@stll/anonymize 1.5.0 → 2.0.0-alpha.1

package/dist/person-stopwords.mjs

@@ -1,205 +1,211 @@
 //#region src/data/person-stopwords.json
+var _comment = "Words that are valid in other labels (address, org) but should never be classified as person. Checked only in person chain scoring.";
+var cs = ["cena"];
+var en = ["dodd-frank"];
+var words = [
+	"addendum",
+	"agent",
+	"agreement",
+	"alameda",
+	"amendment",
+	"annex",
+	"april",
+	"article",
+	"august",
+	"austria",
+	"austrian",
+	"avenida",
+	"basic",
+	"belgian",
+	"borrower",
+	"branch",
+	"british",
+	"bulgarian",
+	"buyer",
+	"california",
+	"center",
+	"consent",
+	"consultant",
+	"contract",
+	"contractor",
+	"croatian",
+	"customer",
+	"cypriot",
+	"czech",
+	"danish",
+	"december",
+	"default",
+	"delaware",
+	"delivery",
+	"department",
+	"division",
+	"dry",
+	"dutch",
+	"english",
+	"estonian",
+	"estrada",
+	"european",
+	"event",
+	"exhibit",
+	"february",
+	"finnish",
+	"french",
+	"german",
+	"germany",
+	"greek",
+	"guarantor",
+	"headquarters",
+	"hungarian",
+	"indiana",
+	"interest",
+	"irish",
+	"israel",
+	"italian",
+	"january",
+	"july",
+	"june",
+	"landlord",
+	"largo",
+	"latvian",
+	"lender",
+	"lessee",
+	"lessor",
+	"license",
+	"licensee",
+	"licensor",
+	"lithuanian",
+	"louisiana",
+	"luxembourgish",
+	"maltese",
+	"march",
+	"maryland",
+	"memorandum",
+	"missouri",
+	"nevada",
+	"notice",
+	"november",
+	"october",
+	"office",
+	"ohio",
+	"opportunity",
+	"owner",
+	"page",
+	"party",
+	"payment",
+	"polish",
+	"portuguese",
+	"praca",
+	"praça",
+	"price",
+	"principal",
+	"purchaser",
+	"rate",
+	"rodovia",
+	"romanian",
+	"rua",
+	"sale",
+	"schedule",
+	"section",
+	"seller",
+	"september",
+	"subsidiary",
+	"slovak",
+	"slovenian",
+	"spanish",
+	"swedish",
+	"tenant",
+	"term",
+	"termination",
+	"time",
+	"travessa",
+	"union",
+	"university",
+	"vendor",
+	"waiver",
+	"borgo",
+	"contrada",
+	"corso",
+	"indirizzo",
+	"lungarno",
+	"lungomare",
+	"lungotevere",
+	"piazza",
+	"piazzale",
+	"salita",
+	"sede",
+	"sezione",
+	"strada",
+	"stradone",
+	"traversa",
+	"via",
+	"viale",
+	"vico",
+	"vicolo",
+	"tato",
+	"tento",
+	"tito",
+	"tomto",
+	"tomuto",
+	"toto",
+	"touto",
+	"tuto",
+	"tyto",
+	"této",
+	"acheteur",
+	"acheteuse",
+	"acquéreur",
+	"acquéreuse",
+	"allée",
+	"annexe",
+	"avenue",
+	"bailleresse",
+	"bailleur",
+	"boulevard",
+	"cédant",
+	"cédante",
+	"cessionnaire",
+	"chapitre",
+	"chemin",
+	"domicilié",
+	"domiciliée",
+	"employé",
+	"employée",
+	"employeur",
+	"employeuse",
+	"fournisseur",
+	"fournisseuse",
+	"impasse",
+	"locataire",
+	"maître",
+	"mandant",
+	"mandante",
+	"mandataire",
+	"passage",
+	"préambule",
+	"preneur",
+	"preneuse",
+	"prestataire",
+	"prêteur",
+	"prêteuse",
+	"promesse",
+	"quai",
+	"résidence",
+	"ruelle",
+	"salarié",
+	"salariée",
+	"société",
+	"titre",
+	"vendeur",
+	"vendeuse",
+	"voie"
+];
 var person_stopwords_default = {
-	_comment: "Words that are valid in other labels (address, org) but should never be classified as person. Checked only in person chain scoring.",
-	words: [
-		"addendum",
-		"agent",
-		"agreement",
-		"alameda",
-		"amendment",
-		"annex",
-		"april",
-		"article",
-		"august",
-		"austria",
-		"austrian",
-		"avenida",
-		"basic",
-		"belgian",
-		"borrower",
-		"branch",
-		"british",
-		"bulgarian",
-		"buyer",
-		"california",
-		"center",
-		"consent",
-		"consultant",
-		"contract",
-		"contractor",
-		"croatian",
-		"customer",
-		"cypriot",
-		"czech",
-		"danish",
-		"december",
-		"default",
-		"delaware",
-		"delivery",
-		"department",
-		"division",
-		"dry",
-		"dutch",
-		"english",
-		"estonian",
-		"estrada",
-		"european",
-		"event",
-		"exhibit",
-		"february",
-		"finnish",
-		"french",
-		"german",
-		"germany",
-		"greek",
-		"guarantor",
-		"headquarters",
-		"hungarian",
-		"indiana",
-		"interest",
-		"irish",
-		"israel",
-		"italian",
-		"january",
-		"july",
-		"june",
-		"landlord",
-		"largo",
-		"latvian",
-		"lender",
-		"lessee",
-		"lessor",
-		"license",
-		"licensee",
-		"licensor",
-		"lithuanian",
-		"louisiana",
-		"luxembourgish",
-		"maltese",
-		"march",
-		"maryland",
-		"memorandum",
-		"missouri",
-		"nevada",
-		"notice",
-		"november",
-		"october",
-		"office",
-		"ohio",
-		"opportunity",
-		"owner",
-		"page",
-		"party",
-		"payment",
-		"polish",
-		"portuguese",
-		"praca",
-		"praça",
-		"price",
-		"principal",
-		"purchaser",
-		"rate",
-		"rodovia",
-		"romanian",
-		"rua",
-		"sale",
-		"schedule",
-		"section",
-		"seller",
-		"september",
-		"subsidiary",
-		"slovak",
-		"slovenian",
-		"spanish",
-		"swedish",
-		"tenant",
-		"term",
-		"termination",
-		"time",
-		"travessa",
-		"union",
-		"university",
-		"vendor",
-		"waiver",
-		"borgo",
-		"contrada",
-		"corso",
-		"indirizzo",
-		"lungarno",
-		"lungomare",
-		"lungotevere",
-		"piazza",
-		"piazzale",
-		"salita",
-		"sede",
-		"sezione",
-		"strada",
-		"stradone",
-		"traversa",
-		"via",
-		"viale",
-		"vico",
-		"vicolo",
-		"tato",
-		"tento",
-		"tito",
-		"tomto",
-		"tomuto",
-		"toto",
-		"touto",
-		"tuto",
-		"tyto",
-		"této",
-		"acheteur",
-		"acheteuse",
-		"acquéreur",
-		"acquéreuse",
-		"allée",
-		"annexe",
-		"avenue",
-		"bailleresse",
-		"bailleur",
-		"boulevard",
-		"cédant",
-		"cédante",
-		"cessionnaire",
-		"chapitre",
-		"chemin",
-		"domicilié",
-		"domiciliée",
-		"employé",
-		"employée",
-		"employeur",
-		"employeuse",
-		"fournisseur",
-		"fournisseuse",
-		"impasse",
-		"locataire",
-		"maître",
-		"mandant",
-		"mandante",
-		"mandataire",
-		"passage",
-		"préambule",
-		"preneur",
-		"preneuse",
-		"prestataire",
-		"prêteur",
-		"prêteuse",
-		"promesse",
-		"quai",
-		"résidence",
-		"ruelle",
-		"salarié",
-		"salariée",
-		"société",
-		"titre",
-		"vendeur",
-		"vendeuse",
-		"voie"
-	]
+	_comment,
+	cs,
+	en,
+	words
 };
 //#endregion
-export { person_stopwords_default as default };
+export { _comment, cs, person_stopwords_default as default, en, words };
 
 //# sourceMappingURL=person-stopwords.mjs.map

package/dist/signing-clauses.mjs

@@ -1,5 +1,5 @@
 //#region src/data/signing-clauses.json
-var _comment = "Signing clause patterns. Captures the place name from contract signing locations. Each entry: prefix (before place), suffix (after place), prepositions (allowed inside multi-word place names).";
+var _comment = "Signing clause patterns. Captures the place name from contract signing locations. Each entry: prefix/suffix build regexes; prefilterPhrases gate native regex scans; guardPrefixPhrases/guardSuffixPhrases suppress deny-list place hits in the same signing context.";
 var patterns = [
 	{
 		"lang": "cs",
@@ -11,7 +11,10 @@ var patterns = [
 			"u",
 			"ve",
 			"na"
-		]
+		],
+		"prefilterPhrases": ["dne"],
+		"guardPrefixPhrases": ["v", "ve"],
+		"guardSuffixPhrases": ["dne"]
 	},
 	{
 		"lang": "sk",
@@ -21,7 +24,10 @@ var patterns = [
 			"nad",
 			"pod",
 			"pri"
-		]
+		],
+		"prefilterPhrases": ["dňa"],
+		"guardPrefixPhrases": ["v", "vo"],
+		"guardSuffixPhrases": ["dňa"]
 	},
 	{
 		"lang": "de",
@@ -31,19 +37,28 @@ var patterns = [
 			"am",
 			"an",
 			"im"
-		]
+		],
+		"prefilterPhrases": ["den"],
+		"guardPrefixPhrases": [""],
+		"guardSuffixPhrases": ["den"]
 	},
 	{
 		"lang": "fr",
 		"prefix": "(?:Fait\\s+)?[Àà]\\s+",
 		"suffix": "\\s*,?\\s*le",
-		"prepositions": []
+		"prepositions": [],
+		"prefilterPhrases": ["À ", "à "],
+		"guardPrefixPhrases": ["à", "fait à"],
+		"guardSuffixPhrases": ["le"]
 	},
 	{
 		"lang": "en",
 		"prefix": "(?:Signed|Executed)\\s+in\\s+",
 		"suffix": "",
-		"prepositions": []
+		"prepositions": [],
+		"prefilterPhrases": ["Signed in", "Executed in"],
+		"guardPrefixPhrases": ["signed in", "executed in"],
+		"guardSuffixPhrases": [""]
 	},
 	{
 		"lang": "pl",
@@ -53,19 +68,28 @@ var patterns = [
 			"nad",
 			"pod",
 			"przy"
-		]
+		],
+		"prefilterPhrases": ["dnia"],
+		"guardPrefixPhrases": ["w", "we"],
+		"guardSuffixPhrases": ["dnia"]
 	},
 	{
 		"lang": "it",
 		"prefix": "(?:Fatto\\s+)?[Aa]\\s+",
 		"suffix": "\\s*,?\\s*(?:il|lì)",
-		"prepositions": []
+		"prepositions": [],
+		"prefilterPhrases": ["A ", "a "],
+		"guardPrefixPhrases": ["a", "fatto a"],
+		"guardSuffixPhrases": ["il", "lì"]
 	},
 	{
 		"lang": "es",
 		"prefix": "(?:Firmado\\s+)?[Ee]n\\s+",
 		"suffix": "\\s*,?\\s*(?:a|el)",
-		"prepositions": ["de", "del"]
+		"prepositions": ["de", "del"],
+		"prefilterPhrases": ["En ", "en "],
+		"guardPrefixPhrases": ["en", "firmado en"],
+		"guardSuffixPhrases": ["a", "el"]
 	}
 ];
 var signing_clauses_default = {

package/index.cjs

@@ -0,0 +1,3 @@
+"use strict";
+
+module.exports = require("./stella_anonymize_napi.node");

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@stll/anonymize",
-  "version": "1.5.0",
+  "version": "2.0.0-alpha.1",
   "description": "Multi-layer PII detection and anonymization pipeline with regex, NER, deny lists, and coreference resolution",
   "type": "module",
   "exports": {
@@ -13,11 +13,29 @@
       "types": "./dist/constants.d.mts",
       "import": "./dist/constants.mjs",
       "default": "./dist/constants.mjs"
+    },
+    "./native": {
+      "types": "./dist/native.d.mts",
+      "import": "./dist/native.mjs",
+      "default": "./dist/native.mjs"
+    },
+    "./native-node": {
+      "types": "./dist/native-node.d.mts",
+      "import": "./dist/native-node.mjs",
+      "default": "./dist/native-node.mjs"
     }
   },
   "types": "dist/index.d.mts",
+  "bin": {
+    "stella-anonymize-build-native-package": "./scripts/build-native-pipeline-package.mjs"
+  },
   "files": [
-    "dist"
+    "ATTRIBUTION.md",
+    "dist",
+    "index.cjs",
+    "native-pipeline.stlanonpkg",
+    "native-pipeline.*.stlanonpkg",
+    "scripts/build-native-pipeline-package.mjs"
   ],
   "publishConfig": {
     "access": "public"
@@ -28,19 +46,27 @@
   },
   "license": "MIT",
   "scripts": {
-    "build": "tsdown",
+    "build": "tsdown && bun scripts/build-native-node.mjs",
     "prepublishOnly": "bun run build",
     "typecheck": "tsc --noEmit -p tsconfig.json && tsc --noEmit -p tsconfig.test.json && tsc --noEmit -p tsconfig.wasm.json",
     "test": "bun test --preload ./src/__test__/setup.ts --timeout 15000",
     "test:fast": "bun run test src/__test__/*.test.ts",
+    "test:native-fixtures": "ANONYMIZE_TEST_SLOW_NATIVE_FIXTURE_PARITY=1 bun test --preload ./src/__test__/setup.ts --timeout 600000 src/__test__/native-adapter-parity.test.ts",
     "perf:contracts": "bun scripts/contract-perf.mjs",
+    "perf:migration-fixtures": "bun scripts/migration-fixture-perf.mjs",
+    "perf:native-adapters": "bun scripts/native-adapter-perf.mjs",
+    "perf:native-default": "bun scripts/native-default-sdk-perf.mjs",
+    "perf:native-default:summary": "ANONYMIZE_NATIVE_DEFAULT_SDK_PERF_OUTPUT=summary bun scripts/native-default-sdk-perf.mjs",
+    "perf:native-package": "bun scripts/native-package-ux-perf.mjs",
+    "check:v1-native": "bun scripts/native-v1-readiness.mjs",
+    "check:v1-native:full": "ANONYMIZE_V1_READINESS_MODE=full bun scripts/native-v1-readiness.mjs",
     "smoke:dist": "bun scripts/dist-smoke.mjs",
     "format": "oxfmt ."
   },
   "dependencies": {
     "@huggingface/tokenizers": "^0.1.3",
     "@stll/stdnum": "^2.1.1",
-    "@stll/text-search": "^1.0.6"
+    "@stll/text-search": "^1.0.7"
   },
   "peerDependencies": {
     "@stll/anonymize-data": "^0.0.6"
@@ -50,11 +76,19 @@
       "optional": true
     }
   },
+  "optionalDependencies": {
+    "@stll/anonymize-darwin-arm64": "2.0.0-alpha.1",
+    "@stll/anonymize-darwin-x64": "2.0.0-alpha.1",
+    "@stll/anonymize-linux-arm64-gnu": "2.0.0-alpha.1",
+    "@stll/anonymize-linux-x64-gnu": "2.0.0-alpha.1",
+    "@stll/anonymize-win32-x64-msvc": "2.0.0-alpha.1"
+  },
   "devDependencies": {
     "@stll/anonymize-data": "workspace:*",
-    "@stll/text-search-wasm": "^1.0.6",
+    "@stll/text-search-wasm": "^1.0.7",
     "bun-types": "^1.3.14",
-    "tsdown": "^0.22.2",
+    "fast-check": "^4.8.0",
+    "tsdown": "^0.22.3",
     "typescript": "^6.0.3",
     "vite": "^8.0.16"
   }