Support, without the guesswork.

PDF → PNG · mupdf

Render selected PDF pages as separate PNG images.

• Raster output has no selectable text. Maximum 40 megapixels per rendered page.

PDF → JPG · mupdf

Render selected PDF pages as separate JPEG images.

• Raster output has no selectable text.

PDF → TXT · mupdf

Extract embedded text in page order.

• Scans need OCR; this tool does not recognize text in images. Reading order follows the PDF.

PDF → PDF · mupdf

Join PDF files in queue order.

• PDF page content is copied; document outlines, signatures and document-level forms are not merged.

PDF → PDF · mupdf

Export selected pages in the order entered.

• Document signatures become invalid after editing.

PDF → PDF · mupdf

Export each selected page as a separate PDF.

PDF → PDF · mupdf

Rotate selected pages in 90-degree increments.

PDF → PDF · mupdf

Rewrite PDF streams and remove unused objects.

• Lossless structural optimization; image resolution is unchanged and files may grow.

PDF → PDF · mupdf

Save a PDF with AES-256 encryption.

• Use different strong opening and owner passwords. Passwords containing commas or equals signs are unsupported.

PDF → PDF · mupdf

Remove encryption after authenticating with a known password.

PDF → PDF · mupdf

Remove text, image pixels and touched vector art in a selected rectangle.

• Coordinates start at the top-left of the displayed page, in PDF points (72 per inch). Redaction is irreversible in the exported PDF. Check the output before sharing.

PDF → PDF · mupdf

Remove every matching text occurrence on selected pages and insert replacement text.

• Original text is redacted before replacement. Layout is not reflowed; replacement must fit. Standard Latin fonts only.

PDF → PDF · mupdf

Add text at a chosen position and bake it into the page.

• Typed signature is a visual mark, not a cryptographic signature. Helvetica supports standard Latin text.

PDF, PNG, JPG, JPEG → PDF · mupdf

Place a raster image on a PDF page.

• Queue exactly one PDF and one PNG/JPEG image. Image is fitted into the selected rectangle. Signatures are visual marks.

PDF, PNG, JPG, JPEG → PDF · mupdf

Remove images intersecting a selected rectangle, then insert a replacement.

• Queue exactly one PDF and one PNG/JPEG image. Any image touching the rectangle is removed in full before replacement. Text and vector art remain.

PDF → PDF · mupdf

Add a comment, highlight or rectangle annotation.

PDF → PDF · mupdf

Set text and choice fields by field name from a JSON object.

• Supports existing text and choice fields. Checkbox, signature and XFA forms are excluded. Unknown field names are errors.

PNG, JPG, JPEG, WEBP, TIF, TIFF, BMP, GIF → PDF · mupdf

Create one PDF page per input image.

• Each page uses the first image frame. Images retain their pixel dimensions as PDF points.

PDF, PNG, JPG, JPEG, BMP, TIF, TIFF, WEBP → TXT · tesseract

Recognize English text in scanned pages or raster images.

• English model only. Recognition is approximate; inspect the text before relying on it.
• PDF pages are rendered at 144 DPI. Maximum 20 pages per file. Images use the first frame.

PDF, PNG, JPG, JPEG, BMP, TIF, TIFF, WEBP → PDF · tesseract

Recognize English text and export a PDF with an invisible searchable text layer.

• English model only. Recognition is approximate; inspect the text before relying on it.
• Pages are rasterized; original vectors, forms, links and annotations become pixels. Maximum 20 pages per file.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → PNG · imagemagick

Resize an image to fit within a width and height, preserving aspect ratio; download a PNG.

• Only the first frame is processed. Output is PNG; editable layers are not preserved.
• Maximum 40 megapixels per frame, including after resizing and rotation. Metadata and color profiles may change.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → PNG · imagemagick

Crop an image to a rectangle specified in pixels and download a PNG.

• Only the first frame is processed. Output is PNG; editable layers are not preserved.
• Maximum 40 megapixels per frame, including after resizing and rotation. Metadata and color profiles may change.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → PNG · imagemagick

Rotate an image clockwise by an angle in degrees and download a PNG.

• Only the first frame is processed. Output is PNG; editable layers are not preserved.
• Maximum 40 megapixels per frame, including after resizing and rotation. Metadata and color profiles may change.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → PNG · imagemagick

Mirror an image horizontally or flip it vertically, then download a PNG.

• Only the first frame is processed. Output is PNG; editable layers are not preserved.
• Maximum 40 megapixels per frame, including after resizing and rotation. Metadata and color profiles may change.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → PNG · imagemagick

Convert an image to grayscale and download a PNG.

• Only the first frame is processed. Output is PNG; editable layers are not preserved.
• Maximum 40 megapixels per frame, including after resizing and rotation. Metadata and color profiles may change.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → PNG · imagemagick

Remove metadata and color profiles from an image and download a PNG.

• Only the first frame is processed. Output is PNG; editable layers are not preserved.
• Maximum 40 megapixels per frame, including after resizing and rotation. Metadata and color profiles may change.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → JPG · imagemagick

Create a JPEG with adjustable lossy quality. The output may be larger than the original.

• Only the first frame is processed. Output is lossy JPEG; transparency and editable layers are not preserved.
• Maximum 40 megapixels per frame, including after resizing and rotation. Metadata and color profiles may change.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → PNG · imagemagick

Convert raster images to PNG; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → JPG · imagemagick

Convert raster images to JPEG; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → WEBP · imagemagick

Convert raster images to WEBP; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → GIF · imagemagick

Convert raster images to GIF; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → TIFF · imagemagick

Convert raster images to TIFF; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → BMP · imagemagick

Convert raster images to BMP; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → AVIF · imagemagick

Convert raster images to AVIF; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → JXL · imagemagick

Convert raster images to JXL; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → JP2 · imagemagick

Convert raster images to JP2; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → QOI · imagemagick

Convert raster images to QOI; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → PSD · imagemagick

Convert raster images to PSD; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → EXR · imagemagick

Convert raster images to EXR; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → HDR · imagemagick

Convert raster images to HDR; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

PNG, JPG, WEBP, GIF, TIFF, BMP, AVIF, JXL, JP2, QOI, PSD, EXR, HDR, PPM, JPEG, TIF → PPM · imagemagick

Convert raster images to PPM; resize, crop and rotate locally.

• Only verified raster formats are advertised. Layered files are rendered; editable layers are not preserved.
• Metadata and color may change during conversion. Animation is preserved only with All frames and GIF, WebP or TIFF output. Maximum 500 frames and 40 megapixels per frame.

MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → PNG · ffmpeg

Extract video frames as a PNG image sequence with adjustable frame rate, time range and image width.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

PNG, JPG, JPEG, WEBP, BMP → MP4 · ffmpeg

Turn an ordered image sequence into an MP4 slideshow with adjustable frame timing.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → MP4 · ffmpeg

Convert video to MP4, with trimming and resizing controls.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → WEBM · ffmpeg

Convert video to WEBM, with trimming and resizing controls.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → GIF · ffmpeg

Convert video to GIF, with trimming and resizing controls.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → PNG · ffmpeg

Save a still image from a video at a selected timestamp as a PNG.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP3, WAV, FLAC, OGG, OPUS, M4A, AAC, AIFF, WMA, MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → MP3 · ffmpeg

Convert audio or extract the first video audio track as MP3.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP3, WAV, FLAC, OGG, OPUS, M4A, AAC, AIFF, WMA, MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → WAV · ffmpeg

Convert audio or extract the first video audio track as WAV.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP3, WAV, FLAC, OGG, OPUS, M4A, AAC, AIFF, WMA, MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → FLAC · ffmpeg

Convert audio or extract the first video audio track as FLAC.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP3, WAV, FLAC, OGG, OPUS, M4A, AAC, AIFF, WMA, MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → OGG · ffmpeg

Convert audio or extract the first video audio track as OGG.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP3, WAV, FLAC, OGG, OPUS, M4A, AAC, AIFF, WMA, MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → OPUS · ffmpeg

Convert audio or extract the first video audio track as OPUS.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

MP3, WAV, FLAC, OGG, OPUS, M4A, AAC, AIFF, WMA, MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, OGV, TS → M4A · ffmpeg

Convert audio or extract the first video audio track as M4A.

• Single-threaded FFmpeg. Large files can exceed browser memory; total job input is limited to 128 MiB.
• DRM is unsupported. Decoder support depends on the source codec; conversion failure is reported. First video/audio stream only; subtitles, chapters and attachments are not copied.

TXT, MD → DOCX · OOXML / fflate

Create a Word document from plain text paragraphs.

• Plain text paragraphs only; Markdown syntax remains literal.

DOCX → TXT · OOXML / fflate

Extract document body text and table cells from Word files.

• Headers, footers, images, comments, tracked-change interpretation and formatting are not preserved. Encrypted documents are unsupported.

PDF, PNG, JPG, JPEG, BMP, TIF, TIFF, WEBP → DOCX · tesseract

Recognize English text and create a plain-text Office document.

• English model only. Recognition is approximate; inspect the text before relying on it.
• PDF pages are rendered at 144 DPI. Maximum 20 pages per file. Images use the first frame.
• Text-only reconstruction: recognized lines become paragraphs. Original columns, fonts, page layout, tables and images are omitted.

DOC, DOCX, ODT, RTF, TXT, HTML, HTM → PDF · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

DOC, DOCX, ODT, RTF, TXT → DOCX · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

DOC, DOCX, ODT, RTF, TXT → ODT · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

DOC, DOCX, ODT, RTF, TXT → RTF · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

DOC, DOCX, ODT, RTF, TXT → TXT · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.
• Plain text discards layout, images and formatting.

CSV → XLSX · OOXML / fflate

Create a single-sheet Excel workbook from delimited text.

• All cells are written as text, preserving leading zeros and preventing formula injection. No formatting or formula calculation. 100,000 rows / 2 million cells maximum.

TSV → XLSX · OOXML / fflate

Create a single-sheet Excel workbook from delimited text.

• All cells are written as text, preserving leading zeros and preventing formula injection. No formatting or formula calculation. 100,000 rows / 2 million cells maximum.

XLSX → CSV · OOXML / fflate

Export an explicitly selected worksheet from an Excel workbook.

• Enter the exact worksheet name; a missing selection lists available names. Formulas use cached values and fail if none exist. Dates remain Excel serial numbers. Formatting, charts and macros are omitted. JSON is an array of rows. 100,000 rows / 2 million cells maximum.

XLSX → TSV · OOXML / fflate

Export an explicitly selected worksheet from an Excel workbook.

• Enter the exact worksheet name; a missing selection lists available names. Formulas use cached values and fail if none exist. Dates remain Excel serial numbers. Formatting, charts and macros are omitted. JSON is an array of rows. 100,000 rows / 2 million cells maximum.

XLSX → JSON · OOXML / fflate

Export an explicitly selected worksheet from an Excel workbook.

• Enter the exact worksheet name; a missing selection lists available names. Formulas use cached values and fail if none exist. Dates remain Excel serial numbers. Formatting, charts and macros are omitted. JSON is an array of rows. 100,000 rows / 2 million cells maximum.

PDF, PNG, JPG, JPEG, BMP, TIF, TIFF, WEBP → XLSX · tesseract

Recognize English text and infer rows and columns from explicit separators.

• English model only. Recognition is approximate; inspect the text before relying on it.
• PDF pages are rendered at 144 DPI. Maximum 20 pages per file. Images use the first frame.
• Heuristic rows and columns only. No table detection, formulas, types or formatting. Review every cell; all values are stored as text.

XLS, XLSX, ODS, CSV → PDF · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

XLS, XLSX, ODS, CSV → XLSX · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

XLS, XLSX, ODS, CSV → ODS · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

TXT, MD → PPTX · PptxGenJS

Create readable presentation slides from plain text.

• A blank line separates slides. Long slides are split after 12 lines. Plain text only; Markdown remains literal.

PPTX → TXT · OOXML / fflate

Extract slide text in presentation order.

• Text only, in document shape order. Slide layouts, charts, speaker notes, diagrams, animations and image text are omitted.

PDF, PNG, JPG, JPEG, BMP, TIF, TIFF, WEBP → PPTX · tesseract

Recognize English text and create a plain-text Office document.

• English model only. Recognition is approximate; inspect the text before relying on it.
• PDF pages are rendered at 144 DPI. Maximum 20 pages per file. Images use the first frame.
• Text-only reconstruction: a blank line separates slides, long blocks split after 12 lines. Original slide layout, images, charts and animations are omitted.

PPT, PPTX, ODP → PDF · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

PPT, PPTX, ODP → PPTX · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

PPT, PPTX, ODP → ODP · office

Render and export locally with LibreOffice's ZetaOffice browser build.

• Desktop browsers with cross-origin isolation and shared WebAssembly memory are required. The engine downloads about 250 MiB and allocates 1 GiB of shared memory.
• Text input requires UTF-8. Self-contained HTML is accepted for PDF output only. HTML export is unavailable in this stock build.
• Jobs accept at most 32 MiB across 8 files. Each file is limited to 16 MiB; expanded OOXML/OpenDocument content is limited to 64 MiB.
• Macros and external-resource access are disabled. Password-protected files and documents containing external links are rejected.
• The approved stock build includes its fonts. Unavailable fonts are substituted; unsupported Office features and animations may change or be lost.

ZIP, RAR, 7Z, TAR, GZ, BZ2, XZ, TGZ → ZIP · libarchive

Extract and rebuild an archive in a different container.

• One output per input archive. Maximum 2,000 entries / 128 MiB expanded size. Encryption, links, special files, permissions and timestamps are not preserved. RAR is input-only.

ZIP, RAR, 7Z, TAR, GZ, BZ2, XZ, TGZ → GZ · libarchive

Extract and rebuild an archive in a different container.

• One output per input archive. Maximum 2,000 entries / 128 MiB expanded size. Encryption, links, special files, permissions and timestamps are not preserved. RAR is input-only.

RAR, 7Z, TAR, GZ, BZ2, XZ, TGZ → * · libarchive

Extract files from common archive containers locally.

• RAR is input-only. Maximum 2,000 entries / 128 MiB expanded size. Encrypted archives, links and special filesystem entries are rejected.

* → GZ · libarchive

Bundle selected files into a gzip-compressed USTAR archive.

• USTAR filename and size restrictions apply. Duplicate names are rejected. No encryption.

* → ZIP · zip.js

Bundle the selected files into one compressed ZIP archive.

• 128 MiB maximum total input. Duplicate names must be renamed before archiving. ZIP encryption and split archives are unsupported.

ZIP → * · fflate

Extract files from a ZIP archive locally.

• Maximum 2,000 entries and 128 MiB expanded size. Encryption, split archives and compression methods other than stored/deflate are unsupported. Unsafe paths are rejected.

TXT, MD → EPUB · EPUB 3 / fflate

Create a reflowable EPUB 3 book with navigation.

• One chapter. Markdown is treated as plain text. No illustrations or rich text. Language metadata is English.

EPUB → PDF · pdf-lib

Create a searchable, simply formatted PDF from readable text.

• Plain text reflow with Helvetica and Western European characters only. Other scripts fail explicitly. Illustrations, book layout and styling are omitted. Maximum 500 pages.

FB2 → PDF · pdf-lib

Create a searchable, simply formatted PDF from readable text.

• Plain text reflow with Helvetica and Western European characters only. Other scripts fail explicitly. Illustrations, book layout and styling are omitted. Maximum 500 pages.

EPUB → TXT · XML / fflate

Extract readable chapter text in reading order.

• Text extraction only. Layout, styles, illustrations and notes metadata are discarded. DRM and non-XHTML EPUB chapters are unsupported.

FB2 → TXT · XML / fflate

Extract readable chapter text in reading order.

• Text extraction only. Layout, styles, illustrations and notes metadata are discarded. DRM and non-XHTML EPUB chapters are unsupported.

SRT → VTT · Subtitle parser

Convert subtitle timing and plain cue text.

• Styling, positioning, karaoke, speaker metadata and embedded fonts are discarded. ASS output rounds timing to centiseconds.

SRT → ASS · Subtitle parser

Convert subtitle timing and plain cue text.

• Styling, positioning, karaoke, speaker metadata and embedded fonts are discarded. ASS output rounds timing to centiseconds.

SRT → TXT · Subtitle parser

Extract spoken text from subtitle cues.

• Styling, positioning, karaoke, speaker metadata and embedded fonts are discarded. ASS output rounds timing to centiseconds.

VTT → SRT · Subtitle parser

Convert subtitle timing and plain cue text.

• Styling, positioning, karaoke, speaker metadata and embedded fonts are discarded. ASS output rounds timing to centiseconds.

VTT → ASS · Subtitle parser

Convert subtitle timing and plain cue text.

• Styling, positioning, karaoke, speaker metadata and embedded fonts are discarded. ASS output rounds timing to centiseconds.

VTT → TXT · Subtitle parser

Extract spoken text from subtitle cues.

• Styling, positioning, karaoke, speaker metadata and embedded fonts are discarded. ASS output rounds timing to centiseconds.

ASS, SSA → SRT · Subtitle parser

Convert subtitle timing and plain cue text.

• Styling, positioning, karaoke, speaker metadata and embedded fonts are discarded. ASS output rounds timing to centiseconds.

ASS, SSA → VTT · Subtitle parser

Convert subtitle timing and plain cue text.

• Styling, positioning, karaoke, speaker metadata and embedded fonts are discarded. ASS output rounds timing to centiseconds.

ASS, SSA → TXT · Subtitle parser

Extract spoken text from subtitle cues.

• Styling, positioning, karaoke, speaker metadata and embedded fonts are discarded. ASS output rounds timing to centiseconds.

TTF → WOFF · fonteditor-core

Rebuild a font for desktop or web use.

• Variable fonts and color fonts are unsupported. CFF outlines are converted to TrueType outlines. Unsupported tables, layout features, digital signatures and some metadata may be lost. Check your font license before conversion.

TTF → WOFF2 · woff2

Rebuild a font for desktop or web use.

• Variable fonts and color fonts are unsupported. CFF outlines are converted to TrueType outlines. Unsupported tables, layout features, digital signatures and some metadata may be lost. Check your font license before conversion.

OTF → TTF · fonteditor-core

Rebuild a font for desktop or web use.

• Variable fonts and color fonts are unsupported. CFF outlines are converted to TrueType outlines. Unsupported tables, layout features, digital signatures and some metadata may be lost. Check your font license before conversion.

OTF → WOFF · fonteditor-core

Rebuild a font for desktop or web use.

• Variable fonts and color fonts are unsupported. CFF outlines are converted to TrueType outlines. Unsupported tables, layout features, digital signatures and some metadata may be lost. Check your font license before conversion.

OTF → WOFF2 · woff2

Rebuild a font for desktop or web use.

• Variable fonts and color fonts are unsupported. CFF outlines are converted to TrueType outlines. Unsupported tables, layout features, digital signatures and some metadata may be lost. Check your font license before conversion.

WOFF → TTF · fonteditor-core

Rebuild a font for desktop or web use.

• Variable fonts and color fonts are unsupported. CFF outlines are converted to TrueType outlines. Unsupported tables, layout features, digital signatures and some metadata may be lost. Check your font license before conversion.

WOFF → WOFF2 · woff2

Rebuild a font for desktop or web use.

• Variable fonts and color fonts are unsupported. CFF outlines are converted to TrueType outlines. Unsupported tables, layout features, digital signatures and some metadata may be lost. Check your font license before conversion.

WOFF2 → TTF · woff2

Rebuild a font for desktop or web use.

• Variable fonts and color fonts are unsupported. CFF outlines are converted to TrueType outlines. Unsupported tables, layout features, digital signatures and some metadata may be lost. Check your font license before conversion.

WOFF2 → WOFF · woff2

Rebuild a font for desktop or web use.

• Variable fonts and color fonts are unsupported. CFF outlines are converted to TrueType outlines. Unsupported tables, layout features, digital signatures and some metadata may be lost. Check your font license before conversion.

JSON → CSV · Structured text

Convert JSON locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

JSON → TSV · Structured text

Convert JSON locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

JSON → YAML · Structured text

Convert JSON locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

JSON → XML · Structured text

Convert JSON locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.
• XML uses one root property, @_ attributes, and #text content. DTDs and entity declarations are rejected; mixed content order is not preserved.

CSV → JSON · Structured text

Convert CSV locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

CSV → TSV · Structured text

Convert CSV locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

CSV → YAML · Structured text

Convert CSV locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

CSV → XML · Structured text

Convert CSV locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.
• XML uses one root property, @_ attributes, and #text content. DTDs and entity declarations are rejected; mixed content order is not preserved.
• Tabular XML uses a root containing record elements. CSV/TSV export produces records/record/field elements with column names as attributes; XML import also accepts records with named child elements.

TSV → JSON · Structured text

Convert TSV locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

TSV → CSV · Structured text

Convert TSV locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

TSV → YAML · Structured text

Convert TSV locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

TSV → XML · Structured text

Convert TSV locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.
• XML uses one root property, @_ attributes, and #text content. DTDs and entity declarations are rejected; mixed content order is not preserved.
• Tabular XML uses a root containing record elements. CSV/TSV export produces records/record/field elements with column names as attributes; XML import also accepts records with named child elements.

YAML, YML → JSON · Structured text

Convert YAML locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

YAML, YML → CSV · Structured text

Convert YAML locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

YAML, YML → TSV · Structured text

Convert YAML locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.

YAML, YML → XML · Structured text

Convert YAML locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.
• XML uses one root property, @_ attributes, and #text content. DTDs and entity declarations are rejected; mixed content order is not preserved.

XML → JSON · Structured text

Convert XML locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.
• XML uses one root property, @_ attributes, and #text content. DTDs and entity declarations are rejected; mixed content order is not preserved.

XML → CSV · Structured text

Convert XML locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.
• XML uses one root property, @_ attributes, and #text content. DTDs and entity declarations are rejected; mixed content order is not preserved.
• Tabular XML uses a root containing record elements. CSV/TSV export produces records/record/field elements with column names as attributes; XML import also accepts records with named child elements.

XML → TSV · Structured text

Convert XML locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.
• XML uses one root property, @_ attributes, and #text content. DTDs and entity declarations are rejected; mixed content order is not preserved.
• Tabular XML uses a root containing record elements. CSV/TSV export produces records/record/field elements with column names as attributes; XML import also accepts records with named child elements.

XML → YAML · Structured text

Convert XML locally with an explicit structured-data mapping.

• UTF-8 text; maximum total input size is 128 MiB. CSV and TSV use a header row; values stay strings on import, and nested values become JSON strings on export.
• XML uses one root property, @_ attributes, and #text content. DTDs and entity declarations are rejected; mixed content order is not preserved.

ICS → JSON · Local text parser

Export iCalendar event records to JSON without uploading your calendar.

• UTF-8 iCalendar VEVENT records only. Recurrences are reported, not expanded; timestamps and timezone parameters stay literal. Alarms and nested components are omitted. JSON preserves top-level properties; CSV exports selected event fields. Maximum 64 MiB / 100,000 records.

VCF, VCARD → JSON · Local text parser

Export UTF-8 vCard contacts to JSON locally.

• vCard 3.0 and 4.0 only; version 2.1, quoted-printable and embedded binary properties are unsupported. JSON preserves property names, parameters and text; CSV exports selected contact fields, with repeated fields represented as JSON arrays. Maximum 64 MiB / 100,000 contacts.

HAR → JSON · Local text parser

Summarize HAR request URLs, methods, response statuses, timing and sizes as JSON.

• Maximum 64 MiB / 100,000 requests. Request and response bodies, cookies and headers are omitted. URLs may contain sensitive query parameters; the summary stays local. Missing or unavailable HAR sizes retain their original values.

ICS → CSV · Local text parser

Export iCalendar event records to CSV without uploading your calendar.

• UTF-8 iCalendar VEVENT records only. Recurrences are reported, not expanded; timestamps and timezone parameters stay literal. Alarms and nested components are omitted. JSON preserves top-level properties; CSV exports selected event fields. Maximum 64 MiB / 100,000 records.

VCF, VCARD → CSV · Local text parser

Export UTF-8 vCard contacts to CSV locally.

• vCard 3.0 and 4.0 only; version 2.1, quoted-printable and embedded binary properties are unsupported. JSON preserves property names, parameters and text; CSV exports selected contact fields, with repeated fields represented as JSON arrays. Maximum 64 MiB / 100,000 contacts.

HAR → CSV · Local text parser

Summarize HAR request URLs, methods, response statuses, timing and sizes as CSV.

• Maximum 64 MiB / 100,000 requests. Request and response bodies, cookies and headers are omitted. URLs may contain sensitive query parameters; the summary stays local. Missing or unavailable HAR sizes retain their original values.

SQL, DUMP, BACKUP → JSON · PostgreSQL COPY reader

List table names, columns and row counts from a PostgreSQL plain SQL COPY dump.

• Plain UTF-8 SQL dumps with text-mode COPY FROM stdin data only. Custom PGDMP, tar/directory archives, INSERT-based dumps and non-UTF-8 dumps are unsupported; use pg_restore or pg_dump to export plain SQL first. SQL is never executed.
• Maximum 64 MiB, 1,000 tables, 100,000 rows per table and 2 million total cells. Values remain text; JSON preserves SQL NULL, while CSV represents NULL as an empty cell.

SQL, DUMP, BACKUP → CSV · PostgreSQL COPY reader

Extract one table from a PostgreSQL plain SQL COPY dump as CSV.

• Plain UTF-8 SQL dumps with text-mode COPY FROM stdin data only. Custom PGDMP, tar/directory archives, INSERT-based dumps and non-UTF-8 dumps are unsupported; use pg_restore or pg_dump to export plain SQL first. SQL is never executed.
• Maximum 64 MiB, 1,000 tables, 100,000 rows per table and 2 million total cells. Values remain text; JSON preserves SQL NULL, while CSV represents NULL as an empty cell.

SQL, DUMP, BACKUP → JSON · PostgreSQL COPY reader

Extract one table from a PostgreSQL plain SQL COPY dump as JSON.

• Plain UTF-8 SQL dumps with text-mode COPY FROM stdin data only. Custom PGDMP, tar/directory archives, INSERT-based dumps and non-UTF-8 dumps are unsupported; use pg_restore or pg_dump to export plain SQL first. SQL is never executed.
• Maximum 64 MiB, 1,000 tables, 100,000 rows per table and 2 million total cells. Values remain text; JSON preserves SQL NULL, while CSV represents NULL as an empty cell.

SQLITE, SQLITE3, DB → JSON · sqlite

List SQLite tables, columns and stored schema as a local JSON report.

• SQLite 3 files only, up to 64 MiB. Encrypted databases and sidecar WAL files are not supported. Views are listed but not evaluated.

SQLITE, SQLITE3, DB → CSV · sqlite

Export one explicitly selected SQLite table to CSV in your browser.

• SQLite 3 files only, up to 64 MiB; select a table, not a view. BLOB values use an explicit hexadecimal object in JSON and JSON object text in CSV. Integers outside the JavaScript safe range use decimal strings; smaller integers remain numbers.
• Maximum 100,000 rows and 2 million cells. UTF-8 output. No database connection or SQL execution against an external server.

SQLITE, SQLITE3, DB → JSON · sqlite

Export one explicitly selected SQLite table to JSON in your browser.

• SQLite 3 files only, up to 64 MiB; select a table, not a view. BLOB values use an explicit hexadecimal object in JSON and JSON object text in CSV. Integers outside the JavaScript safe range use decimal strings; smaller integers remain numbers.
• Maximum 100,000 rows and 2 million cells. UTF-8 output. No database connection or SQL execution against an external server.

CSV → SQLITE · sqlite

Create a portable, single-table SQLite database from CSV records.

• Creates a table named records with TEXT columns. CSV values stay strings; nested JSON values are JSON text; null remains SQL NULL. This does not reproduce keys, indexes, relationships or an existing schema.
• Maximum 100,000 rows and 2 million cells. UTF-8 output. No database connection or SQL execution against an external server.

JSON → SQLITE · sqlite

Create a portable, single-table SQLite database from JSON records.

• Creates a table named records with TEXT columns. CSV values stay strings; nested JSON values are JSON text; null remains SQL NULL. This does not reproduce keys, indexes, relationships or an existing schema.
• Maximum 100,000 rows and 2 million cells. UTF-8 output. No database connection or SQL execution against an external server.

PARQUET → JSON · Parquet reader

Read Apache Parquet rows and export JSON locally.

• Maximum 64 MiB and 100,000 rows / 2 million values. Uncompressed and Snappy files are supported; other codecs are rejected. Integers outside the JavaScript safe range use decimal strings; smaller integers remain numbers. Timestamps use ISO text and binary values use hexadecimal objects. Nested values become JSON text in CSV.

PARQUET → CSV · Parquet reader

Read Apache Parquet rows and export CSV locally.

• Maximum 64 MiB and 100,000 rows / 2 million values. Uncompressed and Snappy files are supported; other codecs are rejected. Integers outside the JavaScript safe range use decimal strings; smaller integers remain numbers. Timestamps use ISO text and binary values use hexadecimal objects. Nested values become JSON text in CSV.

PARQUET → JSON · Parquet reader

Read Parquet column schema, row groups and metadata without exporting row data.

• Maximum 64 MiB. Reports the stored file metadata; it does not modify or repair the file.

JSONL, NDJSON → JSON · Structured text

Convert newline-delimited JSON records to JSON.

• UTF-8, one JSON object per non-empty line; maximum 64 MiB / 100,000 records. Invalid lines are reported by line number. Nested values become JSON text in CSV.

JSONL, NDJSON → CSV · Structured text

Convert newline-delimited JSON records to CSV.

• UTF-8, one JSON object per non-empty line; maximum 64 MiB / 100,000 records. Invalid lines are reported by line number. Nested values become JSON text in CSV.

JSON → JSONL · Structured text

Write a JSON array of records as one compact object per line for data pipelines.

• UTF-8 input must be an array of objects. Maximum 64 MiB / 100,000 records.

STL, PLY, GLTF, GLB, FBX, DAE, STEP, STP, IGES, IGS, BREP → OBJ · Three.js + Assimp + Open CASCADE

Convert static meshes and FBX/DAE scenes; tessellate STEP, IGES or BREP solids locally.

• Maximum input 80 MB. STEP / IGES tessellation uses millimeters. STL / OBJ / PLY use source units.
• Self-contained uncompressed glTF/GLB only; external buffers/images, skins and morph targets are rejected. Animation is discarded.
• FBX/DAE base geometry must be contained in one file. FBX blend shapes and animation are discarded; skins, COLLADA controllers, points and lines are rejected. Assimp applies scene transforms and importer unit/axis conventions; check dimensions before manufacturing. Maximum 2 million expanded vertices. External textures are ignored and never fetched.
• Triangle geometry export only. CAD solids become triangles; hierarchy, materials, textures, normals, UVs, vertex/face colors, animation and rigging are discarded. CAD export is unavailable.

OBJ, PLY, GLTF, GLB, FBX, DAE, STEP, STP, IGES, IGS, BREP → STL · Three.js + Assimp + Open CASCADE

Convert static meshes and FBX/DAE scenes; tessellate STEP, IGES or BREP solids locally.

• Maximum input 80 MB. STEP / IGES tessellation uses millimeters. STL / OBJ / PLY use source units.
• Self-contained uncompressed glTF/GLB only; external buffers/images, skins and morph targets are rejected. Animation is discarded.
• FBX/DAE base geometry must be contained in one file. FBX blend shapes and animation are discarded; skins, COLLADA controllers, points and lines are rejected. Assimp applies scene transforms and importer unit/axis conventions; check dimensions before manufacturing. Maximum 2 million expanded vertices. External textures are ignored and never fetched.
• Triangle geometry export only. CAD solids become triangles; hierarchy, materials, textures, normals, UVs, vertex/face colors, animation and rigging are discarded. CAD export is unavailable.

OBJ, STL, GLTF, GLB, FBX, DAE, STEP, STP, IGES, IGS, BREP → PLY · Three.js + Assimp + Open CASCADE

Convert static meshes and FBX/DAE scenes; tessellate STEP, IGES or BREP solids locally.

• Maximum input 80 MB. STEP / IGES tessellation uses millimeters. STL / OBJ / PLY use source units.
• Self-contained uncompressed glTF/GLB only; external buffers/images, skins and morph targets are rejected. Animation is discarded.
• FBX/DAE base geometry must be contained in one file. FBX blend shapes and animation are discarded; skins, COLLADA controllers, points and lines are rejected. Assimp applies scene transforms and importer unit/axis conventions; check dimensions before manufacturing. Maximum 2 million expanded vertices. External textures are ignored and never fetched.
• Triangle geometry export only. CAD solids become triangles; hierarchy, materials, textures, normals, UVs, vertex/face colors, animation and rigging are discarded. CAD export is unavailable.

OBJ, STL, PLY, GLTF, FBX, DAE, STEP, STP, IGES, IGS, BREP → GLB · Three.js + Assimp + Open CASCADE

Convert static meshes and FBX/DAE scenes; tessellate STEP, IGES or BREP solids locally.

• Maximum input 80 MB. STEP / IGES tessellation uses millimeters. STL / OBJ / PLY use source units.
• Self-contained uncompressed glTF/GLB only; external buffers/images, skins and morph targets are rejected. Animation is discarded.
• FBX/DAE base geometry must be contained in one file. FBX blend shapes and animation are discarded; skins, COLLADA controllers, points and lines are rejected. Assimp applies scene transforms and importer unit/axis conventions; check dimensions before manufacturing. Maximum 2 million expanded vertices. External textures are ignored and never fetched.
• Triangle geometry export only. CAD solids become triangles; hierarchy, materials, textures, normals, UVs, vertex/face colors, animation and rigging are discarded. CAD export is unavailable.

OBJ, STL, PLY, GLB, FBX, DAE, STEP, STP, IGES, IGS, BREP → GLTF · Three.js + Assimp + Open CASCADE

Convert static meshes and FBX/DAE scenes; tessellate STEP, IGES or BREP solids locally.

• Maximum input 80 MB. STEP / IGES tessellation uses millimeters. STL / OBJ / PLY use source units.
• Self-contained uncompressed glTF/GLB only; external buffers/images, skins and morph targets are rejected. Animation is discarded.
• FBX/DAE base geometry must be contained in one file. FBX blend shapes and animation are discarded; skins, COLLADA controllers, points and lines are rejected. Assimp applies scene transforms and importer unit/axis conventions; check dimensions before manufacturing. Maximum 2 million expanded vertices. External textures are ignored and never fetched.
• Triangle geometry export only. CAD solids become triangles; hierarchy, materials, textures, normals, UVs, vertex/face colors, animation and rigging are discarded. CAD export is unavailable.

GEOJSON, KML, GPKG, ZIP → GEOJSON · GDAL WASM

Convert one selected vector layer and reproject it entirely on your device.

• 64 MB input / ZIP expansion limit. ZIP must contain one complete Shapefile; loose .shp sidecars are not accepted. Explicit layer, source CRS and target CRS choices required.
• Only GeoJSON, KML, GeoPackage and Shapefile drivers are registered. External resources and network access are blocked. Output layer is named converted.
• GeoJSON / KML output is WGS84. Shapefile field names and types are limited; KML renames standard name/description fields and may discard unsupported attributes. Styles and non-selected layers are omitted.

GEOJSON, KML, GPKG, ZIP → GPKG · GDAL WASM

Convert one selected vector layer and reproject it entirely on your device.

• 64 MB input / ZIP expansion limit. ZIP must contain one complete Shapefile; loose .shp sidecars are not accepted. Explicit layer, source CRS and target CRS choices required.
• Only GeoJSON, KML, GeoPackage and Shapefile drivers are registered. External resources and network access are blocked. Output layer is named converted.
• GeoJSON / KML output is WGS84. Shapefile field names and types are limited; KML renames standard name/description fields and may discard unsupported attributes. Styles and non-selected layers are omitted.

GEOJSON, KML, GPKG, ZIP → KML · GDAL WASM

Convert one selected vector layer and reproject it entirely on your device.

• 64 MB input / ZIP expansion limit. ZIP must contain one complete Shapefile; loose .shp sidecars are not accepted. Explicit layer, source CRS and target CRS choices required.
• Only GeoJSON, KML, GeoPackage and Shapefile drivers are registered. External resources and network access are blocked. Output layer is named converted.
• GeoJSON / KML output is WGS84. Shapefile field names and types are limited; KML renames standard name/description fields and may discard unsupported attributes. Styles and non-selected layers are omitted.

GEOJSON, KML, GPKG, ZIP → ZIP · GDAL WASM

Convert one selected vector layer and reproject it entirely on your device.

• 64 MB input / ZIP expansion limit. ZIP must contain one complete Shapefile; loose .shp sidecars are not accepted. Explicit layer, source CRS and target CRS choices required.
• Only GeoJSON, KML, GeoPackage and Shapefile drivers are registered. External resources and network access are blocked. Output layer is named converted.
• GeoJSON / KML output is WGS84. Shapefile field names and types are limited; KML renames standard name/description fields and may discard unsupported attributes. Styles and non-selected layers are omitted.

TIF, TIFF → TIF · GDAL WASM

Select one raster band and export a reprojected, losslessly compressed GeoTIFF.

• 64 MB input and 16 million pixels. Select one band and explicit coordinate systems.
• Nearest-neighbor resampling; output keeps source pixel dimensions. Other bands, pyramids, styling and auxiliary metadata are omitted. Remote resources are blocked.

TIF, TIFF → PNG · GDAL WASM

Preview one raster band as a contrast-scaled grayscale PNG.

• 64 MB input and 16 million pixels. Select one band and explicitly confirm source coordinates and source pixel grid.
• Pixel values are scaled to 8-bit for display. PNG discards georeferencing, original numeric precision, other bands and dataset metadata. No reprojection.

GEOJSON → KML · Native WGS84 vector mapping

Map self-contained WGS84 point, line and polygon features into KML.

• 20 MB / 100,000 features. Explicit WGS84 and layer choices required. No reprojection, Shapefile, GeoPackage or raster drivers.
• Properties become KML ExtendedData strings. Nested properties become JSON text; feature IDs, styles and bbox are omitted.

KML → GEOJSON · Native WGS84 vector mapping

Extract vector placemarks from a self-contained KML document.

• 20 MB / 100,000 features. All selected placemarks form one collection. MultiGeometry becomes GeometryCollection.
• Styles, altitude modes, timestamps, folder hierarchy and feature IDs are omitted. Links, overlays, tracks, 3D models and typed SchemaData are rejected. No KMZ.

H5, HDF5, HDF, NXS → JSON · h5wasm

Select one numeric HDF5 dataset and export its values locally.

• 100 MB input; 2 million selected elements. Explicit dataset path required. Compound, reference, string and variable-length data unsupported.
• JSON preserves dimensions and dtype in an envelope. CSV contains values only, without headers; rank must be 1 or 2. Attributes are omitted; unsafe integers and non-finite values are rejected.

H5, HDF5, HDF, NXS → CSV · h5wasm

Select one numeric HDF5 dataset and export its values locally.

• 100 MB input; 2 million selected elements. Explicit dataset path required. Compound, reference, string and variable-length data unsupported.
• JSON preserves dimensions and dtype in an envelope. CSV contains values only, without headers; rank must be 1 or 2. Attributes are omitted; unsafe integers and non-finite values are rejected.

JSON, CSV → H5 · h5wasm

Create a single selected HDF5 dataset from a numeric matrix or an exported JSON envelope.

• 2 million finite numeric elements. CSV requires a rectangular matrix without a header and writes float64.
• JSON must contain flat data, shape and dtype. Creates a new file with one dataset; does not preserve source attributes.

FASTA, FA, FNA, FFN, FAA, FRN → JSON · Local sequence parser

Summarize FASTA record lengths and sequence symbol counts locally.

• UTF-8, uncompressed files only, maximum 64 MiB, 100,000 records and 20 million sequence symbols. No alignment or biological interpretation.
• Accepts IUPAC nucleotide and protein letters, gaps and stop symbols. Sequence case is preserved. Symbol counts are literal; alphabet and DNA/protein identity are not inferred. Semicolon comments are omitted.

FASTA, FA, FNA, FFN, FAA, FRN → CSV · Local sequence parser

Export FASTA identifiers, descriptions and sequence strings to CSV.

• UTF-8, uncompressed files only, maximum 64 MiB, 100,000 records and 20 million sequence symbols. No alignment or biological interpretation.
• Accepts IUPAC nucleotide and protein letters, gaps and stop symbols. Sequence case is preserved. Symbol counts are literal; alphabet and DNA/protein identity are not inferred. Semicolon comments are omitted.

FASTA, FA, FNA, FFN, FAA, FRN → JSON · Local sequence parser

Export FASTA identifiers, descriptions and sequence strings to JSON.

• UTF-8, uncompressed files only, maximum 64 MiB, 100,000 records and 20 million sequence symbols. No alignment or biological interpretation.
• Accepts IUPAC nucleotide and protein letters, gaps and stop symbols. Sequence case is preserved. Symbol counts are literal; alphabet and DNA/protein identity are not inferred. Semicolon comments are omitted.

FASTA, FA, FNA, FFN, FAA, FRN → FASTA · Local sequence parser

Reverse complement IUPAC DNA sequences in a FASTA file while retaining their headers.

• UTF-8, uncompressed files only, maximum 64 MiB, 100,000 records and 20 million sequence symbols. No alignment or biological interpretation.
• DNA IUPAC symbols only; RNA U, gaps, stop symbols and protein-only symbols are rejected. Case is preserved.

FASTQ, FQ → JSON · Local sequence parser

Inspect FASTQ read lengths and explicit Phred+33 quality statistics in your browser.

• UTF-8, uncompressed files only, maximum 64 MiB, 100,000 records and 20 million sequence symbols. No alignment or biological interpretation.
• Four-line FASTQ records only, no wrapped sequences or quality. IUPAC nucleotide symbols and printable ASCII quality. Sequence/quality lengths and repeated plus-line headers must match. Inspection assumes Phred+33 explicitly; historical Phred+64 and Solexa scores are not detected.

FASTQ, FQ → FASTA · Local sequence parser

Extract FASTQ read headers and sequences as FASTA, discarding quality scores.

• UTF-8, uncompressed files only, maximum 64 MiB, 100,000 records and 20 million sequence symbols. No alignment or biological interpretation.
• Four-line FASTQ records only, no wrapped sequences or quality. IUPAC nucleotide symbols and printable ASCII quality. Sequence/quality lengths and repeated plus-line headers must match. Inspection assumes Phred+33 explicitly; historical Phred+64 and Solexa scores are not detected.

FASTQ, FQ → CSV · Local sequence parser

Export FASTQ headers, sequences and original quality strings to CSV.

• UTF-8, uncompressed files only, maximum 64 MiB, 100,000 records and 20 million sequence symbols. No alignment or biological interpretation.
• Four-line FASTQ records only, no wrapped sequences or quality. IUPAC nucleotide symbols and printable ASCII quality. Sequence/quality lengths and repeated plus-line headers must match. Inspection assumes Phred+33 explicitly; historical Phred+64 and Solexa scores are not detected.

FASTQ, FQ → JSON · Local sequence parser

Export FASTQ headers, sequences and original quality strings to JSON.

• UTF-8, uncompressed files only, maximum 64 MiB, 100,000 records and 20 million sequence symbols. No alignment or biological interpretation.
• Four-line FASTQ records only, no wrapped sequences or quality. IUPAC nucleotide symbols and printable ASCII quality. Sequence/quality lengths and repeated plus-line headers must match. Inspection assumes Phred+33 explicitly; historical Phred+64 and Solexa scores are not detected.

BED → CSV · Local genomic table parser

Export BED intervals as CSV while retaining the source coordinate convention and raw annotation fields.

• UTF-8, uncompressed input only, maximum 64 MiB, 100,000 records and 2 million cells. No assembly or coordinate-system conversion. Missing values remain literal dots; CSV cells are not sanitized for spreadsheet formulas.
• BED3 through BED12 with a consistent column count. Coordinates remain 0-based, half-open. Browser/track/comment lines are omitted from CSV and retained in JSON metadata. BED block fields stay literal comma-separated text.

BED → JSON · Local genomic table parser

Export BED intervals as JSON while retaining the source coordinate convention and raw annotation fields.

• UTF-8, uncompressed input only, maximum 64 MiB, 100,000 records and 2 million cells. No assembly or coordinate-system conversion. Missing values remain literal dots; CSV cells are not sanitized for spreadsheet formulas.
• BED3 through BED12 with a consistent column count. Coordinates remain 0-based, half-open. Browser/track/comment lines are omitted from CSV and retained in JSON metadata. BED block fields stay literal comma-separated text.

GFF, GFF3 → CSV · Local genomic table parser

Export GFF3 features as CSV while retaining the source coordinate convention and raw annotation fields.

• UTF-8, uncompressed input only, maximum 64 MiB, 100,000 records and 2 million cells. No assembly or coordinate-system conversion. Missing values remain literal dots; CSV cells are not sanitized for spreadsheet formulas.
• Requires ##gff-version 3 and nine tab-separated columns. Coordinates remain 1-based, inclusive. Attributes stay literal encoded text. Directives/comments are retained in JSON metadata; embedded ##FASTA sequence is omitted.

GFF, GFF3 → JSON · Local genomic table parser

Export GFF3 features as JSON while retaining the source coordinate convention and raw annotation fields.

• UTF-8, uncompressed input only, maximum 64 MiB, 100,000 records and 2 million cells. No assembly or coordinate-system conversion. Missing values remain literal dots; CSV cells are not sanitized for spreadsheet formulas.
• Requires ##gff-version 3 and nine tab-separated columns. Coordinates remain 1-based, inclusive. Attributes stay literal encoded text. Directives/comments are retained in JSON metadata; embedded ##FASTA sequence is omitted.

GTF → CSV · Local genomic table parser

Export GTF gene annotations as CSV while retaining the source coordinate convention and raw annotation fields.

• UTF-8, uncompressed input only, maximum 64 MiB, 100,000 records and 2 million cells. No assembly or coordinate-system conversion. Missing values remain literal dots; CSV cells are not sanitized for spreadsheet formulas.
• Nine tab-separated columns with quoted GTF attributes. Coordinates remain 1-based, inclusive. Attributes stay literal text; comments are retained only in JSON metadata.

GTF → JSON · Local genomic table parser

Export GTF gene annotations as JSON while retaining the source coordinate convention and raw annotation fields.

• UTF-8, uncompressed input only, maximum 64 MiB, 100,000 records and 2 million cells. No assembly or coordinate-system conversion. Missing values remain literal dots; CSV cells are not sanitized for spreadsheet formulas.
• Nine tab-separated columns with quoted GTF attributes. Coordinates remain 1-based, inclusive. Attributes stay literal text; comments are retained only in JSON metadata.

VCF → CSV · Local genomic table parser

Export Genomic VCF variants as CSV while retaining the source coordinate convention and raw annotation fields.

• UTF-8, uncompressed input only, maximum 64 MiB, 100,000 records and 2 million cells. No assembly or coordinate-system conversion. Missing values remain literal dots; CSV cells are not sanitized for spreadsheet formulas.
• Plain VCF 4.1–4.5 with ##fileformat and #CHROM headers. INFO, FORMAT, ALT and sample genotypes stay literal strings; no variant normalization, allele splitting, reference lookup or genotype interpretation. This is genomic VCF, not a vCard contact file.

VCF → JSON · Local genomic table parser

Export Genomic VCF variants as JSON while retaining the source coordinate convention and raw annotation fields.

• UTF-8, uncompressed input only, maximum 64 MiB, 100,000 records and 2 million cells. No assembly or coordinate-system conversion. Missing values remain literal dots; CSV cells are not sanitized for spreadsheet formulas.
• Plain VCF 4.1–4.5 with ##fileformat and #CHROM headers. INFO, FORMAT, ALT and sample genotypes stay literal strings; no variant normalization, allele splitting, reference lookup or genotype interpretation. This is genomic VCF, not a vCard contact file.

VCF → JSON · Local genomic table parser

List a genomic VCF file’s version, sample names, variant counts and chromosome counts locally.

• Plain UTF-8 VCF 4.1–4.5 only, maximum 64 MiB, 100,000 variants and 2 million cells. Not vCard contacts, binary BCF or compressed VCF. Counts describe records, not normalized alleles; genotypes are not interpreted.

MTX → JSON · Local scientific matrix parser

Inspect Matrix Market sparse coordinates and dimensions locally.

• Maximum 64 MiB and 2 million stored values/coordinates; each dimension is at most 2 million. Only finite values are supported. Dense CSV is capped at 2 million cells.
• Matrix Market coordinate real, integer or pattern; general or symmetric only. JSON retains 1-based coordinates and symmetric storage. Duplicate coordinates are summed, with unsafe integer sums rejected. CSV expands symmetric entries. Complex, array, skew-symmetric and Hermitian formats are unsupported.

MTX → CSV · Local scientific matrix parser

Export Matrix Market sparse data as CSV with explicit shape and precision limits.

• Maximum 64 MiB and 2 million stored values/coordinates; each dimension is at most 2 million. Only finite values are supported. Dense CSV is capped at 2 million cells.
• Matrix Market coordinate real, integer or pattern; general or symmetric only. JSON retains 1-based coordinates and symmetric storage. Duplicate coordinates are summed, with unsafe integer sums rejected. CSV expands symmetric entries. Complex, array, skew-symmetric and Hermitian formats are unsupported.

MTX → JSON · Local scientific matrix parser

Export Matrix Market sparse data as JSON with explicit shape and precision limits.

• Maximum 64 MiB and 2 million stored values/coordinates; each dimension is at most 2 million. Only finite values are supported. Dense CSV is capped at 2 million cells.
• Matrix Market coordinate real, integer or pattern; general or symmetric only. JSON retains 1-based coordinates and symmetric storage. Duplicate coordinates are summed, with unsafe integer sums rejected. CSV expands symmetric entries. Complex, array, skew-symmetric and Hermitian formats are unsupported.

NPY → JSON · Local scientific matrix parser

Inspect NumPy NPY shape, numeric dtype and storage order locally.

• Maximum 64 MiB and 2 million stored values/coordinates; each dimension is at most 2 million. Only finite values are supported. Dense CSV is capped at 2 million cells.
• NPY v1/v2/v3, plain numeric/bool dtype only: signed/unsigned 1/2/4/8-byte integers, float32/64 and bool. Endianness and C/Fortran order are handled. No object/pickle, structured, string, float16, complex, NPZ or masked arrays. JSON stores flat data in original storage order with shape/dtype; unsafe integers are decimal strings. CSV supports rank 0–2 and lays out rows/columns; JSON supports rank up to 32.

NPY → CSV · Local scientific matrix parser

Export NumPy numeric arrays as CSV with explicit shape and precision limits.

• Maximum 64 MiB and 2 million stored values/coordinates; each dimension is at most 2 million. Only finite values are supported. Dense CSV is capped at 2 million cells.
• NPY v1/v2/v3, plain numeric/bool dtype only: signed/unsigned 1/2/4/8-byte integers, float32/64 and bool. Endianness and C/Fortran order are handled. No object/pickle, structured, string, float16, complex, NPZ or masked arrays. JSON stores flat data in original storage order with shape/dtype; unsafe integers are decimal strings. CSV supports rank 0–2 and lays out rows/columns; JSON supports rank up to 32.

NPY → JSON · Local scientific matrix parser

Export NumPy numeric arrays as JSON with explicit shape and precision limits.

• Maximum 64 MiB and 2 million stored values/coordinates; each dimension is at most 2 million. Only finite values are supported. Dense CSV is capped at 2 million cells.
• NPY v1/v2/v3, plain numeric/bool dtype only: signed/unsigned 1/2/4/8-byte integers, float32/64 and bool. Endianness and C/Fortran order are handled. No object/pickle, structured, string, float16, complex, NPZ or masked arrays. JSON stores flat data in original storage order with shape/dtype; unsafe integers are decimal strings. CSV supports rank 0–2 and lays out rows/columns; JSON supports rank up to 32.

CSV → MTX · Local scientific matrix parser

Turn a headerless numeric CSV matrix into sparse Matrix Market coordinate data, omitting zero entries.

• Headerless rectangular UTF-8 CSV, maximum 64 MiB and 2 million cells. Cells must be finite decimal numbers; empty cells and unsafe integers are rejected. Writes real general coordinates with 1-based indices. Row/column labels and metadata are not included.

H5, HDF5, HDF, NXS, LOOM, H5MU, NC → JSON · h5wasm

List datasets, dimensions and datatypes in HDF5, Loom, MuData and HDF5-based NetCDF4 files.

• HDF5 containers only; classic NetCDF3 is unsupported. Maximum 64 MiB, 24 levels and 1,000 entries. External and symbolic links are rejected. No dataset values are exported.

H5AD → JSON · h5wasm

Inspect AnnData groups and datasets, including expression matrices, cell observations and gene variables.

• Modern AnnData HDF5 encodings only. Maximum 64 MiB and 2 million values per dataset. Numeric matrices must be finite and exactly representable; external links and unsupported encodings are rejected.
• Matrix JSON retains sparse indices and pointers; matrix CSV expands to dense values with a 2 million cell cap and no row/column labels. Export obs and var separately for labels. Annotation CSV turns missing values into empty cells. This does not recreate a complete H5AD file.

H5AD → JSON · h5wasm

Export an AnnData matrix as JSON, preserving dense values or sparse CSR and CSC indices and pointers.

• Modern AnnData HDF5 encodings only. Maximum 64 MiB and 2 million values per dataset. Numeric matrices must be finite and exactly representable; external links and unsupported encodings are rejected.
• Matrix JSON retains sparse indices and pointers; matrix CSV expands to dense values with a 2 million cell cap and no row/column labels. Export obs and var separately for labels. Annotation CSV turns missing values into empty cells. This does not recreate a complete H5AD file.

H5AD → CSV · h5wasm

Export a selected AnnData expression matrix as dense CSV, expanding sparse CSR or CSC storage locally.

• Modern AnnData HDF5 encodings only. Maximum 64 MiB and 2 million values per dataset. Numeric matrices must be finite and exactly representable; external links and unsupported encodings are rejected.
• Matrix JSON retains sparse indices and pointers; matrix CSV expands to dense values with a 2 million cell cap and no row/column labels. Export obs and var separately for labels. Annotation CSV turns missing values into empty cells. This does not recreate a complete H5AD file.

H5AD → CSV · h5wasm

Export AnnData cell observations to CSV, decoding categorical annotations and preserving cell index labels.

• Modern AnnData HDF5 encodings only. Maximum 64 MiB and 2 million values per dataset. Numeric matrices must be finite and exactly representable; external links and unsupported encodings are rejected.
• Matrix JSON retains sparse indices and pointers; matrix CSV expands to dense values with a 2 million cell cap and no row/column labels. Export obs and var separately for labels. Annotation CSV turns missing values into empty cells. This does not recreate a complete H5AD file.

H5AD → CSV · h5wasm

Export AnnData gene variables to CSV, decoding categorical annotations and preserving gene index labels.

• Modern AnnData HDF5 encodings only. Maximum 64 MiB and 2 million values per dataset. Numeric matrices must be finite and exactly representable; external links and unsupported encodings are rejected.
• Matrix JSON retains sparse indices and pointers; matrix CSV expands to dense values with a 2 million cell cap and no row/column labels. Export obs and var separately for labels. Annotation CSV turns missing values into empty cells. This does not recreate a complete H5AD file.

TXT, MD → HTML · Native text

Create a standalone UTF-8 HTML page with readable, escaped text.

• Markdown syntax remains literal. No scripts, external fonts or remote resources.

TXT → PDF · pdf-lib

Create a searchable, simply formatted PDF from readable text.

• Plain text reflow with Helvetica and Western European characters only. Other scripts fail explicitly. Illustrations, book layout and styling are omitted. Maximum 500 pages.

IPYNB → MD · Local text parser

Extract Jupyter notebook source cells as a readable Markdown document with fenced code.

• Notebook v4 only, up to 64 MiB / 100,000 cells. Execution outputs, embedded attachments and display widgets are omitted. Cells are never executed.

IPYNB → PY · Local text parser

Extract code cells from a Python Jupyter notebook into a script with cell separators.

• Notebook v4 with a declared Python kernel only. Markdown and outputs are omitted. IPython magics remain literal and may not run as ordinary Python. Code is never executed. Maximum 64 MiB / 100,000 cells.