e3c8979e8d
- cmd/fn/doctor.go - cmd/fn/main.go - cpp/apps/primitives_gallery/playground/tables/CMakeLists.txt - cpp/apps/primitives_gallery/playground/tables/data_table.cpp - cpp/apps/primitives_gallery/playground/tables/data_table_logic.cpp - cpp/apps/primitives_gallery/playground/tables/data_table_logic.h - cpp/apps/primitives_gallery/playground/tables/self_test.cpp - cpp/apps/primitives_gallery/playground/tables/tql.cpp - cpp/apps/primitives_gallery/playground/tables/viz.cpp - cpp/apps/primitives_gallery/playground/tables/viz.h - ... Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
85 lines
3.0 KiB
Markdown
85 lines
3.0 KiB
Markdown
---
|
|
name: vault_index_write
|
|
kind: function
|
|
lang: go
|
|
domain: infra
|
|
version: "1.0.0"
|
|
purity: impure
|
|
signature: "func VaultIndexWrite(db *sql.DB, files []VaultFile, prune bool) (WriteReport, error)"
|
|
description: "Upserta un slice de VaultFile en vault_index.db (tabla files + FTS5 files_fts) dentro de una sola transaccion. Cuenta Inserted/Updated/FTS. Con prune=true elimina filas no presentes en el slice."
|
|
tags: [vault, sqlite, index, write, upsert, fts, infra]
|
|
uses_functions: []
|
|
uses_types: ["vault_file_go_infra"]
|
|
returns: []
|
|
returns_optional: false
|
|
error_type: "error_go_core"
|
|
imports: [database/sql, fmt, strings, time]
|
|
params:
|
|
- name: db
|
|
desc: "*sql.DB abierto sobre vault_index.db (tipicamente retornado por VaultIndexOpen)"
|
|
- name: files
|
|
desc: "slice de VaultFile a insertar/actualizar; puede ser vacio"
|
|
- name: prune
|
|
desc: "si true, elimina de 'files' todas las filas cuyo rel_path no este en el slice (sincronizacion destructiva)"
|
|
output: "WriteReport con conteos Inserted/Updated/Pruned/FTS; error si falla la transaccion"
|
|
tested: true
|
|
tests:
|
|
- "N archivos nuevos — Inserted=N"
|
|
- "re-escritura con mtime distinto — Updated=N"
|
|
- "prune elimina filas ausentes"
|
|
- "sin prune, filas previas persisten"
|
|
- "FTS5 MATCH funciona tras escritura"
|
|
test_file_path: "functions/infra/vault_index_write_test.go"
|
|
file_path: "functions/infra/vault_index_write.go"
|
|
---
|
|
|
|
## Ejemplo
|
|
|
|
```go
|
|
db, _ := VaultIndexOpen("/data/vaults/turismo")
|
|
defer db.Close()
|
|
|
|
files, _ := VaultInventoryScan("/data/vaults/turismo", "turismo_v1", "turismo")
|
|
report, err := VaultIndexWrite(db, files, true)
|
|
if err != nil {
|
|
log.Fatal(err)
|
|
}
|
|
fmt.Printf("inserted=%d updated=%d pruned=%d fts=%d\n",
|
|
report.Inserted, report.Updated, report.Pruned, report.FTS)
|
|
```
|
|
|
|
## Notas
|
|
|
|
### WriteReport
|
|
Struct local al paquete infra:
|
|
```go
|
|
type WriteReport struct {
|
|
Inserted int
|
|
Updated int
|
|
Pruned int
|
|
FTS int
|
|
}
|
|
```
|
|
|
|
### Estrategia de conteo Inserted vs Updated
|
|
Se carga el conjunto de rel_paths existentes en un map antes del loop. Un upsert
|
|
se clasifica como Inserted si el rel_path no estaba en el map, Updated si estaba.
|
|
Esto evita N+1 SELECTs y es correcto porque la transaccion serializa los cambios.
|
|
|
|
### FTS5
|
|
`files_fts` usa `content=''` (tabla de contenido externo vacio). Para cada archivo
|
|
se borra la fila FTS existente y se reinserta con `content_text=''`. Los profilers
|
|
posteriores (csv_profiles, knowledge_docs) son responsables de actualizar
|
|
`content_text` con texto indexable real.
|
|
|
|
### Prune
|
|
Con `prune=true` se construye un IN clause con los rel_paths del slice. La FK con
|
|
`ON DELETE CASCADE` propaga el DELETE a csv_profiles, pdf_extracts y knowledge_docs
|
|
automaticamente. Con slice vacio + prune=true se borra todo (DELETE FROM files).
|
|
|
|
### Escapado SQL
|
|
El IN clause se construye escapando las comillas simples en rel_path (duplicandolas).
|
|
Evita inyeccion en rutas con apostrofos. Para entornos con rutas controladas
|
|
(interior de vaults sin apostrofos) esto es suficiente; para entornos adversariales
|
|
usar parametros binding con VALUES multiples via prepared statement.
|