Match padding and sizing for active page number and next button
to match inactive page buttons.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Simple preferences page with engine selection
- Light/dark theme toggle with localStorage persistence
- Clean form layout without complex JS dependencies
- Add dark theme CSS variables
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Wikipedia returns HTML entities like <span> which were being
double-escaped by Go templates. Now using html.UnescapeString and
template.HTML to render properly.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace React SPA with simple Go templates using search-zen-50
visual style. No JavaScript required - pure HTML/CSS with clean
teal accent color scheme, monospace logo, and minimal design.
- Simplified base.html without HTMX or autocomplete JS
- Clean homepage with centered search box
- Results page with sticky header and category tabs
- Simplified CSS matching search-zen-50 aesthetics
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Creates internal/spa package that:
- Embeds React build output from cmd/kafka/dist/
- Provides HTTP handler for static file serving
- Falls back to index.html for SPA client-side routing
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Reverted CSS to the known-working state at 4b0cde9, then re-applied
only the image grid styles. The duplicate .results-layout block is
intentional — it was present in the working version too.
The old 3-column layout block (referencing .left-sidebar/.right-sidebar
classes that don't exist in the HTML) was overriding the correct layout
defined earlier. Removed the stale duplicate.
Three new image search engines:
- bing_images: Bing Images via RSS endpoint
- ddg_images: DuckDuckGo Images via VQD API
- qwant_images: Qwant Images via v3 search API
Frontend:
- Image grid layout with responsive columns
- image_item template with thumbnail, title, and source metadata
- Hover animations and lazy loading
- Grid activates automatically when category=images
Backend:
- category=images routes to image engines via planner
- Image engines registered in factory and engine allowlist
- extractImgSrc helper for parsing thumbnail URLs from HTML
- IsImageSearch flag on PageData for template layout switching
- results-layout: 3-column grid (1fr | min(768px,100%) | 300px) max-width 1400px, centered
- Widen center results column to 768px max
- Right column (formerly sidebar): sticky, contains knowledge panel + related searches
- Knowledge panel: Wikipedia/infobox summary with optional thumbnail
- Related searches: clickable links to refine the query
- Empty left buffer creates balanced whitespace on large screens
- Responsive: 2-col at 1000px, 1-col at 700px
New brave.go: scrapes https://search.brave.com directly.
Extracts title, URL, snippet, and favicon from Brave's HTML.
No API key required.
Rename existing BraveAPIEngine (was BraveEngine) to avoid collision
with the new scraper. API engine stays as 'braveapi', scraper as 'brave'.
Go's regexp package doesn't support Perl lookahead (?=...). Removing
the unnecessary lookahead since each MjjYud div is self-contained.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Wrap sidebar time/type filters in a form with HTMX attributes so
filter changes trigger partial page updates instead of full reload.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Wikipedia language subdomain was derived from user input without
validation, allowing attackers to redirect requests via malicious
language values like "evil.com.attacker.com". Added a whitelist of
valid Wikipedia language codes to prevent this.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Brave API only supports offset values 0-9. When pageno > 1 with
resultsPerPage=20, offset exceeded this limit causing 422 errors.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Test GET /healthz, /, /search, /autocompleter endpoints.
Verify response codes, content types, JSON decoding, empty-query
redirect, and source URL presence in footer.
Also fix dead code in Search handler: the redirect for empty q
was unreachable because ParseSearchRequest errors on empty q first.
Move the q/format check before ParseSearchRequest to fix the redirect.
Trim or remove comments that:
- State the obvious (function names already convey purpose)
- Repeat what the code clearly shows
- Are excessively long without adding value
Keep comments that explain *why*, not *what*.
Thread source_url through: config.ServerConfig → Handler.sourceURL
→ PageData.SourceURL → template footer. Footer only shows Source
link when source_url is set.
Update LICENSE file and add AGPL header to all source files.
AGPLv3 ensures that if someone runs Kafka as a network service and
modifies it, they must release their source code under the same license.
- New CSS: complete design system with CSS variables, modern color palette
- Homepage: full-viewport hero with centered search, logo, tagline
- Result cards: rounded, shadowed, with favicons via Google Favicon API
- Layout: sidebar + results grid, responsive
- Typography: proper font stack, variable weights
- Settings panel: polished popover with animations
- Autocomplete: modern dropdown with keyboard nav
- Dark mode: full color palette via data-theme attribute
- Favicon: clean search icon SVG
- google.go: use inline (?s) flag instead of regexp.DotAll second arg
- youtube.go: remove Metadata field (not in MainResult contract)
- config_test.go: fix expected engine count from 9 to 11 (google+youtube)
html/template requires template names to be string literals, not field
accesses. Use {{if eq .Template "videos"}} to branch and call the
appropriate template by literal name.
Go html/template doesn't support function calls as template names in
{{template (func .Arg) .}}. Instead, precompute TemplateName in
FromResponse and use {{template .TemplateName .}} in the template.
MainResult: add Thumbnail field (used by YouTube, images, etc.)
video_item.html: new partial for video results with thumbnail display
views.go: add templateForResult func + video_item.html to template parse
results_inner.html: dispatch to video_item when Template="videos"
kafka.css: add .video-result flex layout with thumbnail styling
- Replace document.body.innerHTML with panel.querySelector('.settings-popover-body').innerHTML
- Use theme buttons (.theme-btn) with icons instead of radio buttons
- Use .engine-toggle class for engine checkboxes in 2-column grid
- Include settings-notice paragraph for engine changes
- Use dropdowns for safe search and format with proper ids
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
YouTube Data API v3 engine:
- Add YouTubeConfig to EnginesConfig with api_key field
- Add YOUTUBE_API_KEY env override
- Thread *config.Config through search service to factory
- Factory falls back to env vars if config fields are empty
- Update config.example.toml with youtube section
Also update default local_ported to include google and youtube.
Uses the official YouTube Data API v3. Requires YOUTUBE_API_KEY
environment variable (free from Google Cloud Console).
Returns video results with title, description, channel, publish
date, and thumbnail URL. Falls back gracefully if no API key.
