Importing Documents

Most writing lives in Word documents, spreadsheets, or slide decks — formats that are hard to version-control, collaborate on, or publish as-is. Pourdown lets you import any of them directly into an editable Markdown document, so you can clean up, restructure, and export without manual copy-pasting.

How to import: File → Import → choose your format. The file opens immediately as a new Markdown document.

Import converts content to Markdown — it's one-way. Exporting back to the original format will not restore the original layout exactly.

What gets converted

Word (.docx)

• Headings (styles + outline level), bold / italic / strikethrough
• Nested bullet and numbered lists, tables, hyperlinks
• Embedded images, extracted as sidecar files and shown inline

Limitations: vector images (EMF/WMF) can't be displayed in the editor; tracked changes and comments are dropped; a table of contents becomes a placeholder rather than being reconstructed.

Spreadsheet (.xlsx / .xls / .ods)

• Each sheet becomes a section with a full GFM table
• Date columns are auto-detected (header contains "date" / "日期") and formatted as ISO dates
• Embedded images are extracted

Limitations: capped at 500 rows per sheet (with a note when rows are omitted); images can't be mapped to a specific sheet/cell, so they're listed in a best-effort "Embedded Images" section instead of placed inline.

PDF

• Headings inferred from font-size ratios relative to the page's body text
• Paragraph flow sorted top-to-bottom, left-to-right
• Tables detected via geometry (aligned columns across several consecutive rows) and rendered as GFM tables
• Table of Contents entries (dot-leader lines) render as a bulleted list
• Embedded images extracted in reading order

Limitations: text-based PDFs only — scanned/image PDFs aren't supported; complex multi-column layouts may reorder; a table whose wrapped cell content is itself an indented/bulleted list falls back to plain paragraphs rather than corrupting the table (a deliberate, conservative trade-off).

PowerPoint (.pptx)

• Slide titles become # headings; body text becomes paragraphs, one slide per section (separated by ---)
• Bullet/indent level and basic bold/italic formatting are preserved
• Embedded images are extracted and shown inline

Limitations: animations aren't captured (not representable in Markdown); vector images (EMF/WMF) can't be displayed.

Image handling

Extracted images across all four formats are saved as sidecar files next to the imported document — an assets/ folder while the document is unsaved, relocated to <name>.assets/ alongside the .md file on first save — and render live in the editor. The .md file only ever stores the relative path, so the document and its images stay portable together as one unit you can move, zip, or commit to git.

Vector formats the webview can't render (EMF/WMF, common in Office exports) are replaced with an *(unsupported image)* note instead of a broken image link.

Why Markdown Import?

Files are converted to Markdown before processing to minimize token usage. Community benchmarks show Markdown is roughly 15% more token-efficient than JSON, and up to 96% more efficient than raw PDF, when fed to LLMs.

This approach was inspired by Microsoft's MarkItDown; Pourdown is an independent reimplementation in Rust, not a fork or a port — see markdown-import.md in the repository for the full technical breakdown of how each converter works.