Update MCP server to accept JSON body with audio_base64/audio_url

- Remove multipart form; use JSON with audio_base64 or audio_url.
- Add internal audio_path field for internal use.
- Keep OpenAPI spec consistent with new schema.

scraibe/mcp_server.py

@@ -2,8 +2,8 @@
 MCP-style HTTP server for ScrAIbe.
 
 - Exposes an OpenAPI-compliant endpoint for external LLMs to:
-  - Upload audio
-  - Receive transcript JSON (no summary)
+  - Upload audio (as base64 or via URL) in a JSON body.
+  - Receive transcript JSON (no summary).
 - WebUI remains always enabled; this is additive.
 
 Configuration (env):
@@ -17,12 +17,14 @@ Configuration (env):
 import os
 import time
 import uuid
-import json
+import base64
+import tempfile
 import logging
 from typing import Optional
 
-from fastapi import FastAPI, UploadFile, File, Form, HTTPException
-from fastapi.responses import JSONResponse
+import httpx
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
 
 from .autotranscript import Scraibe
 
@@ -41,28 +43,95 @@ app = FastAPI(
 _mcp_jobs: dict = {}
 
 
+class TranscribeRequest(BaseModel):
+    audio_base64: Optional[str] = None
+    audio_url: Optional[str] = None
+    audio_path: Optional[str] = None
+    language: Optional[str] = None
+    num_speakers: Optional[int] = None
+
+
 def _job_id() -> str:
     return str(uuid.uuid4())
 
 
+def _save_audio_from_request(req: TranscribeRequest) -> str:
+    """
+    Save audio to a temporary file from base64, URL, or path.
+    Returns the local file path.
+    """
+    upload_dir = os.getenv("SCRAIBE_UPLOAD_DIR", "/tmp/scraibe_uploads")
+    os.makedirs(upload_dir, exist_ok=True)
+
+    if req.audio_base64:
+        # base64 upload
+        try:
+            data = base64.b64decode(req.audio_base64)
+        except Exception as e:
+            raise HTTPException(status_code=400, detail=f"Invalid base64 audio: {e}")
+
+        ts = time.strftime("%Y%m%d%H%M%S")
+        tmp_name = f"mcp_upload_{ts}_{uuid.uuid4().hex[:8]}.wav"
+        file_path = os.path.join(upload_dir, tmp_name)
+        with open(file_path, "wb") as f:
+            f.write(data)
+        return file_path
+
+    if req.audio_url:
+        # download from URL
+        try:
+            with httpx.stream("GET", req.audio_url, timeout=60) as resp:
+                if resp.status_code != 200:
+                    raise HTTPException(
+                        status_code=400,
+                        detail=f"Failed to download audio from URL: {resp.status_code}",
+                    )
+                ts = time.strftime("%Y%m%d%H%M%S")
+                tmp_name = f"mcp_url_{ts}_{uuid.uuid4().hex[:8]}.wav"
+                file_path = os.path.join(upload_dir, tmp_name)
+                with open(file_path, "wb") as f:
+                    for chunk in resp.iter_bytes():
+                        f.write(chunk)
+            return file_path
+        except HTTPException:
+            raise
+        except Exception as e:
+            raise HTTPException(status_code=400, detail=f"Error downloading audio: {e}")
+
+    if req.audio_path:
+        # use provided path (for internal use)
+        path = req.audio_path
+        if not os.path.isfile(path):
+            raise HTTPException(status_code=400, detail="audio_path does not exist")
+        return path
+
+    raise HTTPException(
+        status_code=400,
+        detail="Provide one of: audio_base64, audio_url, or audio_path",
+    )
+
+
 @app.get("/health")
 async def health():
     return {"status": "ok"}
 
 
 @app.post("/transcribe")
-async def transcribe(
-    file: UploadFile = File(...),
-    language: Optional[str] = Form(None),
-    num_speakers: Optional[int] = Form(None),
-):
+async def transcribe(req: TranscribeRequest):
     """
-    Upload audio and start transcription.
+    Submit an audio file for transcription.
+
+    Input (JSON body):
+      - audio_base64: base64-encoded audio file
+      - audio_url: URL to audio file
+      - audio_path: local file path (for internal use)
+      - language: (optional)
+      - num_speakers: (optional)
 
     Returns:
       {
         "job_id": "<id>",
-        "status": "queued" | "processing" | "completed" | "error",
+        "status": "queued" | "processing",
         "message": "..."
       }
 
@@ -70,21 +139,11 @@ async def transcribe(
     """
     use_celery = os.getenv("MCP_USE_CELERY", "true").strip().lower() in ("true", "1", "yes")
 
-    # Save uploaded file temporarily
+    # Save audio to a temporary file
     try:
-        import tempfile
-        from pathlib import Path
-
-        upload_dir = Path(os.getenv("SCRAIBE_UPLOAD_DIR", "/tmp/scraibe_uploads"))
-        upload_dir.mkdir(parents=True, exist_ok=True)
-
-        ext = Path(file.filename or "file").suffix or ".wav"
-        ts = time.strftime("%Y%m%d%H%M%S")
-        tmp_name = f"mcp_upload_{ts}_{uuid.uuid4().hex[:8]}{ext}"
-        file_path = upload_dir / tmp_name
-
-        content = await file.read()
-        file_path.write_bytes(content)
+        file_path = _save_audio_from_request(req)
+    except HTTPException:
+        raise
     except Exception as e:
         logger.error("Error saving MCP upload: %s", e)
         raise HTTPException(status_code=500, detail=f"Error saving file: {e}")
@@ -101,17 +160,17 @@ async def transcribe(
         if use_celery:
             try:
                 process_mcp_transcribe_task.delay(
-                    audio_path=str(file_path),
+                    audio_path=file_path,
                     job_id=job_id,
-                    language=language or None,
-                    num_speakers=int(num_speakers) if num_speakers else None,
+                    language=req.language or None,
+                    num_speakers=int(req.num_speakers) if req.num_speakers else None,
                 )
             except Exception as e:
                 logger.error("Error enqueuing MCP job: %s", e)
                 _mcp_jobs[job_id] = {
                     "status": "error",
                     "message": f"Error enqueuing job: {e}",
-                    "file_path": str(file_path),
+                    "file_path": file_path,
                 }
                 return {
                     "job_id": job_id,
@@ -122,7 +181,7 @@ async def transcribe(
             _mcp_jobs[job_id] = {
                 "status": "queued",
                 "message": "Job queued for processing.",
-                "file_path": str(file_path),
+                "file_path": file_path,
             }
             return {
                 "job_id": job_id,
@@ -134,16 +193,16 @@ async def transcribe(
     _mcp_jobs[job_id] = {
         "status": "processing",
         "message": "Transcription started (synchronous).",
-        "file_path": str(file_path),
+        "file_path": file_path,
     }
 
     def _run_sync():
         try:
             scraibe = Scraibe(verbose=False)
             result = scraibe.transcribe(
-                audio_file=str(file_path),
-                language=language or None,
-                num_speakers=int(num_speakers) if num_speakers else None,
+                audio_file=file_path,
+                language=req.language or None,
+                num_speakers=int(req.num_speakers) if req.num_speakers else None,
                 verbose=False,
                 for_export=True,
             )
@@ -196,10 +255,8 @@ async def get_json(job_id: str):
     transcript_text = job.get("transcript", "")
     segments = job.get("segments", [])
 
-    return JSONResponse(
-        content={
+    return {
         "job_id": job_id,
         "transcript": transcript_text,
         "segments": segments,
     }
-    )