# Implementation Plan: LLM Table Translation Service

**Branch**: `028-llm-datasource-supeset` | **Date**: 2026-05-08 | **Spec**: [spec.md](./spec.md)
**Input**: Feature specification from `/specs/028-llm-datasource-supeset/spec.md` (updated 2026-05-14 with multi-language & correction optimization)

## Summary

Implement an LLM-powered table translation service as a new backend plugin (`TranslationPlugin`) with companion API routes, ORM models, Pydantic schemas, and Svelte frontend components. The service reads rows from a Superset datasource (translation column + context columns), auto-detects source language per row via LLM, sends batches to a configured LLM provider for ALL target languages simultaneously (multi-target in single structured JSON response), with optional per-batch filtered multilingual terminology dictionary context (language-pair-aware). Generates safe PostgreSQL INSERT/UPSERT SQL, submits to Superset via `/api/v1/sqllab/execute/` with status polling. Supporting capabilities: multilingual terminology dictionaries (source_language + target_language per entry), configurable preview (1-100 rows), inline correction from any run result with dictionary submission, bulk find-and-replace, scheduled execution via APScheduler (new-key-only with baseline_expired fallback), structured event logging with per-language MetricSnapshot persistence, RBAC-gated access control, and 90-day retention with metric continuity.

**Total implemented**: 72 files, 29 128 LOC across backend (Python/FastAPI) and frontend (Svelte 5/SvelteKit).  
**Actual complexity distribution**: 2× C5, 6× C4, 12× C3, 7× C2, 4× C1.  
**Note**: Архитектура — service-based, не plugin-based. `TranslatePlugin` — только регистрационный скелет (C2). Бизнес-логика через независимые service-классы.

## Technical Context

**Language/Version**: Python 3.13+ (backend), JavaScript/TypeScript (frontend Svelte 5)  
**Primary Dependencies**: FastAPI 0.115+, SQLAlchemy 2.0+, APScheduler 3.x, Pydantic v2 (backend); SvelteKit 2.x, Svelte 5.43+, Vite 7.x, Tailwind CSS 3.x (frontend)  
**Storage**: PostgreSQL 16 (shared with existing ss-tools schema)  
**Testing**: pytest 8.x + pytest-asyncio (backend); vitest 4.x + @testing-library/svelte 5.x (frontend)  
**Target Platform**: Linux server (Docker Compose: backend + frontend + PostgreSQL)  
**Project Type**: web application — FastAPI REST + WebSocket backend, SvelteKit SPA frontend  
**Performance Goals**: Preview of 10 rows within 30s (LLM-dependent); INSERT generation <5s for 1000 rows; schedule trigger precision ±60s; event recording <10s after occurrence  
**Constraints**: RBAC enforcement via access-control matrix; dictionary per-batch filtering (case-insensitive, word-boundary-aware); 90-day detailed data retention with MetricSnapshot persistence; plugin lifecycle compatible with `PluginBase`; dialect-aware SQL generation (PostgreSQL/Greenplum, ClickHouse supported for MVP); snapshot isolation for in-progress runs; structured JSON LLM output required  
**Scale/Scope**: Tens of thousands of rows per run; dictionaries of any size; 50 concurrent users; 5–10 active translation jobs per deployment

## Constitution Check

*GATE: Must pass before Phase 0 research. Re-check after Phase 1 design.*

| Principle | Status | Evidence |
|-----------|--------|----------|
| **I. Plugin Architecture** | ✅ PASS | Feature implemented as a new plugin inheriting from `PluginBase` (`backend/src/plugins/translate/`), consistent with existing `llm_analysis`, `git`, `storage` plugin pattern. |
| **II. API-First Design** | ✅ PASS | All operations exposed via REST endpoints under `/api/translate/` (jobs, dictionaries, runs, schedules, history) and WebSocket for run progress. Follows existing FastAPI route conventions. |
| **III. Test-First** | ✅ PASS | Acceptance criteria per user story defined in spec; test strategy in research.md covers unit (pytest), integration (API + DB), and component (vitest + Svelte Testing Library) layers. |
| **IV. RBAC & Security** | ✅ PASS | Granular permissions (`translate.job.*`, `translate.dictionary.*`, `translate.schedule.manage`, `translate.history.view`) via existing RBAC model. API keys encrypted via `EncryptionManager`. PII masking for LLM-facing context per existing patterns. |
| **V. Observability & Retention** | ✅ PASS | Structured Translation Events (FR-046), per-job metrics (FR-047), notification integration (FR-048), 90-day retention with pruning (FR-049). C4+ flows instrumented with `belief_scope`/`reason`/`reflect`/`explore` markers. |
| **Semantic Protocol Compliance** | ✅ PASS | All planned modules assigned `@COMPLEXITY` 1–5 with appropriate tag density per GRACE complexity scale (C1: anchor only, C2: +BRIEF, C3: +RELATION, C4: +PRE/POST/SIDE_EFFECT, C5: +DATA_CONTRACT/INVARIANT/RATIONALE/REJECTED). `@RATIONALE`/`@REJECTED` reserved for C5 contracts only. Canonical `@RELATION PREDICATE -> TARGET_ID` syntax. Anchor format: `#region`/`#endregion` for Python, `<!-- #region -->`/`<!-- #endregion -->` for Svelte markup. |
| **Fractal Limit (INV_7)** | ✅ PASS | Planned modules kept under 400 lines each. No planned contract exceeds 150 lines or CC>10. Decomposition strategy documented in contracts/modules.md. |

**Gate Result**: ✅ ALL PASS — no blocking constitutional or semantic conflicts.

## Project Structure

### Documentation (this feature)

```text
specs/028-llm-datasource-supeset/
├── plan.md              # This file
├── research.md          # Phase 0 output
├── data-model.md        # Phase 1 output
├── quickstart.md        # Phase 1 output
├── contracts/           # Phase 1 output
│   └── modules.md       # Semantic contract design
├── spec.md              # Feature specification
├── ux_reference.md      # Interaction reference
└── checklists/
    └── requirements.md  # Quality validation
```

### Source Code (repository root — ACTUAL as implemented)

```text
backend/
├── src/
│   ├── api/routes/
│   │   └── translate/             # [ACTUAL] Пакет, а не один файл
│   │       ├── __init__.py        # Инициализация, re-export router
│   │       ├── _router.py         # APIRouter(prefix="/api/translate")
│   │       ├── _job_routes.py     # Job CRUD (249 строк)
│   │       ├── _run_routes.py     # Run execution, cancel, inline edit (554 строки)
│   │       ├── _run_list_routes.py # Run list, detail, CSV download (225 строк)
│   │       ├── _preview_routes.py  # Preview CRUD (194 строки)
│   │       ├── _dictionary_routes.py # Dictionary CRUD + import (391 строка)
│   │       ├── _correction_routes.py # Correction submission (100 строк)
│   │       ├── _schedule_routes.py   # Schedule CRUD (239 строк)
│   │       ├── _metrics_routes.py    # Metrics endpoints (65 строк)
│   │       └── _helpers.py           # Response helpers (71 строка)
│   ├── core/
│   │   ├── scheduler.py           # [REUSE] APScheduler
│   │   └── superset_client/       # [REUSE] Superset API client
│   ├── models/
│   │   └── translate.py           # [NEW] 15 ORM models (397 строк)
│   ├── plugins/
│   │   └── translate/             # [NEW] TranslationPlugin
│   │       ├── __init__.py        # Package (2 строки)
│   │       ├── plugin.py          # Plugin skeleton C2 (61 строка, регистрация)
│   │       ├── orchestrator.py    # C5 Run lifecycle (1 137 строк) ⭐
│   │       ├── executor.py        # C4 Batch LLM (1 974 строки) ⭐
│   │       ├── service.py         # C4 TranslateJobService + InlineCorrection + BulkReplace (1 052 строки) ⭐
│   │       ├── preview.py         # C4 Preview engine (1 303 строки) ⭐
│   │       ├── dictionary.py      # C4 Dictionary CRUD + filtering (1 007 строк) ⭐
│   │       ├── scheduler.py       # C4 Schedule management (436 строк)
│   │       ├── sql_generator.py   # C3 Dialect-aware SQL (378 строк)
│   │       ├── superset_executor.py # C3 Superset SQL Lab API (420 строк) [NEW в реализации]
│   │       ├── prompt_builder.py  # C3 Context-aware prompt (153 строки) [NEW]
│   │       ├── events.py          # C5 Event log + pruning (272 строки)
│   │       ├── metrics.py         # C3 Metrics aggregation (192 строки)
│   │       ├── _token_budget.py   # C2 Token estimation (341 строка) [NEW]
│   │       └── _utils.py          # C1 Normalization (33 строки) [NEW]
│   │       └── __tests__/         # Plugin-level unit tests (8 файлов, 5 645 строк)
│   ├── schemas/
│   │   └── translate.py           # [NEW] 35+ Pydantic schemas (688 строк)
│   └── services/
│       ├── llm_provider.py        # [REUSE] LLM provider management
│       └── llm_prompt_templates.py # [REUSE] Prompt rendering
└── tests/                         # Integration tests (7 файлов, 2 458 строк)

frontend/
├── src/
│   ├── routes/
│   │   └── translate/             # [NEW] SvelteKit routes
│   │       ├── +page.svelte       # Job list (279 строк)
│   │       ├── [id]/
│   │       │   └── +page.svelte   # Job config + preview + run (1 363 строки) ⭐
│   │       ├── dictionaries/
│   │       │   ├── +page.svelte   # Dictionary list (270 строк)
│   │       │   └── [id]/
│   │       │       └── +page.svelte # Dictionary editor (821 строка)
│   │       └── history/
│   │           └── +page.svelte   # Run history (523 строки)
│   └── lib/
│       ├── components/
│       │   └── translate/         # [NEW] Translation components (13 файлов)
│       │       ├── TranslationPreview.svelte       # C4 (561 строка)
│       │       ├── TranslationRunResult.svelte     # C4 (464 строки)
│       │       ├── BulkReplaceModal.svelte         # C3 (396 строк) [NEW]
│       │       ├── TranslationRunProgress.svelte   # C4 (310 строк)
│       │       ├── ScheduleConfig.svelte           # C3 (286 строк)
│       │       ├── BulkCorrectionSidebar.svelte    # C3 (285 строк)
│       │       ├── TermCorrectionPopup.svelte      # C3 (342 строки)
│       │       ├── CorrectionCell.svelte           # C3 (326 строк) [NEW]
│       │       ├── TranslationRunGlobalIndicator.svelte # C2 (198 строк) [NEW]
│       │       ├── TranslationMetricsDashboard.svelte   # C2 (184 строки) [NEW]
│       │       ├── DictionaryEditor.svelte         # C3 (на странице)
│       │       └── DictionaryList.svelte           # C3 (на странице)
│       ├── stores/
│       │   ├── translate.js       # [NEW] rune store (196 строк) [NEW в реализации]
│       │   └── translationRun.js   # [NEW] run progress store
│       └── api/
│           └── translate.js       # [NEW] API client (664 строки)
└── e2e/
    └── tests/
        └── translation.e2e.js     # [NEW] E2E тесты (142 строки)
```

## Semantic Contract Guidance

- All planned contracts follow GRACE-Poly v2.6 protocol with `#region`/`#endregion` anchors (or `[DEF:...]`/`[/DEF:...]` for backward compatibility in docs).
- Python backend uses `# #region`/`# #endregion` comment-anchor syntax.
- Svelte components use `<!-- #region -->`/`<!-- #endregion -->` in markup and `// #region`/`// #endregion` in script blocks.
- Complexity assignments follow strict tag density per GRACE complexity scale (detailed in `contracts/modules.md`).
- C4+ Python modules instrumented with `belief_scope()` + `reason()`/`reflect()`/`explore()` markers.
- C5 contracts carry `@RATIONALE` and `@REJECTED` decision memory (C1-C4 MUST NOT have these tags).
- `@RELATION` targets reference existing contracts where applicable (e.g., `[LLMProviderService:Module]`, `[SchedulerService:Class]`, `[SupersetClient:Module]`, `[NotificationService:Module]`).

## Complexity Tracking

No constitutional violations detected. **UPDATED 2026-05-17**: Complexity скорректирована по факту реализации.

| Contract | Plan (C?) | Actual (C?) | Файл | Примечание |
|----------|:---------:|:-----------:|------|------------|
| `TranslatePlugin:Module` | C3 | **C2** | `plugin.py` | Только регистрация; execute() — NotImplementedError |
| `TranslationOrchestrator` | C5 | **C5** ✅ | `orchestrator.py` | Stateful lifecycle, PRE/POST, multi-step coordination |
| `TranslationEventLog` | C5 | **C5** ✅ | `events.py` | Immutable event log + retention + MetricSnapshot |
| `TranslationExecutor` | C4 | **C4** ✅ | `executor.py` | Batch LLM, caching, parsing, retry |
| `TranslationPreview` | C4 | **C4** ✅ | `preview.py` | Quality gate, approve/edit/reject |
| `DictionaryManager` | C4 | **C4** ✅ | `dictionary.py` | CRUD + import + per-batch filtering |
| `TranslationScheduler` | C4 | **C4** ✅ | `scheduler.py` | APScheduler integration |
| `TranslateJobService` | — | **C4** [NEW] | `service.py` | Job CRUD + InlineCorrection + BulkReplace |
| `TranslateRoutes` | C3 | **C4** 🔴 | `__init__.py` | Пакет с 11 подмодулями |
| `SQLGenerator` | C4 | **C3** 🟢 | `sql_generator.py` | Pure function, нет I/O |
| `SupersetSqlLabExecutor` | — | **C3** [NEW] | `superset_executor.py` | Superset SQL Lab API клиент |
| `ContextAwarePromptBuilder` | — | **C3** [NEW] | `prompt_builder.py` | Jaccard similarity, priority |
| `InlineCorrectionService` | C4 | **C3** 🟢 | `service.py` | Context capture, origin tracking |
| `BulkFindReplaceService` | C4 | **C3** 🟢 | `service.py` | Regex, preview, atomic apply |
| `TranslationMetrics` | C3 | **C3** ✅ | `metrics.py` | Metrics aggregation |
| `TranslationModels` | C2 | **C2** ✅ | `models/translate.py` | 15 ORM models |
| `TranslateSchemas` | C2 | **C2** ✅ | `schemas/translate.py` | 35+ Pydantic schemas |
| `TranslateApiClient` | C2 | **C2** ✅ | `api/translate.js` | API client |
| `translateStore` | C3 | **C3** ✅ | `translationRun.js` | Svelte 5 rune store |
| `TranslationLanguage` | C1 | **C1** ✅ | `models/translate.py` | DTO |
| `TranslationRunLanguageStats` | C1 | **C1** ✅ | `models/translate.py` | DTO |
| `CorrectionCell` (Svelte) | — | **C3** [NEW] | `CorrectionCell.svelte` | Inline edit + submit |
| `BulkReplaceModal` (Svelte) | — | **C3** [NEW] | `BulkReplaceModal.svelte` | Find & replace modal |
| Other Svelte components | C3 | **C3** ✅ | — | Все 10 компонентов |