mirror of
https://github.com/awslabs/amazon-bedrock-agentcore-samples.git
synced 2025-09-08 20:50:46 +00:00
a48b6d9bd3
* Add a new Farm management use case to 02-use-cases * feat:modify readme * feat:modify solution diagram * feat:modify cleanup * remove datastore * remove all datastore * remove unnecessary files * modify file structure * remove ds_store
127 lines
4.3 KiB
Python
127 lines
4.3 KiB
Python
# Web search Lambda code - uses Tavily search + Nova Pro with LLM fallback
|
|
|
|
import json
|
|
import boto3
|
|
import urllib3
|
|
|
|
def lambda_handler(event, context):
|
|
try:
|
|
print(f"Received event: {json.dumps(event, default=str)}")
|
|
|
|
plant_name = event.get('plant_name', 'unknown plant')
|
|
health_status = event.get('health_status', '')
|
|
|
|
# Try Tavily search first
|
|
try:
|
|
# Get Tavily API key from environment
|
|
api_key = os.environ.get('TAVILY_API_KEY')
|
|
|
|
|
|
if health_status:
|
|
query = f"{plant_name} plant disease {health_status} treatment care"
|
|
else:
|
|
query = f"{plant_name} plant care fertilizer growing tips"
|
|
|
|
print(f"🔍 Searching Tavily for: {query}")
|
|
|
|
# Use urllib3 for Tavily API
|
|
http = urllib3.PoolManager()
|
|
|
|
search_data = {
|
|
'api_key': api_key,
|
|
'query': query,
|
|
'max_results': 3
|
|
}
|
|
|
|
response = http.request(
|
|
'POST',
|
|
'https://api.tavily.com/search',
|
|
body=json.dumps(search_data),
|
|
headers={'Content-Type': 'application/json'},
|
|
timeout=15
|
|
)
|
|
|
|
if response.status == 200:
|
|
search_result = json.loads(response.data.decode('utf-8'))
|
|
raw_results = search_result.get('results', [])
|
|
|
|
if raw_results:
|
|
search_text = "\\n\\n".join([f"**{r.get('title', 'No title')}**\\n{r.get('content', 'No content')}" for r in raw_results])
|
|
|
|
prompt = f"""Based on web search results about {plant_name} plant care:
|
|
|
|
{search_text}
|
|
|
|
Plant: {plant_name}
|
|
Health Issues: {health_status or 'No specific issues'}
|
|
|
|
Provide comprehensive advice including:
|
|
1) Disease treatment (if applicable)
|
|
2) Fertilizer recommendations
|
|
3) Care tips and best practices
|
|
4) Prevention strategies
|
|
5) Recovery timeline (if diseased)"""
|
|
|
|
print("✅ Using Tavily search results")
|
|
use_tavily = True
|
|
else:
|
|
raise Exception("No search results from Tavily")
|
|
else:
|
|
raise Exception(f"Tavily API returned status {response.status}")
|
|
|
|
except Exception as e:
|
|
print(f"⚠️ Tavily failed: {e}, falling back to Nova")
|
|
use_tavily = False
|
|
|
|
if health_status:
|
|
prompt = f"""Provide comprehensive plant care advice for {plant_name} with health issues: {health_status}
|
|
|
|
Include detailed information on:
|
|
1) Specific treatment for {health_status}
|
|
2) Recommended fertilizers and nutrients
|
|
3) Watering and care schedule
|
|
4) Prevention strategies
|
|
5) Expected recovery timeline
|
|
6) Signs of improvement to watch for"""
|
|
else:
|
|
prompt = f"""Provide comprehensive care guide for {plant_name}:
|
|
|
|
1) Optimal growing conditions (light, temperature, humidity)
|
|
2) Fertilizer schedule and nutrient requirements
|
|
3) Watering frequency and techniques
|
|
4) Common problems and solutions
|
|
5) Seasonal care variations
|
|
6) Harvesting tips (if applicable)
|
|
7) Companion planting suggestions"""
|
|
|
|
# Use Nova to process the prompt
|
|
bedrock = boto3.client("bedrock-runtime", region_name="us-east-1")
|
|
|
|
response = bedrock.converse(
|
|
modelId="us.amazon.nova-pro-v1:0",
|
|
messages=[{"role": "user", "content": [{"text": prompt}]}],
|
|
inferenceConfig={"temperature": 0.1, "maxTokens": 1500}
|
|
)
|
|
|
|
web_advice = response['output']['message']['content'][0]['text']
|
|
|
|
return {
|
|
'statusCode': 200,
|
|
'body': json.dumps({
|
|
'web_search_results': web_advice,
|
|
'plant_name': plant_name,
|
|
'health_status': health_status,
|
|
'search_source': 'tavily+nova' if use_tavily else 'nova_only'
|
|
})
|
|
}
|
|
|
|
except Exception as e:
|
|
print(f"Lambda error: {str(e)}")
|
|
return {
|
|
'statusCode': 500,
|
|
'body': json.dumps({
|
|
'error': str(e),
|
|
'web_search_results': f"Search failed: {str(e)}"
|
|
})
|
|
}
|