olca 0.2.0__tar.gz → 0.2.1__tar.gz

{olca-0.2.0 → olca-0.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: olca
-Version: 0.2.0
+Version: 0.2.1
 Summary: A Python package for experimental usage of Langchain and Human-in-the-Loop
 Home-page: https://github.com/jgwill/olca
 Author: Jean GUillaume ISabelle

olca-0.2.1/olca/__init__.py

@@ -0,0 +1 @@
+#left blank onnpurpose

olca-0.2.1/olca/olcacli.py

@@ -0,0 +1,296 @@
+#%%
+import os
+from click import prompt
+import dotenv
+from langchain import hub
+import argparse
+import yaml
+
+import langsmith 
+# Initialize LangSmith client
+LANGSMITH_API_KEY = os.getenv("LANGSMITH_API_KEY")
+if not LANGSMITH_API_KEY:
+    print("Error: LANGSMITH_API_KEY environment variable is not set.")
+    exit(1)
+client = langsmith.Client(api_key=LANGSMITH_API_KEY)
+
+
+#jgwill/olca1
+#olca1_prompt = hub.pull("jgwill/olca1") #Future use
+#https://smith.langchain.com/hub/jgwill/olca1
+SYSTEM_PROMPT_APPEND="""
+You do what is above and consider the following when doing the task:
+---
+* Create any non-existent directories. (./reports, ./log, ./.olca)
+* create a subfolder .olca in the current directory if dont exist and :
+* Write or append instructions to ./.olca/instructions.txt on how to perform tasks (in terms of reflection and action steps. example: cli syntax to execute something).  You read that file to learn how to do your job if it exists.  You update it with what you learned and keep it up to date. You dont erase the content of the file.  If what you find in it is not related to your task, you ignore it. Dont output all these details you do internally to the standard output.  You read it back from time to time to remember what you did and how to make it and what the user wants.  
+* Produce a report in ./reports/report-olca-[TIMESTAMP].md
+* Stop and report to the user if repeated errors are encountered.
+* Output the iteration number and plan to the standard output during each loop.
+* If you are given an issue_id, you commit the work to that issue and push your results (dont output the git push to the standard output).
+* Do not comment on your final results; just output them but keep logging your action steps you do internally (all reflection and action steps).
+* Dont modify the file: olca.yml 
+* If github issue related, dont checkout a branch except if asked otherwise. 
+* You dont checkout branches, I repeat, you dont checkout branches.
+* Generate your TIMESTAMP with the following command: date +'%Y%m%d%H%M%S' only once at the beginning of your script.
+* Make sure if you Switched to branch, you switch back to main before the end of your script.
+* Try to observe that you keep doing the same thing over and over again and stop right away if you see that (dont do that if you are developping a story)
+* Be quiet with trivial output in the terminal.
+* Write your plan in ./.olca/plan.md
+* You watch out for basic syntax errors with your args when executing echo commands. (example: Syntax error: Unterminated quoted string)
+----
+REMEMBER: Dont introduce nor conclude, just output results. No comments. you  present in a coherent format without preambles or fluff. Never use the word "determination".
+"""
+
+HUMAN_APPEND_PROMPT = """
+* Utilize the 'human' tool for interactions as directed.
+* Communicate clearly and simply, avoiding exaggeration.
+Example Interaction:
+<example>
+'==============================================
+{ PURPOSE_OF_THE_MESSAGE_SHORT }
+==============================================
+{ CURRENT_STATUS_OR_MESSAGE_CONTENT }
+==============================================
+{ PROMPT_FOR_USER_INPUT_SHORT } :
+</example>
+"""
+def get_input() -> str:
+    print("----------------------")
+    contents = []
+    while True:
+        try:
+            line = input()
+        except EOFError:
+            break
+        if line == "q":
+            break
+        contents.append(line)
+    return "\n".join(contents)
+
+
+#try loaging .env and see if a key is there for OLCA_SYSTEM_PROMPT_APPEND
+#If it is there, use it instead of the default
+try:
+    OLCA_SYSTEM_PROMPT_APPEND = os.getenv("OLCA_SYSTEM_PROMPT_APPEND")
+    if OLCA_SYSTEM_PROMPT_APPEND is not None:
+        SYSTEM_PROMPT_APPEND = OLCA_SYSTEM_PROMPT_APPEND
+except:
+    pass
+
+def load_config(config_file):
+    with open(config_file, 'r') as file:
+        config = yaml.safe_load(file)
+    return config
+
+
+#%%
+
+dotenv.load_dotenv()
+
+#%%
+#from dotenv in ./.env , load key
+
+#%%
+
+# First we initialize the model we want to use.
+from json import load
+from langchain_openai import ChatOpenAI,OpenAI
+from langchain.agents import AgentExecutor, create_react_agent
+
+from langchain_community.agent_toolkits.load_tools import load_tools
+
+import warnings
+#
+
+# Suppress the specific UserWarning
+warnings.filterwarnings("ignore", category=UserWarning, message="The shell tool has no safeguards by default. Use at your own risk.")
+
+from langchain_community.tools.shell import ShellTool
+
+from typing import Literal
+
+from langchain_core.tools import tool
+
+from langgraph.prebuilt import create_react_agent
+from langgraph.errors import GraphRecursionError
+
+@tool
+def get_weather(city: Literal["nyc", "sf"]):
+    """Use this to get weather information."""
+    if city == "nyc":
+        return "It might be cloudy in nyc"
+    elif city == "sf":
+        return "It's always sunny in sf"
+    else:
+        raise AssertionError("Unknown city")
+
+
+def print_stream(stream):
+    for s in stream:
+        message = s["messages"][-1]
+        if isinstance(message, tuple):
+            print(message)
+        else:
+            message.pretty_print()
+
+def prepare_input(user_input, system_instructions,append_prompt=True, human=False):
+    appended_prompt = system_instructions + SYSTEM_PROMPT_APPEND if append_prompt else system_instructions
+    appended_prompt = appended_prompt + HUMAN_APPEND_PROMPT if human else appended_prompt
+    
+    inputs = {"messages": [
+    ("system",
+     appended_prompt),
+    ("user", user_input     )
+    ]}
+        
+    return inputs
+
+OLCA_DESCRIPTION = "OlCA (Orpheus Langchain CLI Assistant) (very Experimental and dangerous)"
+OLCA_EPILOG = "For more information: https://github.com/jgwill/orpheuspypractice/wiki/olca"
+OLCA_USAGE="olca [-D] [-H] [-M] [-T] [init] [-y]"
+def _parse_args():
+    parser = argparse.ArgumentParser(description=OLCA_DESCRIPTION, epilog=OLCA_EPILOG,usage=OLCA_USAGE)
+    parser.add_argument("-D", "--disable-system-append", action="store_true", help="Disable prompt appended to system instructions")
+    parser.add_argument("-H", "--human", action="store_true", help="Human in the loop mode")
+    parser.add_argument("-M", "--math", action="store_true", help="Enable math tool")
+    parser.add_argument("-T", "--tracing", action="store_true", help="Enable tracing")
+    parser.add_argument("init", nargs='?', help="Initialize olca interactive mode")
+    parser.add_argument("-y", "--yes", action="store_true", help="Accept the new file olca.yml")
+    return parser.parse_args()
+
+def main():
+    args = _parse_args()
+    olca_config_file = 'olca_config.yaml'
+    olca_new_config_file = 'olca.yml'
+    
+    if args.init:
+        if os.path.exists(olca_new_config_file) or os.path.exists(olca_config_file):
+            print("Error: Configuration file already exists. Cannot run 'olca init'.")
+            return
+        if args.yes:
+            olca_config_file = olca_new_config_file
+        else:
+            generate_config_example()
+            return
+    
+    if os.path.exists(olca_new_config_file):
+        olca_config_file = olca_new_config_file
+    elif os.path.exists(olca_config_file):
+        print("Warning: 'olca_config.yaml' is deprecated. Please use 'olca.yml' instead.")
+    else:
+        generate_config_example()
+        return
+
+    config = load_config(olca_config_file)
+    
+    # Check for tracing flag in config and CLI
+    tracing_enabled = config.get('tracing', False) or args.tracing
+    if tracing_enabled:
+        os.environ["LANGCHAIN_TRACING_V2"] = "true"
+        if not os.getenv("LANGCHAIN_API_KEY"):
+            print("Error: LANGCHAIN_API_KEY environment variable is required for tracing. Please set it up at : https://smith.langchain.com/settings")
+            exit(1)
+    
+    try:
+            
+        api_key_variable = "OPENAI_API_KEY"
+        api_keyname=config.get('api_keyname',"OPENAI_API_KEY_olca")
+
+        api_key_lcpractices2409 = os.getenv(api_keyname)
+        #print(api_key_lcpractices2409)
+        os.environ[api_key_variable] = api_key_lcpractices2409
+    except :
+        #load .env file in current dir or HOME and find OPENAI_API_KEY
+        try:
+            dotenv.load_dotenv()
+        except:
+            #load in HOME
+            try:
+                dotenv.load_dotenv(dotenv.find_dotenv(usecwd=False))
+            except:
+                print("Error: Could not load .env file")
+                exit(1)
+
+
+       
+
+        
+        
+    
+    system_instructions = config.get('system_instructions', '')
+    user_input = config.get('user_input', '')
+    model_name=config.get('model_name', "gpt-4o-mini")
+    recursion_limit=config.get('recursion_limit', 15)
+    disable_system_append = _parse_args().disable_system_append
+    
+    # Use the system_instructions and user_input in your CLI logic
+    print("System Instructions:", system_instructions)
+    print("User Input:", user_input)
+    print("Model Name:", model_name)
+    print("Recursion Limit:", recursion_limit)
+    print("Trace:", tracing_enabled)
+    
+    
+    
+    
+    model = ChatOpenAI(model=model_name, temperature=0)
+    selected_tools = [ "terminal"]
+    
+    human_switch = args.human
+    #look in olca_config.yaml for human: true
+    if "human" in config:
+        human_switch = config["human"]
+        
+    if human_switch:
+        selected_tools.append("human")
+    
+    if args.math:
+        math_llm=OpenAI()
+        selected_tools.append("llm-math")
+        if human_switch:
+            tools = load_tools(    selected_tools, llm=math_llm,   allow_dangerous_tools=True, input_func=get_input)
+        else:
+            tools = load_tools(    selected_tools, llm=math_llm,   allow_dangerous_tools=True)
+    else:
+        if human_switch:
+            tools = load_tools(    selected_tools,    allow_dangerous_tools=True, input_func=get_input)
+        else:
+            tools = load_tools(    selected_tools,    allow_dangerous_tools=True)
+    
+    
+    # Define the graph
+    graph = create_react_agent(model, tools=tools)
+    
+    if graph.config is None:
+      graph.config = {}
+    graph.config["recursion_limit"] = recursion_limit
+    
+    inputs = prepare_input(user_input, system_instructions, not disable_system_append, human_switch)
+    
+
+    try:
+        os.makedirs('.olca', exist_ok=True)
+        print_stream(graph.stream(inputs, stream_mode="values"))
+    except GraphRecursionError as e:
+        #print(f"Error: {e}")
+        print("Recursion limit reached. Please increase the 'recursion_limit' in the olca_config.yaml file.")
+        print("For troubleshooting, visit: https://python.langchain.com/docs/troubleshooting/errors/GRAPH_RECURSION_LIMIT")
+
+def generate_config_example():
+    config = {
+        "api_keyname": input("api_keyname [OPENAI_API_KEY_olca]: ") or "OPENAI_API_KEY_olca",
+        "model_name": input("model_name [gpt-4o-mini]: ") or "gpt-4o-mini",
+        "recursion_limit": int(input("recursion_limit [12]: ") or 12),
+        "temperature": float(input("temperature [0]: ") or 0),
+        "human": input("human [true]: ").lower() in ["true", "yes", "y", "1", ""] or True,
+        "tracing": input("tracing [true]: ").lower() in ["true", "yes", "y", "1", ""] or True,
+        "system_instructions": input("system_instructions [Hello, I am a chatbot. How can I help you today?]: ") or "Hello, I am a chatbot. How can I help you today?",
+        "user_input": input("user_input [What is the weather in NYC?]: ") or "What is the weather in NYC?"
+    }
+    with open('olca.yml', 'w') as file:
+        yaml.dump(config, file)
+    print("Configuration file 'olca.yml' created successfully.")
+
+if __name__ == "__main__":
+    main()

{olca-0.2.0 → olca-0.2.1}/olca.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: olca
-Version: 0.2.0
+Version: 0.2.1
 Summary: A Python package for experimental usage of Langchain and Human-in-the-Loop
 Home-page: https://github.com/jgwill/olca
 Author: Jean GUillaume ISabelle

{olca-0.2.0 → olca-0.2.1}/olca.egg-info/SOURCES.txt

@@ -2,6 +2,8 @@ LICENSE
 README.md
 pyproject.toml
 setup.py
+olca/__init__.py
+olca/olcacli.py
 olca.egg-info/PKG-INFO
 olca.egg-info/SOURCES.txt
 olca.egg-info/dependency_links.txt

olca-0.2.1/olca.egg-info/top_level.txt

@@ -0,0 +1 @@
+olca

{olca-0.2.0 → olca-0.2.1}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "olca"
-version = "0.2.0"
+version = "0.2.1"
 description = "A Python package for experimental usage of Langchain and Human-in-the-Loop"
 readme = "README.md"
 requires-python = ">=3.6"

{olca-0.2.0 → olca-0.2.1}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name='olca',
-    version = "0.2.0",
+    version = "0.2.1",
     author='Jean GUillaume ISabelle',
     author_email='jgi@jgwill.com',
     description='A Python package for experimenting with Langchain agent and interactivity in Terminal modalities.',
@@ -10,7 +10,7 @@ setup(
     long_description_content_type='text/markdown',
     url='https://github.com/jgwill/olca',
     packages=find_packages(
-        include=["olca2", "test-*.py"], exclude=["test*log", "*test*csv", "*test*png"]
+        include=["olca", "test-*.py"], exclude=["test*log", "*test*csv", "*test*png"]
     ),
     #package_dir={'': 'coaiapy'},
     install_requires=[

olca-0.2.0/olca.egg-info/top_level.txt

@@ -1 +0,0 @@
-