This package is aimed at taking snapshots of pages to ensure Tarsier outputs are consistent and performant over time.

We use MHTML pages found in our bananalyzer repo as site targets. For each site on bananalyzer, we store both the screenshot of the page along with the OCR text output. These sites will not change overtime therefore they serve as ideal candidates for snapshotting.

• Create .env and add your credentials
• Run poetry install
• Run poetry run python tarsier_snapshots/snapshots.py