Ajoute un script pour extraire les données du fichier NomenclatureGuichet

converters/données-NomenclatureGuichet/convert.ts

@@ -0,0 +1,48 @@
+import parseCsv from 'csv-parser'
+import fs from 'fs'
+import { join } from 'path'
+import { fileURLToPath } from 'url'
+
+const FILENAME = 'NomenclatureGuichet_v1_26resana.csv'
+
+const __dirname = fileURLToPath(new URL('.', import.meta.url))
+const results = [] as Array<Activity>
+
+fs.createReadStream(join(__dirname, FILENAME))
+	.pipe(
+		parseCsv({
+			separator: ';',
+			skipLines: 1,
+			mapHeaders: ({ header }) =>
+				(header.match(/.+?(?=\n)/)?.[0] ?? header).trim(),
+			mapValues: ({ header, value }) =>
+				(header === 'Codes APE compatibles' && value.split(' ; ')) || value,
+		})
+	)
+	.on('data', (data) => results.push(data))
+	.on('end', () => {
+		fs.writeFileSync(
+			join(__dirname, 'raw_output.json'),
+			JSON.stringify(results)
+		)
+
+		fs.writeFileSync(
+			join(__dirname, 'ape_tags.json'),
+			JSON.stringify(computeAPETag(results))
+		)
+	})
+
+type CodeAPE = string
+type Activity = {
+	'Niv. 1': string
+	'Niv. 2': string
+	'Niv. 3': string
+	'Niv. 4': string
+	'Codes APE compatibles': Array<CodeAPE>
+}
+
+function computeAPETag(
+	results: Array<Activity>
+): Record<CodeAPE, Array<string>> {
+	return {}
+}

converters/package.json

@@ -12,6 +12,7 @@
 	"scripts": {
 		"test": "echo \"Error: no test specified\" && exit 1",
 		"start:données-NAF-CPF-APE": "ts-node-esm ./données-NAF-CPF-APE/convert-pdf.ts",
+		"start:données-NomenclatureGuichet": "ts-node-esm ./données-NomenclatureGuichet/convert.ts",
 		"start:nombre-etablissements-par-code-ape-et-departement": "ts-node-esm ./nombre-etablissements-par-code-ape-et-departement/convert-json.ts",
 		"start:données-code-APE": "ts-node-esm ./données-code-APE/reduce-json.ts",
 		"start:watch:données-NAF-CPF-APE": "nodemon -x 'yarn start:données-NAF-CPF-APE' -e 'ts'",
@@ -20,6 +21,8 @@
 	},
 	"dependencies": {
 		"@types/stream-json": "^1.7.2",
+		"csv": "^6.2.7",
+		"csv-parser": "^3.0.0",
 		"got": "^12.5.3",
 		"pdfdataextract": "^3.2.0",
 		"stream-json": "^1.7.5",

converters/tsconfig.json

@@ -35,7 +35,8 @@
 	"include": [
 		"données-code-APE/reduce-json.ts",
 		"données-NAF-CPF-APE/convert.ts",
-		"nombre-etablissements-par-code-ape-et-departement/convert-json.ts"
+		"nombre-etablissements-par-code-ape-et-departement/convert-json.ts",
+		"données-NomenclatureGuichet/convert.ts"
 	],
 	"exclude": ["**/node_modules"]
 }

yarn.lock

@@ -12820,6 +12820,8 @@ __metadata:
   resolution: "converters@workspace:converters"
   dependencies:
     "@types/stream-json": ^1.7.2
+    csv: ^6.2.7
+    csv-parser: ^3.0.0
     got: ^12.5.3
     nodemon: ^2.0.20
     pdfdataextract: ^3.2.0
@@ -13297,6 +13299,50 @@ __metadata:
   languageName: node
   linkType: hard
 
+"csv-generate@npm:^4.2.2":
+  version: 4.2.2
+  resolution: "csv-generate@npm:4.2.2"
+  checksum: f262698c92376261c928779335850c583f5d6ea9f2e476ba476ece7c9fb70bdafe634a4f3024721d44e46b7c50d2135d5932fa383f4fbe48dcaa642635052b2f
+  languageName: node
+  linkType: hard
+
+"csv-parse@npm:^5.3.5":
+  version: 5.3.5
+  resolution: "csv-parse@npm:5.3.5"
+  checksum: 077e010db2ebbc9db708b719f2f650577c5ec8d176cd4363235d6644c36c642811d05ef1b6907390dd7466bde20a8dbb0f34994bc0e101521eb09e6691a02206
+  languageName: node
+  linkType: hard
+
+"csv-parser@npm:^3.0.0":
+  version: 3.0.0
+  resolution: "csv-parser@npm:3.0.0"
+  dependencies:
+    minimist: ^1.2.0
+  bin:
+    csv-parser: bin/csv-parser
+  checksum: adc9d67d9f185249825570778c24d13004625301655330f6b735a052b9fdfbe1a239a014afb1f89939e0626ee573718f71f9f14164db7c17e4bcb2f38d6a162b
+  languageName: node
+  linkType: hard
+
+"csv-stringify@npm:^6.2.4":
+  version: 6.2.4
+  resolution: "csv-stringify@npm:6.2.4"
+  checksum: 12d8ab299735ad6c1ca90b37d900d66ef0fcc4289f565df7c67cbeed33f2de56daeae19094a215aa7d3bde8877c0a105f1aeb531e4f582d7561f213b0680c99d
+  languageName: node
+  linkType: hard
+
+"csv@npm:^6.2.7":
+  version: 6.2.7
+  resolution: "csv@npm:6.2.7"
+  dependencies:
+    csv-generate: ^4.2.2
+    csv-parse: ^5.3.5
+    csv-stringify: ^6.2.4
+    stream-transform: ^3.2.2
+  checksum: 0c8a310ddbe79ad78ff82ceeca72f2bc1090e876803c00fa43886c87a0bf5bd2ec5a51977dafe533acdec9da410ce84e70b66c756b47abe56b215c1379eaf92f
+  languageName: node
+  linkType: hard
+
 "currently-unhandled@npm:^0.4.1":
   version: 0.4.1
   resolution: "currently-unhandled@npm:0.4.1"
@@ -26975,6 +27021,13 @@ __metadata:
   languageName: node
   linkType: hard
 
+"stream-transform@npm:^3.2.2":
+  version: 3.2.2
+  resolution: "stream-transform@npm:3.2.2"
+  checksum: bc368a623b65f91d4d7faff90c46b740401fddedd79a4b325e5f9319d227894a84aca0e459d92177a679d5006d7ada3f46ebb514a1712da60b2e8beca54111c4
+  languageName: node
+  linkType: hard
+
 "streamx@npm:^2.12.5":
   version: 2.12.5
   resolution: "streamx@npm:2.12.5"