screencastgen

Reader Assets

Builds the browser-reader bundle consumed by the React reader.

Source: screencastgen/reader_assets.py

Overview

build_reader_assets() converts aligned chunks into the hosted assets consumed by the React reader. The Offline Reader packages these same assets into the self-contained downloadable ZIP.

File	Purpose
`reader_manifest.json`	Metadata, global word timings, chunk text, page mapping, and optional presenter filename
`reader_audio.mp3`	Concatenated chunk audio used by highlight-only reader output
`presenter.mp4`	Optional lip-sync presenter video built by Lipsync Pipeline
`pages/page-0001.jpg`	Optional rendered PDF page images for the active-page side panel

All word timings in the manifest are global timestamps relative to the concatenated audio/presenter timeline.

Functions

`build_reader_assets(aligned_chunks, output_dir, pdf_path, title, language, presenter) -> str | None`

Concatenates chunk audio into reader_audio.mp3.
Computes per-chunk offsets.
Renders PDF page images when page numbers are available and the source is a PDF.
Writes reader_manifest.json.
Returns the manifest path, or None if no chunks were available.

`reader_asset_names(page_files=None, presenter=False) -> list[str]`

Returns relative paths for manifest/audio/presenter/page files so storage upload helpers can mirror the full bundle.

Manifest Shape

{
  "version": 1,
  "title": "Book title",
  "language": "en-US",
  "source_type": "pdf",
  "duration": 123.456,
  "audio": "reader_audio.mp3",
  "presenter": "presenter.mp4",
  "pages": {
    "dir": "pages",
    "image_width": 1400,
    "files": {"1": "page-0001.jpg"}
  },
  "chunks": [
    {
      "chunk_num": 1,
      "text": "...",
      "offset": 0.0,
      "pages": [1],
      "words": [{"word": "Hello", "start": 0.12, "end": 0.4}]
    }
  ]
}

Dependencies

Reader Assets
├── Types             (AlignedChunk)
├── Extractor         (render_page_image_with_zoom for PDFs)
└── pydub / ffprobe      (duration and MP3 concatenation)