feat: add MuAPI video generation tool

backend/director/tools/muapi_video.py

@@ -0,0 +1,210 @@
+import os
+import time
+import requests
+from typing import Optional
+
+
+PARAMS_CONFIG = {
+    "text_to_video": {
+        "model_name": {
+            "type": "string",
+            "description": "The muapi.ai model to use for text-to-video generation",
+            "default": "veo3-fast",
+            "enum": [
+                "veo3",
+                "veo3-fast",
+                "kling-master",
+                "wan2.1",
+                "wan2.2",
+                "seedance-pro",
+                "seedance-pro-fast",
+                "runway",
+                "pixverse",
+                "hunyuan",
+                "minimax-hailuo-02-std",
+                "minimax-hailuo-02-pro",
+            ],
+        },
+        "duration": {
+            "type": "integer",
+            "description": "Duration of the video in seconds",
+            "default": 5,
+            "minimum": 3,
+            "maximum": 60,
+        },
+        "aspect_ratio": {
+            "type": "string",
+            "description": "Aspect ratio of the generated video",
+            "default": "16:9",
+            "enum": ["16:9", "9:16", "1:1", "4:3", "3:4"],
+        },
+    },
+    "image_to_video": {
+        "model_name": {
+            "type": "string",
+            "description": "The muapi.ai model to use for image-to-video generation",
+            "default": "kling-master",
+            "enum": [
+                "kling-master",
+                "kling-v2.5-pro",
+                "wan2.1",
+                "wan2.2",
+                "seedance-pro",
+                "seedance-pro-fast",
+                "runway",
+                "pixverse",
+                "hunyuan",
+                "vidu",
+            ],
+        },
+        "duration": {
+            "type": "integer",
+            "description": "Duration of the video in seconds",
+            "default": 5,
+            "minimum": 3,
+            "maximum": 60,
+        },
+        "aspect_ratio": {
+            "type": "string",
+            "description": "Aspect ratio of the generated video",
+            "default": "16:9",
+            "enum": ["16:9", "9:16", "1:1", "4:3", "3:4"],
+        },
+    },
+}
+
+# Endpoints are the model IDs themselves for muapi
+T2V_ENDPOINTS = {
+    "veo3": "veo3",
+    "veo3-fast": "veo3-fast",
+    "kling-master": "kling",
+    "wan2.1": "wan2.1",
+    "wan2.2": "wan2.2",
+    "seedance-pro": "seedance-pro",
+    "seedance-pro-fast": "seedance-pro-fast",
+    "runway": "runway",
+    "pixverse": "pixverse",
+    "hunyuan": "hunyuan",
+    "minimax-hailuo-02-std": "minimax-hailuo-02-std",
+    "minimax-hailuo-02-pro": "minimax-hailuo-02-pro",
+}
+
+I2V_ENDPOINTS = {
+    "kling-master": "kling-i2v",
+    "kling-v2.5-pro": "kling-v2.5-pro-i2v",
+    "wan2.1": "wan2.1-i2v",
+    "wan2.2": "wan2.2-i2v",
+    "seedance-pro": "seedance-pro-i2v",
+    "seedance-pro-fast": "seedance-pro-fast-i2v",
+    "runway": "runway-i2v",
+    "pixverse": "pixverse-i2v",
+    "hunyuan": "hunyuan-i2v",
+    "vidu": "vidu-i2v",
+}
+
+
+class MuApiVideoGenerationTool:
+    """Video generation tool using muapi.ai's 400+ model aggregator API."""
+
+    BASE_URL = "https://api.muapi.ai/api/v1"
+    POLL_INTERVAL = 5  # seconds
+
+    def __init__(self, api_key: str):
+        if not api_key:
+            raise Exception(
+                "MUAPI_API_KEY not found. Get one at https://muapi.ai/dashboard/api-keys"
+            )
+        self.api_key = api_key
+        self.session = requests.Session()
+        self.session.headers.update(
+            {"x-api-key": self.api_key, "Content-Type": "application/json"}
+        )
+
+    def _submit(self, endpoint: str, payload: dict) -> str:
+        """Submit a generation request and return the request_id."""
+        resp = self.session.post(f"{self.BASE_URL}/{endpoint}", json=payload, timeout=30)
+        resp.raise_for_status()
+        try:
+            request_id = resp.json()["request_id"]
+        except KeyError as e:
+            raise Exception(
+                f"MuAPI did not return a request_id. Response: {resp.text}"
+            ) from e
+        return request_id
+
+    def _poll(self, request_id: str, timeout: int = 600) -> str:
+        """Poll until the generation completes and return the output URL."""
+        deadline = time.time() + timeout
+        while time.time() < deadline:
+            resp = self.session.get(
+                f"{self.BASE_URL}/predictions/{request_id}/result", timeout=15
+            )
+            resp.raise_for_status()
+            data = resp.json()
+            status = data.get("status", "pending")
+            if status == "completed":
+                outputs = data.get("outputs", [])
+                if not outputs:
+                    raise Exception("Completed but no outputs returned")
+                return outputs[0]
+            if status in ("failed", "cancelled"):
+                raise Exception(f"Video generation {status}: {data.get('error', '')}")
+            time.sleep(self.POLL_INTERVAL)
+        raise Exception(f"Video generation timed out after {timeout}s")
+
+    def text_to_video(
+        self, prompt: str, save_at: str, duration: float, config: dict
+    ) -> dict:
+        """Generate a video from a text prompt using muapi.ai."""
+        model_name = config.get("model_name", "veo3-fast")
+        endpoint = T2V_ENDPOINTS.get(model_name, model_name)
+
+        payload = {
+            "prompt": prompt,
+            "duration": int(duration),
+            "aspect_ratio": config.get("aspect_ratio", "16:9"),
+        }
+
+        try:
+            request_id = self._submit(endpoint, payload)
+            video_url = self._poll(request_id)
+            video_data = requests.get(video_url, timeout=120)
+            video_data.raise_for_status()
+            with open(save_at, "wb") as f:
+                f.write(video_data.content)
+        except Exception as e:
+            raise Exception(f"Error generating video: {type(e).__name__}: {str(e)}") from e
+
+        return {"status": "success", "video_path": save_at}
+
+    def image_to_video(
+        self,
+        image_url: str,
+        save_at: str,
+        duration: float,
+        config: dict,
+        prompt: Optional[str] = None,
+    ) -> dict:
+        """Generate a video from an image URL using muapi.ai."""
+        model_name = config.get("model_name", "kling-master")
+        endpoint = I2V_ENDPOINTS.get(model_name, f"{model_name}-i2v")
+
+        payload: dict = {
+            "image_url": image_url,
+            "duration": int(duration),
+            "aspect_ratio": config.get("aspect_ratio", "16:9"),
+        }
+        if prompt:
+            payload["prompt"] = prompt
+
+        try:
+            request_id = self._submit(endpoint, payload)
+            video_url = self._poll(request_id)
+            video_data = requests.get(video_url, timeout=120)
+            video_data.raise_for_status()
+            with open(save_at, "wb") as f:
+                f.write(video_data.content)
+        except Exception as e:
+            raise Exception(f"Error generating video: {type(e).__name__}: {str(e)}") from e
+
+        return {"status": "success", "video_path": save_at}