AllergyTJ

Pollen risk & air quality tracker for Tajikistan — combines seasonal flora data, live weather, and real-time air quality monitoring for 30 cities.

To my knowledge, Tajikistan does not yet have a public pollen monitoring network. AllergyTJ bridges that gap by estimating pollen risk from curated botanical data and real-time weather conditions, alongside live air quality data (US AQI, PM2.5, PM10, and 4 other pollutants) from the EU Copernicus CAMS satellite model.

Live: allergy.tj

Screenshots

Dashboard	Forecast

Regions	Insights

---

Features

• 30 cities across all regions of Tajikistan
• 4-tab app — Dashboard, Forecast, Regions, Insights with clean path-based routing (/en/forecast, /ru/regions, /insights)
• Dashboard — circular SVG risk gauge, context-aware weather recommendations, hero weather block, top allergen triggers with severity bars, live AQI card with pollutant breakdown, daily health tip, nearby regions strip
• Pollen risk estimation — combines seasonal pollen calendar with real-time weather (temperature, humidity, wind, precipitation)
• 5-day forecast — daily pollen risk outlook with allergen chips, trend chart, and contextual insight tips
• Regional pollen map — interactive SVG map of Tajikistan with GADM boundaries, 30 city dots, per-region weather, city sidebar sorted by risk
• Push notifications — browser-only pollen, AQI, and UV alerts with quiet hours, bell icon settings overlay
• Hourly breakdown — morning/afternoon/evening risk scores using per-period weather
• Air quality — live US AQI with full 6-pollutant breakdown (O₃, PM2.5, PM10, NO₂, SO₂, CO)
• UV index — real-time UV data from Open-Meteo, color-coded by severity
• Insights tab — "What's Active Now" allergen cards, 12-month timeline calendar, 12 knowledge cards (incl. Cross-Reactivity Guide, Air Pollutants Guide, AQI methodology)
• GPS detection — auto-finds nearest city (HTTPS required)
• Zero config — no API keys, no build step, no framework, no external fonts
• Trilingual — English, Russian, and Tajik with browser auto-detection and URL-based language routing (/en/, /ru/, /tj/)
• PWA — installable on mobile/desktop, offline support, service worker caching
• Auto-refresh — weather data refreshes every 15 minutes and on returning to a stale tab; "Updated X min ago" timestamp on Dashboard
• 3-tier weather fallback — automatically switches to backup weather source (wttr.in) or seasonal-only mode if the primary API is down
• Dark mode — toggle with OS preference detection, persisted across sessions
• SEO-ready — Open Graph image, JSON-LD (WebApplication + FAQPage + BreadcrumbList), canonical URLs with language prefixes, geo meta tags, dynamic page title per city, translation preload hints, robots.txt, sitemap.xml with language alternates, hreflang tags, brand-matched SVG favicon
• Secure — A+ Mozilla HTTP Observatory; strict CSP with SHA-256 script hashes, HSTS preload, zero inline event handlers
• Accessible — WCAG AA color contrast, semantic HTML landmarks, ARIA labels
• Personal allergen profiles — species-level selector (9 allergens), personalized risk score, annotated triggers, profile-aware notifications
• Shareable daily snapshots — branded 1080×1080 PNG card (Canvas API) with risk score, allergen breakdown, AQI, weather, UV; Web Share API on mobile, download + clipboard fallback on desktop
• Telegram channel — @allergytj daily trilingual pollen and air quality updates for Dushanbe, auto-posted at 07:00 via Vercel Cron

How It Works

AllergyTJ estimates pollen risk by combining three data sources:

1. Seasonal pollen calendar — monthly intensity data for Tajikistan's known allergenic flora (poplar, cottonwood, willow, birch, plane tree, mulberry, grasses, mugwort, chenopod, ragweed)
2. Live weather from Open-Meteo — temperature, humidity, wind, and precipitation adjust the seasonal baseline (hot + dry + windy = higher risk; rain/snow = lower)
3. Air quality from Open-Meteo CAMS — real PM2.5, PM10, US AQI data

All computation runs client-side. No backend is needed in production.

Quick Start

Run locally (Python)

pip install -r requirements.txt
python api_proxy.py

Open http://localhost:5000

Run locally (no Python)

Open index.html directly in a browser. City selection works immediately; GPS requires HTTPS.

Project Structure

AllergyTJ/
├── index.html          # Frontend structure (HTML + CSS + structured data)
├── app.js              # All application JavaScript (rendering, routing, API calls)
├── sw.js               # Service worker (PWA caching, offline support)
├── manifest.json       # PWA manifest (app identity, icons, display mode)
├── build_html.py       # Build script: generates per-language HTML + copies assets to dist/
├── package.json        # Node config (ESM for serverless function)
├── lang/
│   ├── en.json         # English translations (~402 keys)
│   ├── ru.json         # Russian translations
│   └── tj.json         # Tajik translations
├── icons/              # PWA icons (10 PNGs + 2 master SVGs)
│   ├── icon.svg        # Standard icon source
│   └── icon-maskable.svg # Android adaptive icon (80% safe zone)
├── api/
│   ├── og.js           # Vercel serverless function (bot-aware OG tags per language)
│   └── telegram-post.js # Vercel serverless function (daily Telegram channel post via cron)
├── api_proxy.py        # Local dev server (Flask, serves index.html)
├── build_icons.py      # Dev tool: rasterizes SVG icons to all PWA PNGs via cairosvg
├── indexnow_submit.py  # Dev tool: batch-submits all 102 URLs to IndexNow (Bing, Yandex)
├── vercel.json         # Vercel deployment config + security headers + language routing
├── robots.txt          # Search engine crawl directives (incl. AI crawler allow rules)
├── sitemap.xml         # Generated sitemap (102 URLs with hreflang alternates)
├── og-image.png        # Tajik OG image (1200×630)
├── og-image-en.png     # English OG image
├── og-image-ru.png     # Russian OG image
├── requirements.txt    # Python dependencies (flask)
├── testing/
│   └── build_map_paths.py  # Dev tool: GADM polygon simplification + city containment
├── .gitignore
├── CODE_OF_CONDUCT.md
├── CONTRIBUTING.md
├── LICENSE             # MIT
└── README.md

Deployment

Vercel (production)

The site deploys as a static site + one serverless function on Vercel. Pushing to master triggers auto-deploy.

Configuration in vercel.json:

• framework: null (prevents auto-detection)
• buildCommand: "python3 build_html.py" — generates per-language HTML files + copies static assets to dist/
• outputDirectory: "dist" — Vercel serves from the build output
• Bot-detecting rewrites — social media bots (Telegram, Facebook, Twitter, etc.) get language-specific OG tags via api/og.js serverless function
• Language routing rewrites (/en/*, /ru/* → pre-rendered HTML)
• Security headers (CSP A+ with SHA-256 hashes, HSTS preload, X-Frame-Options, etc.)

Manual / other hosting

Copy index.html and app.js to any static file server. Requirements:

• Outbound HTTPS to api.open-meteo.com, air-quality-api.open-meteo.com, and wttr.in
• HTTPS for GPS geolocation (optional but recommended)

Data Sources

Source	Provides	Role	Key required
Open-Meteo Forecast API	Temperature, humidity, wind, precipitation	Primary weather	No
wttr.in	Temperature, humidity, wind, precipitation	Weather fallback	No
Open-Meteo Air Quality API	PM2.5, PM10, US AQI	Air quality	No
Curated seasonal calendar	Monthly pollen intensity by allergen	Baseline data	N/A

Tech Stack

• Frontend: Vanilla HTML + CSS + JS (two files, zero dependencies)
• Local dev: Python / Flask
• Hosting: Vercel (static)
• APIs: Open-Meteo (free, no key)

License

MIT © 2026 Eraj Ismatulloev

Built solo with AI-assisted development.