lots of improvements standby for the results

2026-09-27 01:09:57 -06:00
parent 812ac62393
commit 150e5e8636
16 changed files with 2726 additions and 9 deletions
@@ -0,0 +1,85 @@
+#pragma once
+
+#include <cstdint>
+#include <string_view>
+
+namespace xicu {
+
+class PropTable {
+public:
+    PropTable() = default;
+    explicit PropTable(const char* path);
+    ~PropTable();
+
+    PropTable(const PropTable&) = delete;
+    PropTable& operator=(const PropTable&) = delete;
+    PropTable(PropTable&& other) noexcept;
+    PropTable& operator=(PropTable&& other) noexcept;
+
+    bool load(const char* path);
+    bool isLoaded() const { return data_ != nullptr; }
+    std::string_view getName(uint32_t cp) const;
+    bool getPunctuation(uint32_t cp) const;
+    bool getLetter(uint32_t cp) const;
+    int32_t getUppercase_map(uint32_t cp) const;
+    int32_t getLowercase_map(uint32_t cp) const;
+    bool getWhitespace(uint32_t cp) const;
+    bool getSpace(uint32_t cp) const;
+    bool getDecimal(uint32_t cp) const;
+    int32_t getDecimal_val(uint32_t cp) const;
+
+private:
+    const uint8_t* data_ = nullptr;
+    size_t data_size_ = 0;
+    bool owns_data_ = false;
+
+    static uint32_t readVarint(const uint8_t*& ptr);
+    const uint8_t* findFieldData(const char* name, uint8_t& out_type, uint8_t& out_strat) const;
+    const uint8_t* findStringField(const char* name) const;
+
+    // Decoders
+    bool decodeBitmap(const uint8_t* data, uint32_t cp) const;
+    bool decodeDeltaRangesBool(const uint8_t* data, const uint8_t* data_end, uint32_t cp) const;
+    int32_t decodeDeltaRangesInt(const uint8_t* data, const uint8_t* data_end, uint32_t cp) const;
+    bool decodeSparseMapBool(const uint8_t* data, const uint8_t* data_end, uint32_t cp) const;
+    int32_t decodeSparseMapInt(const uint8_t* data, const uint8_t* data_end, uint32_t cp) const;
+};
+
+inline std::string_view getName(uint32_t cp) {
+    static PropTable table("../out/xicu.bin");
+    return table.getName(cp);
+}
+inline bool isPunctuation(uint32_t cp) {
+    static PropTable table("../out/xicu.bin");
+    return table.getPunctuation(cp);
+}
+inline bool isLetter(uint32_t cp) {
+    static PropTable table("../out/xicu.bin");
+    return table.getLetter(cp);
+}
+inline int32_t getUppercase_map(uint32_t cp) {
+    static PropTable table("../out/xicu.bin");
+    return table.getUppercase_map(cp);
+}
+inline int32_t getLowercase_map(uint32_t cp) {
+    static PropTable table("../out/xicu.bin");
+    return table.getLowercase_map(cp);
+}
+inline bool isWhitespace(uint32_t cp) {
+    static PropTable table("../out/xicu.bin");
+    return table.getWhitespace(cp);
+}
+inline bool isSpace(uint32_t cp) {
+    static PropTable table("../out/xicu.bin");
+    return table.getSpace(cp);
+}
+inline bool isDecimal(uint32_t cp) {
+    static PropTable table("../out/xicu.bin");
+    return table.getDecimal(cp);
+}
+inline int32_t getDecimal_val(uint32_t cp) {
+    static PropTable table("../out/xicu.bin");
+    return table.getDecimal_val(cp);
+}
+
+} // namespace xicu