Summary
Page.printToPDF produces a raster-only PDF: zero embedded fonts and an empty text layer. The PDF cannot be searched, copied from, or read by assistive technology.
Environment
- moli 1.0.6, also verified on local build at commit a1385e0 (main, 2026-08-30)
- Chromium baseline: chrome-headless-shell 131.0.6778.204, same probe
Reproduction
Serve this page:
<!DOCTYPE html><html><head><title>v6</title></head><body><h1>Font Text 123</h1></body></html>
Then:
moli fetch http://127.0.0.1:8765/v6.html --dump=pdf --layout > out.pdf
pdffonts out.pdf # -> 0 font rows
pdftotext out.pdf - | tr -d '[:space:]' | wc -c # -> 0
Observed (moli)
pdffonts: 0 fonts embeddedpdftotext: 0 characters in the text layer- Visual rendering is correct: non-white pixel coverage matches Chrome, and the PDF opens as an image-like page
Expected (Chromium 131)
- Same page →
pdftotext recovers 152 characters ("Font Text 123" etc.)
- Fonts are embedded as subsetted font programs
Impact
PDF output is unusable for search, copy-paste, indexing, or accessibility. Anything consuming moli's PDF export (archival, RAG ingestion, agent tooling) gets a picture, not a document.
Verified on a1385e0: still raster-only (0 fonts / 0 text chars).
Summary
Page.printToPDFproduces a raster-only PDF: zero embedded fonts and an empty text layer. The PDF cannot be searched, copied from, or read by assistive technology.Environment
Reproduction
Serve this page:
Then:
Observed (moli)
pdffonts: 0 fonts embeddedpdftotext: 0 characters in the text layerExpected (Chromium 131)
pdftotextrecovers 152 characters ("Font Text 123" etc.)Impact
PDF output is unusable for search, copy-paste, indexing, or accessibility. Anything consuming moli's PDF export (archival, RAG ingestion, agent tooling) gets a picture, not a document.
Verified on a1385e0: still raster-only (0 fonts / 0 text chars).