glmv-pdf-to-ppt

PDF → HTML PPT Skill

Convert any PDF into a multi-slide HTML presentation. Pages are converted to images at DPI 120, read sequentially to understand the content, then a structured outline.json is saved, images are cropped locally (no cloud upload), slides are rendered one by one, and finally a summary.md is generated.

Scripts are in: {SKILL_DIR}/scripts/

Dependencies

Python packages (install once):

pip install pymupdf pillow

System tools: curl (pre-installed on macOS/Linux).

When to Use

Trigger when the user asks to make slides or a presentation from a PDF — phrases like: "make a PPT from a PDF", "convert PDF to slides", "create a presentation from this paper", "根据pdf做ppt", "根据论文做幻灯片", "做PPT", "做幻灯片", "生成演示文稿", "把这个pdf转成ppt", or any similar intent in Chinese or English.

Output Directory Convention

All output goes under {WORKSPACE}/ppt/<pdf_stem>_<timestamp>/:

ppt/
└── <pdf_stem>_<timestamp>/
    ├── outline.json        ← structured slide plan (SlidesPlan schema)
    ├── crops/              ← locally-saved cropped images
    │   ├── slide3_method_crop.png
    │   └── slide5_results_crop.png
    ├── slide_01.html
    ├── slide_02.html
    ├── ...
    └── summary.md          ← final summary document

• <pdf_stem> = PDF filename without extension
• <timestamp> = format YYYYMMDD_HHMMSS (e.g. 20240119_143022)
• Cropped images go in crops/ subfolder
• Each slide HTML references images via relative path crops/<name>.png

Input

$ARGUMENTS is the path to the PDF file (local) or an HTTP/HTTPS URL.

• If user provides a URL: download with curl first, then convert
• If user provides a local PDF path: convert directly

Workflow

Phase 0 — Create Output Directory

Compute the output path:

import os, datetime

Create it immediately:

Record out_dir — use it for all subsequent phases.

Phase 1 — Convert PDF Pages to Images (DPI 120)

If the input is a URL, download it first:

Then convert (pass either the downloaded path or the original local path):

Outputs JSON to stdout:

Parse and store the full page → path map. These local paths are used for viewing pages and as --path input to crop.py.

Phase 2 — Read All Pages in Order

View all page images sequentially before planning anything. Your goal here is pure understanding — absorb the full structure, content, figures, and arguments of the document.

While reading, note:

• What figures, charts, or tables appear on which pages
• The overall arc (intro → method → results → conclusion for papers; or logical structure for other doc types)
• Candidate visuals worth cropping for slides (page number + rough region)

Do NOT plan or write slides yet — just read and understand all pages first.

Phase 3 — Plan Outline & Save outline.json

After reading all pages, plan 8–15 slides (adapt freely for non-academic documents).

For each slide that needs a visual, identify:

• Which page it comes from (the local page path from Phase 1)
• A description of what the visual shows and why it belongs on this slide

Save the outline as <out_dir>/outline.json using exactly this schema:

Field notes:

• lang: "Chinese" or "English" — match the PDF language

• template_id: always null

• required_crops: empty array [] if this slide needs no images

• url in each crop: the local file path of the source page image (from Phase 1 path field) — this is what crop.py will open and crop from

• visual_description: what the visual shows, including figure/table number if available

• usage_reason: why this visual belongs on this particular slide

• For images that need cropping, note the approximate region — exact crop boxes are determined in Phase 4

Write outline.json using the Write tool to <out_dir>/outline.json.

Phase 4 — Crop Required Images (Grounding + Subagent)

IMPORTANT: You MUST delegate ALL cropping to a clean subagent using the Agent tool. By this phase your context is very long (all page images + outline), which degrades visual coordinate accuracy. A fresh subagent with only the target image produces much more precise coordinates.

IMPORTANT: You MUST use the provided {SKILL_DIR}/scripts/crop.py script for ALL image cropping. Do NOT write your own cropping code, do NOT use PIL/Pillow directly, do NOT use any other method.

Read outline.json. Collect all crops needed, then launch one subagent per source page (or one per crop if pages differ). The subagent uses grounding-style localization — it views the image, locates the target element, and outputs a precise bounding box in normalized 0–999 coordinates.

Use the Agent tool like this:

Replace <page_image_path>, <SKILL_DIR>, <out_dir>, and crop details with actual values from your context.

The crop.py script outputs JSON: {"path": "/abs/path/slide3_method_crop.png"}

Collect results from all subagents and build the mapping: slide_index → [crop filename, ...] to reference in HTML. The filename will be <name>_crop.png.

Launch subagents for independent pages in parallel when possible. Wait for all to complete before proceeding.

Phase 5 — Measure Cropped Image Dimensions

After cropping, get pixel dimensions:

Use aspect ratios to pick each slide's layout:

Phase 6 — Generate Slides One by One

For each slide, write the HTML, save it to a temp file, then call generate_slide.py.

Step A — Write HTML to /tmp/slide_N.html

• All <img src="..."> must use relative paths: crops/<name>_crop.png
• Do NOT use absolute paths or URLs for cropped images
• Navigation is click-area based — no buttons needed:
  • Clicking the left half of the slide navigates to the previous slide
  • Clicking the right half of the slide navigates to the next slide
  • On slide 1, left click does nothing; on the last slide, right click does nothing
  • Keyboard ← / → arrows also navigate
  • Implement with two transparent <div> overlays covering each half, positioned absolute over the slide canvas

Step B — Save slide:

Repeat until all slides are saved.

Phase 7 — Generate summary.md

Write <out_dir>/summary.md in the same language as the slides (lang from outline.json).

Include:

• Document title and basic info (authors, venue, year if applicable)
• Brief abstract/overview (2–3 sentences)
• Per-slide breakdown table: slide number, title, 1–2 sentence summary
• Main contributions or takeaways (bullet list)
• Link to slide_01.html to open the first slide

Example structure:

HTML Slide Spec

Each slide is a standalone HTML file — full <html>…</html> with embedded CSS only.

Canvas: fixed 1280 × 720 px, overflow: hidden — nothing scrolls.

Consistent design across all slides:

• Choose a visual style that fits the document's domain and tone — no fixed palette or font required
• If the user specifies a style, follow it exactly; otherwise infer from the content (e.g. a ML paper → clean modern; a historical report → editorial serif; a product pitch → bold and branded)
• Same fonts, colors, and spacing system applied uniformly to every slide
• Every slide shows: slide title, page counter (bottom-right corner), presentation title (subtle footer)

Navigation on each slide:

• Two transparent click areas cover the full slide height: left 50% → previous slide, right 50% → next slide
• On slide 1 the left area is inert; on the last slide the right area is inert
• Keyboard ← / → arrows also navigate
• No visible buttons needed — optionally show a subtle ‹ / › hint at the edges that fades in on hover

Layout patterns:

• title-card — centered hero, large title, authors/venue below
• text-only — structured bullet points, max 5–6 items, generous whitespace
• text + image — image right or left, text opposite
• full-image — image fills canvas, minimal text overlay
• grid — 2×2 or 3-column figures with captions

Images:

• Use relative paths: crops/<name>_crop.png
• Add style="object-fit: contain; max-width: 100%; max-height: 100%;"
• Add captions below in small italic text

Do NOT:

• Use external JS frameworks or icon CDNs
• Use placeholder/stock images — only the cropped PDFs
• Generate generic purple-gradient-on-white slides
• Let content overflow the 720px height

Quality Checklist

• Output directory named <pdf_stem>_<timestamp>/
• outline.json saved with valid SlidesPlan schema
• All crops saved to crops/ (local only, no cloud upload)
• Each slide fits within 1280×720, nothing overflows
• Consistent theme across all slides
• Crop images referenced via relative path crops/<name>_crop.png
• Slide number and presentation title visible on every slide
• Left/right click-area navigation works, keyboard arrows work
• summary.md written in the correct language, links to slide_01.html

Language

Match the PDF language. Chinese PDF → Chinese slides and summary. English → English. No mixing.