Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/RubenDarioGuerreroNeira/Ecosistema-IA-Colombia/llms.txt

Use this file to discover all available pages before exploring further.

Salud IA Bot stores all public health data in a local SQLite database (data/salud-ia-bot.db) rather than parsing XML files at runtime. Seeding is the one-time (or periodic) process that reads the raw XML files from the data/ directory, maps them to TypeORM entities, and writes the results into SQLite. Once the database is populated, the application never touches the XML files again — making production start-up fast and memory-efficient.

​
Seeding Workflow Overview

The seeding pipeline works as follows:
  1. XML source files are placed in the data/ directory.
  2. TypeScript seed scripts in scripts/ use fast-xml-parser or xml2js to parse each file.
  3. Parsed rows are mapped to TypeORM entity classes and bulk-inserted into data/salud-ia-bot.db in chunks of 100 records.
  4. The resulting SQLite file is the only artefact the running application needs.
# One-time seeding order (first-time setup)
npm run import:data       # All main datasets
npm run seed:antioquia    # Antioquia health providers
npm run seed:vaccination  # PAI vaccination coverage

​
Available Seed Commands

Three npm scripts cover the full dataset:

import:data

Imports all main datasets in parallel: SIVIGILA public health events, mental health (CIE-10), sexual health Q&A, and regional providers for Boyacá, Cali, Antioquia, and Yopal. Runs scripts/import-data.ts via ts-node.

seed:antioquia

Seeds the Antioquia health provider dataset from Prestadores_de_Salud_Departamento_de_Antioquia.xml into the antioquia_provider table. Runs scripts/seed-antioquia.ts via ts-node.

seed:vaccination

Seeds PAI vaccination coverage from three XML files: national departmental coverage, Valle del Cauca coverage, and children’s vaccination data. Runs scripts/seed-vaccination.ts via ts-node.

​
Dataset Reference

The table below maps each seed command to its XML source files and the database tables it creates or populates:
ScriptXML Source FilesTables Created/Populated
import:dataEventos_de_Interés_en_Salud_Pública_20260514.xmlhealth_event
import:dataSalud_Mental.xmlmental_health
import:dataSalud_sexual_-_preguntas.xmlsexual_health
import:dataCentros_de_salud_Yopal._.xmlyopal_provider
import:dataSERVICIOS_OFERTADOS_RED_DE_SALUD_DEL_CENTRO_ESE_POR_SEDE_CALI.xmlcali_provider
import:dataservicios_salud_boyaca.xmlboyaca_provider
import:dataPrestadores_de_Salud_Departamento_de_Antioquia.xmlantioquia_provider
seed:antioquiaPrestadores_de_Salud_Departamento_de_Antioquia.xmlantioquia_provider
seed:vaccinationCoberturas_administrativas_de_vacunación_por_departamento_20260528.xmlvaccination
seed:vaccinationCobertura_de_Vacunación_PAI_en_el_Valle_del_Cauca.xmlvaccination
seed:vaccinationDATOS_DE_VACUNACIÓN_EN_NIÑOS_Y_NIÑAS.xmlvaccination

​
XML File Locations

All XML source files must be placed inside the data/ directory at the project root before running any seed script. The scripts resolve file paths relative to process.cwd(): 
data/
├── salud-ia-bot.db                                              ← generated output
├── Eventos_de_Interés_en_Salud_Pública_20260514.xml
├── Salud_Mental.xml
├── Salud_sexual_-_preguntas.xml
├── Prestadores_de_Salud_Departamento_de_Antioquia.xml
├── Centros_de_salud_Yopal._.xml
├── SERVICIOS_OFERTADOS_RED_DE_SALUD_DEL_CENTRO_ESE_POR_SEDE_CALI.xml
├── servicios_salud_boyaca.xml
├── Coberturas_administrativas_de_vacunación_por_departamento_20260528.xml
├── Cobertura_de_Vacunación_PAI_en_el_Valle_del_Cauca.xml
└── DATOS_DE_VACUNACIÓN_EN_NIÑOS_Y_NIÑAS.xml
If a file is missing when a script runs, the importer logs ⏭️ Archivo no encontrado, saltando. and skips that dataset without failing — allowing partial imports when only some XML files are available.
1

Place all XML files in data/

Copy every XML file listed in the table above into the data/ directory. Ensure filenames are byte-for-byte identical to those listed (accented characters included).
2

Run import:data

Import all main datasets:
npm run import:data
This script opens a TypeORM DataSource with synchronize: true to create the schema on first run, then imports Boyacá, Antioquia, Cali, Yopal, vaccination, mental health, sexual health, and SIVIGILA health events in parallel. On completion it prints a summary:
📊 RESUMEN DE IMPORTACIÓN
Boyacá:          N registros
Antioquia:       N registros
Cali:            N registros
Yopal:           N registros
Vacunación:      N registros
Salud Mental:    N registros
Salud Sexual:    N registros
Eventos Salud:   N registros
TOTAL:           N registros
⏱️  Tiempo:       X.XX segundos
3

Run seed:antioquia

Seed the Antioquia provider dataset:
npm run seed:antioquia
This script uses xml2js with explicitArray: false to parse Prestadores_de_Salud_Departamento_de_Antioquia.xml and upserts records into the antioquia_provider table via repo.save(entities, { chunk: 100 }).
4

Run seed:vaccination

Seed PAI vaccination coverage data:
npm run seed:vaccination
Processes all three vaccination XML files. Each file uses a dedicated row mapper to normalise field names to the shared Vaccination entity (fields: coddepto, departamento, a_o, biol_gico, cobertura_de_vacunaci_n, and up to six indicator columns for children’s data).
5

Verify by starting the development server

Start the app and test a query against the bot:
npm run start:dev
Send a Telegram message such as "¿Cuántos casos de dengue hay en Cali?" and confirm that SIVIGILA statistics are returned. If the bot responds with live data, the seeding was successful.

​
How the Seed Scripts Work

The seed scripts are standalone TypeScript files executed by ts-node. Here is the core pattern used by all importers (from scripts/import-data.ts): 
const parser = new XMLParser();
const data = parser.parse(xmlContent);
const rows = ensureArray(data?.response?.rows?.row);

const entities = rows.map((row) => {
  const entity = new HealthEvent();
  entity.departamento = row.departamento || null;
  entity.nombre_del_evento = row.nombre_del_evento || null;
  entity.total_de_eventos = Number(row.total_de_eventos) || 0;
  // ... additional field mappings
  return entity;
});

await repo.save(entities, { chunk: 100 });
The ensureArray helper handles both single-record XML responses (where row is an object) and multi-record responses (where row is an array), ensuring consistent mapping in both cases.

​
Production Notes

After seeding locally, transfer data/salud-ia-bot.db to your production environment. The application does not re-parse XML at startup — services query SQLite directly using TypeORM repositories and better-sqlite3. 
# Example: copy the database to a remote server via scp
scp data/salud-ia-bot.db user@your-server:/app/data/salud-ia-bot.db
For Render or Railway, copy the file to the mounted persistent disk before the first deploy (see Deployment).
Seed scripts use ts-node to run TypeScript directly. If you see module resolution errors such as Cannot find module '@bot/...' or Cannot find module '@shared/...', ensure tsconfig.json path aliases (@bot/* -> src/bot/*, @shared/* -> src/shared/*) are correctly configured and that tsconfig-paths is available. Install it with npm install --save-dev tsconfig-paths and invoke the script as ts-node -r tsconfig-paths/register scripts/import-data.ts if aliases are needed at script level.

Build docs developers (and LLMs) love

Get started for freeTalk to us