@form-engine-ts/privacy 2.7.0 → 2.8.0

package/README.md

@@ -1,9 +1,11 @@
 # @form-engine-ts/privacy
 
 Framework-independent sensitive-data candidate detection for form-engine-ts. The standard detector scans only text and
-textarea answers for email addresses, phone numbers, HTTP(S) URLs, and postal codes. Rules and custom detectors are
+textarea answers for email addresses, phone numbers, HTTP(S)/`www.` URLs, and postal codes. Rules and custom detectors are
 injectable, and findings are advisory so applications can choose confirm, block, or audit behavior.
 
+Overlapping or contained findings with the same field and type are merged into one source range.
+
 ```ts
 import { createStandardPrivacyDetector } from "@form-engine-ts/privacy";
 

package/dist/index.cjs

@@ -25,7 +25,7 @@ __export(index_exports, {
 module.exports = __toCommonJS(index_exports);
 var STANDARD_RULES = [
   { type: "email", pattern: /\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,}\b/giu },
-  { type: "url", pattern: /\bhttps?:\/\/[^\s<>"']+[^\s<>"'.,;:!?)]/giu },
+  { type: "url", pattern: /\b(?:https?:\/\/|www\.)[^\s<>"']*[A-Z0-9/#]/giu },
   { type: "phone", pattern: /(?:\+?\d[\d\s().-]{6,}\d)/gu },
   { type: "postal_code", pattern: /(?:〒\s*)?\b\d{3}-?\d{4}\b/gu }
 ];
@@ -57,6 +57,42 @@ function detectRule(fieldId, text, rule) {
   }
   return findings;
 }
+function findingsOverlap(left, right) {
+  if (left.fieldId !== right.fieldId || left.type !== right.type) return false;
+  if (left.start === void 0 || left.end === void 0 || right.start === void 0 || right.end === void 0) {
+    return left.matchedText !== void 0 && left.matchedText === right.matchedText;
+  }
+  return left.start < right.end && right.start < left.end;
+}
+function deduplicateFindings(findings, values) {
+  const deduplicated = [];
+  for (const finding of findings) {
+    const overlappingIndexes = deduplicated.flatMap(
+      (candidate, index) => findingsOverlap(candidate, finding) ? [index] : []
+    );
+    if (overlappingIndexes.length === 0) {
+      deduplicated.push(finding);
+      continue;
+    }
+    const overlapping = overlappingIndexes.map((index) => deduplicated[index]).filter((value) => value !== void 0);
+    const starts = [finding, ...overlapping].flatMap((value) => value.start === void 0 ? [] : [value.start]);
+    const ends = [finding, ...overlapping].flatMap((value) => value.end === void 0 ? [] : [value.end]);
+    const start = starts.length === 0 ? void 0 : Math.min(...starts);
+    const end = ends.length === 0 ? void 0 : Math.max(...ends);
+    const text = values[finding.fieldId];
+    const merged = {
+      fieldId: finding.fieldId,
+      type: finding.type,
+      ...start === void 0 ? {} : { start },
+      ...end === void 0 ? {} : { end },
+      ...start !== void 0 && end !== void 0 && typeof text === "string" ? { matchedText: text.slice(start, end) } : finding.matchedText === void 0 ? {} : { matchedText: finding.matchedText }
+    };
+    const insertionIndex = overlappingIndexes[0] ?? deduplicated.length;
+    for (const index of [...overlappingIndexes].sort((left, right) => right - left)) deduplicated.splice(index, 1);
+    deduplicated.splice(insertionIndex, 0, merged);
+  }
+  return deduplicated;
+}
 function createStandardPrivacyDetector(config = {}) {
   const rules = configuredRules(config);
   const customDetectors = [...config.customDetectors ?? []];
@@ -73,7 +109,7 @@ function createStandardPrivacyDetector(config = {}) {
         for (const rule of rules) findings.push(...detectRule(field.id, value, rule));
         for (const detector of customDetectors) findings.push(...detector(field.id, value));
       }
-      return findings;
+      return deduplicateFindings(findings, values);
     }
   };
 }

package/dist/index.js

@@ -1,7 +1,7 @@
 // src/index.ts
 var STANDARD_RULES = [
   { type: "email", pattern: /\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,}\b/giu },
-  { type: "url", pattern: /\bhttps?:\/\/[^\s<>"']+[^\s<>"'.,;:!?)]/giu },
+  { type: "url", pattern: /\b(?:https?:\/\/|www\.)[^\s<>"']*[A-Z0-9/#]/giu },
   { type: "phone", pattern: /(?:\+?\d[\d\s().-]{6,}\d)/gu },
   { type: "postal_code", pattern: /(?:〒\s*)?\b\d{3}-?\d{4}\b/gu }
 ];
@@ -33,6 +33,42 @@ function detectRule(fieldId, text, rule) {
   }
   return findings;
 }
+function findingsOverlap(left, right) {
+  if (left.fieldId !== right.fieldId || left.type !== right.type) return false;
+  if (left.start === void 0 || left.end === void 0 || right.start === void 0 || right.end === void 0) {
+    return left.matchedText !== void 0 && left.matchedText === right.matchedText;
+  }
+  return left.start < right.end && right.start < left.end;
+}
+function deduplicateFindings(findings, values) {
+  const deduplicated = [];
+  for (const finding of findings) {
+    const overlappingIndexes = deduplicated.flatMap(
+      (candidate, index) => findingsOverlap(candidate, finding) ? [index] : []
+    );
+    if (overlappingIndexes.length === 0) {
+      deduplicated.push(finding);
+      continue;
+    }
+    const overlapping = overlappingIndexes.map((index) => deduplicated[index]).filter((value) => value !== void 0);
+    const starts = [finding, ...overlapping].flatMap((value) => value.start === void 0 ? [] : [value.start]);
+    const ends = [finding, ...overlapping].flatMap((value) => value.end === void 0 ? [] : [value.end]);
+    const start = starts.length === 0 ? void 0 : Math.min(...starts);
+    const end = ends.length === 0 ? void 0 : Math.max(...ends);
+    const text = values[finding.fieldId];
+    const merged = {
+      fieldId: finding.fieldId,
+      type: finding.type,
+      ...start === void 0 ? {} : { start },
+      ...end === void 0 ? {} : { end },
+      ...start !== void 0 && end !== void 0 && typeof text === "string" ? { matchedText: text.slice(start, end) } : finding.matchedText === void 0 ? {} : { matchedText: finding.matchedText }
+    };
+    const insertionIndex = overlappingIndexes[0] ?? deduplicated.length;
+    for (const index of [...overlappingIndexes].sort((left, right) => right - left)) deduplicated.splice(index, 1);
+    deduplicated.splice(insertionIndex, 0, merged);
+  }
+  return deduplicated;
+}
 function createStandardPrivacyDetector(config = {}) {
   const rules = configuredRules(config);
   const customDetectors = [...config.customDetectors ?? []];
@@ -49,7 +85,7 @@ function createStandardPrivacyDetector(config = {}) {
         for (const rule of rules) findings.push(...detectRule(field.id, value, rule));
         for (const detector of customDetectors) findings.push(...detector(field.id, value));
       }
-      return findings;
+      return deduplicateFindings(findings, values);
     }
   };
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@form-engine-ts/privacy",
-  "version": "2.7.0",
+  "version": "2.8.0",
   "publishConfig": {
     "access": "public"
   },
@@ -38,7 +38,7 @@
     "typescript"
   ],
   "dependencies": {
-    "@form-engine-ts/core": "2.7.0"
+    "@form-engine-ts/core": "2.8.0"
   },
   "scripts": {
     "build": "tsup src/index.ts --format esm,cjs --dts --clean --external @form-engine-ts/core",