n8n-nodes-redactor 5.1.0 → 5.2.0

package/dist/nodes/Redactor/Redactor.node.js

@@ -767,7 +767,7 @@ class Redactor {
         };
     }
     async execute() {
-        var _a, _b, _c, _d, _e, _f, _g;
+        var _a, _b, _c, _d, _e, _f, _g, _h;
         const items = this.getInputData();
         const allPaired = items.map((_, i) => ({ item: i }));
         const operation = this.getNodeParameter('operation', 0);
@@ -1041,6 +1041,7 @@ class Redactor {
                 keyOrigins: new WeakMap(),
                 inputText: (0, engine_1.collectInputText)(items.map((it) => it.json)),
                 lockedSpans: new WeakMap(),
+                priorTokens: (0, engine_1.priorTokenMatcher)((_e = vault.getSession(sessionId)) === null || _e === void 0 ? void 0 : _e.entries),
             };
             const allHits = [];
             const returnData = [];
@@ -1091,7 +1092,7 @@ class Redactor {
                 if (includeReport && this.getNodeParameter('includeRopa', 0, false)) {
                     const d = this.getNodeParameter('ropa', 0, {});
                     const list = (v) => String(v !== null && v !== void 0 ? v : '').split(',').map((x) => x.trim()).filter(Boolean);
-                    report.ropa = (0, ropa_1.generateRopaRecord)(allHits, (0, classification_1.classifyData)(allHits), items.length, enabledPatterns.length, mode, sessionTtl, !!vaultPassphrase, auditLog, { name: String((_e = d.controllerName) !== null && _e !== void 0 ? _e : ''), contactEmail: String((_f = d.controllerContact) !== null && _f !== void 0 ? _f : ''), dpo: d.dpo ? String(d.dpo) : undefined }, list(d.purposes), String((_g = d.legalBasis) !== null && _g !== void 0 ? _g : 'legitimate_interest'), list(d.dataSubjects), list(d.recipients), []);
+                    report.ropa = (0, ropa_1.generateRopaRecord)(allHits, (0, classification_1.classifyData)(allHits), items.length, enabledPatterns.length, mode, sessionTtl, !!vaultPassphrase, auditLog, { name: String((_f = d.controllerName) !== null && _f !== void 0 ? _f : ''), contactEmail: String((_g = d.controllerContact) !== null && _g !== void 0 ? _g : ''), dpo: d.dpo ? String(d.dpo) : undefined }, list(d.purposes), String((_h = d.legalBasis) !== null && _h !== void 0 ? _h : 'legitimate_interest'), list(d.dataSubjects), list(d.recipients), []);
                 }
                 return [returnData, [{ json: report, pairedItem: allPaired }]];
             }

package/dist/nodes/Redactor/engine.js

@@ -36,6 +36,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.collectTokenLike = collectTokenLike;
 exports.collectInputText = collectInputText;
 exports.redactValue = redactValue;
+exports.priorTokenMatcher = priorTokenMatcher;
 exports.restoreValue = restoreValue;
 exports.enhanceWithPresidio = enhanceWithPresidio;
 exports.buildReport = buildReport;
@@ -548,33 +549,56 @@ function assertUserRegexInput(text, what) {
     throw new Error(`${what} cannot scan a value of ${text.length} characters, the limit is ${USER_REGEX_MAX_INPUT}. ` +
         'Exclude the field with field rules, or remove the regex.');
 }
-function freeSegmentAt(segs, start, end) {
-    return segs.findIndex((seg) => !seg.locked && seg.src <= start && end <= seg.src + seg.text.length);
-}
-function lockRange(segs, index, start, end, replacement) {
-    const seg = segs[index];
-    const parts = [];
-    if (start > seg.src)
-        parts.push({ text: seg.text.slice(0, start - seg.src), locked: false, src: seg.src });
-    parts.push({ text: replacement, locked: true, src: -1 });
-    const tail = end - seg.src;
-    if (tail < seg.text.length)
-        parts.push({ text: seg.text.slice(tail), locked: false, src: end });
-    segs.splice(index, 1, ...parts);
+class SegmentCursor {
+    constructor(segs) {
+        this.segs = segs;
+        this.next = [];
+        this.i = 0;
+        this.cur = segs[0];
+    }
+    freeAt(start, end) {
+        while (this.cur && (this.cur.locked || this.cur.src + this.cur.text.length <= start))
+            this.take();
+        const cur = this.cur;
+        return cur && cur.src <= start && end <= cur.src + cur.text.length ? cur : undefined;
+    }
+    lock(start, end, replacement) {
+        const cur = this.cur;
+        if (start > cur.src)
+            this.next.push({ text: cur.text.slice(0, start - cur.src), locked: false, src: cur.src });
+        this.next.push({ text: replacement, locked: true, src: -1 });
+        const tail = end - cur.src;
+        if (tail < cur.text.length)
+            this.cur = { text: cur.text.slice(tail), locked: false, src: end };
+        else
+            this.cur = this.segs[++this.i];
+    }
+    finish() {
+        while (this.cur)
+            this.take();
+        return this.next;
+    }
+    take() {
+        this.next.push(this.cur);
+        this.cur = this.segs[++this.i];
+    }
 }
 function replaceFree(pass, re, fn) {
-    var _a;
-    const flags = re.flags.includes('g') ? re.flags : `${re.flags}g`;
-    for (const m of pass.source.matchAll(new RegExp(re.source, flags))) {
-        const start = (_a = m.index) !== null && _a !== void 0 ? _a : 0;
+    const global = new RegExp(re.source, re.flags.includes('g') ? re.flags : `${re.flags}g`);
+    const cursor = new SegmentCursor(pass.segs);
+    let m;
+    while ((m = global.exec(pass.source)) !== null) {
+        const start = m.index;
         const end = start + m[0].length;
-        const index = m[0] === '' ? -1 : freeSegmentAt(pass.segs, start, end);
-        if (index < 0)
+        if (m[0] === '' || !cursor.freeAt(start, end)) {
+            global.lastIndex = start + 1;
             continue;
+        }
         const replacement = fn(m[0], start);
         if (replacement !== null)
-            lockRange(pass.segs, index, start, end, replacement);
+            cursor.lock(start, end, replacement);
     }
+    pass.segs = cursor.finish();
 }
 function redactMatch(pass, match, pattern, detail) {
     const replacement = applyRedaction(match, pattern, pass.ctx, pass.vault, pass.sessionId);
@@ -625,6 +649,10 @@ function builtInStage(pass) {
 }
 function sourceNames(pass) {
     const names = new Set((0, names_1.detectNamesInText)(pass.source).map((nm) => nm.name));
+    for (const seg of pass.segs)
+        if (!seg.locked)
+            for (const nm of (0, names_1.detectNamesInText)(seg.text))
+                names.add(nm.name);
     return [...names].sort((x, y) => y.length - x.length);
 }
 function nameStage(pass) {
@@ -686,12 +714,27 @@ function lockChild(ctx, container, key, compute) {
     ctx.lastLocked = undefined;
     return out;
 }
+function priorTokenMatcher(entries) {
+    const tokens = Object.keys(entries !== null && entries !== void 0 ? entries : {}).filter((t) => t.length > 0).sort((a, b) => b.length - a.length);
+    return tokens.length > 0 ? new RegExp(tokens.map(escapeRe).join('|'), 'g') : undefined;
+}
+function literalStage(pass) {
+    const { ctx, vault, sessionId } = pass;
+    if (!ctx.priorTokens || (ctx.mode !== 'token' && ctx.mode !== 'pseudonymize'))
+        return;
+    replaceFree(pass, ctx.priorTokens, (match) => {
+        const token = freshToken('LITERAL', vault, sessionId, ctx.reservedTokens);
+        vault.addEntry(sessionId, { token, original: match, patternLabel: 'LITERAL', category: 'other', createdAt: new Date().toISOString() });
+        return token;
+    });
+}
 function redactText(text, ctx, vault, sessionId, hits, itemIndex, fieldPath) {
     var _a;
     const pass = {
         source: text, segs: [{ text, locked: false, src: 0 }], threshold: (_a = ctx.confidenceThreshold) !== null && _a !== void 0 ? _a : 0,
         ctx, vault, sessionId, hits, itemIndex, fieldPath,
     };
+    literalStage(pass);
     denyListStage(pass);
     builtInStage(pass);
     nameStage(pass);

package/dist/nodes/Redactor/regexCharset.js

@@ -93,6 +93,8 @@ function escapeSet(src, i) {
 function classMember(body, j) {
     if (body[j] === '\\' && body[j + 1] === 'b')
         return { set: codeSet(8), next: j + 2, code: 8 };
+    if (body[j] === '\\' && body[j + 1] === 'B')
+        return { set: codeSet(66), next: j + 2, code: 66 };
     if (body[j] === '\\') {
         const { length, code } = readEscape(body, j);
         return { set: escapeSet(body, j), next: j + length, code };

package/dist/nodes/Redactor/regexSafety.js

@@ -3,9 +3,9 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.checkRegex = checkRegex;
 const regexCharset_1 = require("./regexCharset");
 const MAX_REGEX_LENGTH = 500;
-const newFrame = (zeroWidth = false) => ({
+const newFrame = (zeroWidth = false, inLook = zeroWidth) => ({
     hasVariableQuantifier: false, hasAlternation: false, run: null, chars: (0, regexCharset_1.emptySet)(), zeroWidth,
-    branchConsumes: false, allConsume: true, nullable: false,
+    branchConsumes: false, allConsume: true, nullable: false, inLook,
 });
 function endBranch(frame) {
     frame.allConsume && (frame.allConsume = frame.branchConsumes);
@@ -48,7 +48,7 @@ function skipClass(src, i) {
 function groupPrefixLength(src, i) {
     if (src[i + 1] !== '?')
         return 1;
-    const named = /^\(\?<[A-Za-z_$][\w$]*>/.exec(src.slice(i));
+    const named = /^\(\?<[^=!>][^>]*>/.exec(src.slice(i));
     if (named)
         return named[0].length;
     const look = /^\(\?(?::|=|!|<=|<!)/.exec(src.slice(i));
@@ -79,7 +79,8 @@ function readToken(src, i, stack) {
     if (c === '[')
         return { atomEnd: skipClass(src, i), closed: null };
     if (c === '(') {
-        stack.push(newFrame(/^\(\?(?:=|!|<=|<!)/.test(src.slice(i))));
+        const look = /^\(\?(?:=|!|<=|<!)/.test(src.slice(i));
+        stack.push(newFrame(look, look || stack[stack.length - 1].inLook));
         return { atomEnd: i, closed: null, next: i + groupPrefixLength(src, i) };
     }
     if (c === '|')
@@ -127,8 +128,11 @@ function consumes(q, chars, closed) {
         return false;
     return !(closed === null || closed === void 0 ? void 0 : closed.nullable);
 }
+const IN_LOOK = 'repeats a variable part inside a lookahead or lookbehind, which is not supported';
 function stepProblem(src, i, step, q, top) {
     var _a;
+    if (top.inLook && isVariable(q))
+        return IN_LOOK;
     const problem = applyQuantifier(q, step.closed, top);
     if (problem || (q === null || q === void 0 ? void 0 : q.max) === 0)
         return problem;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "n8n-nodes-redactor",
-  "version": "5.1.0",
+  "version": "5.2.0",
   "description": "Redact, mask and anonymize PII before it reaches an LLM, then restore it. 216 patterns for emails, phones, addresses, IBAN, cards, names, SSN, passport and tax IDs. Reversible encrypted vault, GDPR Art.30 ROPA record, prompt-injection check. Runs on n8n Cloud and self-hosted n8n 2.x.",
   "keywords": [
     "n8n-community-node-package",