Replace username/password authentication with FIDO2/WebAuthn passkey-based
auth using webauthn-rs. Sessions and bearer tokens are unchanged — only the
way they are created changes.
- Add webauthn-rs, uuid, open, url deps; remove argon2, rpassword
- Add passkey_credentials and registration_tokens tables (migrations 17-18)
- Add domain entities, typed IDs, and repository traits for passkeys/tokens
- Add SQL repository implementations for passkeys and registration tokens
- Add ChallengeStore for in-memory WebAuthn ceremony state
- Add WebAuthn route handlers (register/auth start+finish ceremonies)
- Add CLI browser handoff for token creation (opens browser, local callback)
- Replace login form with "Sign in with Passkey" button
- Add registration page for first-user bootstrap via one-time token
- Replace BREWLOG_ADMIN_USERNAME/PASSWORD with BREWLOG_RP_ID/RP_ORIGIN
- Change default BREWLOG_URL from 127.0.0.1 to localhost (WebAuthn requires it)
- Extract 35-line filter_map closure into parse_cafe() in foursquare.rs
- Unify triple-duplicated search branching in pagination.rs into
fetch_records() and fetch_count() helpers
- DRY three identical gear-loading blocks in brews.rs via
load_gear_options() helper
- Flatten nested if/else-if in timeline map_details() using match,
extract shared is_blank() helper
- Deduplicate JSON decode pattern in backup.rs with decode_json_vec()
and decode_json_opt() helpers
- Add UpdatedAt variant to BagSortKey
- Sort home page bags by updated_at desc instead of roast_date
- Limit to 3 bags instead of showing all open bags
- Add migration 0016 to drop notes column from roasters table
- Remove notes from domain structs, repositories, views, CLI, routes
- Remove notes from AI extraction prompts and ExtractedRoaster struct
- Remove notes from scan page form and submission handler
- Update templates, bootstrap script, backup module, and tests
- Add migration 0014 to drop the notes column from cafes table
- Remove notes from domain structs, repository, view model, CLI, routes
- Remove notes textarea/column from web templates
- Update bootstrap script and all test files
- Implement CupRepository trait for SqlCupRepository
- Transactional insert creates cup and timeline event atomically
- Support filtering by cafe_id/roast_id, search across roast/roaster/cafe names
- Use QueryBuilder with push_update_field! macro for updates
Position is already shown as a clickable detail row in the card body,
making the title icon redundant. Also removes the unused map_link field
from TimelineEventView.
Implement SqlCafeRepository with CRUD operations and timeline event
creation. Add CafesClient for CLI HTTP access. Wire cafes into backup
export/restore and empty-database verification.
- Add empty_to_none() to replace four repeated empty-string-to-None blocks
- Extract insert_timeline_event() from insert() to isolate timeline logic
- Remove #[allow(clippy::too_many_lines)] from RoastRepository impl
- Combine Roast/Roaster into single column with roaster as subtext on desktop
- Combine Origin/Producer into single column with producer as subtext on desktop
- Add Roaster and Producer as separate mobile-only fields for card layout
- Replace tasting notes pill badges with comma-separated list
- Rename "Notes" column to "Tasting Notes" with right-aligned mobile text
- Include tasting_notes in repository search filter columns
Add SearchFilter struct for LIKE-based search with dynamic column
matching. Add DatabaseDriver type alias to support QueryBuilder
generics across SQLite and Postgres feature flags.
- Add filter_paper_id to SQL queries (SELECT, INSERT, LEFT JOIN)
- Add filter_paper_name via LEFT JOIN on gear table
- Update BrewRecord and BrewWithDetailsRecord structs
- Add filter_paper_id parameter to HTTP client create method
Database:
- Add slug, roaster_slug, brew_data_json columns to timeline_events
- Migration backfills existing data from related tables
- Remove 9-way LEFT JOIN from list query, read directly from columns
Frontend:
- Use CSS :nth-of-type(odd/even) for alternating timeline layout
- Remove JavaScript class manipulation when appending month events
- Simplify infinite scroll month-merging logic
This eliminates query-time JOINs across 5 tables and ~20 lines of
client-side JavaScript for pattern maintenance.
- Simplify timeline layout to single-column without sidebar
- Update tests to explicitly pass page_size for pagination testing
- Include sticky headers and a neater alternating, side-by-side
timeline design
- Add TimelineBrewData struct to domain layer to carry raw brew
parameters needed for repeating a brew
- Extend timeline SQL query to LEFT JOIN with brews table and fetch
brew data (bag_id, grinder_id, brewer_id, coffee_weight, etc.)
- Add brew_data field to TimelineEventView for template access
- Add is_authenticated to TimelineChunkTemplate for auth-aware chunks
- Add "brew again" button to timeline month partial, visible only for
authenticated users on brew events
- Server returns reload script when brew is created from timeline page,
using Datastar's fragment patching to append script to body
- Add SqlBrewRepository with transactional insert that deducts from bag
- Return Conflict error when insufficient coffee in bag
- Add BrewsClient for CLI HTTP operations
- Add [lints.clippy] section to Cargo.toml with pedantic group
- Cherry-pick restriction lints: dbg_macro (deny), todo, unwrap_used, expect_used (warn)
- Allow noisy pedantic lints (missing_errors_doc, module_name_repetitions, etc.)
- Fix unnecessary Result wrappers in token and user repository to_domain functions
- Merge duplicate match arms in TimelineEventViewModel
- Add justified #[allow] attributes for startup code and tests
Previously, list endpoints returned enriched types (RoastWithRoaster,
BagWithRoast) with related entity names, while get/create/update
endpoints returned bare types without this information.
This change makes all endpoints consistent by returning enriched types:
- Added get_with_roaster and get_with_roast repository methods
- Created define_enriched_get_handler! macro for custom getter methods
- Updated create and update handlers to fetch enriched data after write
- Updated CLI client to expect enriched types
The notes field was not providing enough value to justify its presence.
Simplified the Gear entity by removing notes from:
- Domain structs (Gear, NewGear, UpdateGear)
- SQL repository queries and GearRecord
- HTTP client methods
- CLI commands (--notes flag)
- Web views and templates
- All related tests
Added migration 0008_remove_gear_notes.sql to drop the column.
- Allow too_many_arguments for AppState::new since 8 repos are needed
- Implement FromStr trait for GearCategory instead of custom from_str
method to follow Rust conventions
- Update callers to use map_err for Result handling
Implement complete web interface for the Gear entity:
- Add GearView model with category badges and formatted display
- Create GearTemplate and GearListTemplate for Askama rendering
- Build main gear page with collapsible add form (Datastar-powered)
- Implement gear list table with sortable columns and pagination
- Use trash icon for delete actions matching roasts table design
- Add Gear navigation link in main menu between Bags and Timeline
- Integrate gear events into timeline view with proper labels and links
The web UI follows the established patterns from other entities with
Datastar for reactive fragment updates and proper authentication gating.
Apply code formatting fixes across all gear-related modules.
Replace string matching on "UNIQUE constraint failed" with SQLx's
type-safe is_unique_violation() method in roasters.rs and roasts.rs.
This aligns with users.rs and tokens.rs, and ensures the check works
across different database backends (SQLite, PostgreSQL) without
depending on error message formatting.
- Use method access (sort_key(), sort_direction()) instead of field
access for consistency with roasters.rs and roasts.rs
- Add secondary sort clauses (e.g., created_at DESC) as tiebreakers
to ensure deterministic pagination results
Document why direct format!() interpolation is acceptable in
build_where_clause(): the values are type-safe (bool outputs literal
TRUE/FALSE, roast_id is i64 from typed wrapper). Warns future
developers to use parameterized queries if string fields are added.
- Add BagFilter struct with constructor methods (all, open, closed, for_roast)
- Replace 5 repository methods with single list(filter, request) method
- Add build_where_clause helper for dynamic WHERE clause construction
- Update all callers in bags and roasts routes
This eliminates method explosion when adding new filters - now only
BagFilter and build_where_clause need updating instead of adding
new repository methods.
- Add push_update_field! macro to reduce UPDATE query boilerplate
- Rename sort_clause() to order_clause() for consistency
- Convert bags.rs update method from string concatenation to QueryBuilder
- Apply macro to roasters.rs, roasts.rs, and bags.rs update methods
This reduces ~100 lines of repetitive code and ensures consistent
patterns for building dynamic UPDATE queries across all repositories.
Code Quality Improvements:
- Fix hex literal grouping in ID generator (0xB10C_1D -> 0x00B1_0C1D)
- Rename ListQuery::default() to default_query() to avoid confusion with Default trait
- Use div_ceil() instead of manual ceiling division
- Remove unnecessary borrows in auth token generation and hashing
- Simplify nested if statements in error handling
Documentation:
- Add comprehensive authentication section to README
- Document environment variables for server and CLI
- Add security best practices and considerations
- Document password hashing (Argon2id), token storage (SHA-256), and session management
- Include step-by-step authentication setup guide
- Add production deployment recommendations
All 70 tests pass (8 unit + 46 server + 16 CLI) ✅
Co-authored-by: jnsgruk <668505+jnsgruk@users.noreply.github.com>
- Add sessions table to store session tokens with expiration
- Create Session domain model and SessionRepository trait
- Implement SqlSessionRepository for session persistence
- Update is_authenticated() to validate tokens against database
- Sessions expire after 30 days
- Session tokens hashed with SHA-256 before storage
- Delete sessions from database on logout
- Update all page handlers to properly validate sessions
This prevents session hijacking by ensuring only valid, unexpired
tokens stored in the database can authenticate requests.
Co-authored-by: jnsgruk <668505+jnsgruk@users.noreply.github.com>
