Files
stegasoo/PLAN-4.1.2.md
Aaron D. Lee 7a35ac3df7 Update plan: mark #6 Smoke Test Benchmarking as done
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-05 19:15:05 -05:00

251 lines
7.2 KiB
Markdown

# Stegasoo 4.1.2 Plan
## Release Theme
Polish and UX improvements after the 4.1.1 stability release.
---
## 1. Real Progress Bar for Encode/Decode
**Status:** Planned
**Problem:** Users see elapsed time but no indication of how far along the operation is. Long DCT encodes on Pi can take 2-3 minutes with no feedback.
**Solution:** Polling + progress file approach
### Backend Changes
1. **dct_steganography.py** - Write progress during block loop:
```python
if progress_file and block_num % 50 == 0:
with open(progress_file, 'w') as f:
json.dump({"current": block_num, "total": total_blocks, "phase": "embedding"}, f)
```
2. **app.py** - New endpoints:
- `POST /encode` returns `job_id`, starts subprocess
- `GET /encode/progress/<job_id>` returns progress JSON
- `GET /encode/result/<job_id>` returns final result when done
3. **Subprocess wrapper** - Pass progress file path to encode/decode functions
### Frontend Changes
1. **stegasoo.js** - After form submit:
- Show progress bar (Bootstrap progress component)
- Poll `/encode/progress/{job_id}` every 500ms
- Update bar width and percentage text
- Show phase (hashing, embedding, encoding, etc.)
2. **Templates** - Add progress bar markup to encode.html and decode.html
### Files to Modify
- `src/stegasoo/dct_steganography.py`
- `frontends/web/app.py`
- `frontends/web/static/js/stegasoo.js`
- `frontends/web/templates/encode.html`
- `frontends/web/templates/decode.html`
---
## 2. Granular Decode Error Messages
**Status:** Done
**Problem:** Decode failures show generic "Decryption failed" - users don't know if it's wrong photo, wrong passphrase, wrong PIN, corrupted image, or format mismatch.
**Solution:** Bubble up specific error types from library to UI
### Implementation
- Added new exceptions: InvalidMagicBytesError, ReedSolomonError, NoDataFoundError, ModeMismatchError
- DCT decode now raises InvalidMagicBytesError for wrong magic bytes
- DCT decode now raises ReedSolomonError (renamed from reedsolo's) for corruption
- app.py catches specific exceptions with user-friendly messages:
- Invalid magic → "Try a different mode (LSB/DCT)"
- RS error → "Image too corrupted, may have been re-saved"
- Invalid header → "Image may have been modified"
- Decryption error → "Wrong credentials"
### Files Modified
- `src/stegasoo/exceptions.py` (new exceptions)
- `src/stegasoo/__init__.py` (exports)
- `src/stegasoo/dct_steganography.py` (raise specific exceptions)
- `frontends/web/app.py` (catch and display)
---
## 3. Mobile-Responsive Polish
**Status:** Planned
**Problem:** UI works on mobile but has rough edges - cramped buttons, hard-to-tap targets, awkward layouts on small screens.
**Solution:** Targeted CSS/layout fixes for mobile breakpoints
### Areas to Improve
1. **Encode/Decode Forms:**
- Stack image drop zones vertically on mobile (currently side-by-side)
- Larger touch targets for file inputs
- Full-width buttons on small screens
- Passphrase input readable at smaller sizes
2. **Navigation:**
- Hamburger menu for mobile navbar (if not already)
- Sticky header doesn't eat too much screen
- Easy thumb reach for main actions
3. **Results/Output:**
- Download buttons full-width on mobile
- QR codes sized appropriately
- Click-to-copy message box works well with touch
4. **Drop Zones:**
- Larger tap targets
- Visual feedback for touch (not just hover)
- Camera integration hint on mobile ("Tap to take photo or choose file")
### Testing Targets
- iPhone SE (small)
- iPhone 14 (medium)
- iPad (tablet)
- Android Chrome
### Files to Modify
- `frontends/web/static/css/style.css` (or new mobile.css)
- `frontends/web/templates/encode.html`
- `frontends/web/templates/decode.html`
- `frontends/web/templates/base.html` (navbar)
---
## Testing Checklist
- [ ] Progress bar works on localhost
- [ ] Progress bar works on Pi (slower, more visible)
- [ ] Cancellation handling (what if user navigates away?)
- [ ] Error states display correctly
- [ ] Smoke test passes
---
## 4. Forced First-Login Setup
**Status:** Done
**Problem:** Users can navigate the app without creating an admin account first. Should force password setup before anything else.
**Solution:** Middleware/decorator that redirects to setup page if no users exist.
### Implementation
- Added `@app.before_request` hook that redirects to /setup if no users exist
- Skips redirect for static files and setup-related routes
### Files Modified
- `frontends/web/app.py` (added require_setup before_request hook)
---
## 5. Dropzone UX Fixes
**Status:** Done
**Problem:** Dropzone has some interaction bugs:
- Dropzone doesn't clear properly if first QR image fails
- Can't click on image preview to replace file (have to click surrounding border)
**Solution:** Fix JS event handling and state management
### Implementation
- Added click handler on preview images to trigger file input
- Made entire drop zone clickable (not just label)
- QR zone now resets after 2 seconds on error, allowing retry
- Clear file input on QR error so same file can be re-selected
### Files Modified
- `frontends/web/static/js/stegasoo.js`
---
## 6. Smoke Test Benchmarking
**Status:** Done
**Problem:** No way to measure encode/decode performance or track regressions.
**Solution:** Add timing to smoke tests using `hyperfine` or `time`.
### Implementation
- Added `--benchmark` flag to run encode/decode benchmarks after tests
- Added `--runs=N` flag to customize number of benchmark runs (default: 5)
- Uses hyperfine if available for precise timing with warmup
- Falls back to manual timing with bc if hyperfine not installed
- Outputs min/max/avg stats for both encode and decode operations
### Files Modified
- `tests/smoke-test.sh`
---
## 7. Docker Cleanup
**Status:** Done (4.1.1)
**Problem:** Docker build context is larger than needed (includes test images, rpi scripts, etc.)
**Solution:** Added `.dockerignore` and fixed volume permissions in Dockerfile
### Files Modified
- `.dockerignore` (created)
- `Dockerfile` (instance dir permissions)
---
## 8. Release Validation Script
**Status:** Planned
**Problem:** Manual release checklist is error-prone. Need automated validation.
**Solution:** Script that runs through testable checklist items
### Features
- Run pytest
- Build and test Docker image
- SSH to Pi and run smoke test (optional, if PI_IP provided)
- Report pass/fail summary
### Files to Create
- `scripts/validate-release.sh`
---
## 9. Smoke Test Docker Support
**Status:** Done
**Problem:** Smoke test expects systemd service, doesn't auto-create admin for Docker.
**Solution:** Make smoke test Docker-aware
### Features
- Skip systemd checks if not on Pi/Linux with systemd
- Auto-detect fresh Docker (no users) and create admin via /setup
- Add `--docker` flag to skip Pi-specific checks
### Implementation
- Added `--docker` flag that sets localhost and skips SSH/systemd checks
- Docker health check verifies container responds with HTTP 200/302
- Header shows "Docker Smoke Test" in Docker mode
### Files Modified
- `rpi/smoke-test.sh`
---
## Notes
- Keep 4.1.2 focused - 9 features (6 done)
- Don't break DCT compatibility (4.1.1 RS format is stable)
- Test on Pi before release