View

View

Vue 3 + Vite + D3, served from FastAPI at http://127.0.0.1:8180/

The header shows urls, fetched, frontier, filtered, errors, seeds, hosts, and top MIME. Those numbers poll about every 0.8s while a crawl is running — you do not need to switch tabs.

Tab	What it is
Jobs	Crawl jobs and counts. Clear on a card drops that job.
Seeds	Seed URLs. Type to search/highlight. Click a seed to open Documents with that URL as the query.
Documents	Catalog. Click a row for Tika metadata and extracted text. Label not / relevant / highly. Load more 25.
Frontier	UNFETCHED queue, score then depth.
Stats	D3 board (hover everything for a tooltip).

Stats board

• KPIs: crawl span, pages with Tika metadata, avg fetch ms, avg Tika parse ms, distinct Tika fields
• Crawl time × depth — heatmap. Columns are time buckets; rows are depth bins (0…3, 4–7, 8–15, 16+) so blog pagination does not make 100 rows
• Fetch rate — area + line
• Status, MIME, language, encoding — donuts
• Hosts, Tika parsers — packed bubbles (area ∝ pages)
• Tika metadata fields — treemap (tile size = how many pages carry that key)
• Fetch latency — histogram of response_time

There is no Banana dashboard and no Solr Admin iframe. Raw Solr is still at http://127.0.0.1:8983/solr/#/crawldb/query if you want it.

Clear

Clear job / Clear all in the bar. Confirm dialogs. Use the job id, not a click event.