Skip to content

query.md

Latest commit

 

History

History
59 lines (46 loc) · 1.81 KB

query.md

File metadata and controls

59 lines (46 loc) · 1.81 KB

NLP/SQL Query

Przegląd

Przeszukuj historię wymian z LLM za pomocą:

  1. NLP queries — pytanie naturalne → SQL (local parser + LLM fallback)
  2. SQL queries — bezpośrednie zapytania SELECT

NLP Query — przykłady

toonic> query "show all errors from the last hour"
toonic> query "which files were fixed today?"
toonic> query "camera events with high confidence"
toonic> query "total tokens used by gemini model"
toonic> query "last 5 video analyses"
toonic> query "count errors per category"

Wbudowany local parser obsługuje:

  • Filtry czasu: last hour, today, last week
  • Kategorie: video, code, logs, audio
  • Status: error, success
  • Akcje: fix, alert, report
  • Modele: gemini, claude, gpt
  • Agregacje: count, total, how many
  • Content search: about "keyword"

Jeśli local parser nie rozpozna zapytania — używa LLM do generowania SQL.

SQL Query — przykłady

toonic> sql SELECT * FROM exchanges WHERE category='video' ORDER BY timestamp DESC LIMIT 10
toonic> sql SELECT model, COUNT(*), SUM(tokens_used) FROM exchanges GROUP BY model
toonic> sql SELECT action_type, AVG(confidence) FROM exchanges GROUP BY action_type
toonic> sql SELECT * FROM exchanges WHERE content LIKE '%authentication%'
toonic> sql SELECT date(timestamp, 'unixepoch') as day, COUNT(*) FROM exchanges GROUP BY day

REST API

# NLP query
curl -X POST http://localhost:8900/api/query \
  -H "Content-Type: application/json" \
  -d '{"question": "show errors from last hour"}'

# SQL query
curl -X POST http://localhost:8900/api/sql \
  -H "Content-Type: application/json" \
  -d '{"sql": "SELECT * FROM exchanges LIMIT 10"}'

Bezpieczeństwo

  • Tylko zapytania SELECT są dozwolone
  • INSERT/UPDATE/DELETE są blokowane
  • Wyniki limitowane do 50 wierszy domyślnie