azure-search-documents

Ensure you are using an instance of `SearchClient` for document operations. You can obtain a `SearchClient` from a `SearchIndexClient` using `get_search_client()` or instantiate it directly.
```python
from azure.search.documents import SearchClient
from azure.core.credentials import AzureKeyCredential

service_endpoint = "YOUR_SEARCH_SERVICE_ENDPOINT"
index_name = "YOUR_INDEX_NAME"
key = "YOUR_ADMIN_API_KEY"

# Instantiate SearchClient directly for document operations
search_client = SearchClient(service_endpoint, index_name, AzureKeyCredential(key))

documents = [
    {"id": "1", "text": "This is a sample document."}
]
result = search_client.upload_documents(documents=documents)
# or for more granular control, use index_documents
# from azure.search.documents.models import IndexAction, IndexActionType
# actions = [IndexAction(document=doc, action_type=IndexActionType.UPLOAD) for doc in documents]
# result = search_client.index_documents(actions=actions)
```

Install the specific client library directly using pip. If you see this error, it means the required `azure-search-documents` (or its dependencies like `azure-core`) is not correctly installed or recognized in your environment.
```bash
pip install azure-search-documents
# If you also need core components or other Azure SDKs, install them explicitly:
pip install azure-core azure-identity
```

Verify that your API key is correct and has the necessary permissions (admin key for index management/document write, query key for document read). If using Azure AD, ensure the Managed Identity or Service Principal has been assigned appropriate roles (e.g., 'Search Service Contributor' for index creation, 'Search Index Data Contributor' for document write, 'Search Index Data Reader' for document read) on the Azure AI Search service.
```python
from azure.core.credentials import AzureKeyCredential
from azure.search.documents.indexes import SearchIndexClient

service_endpoint = "https://<your-search-service-name>.search.windows.net"
admin_key = "YOUR_ADMIN_API_KEY_FROM_PORTAL" # Ensure this is an admin key for index creation

credential = AzureKeyCredential(admin_key)
index_client = SearchIndexClient(endpoint=service_endpoint, credential=credential)

# Example: Attempting to create an index requires appropriate permissions
# index_client.create_index(SearchIndex(...))
```

Replace `SearchFieldDataType.Single` with `SearchFieldDataType.Double` when defining fields in your search index schema.
```python
from azure.search.documents.indexes.models import SearchField, SearchFieldDataType

# Incorrect
# SearchField(name="price", type=SearchFieldDataType.Single)

# Correct
SearchField(name="price", type=SearchFieldDataType.Double, searchable=True, filterable=True)
```

Specify the `content_key` parameter in the `AzureCognitiveSearchRetriever` constructor to match the actual name of the field in your Azure AI Search index that holds the main document content.
```python
import os
from langchain_community.retrievers import AzureCognitiveSearchRetriever

# Assume these are set as environment variables or directly defined
cognitive_search_name = os.environ.get("AZURE_COGNITIVE_SEARCH_SERVICE_NAME")
index_name = os.environ.get("AZURE_COGNITIVE_SEARCH_INDEX_NAME")
vector_store_address = f"https://{cognitive_search_name}.search.windows.net/"
vector_store_password = os.environ.get("AZURE_COGNITIVE_SEARCH_API_KEY")

# If your index field for content is 'text' instead of 'content'
retriever = AzureCognitiveSearchRetriever(
    service_name=cognitive_search_name,
    index_name=index_name,
    api_key=vector_store_password,
    content_key="text" # Change 'text' to whatever your actual content field name is
)

# Example usage:
# documents = retriever.invoke("your query")
```