pdf/engine/include/pdfengine/pdf_document.hpp

#pragma once

#include <expected>
#include <memory>
#include <string>
#include <vector>
#include <cstdint>

namespace pdfengine {

enum class EngineError {
    FileNotFound,
    InvalidFormat,
    PasswordRequired,
    InvalidPassword,
    PageOutOfBounds,
    RenderFailed,
    WriteFailed,
    Unknown
};

struct DocumentMetadata {
    std::string title;
    std::string author;
    std::string creator;
    std::string producer;
    std::string creationDate;
    std::string modificationDate;
};

struct DocumentPermissions {
    bool isEncrypted = false;
    std::string encryption = "None";
    int securityRevision = -1;
    bool ownerUnlocked = false;

    bool canPrint = true;
    bool canPrintHighRes = true;
    bool canModify = true;
    bool canCopy = true;
    bool canAnnotate = true;
    bool canFillForms = true;
    bool canExtractForAccessibility = true;
    bool canAssemble = true;
};

struct PageImage {
    int width;
    int height;
    std::vector<uint8_t> data;
};

struct Point2D {
    double x;
    double y;
};

struct DevicePoint {
    int x;
    int y;
};

struct GlyphBounds {
    std::string text;
    double x;
    double y;
    double w;
    double h;
    double fontSize;
};

struct HitResult {
    int glyphIndex = -1;
    int caret = 0;
    int line = -1;
};

struct TextSelection {
    int startGlyph = 0;
    int endGlyph = 0;
    std::string text;
    std::vector<GlyphBounds> rects;
};

struct FontInfo {
    std::string fontName;
    std::string type;
    bool isEmbedded = false;
    bool isSubset = false;
    bool isVertical = false;

    std::string encoding;
    bool hasToUnicode = false;
    std::string cmapName;
    std::string cidSystemInfo;
    std::string subsetTag;
    std::string sourceType;
    std::string substitutedFrom;
    std::string substitutedTo;
    std::string normalizedFamily;
    std::string internalFontId;
    uint32_t flags = 0;
    double ascent = 0.0;
    double descent = 0.0;
    double capHeight = 0.0;
};

struct Glyph {
    std::string text;
    uint32_t unicode = 0;
    
    std::string fontName;
    uint32_t flags = 0;
    double fontSize = 0.0;

    double originX = 0.0; 
    double originY = 0.0; 
    double bboxX = 0.0, bboxY = 0.0, bboxW = 0.0, bboxH = 0.0;
    double angle = 0.0;
    int pageObjectIndex = -1;
    int srcIndex = 0;
};

struct TextRun {
    std::string text;
    std::string fontName;
    uint32_t flags = 0;
    double fontSize = 0.0;
    std::string internalFontId;
    bool isEmbedded = false;
    std::string type;
    std::vector<Glyph> glyphs;
    std::vector<int> objectIndices;
    double x = 0.0, y = 0.0, w = 0.0, h = 0.0;
    std::string fillColor = "#000000";
    std::string paraId;
    std::string fontFidelity = "exact";
};

struct TextLine {
    std::vector<TextRun> runs;
    std::vector<Glyph> glyphs;
    double angle = 0.0;
    double baselineY = 0.0;
    double x = 0.0, y = 0.0, w = 0.0, h = 0.0;
};

struct Paragraph {
    std::vector<TextLine> lines;
    double x = 0.0, y = 0.0, w = 0.0, h = 0.0;
};

struct PageModel {
    std::vector<Paragraph> paragraphs;
    double width = 0.0;
    double height = 0.0;
    int pageIndex = 0;
};

class PdfPage {
public:
    virtual ~PdfPage() = default;

    [[nodiscard]] virtual double width() const noexcept = 0;
    [[nodiscard]] virtual double height() const noexcept = 0;

    [[nodiscard]] virtual std::expected<PageImage, EngineError> render(int dpi = 96) const = 0;

    [[nodiscard]] virtual std::expected<PageImage, EngineError>
    renderRegionRaw(int dpi, double yTopPt, double heightPt) const {
        (void)dpi; (void)yTopPt; (void)heightPt;
        return std::unexpected(EngineError::Unknown);
    }

    [[nodiscard]] virtual std::expected<std::string, EngineError> extractText() const = 0;

    [[nodiscard]] virtual std::expected<std::vector<GlyphBounds>, EngineError> extractTextWithBounds() const = 0;

    [[nodiscard]] std::expected<std::vector<GlyphBounds>, EngineError> orderedGlyphs() const;

    [[nodiscard]] std::expected<HitResult, EngineError> hitGlyph(double x, double y) const;

    [[nodiscard]] std::expected<TextSelection, EngineError>
    selectRange(double ax, double ay, double bx, double by) const;

    [[nodiscard]] virtual std::expected<PageModel, EngineError> extractDocumentModel() const = 0;

    [[nodiscard]] virtual std::expected<std::vector<FontInfo>, EngineError> getFonts() const = 0;
    [[nodiscard]] virtual std::expected<std::vector<std::string>, EngineError> extractAnnotationsText() const = 0;
    
    struct AnnotationInfo {
        std::string id;
        std::string type;
        double x = 0.0, y = 0.0, width = 0.0, height = 0.0;
        std::string color;
        std::string author;
        std::string content;
        std::string timestamp;
        int pageIndex = 0;
        std::vector<std::vector<Point2D>> paths;

        std::string fieldName;
        std::string fieldValue;
        std::string fieldType;
        int fieldFlags = 0;
        std::vector<std::string> fieldOptions;
    };
    [[nodiscard]] virtual std::expected<std::vector<AnnotationInfo>, EngineError> extractAnnotations() const = 0;

    [[nodiscard]] virtual std::expected<double, EngineError> getGlyphWidth(const std::string& fontName, uint32_t charcode, double fontSize) const = 0;

    [[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; }

class PdfDocument {
public:
    virtual ~PdfDocument() = default;

    [[nodiscard]] static std::expected<std::shared_ptr<PdfDocument>, EngineError> 
    loadFromFile(const std::string& path, const std::string& password = "");

    [[nodiscard]] static std::expected<std::shared_ptr<PdfDocument>, EngineError> 
    loadFromMemory(const std::vector<uint8_t>& data, const std::string& password = "");

    [[nodiscard]] virtual int pageCount() const noexcept = 0;
    [[nodiscard]] virtual DocumentMetadata metadata() const noexcept = 0;
    [[nodiscard]] virtual DocumentPermissions permissions() const noexcept = 0;

    struct OutlineItem {
        std::string title;
        int pageIndex = -1;
        int level = 0;
    };
    [[nodiscard]] virtual std::expected<std::vector<OutlineItem>, EngineError> extractOutline() const = 0;

    [[nodiscard]] virtual std::expected<std::shared_ptr<PdfPage>, EngineError>
    getPage(int pageIndex) = 0;

    [[nodiscard]] virtual std::expected<std::vector<FontInfo>, EngineError> 
    getFonts(int startPage = 0, int endPage = -1) const = 0;

    [[nodiscard]] virtual std::expected<std::vector<uint8_t>, EngineError>
    getFontData(const std::string& internalFontId) const = 0;

    [[nodiscard]] virtual std::expected<std::vector<uint8_t>, EngineError>
    getReconstructedFontData(const std::string& internalFontId) { (void)internalFontId; return std::unexpected(EngineError::Unknown); }

    virtual void registerAuxFont(const std::string& internalFontId, const std::vector<uint8_t>& sfnt) { (void)internalFontId; (void)sfnt; }

    [[nodiscard]] virtual std::expected<std::shared_ptr<fonts::pdf_fonts::Font>, std::string>
    getResolvedFont(const FontInfo& fontInfo) = 0;

    virtual std::expected<void, EngineError> applyEdits(const std::string& editsJson) = 0;

    [[nodiscard]] virtual std::string lastReflowLayout() const { return {}; }

    [[nodiscard]] virtual bool lastReflowOverflowed() const { return false; }

    [[nodiscard]] virtual std::expected<std::vector<uint8_t>, EngineError> 
    saveIncremental() const = 0;

    [[nodiscard]] virtual std::expected<std::vector<uint8_t>, EngineError>
    saveFull() const = 0;

    [[nodiscard]] virtual std::expected<std::vector<uint8_t>, EngineError>
    saveFullForExport() const { return saveFull(); }
};

}