paperclip-plugin-supermemory-otomatik 0.2.8 → 0.2.9

package/dist/worker.js

@@ -77595,6 +77595,32 @@ async function remoteTextDocument(ctx, base, key, assignment, docId, title, cont
   if (!id) throw new Error("Supermemory Local non ha restituito l\u2019ID del documento.");
   return { id, status: typeof parsed.status === "string" ? parsed.status : "queued" };
 }
+async function indexPdfTextDocuments(ctx, base, key, assignment, parentDocumentId, fileName, bytes) {
+  const extractedText = await extractPdfText(bytes);
+  if (extractedText.length < 40) throw new Error("Il PDF non contiene abbastanza testo selezionabile. I PDF scansionati richiedono OCR.");
+  const chunks = splitUtf8Text(extractedText);
+  const ids = [];
+  const statuses = [];
+  for (let index = 0; index < chunks.length; index += 1) {
+    try {
+      const indexed = await remoteTextDocument(
+        ctx,
+        base,
+        key,
+        assignment,
+        null,
+        chunks.length === 1 ? `${fileName} \u2014 testo estratto` : `${fileName} \u2014 estratto ${index + 1}/${chunks.length}`,
+        chunks[index]
+      );
+      ids.push(indexed.id);
+      statuses.push(indexed.status);
+    } catch (error62) {
+      ctx.logger.warn("PDF text companion could not be indexed", { namespace: assignment.namespace, parentDocumentId, error: errorText(error62) });
+    }
+  }
+  if (!ids.length) throw new Error("Supermemory non ha accettato l\u2019estratto testuale del PDF.");
+  return { ids, status: statuses.every((status) => status === "done") ? "done" : statuses[0] ?? "queued" };
+}
 var plugin = definePlugin({
   async setup(ctx) {
     ctx.tools.register(SEARCH_TOOL, {
@@ -77787,31 +77813,13 @@ var plugin = definePlugin({
       if (!documentId) throw new Error("Il gateway Supermemory non ha restituito l\u2019ID del file caricato.");
       const now = (/* @__PURE__ */ new Date()).toISOString();
       const binaryStatus = typeof payload.status === "string" ? payload.status : "queued";
-      const indexedTextDocumentIds = [];
-      const indexedTextStatuses = [];
+      let indexedTextDocumentIds = [];
+      let indexedTextStatus = null;
       if (mimeType === "application/pdf") {
         try {
-          const extractedText = await extractPdfText(bytes);
-          if (extractedText.length >= 40) {
-            const chunks = splitUtf8Text(extractedText);
-            for (let index = 0; index < chunks.length; index += 1) {
-              try {
-                const indexed = await remoteTextDocument(
-                  ctx,
-                  base,
-                  key,
-                  assignment,
-                  null,
-                  chunks.length === 1 ? `${fileName} \u2014 testo estratto` : `${fileName} \u2014 estratto ${index + 1}/${chunks.length}`,
-                  chunks[index]
-                );
-                indexedTextDocumentIds.push(indexed.id);
-                indexedTextStatuses.push(indexed.status);
-              } catch (error62) {
-                ctx.logger.warn("PDF text companion could not be indexed", { companyId, namespace: assignment.namespace, documentId, error: errorText(error62) });
-              }
-            }
-          }
+          const indexed = await indexPdfTextDocuments(ctx, base, key, assignment, documentId, fileName, bytes);
+          indexedTextDocumentIds = indexed.ids;
+          indexedTextStatus = indexed.status;
         } catch (error62) {
           ctx.logger.warn("Local PDF text extraction failed; keeping the uploaded binary", { companyId, namespace: assignment.namespace, documentId, error: errorText(error62) });
         }
@@ -77826,9 +77834,9 @@ var plugin = definePlugin({
         mimeType,
         fileName,
         sourceUrl: null,
-        status: indexedTextDocumentIds.length ? indexedTextStatuses[0] ?? "queued" : binaryStatus,
+        status: indexedTextStatus ?? binaryStatus,
         indexedTextDocumentIds,
-        indexedTextStatus: indexedTextStatuses.length ? indexedTextStatuses.every((status) => status === "done") ? "done" : indexedTextStatuses[0] ?? "queued" : null,
+        indexedTextStatus,
         binaryProcessingStatus: binaryStatus,
         createdAt: now,
         updatedAt: now
@@ -77836,6 +77844,56 @@ var plugin = definePlugin({
       await saveDocumentEntity(ctx, companyId, assignment, data);
       return { document: { ...data, content: void 0 } };
     });
+    ctx.actions.register("memory-file-reindex-text", async (params, actionCtx) => {
+      const companyId = actionCompany(params, actionCtx);
+      await ownerFor(ctx, companyId, actionCtx);
+      const assignment = await requireAssignment(ctx, companyId, String(params.assignmentId ?? ""));
+      const documentId = typeof params.documentId === "string" ? params.documentId.trim() : "";
+      const fileName = requireTitle(params.fileName);
+      const mimeType = typeof params.mimeType === "string" ? params.mimeType : "";
+      const base643 = typeof params.base64 === "string" ? params.base64 : "";
+      if (!documentId || mimeType !== "application/pdf" || !fileName.toLowerCase().endsWith(".pdf")) throw new Error("Seleziona un PDF per collegare il testo al documento esistente.");
+      const { row, data: record2 } = await sourceDocumentEntry(ctx, companyId, assignment.namespace, documentId);
+      if (record2.kind !== "file" || record2.mimeType !== "application/pdf" || record2.fileName !== fileName) throw new Error("Il nome del file deve corrispondere al PDF gi\xE0 selezionato nella libreria.");
+      const bytes = Buffer.from(base643, "base64");
+      if (!bytes.byteLength || bytes.byteLength > MAX_BINARY_BYTES) throw new Error("Il PDF deve contenere dati e non superare 6 MiB.");
+      if (!/^[A-Za-z0-9+/]*={0,2}$/.test(base643) || bytes.toString("base64").replace(/=+$/, "") !== base643.replace(/=+$/, "")) throw new Error("Contenuto file base64 non valido.");
+      const { base, key } = await adminConfig(ctx, companyId);
+      const indexed = await indexPdfTextDocuments(ctx, base, key, assignment, documentId, fileName, bytes);
+      const oldIds = Array.isArray(record2.indexedTextDocumentIds) ? record2.indexedTextDocumentIds.filter((id) => typeof id === "string") : [];
+      if (oldIds.length) {
+        await ownerGatewayRequest(ctx, base, key, localNamespacePath(assignment.namespace, "/document"), {
+          method: "DELETE",
+          headers: { "content-type": "application/json" },
+          body: JSON.stringify({ ids: [...new Set(oldIds)] })
+        });
+      }
+      const data = {
+        ...record2,
+        indexedTextDocumentIds: indexed.ids,
+        indexedTextStatus: indexed.status,
+        processingStatus: indexed.status,
+        updatedAt: (/* @__PURE__ */ new Date()).toISOString()
+      };
+      const externalId = typeof row.externalId === "string" ? row.externalId : documentExternalId(assignment.namespace, documentId);
+      await ctx.entities.upsert({ entityType: DOCUMENT_ENTITY, scopeKind: "company", scopeId: companyId, externalId, title: row.title ?? fileName, status: "active", data });
+      return { document: {
+        id: documentId,
+        namespace: assignment.namespace,
+        assignmentId: String(data.assignmentId ?? assignment.id),
+        title: String(data.title ?? fileName),
+        kind: "file",
+        mimeType: "application/pdf",
+        fileName,
+        sourceUrl: null,
+        status: indexed.status,
+        indexedTextDocumentIds: indexed.ids,
+        indexedTextStatus: indexed.status,
+        binaryProcessingStatus: typeof data.binaryProcessingStatus === "string" ? data.binaryProcessingStatus : null,
+        createdAt: String(data.createdAt ?? ""),
+        updatedAt: String(data.updatedAt)
+      } };
+    });
     ctx.actions.register("memory-document-delete", async (params, actionCtx) => {
       const companyId = actionCompany(params, actionCtx);
       await ownerFor(ctx, companyId, actionCtx);