pptx

Run python -m markitdown presentation.pptx to extract all text content from a .pptx file. For a visual overview, use the thumbnail.py script to generate a labeled grid of slide images, or unpack the file to inspect raw slide XML.

Unpack the .pptx with scripts/office/unpack.py, then duplicate, reorder, or delete slides by editing presentation.xml and using add_slide.py. Edit each slide's XML content, run clean.py to remove orphaned files, and repack with pack.py for validation.

Use the pptxgenjs library: create a new presentation, set a layout like LAYOUT_16x9, and add slides with text, shapes, images, tables, and charts. Never use # prefixes in hex colors or unicode bullet characters, as these corrupt the output file.

Common causes include hex colors with # prefixes, opacity encoded in 8-character color strings, negative shadow offsets, and broken relationship references after manual slide copying. Use pack.py with validation to auto-repair issues and catch broken references.

Convert the .pptx to PDF with soffice, then to images with pdftoppm at 150 DPI. Inspect the images for overlapping elements, text overflow, low contrast, uneven spacing, and leftover placeholder text, then fix and re-render affected slides.

You need markitdown[pptx] for text extraction, Pillow for thumbnail grids, pptxgenjs via npm for creating decks, LibreOffice (soffice) for PDF conversion, and Poppler's pdftoppm for rendering slide images.