unstructured-ingest 0.3.11__py3-none-any.whl → 0.3.12__py3-none-any.whl

test/integration/connectors/test_milvus.py

@@ -174,6 +174,19 @@ def test_precheck_fails_on_nonexistent_collection(collection: str):
         uploader.precheck()
 
 
+@pytest.mark.tags(CONNECTOR_TYPE, DESTINATION_TAG)
+def test_precheck_fails_on_nonexisting_db(collection: str):
+    uploader = MilvusUploader(
+        connection_config=MilvusConnectionConfig(uri=DB_URI),
+        upload_config=MilvusUploaderConfig(db_name="nonexisting_db", collection_name=collection),
+    )
+    with pytest.raises(
+        DestinationConnectionError,
+        match="database not found",
+    ):
+        uploader.precheck()
+
+
 @pytest.mark.parametrize("upload_file_str", ["upload_file_ndjson", "upload_file"])
 def test_milvus_stager(
     request: TopRequest,

test/integration/connectors/test_onedrive.py

@@ -20,6 +20,9 @@ from unstructured_ingest.v2.processes.connectors.onedrive import (
 
 
 @pytest.fixture
+@pytest.mark.xfail(
+    reason="Issues with test setup on the provider side."
+)  # TODO: remove line when issues are addressed
 def onedrive_test_folder() -> str:
     """
     Pytest fixture that creates a test folder in OneDrive and deletes it after test run.
@@ -66,6 +69,9 @@ def get_connection_config():
 
 @pytest.mark.tags(CONNECTOR_TYPE, DESTINATION_TAG)
 @requires_env("MS_CLIENT_CRED", "MS_CLIENT_ID", "MS_TENANT_ID", "MS_USER_PNAME")
+@pytest.mark.xfail(
+    reason="Issues with test setup on the provider side."
+)  # TODO: remove line when issues are addressed
 def test_onedrive_destination(upload_file: Path, onedrive_test_folder: str):
     """
     Integration test for the OneDrive destination connector.

test/integration/connectors/test_redis.py

@@ -0,0 +1,119 @@
+import asyncio
+import json
+import os
+from pathlib import Path
+from typing import Optional
+
+import numpy as np
+import pytest
+from redis import exceptions as redis_exceptions
+from redis.asyncio import Redis, from_url
+
+from test.integration.connectors.utils.constants import DESTINATION_TAG
+from test.integration.utils import requires_env
+from unstructured_ingest.v2.interfaces.file_data import FileData, SourceIdentifiers
+from unstructured_ingest.v2.processes.connectors.redisdb import (
+    CONNECTOR_TYPE as REDIS_CONNECTOR_TYPE,
+)
+from unstructured_ingest.v2.processes.connectors.redisdb import (
+    RedisAccessConfig,
+    RedisConnectionConfig,
+    RedisUploader,
+    RedisUploaderConfig,
+)
+
+
+async def delete_record(client: Redis, element_id: str) -> None:
+    await client.delete(element_id)
+
+
+async def validate_upload(client: Redis, first_element: dict):
+    element_id = first_element["element_id"]
+    expected_text = first_element["text"]
+    expected_embeddings = first_element["embeddings"]
+    async with client.pipeline(transaction=True) as pipe:
+        try:
+            response = await pipe.json().get(element_id, "$").execute()
+            response = response[0][0]
+        except redis_exceptions.ResponseError:
+            response = await pipe.get(element_id).execute()
+            response = json.loads(response[0])
+
+    embedding_similarity = np.linalg.norm(
+        np.array(response["embeddings"]) - np.array(expected_embeddings)
+    )
+
+    assert response is not None
+    assert response["element_id"] == element_id
+    assert response["text"] == expected_text
+    assert embedding_similarity < 1e-10
+
+
+async def redis_destination_test(
+    upload_file: Path,
+    tmp_path: Path,
+    connection_kwargs: dict,
+    uri: Optional[str] = None,
+    password: Optional[str] = None,
+):
+    uploader = RedisUploader(
+        connection_config=RedisConnectionConfig(
+            **connection_kwargs, access_config=RedisAccessConfig(uri=uri, password=password)
+        ),
+        upload_config=RedisUploaderConfig(batch_size=10),
+    )
+
+    file_data = FileData(
+        source_identifiers=SourceIdentifiers(fullpath=upload_file.name, filename=upload_file.name),
+        connector_type=REDIS_CONNECTOR_TYPE,
+        identifier="mock-file-data",
+    )
+    with upload_file.open() as upload_fp:
+        elements = json.load(upload_fp)
+    first_element = elements[0]
+
+    try:
+        if uploader.is_async():
+            await uploader.run_data_async(data=elements, file_data=file_data)
+
+        if uri:
+            async with from_url(uri) as client:
+                await validate_upload(client=client, first_element=first_element)
+        else:
+            async with Redis(**connection_kwargs, password=password) as client:
+                await validate_upload(client=client, first_element=first_element)
+    except Exception as e:
+        raise e
+    finally:
+        if uri:
+            async with from_url(uri) as client:
+                tasks = [delete_record(client, element["element_id"]) for element in elements]
+                await asyncio.gather(*tasks)
+        else:
+            async with Redis(**connection_kwargs, password=password) as client:
+                tasks = [delete_record(client, element["element_id"]) for element in elements]
+                await asyncio.gather(*tasks)
+
+
+@pytest.mark.asyncio
+@pytest.mark.tags(REDIS_CONNECTOR_TYPE, DESTINATION_TAG)
+@requires_env("AZURE_REDIS_INGEST_TEST_PASSWORD")
+async def test_redis_destination_azure_with_password(upload_file: Path, tmp_path: Path):
+    connection_kwargs = {
+        "host": "utic-dashboard-dev.redis.cache.windows.net",
+        "port": 6380,
+        "db": 0,
+        "ssl": True,
+    }
+    redis_pw = os.environ["AZURE_REDIS_INGEST_TEST_PASSWORD"]
+    await redis_destination_test(upload_file, tmp_path, connection_kwargs, password=redis_pw)
+
+
+@pytest.mark.asyncio
+@pytest.mark.tags(REDIS_CONNECTOR_TYPE, DESTINATION_TAG, "redis")
+@requires_env("AZURE_REDIS_INGEST_TEST_PASSWORD")
+async def test_redis_destination_azure_with_uri(upload_file: Path, tmp_path: Path):
+    connection_kwargs = {}
+    redis_pw = os.environ["AZURE_REDIS_INGEST_TEST_PASSWORD"]
+    uri = f"rediss://:{redis_pw}@utic-dashboard-dev.redis.cache.windows.net:6380/0"
+    await redis_destination_test(upload_file, tmp_path, connection_kwargs, uri=uri)

test/integration/connectors/test_vectara.py

@@ -0,0 +1,270 @@
+import json
+import os
+import time
+from pathlib import Path
+from typing import Generator
+from uuid import uuid4
+
+import pytest
+import requests
+
+from test.integration.connectors.utils.constants import DESTINATION_TAG
+from test.integration.utils import requires_env
+from unstructured_ingest.v2.interfaces.file_data import FileData, SourceIdentifiers
+from unstructured_ingest.v2.logger import logger
+from unstructured_ingest.v2.processes.connectors.vectara import (
+    CONNECTOR_TYPE as VECTARA_CONNECTOR_TYPE,
+)
+from unstructured_ingest.v2.processes.connectors.vectara import (
+    VectaraAccessConfig,
+    VectaraConnectionConfig,
+    VectaraUploader,
+    VectaraUploaderConfig,
+    VectaraUploadStager,
+    VectaraUploadStagerConfig,
+)
+
+
+def validate_upload(response: dict, expected_data: dict):
+    element_id = expected_data["element_id"]
+    expected_text = expected_data["text"]
+    filename = expected_data["metadata"]["filename"]
+    filetype = expected_data["metadata"]["filetype"]
+    page_number = expected_data["metadata"]["page_number"]
+
+    response = response["search_results"][0]
+
+    assert response is not None
+    assert response["text"] == expected_text
+    assert response["part_metadata"]["element_id"] == element_id
+    assert response["part_metadata"]["filename"] == filename
+    assert response["part_metadata"]["filetype"] == filetype
+    assert response["part_metadata"]["page_number"] == page_number
+
+
+@requires_env("VECTARA_OAUTH_CLIENT_ID", "VECTARA_OAUTH_SECRET", "VECTARA_CUSTOMER_ID")
+def _get_jwt_token():
+    """Connect to the server and get a JWT token."""
+    customer_id = os.environ["VECTARA_CUSTOMER_ID"]
+    token_endpoint = (
+        f"https://vectara-prod-{customer_id}.auth.us-west-2.amazoncognito.com/oauth2/token"
+    )
+    headers = {
+        "Content-Type": "application/x-www-form-urlencoded",
+    }
+    data = {
+        "grant_type": "client_credentials",
+        "client_id": os.environ["VECTARA_OAUTH_CLIENT_ID"],
+        "client_secret": os.environ["VECTARA_OAUTH_SECRET"],
+    }
+
+    response = requests.post(token_endpoint, headers=headers, data=data)
+    response.raise_for_status()
+    response_json = response.json()
+
+    return response_json.get("access_token")
+
+
+def query_data(corpus_key: str, element_id: str) -> dict:
+
+    url = f"https://api.vectara.io/v2/corpora/{corpus_key}/query"
+
+    # the query below requires the corpus to have filter attributes for element_id
+
+    data = json.dumps(
+        {
+            "query": "string",
+            "search": {
+                "metadata_filter": f"part.element_id = '{element_id}'",
+                "lexical_interpolation": 1,
+                "limit": 10,
+            },
+        }
+    )
+
+    jwt_token = _get_jwt_token()
+    headers = {
+        "Content-Type": "application/json",
+        "Accept": "application/json",
+        "Authorization": f"Bearer {jwt_token}",
+        "X-source": "unstructured",
+    }
+
+    response = requests.post(url, headers=headers, data=data)
+    response.raise_for_status()
+    response_json = response.json()
+
+    return response_json
+
+
+def create_corpora(corpus_key: str, corpus_name: str) -> None:
+    url = "https://api.vectara.io/v2/corpora"
+    data = json.dumps({"key": corpus_key, "name": corpus_name, "description": "integration test"})
+    jwt_token = _get_jwt_token()
+    headers = {
+        "Content-Type": "application/json",
+        "Accept": "application/json",
+        "Authorization": f"Bearer {jwt_token}",
+        "X-source": "unstructured",
+    }
+
+    response = requests.post(url, headers=headers, data=data)
+    response.raise_for_status()
+
+
+def replace_filter_attributes(corpus_key: str) -> None:
+    url = f"https://api.vectara.io/v2/corpora/{corpus_key}/replace_filter_attributes"
+    data = json.dumps(
+        {
+            "filter_attributes": [
+                {"name": "element_id", "level": "part", "indexed": True, "type": "text"}
+            ]
+        }
+    )
+    jwt_token = _get_jwt_token()
+    headers = {
+        "Content-Type": "application/json",
+        "Accept": "application/json",
+        "Authorization": f"Bearer {jwt_token}",
+        "X-source": "unstructured",
+    }
+
+    response = requests.post(url, headers=headers, data=data)
+    response.raise_for_status()
+
+
+def delete_corpora(corpus_key: str) -> None:
+    url = f"https://api.vectara.io/v2/corpora/{corpus_key}"
+
+    jwt_token = _get_jwt_token()
+    headers = {
+        "Content-Type": "application/json",
+        "Accept": "application/json",
+        "Authorization": f"Bearer {jwt_token}",
+        "X-source": "unstructured",
+    }
+
+    response = requests.delete(url, headers=headers)
+    response.raise_for_status()
+
+
+def list_corpora() -> list:
+    url = "https://api.vectara.io/v2/corpora?limit=100"
+    jwt_token = _get_jwt_token()
+    headers = {
+        "Content-Type": "application/json",
+        "Accept": "application/json",
+        "Authorization": f"Bearer {jwt_token}",
+        "X-source": "unstructured",
+    }
+    response = requests.get(url, headers=headers)
+    response.raise_for_status()
+    response_json = response.json()
+    if response_json.get("corpora"):
+        return [item["key"] for item in response_json.get("corpora")]
+    else:
+        return []
+
+
+def wait_for_ready(corpus_key: str, timeout=60, interval=2) -> None:
+    def is_ready_status():
+        corpora_list = list_corpora()
+        return corpus_key in corpora_list
+
+    start = time.time()
+    is_ready = is_ready_status()
+    while not is_ready and time.time() - start < timeout:
+        time.sleep(interval)
+        is_ready = is_ready_status()
+    if not is_ready:
+        raise TimeoutError("time out waiting for corpus to be ready")
+
+
+def wait_for_delete(corpus_key: str, timeout=60, interval=2) -> None:
+    start = time.time()
+    while time.time() - start < timeout:
+        corpora_list = list_corpora()
+        if corpus_key not in corpora_list:
+            return
+        time.sleep(interval)
+
+    raise TimeoutError("time out waiting for corpus to delete")
+
+
+@pytest.fixture
+def corpora_util() -> Generator[str, None, None]:
+    random_id = str(uuid4()).split("-")[0]
+    corpus_key = f"ingest-test-{random_id}"
+    corpus_name = "ingest-test"
+    logger.info(f"Creating corpus with key: {corpus_key}")
+    try:
+        create_corpora(corpus_key, corpus_name)
+        replace_filter_attributes(corpus_key)
+        wait_for_ready(corpus_key=corpus_key)
+        yield corpus_key
+    except Exception as e:
+        logger.error(f"failed to create corpus {corpus_key}: {e}")
+    finally:
+        logger.info(f"deleting corpus: {corpus_key}")
+        delete_corpora(corpus_key)
+        wait_for_delete(corpus_key=corpus_key)
+
+
+@pytest.mark.asyncio
+@pytest.mark.tags(VECTARA_CONNECTOR_TYPE, DESTINATION_TAG, "vectara")
+@requires_env("VECTARA_OAUTH_CLIENT_ID", "VECTARA_OAUTH_SECRET", "VECTARA_CUSTOMER_ID")
+async def test_vectara_destination(
+    upload_file: Path, tmp_path: Path, corpora_util: str, retries=30, interval=10
+):
+    corpus_key = corpora_util
+    connection_kwargs = {
+        "customer_id": os.environ["VECTARA_CUSTOMER_ID"],
+        "corpus_key": corpus_key,
+    }
+
+    oauth_client_id = os.environ["VECTARA_OAUTH_CLIENT_ID"]
+    oauth_secret = os.environ["VECTARA_OAUTH_SECRET"]
+
+    file_data = FileData(
+        source_identifiers=SourceIdentifiers(fullpath=upload_file.name, filename=upload_file.name),
+        connector_type=VECTARA_CONNECTOR_TYPE,
+        identifier="mock-file-data",
+    )
+
+    stager_config = VectaraUploadStagerConfig(batch_size=10)
+    stager = VectaraUploadStager(upload_stager_config=stager_config)
+    new_upload_file = stager.run(
+        elements_filepath=upload_file,
+        output_dir=tmp_path,
+        output_filename=upload_file.name,
+        file_data=file_data,
+    )
+
+    uploader = VectaraUploader(
+        connection_config=VectaraConnectionConfig(
+            **connection_kwargs,
+            access_config=VectaraAccessConfig(
+                oauth_client_id=oauth_client_id, oauth_secret=oauth_secret
+            ),
+        ),
+        upload_config=VectaraUploaderConfig(),
+    )
+
+    with new_upload_file.open() as new_upload_fp:
+        elements_stager = json.load(new_upload_fp)
+
+    if uploader.is_async():
+        await uploader.run_data_async(data=elements_stager, file_data=file_data)
+
+    with upload_file.open() as upload_fp:
+        elements = json.load(upload_fp)
+    first_element = elements[0]
+
+    for i in range(retries):
+        response = query_data(corpus_key, first_element["element_id"])
+        if not response["search_results"]:
+            time.sleep(interval)
+        else:
+            break
+
+    validate_upload(response=response, expected_data=first_element)

test/integration/embedders/test_bedrock.py

@@ -2,9 +2,12 @@ import json
 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 json
 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 json
 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 json
 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()

test/integration/partitioners/test_partitioner.py

@@ -3,9 +3,9 @@ import os
 from pathlib import Path
 
 import pytest
-from unstructured_client.models.errors.sdkerror import SDKError
 
 from test.integration.utils import requires_env
+from unstructured_ingest.v2.errors import UserError
 from unstructured_ingest.v2.processes.partitioner import Partitioner, PartitionerConfig
 
 int_test_dir = Path(__file__).parent
@@ -71,5 +71,5 @@ async def test_partitioner_api_fast_error(partition_file: Path):
         strategy="fast", partition_by_api=True, api_key=api_key, partition_endpoint=api_url
     )
     partitioner = Partitioner(config=partitioner_config)
-    with pytest.raises(SDKError):
+    with pytest.raises(UserError):
         await partitioner.run_async(filename=partition_file)

test/unit/embed/test_octoai.py

@@ -4,7 +4,14 @@ from unstructured_ingest.embed.octoai import OctoAiEmbeddingConfig, OctoAIEmbedd
 def test_embed_documents_does_not_break_element_to_dict(mocker):
     # Mocked client with the desired behavior for embed_documents
     mock_client = mocker.MagicMock()
-    mock_client.embed_documents.return_value = [1, 2]
+    mock_data = []
+    for i in range(2):
+        data = mocker.MagicMock()
+        data.embedding = [1, 2]
+        mock_data.append(data)
+    mock_response = mocker.MagicMock()
+    mock_response.data = mock_data
+    mock_client.embeddings.create.return_value = mock_response
 
     # Mock get_client to return our mock_client
     mocker.patch.object(OctoAiEmbeddingConfig, "get_client", return_value=mock_client)

unstructured_ingest/__version__.py

@@ -1 +1 @@
-__version__ = "0.3.11"  # pragma: no cover
+__version__ = "0.3.12"  # pragma: no cover

unstructured_ingest/embed/bedrock.py

@@ -6,7 +6,9 @@ from typing import TYPE_CHECKING
 from pydantic import Field, SecretStr
 
 from unstructured_ingest.embed.interfaces import BaseEmbeddingEncoder, EmbeddingConfig
+from unstructured_ingest.logger import logger
 from unstructured_ingest.utils.dep_check import requires_dependencies
+from unstructured_ingest.v2.errors import ProviderError, RateLimitError, UserAuthError, UserError
 
 if TYPE_CHECKING:
     from botocore.client import BaseClient
@@ -44,6 +46,32 @@ class BedrockEmbeddingConfig(EmbeddingConfig):
 class BedrockEmbeddingEncoder(BaseEmbeddingEncoder):
     config: BedrockEmbeddingConfig
 
+    def wrap_error(self, e: Exception) -> Exception:
+        from botocore.exceptions import ClientError
+
+        if isinstance(e, ClientError):
+            # https://docs.aws.amazon.com/awssupport/latest/APIReference/CommonErrors.html
+            http_response = e.response
+            meta = http_response["ResponseMetadata"]
+            http_response_code = meta["HTTPStatusCode"]
+            error_code = http_response["Error"]["Code"]
+            if http_response_code == 400:
+                if error_code == "ValidationError":
+                    return UserError(http_response["Error"])
+                elif error_code == "ThrottlingException":
+                    return RateLimitError(http_response["Error"])
+                elif error_code == "NotAuthorized" or error_code == "AccessDeniedException":
+                    return UserAuthError(http_response["Error"])
+            if http_response_code == 403:
+                return UserAuthError(http_response["Error"])
+            if 400 <= http_response_code < 500:
+                return UserError(http_response["Error"])
+            if http_response_code >= 500:
+                return ProviderError(http_response["Error"])
+
+        logger.error(f"unhandled exception from bedrock: {e}", exc_info=True)
+        return e
+
     def embed_query(self, query: str) -> list[float]:
         """Call out to Bedrock embedding endpoint."""
         # replace newlines, which can negatively affect performance.
@@ -61,25 +89,25 @@ class BedrockEmbeddingEncoder(BaseEmbeddingEncoder):
             input_body["inputText"] = text
         body = json.dumps(input_body)
 
+        bedrock_client = self.config.get_client()
+        # invoke bedrock API
         try:
-            bedrock_client = self.config.get_client()
-            # invoke bedrock API
             response = bedrock_client.invoke_model(
                 body=body,
                 modelId=self.config.embed_model_name,
                 accept="application/json",
                 contentType="application/json",
             )
-
-            # format output based on provider
-            response_body = json.loads(response.get("body").read())
-            if provider == "cohere":
-                return response_body.get("embeddings")[0]
-            else:
-                # includes common provider == "amazon"
-                return response_body.get("embedding")
         except Exception as e:
-            raise ValueError(f"Error raised by inference endpoint: {e}")
+            raise self.wrap_error(e=e)
+
+        # format output based on provider
+        response_body = json.loads(response.get("body").read())
+        if provider == "cohere":
+            return response_body.get("embeddings")[0]
+        else:
+            # includes common provider == "amazon"
+            return response_body.get("embedding")
 
     def embed_documents(self, elements: list[dict]) -> list[dict]:
         embeddings = [self.embed_query(query=e.get("text", "")) for e in elements]

unstructured_ingest/embed/interfaces.py

@@ -17,6 +17,11 @@ class BaseEmbeddingEncoder(ABC):
         """Initializes the embedding encoder class. Should also validate the instance
         is properly configured: e.g., embed a single a element"""
 
+    def wrap_error(self, e: Exception) -> Exception:
+        """Handle errors from the embedding service. Should raise a more informative error
+        if possible"""
+        return e
+
     @property
     def num_of_dimensions(self) -> tuple[int, ...]:
         exemplary_embedding = self.get_exemplary_embedding()