feat/add better custom errors (#315)

* add better custom errors

* share same error handling code between async and non async partitioner code

* Fix async call

* Add error handling to openai embedder

* Add error handling to bedrock embedder

* Add error handling to octoai embedder

* Add error handling to togetherai embedder

* differentiate from rate limit and quote max in openai

* Update method to wrap error rather than raise to help with typing

* Add error handling to voyageai embedder

* fix unit test

Author: rbiseck3

CHANGELOG.md

@@ -1,9 +1,4 @@
-## 0.3.12-dev1
-
-* **Bypass asyncio exception grouping to return more meaningful errors from OneDrive indexer**
-
-
-## 0.3.12-dev0
+## 0.3.12-dev2
 
 ### Fixes
 
@@ -12,6 +7,9 @@
 ### Enhancements
 
 * **Kafka destination connector checks for existence of topic**
+* **Create more reflective custom errors** Provide errors to indicate if the error was due to something user provided or due to a provider issue, applicable to all steps in the pipeline.
+* **Bypass asyncio exception grouping to return more meaningful errors from OneDrive indexer**
+
 
 ## 0.3.11
 

test/integration/embedders/test_bedrock.py

@@ -2,9 +2,12 @@
 import os
 from pathlib import Path
 
+import pytest
+
 from test.integration.embedders.utils import validate_embedding_output, validate_raw_embedder
 from test.integration.utils import requires_env
 from unstructured_ingest.embed.bedrock import BedrockEmbeddingConfig, BedrockEmbeddingEncoder
+from unstructured_ingest.v2.errors import UserAuthError, UserError
 from unstructured_ingest.v2.processes.embedder import Embedder, EmbedderConfig
 
 
@@ -47,3 +50,28 @@ def test_raw_bedrock_embedder(embedder_file: Path):
         expected_dimensions=(1536,),
         expected_is_unit_vector=False,
     )
+
+
+def test_raw_bedrock_embedder_invalid_credentials(embedder_file: Path):
+    embedder = BedrockEmbeddingEncoder(
+        config=BedrockEmbeddingConfig(
+            aws_access_key_id="no_key",
+            aws_secret_access_key="no_secret",
+        )
+    )
+    with pytest.raises(UserAuthError):
+        embedder.get_exemplary_embedding()
+
+
+@requires_env("AWS_ACCESS_KEY_ID", "AWS_SECRET_ACCESS_KEY")
+def test_raw_bedrock_embedder_invalid_model(embedder_file: Path):
+    aws_credentials = get_aws_credentials()
+    embedder = BedrockEmbeddingEncoder(
+        config=BedrockEmbeddingConfig(
+            aws_access_key_id=aws_credentials["aws_access_key_id"],
+            aws_secret_access_key=aws_credentials["aws_secret_access_key"],
+            model_name="invalid_model",
+        )
+    )
+    with pytest.raises(UserError):
+        embedder.get_exemplary_embedding()

test/integration/embedders/test_octoai.py

@@ -2,9 +2,12 @@
 import os
 from pathlib import Path
 
+import pytest
+
 from test.integration.embedders.utils import validate_embedding_output, validate_raw_embedder
 from test.integration.utils import requires_env
 from unstructured_ingest.embed.octoai import OctoAiEmbeddingConfig, OctoAIEmbeddingEncoder
+from unstructured_ingest.v2.errors import UserAuthError
 from unstructured_ingest.v2.processes.embedder import Embedder, EmbedderConfig
 
 API_KEY = "OCTOAI_API_KEY"
@@ -39,3 +42,14 @@ def test_raw_octoai_embedder(embedder_file: Path):
     validate_raw_embedder(
         embedder=embedder, embedder_file=embedder_file, expected_dimensions=(1024,)
     )
+
+
+@pytest.mark.skip(reason="Unexpected connection error at the moment")
+def test_raw_octoai_embedder_invalid_credentials():
+    embedder = OctoAIEmbeddingEncoder(
+        config=OctoAiEmbeddingConfig(
+            api_key="fake_api_key",
+        )
+    )
+    with pytest.raises(UserAuthError):
+        embedder.get_exemplary_embedding()

test/integration/embedders/test_openai.py

@@ -2,9 +2,12 @@
 import os
 from pathlib import Path
 
+import pytest
+
 from test.integration.embedders.utils import validate_embedding_output, validate_raw_embedder
 from test.integration.utils import requires_env
 from unstructured_ingest.embed.openai import OpenAIEmbeddingConfig, OpenAIEmbeddingEncoder
+from unstructured_ingest.v2.errors import UserAuthError
 from unstructured_ingest.v2.processes.embedder import Embedder, EmbedderConfig
 
 API_KEY = "OPENAI_API_KEY"
@@ -39,3 +42,13 @@ def test_raw_openai_embedder(embedder_file: Path):
     validate_raw_embedder(
         embedder=embedder, embedder_file=embedder_file, expected_dimensions=(1536,)
     )
+
+
+def test_raw_openai_embedder_invalid_credentials():
+    embedder = OpenAIEmbeddingEncoder(
+        config=OpenAIEmbeddingConfig(
+            api_key="fake_api_key",
+        )
+    )
+    with pytest.raises(UserAuthError):
+        embedder.get_exemplary_embedding()

test/integration/embedders/test_togetherai.py

@@ -2,12 +2,15 @@
 import os
 from pathlib import Path
 
+import pytest
+
 from test.integration.embedders.utils import validate_embedding_output, validate_raw_embedder
 from test.integration.utils import requires_env
 from unstructured_ingest.embed.togetherai import (
     TogetherAIEmbeddingConfig,
     TogetherAIEmbeddingEncoder,
 )
+from unstructured_ingest.v2.errors import UserAuthError
 from unstructured_ingest.v2.processes.embedder import Embedder, EmbedderConfig
 
 API_KEY = "TOGETHERAI_API_KEY"
@@ -41,3 +44,10 @@ def test_raw_togetherai_embedder(embedder_file: Path):
         expected_dimensions=(768,),
         expected_is_unit_vector=False,
     )
+
+
+def test_raw_togetherai_embedder_invalid_credentials():
+    embedder = TogetherAIEmbeddingEncoder(config=TogetherAIEmbeddingConfig(api_key="fake_api_key"))
+
+    with pytest.raises(UserAuthError):
+        embedder.get_exemplary_embedding()