@stll/anonymize-wasm 1.1.1 → 1.2.0

package/dist/wasm.mjs

@@ -2021,19 +2021,21 @@ const processRegexMatches = (allMatches, sliceStart, sliceEnd, meta_) => {
 		if (idx < sliceStart || idx >= sliceEnd) continue;
 		const meta = meta_[idx - sliceStart];
 		if (!meta) continue;
-		if (meta.label === "phone number" && match.text.length < MIN_PHONE_LENGTH) continue;
+		if (meta.sourceDetail !== "custom-regex" && meta.label === "phone number" && match.text.length < MIN_PHONE_LENGTH) continue;
 		if (meta.validator) {
 			const compacted = meta.validator.compact(match.text);
 			if (!meta.validator.validate(compacted).valid) continue;
 		}
-		results.push({
+		const entity = {
 			start: match.start,
 			end: match.end,
 			label: meta.label,
 			text: match.text,
 			score: meta.score,
 			source: DETECTION_SOURCES.REGEX
-		});
+		};
+		if (meta.sourceDetail) entity.sourceDetail = meta.sourceDetail;
+		results.push(entity);
 	}
 	return results;
 };
@@ -3055,6 +3057,7 @@ const loadGenericRoles = (ctx = defaultContext) => {
 };
 /** Sync accessor — returns empty set before init. */
 const getGenericRoles = (ctx) => ctx.genericRoles ?? EMPTY_GENERIC_ROLES;
+const isCallerOwnedEntity$3 = (entity) => entity.sourceDetail === "custom-deny-list" || entity.sourceDetail === "custom-regex";
 /**
 * Filter out entities that are likely false positives:
 * template placeholders, clause/section numbers,
@@ -3067,9 +3070,17 @@ const filterFalsePositives = (entities, ctx = defaultContext, fullText) => {
 	const filtered = [];
 	const roles = getGenericRoles(ctx);
 	for (const entity of entities) {
+		if (isCallerOwnedEntity$3(entity)) {
+			filtered.push(entity);
+			continue;
+		}
 		const normalized = normalizeEntity(entity);
 		if (!normalized) continue;
 		const trimmed = normalized.text;
+		if (isCallerOwnedEntity$3(normalized)) {
+			filtered.push(normalized);
+			continue;
+		}
 		if (TEMPLATE_PLACEHOLDER_RE.test(trimmed)) continue;
 		const maxLen = MAX_ENTITY_LENGTH[normalized.label];
 		if (maxLen && trimmed.length > maxLen && normalized.source !== "legal-form") continue;
@@ -3244,6 +3255,7 @@ const EMPTY_PERSON_STOPWORDS = /* @__PURE__ */ new Set();
 /** Sync accessor — returns empty set before init. */
 const getPersonStopwords = (ctx) => ctx.personStopwords ?? EMPTY_PERSON_STOPWORDS;
 const PERSON_CHAIN_BREAK_RE = /[!?;:]|,/u;
+const WORD_CHAR_RE$1 = /[\p{L}\p{N}]/u;
 const isInitialContinuationGap = (text, gap) => /^\p{Lu}$/u.test(text) && /^\.[^\S\n]{1,2}$/u.test(gap) || /^[^\S\n]{1,2}(?:\p{Lu}\.[^\S\n]{1,2})+$/u.test(gap);
 /**
 * Resolve which dictionaries to load based on country
@@ -3266,27 +3278,31 @@ const buildDenyList = async (config, ctx = defaultContext) => {
 	const dictionaries = config.dictionaries;
 	const hasDenyList = dictionaries?.denyList && dictionaries?.denyListMeta;
 	const hasCities = dictionaries?.cities && dictionaries.cities.length > 0;
-	if (!hasDenyList && !hasCities) return buildNameCorpusOnly(config, ctx);
+	const hasCustomDenyList = config.customDenyList !== void 0 && config.customDenyList.length > 0;
+	if (!hasDenyList && !hasCities && !hasCustomDenyList) return buildNameCorpusOnly(config, ctx);
 	const allowedCountries = resolveCountries(config.denyListRegions, config.denyListCountries);
 	const excluded = config.denyListExcludeCategories;
 	const excludeCategories = excluded ? new Set(excluded) : /* @__PURE__ */ new Set();
 	const patternList = [];
 	const labelList = [];
+	const customLabelList = [];
 	const sourceList = [];
 	const patternIndex = /* @__PURE__ */ new Map();
-	const addDenyListEntry = (entry, label) => {
-		const normalized = normalizeForSearch(entry).replace(/[|\\]/g, "");
+	const addDenyListEntry = (entry, label, source = "deny-list") => {
+		const normalized = source === "custom-deny-list" ? normalizeForSearch(entry) : normalizeForSearch(entry).replace(/[|\\]/g, "");
 		if (normalized.length === 0) return;
 		const lower = normalized.toLowerCase();
 		const existing = patternIndex.get(lower);
 		if (existing !== void 0) {
 			if (!labelList[existing].includes(label)) labelList[existing].push(label);
-			if (!sourceList[existing].includes("deny-list")) sourceList[existing].push("deny-list");
+			if (!sourceList[existing].includes(source)) sourceList[existing].push(source);
+			if (source === "custom-deny-list" && !customLabelList[existing].includes(label)) customLabelList[existing].push(label);
 		} else {
 			patternIndex.set(lower, patternList.length);
 			patternList.push(normalized);
 			labelList.push([label]);
-			sourceList.push(["deny-list"]);
+			customLabelList.push(source === "custom-deny-list" ? [label] : []);
+			sourceList.push([source]);
 		}
 	};
 	if (hasDenyList) {
@@ -3304,10 +3320,15 @@ const buildDenyList = async (config, ctx = defaultContext) => {
 		}
 	}
 	if (hasCities && !excludeCategories.has("Places")) for (const entry of dictionaries.cities) addDenyListEntry(entry, "address");
+	if (hasCustomDenyList) for (const entry of config.customDenyList) {
+		addDenyListEntry(entry.value, entry.label, "custom-deny-list");
+		for (const variant of entry.variants ?? []) addDenyListEntry(variant, entry.label, "custom-deny-list");
+	}
 	appendNameCorpusEntries(config, ctx, patternList, labelList, sourceList, patternIndex);
 	if (patternList.length === 0) return null;
 	return {
 		labels: labelList,
+		customLabels: customLabelList,
 		originals: patternList,
 		sources: sourceList
 	};
@@ -3324,11 +3345,13 @@ const buildNameCorpusOnly = (config, ctx) => {
 	if ((excluded ? new Set(excluded) : /* @__PURE__ */ new Set()).has("Names")) return null;
 	const patternList = [];
 	const labelList = [];
+	const customLabelList = [];
 	const sourceList = [];
 	appendNameCorpusEntries(config, ctx, patternList, labelList, sourceList, /* @__PURE__ */ new Map());
 	if (patternList.length === 0) return null;
 	return {
 		labels: labelList,
+		customLabels: customLabelList,
 		originals: patternList,
 		sources: sourceList
 	};
@@ -3374,6 +3397,14 @@ const appendNameCorpusEntries = (config, ctx, patternList, labelList, sourceList
 		}
 	}
 };
+const customMatchHasValidEdges = (fullText, start, end, pattern) => {
+	if (!WORD_CHAR_RE$1.test(pattern)) return true;
+	const prev = fullText[start - 1] ?? "";
+	const next = fullText[end] ?? "";
+	if (WORD_CHAR_RE$1.test(prev)) return false;
+	if (WORD_CHAR_RE$1.test(next)) return false;
+	return true;
+};
 /**
 * Ensure all deny-list support data (stopwords, allow
 * list, person stopwords, generic roles) is loaded on
@@ -3411,18 +3442,24 @@ const processDenyListMatches = (allMatches, sliceStart, sliceEnd, fullText, data
 		const idx = match.pattern;
 		if (idx < sliceStart || idx >= sliceEnd) continue;
 		const localIdx = idx - sliceStart;
-		const sourceChar = fullText[match.start] ?? "";
-		if (!UPPER_START_RE.test(sourceChar)) continue;
+		const sources = data.sources[localIdx] ?? [];
 		const matchText = fullText.slice(match.start, match.end);
+		const sourceChar = fullText[match.start] ?? "";
 		const keyword = matchText.toLowerCase();
-		if (getStopwords(ctx).has(keyword) || getAllowList(ctx).has(keyword)) continue;
-		if (ALL_UPPER_RE.test(matchText)) continue;
 		const labels = data.labels[localIdx];
-		if (!labels || labels.length === 0) continue;
+		const pattern = data.originals[localIdx] ?? "";
+		const customPatternLabels = data.customLabels[localIdx] ?? [];
+		const customEdgesAreValid = customMatchHasValidEdges(fullText, match.start, match.end, pattern);
+		const customLabels = customEdgesAreValid ? customPatternLabels : [];
+		if ((!labels || labels.length === 0) && customLabels.length === 0) continue;
+		const curatedLabels = UPPER_START_RE.test(sourceChar) && !getStopwords(ctx).has(keyword) && !getAllowList(ctx).has(keyword) && !ALL_UPPER_RE.test(matchText) ? (labels ?? []).filter((label) => !customPatternLabels.includes(label) && customEdgesAreValid) : [];
+		if (curatedLabels.length === 0 && customLabels.length === 0) continue;
 		const entry = {
 			start: match.start,
 			end: match.end,
-			labels,
+			labels: curatedLabels,
+			customLabels,
+			sources,
 			text: matchText,
 			patternIdx: localIdx
 		};
@@ -3433,22 +3470,31 @@ const processDenyListMatches = (allMatches, sliceStart, sliceEnd, fullText, data
 	const results = [];
 	const nameHits = [];
 	for (const [, matches] of Array.from(matchesByPattern)) {
-		const first = matches[0];
-		if (!first) continue;
-		const hasPerson = first.labels.includes("person");
-		const nonPersonLabels = first.labels.filter((l) => l !== "person");
-		if (hasPerson) {
-			const keyword = first.text.toLowerCase();
-			if (!getPersonStopwords(ctx).has(keyword)) for (const m of matches) nameHits.push(m);
-		}
-		for (const m of matches) for (const label of nonPersonLabels) results.push({
+		if (!matches[0]) continue;
+		for (const m of matches) for (const label of m.customLabels) results.push({
 			start: m.start,
 			end: m.end,
 			label,
 			text: m.text,
 			score: .9,
-			source: DETECTION_SOURCES.DENY_LIST
+			source: DETECTION_SOURCES.DENY_LIST,
+			sourceDetail: "custom-deny-list"
 		});
+		for (const m of matches) {
+			if (m.labels.includes("person")) {
+				const keyword = m.text.toLowerCase();
+				if (!getPersonStopwords(ctx).has(keyword)) nameHits.push(m);
+			}
+			const nonPersonLabels = m.labels.filter((l) => l !== "person");
+			for (const label of nonPersonLabels) results.push({
+				start: m.start,
+				end: m.end,
+				label,
+				text: m.text,
+				score: .9,
+				source: DETECTION_SOURCES.DENY_LIST
+			});
+		}
 	}
 	nameHits.sort((a, b) => a.start - b.start);
 	const nameConsumed = /* @__PURE__ */ new Set();
@@ -3528,6 +3574,7 @@ const TRAILING_WORD_EXCLUSIONS = new Set([
 const extendCityDistricts = (entities, fullText) => {
 	for (const entity of entities) {
 		if (entity.label !== "address") continue;
+		if (entity.sourceDetail === "custom-deny-list") continue;
 		const afterMatch = fullText.slice(entity.end);
 		const suffixM = DISTRICT_SUFFIX_RE.exec(afterMatch);
 		if (suffixM) {
@@ -3646,6 +3693,7 @@ const collectSeeds = (allMatches, sliceStart, sliceEnd, fullText, existingEntiti
 	}
 	for (const e of existingEntities) {
 		if (e.label !== "address") continue;
+		if (e.sourceDetail === "custom-deny-list") continue;
 		if (e.source === "deny-list") seeds.push({
 			type: "city",
 			start: e.start,
@@ -3812,6 +3860,7 @@ const processAddressSeeds = async (allMatches, sliceStart, sliceEnd, fullText, e
 const TRAILING_SEP = /[,\s]+$/;
 const WORD_CHAR_RE = /[\p{L}\p{N}]/u;
 const ORG_PROPAGATION_SCORE = .9;
+const isCallerOwnedEntity$2 = (entity) => entity.sourceDetail === "custom-deny-list" || entity.sourceDetail === "custom-regex";
 /**
 * After the main detection pass, collect organization
 * entities with a legal form suffix, strip the suffix
@@ -3824,6 +3873,7 @@ const propagateOrgNames = (entities, fullText) => {
 	const seenBases = /* @__PURE__ */ new Set();
 	for (const e of entities) {
 		if (e.label !== "organization") continue;
+		if (isCallerOwnedEntity$2(e)) continue;
 		for (const suffix of LEGAL_SUFFIXES) if (e.text.endsWith(suffix)) {
 			const base = e.text.slice(0, -suffix.length).replace(TRAILING_SEP, "").trim();
 			if (base.length >= 3 && !seenBases.has(base)) {
@@ -3903,6 +3953,7 @@ const NEAR_MISS_BAND = .15;
 const BOOST_PER_NEIGHBOUR = .05;
 const CONTEXT_WINDOW_CHARS = 150;
 const HIGH_CONFIDENCE_FLOOR = .9;
+const isCallerOwnedEntity$1 = (entity) => entity.sourceDetail === "custom-deny-list" || entity.sourceDetail === "custom-regex";
 /**
 * Boost confidence of near-miss NER entities that appear
 * near high-confidence detections (regex, trigger phrase).
@@ -4011,7 +4062,7 @@ const getStreetAbbrevs = () => _streetAbbrevs ?? /* @__PURE__ */ new Set();
 */
 const detectStreetPatternsNearAddresses = (fullText, existingEntities) => {
 	const results = [];
-	const addressEntities = existingEntities.filter((e) => e.label === "address");
+	const addressEntities = existingEntities.filter((entity) => !(entity.label === "address" && isCallerOwnedEntity$1(entity))).filter((e) => e.label === "address");
 	const headerEnd = Math.floor(fullText.length * HEADER_ZONE_FRACTION);
 	const houseNumRe = /\b(?:\d{1,4}\/\d+[a-zA-Z]\b|\d{3,4}\/\d+\b|(?:1[3-9]|[2-9]\d)\/\d{3,}\b)/g;
 	houseNumRe.lastIndex = 0;
@@ -4107,6 +4158,7 @@ const ORPHAN_STREET_RE = /^\s*(\p{Lu}[\p{Ll}\p{Lu}]+(?:\s+[\p{Lu}\p{Ll}][\p{Ll}]
 */
 const detectOrphanStreetLines = (fullText, existingEntities) => {
 	const headerEnd = Math.floor(fullText.length * HEADER_ZONE_FRACTION);
+	const contextEntities = existingEntities.filter((entity) => !(entity.label === "address" && isCallerOwnedEntity$1(entity)));
 	const results = [];
 	ORPHAN_STREET_RE.lastIndex = 0;
 	for (let m = ORPHAN_STREET_RE.exec(fullText); m !== null; m = ORPHAN_STREET_RE.exec(fullText)) {
@@ -4116,7 +4168,7 @@ const detectOrphanStreetLines = (fullText, existingEntities) => {
 		const end = start + captured.length;
 		if (start >= headerEnd) continue;
 		if (existingEntities.some((e) => e.start <= start && e.end >= end)) continue;
-		if (!existingEntities.some((e) => Math.abs(e.start - end) < 200 || Math.abs(e.end - start) < 200)) continue;
+		if (!contextEntities.some((e) => Math.abs(e.start - end) < 200 || Math.abs(e.end - start) < 200)) continue;
 		results.push({
 			start,
 			end,
@@ -4464,6 +4516,7 @@ const applyHotwordRules = (entities, fullText) => {
 //#region src/filters/boundary-consistency.ts
 /** Max gap (in chars) between entities to merge. */
 const MAX_GAP = 3;
+const hasLockedBoundary$1 = (entity) => entity.sourceDetail === "custom-deny-list" || entity.sourceDetail === "custom-regex";
 /**
 * Characters allowed in the gap between two adjacent
 * same-label entities that should be merged: spaces,
@@ -4579,6 +4632,11 @@ const mergeAdjacent = (entities, fullText) => {
 	const result = [];
 	const lastByLabel = /* @__PURE__ */ new Map();
 	for (const entity of sorted) {
+		if (hasLockedBoundary$1(entity)) {
+			const copy = { ...entity };
+			result.push(copy);
+			continue;
+		}
 		const prev = lastByLabel.get(entity.label);
 		if (!prev) {
 			const copy = { ...entity };
@@ -4586,7 +4644,7 @@ const mergeAdjacent = (entities, fullText) => {
 			lastByLabel.set(entity.label, copy);
 			continue;
 		}
-		if (entity.start < prev.end) {
+		if (!hasLockedBoundary$1(prev) && entity.start < prev.end) {
 			prev.end = Math.max(prev.end, entity.end);
 			prev.text = fullText.slice(prev.start, prev.end);
 			prev.score = Math.max(prev.score, entity.score);
@@ -4605,7 +4663,7 @@ const mergeAdjacent = (entities, fullText) => {
 				break;
 			}
 		}
-		if (!gapOccupied && gap.length <= MAX_GAP && GAP_PATTERN.test(gap)) {
+		if (!hasLockedBoundary$1(prev) && !gapOccupied && gap.length <= MAX_GAP && GAP_PATTERN.test(gap)) {
 			prev.end = entity.end;
 			prev.text = fullText.slice(prev.start, prev.end);
 			prev.score = Math.max(prev.score, entity.score);
@@ -4637,6 +4695,7 @@ const fixPartialWords = (entities, fullText) => {
 	})).sort((a, b) => a.entity.end - b.entity.end);
 	const endPositions = byEnd.map((x) => x.entity.end);
 	return sorted.map((e, eIdx) => {
+		if (hasLockedBoundary$1(e)) return e;
 		let newStart = wordStartAt(e.start, boundaries, fullText);
 		let newEnd = wordEndAt(e.end, boundaries, fullText);
 		let lo = 0;
@@ -4731,7 +4790,8 @@ const resolveCrossLabelOverlaps = (entities, fullText) => {
 		if (aContainsB || bContainsA) continue;
 		const aLen = a.end - a.start;
 		const bLen = b.end - b.start;
-		if (a.score > b.score || a.score === b.score && aLen >= bLen) {
+		const aLocked = hasLockedBoundary$1(a);
+		if (aLocked !== hasLockedBoundary$1(b) ? aLocked : a.score > b.score || a.score === b.score && aLen >= bLen) {
 			b.start = a.end;
 			b.text = fullText.slice(b.start, b.end);
 		} else {
@@ -4759,8 +4819,11 @@ const enforceBoundaryConsistency = (entities, fullText) => {
 };
 //#endregion
 //#region src/build-unified-search.ts
+const DEFAULT_CUSTOM_REGEX_SCORE$1 = .9;
+const ALNUM_RE = /[\p{L}\p{N}]/u;
 const buildUnifiedSearch = async (config, gazetteerEntries = [], ctx = defaultContext) => {
 	const legalFormsEnabled = isLegalFormsEnabled(config);
+	const customRegexes = config.enableRegex ? config.customRegexes ?? [] : [];
 	const [legalForms, triggers, denyListData, streetTypes, currencyPatterns, datePatterns, signingPatterns] = await Promise.all([
 		legalFormsEnabled ? buildLegalFormPatterns() : Promise.resolve([]),
 		config.enableTriggerPhrases ? buildTriggerPatterns() : Promise.resolve({
@@ -4785,12 +4848,21 @@ const buildUnifiedSearch = async (config, gazetteerEntries = [], ctx = defaultCo
 		...datePatterns.map(() => DATE_PATTERN_META),
 		...signingPatterns.map(() => SIGNING_CLAUSE_META)
 	];
+	const customRegexMeta = customRegexes.map((entry) => ({
+		label: entry.label,
+		score: entry.score ?? DEFAULT_CUSTOM_REGEX_SCORE$1,
+		sourceDetail: "custom-regex"
+	}));
 	let offset = 0;
 	const regexSlice = {
 		start: offset,
 		end: offset + allRegex.length
 	};
 	offset = regexSlice.end;
+	const customRegexSlice = {
+		start: 0,
+		end: customRegexes.length
+	};
 	const legalFormsSlice = {
 		start: offset,
 		end: offset + legalForms.length
@@ -4811,6 +4883,7 @@ const buildUnifiedSearch = async (config, gazetteerEntries = [], ctx = defaultCo
 		...triggerEntries
 	];
 	const tsRegex = new (getTextSearch())(regexAllPatterns);
+	const tsCustomRegex = new (getTextSearch())(customRegexes.map((entry) => entry.pattern), { overlapStrategy: "all" });
 	offset = 0;
 	const denyListOriginals = denyListData?.originals ?? [];
 	const denyListSlice = {
@@ -4829,24 +4902,31 @@ const buildUnifiedSearch = async (config, gazetteerEntries = [], ctx = defaultCo
 		end: offset + (gazResult?.patterns.length ?? 0)
 	};
 	offset = gazetteerSlice.end;
-	const wrapWholeWord = (s) => ({
+	const wrapWholeWord = (s, wholeWords) => ({
 		pattern: s,
 		literal: true,
-		wholeWords: true
+		wholeWords
 	});
+	const customDenyListNeedsWholeWords = (pattern) => {
+		const first = pattern.at(0) ?? "";
+		const last = pattern.at(-1) ?? "";
+		return ALNUM_RE.test(first) && ALNUM_RE.test(last);
+	};
 	const literalAllPatterns = [
-		...denyListOriginals.map(wrapWholeWord),
-		...streetTypes.map(wrapWholeWord),
+		...denyListOriginals.map((pattern, index) => wrapWholeWord(pattern, (denyListData?.sources[index] ?? []).includes("custom-deny-list") ? customDenyListNeedsWholeWords(pattern) : true)),
+		...streetTypes.map((pattern) => wrapWholeWord(pattern, true)),
 		...gazResult?.patterns ?? []
 	];
 	return {
 		tsRegex,
+		tsCustomRegex,
 		tsLiterals: literalAllPatterns.length > 0 ? new (getTextSearch())(literalAllPatterns, {
 			caseInsensitive: true,
 			overlapStrategy: "all"
 		}) : new (getTextSearch())([]),
 		slices: {
 			regex: regexSlice,
+			customRegex: customRegexSlice,
 			legalForms: legalFormsSlice,
 			triggers: triggersSlice,
 			denyList: denyListSlice,
@@ -4854,6 +4934,7 @@ const buildUnifiedSearch = async (config, gazetteerEntries = [], ctx = defaultCo
 			gazetteer: gazetteerSlice
 		},
 		regexMeta,
+		customRegexMeta,
 		triggerRules: triggers.rules,
 		denyListData,
 		gazetteerData: gazResult?.data ?? null
@@ -4863,9 +4944,11 @@ const buildUnifiedSearch = async (config, gazetteerEntries = [], ctx = defaultCo
 //#region src/unified-search.ts
 const runUnifiedSearch = (instance, fullText) => {
 	const regexMatches = instance.tsRegex.findIter(fullText);
+	const customRegexMatches = instance.tsCustomRegex.findIter(fullText);
 	const normalized = normalizeForSearch(fullText);
 	return {
 		regexMatches,
+		customRegexMatches,
 		literalMatches: instance.tsLiterals.findIter(normalized)
 	};
 };
@@ -4978,9 +5061,13 @@ const unmaskNerEntities = (nerEntities, maskResult, fullText) => {
 * so the containment rule trusts their length.
 */
 const LITERAL_SOURCES = new Set(["deny-list", "gazetteer"]);
+const isCallerOwnedEntity = (entity) => entity.sourceDetail === "custom-deny-list" || entity.sourceDetail === "custom-regex";
+const hasLockedBoundary = (entity) => isCallerOwnedEntity(entity);
 const shouldReplace = (a, b) => {
 	const aLen = a.end - a.start;
 	const bLen = b.end - b.start;
+	const aCallerOwned = isCallerOwnedEntity(a);
+	if (aCallerOwned !== isCallerOwnedEntity(b)) return aCallerOwned;
 	if (a.label === b.label && LITERAL_SOURCES.has(a.source) && a.start <= b.start && a.end >= b.end && aLen > bLen) return true;
 	if (a.label === b.label && LITERAL_SOURCES.has(b.source) && b.start <= a.start && b.end >= a.end && bLen > aLen) return false;
 	const aPri = DETECTOR_PRIORITY[a.source] ?? 0;
@@ -4992,6 +5079,7 @@ const shouldReplace = (a, b) => {
 const COLON_LABELS = new Set(["ip address", "mac address"]);
 /** Strip leading/trailing whitespace and punctuation. */
 const sanitizeEntities = (entities) => entities.flatMap((e) => {
+	if (hasLockedBoundary(e)) return [e];
 	const strip = COLON_LABELS.has(e.label) ? /[\s,;]+/ : /[\s:,;]+/;
 	const leadTrimmed = e.text.replace(/^(?:\.\s)+/, "").replace(new RegExp(`^${strip.source}`, strip.flags), "");
 	const lead = e.text.length - leadTrimmed.length;
@@ -5019,8 +5107,30 @@ const mergeAndDedup = (...layers) => {
 		const entity = sorted[i];
 		const last = merged[merged.length - 1];
 		if (!entity || !last) continue;
-		if (last.end <= entity.start) merged.push({ ...entity });
-		else if (shouldReplace(entity, last)) merged[merged.length - 1] = { ...entity };
+		if (last.end <= entity.start) {
+			merged.push({ ...entity });
+			continue;
+		}
+		const overlapStart = (() => {
+			for (let j = merged.length - 1; j >= 0; j--) {
+				const existing = merged[j];
+				if (!existing || existing.end <= entity.start) return j + 1;
+			}
+			return 0;
+		})();
+		const overlaps = merged.slice(overlapStart);
+		if (!overlaps.some((existing) => existing.start !== entity.start || existing.end !== entity.end)) {
+			const sameLabelIndex = overlaps.findIndex((existing) => existing.label === entity.label);
+			if (sameLabelIndex === -1) {
+				merged.push({ ...entity });
+				continue;
+			}
+			const actualIndex = overlapStart + sameLabelIndex;
+			const sameLabel = merged[actualIndex];
+			if (sameLabel && shouldReplace(entity, sameLabel)) merged[actualIndex] = { ...entity };
+			continue;
+		}
+		if (overlaps.every((existing) => shouldReplace(entity, existing))) merged.splice(overlapStart, overlaps.length, { ...entity });
 	}
 	return sanitizeEntities(merged);
 };
@@ -5039,7 +5149,7 @@ const getAmountWordsRe = async () => {
 	return amountWordsRe;
 };
 const extendMonetaryAmountWords = (entities, fullText, re) => entities.map((e) => {
-	if (e.label !== "monetary amount") return e;
+	if (e.label !== "monetary amount" || isCallerOwnedEntity(e)) return e;
 	const after = fullText.slice(e.end);
 	const m = re.exec(after);
 	if (!m) return e;
@@ -5052,6 +5162,7 @@ const extendMonetaryAmountWords = (entities, fullText, re) => entities.map((e) =
 });
 const createAllowedLabelSetFromLabels = (labels) => labels.length > 0 ? new Set(labels) : null;
 const createAllowedLabelSet = (config) => createAllowedLabelSetFromLabels(config.labels);
+const DEFAULT_CUSTOM_REGEX_SCORE = .9;
 const filterAllowedLabels = (entities, allowedLabels) => {
 	if (!allowedLabels) return entities;
 	return entities.filter((e) => allowedLabels.has(e.label));
@@ -5066,8 +5177,18 @@ const checkAbort = (signal) => {
 };
 const configKey = (config, gazetteerEntries) => {
 	const legalFormsEnabled = isLegalFormsEnabled(config);
+	const customDenyFingerprint = config.enableDenyList && config.customDenyList ? config.customDenyList.map((entry) => JSON.stringify({
+		label: entry.label,
+		value: entry.value,
+		variants: [...entry.variants ?? []].sort()
+	})).sort().join("\n") : "";
+	const customRegexFingerprint = config.enableRegex && config.customRegexes ? config.customRegexes.map((entry) => JSON.stringify({
+		label: entry.label,
+		pattern: entry.pattern,
+		score: entry.score ?? DEFAULT_CUSTOM_REGEX_SCORE
+	})).sort().join("\n") : "";
 	const gazFingerprint = config.enableGazetteer && gazetteerEntries.length > 0 ? gazetteerEntries.map((e) => `${e.id}:${e.canonical}:${e.label}:${[...e.variants].sort().join(",")}`).toSorted().join(";") : "";
-	return `${config.enableDenyList}:${config.enableTriggerPhrases}:${legalFormsEnabled}:${config.enableNameCorpus}:${config.denyListCountries?.toSorted().join(",") ?? ""}:${config.denyListRegions?.toSorted().join(",") ?? ""}:${config.denyListExcludeCategories?.toSorted().join(",") ?? ""}:${config.enableGazetteer}:${gazFingerprint}`;
+	return `${config.enableDenyList}:${config.enableTriggerPhrases}:${legalFormsEnabled}:${config.enableNameCorpus}:${config.enableRegex}:${config.denyListCountries?.toSorted().join(",") ?? ""}:${config.denyListRegions?.toSorted().join(",") ?? ""}:${config.denyListExcludeCategories?.toSorted().join(",") ?? ""}:${customDenyFingerprint}:${customRegexFingerprint}:${config.enableGazetteer}:${gazFingerprint}`;
 };
 /**
 * Get or build a cached search instance. Cache state
@@ -5150,10 +5271,11 @@ const runPipeline = async (options) => {
 	const preHotwordAllowedLabels = hotwordsActive ? createAllowedLabelSetFromLabels(expandLabelsForHotwordRules(config.labels)) : allowedLabels;
 	const search = cachedSearch ?? await getCachedSearch(config, gazetteerEntries, ctx);
 	checkAbort(signal);
-	const { regexMatches, literalMatches } = runUnifiedSearch(search, fullText);
+	const { regexMatches, customRegexMatches, literalMatches } = runUnifiedSearch(search, fullText);
 	const { slices } = search;
 	const rawRegexEntities = config.enableRegex ? processRegexMatches(regexMatches, slices.regex.start, slices.regex.end, search.regexMeta) : [];
-	const regexEntities = filterAllowedLabels(rawRegexEntities, preHotwordAllowedLabels);
+	const customRegexEntities = filterAllowedLabels(config.enableRegex ? processRegexMatches(customRegexMatches, slices.customRegex.start, slices.customRegex.end, search.customRegexMeta) : [], preHotwordAllowedLabels);
+	const regexEntities = filterAllowedLabels([...rawRegexEntities, ...customRegexEntities], preHotwordAllowedLabels);
 	if (regexEntities.length > 0) log("regex", `${regexEntities.length} matches`);
 	const rawLegalFormEntities = legalFormsEnabled ? processLegalFormMatches(regexMatches, slices.legalForms.start, slices.legalForms.end, fullText) : [];
 	const legalFormEntities = filterAllowedLabels(rawLegalFormEntities, preHotwordAllowedLabels);
@@ -5178,18 +5300,33 @@ const runPipeline = async (options) => {
 	const gazetteerEntities = filterAllowedLabels(rawGazetteerEntities, preHotwordAllowedLabels);
 	if (gazetteerEntities.length > 0) log("gazetteer", `${gazetteerEntities.length} matches`);
 	checkAbort(signal);
+	const rawCustomDenyListEntities = rawDenyListEntities.filter((entity) => isCallerOwnedEntity(entity));
+	const rawCuratedDenyListEntities = rawDenyListEntities.filter((entity) => !isCallerOwnedEntity(entity));
+	const customDenyListEntities = filterAllowedLabels(rawCustomDenyListEntities, preHotwordAllowedLabels);
 	const ruleContextEntities = [
 		...rawTriggerEntities,
 		...rawRegexEntities,
+		...customRegexEntities,
 		...rawLegalFormEntities,
 		...rawNameCorpusEntities,
-		...rawDenyListEntities,
+		...rawCuratedDenyListEntities,
+		...customDenyListEntities,
+		...rawGazetteerEntities
+	];
+	const nerMaskEntities = [
+		...rawTriggerEntities,
+		...rawRegexEntities,
+		...customRegexEntities,
+		...rawLegalFormEntities,
+		...rawNameCorpusEntities,
+		...rawCuratedDenyListEntities,
+		...customDenyListEntities,
 		...rawGazetteerEntities
 	];
 	let rawNerEntities = [];
 	let nerEntities = [];
 	if (config.enableNer && nerInference) {
-		const maskResult = maskDetectedSpans(fullText, ruleContextEntities);
+		const maskResult = maskDetectedSpans(fullText, nerMaskEntities);
 		log("ner", "running inference...");
 		const rawNer = await nerInference(maskResult.maskedText, [...getRequestedNerLabels(config, hotwordsActive)], config.threshold, signal);
 		rawNerEntities = unmaskNerEntities(rawNer, maskResult, fullText);
@@ -5212,7 +5349,12 @@ const runPipeline = async (options) => {
 	if (addressSeedEntities.length > 0) log("address-seeds", `${addressSeedEntities.length} expanded`);
 	checkAbort(signal);
 	const zoneAdjusted = applyZoneAdjustments([...preAddressEntities, ...addressSeedEntities], zones);
-	const preBoostEntities = hotwordsActive ? filterAllowedLabels(applyHotwordRules(zoneAdjusted, fullText), allowedLabels) : zoneAdjusted;
+	const preBoostEntities = (() => {
+		if (!hotwordsActive) return zoneAdjusted;
+		const hotwordCandidates = zoneAdjusted.filter((entity) => !isCallerOwnedEntity(entity));
+		const callerOwnedEntities = zoneAdjusted.filter(isCallerOwnedEntity);
+		return [...filterAllowedLabels(applyHotwordRules(hotwordCandidates, fullText), allowedLabels), ...callerOwnedEntities];
+	})();
 	let allEntities;
 	if (config.enableConfidenceBoost) {
 		allEntities = boostNearMissEntities(preBoostEntities, config.threshold);
@@ -5247,7 +5389,7 @@ const runPipeline = async (options) => {
 	checkAbort(signal);
 	ctx.corefSourceMap.clear();
 	if (config.enableCoreference) {
-		const terms = await extractDefinedTerms(fullText, merged, ctx);
+		const terms = await extractDefinedTerms(fullText, merged.filter((entity) => !isCallerOwnedEntity(entity)), ctx);
 		if (terms.length > 0) {
 			log("coreference", `${terms.length} defined terms`);
 			const corefSpans = findCoreferenceSpans(fullText, terms, ctx);