Files
odr_console/app.md
T
2026-05-09 18:11:22 +02:00

70 lines
3.3 KiB
Markdown

---
name: odr_console
lang: cpp
domain: tools
description: "Lanzador GUI de funciones del registry para recolectar datos online. Panel de busqueda FTS5, jobs queue async (workers concurrentes), pipeline builder DAG, browser DuckDB, assertions/proposals. Aplica bucle reactivo de 5 pasos sobre operations.db propia."
tags: [imgui, data-collection, scraping, duckdb, jobs, cdp]
uses_functions: []
uses_types: []
framework: "imgui"
entry_point: "main.cpp"
dir_path: "projects/online_data_recopilation/apps/odr_console"
repo_url: ""
---
## Notas
App C++ ImGui que orquesta:
1. **Launcher panel** — busqueda FTS5 sobre `registry.db`. Lanza cualquier funcion/pipeline con form auto-generado desde `params_schema`.
2. **Pipeline builder** — DAG visual con `imgui_node_editor`. Compone collectors validando composabilidad (`returns``uses_types`). Persiste en `operations.relations` con status `designed`.
3. **Jobs queue** — Pool de N workers (default 4). Cada job = subprocess Python collector. Live progress por panel (`PROGRESS:` en stderr). Reusa `jobs_pool_cpp_core` (extraido de osint_graph en issue 0065).
4. **Datasets browser** — Panel DuckDB embebido. Query editor + tabla preview + ImPlot para charts. Lee parquet de `vaults/odr_data/`.
5. **Entities + assertions** — Vista de `operations.entities` por dataset. Editor SQL para assertions. Boton "Eval --react" lanza paso 4 del bucle.
6. **Proposals inbox** — Lista pending de `registry.proposals` originadas por assertions fallidas.
### Estructura
```
odr_console/
main.cpp # fn::run_app + render() + paneles
views_launcher.cpp # Panel 1
views_pipelines.cpp # Panel 2
views_jobs.cpp # Panel 3
views_datasets.cpp # Panel 4
views_assertions.cpp # Panel 5
views_proposals.cpp # Panel 6
data_registry.cpp # Lee registry.db (FTS5, funcs, types)
data_operations.cpp # CRUD operations.db (relations/executions/entities/assertions)
data_duck.cpp # DuckDB connector + ingest
collectors/ # Mismo schema que enrichers de graph_explorer
api_hn_top/ # MVP: HackerNews top stories via API
manifest.yaml
run.py
migrations/ # operations.db migrations (esquema 5-pasos)
CMakeLists.txt
```
### Local files (regla cpp_apps §7)
- `local_files/odr_console.ini` — settings persistidas
- `local_files/imgui.ini` — layout
- `local_files/odr.duckdb` — DuckDB embebido (datos crudos pequeños)
- `local_files/cache/<sha[0:2]>/<sha>.{html,json,parquet}` — cache addressable
- `operations.db` queda en el dir del exe (consultable por `fn ops`)
### Decisiones tomadas
| Tema | Decision |
|---|---|
| Workers default | 4 (mas que graph_explorer porque crawls esperan red) |
| operations.db | Una unica por la app |
| DuckDB | Embebido (linkar libduckdb), no subprocess |
| Collectors lang inicial | Python (espejo graph_explorer enrichers) |
| Browser | CDP via `cdp-cli` Go (issue 0038) cuando aplica |
### Decisiones pendientes
- Refactor jobs system: en paralelo a MVP. Ver issue 0065.
- Schema operations.db: requiere migracion 001 con relations/executions/entities/types_snapshot/assertions/assertion_results (ver `fn_operations/migrations/`).