ā” Programmatic API Access with `gradio_client`
Every Gradio application automatically exposes an API endpoint. Using the gradio_client library, Python scripts, backend web servers (like PHP or FastAPI), and automated bots can connect to any hosted Gradio space or local server to execute predictions programmatically.
Example 1: Definition & Basic `Client()` Connection
Connects to a remote or local Gradio application instance using the
Client class.
Python (client_ex1.py)
from gradio_client import Client
# Connect to a hosted Gradio application or Hugging Face Space
client = Client("abidlabs/huggingface-logo")
# Inspect available API endpoints and parameters
print(client.view_api())
š„ļø Live Interactive Web UI Simulator
### Basic Client Connection
Click 'Inspect API Schema' to view endpoint definitions.
Example 2: Executing Predictions via `client.predict()`
Sends parameters to a specific API endpoint index or named route and retrieves prediction results programmatically.
Python (client_ex2.py)
from gradio_client import Client
client = Client("abidlabs/en-to-es-translator")
# Execute prediction by passing parameters positionally
result = client.predict(
text="Welcome to Nikita-Infotech masterclass.",
api_name="/predict"
)
print("Translation Result:", result)
š„ļø Live Interactive Web UI Simulator
### Predict API Execution
Click 'Execute client.predict()' to run remote prediction.
Example 3: Streaming API Responses with `client.submit()`
Initiates an asynchronous background prediction job using
client.submit() and iterates through streaming updates.
Python (client_ex3.py)
from gradio_client import Client
client = Client("nasa/apod-generator")
# Submit job for streaming output
job = client.submit(prompt="Nebula galaxy", api_name="/generate")
# Iterate over streaming results as they complete
for response in job:
print("Stream update:", response)
š„ļø Live Interactive Web UI Simulator
### Streaming Job Submission
Click 'Submit & Stream Job' to test job iteration.
Example 4: Handling File Uploads in API Client Calls
Passes local file paths (images, PDFs, CSVs) directly into remote Gradio API endpoints.
Python (client_ex4.py)
from gradio_client import Client, handle_file
client = Client("akhaliq/vit-image-classifier")
# Use handle_file() to upload local assets securely
result = client.predict(
image=handle_file("sample_cat.jpg"),
api_name="/predict"
)
print("Classification:", result)
š„ļø Live Interactive Web UI Simulator
### File Upload API Client
Click 'Upload & Predict' to test handle_file() transmission.
Example 5: Connecting to Private Hugging Face Spaces with Token Authentication
Authenticates API client requests using Hugging Face access tokens for private spaces.
Python (client_ex5.py)
from gradio_client import Client
# Pass hf_token for private space authentication
client = Client("nikita-infotech/private-enterprise-ai", hf_token="hf_YourAccessTokenHere")
result = client.predict(message="Run confidential audit", api_name="/chat")
print(result)
š„ļø Live Interactive Web UI Simulator
### Authenticated Private Client
Click 'Connect & Authenticate' to test token validation.
Example 6: Multimodal API Client Calls (Text + Image)
Sends structured multimodal dictionaries containing both text prompts and image file references to vision-enabled APIs.
Python (client_ex6.py)
from gradio_client import Client, handle_file
client = Client("llava-hf/llava-1.5-7b-hf")
# Construct multimodal input payload
payload = {
"text": "Describe this architectural diagram.",
"files": [handle_file("diagram.png")]
}
result = client.predict(message=payload, api_name="/chat")
print("Vision Response:", result)
š„ļø Live Interactive Web UI Simulator
### Multimodal API Client
Click 'Send Multimodal Payload' to test vision client.
Example 7: Batch Processing Multiple Requests Concurrently
Iterates through a list of input prompts and dispatches concurrent API predictions using Python threading.
Python (client_ex7.py)
from gradio_client import Client
import concurrent.futures
client = Client("huggingface/CodeLlama-7b-Instruct")
prompts = ["Write a PHP loop", "Write a Python generator", "Write a SQL join"]
def worker(prompt):
return client.predict(message=prompt, api_name="/chat")
with concurrent.futures.ThreadPoolExecutor(max_workers=3) as executor:
results = list(executor.map(worker, prompts))
for i, res in enumerate(results):
print(f"Result {i+1}:", res[:100])
š„ļø Live Interactive Web UI Simulator
### Batch ThreadPool Client
Click 'Run Batch ThreadPool' to execute parallel requests.
Example 8: API Error Handling & Connection Timeout Management
Implements robust exception handling (`httpx.TimeoutException`, connection errors) when calling remote APIs.
Python (client_ex8.py)
from gradio_client import Client
from gradio_client.exceptions import Error
try:
client = Client("non-existent-space-url-xyz", src="https://huggingface.co/spaces/invalid/space")
result = client.predict(text="test", api_name="/predict")
except Exception as e:
print(f"ā ļø API Connection Failed: {str(e)}")
print("š Fallback to local backup server activated.")
š„ļø Live Interactive Web UI Simulator
### API Error & Timeout Guard
Click 'Test Error Handling' to simulate connection recovery.
Example 9: Automated CI/CD Pipeline Testing via Gradio Client
Writes automated unit test assertions verifying that deployed Gradio endpoints return expected output types and values.
Python (client_ex9.py)
from gradio_client import Client
def test_deployed_gradio_app():
client = Client("http://localhost:7860")
response = client.predict(name="Nikita", api_name="/predict")
assert "Nikita" in response, "Assertion Failed: Name missing in response"
assert isinstance(response, str), "Assertion Failed: Output must be string"
print("ā
All CI/CD API Integration Tests Passed!")
if __name__ == "__main__":
test_deployed_gradio_app()
š„ļø Live Interactive Web UI Simulator
### CI/CD Integration Test Runner
Click 'Run Unit Test Suite' to execute client assertions.
Example 10: Enterprise Multi-Client API Integration Hub (`gr.Blocks`)
Combines multiple remote client endpoints (Translation, Summarization, Sentiment) into an enterprise management dashboard.
Python (client_ex10.py)
import gradio as gr
from gradio_client import Client
def enterprise_hub_query(prompt):
# Simulated multi-client aggregation hub
return f"Hub Request Processed Successfully š\n- Endpoint 1 (GPT-4o): Answer generated.\n- Endpoint 2 (Classifier): Positive (0.99).\n- Query: '{prompt}'"
with gr.Blocks() as demo:
gr.Markdown("# š Enterprise Multi-Client API Gateway")
with gr.Row():
status = gr.Textbox(label="Client Gateway Status", value="Connected to 3 Remote Spaces š¢")
prompt_in = gr.Textbox(label="Global Prompt Dispatcher", value="Audit enterprise software architecture.")
btn = gr.Button("Dispatch to All Clients", variant="primary")
out_box = gr.Textbox(label="Aggregated Hub Response", lines=4)
btn.click(enterprise_hub_query, inputs=prompt_in, outputs=out_box)
if __name__ == "__main__":
demo.launch()
š„ļø Live Interactive Web UI Simulator
# š Enterprise Multi-Client API Gateway
Status: Connected to 3 Remote Spaces š¢
Click 'Dispatch to All Clients' to test enterprise gateway.