@cj-tech-master/excelts 9.1.0 → 9.2.0

package/dist/cjs/modules/pdf/builder/pdf-editor.js

@@ -0,0 +1,1612 @@
+"use strict";
+/**
+ * PDF editor — modify existing PDF documents.
+ *
+ * Supports:
+ * - Adding new pages with free-form content
+ * - Adding text/shapes/images to existing pages (overlay)
+ * - Filling form fields (AcroForm)
+ * - Copying pages from other PDFs (merge)
+ * - Preserving page properties (Rotate, CropBox, etc.) and metadata
+ *
+ * Note: save() rebuilds the PDF from scratch rather than using incremental
+ * updates. This is simpler and more reliable but means object numbers change
+ * and existing digital signatures will be invalidated.
+ *
+ * @example Edit an existing PDF:
+ * ```typescript
+ * import { PdfEditor } from "@cj-tech-master/excelts/pdf";
+ *
+ * const editor = PdfEditor.load(existingPdfBytes);
+ * editor.getPage(0).drawText("APPROVED", { x: 200, y: 400, fontSize: 48, color: { r: 0, g: 0.5, b: 0 } });
+ * editor.setFormField("name", "John Doe");
+ * const result = await editor.save();
+ * ```
+ */
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.PdfEditor = exports.PdfEditorPage = void 0;
+const pdf_document_1 = require("../reader/pdf-document");
+const pdf_writer_1 = require("../core/pdf-writer");
+const pdf_object_1 = require("../core/pdf-object");
+const font_manager_1 = require("../font/font-manager");
+const ttf_parser_1 = require("../font/ttf-parser");
+const pdf_decrypt_1 = require("../reader/pdf-decrypt");
+const form_extractor_1 = require("../reader/form-extractor");
+const metadata_reader_1 = require("../reader/metadata-reader");
+const pdf_parser_1 = require("../reader/pdf-parser");
+const document_builder_1 = require("./document-builder");
+const errors_1 = require("../errors");
+const image_utils_1 = require("./image-utils");
+const form_appearance_1 = require("./form-appearance");
+const resource_merger_1 = require("./resource-merger");
+// =============================================================================
+// PdfEditorPage
+// =============================================================================
+/**
+ * Proxy for an existing page that allows overlaying new content.
+ * New content is drawn on top of existing content via a separate content stream.
+ */
+class PdfEditorPage {
+    /** @internal */
+    constructor(pageIndex, width, height, fontManager) {
+        this._pageIndex = pageIndex;
+        this._width = width;
+        this._height = height;
+        this._overlay = new document_builder_1.PdfPageBuilder(width, height, fontManager);
+    }
+    /** Page width in points. */
+    get width() {
+        return this._width;
+    }
+    /** Page height in points. */
+    get height() {
+        return this._height;
+    }
+    /**
+     * Measure the width of a text string in points.
+     */
+    measureText(text, options) {
+        return this._overlay.measureText(text, options);
+    }
+    /**
+     * Draw text on this existing page (overlaid on top).
+     */
+    drawText(text, options) {
+        this._overlay.drawText(text, options);
+        return this;
+    }
+    /**
+     * Draw a rectangle on this existing page.
+     */
+    drawRect(options) {
+        this._overlay.drawRect(options);
+        return this;
+    }
+    /**
+     * Draw a circle on this existing page.
+     */
+    drawCircle(options) {
+        this._overlay.drawCircle(options);
+        return this;
+    }
+    /**
+     * Draw an ellipse on this existing page.
+     */
+    drawEllipse(options) {
+        this._overlay.drawEllipse(options);
+        return this;
+    }
+    /**
+     * Draw a line on this existing page.
+     */
+    drawLine(options) {
+        this._overlay.drawLine(options);
+        return this;
+    }
+    /**
+     * Draw an image on this existing page.
+     */
+    drawImage(options) {
+        this._overlay.drawImage(options);
+        return this;
+    }
+    /**
+     * Get the raw overlay content stream.
+     */
+    getContentStream() {
+        return this._overlay._stream;
+    }
+    /**
+     * Add an annotation to this existing page (Highlight, Text, FreeText, Stamp, etc.).
+     */
+    addAnnotation(options) {
+        this._overlay.addAnnotation(options);
+        return this;
+    }
+    /**
+     * Add a form field to this existing page.
+     */
+    addFormField(options) {
+        this._overlay.addFormField(options);
+        return this;
+    }
+    /**
+     * Draw an SVG path on this existing page.
+     */
+    drawSvgPath(d, options) {
+        this._overlay.drawSvgPath(d, options);
+        return this;
+    }
+    /**
+     * Draw a complex path from a list of path operations.
+     */
+    drawPath(ops, options) {
+        this._overlay.drawPath(ops, options);
+        return this;
+    }
+    /** @internal */
+    _hasOverlay() {
+        return (this._overlay._stream.toString().length > 0 ||
+            this._overlay._images.length > 0 ||
+            this._overlay._builderAnnotations.length > 0 ||
+            this._overlay._formFields.length > 0);
+    }
+}
+exports.PdfEditorPage = PdfEditorPage;
+// =============================================================================
+// PdfEditor
+// =============================================================================
+/**
+ * Editor for modifying existing PDF documents.
+ *
+ * Load an existing PDF, overlay content on existing pages, fill form fields,
+ * add new pages, copy pages from other documents, and save.
+ */
+class PdfEditor {
+    constructor(data, password) {
+        this._pages = [];
+        this._newPages = [];
+        this._fontManager = new font_manager_1.FontManager();
+        this._formFieldUpdates = new Map();
+        this._copiedPages = [];
+        /** @internal - Indices of original pages to remove on save */
+        this._removedPageIndices = new Set();
+        /** @internal - True during saveIncremental() to preserve original refs */
+        this._isIncrementalSave = false;
+        /** @internal - Rotation overrides for original pages: index → degrees (0/90/180/270) */
+        this._rotationOverrides = new Map();
+        this._signaturePlaceholder = null;
+        /** @internal - Writer reference during save(), for deep-clone */
+        this._writerForSave = null;
+        /** @internal - Cache of cloned indirect refs: "objNum:gen" → new objNum in writer */
+        this._clonedRefs = new Map();
+        this._doc = new pdf_document_1.PdfDocument(data);
+        this._password = password;
+        // Handle encryption
+        if ((0, pdf_decrypt_1.isEncrypted)(this._doc)) {
+            const success = (0, pdf_decrypt_1.initDecryption)(this._doc, password);
+            if (!success) {
+                throw new errors_1.PdfStructureError("Failed to decrypt PDF: incorrect password");
+            }
+        }
+        // Initialize page proxies
+        const pagesInfo = this._doc.getPagesWithObjInfo();
+        for (let i = 0; i < pagesInfo.length; i++) {
+            const { dict } = pagesInfo[i];
+            const dims = this._doc.resolvePageBox(dict) ?? {
+                width: 612,
+                height: 792
+            };
+            this._pages.push(new PdfEditorPage(i, dims.width, dims.height, this._fontManager));
+        }
+    }
+    /**
+     * Load a PDF for editing.
+     *
+     * @param data - Raw PDF file bytes
+     * @param options - Load options (e.g., password)
+     * @returns A PdfEditor instance
+     */
+    static load(data, options) {
+        return new PdfEditor(data, options?.password ?? "");
+    }
+    /** Number of existing pages. */
+    get pageCount() {
+        return this._pages.length;
+    }
+    /**
+     * Get an existing page for editing (overlaying content).
+     *
+     * @param index - 0-based page index
+     */
+    getPage(index) {
+        if (index < 0 || index >= this._pages.length) {
+            throw new errors_1.PdfStructureError(`Page index ${index} out of range (0-${this._pages.length - 1})`);
+        }
+        return this._pages[index];
+    }
+    /**
+     * Add a new blank page to the end of the document.
+     */
+    addPage(options) {
+        const width = options?.width ?? 595.28;
+        const height = options?.height ?? 841.89;
+        const page = new document_builder_1.PdfPageBuilder(width, height, this._fontManager);
+        this._newPages.push(page);
+        return page;
+    }
+    /**
+     * Remove a page from the document.
+     *
+     * @param index - 0-based page index (of original pages only)
+     */
+    removePage(index) {
+        if (index < 0 || index >= this._pages.length) {
+            throw new errors_1.PdfStructureError(`Page index ${index} out of range (0-${this._pages.length - 1})`);
+        }
+        this._removedPageIndices.add(index);
+        return this;
+    }
+    /**
+     * Set the rotation of an existing page.
+     *
+     * @param index - 0-based page index (of original pages only)
+     * @param degrees - Rotation in degrees (must be 0, 90, 180, or 270)
+     */
+    rotatePage(index, degrees) {
+        if (index < 0 || index >= this._pages.length) {
+            throw new errors_1.PdfStructureError(`Page index ${index} out of range (0-${this._pages.length - 1})`);
+        }
+        if (degrees !== 0 && degrees !== 90 && degrees !== 180 && degrees !== 270) {
+            throw new errors_1.PdfStructureError(`Invalid rotation ${degrees}: must be 0, 90, 180, or 270`);
+        }
+        this._rotationOverrides.set(index, degrees);
+        return this;
+    }
+    /**
+     * Split the document: save each page (or a subset) as a separate PDF.
+     *
+     * @param pageIndices - 0-based page indices to extract. Omit to split all pages.
+     * @returns Array of Uint8Array, one per requested page.
+     */
+    async splitPages(pageIndices) {
+        const pagesInfo = this._doc.getPagesWithObjInfo();
+        const indices = pageIndices ?? Array.from({ length: pagesInfo.length }, (_, i) => i);
+        const results = [];
+        for (const idx of indices) {
+            if (idx < 0 || idx >= pagesInfo.length) {
+                continue;
+            }
+            // Create a single-page editor from original bytes and save it
+            const singlePageEditor = PdfEditor.load(this._doc.data, {
+                password: this._password
+            });
+            // Remove all pages except the one we want
+            for (let i = 0; i < pagesInfo.length; i++) {
+                if (i !== idx) {
+                    singlePageEditor.removePage(i);
+                }
+            }
+            results.push(await singlePageEditor.save());
+        }
+        return results;
+    }
+    /**
+     * Embed a TrueType font for Unicode/CJK support.
+     */
+    embedFont(fontBytes) {
+        const ttfFont = (0, ttf_parser_1.parseTtf)(fontBytes);
+        this._fontManager.registerEmbeddedFont(ttfFont);
+        return this;
+    }
+    // ===========================================================================
+    // Form Fields
+    // ===========================================================================
+    /**
+     * Set the value of a form field.
+     * The field is identified by its fully qualified name (e.g., "form.address.city").
+     *
+     * @param fieldName - Fully qualified field name
+     * @param value - New value to set
+     */
+    setFormField(fieldName, value) {
+        this._formFieldUpdates.set(fieldName, value);
+        return this;
+    }
+    /**
+     * Set multiple form field values at once.
+     *
+     * @param fields - Object mapping field names to values
+     */
+    setFormFields(fields) {
+        for (const [name, value] of Object.entries(fields)) {
+            this._formFieldUpdates.set(name, value);
+        }
+        return this;
+    }
+    /**
+     * Get current form fields (before any modifications).
+     */
+    getFormFields() {
+        return (0, form_extractor_1.extractFormFields)(this._doc);
+    }
+    // ===========================================================================
+    // Page Copy / Merge
+    // ===========================================================================
+    /**
+     * Copy pages from another PDF document into this document.
+     *
+     * @param sourcePdf - Raw bytes of the source PDF
+     * @param pageIndices - 0-based page indices to copy. Omit to copy all pages.
+     * @param options - Load options for the source PDF (e.g., password)
+     */
+    copyPagesFrom(sourcePdf, pageIndices, options) {
+        const sourceDoc = new pdf_document_1.PdfDocument(sourcePdf);
+        if ((0, pdf_decrypt_1.isEncrypted)(sourceDoc)) {
+            const success = (0, pdf_decrypt_1.initDecryption)(sourceDoc, options?.password ?? "");
+            if (!success) {
+                throw new errors_1.PdfStructureError("Failed to decrypt source PDF for page copy");
+            }
+        }
+        const sourcePagesInfo = sourceDoc.getPagesWithObjInfo();
+        const indices = pageIndices ?? Array.from({ length: sourcePagesInfo.length }, (_, i) => i);
+        for (const idx of indices) {
+            if (idx < 0 || idx >= sourcePagesInfo.length) {
+                continue;
+            }
+            const { dict: pageDict } = sourcePagesInfo[idx];
+            const dims = sourceDoc.resolvePageBox(pageDict) ?? {
+                width: 612,
+                height: 792
+            };
+            // Collect all content streams from the source page
+            const contentStreams = this._collectContentStreams(sourceDoc, pageDict);
+            this._copiedPages.push({
+                width: dims.width,
+                height: dims.height,
+                contentStreams,
+                sourceDoc,
+                sourcePageDict: pageDict
+            });
+        }
+        return this;
+    }
+    // ===========================================================================
+    // Save
+    // ===========================================================================
+    /**
+     * Save the modified PDF.
+     *
+     * Rebuilds the PDF from scratch — content streams, resources, and page
+     * properties are deep-cloned into a new document. Original metadata and
+     * XMP streams are preserved. Digital signatures will be invalidated.
+     *
+     * @returns The modified PDF as Uint8Array
+     */
+    async save() {
+        // Rebuild the PDF (not incremental update — simpler and more reliable)
+        const writer = new pdf_writer_1.PdfWriter();
+        this._writerForSave = writer;
+        this._clonedRefs = new Map();
+        try {
+            return await this._buildFullSave(writer);
+        }
+        finally {
+            this._writerForSave = null;
+            this._clonedRefs.clear();
+        }
+    }
+    /** @internal Full rebuild implementation, extracted for try/finally cleanup. */
+    async _buildFullSave(writer) {
+        // Write font resources for any overlay content
+        const fontObjectMap = this._fontManager.writeFontResources(writer);
+        const fontDictStr = this._fontManager.buildFontDictString(fontObjectMap);
+        const pagesTreeObjNum = writer.allocObject();
+        const pageObjNums = [];
+        // Re-emit existing pages
+        const pagesInfo = this._doc.getPagesWithObjInfo();
+        for (let i = 0; i < pagesInfo.length; i++) {
+            // Skip removed pages
+            if (this._removedPageIndices.has(i)) {
+                continue;
+            }
+            const { dict: pageDict } = pagesInfo[i];
+            const dims = this._doc.resolvePageBox(pageDict) ?? {
+                width: 612,
+                height: 792
+            };
+            const editorPage = this._pages[i];
+            // Get original content streams
+            const originalStreams = this._collectContentStreams(this._doc, pageDict);
+            const originalResources = this._serializeResources(this._doc, pageDict);
+            // Combine original + overlay content
+            const allContentRefs = [];
+            // Write original content streams
+            for (const streamData of originalStreams) {
+                const objNum = writer.allocObject();
+                writer.addStreamObject(objNum, new pdf_object_1.PdfDict(), streamData);
+                allContentRefs.push(objNum);
+            }
+            // Write overlay content stream if present
+            let overlayResourcesStr = "";
+            if (editorPage._hasOverlay()) {
+                const overlayObjNum = writer.allocObject();
+                writer.addStreamObject(overlayObjNum, new pdf_object_1.PdfDict(), editorPage._overlay._stream);
+                allContentRefs.push(overlayObjNum);
+                // Build overlay-specific resources (fonts, images) as structured dict
+                const imageXObjectMap = this._writeOverlayImages(writer, editorPage._overlay);
+                const overlayDict = this._buildOverlayResourceDict(fontDictStr, imageXObjectMap);
+                overlayResourcesStr = (0, resource_merger_1.serializeResourceDict)(overlayDict);
+            }
+            // Write resources (merge original + overlay)
+            const resourcesObjNum = writer.allocObject();
+            const mergedResources = this._mergeResourceStrings(originalResources, overlayResourcesStr);
+            writer.addObject(resourcesObjNum, mergedResources || "<< >>");
+            // Write page dict
+            const contentsStr = allContentRefs.length === 1
+                ? (0, pdf_object_1.pdfRef)(allContentRefs[0])
+                : `[${allContentRefs.map(r => (0, pdf_object_1.pdfRef)(r)).join(" ")}]`;
+            // Apply form field updates to annotations
+            const annotRefs = this._writeFormFieldUpdates(writer, pageDict, i);
+            // Write overlay builder annotations (Highlight, Text, FreeText, Stamp, etc.)
+            for (const annot of editorPage._overlay._builderAnnotations) {
+                const annotObjNum = writer.allocObject();
+                const rect = `[${(0, pdf_object_1.pdfNumber)(annot.rect[0])} ${(0, pdf_object_1.pdfNumber)(annot.rect[1])} ${(0, pdf_object_1.pdfNumber)(annot.rect[2])} ${(0, pdf_object_1.pdfNumber)(annot.rect[3])}]`;
+                const annotDict = new pdf_object_1.PdfDict()
+                    .set("Type", "/Annot")
+                    .set("Subtype", `/${annot.subtype}`)
+                    .set("Rect", rect)
+                    .set("F", "4");
+                for (const [key, value] of annot.entries) {
+                    annotDict.set(key, value);
+                }
+                writer.addObject(annotObjNum, annotDict);
+                annotRefs.push(annotObjNum);
+            }
+            // Write overlay form fields
+            for (const field of editorPage._overlay._formFields) {
+                if (field.options.type === "radio") {
+                    // Radio group: parent field + child widgets
+                    const parentObjNum = writer.allocObject();
+                    const childRefs = [];
+                    let ff = 1 << 15; // Radio
+                    ff |= 1 << 14; // NoToggleToOff
+                    if (field.options.readOnly) {
+                        ff |= 1;
+                    }
+                    if (field.options.required) {
+                        ff |= 1 << 1;
+                    }
+                    for (const btn of field.options.buttons) {
+                        const childObjNum = writer.allocObject();
+                        const rect = `[${btn.rect.map(v => (0, pdf_object_1.pdfNumber)(v)).join(" ")}]`;
+                        const isSelected = field.options.selected === btn.value;
+                        const apState = isSelected ? `/${btn.value}` : "/Off";
+                        const childDict = new pdf_object_1.PdfDict()
+                            .set("Type", "/Annot")
+                            .set("Subtype", "/Widget")
+                            .set("Rect", rect)
+                            .set("Parent", (0, pdf_object_1.pdfRef)(parentObjNum))
+                            .set("AS", apState)
+                            .set("AP", `<< /N << /${btn.value} null /Off null >> >>`);
+                        writer.addObject(childObjNum, childDict);
+                        childRefs.push(childObjNum);
+                        annotRefs.push(childObjNum); // Children go into page /Annots
+                    }
+                    const parentDict = new pdf_object_1.PdfDict()
+                        .set("FT", "/Btn")
+                        .set("T", (0, pdf_object_1.pdfString)(field.options.name))
+                        .set("Ff", String(ff))
+                        .set("Kids", `[${childRefs.map(r => (0, pdf_object_1.pdfRef)(r)).join(" ")}]`);
+                    if (field.options.selected) {
+                        parentDict.set("V", `/${field.options.selected}`);
+                    }
+                    writer.addObject(parentObjNum, parentDict);
+                    // Parent does NOT go into annotRefs (not a visual annotation)
+                }
+                else {
+                    // Single-widget fields: text, checkbox, dropdown
+                    const fieldObjNum = writer.allocObject();
+                    const r = field.options.rect;
+                    const rect = `[${(0, pdf_object_1.pdfNumber)(r[0])} ${(0, pdf_object_1.pdfNumber)(r[1])} ${(0, pdf_object_1.pdfNumber)(r[2])} ${(0, pdf_object_1.pdfNumber)(r[3])}]`;
+                    const fieldDict = new pdf_object_1.PdfDict()
+                        .set("Type", "/Annot")
+                        .set("Subtype", "/Widget")
+                        .set("Rect", rect)
+                        .set("T", (0, pdf_object_1.pdfString)(field.options.name))
+                        .set("FT", field.options.type === "text"
+                        ? "/Tx"
+                        : field.options.type === "checkbox"
+                            ? "/Btn"
+                            : "/Ch");
+                    if (field.options.value) {
+                        fieldDict.set("V", (0, pdf_object_1.pdfString)(field.options.value));
+                    }
+                    fieldDict.set("DA", (0, pdf_object_1.pdfString)("/Helv 12 Tf 0 g"));
+                    writer.addObject(fieldObjNum, fieldDict);
+                    annotRefs.push(fieldObjNum);
+                }
+            }
+            const pageDict2 = new pdf_object_1.PdfDict()
+                .set("Type", "/Page")
+                .set("Parent", (0, pdf_object_1.pdfRef)(pagesTreeObjNum))
+                .set("MediaBox", `[0 0 ${(0, pdf_object_1.pdfNumber)(dims.width)} ${(0, pdf_object_1.pdfNumber)(dims.height)}]`)
+                .set("Contents", contentsStr)
+                .set("Resources", (0, pdf_object_1.pdfRef)(resourcesObjNum));
+            // Preserve page-level properties from the original page dict
+            this._copyPageProperties(pageDict, pageDict2);
+            // Apply rotation override if set
+            const rotationOverride = this._rotationOverrides.get(i);
+            if (rotationOverride !== undefined) {
+                if (rotationOverride === 0) {
+                    // Remove rotation (unset the key)
+                    pageDict2.delete("Rotate");
+                }
+                else {
+                    pageDict2.set("Rotate", String(rotationOverride));
+                }
+            }
+            if (annotRefs.length > 0) {
+                pageDict2.set("Annots", `[${annotRefs.map(r => (0, pdf_object_1.pdfRef)(r)).join(" ")}]`);
+            }
+            const pageObjNum = writer.allocObject();
+            writer.addObject(pageObjNum, pageDict2);
+            pageObjNums.push(pageObjNum);
+        }
+        // Write copied pages
+        for (const copied of this._copiedPages) {
+            const contentRefs = [];
+            for (const streamData of copied.contentStreams) {
+                const objNum = writer.allocObject();
+                writer.addStreamObject(objNum, new pdf_object_1.PdfDict(), streamData);
+                contentRefs.push(objNum);
+            }
+            // Deep-clone resources from the source document
+            const resourcesStr = this._serializeResources(copied.sourceDoc, copied.sourcePageDict);
+            const resourcesObjNum = writer.allocObject();
+            writer.addObject(resourcesObjNum, resourcesStr || "<< >>");
+            const contentsStr = contentRefs.length === 1
+                ? (0, pdf_object_1.pdfRef)(contentRefs[0])
+                : `[${contentRefs.map(r => (0, pdf_object_1.pdfRef)(r)).join(" ")}]`;
+            const pageDict = new pdf_object_1.PdfDict()
+                .set("Type", "/Page")
+                .set("Parent", (0, pdf_object_1.pdfRef)(pagesTreeObjNum))
+                .set("MediaBox", `[0 0 ${(0, pdf_object_1.pdfNumber)(copied.width)} ${(0, pdf_object_1.pdfNumber)(copied.height)}]`)
+                .set("Contents", contentsStr)
+                .set("Resources", (0, pdf_object_1.pdfRef)(resourcesObjNum));
+            // Preserve page properties from source page
+            this._copyPageProperties(copied.sourcePageDict, pageDict, copied.sourceDoc);
+            const pageObjNum = writer.allocObject();
+            writer.addObject(pageObjNum, pageDict);
+            pageObjNums.push(pageObjNum);
+        }
+        // Write new pages
+        for (const page of this._newPages) {
+            const imageXObjectMap = this._writeOverlayImages(writer, page);
+            let xobjDictStr = "";
+            if (imageXObjectMap.size > 0) {
+                const entries = [...imageXObjectMap.entries()]
+                    .map(([name, objNum]) => `/${name} ${(0, pdf_object_1.pdfRef)(objNum)}`)
+                    .join(" ");
+                xobjDictStr = `<< ${entries} >>`;
+            }
+            const contentObjNum = writer.allocObject();
+            writer.addStreamObject(contentObjNum, new pdf_object_1.PdfDict(), page._stream);
+            const resourcesObjNum = writer.allocObject();
+            let resStr = "<< ";
+            if (fontDictStr) {
+                resStr += `/Font ${fontDictStr} `;
+            }
+            if (xobjDictStr) {
+                resStr += `/XObject ${xobjDictStr} `;
+            }
+            resStr += ">>";
+            writer.addObject(resourcesObjNum, resStr);
+            const pageObjNum = writer.addPage({
+                parentRef: pagesTreeObjNum,
+                width: page._width,
+                height: page._height,
+                contentsRef: contentObjNum,
+                resourcesRef: resourcesObjNum
+            });
+            pageObjNums.push(pageObjNum);
+        }
+        // Pages tree
+        const kidsStr = pageObjNums.map(n => (0, pdf_object_1.pdfRef)(n)).join(" ");
+        writer.addObject(pagesTreeObjNum, new pdf_object_1.PdfDict()
+            .set("Type", "/Pages")
+            .set("Kids", `[${kidsStr}]`)
+            .set("Count", String(pageObjNums.length)));
+        // Catalog — with optional AcroForm
+        const catalogObjNum = writer.allocObject();
+        const catalogDict = new pdf_object_1.PdfDict().set("Type", "/Catalog").set("Pages", (0, pdf_object_1.pdfRef)(pagesTreeObjNum));
+        // If we have form field updates, write an AcroForm dict
+        if (this._formFieldUpdates.size > 0) {
+            try {
+                const catalog = this._doc.getCatalog();
+                const acroFormRef = catalog.get("AcroForm");
+                if (acroFormRef) {
+                    const acroFormStr = this._deepSerialize(this._doc, acroFormRef);
+                    // Insert /NeedAppearances into the cloned dict
+                    if (acroFormStr && acroFormStr.startsWith("<<")) {
+                        catalogDict.set("AcroForm", acroFormStr.replace("<<", "<< /NeedAppearances true"));
+                    }
+                    else {
+                        catalogDict.set("AcroForm", "<< /NeedAppearances true >>");
+                    }
+                }
+                else {
+                    catalogDict.set("AcroForm", "<< /NeedAppearances true >>");
+                }
+            }
+            catch {
+                catalogDict.set("AcroForm", "<< /NeedAppearances true >>");
+            }
+        }
+        // Preserve XMP metadata stream from original catalog
+        try {
+            const originalCatalog = this._doc.getCatalog();
+            const metadataRef = originalCatalog.get("Metadata");
+            if (metadataRef) {
+                const clonedRef = this._deepSerialize(this._doc, metadataRef);
+                if (clonedRef) {
+                    catalogDict.set("Metadata", clonedRef);
+                }
+            }
+        }
+        catch {
+            // If catalog is inaccessible, skip XMP preservation
+        }
+        writer.addObject(catalogObjNum, catalogDict);
+        writer.setCatalog(catalogObjNum);
+        // Inject signature placeholder if signing is in progress
+        if (this._signaturePlaceholder) {
+            const sigDictObjNum = writer.allocObject();
+            writer.addObject(sigDictObjNum, this._signaturePlaceholder);
+            const sigWidgetObjNum = writer.allocObject();
+            const sigWidgetDict = new pdf_object_1.PdfDict()
+                .set("Type", "/Annot")
+                .set("Subtype", "/Widget")
+                .set("FT", "/Sig")
+                .set("Rect", "[0 0 0 0]")
+                .set("T", (0, pdf_object_1.pdfString)("Signature1"))
+                .set("V", (0, pdf_object_1.pdfRef)(sigDictObjNum))
+                .set("F", "4");
+            writer.addObject(sigWidgetObjNum, sigWidgetDict);
+            // Patch catalog to include signature widget in AcroForm with SigFlags
+            // Grab existing fields from the catalog dict string representation
+            const catalogStr = catalogDict.toString();
+            const existingFieldsMatch = catalogStr.match(/\/Fields\s*\[([^\]]*)\]/);
+            const existingFields = existingFieldsMatch ? existingFieldsMatch[1].trim() : "";
+            const sigFieldRef = (0, pdf_object_1.pdfRef)(sigWidgetObjNum);
+            const allFieldRefs = existingFields ? `${existingFields} ${sigFieldRef}` : sigFieldRef;
+            catalogDict.set("AcroForm", `<< /Fields [${allFieldRefs}] /SigFlags 3 >>`);
+            writer.addObject(catalogObjNum, catalogDict);
+        }
+        // Info dict — preserve original metadata
+        const originalMeta = (0, metadata_reader_1.extractMetadata)(this._doc);
+        writer.addInfoDict({
+            title: originalMeta.title || undefined,
+            author: originalMeta.author || undefined,
+            subject: originalMeta.subject || undefined,
+            creator: originalMeta.creator || "excelts"
+        });
+        return writer.build();
+    }
+    /**
+     * Save the modified PDF using incremental update.
+     *
+     * Appends new/modified objects after the original PDF bytes, preserving the
+     * original data byte-for-byte. This is ideal for overlays and form field
+     * updates on existing pages — it preserves digital signatures on unmodified
+     * content and produces smaller output.
+     *
+     * Falls back to {@link save} (full rebuild) if structural changes are
+     * present (new pages, copied pages, or removed pages).
+     *
+     * @returns The modified PDF as Uint8Array
+     */
+    async saveIncremental() {
+        // Fall back to full rebuild if structural changes are present
+        if (this._newPages.length > 0 ||
+            this._copiedPages.length > 0 ||
+            this._removedPageIndices.size > 0) {
+            return this.save();
+        }
+        // Fall back to full rebuild if rotation overrides are present
+        if (this._rotationOverrides.size > 0) {
+            return this.save();
+        }
+        // Fall back to full rebuild for xref-stream PDFs (no "trailer" keyword)
+        const tailBytes = this._doc.data.subarray(Math.max(0, this._doc.data.length - 1024));
+        const tailStr = new TextDecoder().decode(tailBytes);
+        if (!tailStr.includes("trailer")) {
+            return this.save();
+        }
+        // Check if there are any modifications at all
+        const hasOverlays = this._pages.some(p => p._hasOverlay());
+        const hasFormUpdates = this._formFieldUpdates.size > 0;
+        if (!hasOverlays && !hasFormUpdates) {
+            // No modifications — return the original bytes
+            return this._doc.data;
+        }
+        this._isIncrementalSave = true;
+        try {
+            return await this._buildIncrementalUpdate();
+        }
+        finally {
+            this._isIncrementalSave = false;
+            this._writerForSave = null;
+            this._clonedRefs.clear();
+        }
+    }
+    /** @internal — Core incremental update logic, separated for try/finally cleanup. */
+    async _buildIncrementalUpdate() {
+        // Determine the next available object number from the original PDF's /Size
+        const originalSize = (0, pdf_parser_1.dictGetNumber)(this._doc.trailer, "Size") ?? 1;
+        let nextObjNum = originalSize;
+        // Collect modified objects: objNum → serialized content
+        const modifiedObjects = new Map();
+        // Check what kinds of modifications exist
+        const hasOverlays = this._pages.some(p => p._hasOverlay());
+        // We need a temporary PdfWriter for font resources used in overlays
+        const writer = new pdf_writer_1.PdfWriter();
+        this._writerForSave = writer;
+        this._clonedRefs = new Map();
+        let fontDictStr = "";
+        // Map of writer objNum → actual new objNum we'll use in the incremental update
+        const writerFontObjRemap = new Map();
+        if (hasOverlays) {
+            // Write font resources via the writer (to serialize font objects)
+            const fontObjectMap = this._fontManager.writeFontResources(writer);
+            // Remap all writer-allocated objects (fonts + their dependencies like
+            // CID font descriptors, ToUnicode CMaps, etc.) into the incremental
+            // update's object number space.
+            const writerObjects = writer.getObjects();
+            for (const obj of writerObjects) {
+                if (!writerFontObjRemap.has(obj.objectNumber)) {
+                    writerFontObjRemap.set(obj.objectNumber, nextObjNum++);
+                }
+            }
+            // Write all font objects into modifiedObjects with remapped refs
+            for (const obj of writerObjects) {
+                const newObjNum = writerFontObjRemap.get(obj.objectNumber);
+                const remappedContent = this._remapRefsInString(obj.content, writerFontObjRemap);
+                if (obj.streamData) {
+                    // Parse the remapped content back into a PdfDict for stream objects
+                    modifiedObjects.set(newObjNum, {
+                        dict: pdf_object_1.PdfDict.fromRawString(remappedContent),
+                        data: obj.streamData
+                    });
+                }
+                else {
+                    modifiedObjects.set(newObjNum, remappedContent);
+                }
+            }
+            // Build a remapped font dict string
+            const rawFontDict = this._fontManager.buildFontDictString(fontObjectMap);
+            fontDictStr = this._remapRefsInString(rawFontDict, writerFontObjRemap);
+        }
+        const pagesInfo = this._doc.getPagesWithObjInfo();
+        for (let i = 0; i < pagesInfo.length; i++) {
+            const editorPage = this._pages[i];
+            const { dict: pageDict, objNum: pageObjNum } = pagesInfo[i];
+            if (pageObjNum === 0) {
+                // Can't do incremental update without knowing the page object number.
+                // Fall back to full rebuild (finally block handles cleanup).
+                return this.save();
+            }
+            const pageHasOverlay = editorPage._hasOverlay();
+            const pageHasFormUpdates = this._hasFormUpdatesForPage(pageDict);
+            if (!pageHasOverlay && !pageHasFormUpdates) {
+                continue; // Page is unchanged
+            }
+            // Build the updated page dictionary. We start from the original page
+            // dict's entries and modify only what's necessary.
+            const updatedPageDict = new pdf_object_1.PdfDict();
+            // Copy all existing entries from the original page dict
+            for (const [key, val] of pageDict.entries()) {
+                if (key === "Contents" && pageHasOverlay) {
+                    continue; // We'll rewrite /Contents below
+                }
+                if (key === "Resources" && pageHasOverlay) {
+                    continue; // We'll rewrite /Resources below
+                }
+                if (key === "Annots" && pageHasFormUpdates) {
+                    continue; // We'll rewrite /Annots below
+                }
+                updatedPageDict.set(key, this._serializeOriginalValue(val));
+            }
+            if (pageHasOverlay) {
+                // Create a new content stream object for the overlay
+                const overlayStreamData = editorPage._overlay._stream.toUint8Array();
+                const overlayObjNum = nextObjNum++;
+                modifiedObjects.set(overlayObjNum, {
+                    dict: new pdf_object_1.PdfDict(),
+                    data: overlayStreamData
+                });
+                // Build the new /Contents array: original refs + overlay ref
+                const originalContentsObj = pageDict.get("Contents");
+                const contentRefs = this._collectOriginalContentRefs(originalContentsObj);
+                contentRefs.push((0, pdf_object_1.pdfRef)(overlayObjNum));
+                const contentsStr = contentRefs.length === 1 ? contentRefs[0] : `[${contentRefs.join(" ")}]`;
+                updatedPageDict.set("Contents", contentsStr);
+                // Build merged resources: original + overlay fonts/images
+                const originalResources = this._serializeOriginalResources(this._doc, pageDict);
+                // For overlay images, write them as new objects
+                const imageObjMap = new Map();
+                for (let imgIdx = 0; imgIdx < editorPage._overlay._images.length; imgIdx++) {
+                    const img = editorPage._overlay._images[imgIdx];
+                    const imgName = `Im${imgIdx + 1}`;
+                    const imgObjNum = nextObjNum++;
+                    imageObjMap.set(imgName, imgObjNum);
+                    modifiedObjects.set(imgObjNum, this._buildImageXObjectForIncremental(img.data, img.format));
+                }
+                // Build overlay resource string
+                let overlayStr = "<< ";
+                if (fontDictStr) {
+                    overlayStr += `/Font ${fontDictStr} `;
+                }
+                if (imageObjMap.size > 0) {
+                    const entries = [...imageObjMap.entries()]
+                        .map(([name, objNum]) => `/${name} ${(0, pdf_object_1.pdfRef)(objNum)}`)
+                        .join(" ");
+                    overlayStr += `/XObject << ${entries} >> `;
+                }
+                overlayStr += ">>";
+                const mergedResources = this._mergeResourceStrings(originalResources, overlayStr);
+                // Write merged resources as a new object
+                const resourcesObjNum = nextObjNum++;
+                modifiedObjects.set(resourcesObjNum, mergedResources || "<< >>");
+                updatedPageDict.set("Resources", (0, pdf_object_1.pdfRef)(resourcesObjNum));
+            }
+            if (pageHasFormUpdates) {
+                // Build updated annotations list
+                const annotsResult = this._buildIncrementalAnnots(pageDict, modifiedObjects, nextObjNum);
+                nextObjNum = annotsResult.nextObjNum;
+                if (annotsResult.annotRefs.length > 0) {
+                    updatedPageDict.set("Annots", `[${annotsResult.annotRefs.map(r => (0, pdf_object_1.pdfRef)(r)).join(" ")}]`);
+                }
+            }
+            // Write the updated page dict as a modified object (same object number as original)
+            modifiedObjects.set(pageObjNum, updatedPageDict.toString());
+        }
+        if (modifiedObjects.size === 0) {
+            return this._doc.data;
+        }
+        return (0, pdf_writer_1.buildIncremental)(this._doc.data, modifiedObjects, new Map());
+    }
+    /**
+     * Check if a page has form field updates in its annotations.
+     * @internal
+     */
+    _hasFormUpdatesForPage(pageDict) {
+        if (this._formFieldUpdates.size === 0) {
+            return false;
+        }
+        const annotsObj = pageDict.get("Annots");
+        if (!annotsObj) {
+            return false;
+        }
+        const annotsResolved = this._doc.deref(annotsObj);
+        if (!(0, pdf_parser_1.isPdfArray)(annotsResolved)) {
+            return false;
+        }
+        for (const annotRef of annotsResolved) {
+            const annotDict = this._doc.derefDict(annotRef);
+            if (!annotDict) {
+                continue;
+            }
+            const subtype = (0, pdf_parser_1.dictGetName)(annotDict, "Subtype");
+            if (subtype === "Widget") {
+                const fieldName = this._resolveFieldName(annotDict);
+                if (fieldName && this._formFieldUpdates.has(fieldName)) {
+                    return true;
+                }
+            }
+        }
+        return false;
+    }
+    /**
+     * Collect original /Contents refs as serialized ref strings.
+     * For incremental update — preserves original object references.
+     * @internal
+     */
+    _collectOriginalContentRefs(contentsObj) {
+        if (!contentsObj) {
+            return [];
+        }
+        if ((0, pdf_parser_1.isPdfRef)(contentsObj)) {
+            // Could be a single ref to a stream, or a ref to an array
+            const resolved = this._doc.deref(contentsObj);
+            if ((0, pdf_parser_1.isPdfArray)(resolved)) {
+                return resolved
+                    .filter((item) => (0, pdf_parser_1.isPdfRef)(item))
+                    .map(item => (0, pdf_object_1.pdfRef)(item.objNum, item.gen));
+            }
+            return [(0, pdf_object_1.pdfRef)(contentsObj.objNum, contentsObj.gen)];
+        }
+        if ((0, pdf_parser_1.isPdfArray)(contentsObj)) {
+            return contentsObj
+                .filter((item) => (0, pdf_parser_1.isPdfRef)(item))
+                .map(item => (0, pdf_object_1.pdfRef)(item.objNum, item.gen));
+        }
+        return [];
+    }
+    /**
+     * Serialize a page's Resources dict preserving original object references.
+     * Unlike _serializeResources which deep-clones into the writer, this keeps
+     * the original object numbers intact for incremental updates.
+     * @internal
+     */
+    _serializeOriginalResources(doc, pageDict) {
+        const resourcesDict = doc.resolvePageResources(pageDict);
+        if (!resourcesDict || resourcesDict.size === 0) {
+            return "<< >>";
+        }
+        const parts = ["<<"];
+        for (const [key, val] of resourcesDict.entries()) {
+            const serialized = this._serializeOriginalValue(val);
+            if (serialized) {
+                parts.push(`/${key} ${serialized}`);
+            }
+        }
+        parts.push(">>");
+        return parts.join(" ");
+    }
+    /**
+     * Serialize a PDF value from the original document, preserving original refs.
+     * Unlike _deepSerialize which clones refs into a new writer, this keeps
+     * the original object numbers intact.
+     * @internal
+     */
+    _serializeOriginalValue(val) {
+        if (val === null || val === undefined) {
+            return "null";
+        }
+        if (typeof val === "string") {
+            // In parsed PDF objects, string type = PDF name (without leading /)
+            // Uint8Array = PDF string literal
+            return "/" + val;
+        }
+        if (typeof val === "number") {
+            return (0, pdf_object_1.pdfNumber)(val);
+        }
+        if (typeof val === "boolean") {
+            return val ? "true" : "false";
+        }
+        if (val instanceof Uint8Array) {
+            return (0, pdf_object_1.pdfHexString)(val);
+        }
+        if ((0, pdf_parser_1.isPdfRef)(val)) {
+            return (0, pdf_object_1.pdfRef)(val.objNum, val.gen);
+        }
+        if ((0, pdf_parser_1.isPdfArray)(val)) {
+            const items = val.map(item => this._serializeOriginalValue(item));
+            return `[${items.join(" ")}]`;
+        }
+        if (val instanceof Map) {
+            const parts = ["<<"];
+            for (const [k, v] of val.entries()) {
+                parts.push(`/${k} ${this._serializeOriginalValue(v)}`);
+            }
+            parts.push(">>");
+            return parts.join(" ");
+        }
+        return "";
+    }
+    /**
+     * Replace object number references in a serialized string.
+     * Maps old obj numbers (from temporary writer) to new obj numbers.
+     * @internal
+     */
+    _remapRefsInString(str, remap) {
+        // Replace patterns like "N 0 R" where N is in the remap map
+        return str.replace(/(\d+) (\d+) R/g, (match, objNumStr, genStr) => {
+            const objNum = parseInt(objNumStr, 10);
+            const remapped = remap.get(objNum);
+            if (remapped !== undefined) {
+                return `${remapped} ${genStr} R`;
+            }
+            return match;
+        });
+    }
+    /**
+     * Build image XObject content for incremental update.
+     * @internal
+     */
+    _buildImageXObjectForIncremental(data, format) {
+        const dims = (0, image_utils_1.parseImageDimensions)(data, format);
+        const dict = new pdf_object_1.PdfDict()
+            .set("Type", "/XObject")
+            .set("Subtype", "/Image")
+            .set("Width", (0, pdf_object_1.pdfNumber)(dims.width))
+            .set("Height", (0, pdf_object_1.pdfNumber)(dims.height))
+            .set("BitsPerComponent", "8")
+            .set("ColorSpace", "/DeviceRGB");
+        if (format === "jpeg") {
+            dict.set("Filter", "/DCTDecode");
+        }
+        return { dict, data };
+    }
+    /**
+     * Build updated annotations for incremental save.
+     * Modified widgets get rewritten at their original object number;
+     * unmodified annots keep original refs.
+     * @internal
+     */
+    _buildIncrementalAnnots(pageDict, modifiedObjects, nextObjNum) {
+        const annotsObj = pageDict.get("Annots");
+        if (!annotsObj) {
+            return { annotRefs: [], nextObjNum };
+        }
+        const annotsResolved = this._doc.deref(annotsObj);
+        if (!(0, pdf_parser_1.isPdfArray)(annotsResolved)) {
+            return { annotRefs: [], nextObjNum };
+        }
+        const annotRefs = [];
+        for (const annotRef of annotsResolved) {
+            const annotDict = this._doc.derefDict(annotRef);
+            if (!annotDict) {
+                // Keep original ref if we can't resolve
+                if ((0, pdf_parser_1.isPdfRef)(annotRef)) {
+                    annotRefs.push(annotRef.objNum);
+                }
+                continue;
+            }
+            const subtype = (0, pdf_parser_1.dictGetName)(annotDict, "Subtype");
+            if (subtype === "Widget" && this._formFieldUpdates.size > 0) {
+                const fieldName = this._resolveFieldName(annotDict);
+                const newValue = fieldName ? this._formFieldUpdates.get(fieldName) : undefined;
+                if (newValue !== undefined) {
+                    // Rewrite the annotation at its original object number
+                    const annotObjNum = (0, pdf_parser_1.isPdfRef)(annotRef) ? annotRef.objNum : nextObjNum++;
+                    const newDict = this._buildModifiedWidgetDict(annotDict, newValue);
+                    modifiedObjects.set(annotObjNum, newDict);
+                    annotRefs.push(annotObjNum);
+                    continue;
+                }
+            }
+            // Keep original annotation reference
+            if ((0, pdf_parser_1.isPdfRef)(annotRef)) {
+                annotRefs.push(annotRef.objNum);
+            }
+            else {
+                // Inline annotation — write as new object
+                const annotObjNum = nextObjNum++;
+                const serialized = this._serializeAnnotDict(annotDict);
+                modifiedObjects.set(annotObjNum, serialized);
+                annotRefs.push(annotObjNum);
+            }
+        }
+        return { annotRefs, nextObjNum };
+    }
+    /** @internal - Copy preserved page properties from source to target dict. */
+    _copyPageProperties(source, target, doc) {
+        const resolveDoc = doc ?? this._doc;
+        for (const key of PdfEditor._PAGE_PRESERVE_KEYS) {
+            const val = source.get(key);
+            if (val !== undefined && val !== null) {
+                target.set(key, this._deepSerialize(resolveDoc, val));
+            }
+        }
+    }
+    /**
+     * Sign this PDF with a digital signature.
+     *
+     * Performs a full save with an embedded PKCS#7 signature placeholder,
+     * then fills in the real CMS SignedData.
+     *
+     * @param options - Certificate, private key, and optional signer metadata
+     * @returns The signed PDF as Uint8Array
+     *
+     * @example
+     * ```typescript
+     * const editor = PdfEditor.load(pdfBytes);
+     * const signed = await editor.sign({
+     *   certificate: certDerBytes,
+     *   privateKey: pkcs8DerBytes,
+     *   name: "Jane Doe",
+     *   reason: "Approval"
+     * });
+     * ```
+     */
+    async sign(options) {
+        const { buildSignatureDictPlaceholder, signPdf } = await Promise.resolve().then(() => __importStar(require("../core/digital-signature")));
+        const { dictString } = buildSignatureDictPlaceholder({
+            name: options.name,
+            reason: options.reason,
+            location: options.location,
+            contactInfo: options.contactInfo
+        });
+        // Inject the signature placeholder into the form field updates
+        // so that save() includes it in the output
+        this._signaturePlaceholder = dictString;
+        let pdfWithPlaceholder;
+        try {
+            pdfWithPlaceholder = await this.save();
+        }
+        finally {
+            this._signaturePlaceholder = null;
+        }
+        return signPdf(pdfWithPlaceholder, options.certificate, options.privateKey);
+    }
+    /** @internal - Collect decoded content stream bytes from a page dict. */
+    _collectContentStreams(doc, pageDict) {
+        const contentsObj = pageDict.get("Contents");
+        if (!contentsObj) {
+            return [];
+        }
+        if ((0, pdf_parser_1.isPdfRef)(contentsObj)) {
+            const result = doc.derefStreamWithObjNum(contentsObj);
+            if (result) {
+                return [doc.getStreamData(result.stream, result.objNum, result.gen)];
+            }
+            // Could be a ref to an array
+            const resolved = doc.deref(contentsObj);
+            if ((0, pdf_parser_1.isPdfArray)(resolved)) {
+                return this._resolveStreamArray(doc, resolved);
+            }
+            return [];
+        }
+        if ((0, pdf_parser_1.isPdfArray)(contentsObj)) {
+            return this._resolveStreamArray(doc, contentsObj);
+        }
+        return [];
+    }
+    /** @internal */
+    _resolveStreamArray(doc, arr) {
+        const result = [];
+        for (const item of arr) {
+            const r = doc.derefStreamWithObjNum(item);
+            if (r) {
+                result.push(doc.getStreamData(r.stream, r.objNum, r.gen));
+            }
+        }
+        return result;
+    }
+    /** @internal - Serialize a page's Resources dict by deep-cloning objects into the writer. */
+    _serializeResources(doc, pageDict) {
+        const resourcesDict = doc.resolvePageResources(pageDict);
+        if (!resourcesDict || resourcesDict.size === 0) {
+            return "<< >>";
+        }
+        // Deep-clone the resources dict into the writer.
+        // We need to re-emit Font, XObject, ExtGState sub-dicts with their
+        // referenced objects written as new indirect objects in the writer.
+        return this._serializeDictToWriter(doc, resourcesDict);
+    }
+    /**
+     * @internal - Recursively serialize a PdfDictValue, writing stream objects
+     * as new indirect objects and converting refs to new writer refs.
+     */
+    _serializeDictToWriter(doc, dict) {
+        const parts = ["<<"];
+        for (const [key, val] of dict.entries()) {
+            const serialized = this._deepSerialize(doc, val);
+            if (serialized) {
+                parts.push(`/${key} ${serialized}`);
+            }
+        }
+        parts.push(">>");
+        return parts.join(" ");
+    }
+    /**
+     * @internal - Deep-serialize a PDF value.
+     * For indirect refs: resolve the target, write it as a new object in the writer,
+     * and return a ref to the new object.
+     * For dicts/arrays: recurse.
+     */
+    _deepSerialize(doc, val) {
+        if (val === null || val === undefined) {
+            return "null";
+        }
+        if (typeof val === "string") {
+            // In parsed PDF objects, string type = PDF name (without leading /)
+            return "/" + val;
+        }
+        if (typeof val === "number") {
+            return (0, pdf_object_1.pdfNumber)(val);
+        }
+        if (typeof val === "boolean") {
+            return val ? "true" : "false";
+        }
+        if (val instanceof Uint8Array) {
+            return (0, pdf_object_1.pdfHexString)(val);
+        }
+        if ((0, pdf_parser_1.isPdfRef)(val)) {
+            // Check if this ref has already been cloned
+            const docId = doc === this._doc ? "main" : `src${doc.data.length}`;
+            const cacheKey = `${docId}:${val.objNum}:${val.gen}`;
+            const cached = this._clonedRefs.get(cacheKey);
+            if (cached !== undefined) {
+                return (0, pdf_object_1.pdfRef)(cached);
+            }
+            // Try as stream first
+            const streamResult = doc.derefStreamWithObjNum(val);
+            if (streamResult) {
+                const newObjNum = this._writerForSave.allocObject();
+                this._clonedRefs.set(cacheKey, newObjNum);
+                // Get the decoded stream data
+                const streamData = doc.getStreamData(streamResult.stream, streamResult.objNum, streamResult.gen);
+                // Serialize the stream's dict (excluding /Length which will be set automatically)
+                const streamDict = new pdf_object_1.PdfDict();
+                for (const [k, v] of streamResult.stream.dict.entries()) {
+                    if (k === "Length" || k === "Filter" || k === "DecodeParms") {
+                        // Skip — the writer will re-compress and set these
+                        continue;
+                    }
+                    const sv = this._deepSerialize(doc, v);
+                    if (sv) {
+                        streamDict.set(k, sv);
+                    }
+                }
+                this._writerForSave.addStreamObject(newObjNum, streamDict, streamData);
+                return (0, pdf_object_1.pdfRef)(newObjNum);
+            }
+            // Try as regular dict/value
+            const resolved = doc.deref(val);
+            if (resolved instanceof Map) {
+                const newObjNum = this._writerForSave.allocObject();
+                this._clonedRefs.set(cacheKey, newObjNum);
+                const dictStr = this._serializeDictToWriter(doc, resolved);
+                this._writerForSave.addObject(newObjNum, dictStr);
+                return (0, pdf_object_1.pdfRef)(newObjNum);
+            }
+            // Primitive value behind a ref — just inline it
+            return this._deepSerialize(doc, resolved);
+        }
+        if ((0, pdf_parser_1.isPdfArray)(val)) {
+            const items = val.map(item => this._deepSerialize(doc, item));
+            return `[${items.join(" ")}]`;
+        }
+        if (val instanceof Map) {
+            return this._serializeDictToWriter(doc, val);
+        }
+        return "";
+    }
+    /** @internal */
+    _writeOverlayImages(writer, page) {
+        const map = new Map();
+        for (let i = 0; i < page._images.length; i++) {
+            const img = page._images[i];
+            const imgName = `Im${i + 1}`;
+            const objNum = (0, image_utils_1.writeImageXObject)(writer, img.data, img.format);
+            map.set(imgName, objNum);
+        }
+        return map;
+    }
+    /** @internal - Build overlay resources as a structured PdfResourceDict. */
+    _buildOverlayResourceDict(fontDictStr, imageXObjectMap) {
+        const dict = new Map();
+        if (fontDictStr) {
+            // Parse the font dict string into structured entries
+            // fontDictStr is already a `<< /F1 3 0 R ... >>` string
+            const fontInner = fontDictStr.trim();
+            if (fontInner.startsWith("<<") && fontInner.endsWith(">>")) {
+                const parsed = (0, resource_merger_1.parseResourceDict)(`<< /Font ${fontInner} >>`);
+                const fontMap = parsed.get("Font");
+                if (fontMap) {
+                    dict.set("Font", fontMap);
+                }
+            }
+        }
+        if (imageXObjectMap.size > 0) {
+            const xobjMap = new Map();
+            for (const [name, objNum] of imageXObjectMap) {
+                xobjMap.set(name, (0, pdf_object_1.pdfRef)(objNum));
+            }
+            dict.set("XObject", xobjMap);
+        }
+        return dict;
+    }
+    /** @internal - Merge original and overlay resource strings via parsed object graph. */
+    _mergeResourceStrings(original, overlay) {
+        // If no overlay, return original
+        if (!overlay || overlay === "<< >>") {
+            return original;
+        }
+        // If no original, return overlay
+        if (!original || original === "<< >>") {
+            return overlay;
+        }
+        const origDict = (0, resource_merger_1.parseResourceDict)(original);
+        const overlayDict = (0, resource_merger_1.parseResourceDict)(overlay);
+        const merged = (0, resource_merger_1.mergeResourceDicts)(origDict, overlayDict);
+        return (0, resource_merger_1.serializeResourceDict)(merged);
+    }
+    /** @internal - Write form field value updates as annotation objects. */
+    _writeFormFieldUpdates(writer, pageDict, _pageIndex) {
+        if (this._formFieldUpdates.size === 0) {
+            return this._copyExistingAnnots(writer, pageDict);
+        }
+        // Get existing annotations
+        const annotRefs = this._copyExistingAnnots(writer, pageDict);
+        // We handle form field updates by modifying Widget annotations.
+        // This is done during the annotation copy above — if we find a Widget
+        // annotation whose field name matches an update, we modify its /V value.
+        return annotRefs;
+    }
+    /** @internal */
+    _copyExistingAnnots(writer, pageDict) {
+        const annotsObj = pageDict.get("Annots");
+        if (!annotsObj) {
+            return [];
+        }
+        const annotsResolved = this._doc.deref(annotsObj);
+        if (!(0, pdf_parser_1.isPdfArray)(annotsResolved)) {
+            return [];
+        }
+        const annotRefs = [];
+        for (const annotRef of annotsResolved) {
+            const annotDict = this._doc.derefDict(annotRef);
+            if (!annotDict) {
+                continue;
+            }
+            const subtype = (0, pdf_parser_1.dictGetName)(annotDict, "Subtype");
+            const objNum = writer.allocObject();
+            if (subtype === "Widget" && this._formFieldUpdates.size > 0) {
+                // Check if this widget has a field name that matches an update
+                const fieldName = this._resolveFieldName(annotDict);
+                const newValue = fieldName ? this._formFieldUpdates.get(fieldName) : undefined;
+                if (newValue !== undefined) {
+                    // Write a modified widget annotation with the new value
+                    const newDict = this._buildModifiedWidgetDict(annotDict, newValue);
+                    writer.addObject(objNum, newDict);
+                    annotRefs.push(objNum);
+                    continue;
+                }
+            }
+            // Copy annotation as-is (serialize what we can)
+            const serialized = this._serializeAnnotDict(annotDict);
+            writer.addObject(objNum, serialized);
+            annotRefs.push(objNum);
+        }
+        return annotRefs;
+    }
+    /** @internal */
+    _resolveFieldName(dict) {
+        const parts = [];
+        let current = dict;
+        const visited = new Set();
+        while (current) {
+            const tVal = current.get("T");
+            if (tVal) {
+                const resolved = this._doc.deref(tVal);
+                let name = "";
+                if (typeof resolved === "string") {
+                    name = resolved;
+                }
+                else if (resolved instanceof Uint8Array) {
+                    name = (0, pdf_parser_1.decodePdfStringBytes)(resolved);
+                }
+                if (name) {
+                    parts.unshift(name);
+                }
+            }
+            const parentVal = current.get("Parent");
+            if (!parentVal) {
+                break;
+            }
+            // Cycle guard
+            const key = String(parentVal);
+            if (visited.has(key)) {
+                break;
+            }
+            visited.add(key);
+            current = this._doc.derefDict(parentVal);
+        }
+        return parts.join(".");
+    }
+    /** @internal */
+    _buildModifiedWidgetDict(originalDict, newValue) {
+        // Determine the field type (/FT) — may be directly on the widget or inherited from parent
+        const fieldType = this._resolveFieldType(originalDict);
+        // For text fields, generate an appearance stream instead of stripping /AP
+        if (fieldType === "Tx" && this._writerForSave) {
+            return this._buildTextFieldWidgetDict(originalDict, newValue);
+        }
+        // For non-text fields, fall back to stripping /AP (force viewer to regenerate)
+        const parts = ["<<"];
+        for (const [key, val] of originalDict.entries()) {
+            if (key === "V" || key === "AP") {
+                continue;
+            }
+            const serialized = this._serializePdfValue(val);
+            if (serialized) {
+                parts.push(`/${key} ${serialized}`);
+            }
+        }
+        parts.push(`/V ${(0, pdf_object_1.pdfString)(newValue)}`);
+        parts.push(">>");
+        return parts.join(" ");
+    }
+    /**
+     * @internal - Build a modified widget dict for a text field with an inline
+     * appearance stream. The stream renders the field value so it is visible
+     * in all viewers, even those that ignore /NeedAppearances.
+     */
+    _buildTextFieldWidgetDict(originalDict, newValue) {
+        const writer = this._writerForSave;
+        // Extract the widget Rect for sizing the appearance
+        const rect = this._resolveWidgetRect(originalDict);
+        // Determine alignment from /Q entry (0=left, 1=center, 2=right)
+        const qVal = originalDict.get("Q");
+        let alignment = "left";
+        if (qVal === 1) {
+            alignment = "center";
+        }
+        else if (qVal === 2) {
+            alignment = "right";
+        }
+        // Generate the appearance stream
+        const { stream, resources } = (0, form_appearance_1.generateTextFieldAppearance)({
+            value: newValue,
+            rect,
+            alignment
+        });
+        // Write the appearance stream as a Form XObject indirect object
+        const apObjNum = writer.allocObject();
+        const apDict = new pdf_object_1.PdfDict()
+            .set("Type", "/XObject")
+            .set("Subtype", "/Form")
+            .set("BBox", (0, form_appearance_1.buildAppearanceBBox)(rect))
+            .set("Resources", resources);
+        writer.addStreamObject(apObjNum, apDict, stream, { compress: false });
+        // Build the widget dict
+        const parts = ["<<"];
+        for (const [key, val] of originalDict.entries()) {
+            if (key === "V" || key === "AP") {
+                continue;
+            }
+            const serialized = this._serializePdfValue(val);
+            if (serialized) {
+                parts.push(`/${key} ${serialized}`);
+            }
+        }
+        // Set the new value and appearance
+        parts.push(`/V ${(0, pdf_object_1.pdfString)(newValue)}`);
+        parts.push(`/AP << /N ${(0, pdf_object_1.pdfRef)(apObjNum)} >>`);
+        parts.push(">>");
+        return parts.join(" ");
+    }
+    /**
+     * @internal - Resolve the field type (/FT) which may be inherited from a parent dict.
+     */
+    _resolveFieldType(dict) {
+        let current = dict;
+        const visited = new Set();
+        while (current) {
+            const ft = (0, pdf_parser_1.dictGetName)(current, "FT");
+            if (ft) {
+                return ft;
+            }
+            const parentVal = current.get("Parent");
+            if (!parentVal) {
+                break;
+            }
+            const key = String(parentVal);
+            if (visited.has(key)) {
+                break;
+            }
+            visited.add(key);
+            current = this._doc.derefDict(parentVal);
+        }
+        return undefined;
+    }
+    /**
+     * @internal - Extract the /Rect array from a widget annotation dict as [x1, y1, x2, y2].
+     */
+    _resolveWidgetRect(dict) {
+        let rectVal = dict.get("Rect");
+        if (rectVal) {
+            rectVal = this._doc.deref(rectVal);
+        }
+        if (rectVal && (0, pdf_parser_1.isPdfArray)(rectVal)) {
+            return rectVal.map(v => (typeof v === "number" ? v : 0));
+        }
+        // Fallback: 0-size rect
+        return [0, 0, 100, 20];
+    }
+    /** @internal */
+    _serializeAnnotDict(dict) {
+        const parts = ["<<"];
+        for (const [key, val] of dict.entries()) {
+            const serialized = this._serializePdfValue(val);
+            if (serialized) {
+                parts.push(`/${key} ${serialized}`);
+            }
+        }
+        parts.push(">>");
+        return parts.join(" ");
+    }
+    /** @internal */
+    _serializePdfValue(val) {
+        if (val === null || val === undefined) {
+            return "null";
+        }
+        if (typeof val === "string") {
+            // In parsed PDF objects, string type = PDF name (without leading /)
+            return "/" + val;
+        }
+        if (typeof val === "number") {
+            return (0, pdf_object_1.pdfNumber)(val);
+        }
+        if (typeof val === "boolean") {
+            return val ? "true" : "false";
+        }
+        if (val instanceof Uint8Array) {
+            return (0, pdf_object_1.pdfHexString)(val);
+        }
+        if ((0, pdf_parser_1.isPdfRef)(val)) {
+            // During incremental save, original refs remain valid — preserve them as-is.
+            // During full save, deep-clone into the new writer.
+            if (this._writerForSave && !this._isIncrementalSave) {
+                return this._deepSerialize(this._doc, val);
+            }
+            return `${val.objNum} ${val.gen} R`;
+        }
+        if ((0, pdf_parser_1.isPdfArray)(val)) {
+            const items = val.map(item => this._serializePdfValue(item));
+            return `[${items.join(" ")}]`;
+        }
+        if (val instanceof Map) {
+            const parts = ["<<"];
+            for (const [k, v] of val.entries()) {
+                const serialized = this._serializePdfValue(v);
+                parts.push(`/${k} ${serialized}`);
+            }
+            parts.push(">>");
+            return parts.join(" ");
+        }
+        return "";
+    }
+}
+exports.PdfEditor = PdfEditor;
+// ===========================================================================
+// Internal Helpers
+// ===========================================================================
+/** Page-level keys to preserve when rebuilding page dicts. */
+PdfEditor._PAGE_PRESERVE_KEYS = [
+    "Rotate",
+    "CropBox",
+    "BleedBox",
+    "TrimBox",
+    "ArtBox",
+    "Group",
+    "UserUnit",
+    "Tabs"
+];