Commit Graph

10 Commits

Author SHA1 Message Date
c632aa41a2 fix: Node proxy on 3000 → Python dashboard on 8000
run is Node-based, so we need a Node entrypoint. server.js:
1) pip install -r requirements.txt (idempotent, --break-system-packages)
2) spawn python3 -m src.dashboard.run on PORT 8000
3) http proxy 3000 → 8000
This makes the dashboard visible at /proxy/3000/.
2026-09-24 10:31:45 +00:00
8d527c2f1c fix: index.html redirects to Python dashboard (port 3000)
Old Node.js agent (currencies+weather) saved as index.legacy.html.
Preview now shows the new AgroMarket price dashboard with
potato/wheat/milk/apple prices, search, spreads, alerts.
2026-09-24 09:14:03 +00:00
0900a06bd0 feat: product search + potato/onion/carrot/beet in demo + demo-data banner + /api/products
- overview: search field with datalist -> /product/<id> (matches ru label or id)
- demo_data: +4 tubers (potato KZ/TJ/UZ, onion, carrot, beet)
- product page: banner 'Демо-данные' when all sources are demo_*
- /api/products: id->{label} for search
- regenerated demo + report (10 sheets incl tubers_potato)
2026-09-24 07:35:39 +00:00
c6705c1cd9 feat: Excel export (in-browser SheetJS on /countries + CLI makexlsx report) + README
- countries.html: 'Скачать в Excel' button (SheetJS CDN)
- base.html: xlsx.full.min.js from CDNSheetJS
- src/export_excel.py: CLI → data/agromarket_report.xlsx (dynamics per product + spreads + alerts)
- README: upload + export docs
2026-09-24 07:15:28 +00:00
4a928c10d6 feat: manual CSV/HTML upload for sources without public API (Arbuz/Sharyn)
- src/pipeline/upload.py: parse_csv/parse_html → normalize→categorize→validate→upsert
- POST /api/upload (multipart) + form on /sources
- low-confidence records → review queue, not analytics
- +3 tests (16 passed, 1 skipped)
2026-09-24 07:13:33 +00:00
3bd5a322d1 docs: README для Python MVP; robust seeding in run.py (only if DB empty) 2026-09-24 07:10:23 +00:00
6c890fa20a v2: Python MVP — price pipeline + analytics + FastAPI dashboard + digest + tests
- pipeline: fetch/extract/normalize/categorize/validate/llm_extract/raw/orchestrator
- analytics: dynamics, spreads/arbitrage, seasonality, anomalies, alerts, forecast
- dashboard: FastAPI + Plotly (overview/products/product/countries/alerts/quality/sources)
- scheduler (APScheduler 07:30), digest (Telegram/SMTP dry-run), demo_data
- SQLite (PORTABLE) with idempotent upserts, quarantine, alerts dedup
- truthfulness: LLM quote-verified, source link + date on every price
- tests: 13 passed, 1 skipped (robots port)
2026-09-24 07:07:32 +00:00
5f18f7b502 fix: fallback-данные для Предпросмотра (белый экран) 2026-09-24 06:39:10 +00:00
fcd7575ebf v1 - Price Agent: дашборд агро-данных (валюты+погода), ИИ-аналитик, чат 2026-09-24 05:25:39 +00:00
e865a5840c Initial commit 2026-09-24 05:04:03 +00:00