TashaSkyUp
40831eaa28
- Add detailed docstrings to AgentNode, BasicRecursionFilterNode, and DocumentChunkRecursionFilterNode for clarity and documentation. - Improve input validation and error handling in agent nodes. - Update .gitignore to cover more environment, config, and IDE files. - Refactor LiteLLMCustomEndpointProvider for custom API endpoints. - Add conditional imports and dummy functions for LightRAG when unavailable. - Update config.yaml to use environment variables for API keys and clarify usage. - Refine requirements.txt to avoid heavy LightRAG dependencies. - Add tests/__init__.py with test suite metadata; remove redundant test files. - Fix CustomDict import and model config handling in litellmnodes.py. - Improve chunk processing and recursion filter logic in agent nodes. - Update utils/__init__.py to include env_config.
85 lines
3.4 KiB
Python
Executable File
85 lines
3.4 KiB
Python
Executable File
import hashlib
|
|
import json
|
|
|
|
def get_input_hash(**kwargs):
|
|
"""
|
|
Creates a stable hash from input parameters, handling various Python types.
|
|
|
|
Handles:
|
|
- Callable objects (using their class names)
|
|
- Message dictionaries
|
|
- Image data (using shape and mean)
|
|
- Basic types (str, int, float, bool, None)
|
|
- Lists (converting elements to strings)
|
|
- Other types (using string representation)
|
|
|
|
Args:
|
|
**kwargs: Keyword arguments to hash
|
|
|
|
Returns:
|
|
str: MD5 hash of the input parameters
|
|
"""
|
|
# Create a copy of kwargs without the use_last_response flag
|
|
hash_kwargs = kwargs.copy()
|
|
hash_kwargs.pop('use_last_response', None)
|
|
|
|
# Create a list to store hashable components
|
|
hash_components = []
|
|
|
|
# Handle each input type appropriately
|
|
for key, value in sorted(hash_kwargs.items()):
|
|
if key in ('recursion_filter', 'memory_provider'):
|
|
if value is not None:
|
|
try:
|
|
# Get the class name for callable objects
|
|
if hasattr(value, '__self__'):
|
|
# If it's a bound method, get the class name of the instance
|
|
class_name = value.__self__.__class__.__name__
|
|
else:
|
|
# If it's a regular callable, get its class name
|
|
class_name = value.__class__.__name__
|
|
hash_components.append(f"{key}_class:{class_name}")
|
|
except:
|
|
# If we can't get the class name, use a default identifier
|
|
hash_components.append(f"{key}_class:unknown")
|
|
continue
|
|
elif key == 'messages':
|
|
# Convert messages to a stable string representation using json
|
|
try:
|
|
# Convert each message to a tuple of (role, content) and then to JSON
|
|
messages_data = [(m.get('role', ''), m.get('content', '')) for m in value]
|
|
messages_str = json.dumps(messages_data, sort_keys=True)
|
|
hash_components.append(f"messages:{messages_str}")
|
|
except:
|
|
# If messages can't be processed, skip them
|
|
continue
|
|
elif key == 'image':
|
|
# If there's image data, use its shape and mean value
|
|
try:
|
|
if value is not None:
|
|
hash_components.append(f"image_shape:{value.shape}")
|
|
hash_components.append(f"image_mean:{float(value.mean()):.4f}")
|
|
except:
|
|
# If image can't be processed, skip it
|
|
continue
|
|
elif isinstance(value, (str, int, float, bool, type(None))):
|
|
# Handle basic types directly
|
|
hash_components.append(f"{key}:{value}")
|
|
elif isinstance(value, list):
|
|
# Handle lists by converting to JSON for stable string representation
|
|
try:
|
|
list_str = json.dumps(value, sort_keys=True)
|
|
hash_components.append(f"{key}:{list_str}")
|
|
except:
|
|
continue
|
|
else:
|
|
# For any other type, use its string representation
|
|
try:
|
|
hash_components.append(f"{key}:{str(value)}")
|
|
except:
|
|
continue
|
|
|
|
# Join all components and create hash
|
|
hash_string = "|".join(hash_components)
|
|
return hashlib.md5(hash_string.encode()).hexdigest()
|