in the original format is not the RVA itself, but the
-    // difference with the previous entry (in MagicBlob, RVA is sorted from
-    // small to large)
-    // But here, we still store the "real" RVA.
-    rva_t rva;
-    // Do not put the original hash in the entry yet.
-    // hash_t hash;
+    MagicEntry(hash_t hash, rva_t rva) : hash(hash), rva(rva) {}
 
-    constexpr bool is_function() const { return flags[0]; }
-    constexpr bool _unk2() const { return flags[1]; }
-    constexpr bool is_verbose() const { return flags[2]; }
-    constexpr bool _unk4() const { return flags[3]; }
+    virtual void to_json(nlohmann::json& json) const;
 };
 
-void to_json(nlohmann::json& json, const MagicEntry& entry);
-
-// TODO
-// void from_json(const nlohmann::json& json, MagicEntry& entry);
-
 } // namespace di

src/tools/askrva/main.cpp

@@ -2,14 +2,15 @@
 #include "data_format/raw_text.h"
 #include "data_format/typed_symbol_list.h"
 
-#include <argparse/argparse.hpp>
-
 #if DI_USE_NATIVE_SYMBOL_RESOLVER
 #include <pl/SymbolProvider.h>
 #else
 #include "data_format/magic_blob.h"
 #endif
 
+#include <argparse/argparse.hpp>
+#include <magic_enum.hpp>
+
 using namespace di;
 using namespace di::data_format;
 
@@ -20,6 +21,7 @@ using namespace di::data_format;
         std::vector<std::string> m_input_paths;
         std::string              m_output_path;
         std::string              m_magic_blob_path;
+        std::string              m_format_version;
 
         std::optional<std::string> m_output_failed_path;
     } args;
@@ -36,6 +38,15 @@ using namespace di::data_format;
         .help("Path to magic blob (for builtin-symbol-resolver only).")
         .default_value("bedrock_runtime_data")
         .store_into(args.m_magic_blob_path);
+    
+    std::apply([&](auto&&... xs) {
+        program.add_argument("--preloader-version")
+            .help("Choose a compatible PreLoader version. (for builtin-symbol-resolver only).")
+            .choices(xs...)
+            .store_into(args.m_format_version)
+            .required();
+    }, magic_enum::enum_names<MagicBlob::FormatVersion>());
+
 
     program.add_argument("--output", "-o")
         .help("Path to output.")
@@ -75,10 +86,21 @@ int main(int argc, char* argv[]) try {
     );
 
 #if !DI_USE_NATIVE_SYMBOL_RESOLVER
-    MagicBlob magic_blob;
-    magic_blob.read(args.m_magic_blob_path);
+    auto format_version =
+        magic_enum::enum_cast<MagicBlob::FormatVersion>(args.m_format_version);
+    assert(format_version.has_value());
 
-    std::println("{} entries loaded from magicblob.", magic_blob.count());
+    auto magic_blob = MagicBlob::create(*format_version);
+    if (!magic_blob) {
+        std::println(
+            "Format version: {} is not yet implemented.",
+            args.m_format_version
+        );
+    }
+
+    magic_blob->read(args.m_magic_blob_path);
+
+    std::println("{} entries loaded from magicblob.", magic_blob->count());
 #endif
 
     symlist.for_each([&](const TypedSymbol& symbol) {
@@ -94,7 +116,7 @@ int main(int argc, char* argv[]) try {
         ));
         // TODO: imagebase...
 #else
-        if (auto entry = magic_blob.query(sym)) {
+        if (auto entry = magic_blob->query(sym)) {
             rva = entry->rva;
         }
 #endif

src/tools/blob-extractor/main.cpp

@@ -1,6 +1,7 @@
 #include "data_format/magic_blob.h"
 
 #include <argparse/argparse.hpp>
+#include <magic_enum.hpp>
 #include <nlohmann/json.hpp>
 
 using namespace di;
@@ -13,6 +14,8 @@ auto load_args(int argc, char* argv[]) {
     struct {
         std::string m_magic_blob_path;
         fs::path    m_output_path;
+
+        std::string m_format_version;
     } args;
 
     // clang-format off
@@ -26,6 +29,14 @@ auto load_args(int argc, char* argv[]) {
         .help("Path to output symlist.")
         .required();
 
+    std::apply([&](auto&&... xs) {
+        program.add_argument("--preloader-version")
+            .help("Choose a compatible PreLoader version. (for builtin-symbol-resolver only).")
+            .choices(xs...)
+            .store_into(args.m_format_version)
+            .required();
+    }, magic_enum::enum_names<MagicBlob::FormatVersion>());
+
     // clang-format on
 
     program.parse_args(argc, argv);
@@ -39,12 +50,23 @@ int main(int argc, char* argv[]) try {
 
     auto args = load_args(argc, argv);
 
-    MagicBlob blob;
-    blob.read(args.m_magic_blob_path);
+    auto format_version =
+        magic_enum::enum_cast<MagicBlob::FormatVersion>(args.m_format_version);
+    assert(format_version.has_value());
+
+    auto magic_blob = MagicBlob::create(*format_version);
+    if (!magic_blob) {
+        std::println(
+            "Format version: {} is not yet implemented.",
+            args.m_format_version
+        );
+    }
+
+    magic_blob->read(args.m_magic_blob_path);
 
     nlohmann::json data;
-    blob.for_each([&data](hash_t hash, const MagicEntry& entry) {
-        data[std::format("{:#x}", hash)] = entry;
+    magic_blob->for_each([&data](hash_t hash, MagicBlob::shared_entry_t entry) {
+        entry->to_json(data.emplace_back());
     });
 
     std::ofstream ofs(args.m_output_path);

xmake.lua

@@ -4,6 +4,7 @@ add_requires('argparse      3.2')
 add_requires('nlohmann_json 3.12.0')
 add_requires('xxhash        0.8.3')
 add_requires('libllvm       19.1.7')
+add_requires('magic_enum    0.9.7')
 add_requires('boost         1.88.0', {
     system = false, 
     configs = {
@@ -97,7 +98,10 @@ target('askrva')
     set_pcxxheader('src/pch.h')
 
     add_deps('libdi')
-    add_packages('argparse')
+    add_packages(
+        'argparse',
+        'magic_enum'
+    )
 
     if is_config('symbol-resolver', 'native') then
         add_packages('preloader')
@@ -112,7 +116,8 @@ target('blob-extractor')
     add_deps('libdi')
     add_packages(
         'argparse',
-        'nlohmann_json'
+        'nlohmann_json',
+        'magic_enum'
     )
 
 target('dumpsym')