Implements full search functionality for StarPunk v1.1.0.
Search API Endpoint (/api/search):
- GET endpoint with query parameter (q) validation
- Pagination via limit (default 20, max 100) and offset parameters
- JSON response with results count and formatted search results
- Authentication-aware: anonymous users see published notes only
- Graceful handling of FTS5 unavailability (503 error)
- Proper error responses for missing/empty queries
Search Web Interface (/search):
- HTML search results page with Bootstrap-inspired styling
- Search form with HTML5 validation (minlength=2, maxlength=100)
- Results display with title, excerpt, date, and links
- Empty state for no results
- Error state for FTS5 unavailability
- Simple pagination (Next/Previous navigation)
Navigation Integration:
- Added search box to site navigation in base.html
- Preserves query parameter on results page
- Responsive design with emoji search icon
- Accessible with proper ARIA labels
FTS Index Population:
- Added startup check in __init__.py for empty FTS index
- Automatic rebuild from existing notes on first run
- Graceful degradation if population fails
- Logging for troubleshooting
Security Features:
- XSS prevention: HTML in search results properly escaped
- Safe highlighting: FTS5 <mark> tags preserved, user content escaped
- Query validation: empty queries rejected, length limits enforced
- SQL injection prevention via FTS5 query parser
- Authentication filtering: unpublished notes hidden from anonymous users
Testing:
- Added 41 comprehensive tests across 3 test files
- test_search_api.py: 12 tests for API endpoint validation
- test_search_integration.py: 17 tests for UI rendering and integration
- test_search_security.py: 12 tests for XSS, SQL injection, auth filtering
- All tests passing with no regressions
Implementation follows architect specifications from:
- docs/architecture/v1.1.0-validation-report.md
- docs/architecture/v1.1.0-feature-architecture.md
- docs/decisions/ADR-034-full-text-search.md
Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Release v1.1.0 "Searchlight" with search, custom slugs, and RSS fix.
Changes:
- Updated version to 1.1.0 in starpunk/__init__.py
- Updated CHANGELOG.md with v1.1.0 release notes
- Created implementation report in docs/reports/
Release highlights:
- Full-text search with FTS5 (core functionality complete)
- Custom slugs via Micropub mp-slug property
- RSS feed ordering fix (newest first)
- Migration system redesign (INITIAL_SCHEMA_SQL)
All features implemented and tested. Search UI to be completed
in immediate follow-up work.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Implements FTS5-based full-text search for notes as specified in ADR-034.
Changes:
- Created migration 005_add_fts5_search.sql with FTS5 virtual table
- Created starpunk/search.py module with search functions
- Integrated FTS index updates into create_note() and update_note()
- DELETE trigger automatically removes notes from FTS index
- INSERT/UPDATE handled by application code (files not in DB)
Features:
- Porter stemming for better English search
- Unicode normalization for international characters
- Relevance ranking with snippets
- Graceful degradation if FTS5 unavailable
- Helper function to rebuild index if needed
Note: Initial FTS index population needs to be added to app startup.
Part of v1.1.0 (Phase 3).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This aligns with ADR-033's migration system redesign. The initial schema
represents the v1.0.0 baseline and should not be modified. All schema
changes after v1.0.0 must go in migration files.
Changes:
- Renamed SCHEMA_SQL → INITIAL_SCHEMA_SQL in database.py
- Updated all references in migrations.py comments
- Added comment: "DO NOT MODIFY - This represents the v1.0.0 schema state"
- No functional changes, purely documentation improvement
Part of v1.1.0 migration system redesign (Phase 2).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Fixed bug where feedgen library was reversing the order of feed items.
Database returns notes in DESC order (newest first), but feedgen was
displaying them oldest-first in the RSS XML. Added reversed() wrapper
to maintain correct chronological order in the feed.
Added regression test to verify feed order matches database order.
Bug confirmed by testing:
- Database: [Note 2, Note 1, Note 0] (newest first)
- Old feed: [Note 0, Note 1, Note 2] (oldest first) ❌
- New feed: [Note 2, Note 1, Note 0] (newest first) ✅🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Remove leading slash when constructing URLs with SITE_URL
- SITE_URL already includes trailing slash per IndieAuth spec
- Fixes malformed Location header in Micropub responses
- Fixes malformed URLs in Microformats2 query responses
Changes:
- starpunk/micropub.py line 312: f"{site_url}notes/{note.slug}"
- starpunk/micropub.py line 383: f"{site_url}notes/{note.slug}"
- Added comments explaining SITE_URL trailing slash convention
- Updated version to 1.0.1 in starpunk/__init__.py
- Updated CHANGELOG.md with v1.0.1 release notes
Fixes double slash issue reported after v1.0.0 release.
Per ADR-039 and docs/releases/v1.0.1-hotfix-plan.md
Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
First production-ready release of StarPunk - a minimal, self-hosted
IndieWeb CMS with full IndieAuth and Micropub compliance.
Changes:
- Update version to 1.0.0 in starpunk/__init__.py
- Update README.md version references and feature descriptions
- Finalize CHANGELOG.md with comprehensive v1.0.0 release notes
This milestone completes all V1 features:
- W3C IndieAuth specification compliance with endpoint discovery
- W3C Micropub specification implementation
- Robust database migrations with race condition protection
- Production-ready containerized deployment
- 536 tests passing with 87% code coverage
StarPunk is now ready for production use as a personal IndieWeb
publishing platform.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Add comprehensive architectural documentation for the migration race
condition fix, including:
- ADR-022: Architectural decision record for the fix
- migration-race-condition-answers.md: All 23 Q&A answered
- migration-fix-quick-reference.md: Implementation checklist
- migration-race-condition-fix-implementation.md: Detailed guide
These documents guided the implementation in v1.0.0-rc.5.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This major architectural change removes the built-in IndieAuth
authorization server in favor of external provider delegation.
Key changes:
- Remove authorization and token endpoints
- Remove token storage and management code
- Implement external token verification via configured endpoint
- Drop auth_codes and auth_tokens database tables
- Simplify security model by delegating to external providers
Breaking Changes:
- Existing tokens issued by StarPunk will no longer work
- Users must configure TOKEN_ENDPOINT in settings
- Micropub clients must obtain tokens from external providers
Benefits:
- Reduces codebase by ~500 lines of security-critical code
- Eliminates token storage and cryptographic responsibilities
- Maintains full IndieAuth specification compliance
- Simpler security model focused on verification only
Implements: ADR-050 (Remove Authorization Server)
Implements: ADR-030 (External Token Verification)
Migration: Database migrations 003 and 004 included
See docs/reports/indieauth-removal-implementation-report.md for
complete implementation details and migration guide.
Version: 1.0.0-rc.4
Fixed 5 failing tests related to code_verifier column which was
added by migration 001 but removed by migration 003.
Changes:
- Renamed legacy_db_without_code_verifier to legacy_db_basic
- Updated column_exists tests to use 'state' column instead of 'code_verifier'
- Updated test_run_migrations_legacy_database to test with generic column
- Replaced test_actual_migration_001 with test_actual_migration_003
- Fixed test_dev_mode_requires_dev_admin_me to explicitly override DEV_ADMIN_ME
All 551 tests now passing.
Part of Phase 1 completion: IndieAuth authorization server removal
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Fixes database migration detection for partially migrated databases.
This hotfix resolves an issue where migration 002 would fail to detect
existing migrated tables, causing conflicts on databases that had been
partially migrated.
CRITICAL HOTFIX for production deployment failure
Problem:
- Production database had migration 001 applied but not migration 002
- Migration 002's tables (tokens, authorization_codes) already existed from SCHEMA_SQL
- Smart detection only checked when migration_count == 0 (fresh database)
- For partially migrated databases (count > 0), tried to run full migration
- This failed with "table already exists" error
Solution:
- Always check migration 002's state, regardless of migration_count
- If tables exist with correct structure, skip table creation
- Create missing indexes only
- Mark migration as applied
Testing:
- Manual verification with production scenario: SUCCESS
- 561 automated tests passing
- test_run_migrations_partial_applied confirms fix works
Impact:
- Fixes deployment on partially migrated production databases
- No impact on fresh or fully migrated databases
- Backwards compatible with all database states
Version: 1.0.0-rc.2 → 1.0.0-rc.3
Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
Hotfix 1.0.0-rc.2: Critical database migration fix
Resolves index conflict issue where migration 002 would fail on existing
databases due to duplicate index definitions in SCHEMA_SQL.
Fixes critical issue where migration 002 indexes already existed in SCHEMA_SQL,
causing 'index already exists' errors on databases created before v1.0.0-rc.1.
Changes:
- Removed duplicate index definitions from SCHEMA_SQL (database.py)
- Enhanced migration system to detect and handle indexes properly
- Added comprehensive documentation of the fix
Version bumped to 1.0.0-rc.2 with full changelog entry.
Refs: docs/reports/2025-11-24-migration-fix-v1.0.0-rc.2.md
- Update version to 0.9.5 throughout README
- Clarify Micropub as coming in v1.0 (currently in development)
- Add note that database auto-initializes on first run
- Fix deployment documentation link to standards location
- Add .gitignore entry for test.ini temporary file
All changes approved by architect agent.
Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Updated 42 references from indieauth.spec.indieweb.org to www.w3.org/TR/indieauth
- Ensures consistency across all documentation
- Points to the authoritative W3C specification
- No functional changes, documentation update only
Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
Implements token security and management as specified in ADR-029:
Database Changes (BREAKING):
- Add secure tokens table with SHA256 hashed storage
- Add authorization_codes table for IndieAuth token exchange
- Drop old insecure tokens table (invalidates existing tokens)
- Update SCHEMA_SQL to match post-migration state
Token Management (starpunk/tokens.py):
- Generate cryptographically secure tokens
- Hash tokens with SHA256 for secure storage
- Create and verify access tokens
- Create and exchange authorization codes
- PKCE support (optional but recommended)
- Scope validation (V1: only 'create' scope)
- Token expiry and revocation support
Testing:
- Comprehensive test suite for all token operations
- Test authorization code replay protection
- Test PKCE validation
- Test parameter validation
- Test token expiry
Security:
- Tokens never stored in plain text
- Authorization codes single-use with replay protection
- Optional PKCE for enhanced security
- Proper UTC datetime handling for expiry
Related:
- ADR-029: Micropub IndieAuth Integration Strategy
- Migration 002: Secure tokens and authorization codes
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This agent helps maintain documentation organization and ensures
README.md stays current with the codebase.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add comprehensive Micropub endpoint design document
- Define token management approach for IndieAuth
- Specify minimal V1 feature set (create posts, queries)
- Defer media endpoint and advanced features to post-V1
- Add ADR-028 documenting implementation strategy
- 8-10 day implementation timeline to unblock V1
The Micropub endpoint is the final blocker for V1.0.0 release.
Added "Documentation Navigation" section with:
- Clear explanation of docs/ folder structure and purpose of each subdirectory
- Guidelines for finding existing documentation before implementing features
- Practical rules for when to create ADRs, design docs, reports, or standards
- File naming conventions for different document types
This improves agent and developer ability to navigate the documentation
system and maintain proper organization standards.
Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Switched from GitLab CI to Gitea Actions for container builds.
Triggers on version tags, pushes to Gitea Container Registry.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Builds and pushes container images to GitLab Container Registry
when version tags (v*.*.*) are pushed.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
os.getenv() returns empty string instead of using default when env var
is set but empty. This caused SECRET_KEY to be empty when FLASK_SECRET_KEY=""
was in .env, breaking Flask sessions/flash messages.
Now treats empty string same as unset, properly falling back to SESSION_SECRET.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
IndieAuth authentication-only flows should redeem the code at the
authorization endpoint, not the token endpoint. The token endpoint
is only for authorization flows that need access tokens.
- Remove grant_type parameter (only needed for token flows)
- Change endpoint from /token to /authorize
- Update debug logging to reflect code verification flow
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
