infogdl

Download, analyze, sort, and resize infographics from LinkedIn and Twitter — or process a local directory of images.

Images are classified by orientation (vertical/horizontal), number of dominant colors, and information density (fill rate), then sorted into a structured folder tree. Each image is cropped to its content (with a 1px border preserved), scaled to fit within 1920×1080, and compressed if oversized. No rotation is ever applied.

output/
├── horizontal/
│   ├── low_colors/
│   │   ├── sparse_fill/
│   │   ├── moderate_fill/
│   │   └── dense_fill/
│   ├── medium_colors/
│   └── high_colors/
└── vertical/
    └── ... (same structure)

Windows:

setup.bat

Linux / macOS:

./setup.sh

Or manually:

pip install -r requirements.txt

Requires Python 3.10+. Scraping mode also needs Chrome + chromedriver.

python infogdl.py -i /path/to/images -o /path/to/output

Recursively finds all images (png, jpg, webp, gif, bmp, tiff) in the input directory and its subdirectories.

1. Log into LinkedIn and/or Twitter in Chrome or Firefox
2. Edit config.json with the profile URLs you want to scrape
3. Run:

python infogdl.py

The tool borrows your browser session cookies — no credentials are stored or requested.

If you already use gallery-dl, infogdl will automatically pick up your gallery-dl config (~/.config/gallery-dl/config.json on Linux, %APPDATA%\gallery-dl\config.json on Windows) and use the same cookie source (browser or cookie file). Just run infogdl from the same directory.

Each profile's download progress is stored in .infogdl.db. On subsequent runs, only new images are downloaded:

# Normal run — skips already-downloaded images
python infogdl.py

# Full rescan — ignore progress, re-download everything
python infogdl.py --full-rescan

config.json controls all parameters:

1. Authenticate — Extracts cookies directly from browser SQLite databases (Chrome, Firefox, Edge, Brave, Opera) or imports a Netscape cookie file. Sessions are cached and validated before use.
2. Scrape — Selenium opens each profile, scrolls to load content, collects images >200px (skips avatars/icons)
3. Track — Each downloaded URL is recorded in .infogdl.db per profile. Next run skips known URLs.
4. Analyze — Each image is measured for dominant color count (k-means clustering), fill rate (edge density), and orientation
5. Sort — Placed into subfolders based on the analysis
6. Crop — Content bounding box detected via background color sampling; at least 1px border always preserved
7. Resize — Scaled to fill one dimension of the target size, maintaining aspect ratio (max 2× upscale)
8. Compress — PNG if small enough, otherwise JPEG at decreasing quality until under the size limit

MIT