feat: expose PDF page indirect object number via EPDF_GetPageObjNum

.changeset/pink-balloons-make.md

@@ -0,0 +1,7 @@
+---
+'@embedpdf/engines': minor
+'@embedpdf/models': minor
+'@embedpdf/pdfium': minor
+---
+
+Add EPDF_GetPageObjNum to expose PDF page indirect object numbers, populate objectNumber on PdfPageObject via the engine

packages/engines/src/lib/pdfium/engine.ts

@@ -385,13 +385,23 @@ export class PdfiumNative implements IPdfiumExecutor {
 
       const rotation = this.pdfiumModule.EPDF_GetPageRotationByIndex(docPtr, index) as Rotation;
 
-      const page = {
+      let objectNumber: number | undefined;
+      const epdfGetPageObjNum = this.pdfiumModule.EPDF_GetPageObjNum;
+      if (typeof epdfGetPageObjNum === 'function') {
+        const objNum = epdfGetPageObjNum(docPtr, index);
+        if (objNum > 0) {
+          objectNumber = objNum;
+        }
+      }
+
+      const page: PdfPageObject = {
         index,
         size: {
           width: this.pdfiumModule.pdfium.getValue(sizePtr, 'float'),
           height: this.pdfiumModule.pdfium.getValue(sizePtr + 4, 'float'),
         },
         rotation,
+        ...(objectNumber !== undefined ? { objectNumber } : {}),
       };
 
       pages.push(page);

packages/models/src/pdf.ts

@@ -12,6 +12,21 @@ export interface PdfPageObject {
    */
   index: number;
 
+  /**
+   * PDF page dictionary indirect object number.
+   *
+   * Populated by the engine at document open time when the underlying WASM
+   * build exports `EPDF_GetPageObjNum`. Undefined when that export is absent
+   * (e.g. older builds) or when the page dictionary has no valid object number.
+   *
+   * This value is stable across page reorder / insert / delete operations on
+   * the same file, making it suitable as a permanent page reference for
+   * annotations, citations, and similar features.
+   *
+   * @public
+   */
+  objectNumber?: number;
+
   /**
    * Orignal size of this page
    */

packages/pdfium/build/code/cpp/ext_api.h

@@ -21,6 +21,8 @@ void        PDFiumExt_ExitFormFillEnvironment(void* formHandle);
 
 int         PDFiumExt_SaveAsCopy(void* document, void* writer);
 
+int         EPDF_GetPageObjNum(void* document, int page_index);
+
 #ifdef __cplusplus
 }
 #endif

packages/pdfium/build/code/cpp/main.cpp

@@ -3,6 +3,9 @@
 #include <emscripten.h>
 #include "filewriter.h"
 #include "string.h"
+#include "fpdfsdk/cpdfsdk_helpers.h"
+#include "core/fpdfapi/parser/cpdf_document.h"
+#include "core/fpdfapi/parser/cpdf_dictionary.h"
 
 #ifdef __cplusplus
 extern "C"
@@ -23,6 +26,8 @@ extern "C"
 
     EMSCRIPTEN_KEEPALIVE int PDFiumExt_SaveAsCopy(void *document, void *writer);
 
+    EMSCRIPTEN_KEEPALIVE int EPDF_GetPageObjNum(FPDF_DOCUMENT document, int page_index);
+
 #ifdef __cplusplus
 }
 #endif
@@ -91,3 +96,14 @@ void PDFiumExt_ExitFormFillEnvironment(void *form_handle)
 {
     FPDFDOC_ExitFormFillEnvironment(static_cast<FPDF_FORMHANDLE>(form_handle));
 }
+
+int EPDF_GetPageObjNum(FPDF_DOCUMENT document, int page_index)
+{
+    CPDF_Document *pDoc = CPDFDocumentFromFPDFDocument(document);
+    if (!pDoc)
+        return -1;
+    RetainPtr<const CPDF_Dictionary> pDict = pDoc->GetPageDictionary(page_index);
+    if (!pDict)
+        return -1;
+    return static_cast<int>(pDict->GetObjNum());
+}

packages/pdfium/build/compile.esm.sh

@@ -14,6 +14,7 @@ em++ $(ls ./code/cpp/*.cpp) \
   -sEXPORTED_FUNCTIONS=$(cat ./wasm/exported-functions.txt) \
   -lpdfium \
   -L/workspace/packages/pdfium/pdfium-src/out/wasm/obj \
+  -I/workspace/packages/pdfium/pdfium-src \
   -I/workspace/packages/pdfium/pdfium-src/public \
   -std=c++11 \
   -Wall \

packages/pdfium/build/compile.sh

@@ -13,6 +13,7 @@ em++ $(ls ./code/cpp/*.cpp) \
   -sEXPORTED_FUNCTIONS=$(cat ./wasm/exported-functions.txt) \
   -lpdfium \
   -L/workspace/packages/pdfium/pdfium-src/out/wasm/obj \
+  -I/workspace/packages/pdfium/pdfium-src \
   -I/workspace/packages/pdfium/pdfium-src/public \
   -std=c++11 \
   -Wall \

packages/pdfium/src/vendor/functions.ts

@@ -7,6 +7,7 @@ export const functions = {
   ] as const,
   EPDF_GetMetaTrapped: [['number'] as const, 'number'] as const,
   EPDF_GetPageRotationByIndex: [['number', 'number'] as const, 'number'] as const,
+  EPDF_GetPageObjNum: [['number', 'number'] as const, 'number'] as const,
   EPDF_GetPageSizeByIndexNormalized: [['number', 'number', 'number'] as const, 'boolean'] as const,
   EPDF_HasMetaText: [['number', 'string'] as const, 'boolean'] as const,
   EPDF_IsEncrypted: [['number'] as const, 'boolean'] as const,

packages/pdfium/src/vendor/pdfium.js

@@ -30,6 +30,7 @@ var createPdfium = (() => {
       '_EPDF_GetMetaKeyCount',
       '_EPDF_GetMetaKeyName',
       '_EPDF_GetMetaTrapped',
+      '_EPDF_GetPageObjNum',
       '_EPDF_GetPageRotationByIndex',
       '_EPDF_GetPageSizeByIndexNormalized',
       '_EPDF_HasMetaText',
@@ -7709,6 +7710,8 @@ var createPdfium = (() => {
       'FPDF_GetPageSizeByIndexF',
       3,
     ));
+    var _EPDF_GetPageObjNum = (Module['_EPDF_GetPageObjNum'] =
+      createExportWrapper('EPDF_GetPageObjNum', 2));
     var _EPDF_GetPageRotationByIndex = (Module['_EPDF_GetPageRotationByIndex'] =
       createExportWrapper('EPDF_GetPageRotationByIndex', 2));
     var _EPDF_GetPageSizeByIndexNormalized = (Module['_EPDF_GetPageSizeByIndexNormalized'] =