fdt: working on more type driven parser/tokenizer

Signed-off-by: Bartłomiej Burdukiewicz <[email protected]>
dev-0x7C6 · Jun 28, 2024 · be9a3ee · be9a3ee
1 parent 2219b33
commit be9a3ee
Show file tree

Hide file tree

Showing 3 changed files with 170 additions and 21 deletions.
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
@@ -2,28 +2,12 @@ configure_file(config.h.in config.h)
 
 add_subdirectory("submodules/qhexview")
 
+file(GLOB_RECURSE fdt-viewer-headers *.hpp)
+file(GLOB_RECURSE fdt-viewer-sources *.cpp)
+
 add_executable(fdt-viewer
-    dialogs.cpp
-    dialogs.hpp
-    endian-conversions.hpp
-    fdt/fdt-generator-qt.cpp
-    fdt/fdt-generator-qt.hpp
-    fdt/fdt-generator.hpp
-    fdt/fdt-header.hpp
-    fdt/fdt-parser.cpp
-    fdt/fdt-parser.hpp
-    fdt/fdt-property-types.hpp
-    fdt/fdt-view.cpp
-    fdt/fdt-view.hpp
-    main-window.cpp
-    main-window.hpp
-    main-window.ui
-    main.cpp
-    menu-manager.cpp
-    menu-manager.hpp
-    types.hpp
-    viewer-settings.cpp
-    viewer-settings.hpp
+    ${fdt-viewer-headers}
+    ${fdt-viewer-sources}
     ../resources.qrc
 )
 

diff --git a/src/fdt/fdt-parser-v2.hpp b/src/fdt/fdt-parser-v2.hpp
@@ -0,0 +1,47 @@
+#pragma once
+
+#include <type_traits>
+#include <variant>
+#include <string_view>
+#include <vector>
+#include <cstdint>
+
+using u32 = std::uint32_t;
+
+namespace fdt::tokenizer {
+
+namespace types {
+struct node_begin {
+    std::string_view name;
+};
+struct node_end {};
+struct property {
+    std::string_view name;
+    std::string_view data;
+};
+struct nop {};
+struct end {};
+
+constexpr auto id_of(node_begin) -> u32 { return 0x01; };
+constexpr auto id_of(node_end) -> u32 { return 0x02; };
+constexpr auto id_of(property) -> u32 { return 0x03; };
+constexpr auto id_of(nop) -> u32 { return 0x04; };
+constexpr auto id_of(end) -> u32 { return 0x09; };
+
+} // namespace types
+
+using token = std::variant<types::node_begin, types::node_end, types::property, types::nop, types::end>;
+using token_list = std::vector<token>;
+
+struct context {
+    std::string_view structs;
+    std::string_view strings;
+    token_list tokens;
+};
+
+template <typename T>
+concept Tokenizable = requires(T t) {
+    { fdt::tokenizer::types::id_of(t) } -> std::convertible_to<u32>;
+};
+
+} // namespace fdt::tokenizer
diff --git a/src/fdt/fdt-parser.cpp b/src/fdt/fdt-parser.cpp
@@ -1,6 +1,11 @@
 #include "fdt-parser.hpp"
+#include <ostream>
+#include <string_view>
+#include <variant>
 
 #include <cstring>
+#include <concepts>
+#include <algorithm>
 #include <endian-conversions.hpp>
 
 fdt_parser::fdt_parser(const char *data, u64 size, iface_fdt_generator &generator, const QString &default_root_node, const std::vector<fdt_handle_special_property> &handle_special_properties)
@@ -24,6 +29,7 @@ fdt_parser::fdt_parser(const char *data, u64 size, iface_fdt_generator &generato
     }
 }
 
+/*
 void fdt_parser::parse(const fdt::header header, iface_fdt_generator &generator) {
     const auto dt_struct = m_data + header.off_dt_struct;
     const auto dt_strings = m_data + header.off_dt_strings;
@@ -75,3 +81,115 @@ void fdt_parser::parse(const fdt::header header, iface_fdt_generator &generator)
             break;
     }
 }
+*/
+
+#include "fdt-parser-v2.hpp"
+#include <iostream>
+#include <algorithm>
+
+auto align(const std::size_t size) {
+    const auto q = size % sizeof(u32);
+    const auto w = size / sizeof(u32);
+    return std::max<u32>(1u, w + (q ? 1u : 0u));
+};
+
+namespace fdt::parser {
+
+using namespace fdt::tokenizer;
+using namespace fdt::tokenizer::types;
+
+auto parse(node_begin &&token, const char *data, int &skip, context &ctx) -> fdt::tokenizer::token {
+    const auto size = std::strlen(data);
+    token.name = std::string_view(data, size);
+    skip += align(size);
+    return {std::move(token)};
+}
+
+auto parse(node_end &&token, const char *data, int &skip, context &ctx) -> fdt::tokenizer::token {
+    return {std::move(token)};
+}
+
+auto parse(types::property &&token, const char *data, int &skip, context &ctx) -> fdt::tokenizer::token {
+    const auto header = read_data_32be<fdt::property>(data);
+    skip += align(sizeof(header)) + align(header.len);
+    data += sizeof(header);
+
+    token.name = std::string_view();
+    token.data = std::string_view(data, header.len);
+
+    return {std::move(token)};
+}
+
+auto parse(nop &&token, const char *data, int &skip, context &ctx) -> fdt::tokenizer::token {
+    return {std::move(token)};
+}
+
+auto parse(end &&token, const char *data, int &skip, context &ctx) -> fdt::tokenizer::token {
+    return {std::move(token)};
+}
+} // namespace fdt::parser
+
+template <fdt::tokenizer::Tokenizable... Ts>
+auto foreach_token_type(std::variant<Ts...>, const u32 token_id, const char *data, int &skip, fdt::tokenizer::context &ctx) {
+    auto conditional_parse = [&](auto &&token) {
+        if (fdt::tokenizer::types::id_of(token) == token_id) {
+            ctx.tokens.emplace_back(fdt::parser::parse(std::move(token), data, skip, ctx));
+            return true;
+        }
+        return false;
+    };
+    return (conditional_parse(Ts{}) || ...);
+}
+
+void fdt_parser::parse(const fdt::header header, iface_fdt_generator &generator) {
+    const auto dt_struct = m_data + header.off_dt_struct;
+    const auto dt_strings = m_data + header.off_dt_strings;
+
+    auto get_property_name = [&](auto offset) {
+        const auto ptr = dt_strings + offset;
+        return QString::fromUtf8(ptr, std::strlen(ptr));
+    };
+
+    using namespace fdt::tokenizer;
+    using namespace fdt::tokenizer::types;
+    context ctx;
+
+    auto begin = reinterpret_cast<const u32 *>(dt_struct);
+    auto end = reinterpret_cast<const u32 *>(dt_struct) + header.size_dt_struct / sizeof(u32);
+
+    for (auto iter = begin; iter != end;) {
+        const auto id = static_cast<u32>(convert(*iter));
+        iter++;
+        std::cout << id << std::endl;
+
+        auto skip = 0;
+        foreach_token_type(token{}, id, reinterpret_cast<const char *>(iter), skip, ctx);
+        iter += skip;
+    }
+
+    const auto node_begin_count = std::ranges::count_if(ctx.tokens, [](auto &&v) {
+        return std::holds_alternative<node_begin>(v);
+    });
+
+    const auto node_end_count = std::ranges::count_if(ctx.tokens, [](auto &&v) {
+        return std::holds_alternative<node_end>(v);
+    });
+
+    const auto property_count = std::ranges::count_if(ctx.tokens, [](auto &&v) {
+        return std::holds_alternative<property>(v);
+    });
+
+    const auto nop_count = std::ranges::count_if(ctx.tokens, [](auto &&v) {
+        return std::holds_alternative<nop>(v);
+    });
+
+    const auto end_count = std::ranges::count_if(ctx.tokens, [](auto &&v) {
+        return std::holds_alternative<types::end>(v);
+    });
+
+    std::cout << "node begin  count: " << node_begin_count << std::endl;
+    std::cout << "node end    count: " << node_end_count << std::endl;
+    std::cout << "property    count: " << property_count << std::endl;
+    std::cout << "nop         count: " << nop_count << std::endl;
+    std::cout << "end         count: " << end_count << std::endl;
+}