Foxit PDF SDK Changelog v12.0

Windows/Mac/Linux

October 13, 2026

Incompatible Changes​

When upgrading from v11.1, the following changes may require code updates. Projects that are not updated may fail to compile or behave differently.

SearchCallback::NeedToCancel — New Pure Virtual Method​

A new pure virtual method was added to the full-text search callback. Missing implementations fail to compile:

// Earlier versions: method did not exist
// 12.0 and later
virtual bool NeedToCancel() = 0;

Required action: Implement the method on every SearchCallback subclass. If cancellation is not needed, return false. See Full-text search · Cooperative cancel.

WatermarkSettings Constructor and Set — Parameter Expansion​

Two parameters were appended to the constructor and Set:

// Earlier versions
WatermarkSettings(position, offset_x, offset_y, flags, scale_x, scale_y, rotation, opacity);
settings.Set(position, offset_x, offset_y, flags, scale_x, scale_y, rotation, opacity);

// 12.0 and later
WatermarkSettings(position, offset_x, offset_y, flags, scale_x, scale_y, rotation, opacity,
                  is_offset_centered, relative_scale_to_page);
settings.Set(position, offset_x, offset_y, flags, scale_x, scale_y, rotation, opacity,
             is_offset_centered, relative_scale_to_page);

Required action: Supply the new parameters per the updated signatures. When relative scaling is unused, pass relative_scale_to_page = 0 (or your product convention); when centered offset is unused, pass is_offset_centered = false. See Watermark · Relative scale.

FillerAssistCallback::Refresh — Signature Change​

The form-filler assist refresh callback signature changed:

// Earlier versions
virtual void Refresh(const PDFPage& page, const RectF* rect) = 0;

// 12.0 and later
virtual void Refresh(const PDFPage& page, const RectF& rect, bool has_reparse_page) = 0;

Required action: Update every FillerAssistCallback (or equivalent) implementation: take rect by reference and handle has_reparse_page as needed. See AcroForm · Refresh.

Preflight::Analyze / AnalyzeAndFixup — Return Type Change​

In the Compliance module, these APIs changed from returning nothing to returning ResultInformation. Parameters are unchanged; only the return type changed:

Required action: Update call sites to receive and handle the new return type. See Compliance · ResultInformation.

Callbacks required only when using the corresponding new capabilities (no change if unused):

• PreflightGroupLocalCallback: localizes group item names and comments via Preflight::SetPreflightGroupLocalCallback. Implement Release, GetGroupItemLocalName, and GetGroupItemLocalComment.

TaggedPDF / Accessibility APIs — Header Move and StartTagDocument​

PDF Accessibility APIs moved into the accessibility add-on, with AutoTagging entry-point changes:

// Earlier versions
TaggedPDF tagged(doc);
tagged.SetCallback(cb);
tagged.StartTagDocument();

// 12.0 and later
TaggedPDF tagged(doc, false);
tagged.StartTagDocument(NULL, report_stream, provider_callback);  // pause must be explicit

Required action: Update #includes; remove SetCallback and pass the report stream plus optional AutoTagProviderCallback to StartTagDocument; always pass pause explicitly. See Accessibility · Auto tag.

DWG2PDFSettingData::export_hatches_type — Default Value Change​

When left unset, the default hatch export type changed (silent behavior change):

Required action: To keep the earlier behavior, set export_hatches_type = e_DWG2PDFExportHatchesTypeBitmap explicitly; configure the new other/gradient hatch properties as needed. See DWG to PDF · Layout.

JavascriptEventType — Some Enum Numeric Values Shifted​

In earlier versions, an unused integer gap (20) existed between FieldKeyStroke (value 19) and FieldValidate. In 12.0 that gap was assigned to FieldValidate, and subsequent enum values shifted forward by 1:

Required action: Always use enum symbols (for example e_JavascriptEventTypeFieldValidate); do not hard-code integer enum values. If code uses if (type == 21) expecting FieldValidate, integer 21 maps to FieldCalculate in 12.0 and later. See JavaScript · Event types.

New Features and Enhancements​

Platform​

• Node.js (Node-API): Migrated the Node.js native binding from the legacy Node/V8 Binding to Node-API, with prebuilt packages for napi3 / napi6 / napi8, expanding compatibility (approximately Node.js 8.15–26; some historical minor versions excluded). Installation automatically matches the current runtime.

• Simple Demo command-line samples: Simple Demo supports command-line arguments. Samples are published at https://github.com/foxitsoftware/foxit-pdf-sdk-cli-samples.

Conversion​

• Markdown and PDF conversion: Added Markdown to PDF (Md2PDF::ConvertFromMarkdown) and PDF to Markdown (PDF2Md::ConvertToMarkdown). Load the Conversion library and ensure the corresponding module license before use.

• Enhanced DWG/DXF to PDF: Configure layout scope, title-block XML, auto-tiling, line-weight styles, output DPI, and more via DWG2PDFSettingData. See export_hatches_type default change above.

• Unified cross-platform Java Office-to-PDF entry points: fromWord / fromExcel / fromPowerPoint share the same method parameters; the SDK Java binding dispatches by platform internally. Existing platform-specific APIs remain available.

OCR​

• Improved OCR recognition quality (large-image handling, certain Chinese garbled-text scenarios, and related cases). Upgrade to the ABBYY OCR resource pack that ships with this release.

Signature and Security​

• ISO 32004 document integrity protection: Standalone mode (PDFDoc::StartIntegrityProtection) and signature-attached mode (StartIntegrityProtectionWithSignature), with configurable digest algorithm, password, permission bits, and whether to encrypt metadata.

Text and Content​

• Glyph ID lookup by Unicode: Font::GetGlyphID.

• Full-text search ignore whitespace in the pattern: FullTextSearch::SetIgnoreSpace; off by default; no index rebuild required.

• Full-text search supports cooperative cancellation (see NeedToCancel above)

• Watermarks support relative scaling to the target page and offset calculation from the watermark center (see WatermarkSettings above)

Rendering​

• Convert bitmaps to CMYK: Bitmap::ConvertFormat(..., e_DIBCmyk) converts RGB / ARGB / RGB32 (and related) bitmaps directly to CMYK.

• Improved CMYK page rendering performance (same-document retest: about 86% reduction on Win32, about 19% on Win64)

• Improved decoding of halftone JBIG2 images with HENABLESKIP

Forms and Annotations​

• Batch annotation hit-testing by rectangle: PDFPage::GetAnnotsAtRect / GetAnnotsAtDeviceRect, complementing existing point hit-testing; annotations with Hidden / NoView / Invisible flags are not returned.

Standards and Compliance​

• PDF/A-4e / PDF/A-4f: Added validation and conversion. Upgrade to the Compliance resource pack that ships with this release.

Accessibility​

• Accessibility check, fix, navigation, and tag tree: Publicized AccessibilityChecker, AccessibilityFixer, AccessibilityNavigator, ReadingOrder, TagTreeEditor, and related APIs.

• Improved AutoTagging structure recognition: better heading levels (such as H1/H2) and less overly flat structure trees, closer to common reader tagging results.

Document and Page​

• Improved document open performance in specific scenarios (sample about 99.7% reduction)

• Add image frames from a file path: Image::AddFrame accepts an image file path.

• SVG / WMF (EMF) vector graphics into PDF: Write SVG / WMF (EMF) into a PDF page or whole document via VectorGraphics.

Bug Fixes​

Conversion​

• Fixed font drift and incorrect font sizes in DWG2PDF when specific fonts were missing (can be mitigated/fixed by adding the corresponding .shx fonts to the engine Fonts directory)

• Fixed execution failures/crashes in Windows Java DWG2PDF caused by conflicts between the ODA export component and JDK bundled runtimes

• Fixed garbled text in parts of specific DWG drawings after conversion to PDF

• Fixed garbled text in DWG2PDF output from historical versions

• Fixed DWG2PDF output losing original drawing layer information

• Fixed line weight and related attributes not being preserved correctly after DWG2PDF conversion

• Fixed unexpected font substitution for some fonts during PDF to Word / Excel

• Fixed Office2PDF incorrectly reporting “No 3D module right” when fonts were missing; the error message is now accurate

• Fixed garbled text (shown as “?”) after converting specific DWG files on Linux

• Fixed overlapping/squeezed lines after converting specific DWG files

• Fixed crashes in Linux Java when combining OCR and PDF2Word engine workflows

Rendering​

• Fixed incorrect knockout transparency-group shape calculation that made preview results inconsistent with Adobe / WPS

• Fixed rendering of specific XFA documents that was inconsistent with Adobe

• Fixed some text appearing bolder and abnormal anti-aliasing in Output Preview (regression relative to v11.1)

• Fixed bordered link annotations not showing borders during rendering / image export

• Fixed results inconsistent with Adobe after enabling simulated overprint (EnableSimulateOverprint)

Document and Page​

• Fixed excessively long PDF/A validation and incomplete validation output after adding FreeText annotations with embedded fonts

• Fixed garbled output when writing certain Unicode symbols into headers/footers from C# (not reproducible on the C++ side)

• Fixed possible hangs or stack overflows from Bookmark::MoveTo on structurally abnormal PDFs; the API now returns errors more robustly

• Fixed unexpected results when moving a bookmark node to become a sibling of its parent

• Fixed page width/height not updating after setting PageBox when re-read (use the correct path including PDFPage::Normalize() where applicable)

OCR​

• Fixed text position drift and ghosting after OCR on specific documents

• Fixed upside-down OCR results on specific pages

• Fixed flipped or 90°-rotated OCR results on specific pages

Text and Content​

• Fixed intermittent errors from PDFPage::AddText

• Fixed invisible text after replacing Japanese with English using TextSearchReplace

• Fixed rare characters not displaying when creating text objects with SimSun-ExtB, and incorrect IsCharSupported results

• Fixed TextPage.GetText returning different results on .NET Core versus .NET Framework

• Fixed inaccurate Font::IsCharSupported results for some fonts (such as HanaMinA)

Platform​

• Fixed heap-corruption style crashes on Linux (for example Ubuntu 24) in specific licensing scenarios

• Fixed .NET on Linux ARM failing to load fsdk.dll when calling TextPage.GetChars

• Fixed document object misalignment / format corruption under concurrent editing

Signature and Security​

• Fixed incorrect position of the last character in watermark strings after calling embed() on the watermark font

• Fixed Adobe reporting errors on specific pages after EmbedFont-related flows

• Fixed Chinese tiled watermarks not displaying on Linux when fonts were loaded from a path

Printing​

• Fixed possible Access Violation on application exit if PrintManager.Release was not called

• Fixed PrintManager incorrectly using an already-closed printer handle on release/close paths

Forms and Annotations​

• Fixed missing /Length entries when exporting stamps added through the SDK to FDF

• Fixed XFADoc::FlattenTo dropping field values that contain diacritics

Optimization​

• Fixed inconsistent Optimizer handling of already-subset / non-embedded fonts that caused some fonts to be substituted