@manfred-kunze-dev/backbone-mcp-server 2.9.0 → 2.10.0-dev.2

package/dist/tools/conversion.js

@@ -30,9 +30,49 @@ function formatExportDocument(doc) {
         parts.push({ type: "text", text: `--- ${doc.filename} (json) ---\n${JSON.stringify(doc.jsonContent, null, 2)}` });
     return parts;
 }
+const TERMINAL_STATUSES = new Set(["SUCCESS", "PARTIAL_SUCCESS", "FAILURE", "COMPLETED", "FAILED"]);
+/**
+ * Poll an async task until terminal status, then fetch and return the result.
+ */
+async function waitForTask(client, taskId, pollSeconds = 30) {
+    // eslint-disable-next-line no-constant-condition
+    while (true) {
+        const { data } = await client.GET("/v1/convert/tasks/{taskId}", {
+            params: { path: { taskId }, query: { wait: pollSeconds } },
+        });
+        const status = data?.taskStatus;
+        if (status && TERMINAL_STATUSES.has(status.toUpperCase()))
+            break;
+    }
+    const { data } = await client.GET("/v1/convert/tasks/{taskId}/result", {
+        params: { path: { taskId } },
+    });
+    return data;
+}
+/**
+ * Format a ConvertResponse into MCP text content parts.
+ */
+function formatConvertResponse(result) {
+    const parts = [];
+    if (result.documents?.length) {
+        for (const doc of result.documents) {
+            parts.push(...formatExportDocument(doc));
+        }
+    }
+    if (result.errors?.length) {
+        parts.push({
+            type: "text",
+            text: `Errors:\n${result.errors.map((e) => `- ${e.filename ?? "unknown"}: ${e.errorMessage}`).join("\n")}`,
+        });
+    }
+    if (parts.length === 0) {
+        parts.push({ type: "text", text: `Conversion completed with status: ${result.status}` });
+    }
+    return parts;
+}
 export function register(server, client) {
     // ── convert_document ────────────────────────────────────────────────────
-    server.tool("backbone_convert_document", "Convert documents (PDF, DOCX, XLSX, images, etc.) to Markdown, text, HTML, or JSON. Accepts URLs, base64 data, or local file paths. Local files are automatically read and base64-encoded. Use this to convert binary files before passing them to backbone_create_extraction. Pipeline options: 'fast' (default, fast text extraction), 'ocr' (OCR with layout analysis), or 'vlm' (vision language model for image-heavy/complex layouts). Use async mode for large documents.", {
+    server.tool("backbone_convert_document", "Convert documents (PDF, DOCX, XLSX, images, etc.) to Markdown, text, HTML, or JSON. Accepts URLs, base64 data, or local file paths. Local files are automatically read and base64-encoded. Use this to convert binary files before passing them to backbone_create_extraction. Pipeline options: 'fast' (default, fast text extraction), 'ocr' (OCR with layout analysis), or 'vlm' (vision language model for image-heavy/complex layouts). Set async=true with waitForCompletion=true for long-running pipelines like VLM — this submits asynchronously and automatically polls until the result is ready.", {
         sources: z
             .array(z.object({
             type: z.enum(["url", "base64", "file"]).describe("Source type: 'url' for HTTP URLs, 'base64' for base64-encoded data, 'file' for local file paths"),
@@ -64,8 +104,13 @@ export function register(server, client) {
             .boolean()
             .optional()
             .default(false)
-            .describe("If true, submit as async task and return task ID for polling"),
-    }, async ({ sources, pipeline, options: pipelineOptions, async: isAsync }) => {
+            .describe("If true, submit as async task. Combine with waitForCompletion=true to get the final result in a single call."),
+        waitForCompletion: z
+            .boolean()
+            .optional()
+            .default(false)
+            .describe("When async=true, automatically poll until done and return the final result instead of a task ID. Recommended for VLM pipeline."),
+    }, async ({ sources, pipeline, options: pipelineOptions, async: isAsync, waitForCompletion }) => {
         try {
             const apiSources = [];
             for (const src of sources) {
@@ -122,36 +167,17 @@ export function register(server, client) {
             };
             if (isAsync) {
                 const { data } = await client.POST("/v1/convert/source/async", { body });
+                const taskId = data?.taskId;
+                if (waitForCompletion && taskId) {
+                    const result = await waitForTask(client, taskId);
+                    return { content: formatConvertResponse(result) };
+                }
                 return {
-                    content: [
-                        {
-                            type: "text",
-                            text: JSON.stringify(data, null, 2),
-                        },
-                    ],
+                    content: [{ type: "text", text: JSON.stringify(data, null, 2) }],
                 };
             }
             const { data } = await client.POST("/v1/convert/source", { body });
-            const convertResult = data;
-            const parts = [];
-            if (convertResult.documents?.length) {
-                for (const doc of convertResult.documents) {
-                    parts.push(...formatExportDocument(doc));
-                }
-            }
-            if (convertResult.errors?.length) {
-                parts.push({
-                    type: "text",
-                    text: `Errors:\n${convertResult.errors.map((e) => `- ${e.filename ?? "unknown"}: ${e.errorMessage}`).join("\n")}`,
-                });
-            }
-            if (parts.length === 0) {
-                parts.push({
-                    type: "text",
-                    text: `Conversion completed with status: ${convertResult.status}`,
-                });
-            }
-            return { content: parts };
+            return { content: formatConvertResponse(data) };
         }
         catch (error) {
             return {
@@ -194,26 +220,7 @@ export function register(server, client) {
             const { data } = await client.GET("/v1/convert/tasks/{taskId}/result", {
                 params: { path: { taskId } },
             });
-            const result = data;
-            const parts = [];
-            if (result.documents?.length) {
-                for (const doc of result.documents) {
-                    parts.push(...formatExportDocument(doc));
-                }
-            }
-            if (result.errors?.length) {
-                parts.push({
-                    type: "text",
-                    text: `Errors:\n${result.errors.map((e) => `- ${e.filename ?? "unknown"}: ${e.errorMessage}`).join("\n")}`,
-                });
-            }
-            if (parts.length === 0) {
-                parts.push({
-                    type: "text",
-                    text: `Task result status: ${result.status}`,
-                });
-            }
-            return { content: parts };
+            return { content: formatConvertResponse(data) };
         }
         catch (error) {
             return {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@manfred-kunze-dev/backbone-mcp-server",
-  "version": "2.9.0",
+  "version": "2.10.0-dev.2",
   "description": "MCP server for the Backbone AI platform",
   "type": "module",
   "main": "dist/index.js",