Dagochim's tools run in your browser thanks to the open-source software below. Each library and model is served from this site without modification (for a few, only the export line was changed so they can be used as a global variable). Fonts for the HWP viewer and the PDF character maps come in many small slices, so we bundle the unchanged original files into a few pack files and serve them one by one. AI model files are downloaded from the Dagochim family model storage (models.dagotools.com) when you start an AI feature. They are the original files from Hugging Face, uploaded unchanged; if our storage can't be reached, your browser downloads them from Hugging Face instead. Full license texts are included next to each library as LICENSE files in /assets/vendor/.
| Name | Version | License | Used for |
|---|---|---|---|
| pdf.js | 3.11.174 | Apache-2.0 | Opening, previewing and rendering PDFs |
| pdf.js CMaps | 3.11.174 | Apache-2.0 (Adobe CMaps: BSD-style) | Displaying PDFs without embedded Korean/CJK fonts (the 168 original files are bundled into one file and served from this site) |
| pdf-lib | 1.17.1 | MIT | Merging, splitting and editing PDFs |
| @pdf-lib/fontkit | - | MIT | Embedding fonts in PDFs |
| qpdf (qpdf-wasm 0.3.0) | 12.2.0 | Apache-2.0 (wrapper ISC; includes zlib and libjpeg-turbo) | Removing a PDF password you already know |
| rhwp | 0.8.6 | MIT | Reading HWP/HWPX files, editing and saving (HWP editor) |
| libarchive.js | 2.0.2 | MIT (libarchive: BSD and others) | Extracting ZIP, 7Z and RAR archives |
| fflate | 0.8.2 | MIT | Creating ZIP files |
| tesseract.js / tesseract.js-core | 7.0.0 | Apache-2.0 | Text recognition (OCR) |
| tessdata_fast (kor, eng, spa, por, fra, deu) | 4.1.0 | Apache-2.0 | OCR language data (served from this site; only the language you pick is downloaded, on first use) |
| PP-OCRv5 models (PP-OCRv5_mobile_det, korean_PP-OCRv5_mobile_rec) · PaddlePaddle Authors | v5 | Apache-2.0 (full text) | Text detection and reading models for the "Accuracy first" OCR mode and photo/scanned table reading (downloaded only when used) |
| RapidOCR ONNX files · RapidOCR Authors | 3.10.0 | Apache-2.0 (full text) | ONNX files of the models above (used unmodified) |
| ONNX Runtime Web | 1.30.0 | MIT (full text, third-party notices) | Runs the "Accuracy first" OCR and photo/scanned table models |
| mp4box.js | 0.5.2 | BSD-3-Clause | Fast video compression (MP4 demuxing) |
| mp4-muxer | 5.1.3 | MIT | Fast video compression (MP4 muxing) |
| MediaPipe Tasks Vision | 0.10.21 | Apache-2.0 | Face detection and person segmentation (ID photo, face mosaic) |
| MediaPipe models (blaze_face_short_range, selfie_segmenter) | - | Apache-2.0 | Face detection and person segmentation models |
| ONNX Runtime Web | - | MIT | AI model runtime |
| Transformers.js | - | Apache-2.0 | Loading AI models in the browser |
| U-2-Netp | - | Apache-2.0 | Background remover (default model) |
| BiRefNet_lite | - | MIT | Background remover HD model (downloaded only when you turn it on) |
| Qwen2.5-0.5B-Instruct | - | Apache-2.0 | English grammar AI polish (downloaded only when you turn it on) |
| opus-mt-ko-en | - | Apache-2.0 | PDF translate, Korean to English fast mode (downloaded only when used) |
| M2M100 418M | - | MIT | PDF translate, other language pairs (downloaded only when used) |
| Whisper (base, small) | - | MIT | Shorts maker auto captions (downloaded only when used) |
| Mammoth.js (browser bundle) | 1.8.0 | BSD-2-Clause | Reading Word (DOCX) files. The bundle contains the components in the next row |
| Components bundled inside Mammoth: JSZip, pako, lie, immediate, readable-stream, underscore, bluebird, @xmldom/xmldom, xmlbuilder, base64-js, dingbat-to-unicode, lop | as bundled | Mostly MIT. JSZip is dual-licensed MIT or GPL-3.0 and we use it under the MIT license. pako: MIT and zlib terms. dingbat-to-unicode and lop: BSD-2-Clause | Shipped inside the Mammoth bundle, not called separately. Full text: /assets/vendor/mammoth.LICENSE.txt |
| Dagochim MP3 encoder (our own) | 1.0.0 | Our own code, all rights reserved | MP3 export (Huffman table data from go-mp3, Apache-2.0) |
| gifenc | 1.0.3 | MIT | Creating GIFs |
| fix-webm-duration | 1.0.6 | MIT | Fixing duration metadata in screen recordings |
| qrcode-generator | 2.0.4 | MIT | QR code generator |
| jsQR | 1.4.0 | Apache-2.0 | QR code reader |
| JsBarcode | 3.11.6 | MIT | Barcode generator |
| Pretendard | 1.3.9 | SIL OFL 1.1 | Site font (official files served unchanged from this site) |
| Noto Sans KR, Nanum Myeongjo | Fontsource 5.3.0 | SIL OFL 1.1 | Fonts for viewing HWP files (original slices bundled into a few files and served from this site) |
| Nanum Gothic | - | SIL OFL 1.1 | Korean font embedded in PDFs we create |
| Noto Sans (Latin subset) | - | SIL OFL 1.1 | Accented Latin letters (á, ñ, ç, ã, ü and so on) in PDFs we create, used only for characters Nanum Gothic does not have. Full text: /assets/vendor/fonts/OFL-NotoSans.txt |
| Lucide icons | 1.52.0 | ISC | Site icons |
Fonts
All fonts are used under the SIL Open Font License 1.1 and served unchanged from this site. Pretendard declares the Reserved Font Name "Pretendard"; we only distribute its official, unmodified files, so the name is kept as is.
Model mirror (models.dagotools.com) mapping
AI model files are unmodified copies of the Hugging Face originals, kept on our family model store (models.dagotools.com). Tools try the store first and fall back to Hugging Face. A store path equals the Hugging Face URL with huggingface.co replaced by models.dagotools.com (/<repo>/resolve/main/<file>). The full license text of each model is in the LICENSE file or model card at its original source.
| Model repo | License | Used for |
|---|---|---|
| onnx-community/Qwen2.5-0.5B-Instruct (Qwen2.5, © Alibaba Cloud) | Apache-2.0 | English grammar polish |
| Xenova/opus-mt-ko-en (original: Helsinki-NLP/opus-mt-ko-en) | Apache-2.0 (per the original model card) | PDF translation, Korean to English |
| Xenova/m2m100_418M (original: facebook/m2m100_418M) | MIT | PDF translation, other languages |
| Xenova/whisper-base, whisper-small (original: openai/whisper) | MIT (per original Whisper; the converted model cards may not state a license) | Shorts auto-captions |
MP3 export (our own encoder)
MP3 export uses an encoder Dagochim wrote from the MPEG audio specification (ISO/IEC 11172-3 and 13818-3, Layer III): /assets/ttdu/dago-audio.js, our own code, all rights reserved. We don't use GPL/LGPL code such as LAME or lamejs. All MP3-related patents expired in 2017. The Huffman code table values defined by the specification were taken from the table data published by go-mp3 (Apache-2.0). Short blocks are used where the sound changes suddenly, such as percussion, to reduce pre-echo.
HWP document format notice
This product was developed with reference to Hancom's publicly released Hangul document file (.hwp) format documents.
"Hangul", "Hancom", "HWP" and "HWPX" are registered trademarks of Hancom Inc. Dagochim and rhwp are not affiliated with, sponsored or endorsed by Hancom.
Dagochim PC programs
The Windows programs on the PC apps page are written entirely in our own code and contain no third-party libraries or GPL/LGPL code. They only call components already built into Windows (.NET Framework 4.8, Media Foundation, WASAPI and built-in commands). Each download includes a license notice (라이선스.txt).
Trademarks
Company and product names on this page and in our tool descriptions are trademarks of their respective owners and are used only to identify the source.
Korean version: 오픈소스 라이선스 · Last checked: October 6, 2026