From 6f112c2d16f0e31e636411d4280aedd7fec90826 Mon Sep 17 00:00:00 2001
From: WerWolv <werwolv98@gmail.com>
Date: Thu, 15 May 2025 22:00:26 +0200
Subject: [PATCH] feat: Added custom encoding row to data inspector

---
 plugins/builtin/romfs/lang/en_US.json         |  3 +
 .../builtin/source/content/data_inspector.cpp | 64 ++++++++++++++++++-
 2 files changed, 65 insertions(+), 2 deletions(-)

diff --git a/plugins/builtin/romfs/lang/en_US.json b/plugins/builtin/romfs/lang/en_US.json
index 8e0dd5cc5..74038cca1 100644
--- a/plugins/builtin/romfs/lang/en_US.json
+++ b/plugins/builtin/romfs/lang/en_US.json
@@ -79,6 +79,9 @@
         "hex.builtin.inspector.ascii": "ASCII Character",
         "hex.builtin.inspector.binary": "Binary (8 bit)",
         "hex.builtin.inspector.bool": "bool",
+        "hex.builtin.inspector.custom_encoding": "Custom Encoding",
+        "hex.builtin.inspector.custom_encoding.change": "Select encoding",
+        "hex.builtin.inspector.custom_encoding.no_encoding": "No encoding selected",
         "hex.builtin.inspector.dos_date": "DOS Date",
         "hex.builtin.inspector.dos_time": "DOS Time",
         "hex.builtin.inspector.double": "double (64 bit)",
diff --git a/plugins/builtin/source/content/data_inspector.cpp b/plugins/builtin/source/content/data_inspector.cpp
index 515805c53..65c94d8b3 100644
--- a/plugins/builtin/source/content/data_inspector.cpp
+++ b/plugins/builtin/source/content/data_inspector.cpp
@@ -14,7 +14,10 @@
 
 #include <imgui_internal.h>
 #include <fonts/vscode_icons.hpp>
+#include <hex/helpers/default_paths.hpp>
+#include <hex/helpers/encoding_file.hpp>
 #include <hex/ui/imgui_imhex_extensions.h>
+#include <popups/popup_file_chooser.hpp>
 
 namespace hex::plugin::builtin {
 
@@ -448,7 +451,7 @@ namespace hex::plugin::builtin {
                 u32 codepoint           = 0;
 
                 std::memcpy(utf8Buffer, reinterpret_cast<char8_t *>(buffer.data()), 4);
-                u8 codepointSize = ImTextCharFromUtf8(&codepoint, utf8Buffer, utf8Buffer + 4);
+                u8 codepointSize = ImTextCharFromUtf8(&codepoint, utf8Buffer, nullptr);
 
                 std::memcpy(codepointString, utf8Buffer, std::min(codepointSize, u8(4)));
                 auto value = hex::format("'{0}' (U+0x{1:04X})",
@@ -616,6 +619,63 @@ namespace hex::plugin::builtin {
             }
         );
 
+        ContentRegistry::DataInspector::add("hex.builtin.inspector.custom_encoding", 1, [encodingFile = EncodingFile()](const std::vector<u8> &, std::endian, Style) mutable {
+            std::string value, copyValue;
+
+            if (encodingFile.valid()) {
+                auto currSelection = ImHexApi::HexEditor::getSelection();
+
+                if (currSelection.has_value()) {
+                    std::vector<u8> stringBuffer(std::min<size_t>(currSelection->size, 0x1000), 0x00);
+                    ImHexApi::Provider::get()->read(currSelection->address, stringBuffer.data(), stringBuffer.size());
+
+                    u64 offset = 0;
+                    while (offset < stringBuffer.size()) {
+                        const auto [character, size] = encodingFile.getEncodingFor(std::span(stringBuffer).subspan(offset));
+                        value += character;
+                        offset += size;
+                    }
+                    copyValue = value;
+
+
+                    if (value.size() > MaxStringLength) {
+                        value.resize(MaxStringLength);
+                        value += "...";
+                    }
+                } else {
+                    value = "";
+                    copyValue = "";
+                }
+            } else {
+                value = "Invalid";
+            }
+            return [&, value, copyValue]() mutable -> std::string {
+                ContentRegistry::DataInspector::drawMenuItems([&] {
+                    if (ImGui::MenuItemEx("hex.builtin.inspector.custom_encoding.change"_lang, "あ")) {
+                        const auto basePaths = paths::Encodings.read();
+                        std::vector<std::fs::path> paths;
+                        for (const auto &basePath : basePaths) {
+                            for (const auto &entry : std::filesystem::directory_iterator(basePath)) {
+                                paths.push_back(entry.path());
+                            }
+                        }
+
+                        ui::PopupFileChooser::open(basePaths, paths, std::vector<fs::ItemFilter>{ {"Thingy Table File", "tbl"} }, false, [&](const auto &path) {
+                            encodingFile = EncodingFile(EncodingFile::Type::Thingy, path);
+                        });
+                    }
+                });
+
+                if (encodingFile.valid() && !value.empty()) {
+                    ImGuiExt::TextFormatted("({})\"{}\"", encodingFile.getName(), value);
+                } else {
+                    ImGuiExt::TextFormatted("hex.builtin.inspector.custom_encoding.no_encoding"_lang);
+                }
+
+                return copyValue;
+            };
+        });
+
 #if defined(OS_WINDOWS)
 
         ContentRegistry::DataInspector::add("hex.builtin.inspector.time32", sizeof(u32), [](auto buffer, auto endian, auto style) {
@@ -769,4 +829,4 @@ namespace hex::plugin::builtin {
     }
     // clang-format on
 
-}
\ No newline at end of file
+}