create the api for the display list

engine/include/pdfengine/pdf_document.hpp

@@ -209,6 +209,9 @@ public:
 
     [[nodiscard]] virtual DevicePoint pageToDevice(const Point2D& pagePoint, int deviceWidth, int deviceHeight, int rotate = 0) const noexcept = 0;
     [[nodiscard]] virtual Point2D deviceToPage(const DevicePoint& devicePoint, int deviceWidth, int deviceHeight, int rotate = 0) const noexcept = 0;
+
+    [[nodiscard]] virtual std::expected<std::string, EngineError> extractDisplayListJson() const = 0;
+    [[nodiscard]] virtual std::expected<std::vector<uint8_t>, EngineError> extractImageXObject(const std::string& name) const = 0;
 };
 
 namespace fonts::pdf_fonts { class Font; }

engine/src/parser/pdfium_document.cpp

@@ -19,6 +19,9 @@
 #include "fonts/shaping/hb_shaper.hpp"
 #include "fonts/face/font_face.hpp"
 #include "decoration_builder.hpp"
+#include "qpdf/qpdf_extractor.hpp"
+#include "parser/lexer.hpp"
+#include "parser/parser.hpp"
 
 #include <nlohmann/json.hpp>
 #include <spdlog/spdlog.h>
@@ -4328,6 +4331,51 @@ std::expected<std::shared_ptr<fonts::pdf_fonts::Font>, std::string> PdfiumDocume
 #endif
 }
 
+std::expected<std::string, EngineError> PdfiumPage::extractDisplayListJson() const {
+#ifdef PDFENGINE_WITH_PDFIUM
+    if (!ownerDoc_) return std::unexpected(EngineError::Unknown);
+    const auto& buffer = ownerDoc_->getMemoryBuffer();
+    if (buffer.empty()) return std::unexpected(EngineError::FileNotFound);
+
+    qpdf_layer::QpdfExtractor extractor;
+    auto stream = extractor.extractPageStreamFromMemory(buffer, pageIndex_);
+    if (!stream) return std::unexpected(EngineError::InvalidFormat);
+
+    Lexer lexer(stream->decodedContent);
+    auto tokens = lexer.tokenize();
+    ContentParser parser(tokens);
+    auto operations = parser.parse();
+
+    nlohmann::json j_array = nlohmann::json::array();
+    for (const auto& op : operations) {
+        nlohmann::json j_op = nlohmann::json::object();
+        j_op["op"] = op.op;
+        if (!op.operands.empty()) {
+            nlohmann::json j_args = nlohmann::json::array();
+            for (const auto& arg : op.operands) {
+                if (arg->type == AstNodeType::Number) {
+                    j_args.push_back(arg->numberValue);
+                } else if (arg->type == AstNodeType::Name || arg->type == AstNodeType::String) {
+                    j_args.push_back(arg->stringValue);
+                } else {
+                    j_args.push_back("<other>");
+                }
+            }
+            j_op["args"] = j_args;
+        }
+        j_array.push_back(j_op);
+    }
+    return j_array.dump();
+#else
+    return std::unexpected(EngineError::Unknown);
+#endif
+}
+
+std::expected<std::vector<uint8_t>, EngineError> PdfiumPage::extractImageXObject(const std::string& name) const {
+    (void)name;
+    return std::unexpected(EngineError::Unknown); // Stub for now
+}
+
 void PdfiumDocument::invalidateCaches() {
     {
         std::lock_guard<std::mutex> lock(fontsMutex_);

engine/src/parser/pdfium_document.hpp

@@ -55,6 +55,9 @@ public:
 
     std::expected<double, EngineError> getGlyphWidth(const std::string& fontName, uint32_t charcode, double fontSize) const override;
 
+    std::expected<std::string, EngineError> extractDisplayListJson() const override;
+    std::expected<std::vector<uint8_t>, EngineError> extractImageXObject(const std::string& name) const override;
+
     DevicePoint pageToDevice(const Point2D& pagePoint, int deviceWidth, int deviceHeight, int rotate = 0) const noexcept override;
     Point2D deviceToPage(const DevicePoint& devicePoint, int deviceWidth, int deviceHeight, int rotate = 0) const noexcept override;
 
@@ -102,6 +105,8 @@ public:
     std::string lastReflowLayout() const { return lastReflowLayout_; }
     bool lastReflowOverflowed() const { return lastReflowOverflowed_; }
 
+    const std::vector<uint8_t>& getMemoryBuffer() const { return memoryBuffer_; }
+
 private:
     mutable std::string lastReflowLayout_;
     mutable bool lastReflowOverflowed_ = false;