# MicroQA

QA assistant for the Internet Archive's microfiche scanning team.

## Usage

Analyze page statistics for item:

```sh
echo 'micro_IA04244212_1665' | uv run main.py | jq
```

Paste item IDs from clipboard and summarize all (`tr` command collapses input to
a single line so that items are summarized in parallel):

```sh
pbpaste | tr '\n' ',' | uv run main.py --summarize -workers 4 -v | jq
```

Query a pre-populated database for suspect pages:

```sql
select   'https://archive.org/details/' || items.id,
         pages.page,
         pages.orientation_match,
         pages.sharpness,
         pages.text_margin_px
from     items
         join pages on pages.item = items.id
where    pages.orientation_match = 0
         or pages.sharpness < 0.07
         or (pages.text_margin_px > -1 and pages.text_margin_px < 50)
order by items.id;
```

## Test Cases

- Blurry pages: `micro_IA40244209_0984`
- Contrast, page orientation: `micro_IA40244211_2290`
- Crop, low quality fiche: `micro_IA40386420_0689`
- "Bite sized" SCOTUS doc with multiple viewable files and some blurry pages: `micro_IA40386007_0012`