Pptx

Write one HTML file per slide with fixed body dimensions (720pt × 405pt for 16:9), keeping all text inside p, h1-h6, ul, or ol tags. Then call the html2pptx() function with a PptxGenJS instance to convert each file and save with pptx.writeFile().

Unpack the .pptx with the unpack.py script to expose its OOXML structure, edit files like ppt/slides/slideN.xml or notesSlide XML directly, validate with validate.py against the original, then repack with pack.py.

Extract the template's text with markitdown and build a thumbnail grid, then use rearrange.py to duplicate and reorder template slides by index. Finally run inventory.py and replace.py to swap placeholder text while preserving formatting.

The most common cause is using a # prefix on hex colors in PptxGenJS, which corrupts the file. Always write colors like "FF0000" without the hash, and avoid CSS gradients in HTML since they must be rasterized to PNG first.

Yes, but you need raw XML access rather than plain text extraction. Unpack the file and read ppt/notesSlides/notesSlideN.xml for speaker notes and ppt/comments/ for comments, since markitdown only captures slide text.

CSS gradients are unsupported and must be rasterized as PNG images with Sharp first. Backgrounds, borders, and shadows only work on div elements, not text elements, and only web-safe fonts like Arial or Georgia render reliably.