Web Analytics Made Easy - Statcounter
Skip to content

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

Added

  • Unterstützung für UI-Sprachen (Englisch en, Spanisch es, Französisch fr, Deutsch de) mit Sprachumschalter.
  • Mehrsprachige MkDocs-Dokumentation (Englisch, Spanisch, Französisch, Deutsch) unter /api/docs/site/<locale>/.
  • Kategoriebeschriftungen im Dropzone-Panel (Dokumente, Web, Bilder, Daten) jetzt vollständig internationalisiert.

Fixed

  • Die Dokumentationsnavigation zeigt jetzt vollständig lokalisierte Seitennamen in allen unterstützten Sprachen an.
  • Kategoriebeschriftungen für Dateiformate im Dropzone-Panel werden jetzt korrekt basierend auf der ausgewählten Sprache übersetzt.
  • Verbesserte Extraktion von Dokumentationsseitentiteln mit besserem Fallback auf übersetzte Namen.

Changed

  • Backend-Einstiegspunkt von app.py zu duckling.py umbenannt für bessere Klarheit.
  • Flask-Anwendungsname zu "duckling" geändert (zeigt "Serving Flask app 'duckling'").

[0.0.10] - 2026-02-24

Security

  • Frontend-Sicherheitslücken behoben (esbuild GHSA-67mh-4wv8-2f99): Vite 5→7, Vitest 1→4 und zugehörige Abhängigkeiten aktualisiert.

Fixed

  • vitest.config.ts für Vitest 4-Kompatibilität aktualisiert.
  • CI/CD Node.js-Versionsanforderung auf 22 aktualisiert (erforderlich für Vite 7).

[0.0.9] - 2026-01-08

Added

  • Custom Branding: Duckling-Logo und Versionsanzeige in der Kopfzeile.
  • URL-basierte Dokumentenkonvertierung: Konvertierung von URLs mit automatischer Bildextraktion für HTML.
  • Dokumentenanreicherungsoptionen: Code-, Formel-, Bildklassifizierung und Bildbeschreibung.
  • Enrichment-Modell-Vorab-Download: KI-Modelle vor der Verarbeitung herunterladen.
  • Bildvorschau-Galerie: Visuelle Miniaturansichten mit Lightbox-Viewer.
  • OCR-Backend-Auto-Installation: Ein-Klick-Installation für pip-installierbare Backends.
  • Format-spezifische Vorschau: Vorschau-Panel zeigt Inhalt im gewählten Exportformat.
  • Gerendert vs. Roh-Vorschau-Modus: Umschalter für HTML und Markdown.
  • Erweiterte Docker-Unterstützung: Multi-Stage-Dockerfiles, docker-compose-Varianten, Multi-Platform-Builds.

Fixed

  • Multi-Worker-Inhaltsabruf (Bilder, Tabellen, Ergebnisse).
  • HTML-Vorschau in der UI.
  • URL-Bildextraktion für nicht in Anführungszeichen gesetzte src-Attribute.
  • Dokumentations-Panel bedient jetzt vorgefertigte MkDocs-Site.
  • Umgebungsvariablen und .env-Laden.
  • Groß-/kleinschreibungsunabhängige Dateiendungen.
  • Konfidenz-Score und OCR-Backend-Auswahl.

[0.0.8] - 2026-01-07

Changed

  • Renamed: Project renamed from "Docling UI" to "Duckling"
  • Updated all documentation, code, and configuration files
  • Branding updated throughout the application

[0.0.7] - 2026-01-07

Added

  • MkDocs Documentation: Migrated documentation to MkDocs with Material theme
  • Modern, searchable documentation site
  • Dark/light theme toggle
  • Mermaid diagram support
  • Improved navigation and organization

Changed

  • Documentation structure reorganized for better navigation
  • All diagrams converted to Mermaid format for live rendering

[0.0.6] - 2025-12-11

Security

  • CRITICAL: Fixed Flask debug mode enabled by default in production
  • Debug mode now controlled by FLASK_DEBUG environment variable (default: false)
  • Host binding defaults to 127.0.0.1 instead of 0.0.0.0
  • HIGH: Updated vulnerable dependencies
  • flask-cors: 4.0.0 → 6.0.0 (CVE-2024-1681, CVE-2024-6844, CVE-2024-6866, CVE-2024-6839)
  • gunicorn: 21.2.0 → 23.0.0 (CVE-2024-1135, CVE-2024-6827)
  • werkzeug: 3.0.1 → 3.1.4 (CVE-2024-34069, CVE-2024-49766, CVE-2024-49767, CVE-2025-66221)
  • MEDIUM: Added path traversal protection to file serving endpoints
  • Image serving validates paths don't escape allowed directories
  • Blocks directory traversal sequences (..)
  • MEDIUM: Enhanced SQL query sanitization
  • Search queries now escape LIKE wildcards
  • Added query length limits
  • Added comprehensive SECURITY.md with:
  • Security audit summary
  • Production deployment checklist
  • Environment variable documentation
  • Vulnerability reporting guidelines

Changed

  • Backend now uses environment variables for all security-sensitive configuration
  • Default host changed from 0.0.0.0 to 127.0.0.1 for safer defaults

[0.0.5] - 2025-12-10

Added

  • Batch Processing: Upload and convert multiple files at once
  • Toggle batch mode in the header

  • Process multiple documents simultaneously

  • Image & Table Extraction:

  • Extract embedded images from documents
  • Download individual images
  • Extract tables with full data preservation
  • Export tables to CSV format

  • View table previews in the UI

  • RAG/Chunking Support:

  • Document chunking for RAG applications
  • Configurable max tokens per chunk (64-8192)
  • Merge peers option for undersized chunks

  • Download chunks as JSON

  • Additional Export Formats:

  • Document Tokens (.tokens.json)
  • RAG Chunks (.chunks.json)

  • Enhanced DocTags export

  • Advanced OCR Options:

  • Multiple OCR backends: EasyOCR, Tesseract, macOS Vision, RapidOCR
  • GPU acceleration support (EasyOCR)
  • Configurable confidence threshold (0-1)
  • Bitmap area threshold control

  • Support for 28+ languages

  • Table Structure Options:

  • Fast vs Accurate detection modes
  • Cell matching configuration

  • Structure extraction toggle

  • Image Generation Options:

  • Generate page images
  • Extract picture images
  • Extract table images

  • Configurable image scale (0.1x - 4.0x)

  • Performance/Accelerator Options:

  • Device selection: Auto, CPU, CUDA, MPS (Apple Silicon)
  • Thread count configuration (1-32)

  • Document timeout setting

  • New API Endpoints:

  • POST /api/convert/batch - Batch conversion
  • GET /api/convert/<job_id>/images - List extracted images
  • GET /api/convert/<job_id>/images/<id> - Download image
  • GET /api/convert/<job_id>/tables - List extracted tables
  • GET /api/convert/<job_id>/tables/<id>/csv - Download table CSV
  • GET /api/convert/<job_id>/tables/<id>/image - Download table image
  • GET /api/convert/<job_id>/chunks - Get document chunks
  • GET/PUT /api/settings/performance - Performance settings
  • GET/PUT /api/settings/chunking - Chunking settings
  • GET /api/settings/formats - List all supported formats

Changed

  • Settings Panel: Completely redesigned with all new options
  • Export Options: Enhanced with tabs for different content types
  • DropZone: Updated with format categories and batch mode support
  • Converter Service: Major refactoring for dynamic pipeline options

Fixed

  • Confidence score calculation now uses cluster-level predictions
  • Better handling of partial conversion success

[0.0.4] - 2025-12-10

Added

  • OCR Support: Full OCR integration using EasyOCR
  • Support for 14+ languages
  • Force Full Page OCR option
  • Configurable OCR settings
  • Improved Confidence Calculation: Average confidence from layout predictions

Changed

  • Updated converter service to use configurable pipeline options
  • Enhanced settings panel with OCR options

[0.0.3] - 2025-12-10

Added

  • Initial release of Duckling
  • Frontend Features:
  • Drag-and-drop file upload
  • Real-time conversion progress
  • Multi-format export options
  • Settings panel
  • Conversion history panel
  • Dark theme with teal accent
  • Responsive design

  • Animated transitions

  • Backend Features:

  • Flask REST API with CORS
  • Async document conversion
  • SQLite-based history
  • File upload management
  • Configurable settings

  • Health check endpoint

  • Supported Input Formats:

  • PDF, Word, PowerPoint, Excel
  • HTML, Markdown, CSV
  • Images (PNG, JPG, TIFF, etc.)

  • AsciiDoc, XML

  • Export Formats:

  • Markdown, HTML, JSON

  • DocTags, Plain Text

  • Developer Experience:

  • Comprehensive test suites
  • Docker support
  • TypeScript
  • ESLint configuration

Security

  • Input validation for file uploads
  • File type restrictions
  • Maximum file size limits
  • Secure filename handling

[Unreleased]

Planned

  • User authentication
  • Cloud storage integration
  • Conversion templates
  • API rate limiting
  • WebSocket for real-time updates
  • Dark/light theme toggle
  • Keyboard shortcuts
  • Accessibility improvements (WCAG 2.1)