# Implementation Plan: Short Image IDs

**Branch**: `017-short-id-migration` | **Date**: 2026-05-09 | **Spec**: [spec.md](spec.md)
**Input**: Feature specification from `specs/017-short-id-migration/spec.md`

## Summary

Replace hash-based storage keys and UUID-based URL routing with 8-character base62 short IDs. The short ID becomes the canonical identifier in URLs (`/i/:short_id`), storage keys (`{short_id}` / `{short_id}-thumb`), and all API responses. Hash-based deduplication is preserved unchanged. A Python migration script handles existing images: generates short IDs, copies storage objects to new keys, updates DB records, deletes old keys.

## Technical Context

**Language/Version**: Python 3.12+ (API), TypeScript strict (UI)
**Primary Dependencies**: FastAPI, SQLAlchemy 2.x async, Alembic, aiobotocore/boto3, Angular (latest stable)
**Storage**: PostgreSQL (DB), S3-compatible via boto3 (MinIO local / CDN in prod)
**Testing**: pytest + pytest-asyncio (API unit + integration), Karma/Jasmine (Angular)
**Target Platform**: Linux server (k3s), browser SPA
**Project Type**: Web application (FastAPI API + Angular SPA)
**Performance Goals**: Migration script should process all existing images without timeout; no user-facing performance change
**Constraints**: Migration must be idempotent; no data loss; copy-before-delete for all storage operations
**Scale/Scope**: Personal collection (~hundreds to low thousands of images); collision probability negligible

## Constitution Check

*GATE: Must pass before implementation.*

| Principle | Status | Notes |
|-----------|--------|-------|
| §2.5 DB abstraction — all queries through repos | ✅ PASS | New `get_by_short_id()` added to `ImageRepository`; no raw SQL outside repo |
| §2.6 No speculative abstraction | ✅ PASS | `generate_short_id()` is a concrete utility; no new interfaces |
| §3.1 Routes prefixed `/api/v1/` | ✅ PASS | All routes remain under `/api/v1/images/` |
| §3.1 Adding fields is non-breaking | ✅ PASS | `short_id` is additive; `id` UUID retained |
| §4.2 Images immutable after upload | ✅ PASS | File content is copied, not replaced; the operation changes the storage key, not the bytes |
| §4.3 Deduplication by content hash | ✅ PASS | `hash` column retained; `get_by_hash` unchanged |
| §5.1 Tests alongside every implementation task | ✅ PASS | Each task includes tests |
| §5.2 Integration tests use real PostgreSQL + MinIO | ✅ PASS | Existing integration test infrastructure reused |
| §8 Scope boundaries | ✅ PASS | No multi-user, no public sharing feature, no OR/NOT tag logic |

**No violations. Implementation may proceed.**

## Project Structure

### Documentation (this feature)

```text
specs/017-short-id-migration/
├── plan.md              ← this file
├── research.md          ← short ID generation, migration strategy
├── data-model.md        ← Image schema changes, Alembic migrations
├── contracts/
│   └── image-api.md     ← updated ImageRecord schema, route changes
├── quickstart.md        ← manual test scenarios
└── tasks.md             ← generated by /speckit-tasks
```

### Source Code Changes

```text
api/
├── app/
│   ├── models.py                    # Add Image.short_id column
│   ├── utils.py                     # Add generate_short_id()
│   ├── repositories/
│   │   └── image_repo.py            # Add get_by_short_id(), update create()
│   └── routers/
│       └── images.py                # Path params uuid→str, add short_id to response
├── alembic/versions/
│   ├── 003_add_short_id.py          # ADD COLUMN short_id VARCHAR(8) NULLABLE UNIQUE
│   └── 004_short_id_not_null.py     # SET NOT NULL (run after migration script)
├── scripts/
│   └── migrate_to_short_ids.py      # Backfill existing images
└── tests/
    ├── unit/
    │   ├── test_hashing.py          # Add generate_short_id() tests
    │   ├── test_url_construction.py # Update mock images to include short_id
    │   └── test_short_id.py         # NEW: collision retry, charset validation
    └── integration/
        ├── test_upload.py           # Assert short_id in response
        ├── test_search.py           # Update {id} → {short_id} in route calls
        ├── test_delete.py           # Update route params
        ├── test_serving.py          # Update route params
        └── test_tags.py             # Update route params

ui/src/app/
├── app.routes.ts                    # 'images/:id' → 'i/:id'
├── services/
│   └── image.service.ts             # Add short_id to ImageRecord, update service calls
├── library/
│   └── library.component.ts         # Navigate to ['/i', img.short_id]
├── upload/
│   └── upload.component.ts          # Navigate to ['/i', res.short_id] after upload
└── detail/
    └── detail.component.ts          # (no route change needed; reads :id param same way)
```

**Structure Decision**: Existing web application layout. API changes are concentrated in models, repository, router, and a new migration script. UI changes are confined to routes, image service interface, and two navigation calls.

## Implementation Phases

### Phase 1: Backend — Short ID Infrastructure

1. Add `generate_short_id()` to `api/app/utils.py`
   - Base62 charset: `string.ascii_letters + string.digits`
   - Uses `secrets.choice` for cryptographic randomness
   - Returns 8-character string

2. Add Alembic migration `003_add_short_id.py`
   - `ADD COLUMN short_id VARCHAR(8) NULL`
   - `CREATE UNIQUE INDEX ix_images_short_id ON images (short_id)`

3. Update `api/app/models.py`
   - Add `short_id: Mapped[str | None] = mapped_column(String(8), unique=True, nullable=True, index=True)`

4. Update `api/app/repositories/image_repo.py`
   - Add `get_by_short_id(short_id: str) -> Image | None`
   - Update `create()` to accept and persist `short_id` parameter

5. Update `api/app/routers/images.py`
   - Change all `image_id: uuid.UUID` path params to `short_id: str`
   - Add `_validate_short_id(short_id: str)` helper (8 alphanumeric chars, else 422)
   - Replace `get_by_id` calls with `get_by_short_id`
   - Update `_image_to_dict` to include `"short_id": image.short_id` in response
   - Update upload handler: generate `short_id` with collision retry, use as storage key

### Phase 2: Migration Script

`api/scripts/migrate_to_short_ids.py`:

```
for each image where short_id IS NULL:
    generate short_id (retry on DB collision)
    copy {hash} → {short_id} in storage
    if thumbnail_key IS NOT NULL:
        copy {hash}-thumb → {short_id}-thumb in storage
    verify new objects exist (head_object)
    UPDATE images SET short_id={sid}, storage_key={sid}, thumbnail_key={sid}-thumb WHERE id={id}
    delete {hash} from storage
    if thumbnail_key was not null:
        delete {hash}-thumb from storage
    log: "migrated {id} → {short_id}"

print summary: N migrated, M skipped (already had short_id)
```

After script runs with 0 remaining `NULL` short_ids, apply migration `004_short_id_not_null.py`.

### Phase 3: Frontend

1. `app.routes.ts`: `path: 'images/:id'` → `path: 'i/:id'`
2. `image.service.ts`: add `short_id: string` to `ImageRecord`
3. `library.component.ts`: `router.navigate(['/images', img.id])` → `router.navigate(['/i', img.short_id])`
4. `upload.component.ts`: `router.navigate(['/images', res.id])` → `router.navigate(['/i', res.short_id])`

### Phase 4: Polish

- Update all existing API integration tests to use `short_id` in route paths
- Run `ng lint` and `ruff check` across modified files
- Verify `ng build --configuration production` succeeds
- Run full test suites: `make test-unit && make test-integration`

## Key Implementation Notes

### Collision Retry Pattern (upload)

```python
MAX_RETRIES = 10
for attempt in range(MAX_RETRIES):
    short_id = generate_short_id()
    try:
        image = await image_repo.create(..., short_id=short_id)
        break
    except IntegrityError:  # short_id collision
        await db.rollback()
        if attempt == MAX_RETRIES - 1:
            raise RuntimeError("Could not generate unique short_id")
```

### Route Validation

```python
import re
_SHORT_ID_RE = re.compile(r'^[a-zA-Z0-9]{8}$')

def _validate_short_id(short_id: str) -> None:
    if not _SHORT_ID_RE.match(short_id):
        raise HTTPException(422, detail={"detail": "Invalid image ID", "code": "invalid_short_id"})
```

### `_image_to_dict` Update

Add `"short_id": image.short_id` to the returned dict. The `file_url` and `thumbnail_url` generation already uses `image.storage_key` which will now equal `image.short_id` — no formula change needed.

### Migration Script Entry Point

```bash
cd api && python -m scripts.migrate_to_short_ids
```

Reads DB URL and storage config from environment variables (same as the application).