Merge pull request #163 from Lemoncode/feature/integration-cuenca-segura

feature/integration-cuenca-segura

Author: brauliodiez

functions/src/functions/scraping-functions.ts

@@ -34,6 +34,8 @@ export async function scrapingsFunction(
 
     const responseCuencaJucar = await embalsesRepository.actualizarCuencaJucar();
 
+    const responseCuencaSegura = await embalsesRepository.actualizarCuencaSegura();
+
     if (responseCuencaMediterranea) {
       context.log(
         "scrapings-function: Se han actualizado los embalses de la cuenca Mediterránea",
@@ -82,6 +84,14 @@ export async function scrapingsFunction(
       );
     }
 
+    if (responseCuencaSegura) {
+      context.log(`Se han actualizado los embalses de la cuenca Segura`);
+    } else {
+      context.log(
+        "No se han podido actualizar los embalses de la cuenca Segura"
+      );
+    }
+
   } catch (error) {
     context.error("scrapings-function: ERROR", error);
     throw error;

integrations/scraping-cuenca-segura/package.json

@@ -1,15 +1,27 @@
 {
-  "name": "scraping-cuenca-segura",
+"name": "scraping-cuenca-segura",
   "version": "1.0.0",
   "private": true,
   "type": "module",
   "exports": {
-    ".": "./src/index.ts"
+    ".": "./dist/index.js"
   },
+  "main": "./dist/index.js",
+  "types": "./dist/index.d.ts",
   "scripts": {
-    "start": "tsx --watch src/console-runner.ts"
+    "start": "tsx --watch ./src/console-runner.ts",
+    "build": "run-p clean type-check build:scraping-cuenca-segura",
+    "build:scraping-cuenca-segura": "tsc",
+    "clean": "rimraf dist",
+    "type-check": "tsc --noEmit --preserveWatchOutput"
   },
   "dependencies": {
+    "axios": "^1.11.0",
+    "cheerio": "^1.1.2",
     "db-model": "^1.0.0"
+  },
+  "devDependencies": {
+    "ts-node": "^10.9.2",
+    "typescript": "^5.9.2"
   }
 }

integrations/scraping-cuenca-segura/src/api/index.ts

@@ -1,3 +1,3 @@
 // Barrel file for API exports
-export * from './cuenca.api';
-export * from './cuenca.model';
+export * from './cuenca.api.js';
+export * from './cuenca.model.js';

integrations/scraping-cuenca-segura/src/console-runner.ts

@@ -1,8 +1,5 @@
-import { scrapeCuencaSegura } from './integration';
-import { mapToEmbalseUpdateSAIH } from './scraper';
+import { scrapeCuencaSegura } from './integration.js';
 
-const URL = 'https://chsegura.es/es/cuenca/redes-de-control/estadisticas-hidrologicas/estado-de-embalses/';
 console.log('Estado de la Cuenca Segura:');
-const scrapedCuencaSegura = await scrapeCuencaSegura(URL);
-const result = mapToEmbalseUpdateSAIH(scrapedCuencaSegura)
-console.log(result);
+const scrapedCuencaSegura = await scrapeCuencaSegura();
+console.log(scrapedCuencaSegura);

integrations/scraping-cuenca-segura/src/index.ts

@@ -1 +1 @@
-export * from "./integration";
+export * from "./integration.js";

integrations/scraping-cuenca-segura/src/integration.ts

@@ -1,17 +1,20 @@
 import * as cheerio from 'cheerio';
-import { getCuencaPageHTMLContent, EmbalsesSegura } from '@/api';
-import { extractReservoirsFromSeguraPage } from '@/scraper';
+import { getCuencaPageHTMLContent } from './api/index.js';
+import { extractReservoirsFromSeguraPage } from './scraper/index.js';
+import { mapToEmbalseUpdateSAIH } from "./scraper/mapper.js";
+import { EmbalseUpdateSAIHEntity } from 'db-model';
+
+const URL = 'https://chsegura.es/es/cuenca/redes-de-control/estadisticas-hidrologicas/estado-de-embalses/';
 
 /**
  * Scrapes Segura reservoir data and returns it as an array.
  * @param url - The URL to scrape the data from
  */
-export async function scrapeCuencaSegura(
-  url: string
-): Promise<EmbalsesSegura[]> {
-  const html = await getCuencaPageHTMLContent(url);
+export async function scrapeCuencaSegura(): Promise<EmbalseUpdateSAIHEntity[]> {
+  const html = await getCuencaPageHTMLContent(URL);
   const $: cheerio.CheerioAPI = cheerio.load(html);
 
   // Extract and map reservoir data
-  return extractReservoirsFromSeguraPage($);
+  const embalses = extractReservoirsFromSeguraPage($);
+  return mapToEmbalseUpdateSAIH(embalses);
 }

integrations/scraping-cuenca-segura/src/scraper/business.ts

@@ -1,27 +1,27 @@
 import { CheerioAPI } from 'cheerio';
 import type { Element } from 'domhandler';
-import { EmbalsesSegura } from '@/api';
-import { mapEmbalsesToEntities } from '@/scraper'
+import { EmbalsesSegura } from '../api/index.js';
+import { mapEmbalsesToEntities } from './mapper.js'
 
 // Function to extract capacity data from main table
 function getReservoirCapacities($: CheerioAPI): Record<string, { capacity: number; percentage: number }> {
   const capacityMap: Record<string, { capacity: number; percentage: number }> = {};
-  
+
   $('#n0 tbody tr').each((_, row) => {
     const $row = $(row);
     const cols = $row.find('td');
     if (cols.length !== 4) return;
 
     const embalse = $(cols[0]).text().trim();
-    if (!embalse || 
-        embalse.toLowerCase().includes('total') || 
-        embalse.toLowerCase().includes('resto')) {
+    if (!embalse ||
+      embalse.toLowerCase().includes('total') ||
+      embalse.toLowerCase().includes('resto')) {
       return;
     }
 
     const capacidadTotalHm3 = Number($(cols[1]).text().trim());
     const porcentajeActual = Number($(cols[3]).text().trim());
-    
+
     capacityMap[embalse] = {
       capacity: capacidadTotalHm3,
       percentage: porcentajeActual
@@ -67,11 +67,11 @@ function parseAnnualStatsRow(
 export function extractReservoirsFromSeguraPage($: CheerioAPI): EmbalsesSegura[] {
   // Get capacity data from main table (#n0)
   const capacityMap = getReservoirCapacities($);
-  
+
   // Get most recent monthly data from annual table (#n1)
   const reservoirs: EmbalsesSegura[] = [];
   const annualRows = extractAnnualStatsRows($);
-  
+
   // Take only the LAST row (most recent month)
   if (annualRows.length > 0) {
     const lastRow = annualRows[annualRows.length - 1];

integrations/scraping-cuenca-segura/src/scraper/index.ts

@@ -1,2 +1,2 @@
-export * from './business';
-export * from './mapper';
+export * from './business.js';
+export * from './mapper.js';

integrations/scraping-cuenca-segura/src/scraper/mapper.ts

@@ -1,5 +1,5 @@
 import { EmbalseUpdateSAIHEntity } from 'db-model';
-import { EmbalsesSegura } from '@/api';
+import { EmbalsesSegura } from '../api/index.js';
 
 // Province lookup for each reservoir
 const reservoirProvince: Record<string, string> = {

integrations/scraping-cuenca-segura/tsconfig.json

@@ -1,17 +1,16 @@
 {
   "compilerOptions": {
     "target": "ESNext",
-    "module": "ESNext",
-    "moduleResolution": "bundler",
+    "module": "nodenext",
+    "moduleResolution": "nodenext",
+    "outDir": "dist",
     "skipLibCheck": true,
     "isolatedModules": true,
     "esModuleInterop": true,
-    "baseUrl": "./",
-    "paths": {
-      "@/*": ["src/*"],
-      "@/api/*": ["src/api/*"],
-      "@/scraper/*": ["src/scraper/*"]
-    }
+    "verbatimModuleSyntax": false,
+    "declaration": true,
+    "baseUrl": "./"
   },
-  "include": ["src"]
+  "include": ["src/**/*"],
+  "exclude": ["dist", "node_modules"]
 }