fix: allow rerunning imported projects

src/lib/project-review.ts

@@ -89,9 +89,14 @@ export function withoutWhitelistedHits<T extends GroupableHit>(hits: T[], whitel
 }
 
 /** A later run must not turn an existing Velum placeholder plus adjacent text into a new entity. */
-export function withoutPlaceholderHits<T extends { original: string; type: string }>(hits: T[]): T[] {
+export function withoutPlaceholderHits<T extends { original: string; type: string }>(
+  hits: T[], knownPlaceholders: Iterable<string> = [],
+): T[] {
   const placeholder = /(?<![\p{L}\p{N}])[A-Z][A-Z0-9]*_\d{2,}(?![\p{L}\p{N}])/u;
-  return hits.filter((hit) => hit.type === "PASSWORT" || !placeholder.test(hit.original));
+  const known = [...knownPlaceholders].filter(Boolean);
+  return hits.filter((hit) => hit.type === "PASSWORT" || (
+    !placeholder.test(hit.original) && !known.some((key) => hit.original.includes(key))
+  ));
 }
 
 export function withoutWhitelistedRules<T extends { type: string; value: string }>(

src/pseudonymize-project-files.tsx

@@ -101,10 +101,8 @@ async function prepare(root: string, files: ProjectFile[], rules: ProjectRule[],
   const masks = forceMasks(withoutWhitelistedRules(rules, whitelist));
   const latestEntry = (await listRuns(root))[0];
   const latest = latestEntry?.run;
-  if (latest?.origin === "imported") {
-    throw new Error("Dieser Quellordner enthält bereits pseudonymisierte Dateien. Für einen neuen Lauf bitte einen Klartext-Quellordner wählen.");
-  }
   const previousMapping: PlaceholderMapping = latest?.mapping ?? {};
+  const previousPlaceholders = Object.keys(previousMapping);
   let mapping: PlaceholderMapping = previousMapping;
   const drafts: Draft[] = [];
   const retained: Preview["retained"] = [];
@@ -112,6 +110,8 @@ async function prepare(root: string, files: ProjectFile[], rules: ProjectRule[],
   const hits: Hit[] = [];
   const texts = new Map<string, string>();
   const sourceInPlace = Boolean(latest?.inPlace || latest?.sourceInPlace);
+  // A migrated snapshot may differ from the source folder. Recheck its files
+  // unless an earlier run actually replaced them in place.
   const previousOutputs = new Set(sourceInPlace ? latest?.files.map((file) => file.output) : []);
   const previousDirectory = latestEntry?.directory;
   for (const file of files) {
@@ -131,7 +131,7 @@ async function prepare(root: string, files: ProjectFile[], rules: ProjectRule[],
     if (file.kind === "text") texts.set(file.relPath, textFromBytes(data));
   }
   const addHits = (draft: Draft, found: Hit[]) => {
-    draft.hits = withoutWhitelistedHits(withoutPlaceholderHits(found), whitelist);
+    draft.hits = withoutWhitelistedHits(withoutPlaceholderHits(found, previousPlaceholders), whitelist);
     hits.push(...draft.hits);
     if (hits.length > MAX_REVIEW_HITS) throw new Error("Zu viele Treffer. Bitte weniger Kategorien auswählen.");
   };
@@ -175,7 +175,7 @@ async function prepare(root: string, files: ProjectFile[], rules: ProjectRule[],
         source: "Datei- oder Ordnername", kind: "path" as const,
         context: path.value,
       }));
-      path.hits = withoutWhitelistedHits(withoutPlaceholderHits(found), whitelist);
+      path.hits = withoutWhitelistedHits(withoutPlaceholderHits(found, previousPlaceholders), whitelist);
       hits.push(...path.hits);
       if (hits.length > MAX_REVIEW_HITS) throw new Error("Zu viele Treffer. Bitte weniger Kategorien auswählen.");
     }

tests/project-review.test.ts

@@ -67,6 +67,10 @@ it("zeigt beim erneuten Lauf keine Treffer, die einen bestehenden Platzhalter ei
     { type: "PASSWORT", original: "ORT_08!" },
   ];
   expect(withoutPlaceholderHits(hits)).toEqual(hits.slice(2));
+  expect(withoutPlaceholderHits([
+    { type: "PERSON", original: "KUNDE_100005Service" },
+    { type: "PERSON", original: "Roman" },
+  ], ["KUNDE_100005"])).toEqual([{ type: "PERSON", original: "Roman" }]);
 });
 
 it("fasst ausgewählte Personen zu einem kanonischen Platzhalter zusammen und bewahrt alte Zuordnungen", () => {