Add docker-compose

2026-08-08 21:05:14 +10:00 · 2026-03-07 12:15:53 +11:00
parent 102f89a4bb
commit 9c9c52bf8b
3 changed files with 121 additions and 49 deletions
@@ -3,13 +3,15 @@
 ## Search

 - **Web search** — aggregates results from Brave, DuckDuckGo, Google, and Bing concurrently
- **Image search** — Pinterest-style masonry layout respecting natural image aspect ratios, with lightbox viewer and keyboard navigation (←/→ to browse, Escape to close)
- **Image size filter** — filter images by size (Large, Medium, Small) — applied server-side across all engines
- **Autocomplete** — live search suggestions as you type (cascades Google → DuckDuckGo → Brave)
+- **Image search** — masonry grid respecting natural aspect ratios, with lightbox viewer and keyboard navigation (←/→ to browse, Escape to close)
+- **Image size filter** — filter images by size (Large, Medium, Small) applied server-side across all engines
+- **Sort order** — sort results by engine rank (default), newest first, or oldest first
+- **Autocomplete** — live search suggestions as you type (200ms debounce, ↑↓/Enter/Escape keyboard support)
 - **Pagination** — navigate through result pages; URL reflects current state (`?q=...&page=2`)
 - **URL deduplication** — duplicate results from multiple engines are merged automatically
- **Domain exclusion** — exclude specific websites from search results; settings persist in SQLite
- **Search stats** — per-engine result counts, status indicators (ok/error/timeout), collapsible sidebar
+- **Domain exclusion** — exclude specific websites from all search results; settings persist in SQLite
+- **Engine filter** — restrict a search to specific engines via the `engines` query parameter (e.g. `engines=google,bing`)
+- **Per-engine stats** — result counts, status (ok / error / timeout), and error messages per engine shown in a collapsible bar

 ## Search Engines

@@ -18,80 +20,110 @@
 - **DuckDuckGo** — HTML Lite scraping for web, API for images
 - **Brave** — Brave Search API for web and images

-## REST API
-
- Full REST API for search, autocomplete, and engine management
- `/api/search` accepts both **GET and POST** requests with query string parameters
- Compatibility parameters on `/api/search`: `pageNumber` (alias for `page`), `numResults`, `format`, `imageProxy`, `safesearch`
- OpenAPI specification with interactive docs via [Swagger UI](https://swagger.io/tools/swagger-ui/) (`/docs`) and [Redoc](https://github.com/Redocly/redoc) (`/redoc`)
-
 ## Engine Management

 - **4 built-in engines**: Brave, DuckDuckGo, Google, Bing
- Enable or disable engines via the UI or API at runtime
+- Enable or disable engines via the UI or API at runtime; settings persist in SQLite
+- **Drag-and-drop reordering** — set engine priority order; affects result ordering and display badges
 - Each engine supports both web and image search categories
 - Google and Bing include automatic fallback mechanisms for resilience

-## Reliability
+## Search History

- **Retry mechanism** — failed upstream requests are retried with exponential backoff (via [tenacity](https://github.com/jd/tenacity))
- **Error reporting** — when an upstream engine fails or times out, the UI shows a toast notification identifying which engine had issues
- **Graceful degradation** — partial failures don't block results from other engines
+- **Automatic recording** — every search query is saved with timestamp and category
+- **History page** (`/history`) — paginated, date-grouped list of recent searches (Today, Yesterday, etc.)
+- **Re-run any query** — click a history entry to instantly re-search it
+- **Delete entries** — remove individual entries or clear all history at once
+
+## Analytics Dashboard
+
+- **Stats page** (`/stats`) — full analytics dashboard with configurable time range (7 / 30 / 90 days)
+- **KPI cards** — total searches, total clicks, click-through rate, average clicked result position
+- **Daily trend chart** — bar chart of search volume per day
+- **Top queries** — ranked list of most-searched terms with counts
+- **Clicks by engine** — bar chart showing which engines' results get clicked most
+- **Click position distribution** — histogram of which result ranks users click on
+- **Top clicked results** — ranked list of most-clicked URLs with title, engine, and count
+- **Click tracking** — every result click is recorded server-side (fire-and-forget from the frontend)

 ## Caching

- **Redis cache** — identical searches served from Redis cache to reduce upstream load and latency
+- **Redis cache** — identical searches served from Redis to reduce upstream load and latency
 - **Configurable TTL** — cache duration adjustable from 0 (disabled) to 168 hours (1 week); default 6 hours
- **UI controls** — preset buttons + slider in Settings → Cache tab, with flush button
+- **Cache status** — search response includes a `cached` flag; the stats bar indicates cache hits
+- **UI controls** — TTL slider + preset buttons, Redis URL input, and flush button in Settings → Cache tab
 - **Optional** — when `REDIS_URL` is not set or Redis is unreachable, caching is silently disabled
 - **Deterministic keys** — cache key derived from query + category + page + image_size + engines

 ## Bookmarks

- **Bookmark any result** — click the bookmark icon on web or image search results to save them
- **Bookmarks page** — dedicated `/bookmarks` page to browse all saved items
+- **Bookmark any result** — click the bookmark icon on web or image results to save them
+- **Bookmarks page** (`/bookmarks`) — dedicated page to browse all saved items
 - **Filter by type** — tabs to filter All / Web / Images
- **Masonry image grid** — saved images displayed in a masonry layout
+- **Masonry image grid** — saved images displayed in a masonry layout with hover overlays
 - **Remove bookmarks** — delete individual bookmarks from the bookmarks page
 - **Persistent storage** — bookmarks stored in SQLite alongside other app data

+## Background Images
+
+- **Homepage backgrounds** — random background image displayed on the home page (sourced from Unsplash / Picsum)
+- **Local caching** — images are downloaded and stored in `data/backgrounds/` to avoid re-fetching
+- **Auto-refresh** — configurable refresh interval (1–1440 minutes, default 30) fetches new images automatically
+- **Background gallery** (`/backgrounds`) — browse and manage all downloaded backgrounds; click to preview, view file size and source
+- **Manual refresh** — refresh button in the gallery to fetch a new image immediately
+- **Enable / disable** — toggle backgrounds on or off in Settings → Background tab
+- **iOS color sampling** — samples edge pixels from the background to set the `theme-color` meta tag, preventing white/black bars in Safari
+
+## Settings
+
+- **Unified settings modal** — single ⚙ button opens a tabbed modal with four tabs:
+  1. **Engines** — enable/disable and drag-to-reorder search engines
+  2. **Excluded Sites** — add or remove domains from the search blocklist
+  3. **Cache** — view Redis status, set TTL, update Redis URL, flush cache
+  4. **Background** — enable/disable background images, set refresh interval
+
+## REST API
+
+- Full REST API for search, autocomplete, engine management, bookmarks, history, stats, and settings
+- `/api/search` accepts both **GET and POST** requests with query string parameters
+- Compatibility parameters: `pageNumber` (alias for `page`), `numResults`, `format`, `imageProxy`, `safesearch`
+- Every result includes `result_id`, `rank`, `engine`, and `published_date`
+- `has_next` and `total_results` fields for cursor-aware pagination
+- `X-Response-Time-Ms` response header on all endpoints
+- OpenAPI specification with interactive docs via [Swagger UI](https://swagger.io/tools/swagger-ui/) (`/docs`) and [Redoc](https://github.com/Redocly/redoc) (`/redoc`)
+
+## Reliability
+
+- **Retry mechanism** — failed upstream requests are retried with exponential backoff (via [tenacity](https://github.com/jd/tenacity), 2 attempts)
+- **Error reporting** — when an upstream engine fails or times out, the UI shows a toast notification identifying which engine had issues
+- **Graceful degradation** — partial failures don't block results from other engines
+
 ## UI / UX

 - **Mobile-first** responsive design built with React, Tailwind CSS, and shadcn theming
- Clean search home page with branded gradient header
- Web results show favicons, engine badges, and content snippets
- Image results displayed in a masonry grid with hover previews and a full lightbox with arrow-key navigation
- **Visited link colors** — clicked links turn purple to distinguish from unvisited results
- Engine settings modal with toggle switches
- Excluded domains management modal with add/remove UI
- **Unified settings modal** — single Settings button opens a tabbed modal (Engines, Excluded Sites)
+- Web results show favicons, engine badges, publish dates, and content snippets
+- Image results in a masonry grid with hover previews and a full-screen lightbox
+- **Visited link colours** — clicked links turn purple to distinguish from unvisited results
+- **Dropdown navigation menu** — consistent ☰ menu across all pages (search, bookmarks, history, stats, gallery)
 - Keyboard navigation for autocomplete suggestions (↑↓ arrows, Enter, Escape)
- Footer with link to interactive API documentation (Swagger UI)
+- URL-based routing with browser history support — shareable search URLs

 ## Accessibility

 - **Skip to main content** link for keyboard users
- Image cards are real `<a>` links (Cmd/Ctrl-click, right-click, open-in-new-tab)
+- Image cards are real `<a>` links (Cmd/Ctrl-click, right-click, open-in-new-tab all work)
 - `aria-label` on all icon-only buttons and interactive elements
 - Search input with `<label>`, `name`, `type="search"`, and ARIA combobox pattern
 - Visible `focus-visible` ring on all focusable elements
- `aria-live` regions for loading/results/error announcements
+- `aria-live` regions for loading / results / error announcements
 - Proper heading hierarchy (`h1`/`h2`) and landmark elements (`<nav>`, `<main>`, `<aside>`)
 - Dialog semantics on modals and lightbox (`role="dialog"`, `aria-modal`)

-## REST API
-
- Full REST API for search, autocomplete, and engine management
- `/api/search` accepts both **GET and POST** requests with query string parameters
- Compatibility parameters on `/api/search`: `pageNumber` (alias for `page`), `numResults`, `format`, `imageProxy`, `safesearch`
- Every result includes `result_id`, `rank`, `engine`, and `timestamp` for agent integration
- `has_next` and `total_results` fields for cursor-aware pagination
- Standardized error schema (`code`/`message`/`details`/`retry_hint`) on all error responses
- `X-Response-Time-Ms` and rate-limit headers on all responses
- Copy-paste `curl` examples in OpenAPI docs for every endpoint
- OpenAPI specification with interactive docs via [Swagger UI](https://swagger.io/tools/swagger-ui/) (`/docs`) and [Redoc](https://github.com/Redocly/redoc) (`/redoc`)
-
 ## Deployment

 - **Docker** — multi-stage Dockerfile: Node.js frontend build → Python production image
- Backend uses [uv](https://docs.astral.sh/uv/) for fast, reproducible dependency management
+- **GHCR image** — pre-built image published to GitHub Container Registry on every release
+- **Data volume** — `/app/data` stores SQLite databases and background images; mount for persistence
+- **Environment variables** — `DATA_DIR` (data directory path), `REDIS_URL` (optional Redis connection string)
+- Backend uses [uv](https://docs.astral.sh/uv/) for fast, reproducible Python dependency management
+
@@ -53,19 +53,28 @@ just dev
 REDIS_URL=redis://192.168.1.2:6399 just dev
 ```

-### 2. Docker
+### 2. Docker Compose
+
+The easiest way to run HeySearch with Redis caching in one command:

 ```bash
+docker compose up -d
+```

+Then open [http://localhost:8000](http://localhost:8000). Data and Redis are persisted in named volumes automatically.
+
+### 3. Docker (standalone)
+
+```bash
 # Basic — data stored in anonymous volume
-docker run -p 8000:8000  ghcr.io/wahyd4/hey-search
+docker run -p 8000:8000 ghcr.io/wahyd4/hey-search

 # Recommended — mount data directory for persistence
-docker run -p 8000:8000 -v ./hey-search-data:/app/data hey-search
+docker run -p 8000:8000 -v ./hey-search-data:/app/data ghcr.io/wahyd4/hey-search

 # With Redis
 docker run -p 8000:8000 -v ./hey-search-data:/app/data \
-  -e REDIS_URL=redis://your-redis:6379 hey-search
+  -e REDIS_URL=redis://your-redis:6379 ghcr.io/wahyd4/hey-search
 ```

 The `/app/data` volume stores the SQLite database (engine settings, excluded domains, cache config). Mount it to preserve your settings across container restarts.
@@ -0,0 +1,31 @@
+services:
+  app:
+    image: ghcr.io/wahyd4/hey-search:latest
+    build: .
+    ports:
+      - "8000:8000"
+    environment:
+      - REDIS_URL=redis://redis:6379
+      - DATA_DIR=/app/data
+    volumes:
+      - hey-search-data:/app/data
+    depends_on:
+      redis:
+        condition: service_healthy
+    restart: unless-stopped
+
+  redis:
+    image: redis:7-alpine
+    volumes:
+      - hey-search-redis:/data
+    command: redis-server --save 60 1 --loglevel warning
+    healthcheck:
+      test: ["CMD", "redis-cli", "ping"]
+      interval: 5s
+      timeout: 3s
+      retries: 5
+    restart: unless-stopped
+
+volumes:
+  hey-search-data:
+  hey-search-redis: