@vertesia/workflow 1.0.0-dev.20260203.130115Z → 1.0.0-dev.20260225.024852Z

package/src/index.ts

@@ -18,15 +18,15 @@ export * from "./activities/executeInteraction.js";
 export * from "./activities/extractDocumentText.js";
 export * from "./activities/generateDocumentProperties.js";
 export * from "./activities/generateEmbeddings.js";
-export * from "./activities/renditions/generateImageRendition.js";
-export * from "./activities/renditions/generateVideoRendition.js";
 export * from "./activities/generateOrAssignContentType.js";
+export * from "./activities/media/prepareAudio.js";
+export * from "./activities/media/prepareVideo.js";
 export * from "./activities/notifyWebhook.js";
 export * from "./activities/rateLimiter.js";
+export * from "./activities/renditions/generateImageRendition.js";
+export * from "./activities/renditions/generateVideoRendition.js";
 export * from "./activities/setDocumentStatus.js";
 export * from "./iterative-generation/activities/index.js";
-export * from "./activities/media/prepareVideo.js";
-export * from "./activities/media/prepareAudio.js";
 
 export * from "./dsl/setup/ActivityContext.js";
 export * from "./errors.js";
@@ -34,8 +34,9 @@ export * from "./result-types.js";
 export * from "./utils/blobs.js";
 export * from "./utils/client.js";
 export * from "./utils/memory.js";
-export * from "./utils/tokens.js";
 export * from "./utils/renditions.js";
 export * from "./utils/storage.js";
+export * from "./utils/text-preview-utils.js";
+export * from "./utils/tokens.js";
 
 export * from "./conversion/image.js";

package/src/result-types.ts

@@ -5,7 +5,11 @@
  * The result of a text extraction operation.
  */
 export interface TextExtractionResult {
-    objectId: string;
+    objectId?: string;
+    file?: {   
+        source_url: string;
+        result_url?: string;
+    }
     status: TextExtractionStatus;
     hasText: boolean;
     message?: string;

package/src/utils/renditions.ts

@@ -9,6 +9,8 @@ import { imageResizer } from "../conversion/image.js";
 export interface ImageRenditionParams {
     max_hw: number; //maximum size of the longest side of the image
     format: ImageRenditionFormat;
+    /** Custom upload path — overrides default renditions/{etag}/{max_hw}/{page}.{format} path */
+    outputPath?: string;
 }
 
 /**
@@ -53,7 +55,7 @@ export async function uploadRenditionPages(
     params: ImageRenditionParams,
     concurrency?: number,
 ) {
-    log.info(
+    log.debug(
         `Uploading rendition for etag ${contentEtag} with ${files.length} pages (max_hw: ${params.max_hw}, format: ${params.format})`,
         { files },
     );
@@ -61,11 +63,15 @@ export async function uploadRenditionPages(
     const limit = pLimit(concurrency ?? 20);
 
     const uploads = files.map((file, i) => limit(async () => {
-        const pageId = getRenditionPagePath(contentEtag, params, i);
+        const pageId = params.outputPath
+            ? (files.length === 1
+                ? params.outputPath
+                : `${params.outputPath}/${String(i).padStart(4, "0")}.${params.format}`)
+            : getRenditionPagePath(contentEtag, params, i);
         let resizedImagePath = null;
 
         try {
-            log.info(`Resizing image for ${contentEtag} page ${i}`, {
+            log.debug(`Resizing image for ${contentEtag} page ${i}`, {
                 file,
                 params,
             });
@@ -87,7 +93,7 @@ export async function uploadRenditionPages(
                 pageId,
             );
 
-            log.info(
+            log.debug(
                 `Uploading rendition for ${contentEtag} page ${i} with max_hw: ${params.max_hw} and format: ${params.format}`,
                 {
                     resizedImagePath,
@@ -110,7 +116,7 @@ export async function uploadRenditionPages(
                     );
                     return Promise.reject(`Upload failed: ${err.message}`);
                 });
-            log.info(`Rendition uploaded for ${contentEtag} page ${i}`, {
+            log.debug(`Rendition uploaded for ${contentEtag} page ${i}`, {
                 result,
             });
 

package/src/utils/text-preview-utils.ts

@@ -0,0 +1,62 @@
+import { log } from "@temporalio/activity";
+import { VertesiaClient } from "@vertesia/client";
+import { NodeStreamSource } from "@vertesia/client/node";
+import { Readable } from "stream";
+import { TextExtractionResult, TextExtractionStatus } from "../result-types.js";
+
+/**
+ * Uploads extracted text preview to cloud storage
+ */
+export async function uploadTextPreviewToStorage(
+    vertesia: VertesiaClient,
+    text: string,
+    storagePath: string,
+    fileType: string
+): Promise<string> {
+
+    const source = new NodeStreamSource(
+        Readable.from(text),
+        storagePath,
+        'text/markdown',
+        storagePath
+    );
+
+    const uploadedPath = await vertesia.files.uploadFile(source);
+    log.info(`Uploaded ${fileType} preview text to ${uploadedPath}`);
+
+    return uploadedPath;
+}
+
+/**
+ * Creates the TextExtractionResult for file_source mode
+ */
+export function createFileSourceResult(
+    sourceUrl: string,
+    storagePath: string,
+    text: string | null
+): TextExtractionResult {
+    return {
+        hasText: !!text,
+        file: {
+            source_url: sourceUrl,
+            result_url: text && storagePath ? storagePath : undefined,
+        },
+        status: TextExtractionStatus.success,
+    };
+}
+
+/**
+ * Saves extracted text to an object in the object store
+ */
+export async function saveTextToObject(
+    vertesia: VertesiaClient,
+    objectId: string,
+    text: string
+): Promise<void> {
+    const object = await vertesia.objects.retrieve(objectId);
+    await vertesia.objects.update(objectId, {
+        text: text,
+        text_etag: object.content?.etag,
+    });
+    log.info(`Saved text to object ${objectId}`);
+}