wololo/csv_to_geojson.ts

/**
 * csv_to_geojson.ts
 * 
 * Convertir un fichier CSV en GeoJSON
 * 
 * Utilisation:
 * 
 * npx ts-node csv_to_geojson.ts -d etalab_data/panneaux -f panneaux_limite_de_vitesse_fr_panoramax_detections.csv --latColumn 'GPSLatitude' --lonColumn 'GPSLongitude' -h
 * 
 */
import fs from 'fs';
import path from 'path';
import csvParser from 'csv-parser';
import minimist from 'minimist';
import { Feature, FeatureCollection, Point } from 'geojson';

interface Options {
  dir: string;
  file: string;
  latColumn: string;
  lonColumn: string;
  hasHeaders: boolean;
}
let counter = 0;
function csvToGeoJSON(options: Options): FeatureCollection<Point> {
  const { dir, file, latColumn, lonColumn, hasHeaders } = options;

  console.log(`options latColumn:  ${latColumn}`);
  console.log(`options lonColumn:  ${lonColumn}`);

  const filePath = path.join(dir, file);
  const features: Feature<Point>[] = [];
  let geoJSON: FeatureCollection<Point> = {
    type: 'FeatureCollection',
    features,
  };

  let headers: string[] = [];
  let headersFound = false;

  let limitOffset = 100;

  console.log(`hasHeaders:  ${hasHeaders}`);
  fs.createReadStream(filePath)
    .pipe(csvParser({ headers: hasHeaders }))
    .on('data', (row) => {
      counter++;
      if (!headersFound && hasHeaders) {
        let keys = Object.keys(row);
        keys.forEach((key) => {
          headers.push(row[key]);
        });
        headersFound = true;
        console.log(`headers:  ${headers}`);
      }
      if (counter > limitOffset) {
        return;
      }
      if (headers.indexOf(latColumn)) {
        console.log(`latColumn:  ${latColumn}`);
        console.log(`headers latColumn:  ${headers.indexOf(latColumn)}`);
        console.log(`headers.indexOf(latColumn):  `, headers.indexOf(latColumn));
        console.log('row', row);
        const lat = parseFloat(row['_' + headers.indexOf(latColumn)]);
        const lon = parseFloat(row['_' + headers.indexOf(lonColumn)]);

        // Si pas d'entêtes, utiliser des noms de colonnes génériques
        if (!hasHeaders) {
          const properties: { [key: string]: any } = {};
          Object.keys(row).forEach((key, index) => {
            properties[headers[index]] = row[key];
          });
          row = properties;
        } else {
          let keys = Object.keys(row);
          // Utiliser les entêtes comme noms de propriétés
          const properties: { [key: string]: any } = {};
          headers.forEach((header, index) => {

            properties[header] = row['_' + index];
          });
          row = properties;
        }

        // filtrer la ligne du header si présente
        if (hasHeaders && counter > 1 || !hasHeaders || counter > limitOffset) {
          features.push({
            type: 'Feature',
            geometry: {
              type: 'Point',
              coordinates: [lon, lat],
            },
            properties: row,
          });
        }
      }
      else {
        console.log('!!! no latColumn', row);
      }

    })
    .on('end', () => {
      geoJSON = {
        type: 'FeatureCollection',
        features,
      };
      fs.writeFileSync(`${dir}/${file}.geojson`, JSON.stringify(geoJSON, null, 2));
      console.log(`GeoJSON créé avec succès : ${file}.geojson`);
    });

  console.log(`geoJSON lines:  ${counter}`);
  return geoJSON;
}

const args = minimist<Options>(process.argv.slice(2), {
  alias: {
    dir: 'd',
    file: 'f',
    latColumn: 'lat',
    lonColumn: 'lon',
    hasHeaders: 'h',
  },
  default: {
    hasHeaders: true,
  },
});

function checkFile(args: Options) {
  let filePath = path.join(args.dir, args.file);
  let lineCount = 0;

  // Vérifier si le fichier existe
  if (!fs.existsSync(filePath)) {
    throw new Error(`Le fichier CSV ${filePath} n'existe pas`);
  } else {
    console.log(`Le fichier CSV ${filePath} existe`);
  }

  fs.createReadStream(filePath)
    .on('data', () => {
      lineCount++;
    })
    .on('end', () => {
      console.log(`Nombre de lignes dans le fichier CSV : ${Math.floor(lineCount)}`);
    });


}

console.log(`args:  `, args);
checkFile(args);
csvToGeoJSON(args);