feat: implement PDF page parsing, text extraction, and document modeling logic

engine/include/pdfengine/pdf_document.hpp

@@ -5,6 +5,7 @@
 #include <string>
 #include <vector>
 #include <cstdint>
+#include <array>
 
 namespace pdfengine {
 
@@ -198,6 +199,7 @@ public:
         std::string timestamp;
         int pageIndex = 0;
         std::vector<std::vector<Point2D>> paths;
+        std::vector<std::array<Point2D, 4>> quadPoints;
 
         std::string fieldName;
         std::string fieldValue;

engine/src/parser/pdfium_edit_pages.cpp

@@ -133,22 +133,19 @@ std::expected<void, EngineError> PdfiumDocument::applyOp_pageRotation(const nloh
 #endif
 }
 
-std::expected<void, EngineError> PdfiumDocument::applyOp_pageDeletion(const nlohmann::json& op, int pageIndex) {
 #ifdef PDFENGINE_WITH_PDFIUM
+std::expected<void, EngineError> PdfiumDocument::applyOp_pageDeletion(const nlohmann::json& op, int pageIndex) {
                 if (pageCount() <= 1) {
                     spdlog::error("Cannot delete the only page in the document");
                     return std::unexpected(EngineError::Unknown);
                 }
                 FPDFPage_Delete(doc_, pageIndex);
     return {};
-#else
-    (void)op; (void)pageIndex;
-    return std::unexpected(EngineError::Unknown);
-#endif
 }
+#endif
 
-std::expected<void, EngineError> PdfiumDocument::applyOp_pageReorder(const nlohmann::json& op, int pageIndex) {
 #ifdef PDFENGINE_WITH_PDFIUM
+std::expected<void, EngineError> PdfiumDocument::applyOp_pageReorder(const nlohmann::json& op, int pageIndex) {
                 if (!op.contains("data") || !op["data"].is_object()) {
                     spdlog::error("page_reorder operation missing 'data' object");
                     return std::unexpected(EngineError::InvalidFormat);
@@ -170,10 +167,7 @@ std::expected<void, EngineError> PdfiumDocument::applyOp_pageReorder(const nlohm
                     return std::unexpected(EngineError::Unknown);
                 }
     return {};
-#else
-    (void)op; (void)pageIndex;
-    return std::unexpected(EngineError::Unknown);
-#endif
 }
+#endif
 
-} 
+}

engine/src/parser/pdfium_page_model.cpp

@@ -614,6 +614,21 @@ std::expected<std::vector<PdfPage::AnnotationInfo>, EngineError> PdfiumPage::ext
                     }
                     if (!stroke.empty()) info.paths.push_back(std::move(stroke));
                 }
+            } else if (subtype == FPDF_ANNOT_HIGHLIGHT || subtype == FPDF_ANNOT_STRIKEOUT || subtype == FPDF_ANNOT_UNDERLINE || subtype == FPDF_ANNOT_SQUIGGLY) {
+                const double pageH = height();
+                size_t quadCount = FPDFAnnot_CountAttachmentPoints(annot);
+                for (size_t q = 0; q < quadCount; ++q) {
+                    FS_QUADPOINTSF quad;
+                    if (FPDFAnnot_GetAttachmentPoints(annot, q, &quad)) {
+                        std::array<Point2D, 4> pts = {{
+                            {static_cast<double>(quad.x1), pageH - static_cast<double>(quad.y1)},
+                            {static_cast<double>(quad.x2), pageH - static_cast<double>(quad.y2)},
+                            {static_cast<double>(quad.x3), pageH - static_cast<double>(quad.y3)},
+                            {static_cast<double>(quad.x4), pageH - static_cast<double>(quad.y4)}
+                        }};
+                        info.quadPoints.push_back(pts);
+                    }
+                }
             }
 
             result.push_back(info);