Added bal meta clefInterop retrieval and check in export from postgre…

…s to mongo

lib/api/consumers/format-to-legacy-helpers.js

@@ -8,7 +8,13 @@ const districtsGeographicOutlineJsonURL = new URL('../../../geo.json', import.me
 const districtsGeographicOutline = JSON.parse(readFileSync(districtsGeographicOutlineJsonURL))
 
 const districtsAddressesExtraDataJsonURL = new URL(`../../../${process.env.COMMUNES_LOCAUX_ADRESSES_DATA_PATH}`, import.meta.url)
-const districtsAddressesExtraData = JSON.parse(readFileSync(districtsAddressesExtraDataJsonURL))
+let districtsAddressesExtraData = []
+try {
+  districtsAddressesExtraData = JSON.parse(readFileSync(districtsAddressesExtraDataJsonURL))
+} catch (error) {
+  console.error('Error while reading the districts addresses extra data file', error)
+}
+
 const districtsAddressesExtraDataIndex = districtsAddressesExtraData.reduce((acc, commune) => {
   acc[commune.codeCommune] = commune
   return acc
@@ -107,15 +113,7 @@ export const formatCommonToponymDataForLegacy = (commonToponym, district, pseudo
   const codeAncienneCommune = meta?.bal?.codeAncienneCommune
   const legacyCommonToponymFantoirId = meta?.dgfip?.fantoir ? `${cog}_${meta?.dgfip?.fantoir}` : null
 
-  let legacyCommonToponymId = legacyCommonToponymFantoirId
-  // If the legacy common toponym id is already used or not defined, we calculate a pseudo code
-  if (!legacyCommonToponymId || commonToponymLegacyIDSet.has(legacyCommonToponymId)) {
-    legacyCommonToponymId = `${cog}_${pseudoCodeVoieGenerator.getCode(legacyLabelValue, codeAncienneCommune)}`.toLowerCase()
-    // If the pseudo code is already used, we generate a new one with a hash from the common toponym id
-    if (commonToponymLegacyIDSet.has(legacyCommonToponymId)) {
-      legacyCommonToponymId = `${cog}_${createHmac('sha256', 'ban').update(id).digest('hex').slice(0, 6)}`
-    }
-  }
+  const legacyCommonToponymId = getLegacyCommonToponymId(commonToponymLegacyIDSet, legacyCommonToponymFantoirId, meta, {cog, id, pseudoCodeVoieGenerator, legacyLabelValue, codeAncienneCommune})
 
   // Store the legacy common toponym id for each common toponym to then be able to set it on legacy addresses
   commonToponymLegacyIDCommonToponymIDMap.set(id, legacyCommonToponymId)
@@ -212,7 +210,7 @@ export const formatAddressDataForLegacy = (address, district, commonToponymLegac
 
   // Ids
   const legacyCommonToponymId = commonToponymLegacyIDCommonToponymIDMap.get(mainCommonToponymID)
-  const legacyInteropKey = `${legacyCommonToponymId}_${String(number).padStart(5, '0')}${suffix ? `_${suffix}` : ''}`.toLowerCase()
+  const legacyInteropKey = getAddressLegacyInteropKey(meta, {legacyCommonToponymId, number, suffix})
   const legacyID = getAddressLegacyId(addressLegacyIDSet, legacyInteropKey)
   addressLegacyIDSet.add(legacyID)
   const banIdSecondaryCommonToponyms = secondaryCommonToponymIDs && secondaryCommonToponymIDs.length > 0 ? secondaryCommonToponymIDs : null
@@ -332,3 +330,70 @@ const getAddressLegacyId = (addressLegacyIDSet, legacyInteropKey, suffix = 0) =>
 
   return `${legacyInteropKey}`
 }
+
+const getAddressLegacyInteropKey = (meta, {legacyCommonToponymId, number, suffix}) => {
+  const cleInterop = meta.bal?.cleInterop
+  const isValidInteropKey = checkIfCleInteropIsValid(cleInterop)
+  const addressLegacyInteropKey = isValidInteropKey
+    ? cleInterop
+    : `${legacyCommonToponymId}_${String(number).padStart(5, '0')}${suffix ? `_${suffix}` : ''}`.toLowerCase()
+
+  return addressLegacyInteropKey
+}
+
+const getLegacyCommonToponymId = (
+  commonToponymLegacyIDSet,
+  legacyCommonToponymFantoirId,
+  meta,
+  {cog, id, pseudoCodeVoieGenerator, legacyLabelValue, codeAncienneCommune}
+) => {
+  let legacyCommonToponymId
+  const cleInterop = meta.bal?.cleInterop
+  const isCleInteropValid = checkIfCleInteropIsValid(cleInterop)
+  if (isCleInteropValid) {
+    const cleInteropParts = cleInterop.split('_')
+    legacyCommonToponymId = `${cleInteropParts[0]}_${cleInteropParts[1]}`
+  } else {
+    legacyCommonToponymId = legacyCommonToponymFantoirId
+  }
+
+  // If the legacy common toponym id is already used or not defined, we calculate a pseudo code
+  if (!legacyCommonToponymId || commonToponymLegacyIDSet.has(legacyCommonToponymId)) {
+    legacyCommonToponymId = `${cog}_${pseudoCodeVoieGenerator.getCode(legacyLabelValue, codeAncienneCommune)}`.toLowerCase()
+    // If the pseudo code is already used, we generate a new one with a hash from the common toponym id
+    if (commonToponymLegacyIDSet.has(legacyCommonToponymId)) {
+      legacyCommonToponymId = `${cog}_${createHmac('sha256', 'ban').update(id).digest('hex').slice(0, 6)}`
+    }
+  }
+
+  return legacyCommonToponymId
+}
+
+export const checkIfCleInteropIsValid = cleInterop => {
+  if (!cleInterop) {
+    return false
+  }
+
+  // https://aitf-sig-topo.github.io/voies-adresses/files/AITF_SIG_Topo_Format_Base_Adresse_Locale_v1.3.pdf
+  // ● INSEE code with 5 characters
+  // ● Street code: the unique street identifier provided by the national unique service,
+  //   or the FANTOIR DGFiP code, consisting of 4 to 6 alphanumeric characters
+  // ● Address number consisting of 5 characters, prefixed with zeros if necessary
+  // ● Suffix (bis / ter / qua / qui / a / b / c...). The repetition indices “bis, ter…” will
+  //   be coded with 3 characters, and others (a, b, c, a1, b2...) will be in lowercase
+  //   without a fixed number of characters.
+  // ● Each item is separated by an underscore “_”
+  // ● Everything is in lowercase
+
+  const cleInteropFormatRegex = /^[a-z\d]{5}_(?:[a-z\d]{4}|[a-z\d]{6})_\d{5}(_.+)?$/
+  if (!cleInteropFormatRegex.test(cleInterop)) {
+    return false
+  }
+
+  // Check if the street code is 'xxxx' to detect the non-valid pseudo code created by "mes-adresses" service
+  if (cleInterop.split('_')[1] === 'xxxx') {
+    return false
+  }
+
+  return true
+}

lib/api/consumers/format-to-legacy-helpers.spec.js

@@ -0,0 +1,36 @@
+import {checkIfCleInteropIsValid} from './format-to-legacy-helpers.js'
+
+describe('checkIfCleInteropIsValid', () => {
+  test('should return false for an empty string', () => {
+    expect(checkIfCleInteropIsValid('')).toBe(false)
+  })
+
+  test('should return false for a null value', () => {
+    expect(checkIfCleInteropIsValid(null)).toBe(false)
+  })
+
+  test('should return false for a string with "xxxx" as the street code', () => {
+    expect(checkIfCleInteropIsValid('31555_xxxx_00001')).toBe(false)
+  })
+
+  test('should return false for a string not matching the pattern', () => {
+    expect(checkIfCleInteropIsValid('315_7172_00001')).toBe(false)
+    expect(checkIfCleInteropIsValid('31555_7172a_00001')).toBe(false)
+    expect(checkIfCleInteropIsValid('31555_ilwhc0_000001')).toBe(false)
+    expect(checkIfCleInteropIsValid('31555_ilwhc0_0000a')).toBe(false)
+    expect(checkIfCleInteropIsValid('31555_Ilwhc0_00001')).toBe(false)
+  })
+
+  test('should return false for a string with invalid characters', () => {
+    expect(checkIfCleInteropIsValid('31555_ilwhc#_00001')).toBe(false)
+    expect(checkIfCleInteropIsValid('31555_ilwhc!_00001')).toBe(false)
+  })
+
+  test('should return true for a string matching the pattern', () => {
+    expect(checkIfCleInteropIsValid('31555_ilwhc0_00001')).toBe(true)
+    expect(checkIfCleInteropIsValid('31555_7172_00001')).toBe(true)
+    expect(checkIfCleInteropIsValid('31555_7172_00001_bis')).toBe(true)
+    expect(checkIfCleInteropIsValid('31555_7172_00001_a1')).toBe(true)
+    expect(checkIfCleInteropIsValid('31555_7172_00001_bât. a')).toBe(true)
+  })
+})