fix: typologie algo improve (#108)

* fix: check deeper for process typologie algo * fix: add some rules on matcher file * fix: prettier and linter * fix: refactor * fix: linter * fix: email multiple at
anct-cartographie-nationale · Jul 28, 2023 · 457e7ed · 457e7ed
1 parent 953adeb
commit 457e7ed
Show file tree

Hide file tree

Showing 5 changed files with 74 additions and 14 deletions.
diff --git a/src/transformer/fields/contact/clean-operations.ts b/src/transformer/fields/contact/clean-operations.ts
@@ -313,6 +313,12 @@ const removeDashEmail = (field: string): CleanOperation => ({
   field
 });
 
+const removeMultipleAtEmail = (field: string): CleanOperation => ({
+  name: 'multiple at',
+  selector: /@.+@/u,
+  field
+});
+
 const cleanOperationIfAny = (
   cleanOperator: (colonne: string, codePostal?: string) => CleanOperation,
   colonne?: string,
@@ -365,5 +371,6 @@ export const cleanOperations = (
   ...cleanOperationIfAny(fixUnexpectedEmailList, matching.courriel?.colonne),
   ...cleanOperationIfAny(fixObfuscatedAtInEmail, matching.courriel?.colonne),
   ...cleanOperationIfAny(fixMissingEmailExtension, matching.courriel?.colonne),
-  ...cleanOperationIfAny(removeMissingAtInEmail, matching.courriel?.colonne)
+  ...cleanOperationIfAny(removeMissingAtInEmail, matching.courriel?.colonne),
+  ...cleanOperationIfAny(removeMultipleAtEmail, matching.courriel?.colonne)
 ];
diff --git a/src/transformer/fields/contact/contact.field.spec.ts b/src/transformer/fields/contact/contact.field.spec.ts
@@ -1019,4 +1019,15 @@ describe('contact field', (): void => {
       })
     );
   });
+
+  it('should remove email if multiple At', (): void => {
+    const contact: Contact = processContact(Report().entry(0))(
+      {
+        [EMAIL_FIELD]: 'msap@marchaux1@orange.fr'
+      } as DataSource,
+      matching
+    );
+
+    expect(contact).toStrictEqual<Contact>(Contact({}));
+  });
 });
diff --git a/src/transformer/fields/typologies/name-to-typologie.ts b/src/transformer/fields/typologies/name-to-typologie.ts
@@ -6,15 +6,15 @@ import { TypologieMatcher } from './typologies.field';
 export const TYPOLOGIE_MATCHERS: TypologieMatcher[] = [
   {
     typologie: Typologie.MUNI,
-    matchers: [/mairie/iu, /commune/iu, /^ville d[eu]/iu, /hôtel de ville/iu]
+    matchers: [/mairie/iu, /commune/iu, /^ville d[eu]/iu, /h[oô]tel de ville/iu]
   },
   {
     typologie: Typologie.PE,
     matchers: [/p[ôo]le emploi/iu]
   },
   {
     typologie: Typologie.BIB,
-    matchers: [/m[ée]diath[èe]que/iu, /biblioth[èe]que/iu]
+    matchers: [/m[ée]diath[èeé]que/iu, /biblioth[èeé]que/iu]
   },
   {
     typologie: Typologie.TIERS_LIEUX,
@@ -32,14 +32,14 @@ export const TYPOLOGIE_MATCHERS: TypologieMatcher[] = [
     typologie: Typologie.MDS,
     matchers: [
       /maison du d[ée]partement/iu,
-      /(?:maison|espace) départementale? des solidarités/iu,
-      /(?:maison|espace) départementale? de la solidarité/iu,
-      /(?:maison|espace) de la solidarité départementale/iu
+      /(?:maison|espace) d[eé]partementale? des solidarit[eé]s/iu,
+      /(?:maison|espace) d[eé]partementale? de la solidarit[ée]/iu,
+      /(?:maison|espace) de la solidarit[eé] d[eé]partementale/iu
     ]
   },
   {
     typologie: Typologie.CHRS,
-    matchers: [/centre d'hébergement et de réinsertion sociale/iu]
+    matchers: [/centre d'h[eé]bergement et de r[eé]insertion sociale/iu]
   },
   {
     typologie: Typologie.CHU,
@@ -63,7 +63,7 @@ export const TYPOLOGIE_MATCHERS: TypologieMatcher[] = [
       /(?:^|\W)ASS(?:\W|$)/iu,
       /(?:^|\W)ASSOC(?:\W|$)/iu,
       /association/iu,
-      /emmaüs/iu,
+      /emma[üu]s/iu,
       /secours populaire/iu,
       /croix[\s-]Rouge/iu,
       /secours catholique/iu,
@@ -78,11 +78,11 @@ export const TYPOLOGIE_MATCHERS: TypologieMatcher[] = [
   },
   {
     typologie: Typologie.PREF,
-    matchers: [/sous[-\s]prefecture/iu]
+    matchers: [/sous[-\s]pr[ée]fecture/iu]
   },
   {
     typologie: Typologie.CD,
-    matchers: [/conseil départemental/iu]
+    matchers: [/conseil d[eé]partemental/iu]
   },
   {
     typologie: Typologie.CC,
@@ -116,7 +116,7 @@ export const TYPOLOGIE_MATCHERS: TypologieMatcher[] = [
     typologie: Typologie.CCONS,
     matchers: [
       /chambre[\w\s']+agriculture/iu,
-      /chambre[\w\s']+metiers[\w\s']+artisanat/iu,
+      /chambre[\w\s']+m[ée]tiers[\w\s']+artisanat/iu,
       /chambre[\w\s']+commerce[\w\s']+industrie/iu,
       /(?:^|\W)CCI(?:\W|$)/iu
     ]
@@ -134,9 +134,9 @@ export const TYPOLOGIE_MATCHERS: TypologieMatcher[] = [
     matchers: [
       /(?:^|\W)EFS(?:\W|$)/iu,
       /frances?[\s-]services?/iu,
-      /Fixe Bruay-sur-l'Escaut\s{2}\( Département du Nord\)/iu,
-      /Folschviller - Antenne de L'Hôpital/iu,
-      /Communauté de communes Vaison Ventoux/iu
+      /Fixe Bruay-sur-l'Escaut\s{2}\( D[ée]partement du Nord\)/iu,
+      /Folschviller - Antenne de L'H[ôo]pital/iu,
+      /Communaut[eé] de communes Vaison Ventoux/iu
     ]
   },
   {

diff --git a/src/transformer/fields/typologies/typologies.field.ts b/src/transformer/fields/typologies/typologies.field.ts
@@ -60,7 +60,18 @@ const inferTypologies = (source: DataSource, matching: LieuxMediationNumeriqueMa
     ? Typologies([Typologie.RFS])
     : TYPOLOGIE_MATCHERS.reduce(toTypologieMatchingName(source, matching), Typologies([]));
 
+const checkingTypologieSourceValues = (source: DataSource, matching: LieuxMediationNumeriqueMatching): boolean[] | undefined =>
+  matching.typologie?.map(
+    (typo: Choice<Typologie>): boolean =>
+      typo.colonnes != null && source[typo.colonnes[0] ?? ''] != null && source[typo.colonnes[0] ?? ''] !== ''
+  );
+
+const typologiesArePreset = (matching: LieuxMediationNumeriqueMatching): boolean =>
+  matching.typologie?.[0]?.cible != null && matching.typologie[0].colonnes == null && matching.typologie[0].termes == null;
+
 export const processTypologies = (source: DataSource, matching: LieuxMediationNumeriqueMatching): Typologies =>
+  ((checkingTypologieSourceValues(source, matching) ?? []).some((check: boolean): boolean => !check) &&
+    !typologiesArePreset(matching)) ||
   matching.typologie?.at(0)?.cible == null
     ? inferTypologies(source, matching)
     : Typologies(Array.from(new Set(matching.typologie.reduce(appendTypologies(source), []))));
diff --git a/src/transformer/fields/typologies/typologies.spec.ts b/src/transformer/fields/typologies/typologies.spec.ts
@@ -23,4 +23,35 @@ describe('typologies field', (): void => {
 
     expect(typologies).toStrictEqual([Typologie.BIB]);
   });
+
+  it('should process algo research in name even if there is matching', (): void => {
+    const matching: LieuxMediationNumeriqueMatching = {
+      nom: { colonne: 'name' },
+      typologie: [{ colonnes: ['checkboxListeTypelieu'], termes: ['5'], cible: Typologie.BIB }]
+    } as LieuxMediationNumeriqueMatching;
+
+    const typologies: Typologies = processTypologies({ name: 'médiathèque' }, matching);
+
+    expect(typologies).toStrictEqual([Typologie.BIB]);
+  });
+
+  it('should get Bib even if the typo of mediathèque is not good', (): void => {
+    const matching: LieuxMediationNumeriqueMatching = {
+      nom: { colonne: 'name' }
+    } as LieuxMediationNumeriqueMatching;
+
+    const typologies: Typologies = processTypologies({ name: 'médiathéque' }, matching);
+
+    expect(typologies).toStrictEqual([Typologie.BIB]);
+  });
+
+  it('should get Bib even if the typo of bibliothèque is not good', (): void => {
+    const matching: LieuxMediationNumeriqueMatching = {
+      nom: { colonne: 'name' }
+    } as LieuxMediationNumeriqueMatching;
+
+    const typologies: Typologies = processTypologies({ name: 'bibliothéque' }, matching);
+
+    expect(typologies).toStrictEqual([Typologie.BIB]);
+  });
 });