Add prediction field to ModelError (#326)

This PR extends #325 to add the prediction object itself to
`ModelError`, as opposed to just its ID. This makes it convenient to
introspect logs and other information to determine how to handle the
failure.

```python
import replicate
from replicate.exceptions import ModelError

try:
  output = replicate.run("stability-ai/stable-diffusion-3", { "prompt": "..." })
except ModelError as e
  if "(some known issue)" in e.logs:
    pass

  print("Failed prediction: " + e.prediction.id)
```

---------

Signed-off-by: Rohan Mehta <[email protected]>
Signed-off-by: Mattt Zmuda <[email protected]>
Co-authored-by: Rohan Mehta <[email protected]>

README.md

@@ -82,6 +82,24 @@ or a handle to a file on your local device.
 "an astronaut riding a horse"
 ```
 
+`replicate.run` raises `ModelError` if the prediction fails.
+You can access the exception's `prediction` property 
+to get more information about the failure.
+
+```python
+import replicate
+from replicate.exceptions import ModelError
+
+try:
+  output = replicate.run("stability-ai/stable-diffusion-3", { "prompt": "An astronaut riding a rainbow unicorn" })
+except ModelError as e
+  if "(some known issue)" in e.logs:
+    pass
+
+  print("Failed prediction: " + e.prediction.id)
+```
+
+
 ## Run a model and stream its output
 
 Replicate’s API supports server-sent event streams (SSEs) for language models. 

replicate/exceptions.py

@@ -1,7 +1,10 @@
-from typing import Optional
+from typing import TYPE_CHECKING, Optional
 
 import httpx
 
+if TYPE_CHECKING:
+    from replicate.prediction import Prediction
+
 
 class ReplicateException(Exception):
     """A base class for all Replicate exceptions."""
@@ -10,6 +13,12 @@ class ReplicateException(Exception):
 class ModelError(ReplicateException):
     """An error from user's code in a model."""
 
+    prediction: "Prediction"
+
+    def __init__(self, prediction: "Prediction") -> None:
+        self.prediction = prediction
+        super().__init__(prediction.error)
+
 
 class ReplicateError(ReplicateException):
     """

replicate/prediction.py

@@ -249,7 +249,7 @@ def output_iterator(self) -> Iterator[Any]:
             self.reload()
 
         if self.status == "failed":
-            raise ModelError(self.error)
+            raise ModelError(self)
 
         output = self.output or []
         new_output = output[len(previous_output) :]
@@ -272,7 +272,7 @@ async def async_output_iterator(self) -> AsyncIterator[Any]:
             await self.async_reload()
 
         if self.status == "failed":
-            raise ModelError(self.error)
+            raise ModelError(self)
 
         output = self.output or []
         new_output = output[len(previous_output) :]

replicate/run.py

@@ -58,7 +58,7 @@ def run(
     prediction.wait()
 
     if prediction.status == "failed":
-        raise ModelError(prediction.error)
+        raise ModelError(prediction)
 
     return prediction.output
 
@@ -97,7 +97,7 @@ async def async_run(
     await prediction.async_wait()
 
     if prediction.status == "failed":
-        raise ModelError(prediction.error)
+        raise ModelError(prediction)
 
     return prediction.output
 

tests/test_run.py

@@ -7,7 +7,7 @@
 
 import replicate
 from replicate.client import Client
-from replicate.exceptions import ReplicateError
+from replicate.exceptions import ModelError, ReplicateError
 
 
 @pytest.mark.vcr("run.yaml")
@@ -184,3 +184,72 @@ def prediction_with_status(status: str) -> dict:
     )
 
     assert output == "Hello, world!"
+
+
+@pytest.mark.asyncio
+async def test_run_with_model_error(mock_replicate_api_token):
+    def prediction_with_status(status: str) -> dict:
+        return {
+            "id": "p1",
+            "model": "test/example",
+            "version": "v1",
+            "urls": {
+                "get": "https://api.replicate.com/v1/predictions/p1",
+                "cancel": "https://api.replicate.com/v1/predictions/p1/cancel",
+            },
+            "created_at": "2023-10-05T12:00:00.000000Z",
+            "source": "api",
+            "status": status,
+            "input": {"text": "world"},
+            "output": None,
+            "error": "OOM" if status == "failed" else None,
+            "logs": "",
+        }
+
+    router = respx.Router(base_url="https://api.replicate.com/v1")
+    router.route(method="POST", path="/predictions").mock(
+        return_value=httpx.Response(
+            201,
+            json=prediction_with_status("processing"),
+        )
+    )
+    router.route(method="GET", path="/predictions/p1").mock(
+        return_value=httpx.Response(
+            200,
+            json=prediction_with_status("failed"),
+        )
+    )
+    router.route(
+        method="GET",
+        path="/models/test/example/versions/v1",
+    ).mock(
+        return_value=httpx.Response(
+            201,
+            json={
+                "id": "f2d6b24e6002f25f77ae89c2b0a5987daa6d0bf751b858b94b8416e8542434d1",
+                "created_at": "2024-07-18T00:35:56.210272Z",
+                "cog_version": "0.9.10",
+                "openapi_schema": {
+                    "openapi": "3.0.2",
+                },
+            },
+        )
+    )
+    router.route(host="api.replicate.com").pass_through()
+
+    client = Client(
+        api_token="test-token", transport=httpx.MockTransport(router.handler)
+    )
+    client.poll_interval = 0.001
+
+    with pytest.raises(ModelError) as excinfo:
+        client.run(
+            "test/example:v1",
+            input={
+                "text": "Hello, world!",
+            },
+        )
+
+    assert str(excinfo.value) == "OOM"
+    assert excinfo.value.prediction.error == "OOM"
+    assert excinfo.value.prediction.status == "failed"