@stll/anonymize-wasm 1.1.1 → 1.3.0

package/dist/coreference.it.mjs

@@ -0,0 +1,27 @@
+//#region src/data/coreference.it.json
+var coreference_it_default = [
+	{
+		"label": "(in seguito \"X\") or (di seguito \"X\")",
+		"pattern": "\\((?:in\\s+seguito|di\\s+seguito|qui\\s+di\\s+seguito|nel\\s+seguito)\\s+(?:denominata|denominato|detta|detto|indicata|indicato|anche|la|il)?\\s*[\"'“”«»‘’‚]([^\"'“”«»‘’‚]+)[\"'“”«»‘’‚]\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "(denominato/denominata \"X\")",
+		"pattern": "\\((?:denominata|denominato|qui\\s+di\\s+seguito\\s+denominata|qui\\s+di\\s+seguito\\s+denominato)\\s+[\"'“”«»‘’‚]([^\"'“”«»‘’‚]+)[\"'“”«»‘’‚]\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "(la/il \"X\")",
+		"pattern": "\\((?:la|il|lo|le|gli|l')\\s*[\"'“”«»‘’‚]([^\"'“”«»‘’‚]+)[\"'“”«»‘’‚]\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "denominata/denominato \"X\" without parens",
+		"pattern": "(?:in\\s+seguito|di\\s+seguito)\\s+(?:denominata|denominato|detta|detto|indicata|indicato)\\s+[\"'“”«»‘’‚]([^\"'“”«»‘’‚]+)[\"'“”«»‘’‚]",
+		"flags": "gi"
+	}
+];
+//#endregion
+export { coreference_it_default as default };
+
+//# sourceMappingURL=coreference.it.mjs.map

package/dist/coreference.it.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"coreference.it.mjs","names":[],"sources":["../../src/data/coreference.it.json"],"sourcesContent":[""],"mappings":""}

package/dist/coreference.pl.mjs

@@ -0,0 +1,27 @@
+//#region src/data/coreference.pl.json
+var coreference_pl_default = [
+	{
+		"label": "(zwana/y dalej \"X\") parenthetical define-term",
+		"pattern": "zwan(?:ego|emu|ymi|ym|ej|[ayąei])?\\s+dalej\\s+[\"„‚']([^\"”’']+)[\"”’']",
+		"flags": "gi"
+	},
+	{
+		"label": "dalej zwana/y \"X\"",
+		"pattern": "dalej\\s+zwan(?:ego|emu|ymi|ym|ej|[ayąei])?\\s+[\"„‚']([^\"”’']+)[\"”’']",
+		"flags": "gi"
+	},
+	{
+		"label": "(\"X\") parenthetical define-term",
+		"pattern": "\\(\\s*[\"„‚']([^\"”’']+)[\"”’']\\s*\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "zwanymi dalej łącznie \"X\"",
+		"pattern": "zwanymi\\s+dalej\\s+(?:łącznie\\s+)?[\"„‚']([^\"”’']+)[\"”’']",
+		"flags": "gi"
+	}
+];
+//#endregion
+export { coreference_pl_default as default };
+
+//# sourceMappingURL=coreference.pl.mjs.map

package/dist/coreference.pl.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"coreference.pl.mjs","names":[],"sources":["../../src/data/coreference.pl.json"],"sourcesContent":[""],"mappings":""}

package/dist/coreference.pt-br.mjs

@@ -0,0 +1,14 @@
+//#region src/data/coreference.pt-br.json
+var coreference_pt_br_default = [{
+	"label": "(doravante denominada \"X\") / (doravante denominado \"X\")",
+	"pattern": "doravante\\s+denominad[ao]s?(?:\\s+simplesmente)?\\s+[\"'“”„‟‚‘’]?([^\"'“”„‟‚‘’,;.\\n)]+?)[\"'“”„‟‚‘’,;.\\n)]",
+	"flags": "gi"
+}, {
+	"label": "doravante referida como \"X\"",
+	"pattern": "doravante\\s+(?:referid[ao]|design[ao]d[ao])s?(?:\\s+como)?\\s+[\"'“”„‟‚‘’]?([^\"'“”„‟‚‘’,;.\\n)]+?)[\"'“”„‟‚‘’,;.\\n)]",
+	"flags": "gi"
+}];
+//#endregion
+export { coreference_pt_br_default as default };
+
+//# sourceMappingURL=coreference.pt-br.mjs.map

package/dist/coreference.pt-br.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"coreference.pt-br.mjs","names":[],"sources":["../../src/data/coreference.pt-br.json"],"sourcesContent":[""],"mappings":""}

package/dist/coreference.sk.mjs

@@ -1,9 +1,26 @@
 //#region src/data/coreference.sk.json
-var coreference_sk_default = [{
-	"label": "(dalej len \"X\")",
-	"pattern": "\\(ďalej\\s+len\\s+[„\"'‚]([^\"'\"\"]+)[\"\"']\\)",
-	"flags": "gi"
-}];
+var coreference_sk_default = [
+	{
+		"label": "(ďalej len \"X\") in parentheses",
+		"pattern": "\\(ďalej\\s+len\\s+[„\"'‚,]*([^\"'„“,;.\\n)]+?)[\"'„“,;.]*\\s*\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "ďalej len \"X\" without parens",
+		"pattern": "ďalej\\s+len\\s+[„\"'‚,]([^\"'„“,;.\\n]+?)[\"'„“]",
+		"flags": "gi"
+	},
+	{
+		"label": "(ďalej spolu ako \"X\")",
+		"pattern": "\\(ďalej\\s+spolu\\s+(?:len|ako)\\s+[„\"'‚,]*([^\"'„“,;.\\n)]+?)[\"'„“,;.]*\\s*\\)",
+		"flags": "gi"
+	},
+	{
+		"label": "(ďalej tiež \"X\")",
+		"pattern": "\\(ďalej\\s+tiež\\s+[„\"'‚,]*([^\"'„“,;.\\n)]+?)[\"'„“,;.]*\\s*\\)",
+		"flags": "gi"
+	}
+];
 //#endregion
 export { coreference_sk_default as default };
 

package/dist/generic-roles.mjs

@@ -1,59 +1,237 @@
 //#region src/data/generic-roles.json
 var _comment = "Generic legal/business role terms that are not PII. Used to filter false positives from name detection.";
 var roles = [
+	"acreedor",
+	"acreedora",
+	"adjudicatario",
+	"adjudicataria",
 	"arbeitgeber",
 	"arbeitnehmer",
+	"arrendador",
+	"arrendadora",
+	"arrendatario",
+	"arrendataria",
 	"auftraggeber",
 	"auftragnehmer",
 	"bonus",
 	"borrower",
 	"buyer",
+	"capitán",
+	"cedente",
 	"cena",
+	"cesionario",
+	"cesionaria",
 	"client",
+	"cliente",
+	"comprador",
+	"compradora",
 	"committee",
 	"company",
 	"contractor",
+	"contratante",
+	"contratista",
+	"coronel",
 	"customer",
 	"darlehensgeber",
 	"darlehensnehmer",
+	"deudor",
+	"deudora",
 	"dlužník",
+	"dlžník",
 	"dodavatel",
+	"dodávateľ",
+	"dostawca",
+	"dostawcą",
+	"dostawcy",
+	"empleado",
+	"empleador",
+	"empleadora",
+	"empleada",
+	"empresa",
+	"empresario",
+	"empresaria",
 	"employee",
 	"employer",
 	"fax",
+	"fiador",
+	"fiadora",
 	"gesellschaft",
 	"guarantor",
 	"kupující",
+	"kupujúci",
 	"käufer",
 	"landlord",
+	"licitador",
+	"licitadora",
 	"lender",
 	"licensee",
 	"licensor",
 	"mieter",
 	"nájemce",
+	"nájomca",
 	"objednatel",
+	"objednávateľ",
 	"odběratel",
+	"odberateľ",
+	"otorgante",
+	"otorgado",
+	"otorgada",
+	"parte",
+	"partes",
 	"parties",
 	"party",
 	"plátce",
+	"poskytovateľ",
+	"prestador",
+	"prestadora",
+	"prestatario",
+	"prestataria",
+	"prijímateľ",
 	"prodávající",
+	"predávajúci",
 	"pronajímatel",
+	"prenajímateľ",
+	"proveedor",
+	"proveedora",
 	"příjemce",
 	"seller",
+	"sociedad",
+	"spoločnosť",
 	"společnost",
 	"strana",
 	"strany",
 	"supplier",
 	"tenant",
+	"vendedor",
+	"vendedora",
 	"vendor",
 	"vermieter",
 	"verkäufer",
 	"věřitel",
+	"veriteľ",
 	"uchazeč",
+	"ubytovateľ",
+	"usporiadateľ",
 	"zadavatel",
+	"zamawiający",
+	"zamawiającego",
+	"zamawiającym",
 	"zaměstnanec",
+	"zamestnanec",
 	"zaměstnavatel",
-	"zhotovitel"
+	"zamestnávateľ",
+	"zhotovitel",
+	"zhotoviteľ",
+	"acquirente",
+	"amministrazione",
+	"appaltatore",
+	"appaltatrice",
+	"cedente",
+	"cessionario",
+	"cliente",
+	"committente",
+	"compratore",
+	"compratrice",
+	"concedente",
+	"concessionario",
+	"conduttore",
+	"conduttrice",
+	"creditore",
+	"datore",
+	"debitore",
+	"denominazione",
+	"dipendente",
+	"fideiussore",
+	"fornitore",
+	"fornitrice",
+	"garante",
+	"lavoratore",
+	"lavoratrice",
+	"licenziante",
+	"licenziatario",
+	"locatario",
+	"locataria",
+	"locatore",
+	"locatrice",
+	"mandante",
+	"mandatario",
+	"mandataria",
+	"mutuante",
+	"mutuatario",
+	"parte",
+	"parti",
+	"stazione",
+	"subappaltatore",
+	"venditore",
+	"venditrice",
+	"zleceniodawca",
+	"zleceniodawcą",
+	"zleceniodawcy",
+	"zleceniobiorca",
+	"zleceniobiorcą",
+	"zleceniobiorcy",
+	"wykonawca",
+	"wykonawcą",
+	"wykonawcy",
+	"wynajmujący",
+	"wynajmująca",
+	"wynajmującym",
+	"wynajmującej",
+	"najemca",
+	"najemcy",
+	"sprzedawca",
+	"sprzedawcy",
+	"sprzedawcą",
+	"sprzedający",
+	"kupujący",
+	"kupującego",
+	"pracodawca",
+	"pracodawcą",
+	"pracownik",
+	"pracownikiem",
+	"pożyczkodawca",
+	"pożyczkodawcą",
+	"pożyczkobiorca",
+	"pożyczkobiorcą",
+	"odbiorca",
+	"odbiorcą",
+	"acheteur",
+	"acheteuse",
+	"acquéreur",
+	"acquéreuse",
+	"bailleresse",
+	"bailleur",
+	"cédant",
+	"cédante",
+	"cessionnaire",
+	"concédant",
+	"concédante",
+	"concessionnaire",
+	"donneur d'ordre",
+	"emprunteur",
+	"emprunteuse",
+	"employé",
+	"employée",
+	"employeur",
+	"employeuse",
+	"fournisseur",
+	"fournisseuse",
+	"garant",
+	"garante",
+	"locataire",
+	"mandant",
+	"mandante",
+	"mandataire",
+	"preneur",
+	"preneuse",
+	"prestataire",
+	"prêteur",
+	"prêteuse",
+	"salarié",
+	"salariée",
+	"société",
+	"vendeur",
+	"vendeuse"
 ];
 var generic_roles_default = {
 	_comment,

package/dist/hotword-rules.mjs

@@ -7,7 +7,7 @@ var rules = [
 			"birth number",
 			"Geburtsnummer"
 		],
-		"targetLabels": ["czech birth number"],
+		"targetLabels": ["birth number"],
 		"scoreAdjustment": .25,
 		"proximityBefore": 60,
 		"proximityAfter": 60
@@ -87,7 +87,8 @@ var rules = [
 		"hotwords": [
 			"e-mail",
 			"email",
-			"E-Mail"
+			"E-Mail",
+			"pec"
 		],
 		"targetLabels": ["email address"],
 		"scoreAdjustment": .1,
@@ -101,7 +102,10 @@ var rules = [
 			"sídlem",
 			"bytem",
 			"Adresse",
-			"address"
+			"address",
+			"indirizzo",
+			"sede legale",
+			"sede"
 		],
 		"targetLabels": ["address"],
 		"scoreAdjustment": .15,

package/dist/legal-forms.mjs

@@ -282,7 +282,10 @@ var legal_forms_default = {
 		"Empresa Individual de Responsabilidade Limitada",
 		"EPP",
 		"Fundação",
+		"Ltda",
 		"Ltda.",
+		"LTDA",
+		"LTDA.",
 		"ME",
 		"MEI",
 		"Microempreendedor Individual",

package/dist/legal-role-heads.cs.mjs

@@ -0,0 +1,42 @@
+//#region src/data/legal-role-heads.cs.json
+var _comment = "Czech contract/legal role words. When the legal-form detector picks up a phrase whose first word is one of these, the span is sentence noise (\"Prodávající 1 je vlastníkem podílu ve společnosti Acme s.r.o.\") and gets trimmed back to the actual organisation name. Lowercased; matched case-insensitively. Cover nominative + most common declensions.";
+var words = [
+	"prodávající",
+	"prodávajícího",
+	"prodávajícím",
+	"kupující",
+	"kupujícího",
+	"kupujícím",
+	"nájemce",
+	"nájemci",
+	"pronajímatel",
+	"pronajímatele",
+	"pronajímateli",
+	"objednatel",
+	"objednatele",
+	"zhotovitel",
+	"zhotovitele",
+	"zaměstnavatel",
+	"zaměstnavatele",
+	"zaměstnanec",
+	"zaměstnance",
+	"dlužník",
+	"dlužníka",
+	"věřitel",
+	"věřitele",
+	"dodavatel",
+	"dodavatele",
+	"odběratel",
+	"odběratele",
+	"smluvní",
+	"strana",
+	"strany"
+];
+var legal_role_heads_cs_default = {
+	_comment,
+	words
+};
+//#endregion
+export { _comment, legal_role_heads_cs_default as default, words };
+
+//# sourceMappingURL=legal-role-heads.cs.mjs.map

package/dist/legal-role-heads.cs.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"legal-role-heads.cs.mjs","names":[],"sources":["../../src/data/legal-role-heads.cs.json"],"sourcesContent":[""],"mappings":""}

package/dist/legal-role-heads.de.mjs

@@ -0,0 +1,33 @@
+//#region src/data/legal-role-heads.de.json
+var _comment = "German contract/legal role words. Used by the legal-form detector to recognise sentence-level false positives and trim them back to the trailing organisation name.";
+var words = [
+	"verkäufer",
+	"verkaeufer",
+	"käufer",
+	"kaeufer",
+	"vermieter",
+	"mieter",
+	"arbeitgeber",
+	"arbeitnehmer",
+	"auftraggeber",
+	"auftragnehmer",
+	"darlehensgeber",
+	"darlehensnehmer",
+	"lizenzgeber",
+	"lizenznehmer",
+	"verleiher",
+	"entleiher",
+	"vertragspartei",
+	"vertragsparteien",
+	"schuldner",
+	"gläubiger",
+	"glaeubiger"
+];
+var legal_role_heads_de_default = {
+	_comment,
+	words
+};
+//#endregion
+export { _comment, legal_role_heads_de_default as default, words };
+
+//# sourceMappingURL=legal-role-heads.de.mjs.map

package/dist/legal-role-heads.de.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"legal-role-heads.de.mjs","names":[],"sources":["../../src/data/legal-role-heads.de.json"],"sourcesContent":[""],"mappings":""}

package/dist/legal-role-heads.en.mjs

@@ -0,0 +1,37 @@
+//#region src/data/legal-role-heads.en.json
+var _comment = "English contract/legal role words. Used by the legal-form detector to recognise sentence-level false positives and trim them back to the trailing organisation name.";
+var words = [
+	"buyer",
+	"seller",
+	"vendor",
+	"purchaser",
+	"lender",
+	"borrower",
+	"employer",
+	"employee",
+	"contractor",
+	"subcontractor",
+	"client",
+	"customer",
+	"landlord",
+	"tenant",
+	"licensor",
+	"licensee",
+	"lessor",
+	"lessee",
+	"party",
+	"parties",
+	"guarantor",
+	"assignor",
+	"assignee",
+	"disclosing",
+	"receiving"
+];
+var legal_role_heads_en_default = {
+	_comment,
+	words
+};
+//#endregion
+export { _comment, legal_role_heads_en_default as default, words };
+
+//# sourceMappingURL=legal-role-heads.en.mjs.map

package/dist/legal-role-heads.en.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"legal-role-heads.en.mjs","names":[],"sources":["../../src/data/legal-role-heads.en.json"],"sourcesContent":[""],"mappings":""}

package/dist/legal-role-heads.es.mjs

@@ -0,0 +1,54 @@
+//#region src/data/legal-role-heads.es.json
+var _comment = "Spanish contract/legal role words. Used by the legal-form detector to recognise sentence-level false positives and trim them back to the trailing organisation name. Lowercased; matched case-insensitively. Includes common masculine/feminine variants.";
+var words = [
+	"vendedor",
+	"vendedora",
+	"comprador",
+	"compradora",
+	"arrendador",
+	"arrendadora",
+	"arrendatario",
+	"arrendataria",
+	"empleador",
+	"empleadora",
+	"empleado",
+	"empleada",
+	"empresa",
+	"empresario",
+	"empresaria",
+	"cliente",
+	"proveedor",
+	"proveedora",
+	"contratante",
+	"contratista",
+	"prestador",
+	"prestadora",
+	"prestatario",
+	"prestataria",
+	"otorgante",
+	"otorgado",
+	"otorgada",
+	"cedente",
+	"cesionario",
+	"cesionaria",
+	"acreedor",
+	"acreedora",
+	"deudor",
+	"deudora",
+	"fiador",
+	"fiadora",
+	"parte",
+	"partes",
+	"adjudicatario",
+	"adjudicataria",
+	"licitador",
+	"licitadora"
+];
+var legal_role_heads_es_default = {
+	_comment,
+	words
+};
+//#endregion
+export { _comment, legal_role_heads_es_default as default, words };
+
+//# sourceMappingURL=legal-role-heads.es.mjs.map

package/dist/legal-role-heads.es.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"legal-role-heads.es.mjs","names":[],"sources":["../../src/data/legal-role-heads.es.json"],"sourcesContent":[""],"mappings":""}

package/dist/legal-role-heads.fr.mjs

@@ -0,0 +1,72 @@
+//#region src/data/legal-role-heads.fr.json
+var _comment = "French contract/legal role words. Used by the legal-form detector to recognise sentence-level false positives and trim them back to the trailing organisation name.";
+var words = [
+	"acheteur",
+	"acheteuse",
+	"acquéreur",
+	"acquéreuse",
+	"acquereur",
+	"associé",
+	"associée",
+	"associés",
+	"bailleur",
+	"bailleresse",
+	"cédant",
+	"cédante",
+	"cedant",
+	"cessionnaire",
+	"client",
+	"cliente",
+	"co-contractant",
+	"cocontractant",
+	"concédant",
+	"concédante",
+	"concedant",
+	"concessionnaire",
+	"consommateur",
+	"consommatrice",
+	"donneur d'ordre",
+	"emprunteur",
+	"emprunteuse",
+	"employeur",
+	"employeuse",
+	"employée",
+	"employé",
+	"entreprise",
+	"fournisseur",
+	"fournisseuse",
+	"garant",
+	"garante",
+	"locataire",
+	"loueur",
+	"loueuse",
+	"mandant",
+	"mandante",
+	"mandataire",
+	"partie",
+	"parties",
+	"preneur",
+	"preneuse",
+	"prestataire",
+	"preteur",
+	"prêteur",
+	"prêteuse",
+	"promettant",
+	"salarié",
+	"salariée",
+	"société",
+	"sous-traitant",
+	"souscripteur",
+	"souscriptrice",
+	"vendeur",
+	"vendeuse",
+	"venderesse"
+];
+var legal_role_heads_fr_default = {
+	_comment,
+	words
+};
+//#endregion
+export { _comment, legal_role_heads_fr_default as default, words };
+
+//# sourceMappingURL=legal-role-heads.fr.mjs.map

package/dist/legal-role-heads.fr.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"legal-role-heads.fr.mjs","names":[],"sources":["../../src/data/legal-role-heads.fr.json"],"sourcesContent":[""],"mappings":""}