olca 0.2.62__tar.gz → 0.2.64__tar.gz

{olca-0.2.62 → olca-0.2.64}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: olca
-Version: 0.2.62
+Version: 0.2.64
 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
@@ -375,7 +375,6 @@ oLCa is a Python package that provides a CLI tool for Experimenting Langchain wi
 
 ## Features
 
-
 ## Installation
 
 To install the package, you can use pip:
@@ -384,11 +383,31 @@ To install the package, you can use pip:
 pip install olca
 ```
 
+## Quick Start
+
+1. Install the package:
+   ```bash
+   pip install olca
+   ```
+2. Initialize configuration:
+   ```bash
+   olca init
+   ```
+3. Run the CLI with tracing:
+   ```bash
+   olca -T
+   ```
+
+## Environment Variables
+
+Set LANGFUSE_PUBLIC_KEY, LANGFUSE_SECRET_KEY, and LANGFUSE_HOST for tracing with Langfuse.  
+Set LANGCHAIN_API_KEY for LangSmith tracing.  
+Optionally, set OPENAI_API_KEY for OpenAI usage.  
+
 ## Usage
 
 ### CLI Tool
 
-
 #### Help
 
 To see the available commands and options, use the `--help` flag:
@@ -397,8 +416,6 @@ To see the available commands and options, use the `--help` flag:
 olca2 --help
 ```
 
-
-
 ## fusewill
 
 The `fusewill` command is a CLI tool that provides functionalities for interacting with Langfuse, including tracing, dataset management, and prompt operations.
@@ -407,12 +424,11 @@ The `fusewill` command is a CLI tool that provides functionalities for interacti
 
 To see the available commands and options for `fusewill`, use the `--help` flag:
 
-
 ----
+
 IMPORTED README from olca1
 ----
 
-
 ### Olca
 
 The olca.py script is designed to function as a command-line interface (CLI) agent. It performs various tasks based on given inputs and files present in the directory. The agent is capable of creating directories, producing reports, and writing instructions for self-learning. It operates within a GitHub repository environment and can commit and push changes if provided with an issue ID. The script ensures that it logs its internal actions and follows specific guidelines for handling tasks and reporting, without modifying certain configuration files or checking out branches unless explicitly instructed.

{olca-0.2.62 → olca-0.2.64}/README.md

@@ -4,7 +4,6 @@ oLCa is a Python package that provides a CLI tool for Experimenting Langchain wi
 
 ## Features
 
-
 ## Installation
 
 To install the package, you can use pip:
@@ -13,11 +12,31 @@ To install the package, you can use pip:
 pip install olca
 ```
 
+## Quick Start
+
+1. Install the package:
+   ```bash
+   pip install olca
+   ```
+2. Initialize configuration:
+   ```bash
+   olca init
+   ```
+3. Run the CLI with tracing:
+   ```bash
+   olca -T
+   ```
+
+## Environment Variables
+
+Set LANGFUSE_PUBLIC_KEY, LANGFUSE_SECRET_KEY, and LANGFUSE_HOST for tracing with Langfuse.  
+Set LANGCHAIN_API_KEY for LangSmith tracing.  
+Optionally, set OPENAI_API_KEY for OpenAI usage.  
+
 ## Usage
 
 ### CLI Tool
 
-
 #### Help
 
 To see the available commands and options, use the `--help` flag:
@@ -26,8 +45,6 @@ To see the available commands and options, use the `--help` flag:
 olca2 --help
 ```
 
-
-
 ## fusewill
 
 The `fusewill` command is a CLI tool that provides functionalities for interacting with Langfuse, including tracing, dataset management, and prompt operations.
@@ -36,12 +53,11 @@ The `fusewill` command is a CLI tool that provides functionalities for interacti
 
 To see the available commands and options for `fusewill`, use the `--help` flag:
 
-
 ----
+
 IMPORTED README from olca1
 ----
 
-
 ### Olca
 
 The olca.py script is designed to function as a command-line interface (CLI) agent. It performs various tasks based on given inputs and files present in the directory. The agent is capable of creating directories, producing reports, and writing instructions for self-learning. It operates within a GitHub repository environment and can commit and push changes if provided with an issue ID. The script ensures that it logs its internal actions and follows specific guidelines for handling tasks and reporting, without modifying certain configuration files or checking out branches unless explicitly instructed.

{olca-0.2.62 → olca-0.2.64}/olca/fusewill_utils.py

@@ -123,7 +123,7 @@ def list_traces_by_score(score_name, min_value=None, max_value=None, limit=100):
     return filtered_traces
 
 def add_score_to_a_trace(trace_id, generation_id, name, value, data_type="NUMERIC", comment=""):
-    langfuse.score(
+    result_add_score_to_a_trace=langfuse.score(
         trace_id=trace_id,
         observation_id=generation_id,
         name=name,
@@ -131,11 +131,16 @@ def add_score_to_a_trace(trace_id, generation_id, name, value, data_type="NUMERI
         data_type=data_type,
         comment=comment
     )
+    return result_add_score_to_a_trace
 
 def create_score(name, data_type, description="", possible_values=None, min_value=None, max_value=None):
-    langfuse.score(
+    placeholder_value = ""
+    if data_type.upper() == "BOOLEAN":
+        placeholder_value = "1"
+
+    resulting_score = langfuse.score(
         name=name,
-        value="",  # Provide a placeholder value
+        value=placeholder_value,
         data_type=data_type,
         description=description,
         # For categorical:
@@ -143,11 +148,17 @@ def create_score(name, data_type, description="", possible_values=None, min_valu
         # For numeric:
         **({"min_value": min_value, "max_value": max_value} if data_type == "NUMERIC" and min_value is not None and max_value is not None else {})
     )
+    return resulting_score
 
 def score_exists(name):
-    scores = langfuse.get_scores()
-    for score in scores.data:
-        if score.name == name:
+    """
+    Check if a score with the given name exists by calling list_scores().
+    """
+    scores = list_scores()
+    if not scores or scores.get('meta', {}).get('totalItems', 0) == 0:
+        return False
+    for sc in scores:
+        if sc.get("name") == name:
             return True
     return False
 
@@ -265,7 +276,7 @@ def fetch_all_traces(start_date=None, end_date=None):
 
 def export_traces(format='json', output_path=None, start_date=None, end_date=None):
     """
-    Export traces to a given format (json or csv).
+    Export traces along with their full score details.
     """
     try:
         all_traces = fetch_all_traces(start_date=start_date, end_date=end_date)
@@ -277,9 +288,23 @@ def export_traces(format='json', output_path=None, start_date=None, end_date=Non
         if output_dir and not os.path.exists(output_dir):
             os.makedirs(output_dir)
 
+        all_scores=list_scores()
+        exported_data = []
+        for t in all_traces:
+            # fetch full score details
+            score_details = []
+            if t.scores:
+                for s_id in t.scores:
+                    s_detail = get_score_by_id(s_id)
+                    if s_detail:
+                        score_details.append(s_detail)
+            t_dict = t.__dict__
+            t_dict["score_details"] = score_details
+            exported_data.append(t_dict)
+
         if format == 'json':
             with open(output_path, 'w') as f:
-                json.dump([t.__dict__ for t in all_traces], f, indent=2, default=str)
+                json.dump(exported_data, f, indent=2, default=str)
         elif format == 'csv':
             import csv
             fieldnames = ['id', 'name', 'input', 'output', 'createdAt']
@@ -299,7 +324,7 @@ def export_traces(format='json', output_path=None, start_date=None, end_date=Non
             # Sort traces by createdAt to ensure the oldest date is first
             all_traces.sort(key=lambda x: x.createdAt)
             first_trace_date = datetime.datetime.fromisoformat(all_traces[0].createdAt.replace('Z', '+00:00')).strftime('%Y-%m-%d %H:%M:%S')
-            last_trace_date = datetime.datetime.fromisoformat(all_traces[-1].CreatedAt.replace('Z', '+00:00')).strftime('%Y-%m-%d %H:%M:%S')
+            last_trace_date = datetime.datetime.fromisoformat(all_traces[-1].createdAt.replace('Z', '+00:00')).strftime('%Y-%m-%d %H:%M:%S')
             print(f"Traces exported to {output_path}. Total traces exported: {len(all_traces)}")
             print(f"Date range: {first_trace_date} to {last_trace_date}")
         else:
@@ -307,9 +332,29 @@ def export_traces(format='json', output_path=None, start_date=None, end_date=Non
     except Exception as e:
         print(f"Error exporting traces: {e}")
 
+def create_new_trace(name, input_text, output_text, session_id=None, metadata=None, timestamp=None):
+    """
+    Creates a new trace with an optional timestamp.
+    """
+    parsed_timestamp = None
+    if timestamp:
+        try:
+            parsed_timestamp = datetime.datetime.fromisoformat(timestamp.replace('Z', '+00:00'))
+        except ValueError:
+            pass
+    trace_created=langfuse.trace(
+        name=name,
+        input=input_text,
+        output=output_text,
+        session_id=session_id,
+        metadata=metadata,
+        timestamp=parsed_timestamp
+    )
+    return trace_created
+
 def import_traces(format='json', input_path=None):
     """
-    Import traces from a given file (json or csv) into Langfuse.
+    Import traces. If any score doesn't exist, create it and attach it to the trace.
     """
     if not input_path:
         print("No input file provided for importing traces.")
@@ -327,14 +372,51 @@ def import_traces(format='json', input_path=None):
                 for row in reader:
                     data.append(row)
 
+        if isinstance(data, dict):
+            data = [data]
+
         # Create new traces in Langfuse from data
         for item in data:
-            langfuse.create_trace(
+            trace_timestamp = item.get('timestamp') or item.get('createdAt')
+            new_trace = create_new_trace(
                 name=item.get('name', 'Imported Trace'),
-                input=item.get('input', ''),
-                output=item.get('output', '')
-                # pass other fields as needed
+                input_text=item.get('input', ''),
+                output_text=item.get('output', ''),
+                session_id=item.get('session_id'),
+                metadata=item.get('metadata'),
+                timestamp=trace_timestamp
             )
+            # handle imported scores
+            for s_detail in item.get("score_details", []):
+                score_name = s_detail["name"]
+                score_value = str(s_detail.get("value", "0"))
+                score_data_type = s_detail.get("dataType", "NUMERIC")
+                score_comment = s_detail.get("comment", "")
+                score_description = s_detail.get("description", "")
+                score_possible_values = s_detail.get("possible_values")
+                minimum_score_value = s_detail.get("min_value")
+                max_score_value = s_detail.get("max_value")
+                if not score_exists(score_name):
+                    resulting_score=create_score(
+                        name=score_name,
+                        data_type=score_data_type,
+                        description=score_description,
+                        possible_values=score_possible_values,
+                        min_value=minimum_score_value,
+                        max_value=max_score_value
+                        
+                    )
+                result_add_score_to_a_trace=add_score_to_a_trace(
+                    trace_id=new_trace.id,
+                    generation_id=None,  # Replace as needed if your data includes observation IDs
+                    name=score_name,
+                    value=score_value,
+                    data_type=score_data_type,
+                    comment=score_comment
+                )
+                print(f"Added score {score_name} to trace {new_trace.id}")
+                print(result_add_score_to_a_trace)
+                
         print(f"Imported {len(data)} traces from {input_path}")
     except Exception as e:
         print(f"Error importing traces: {e}")

olca-0.2.64/olca/oiv.py

@@ -0,0 +1,145 @@
+#!/bin/env python
+#@STCGoal Create an agent callable with command line
+#@STCIssue The original script was not callable from the command line.
+
+from langchain import hub
+from langchain.agents import AgentExecutor, create_react_agent
+from langchain_community.agent_toolkits.load_tools import load_tools
+from langchain_openai import ChatOpenAI
+import argparse
+import os
+import json
+import tlid
+import warnings
+
+warnings.filterwarnings("ignore", message="The function `loads` is in beta. It is actively being worked on, so the API may change.")
+DEBUG_MODE=False
+
+prompt=None
+def _create_agent_tools(tool_name = "arxiv",temperature = 0.0,tool_hub_tag = "jgwill/react",chatbot_model = "gpt-4o-mini"):
+    global prompt
+    
+    llm = ChatOpenAI(temperature=temperature,name=chatbot_model)
+    
+    tools = load_tools(
+        [tool_name],
+        )
+    if DEBUG_MODE:
+          
+      print("Tools:")
+      print(tools)
+      print("--------------")
+    prompt = hub.pull(tool_hub_tag)
+    
+    if DEBUG_MODE:
+      print("prompt:")
+      print(prompt)
+      print("--------------")
+
+    agent = create_react_agent(llm, tools, prompt)
+    
+    if DEBUG_MODE:
+      print("Agent:")
+      print(agent)
+      print("--------------")
+    return tools,agent
+
+
+def create_agent_executor(tools, agent) -> AgentExecutor:
+    return AgentExecutor(agent=agent, tools=tools, verbose=True,handle_parsing_errors=True)
+   
+
+
+
+# tools, agent=_create_agent_tools()
+
+# agent_executor=create_agent_executor(tools, agent)
+def ask_agent(input_request, agent_executor=None,tool_hub_tag = "jgwill/react",chatbot_model = "gpt-4o-mini"):
+  if agent_executor is None:
+    tools, agent = _create_agent_tools(tool_hub_tag=tool_hub_tag)
+    agent_executor = create_agent_executor(tools, agent)
+  
+  resp = agent_executor.invoke(
+    {
+      "input": input_request,
+    }
+  )
+  
+  return resp
+
+def serialize_response_to_json(resp):
+    return json.dumps(resp)
+
+def serialize_response_to_json_file(resp, filename):
+  json_str=serialize_response_to_json(resp)
+  with open(filename, 'w') as f:
+    f.write(json_str)
+    
+def serialize_response_to_markdown(o):
+    output=o["output"]["output"]
+    string_md="# Output\n"
+    #string_md+=f"## Model\n{o['model']}\n"
+    #string_md+=f"## Prompt\n{o['prompt']['prompt']}\n"
+    string_md+=f"## Input\n{o['input']}\n"
+    string_md+=f"## Output\n{output}\n"
+    return string_md
+
+def serialize_response_to_markdown_file(o, filename):
+    string_md=serialize_response_to_markdown(o)
+    with open(filename, 'w') as f:
+        f.write(string_md)
+
+
+def main():
+    global prompt
+    args = parse_cli_arguments()
+
+    input_request = args.input
+    tool_hub_tag = "jgwill/react" if args.hub_tag is None else args.hub_tag
+    resp = ask_agent(input_request,tool_hub_tag=tool_hub_tag,chatbot_model=args.chatbot_model)
+    outdir=os.path.join(os.getcwd(),"output")
+    os.makedirs(outdir, exist_ok=True)
+    out_filename = f"{args.prefix}output-{tlid.get_minutes()}.json"
+    outfile=os.path.join(outdir,out_filename)
+    o={}
+    prompt_dict = {
+        "prompt": str(prompt)
+    }
+    o["model"]=args.chatbot_model
+    o["prompt"]=prompt_dict
+    o["input"]=input_request
+    o["output"]=resp
+    serialize_response_to_json_file(o, outfile)
+    serialize_response_to_markdown_file(o, outfile.replace(".json",".md"))
+    VERBOSE_RESULT=False
+    if VERBOSE_RESULT:
+        print("==================================")
+        print(input_request)
+        print("============RESPONSE============")
+        print(resp)
+        print("==================================")
+        print("=============INPUT Request=====================")
+        print(input_request)
+        print("==================================")
+        print("============OUTPUT============")
+    output=resp["output"]
+    print(output)
+
+def parse_cli_arguments():
+    parser = argparse.ArgumentParser(description='Process Input request for pattern search.')
+    parser.add_argument('-I','--input', type=str,
+                        help='an input request for the searched article')
+    ##--hub_tag
+    parser.add_argument('-H','-ht','--hub_tag', type=str,
+                        help='The hub tag for the process',default="jgwill/react")
+    #--chatbot_model
+    parser.add_argument('-M','-m','--chatbot_model', type=str,
+                        help='a chatbot model for the processing',default="gpt-4o-mini")
+    #--prefix
+    parser.add_argument('-P','-p','--prefix', type=str,
+                        help='a file prefix for output',default="arxiv-")
+    args = parser.parse_args()
+    return args
+
+if __name__ == "__main__":
+    main()

{olca-0.2.62 → olca-0.2.64}/olca/olcacli.py

@@ -8,7 +8,7 @@ import argparse
 import yaml
 from olca.utils import load_environment, initialize_langfuse
 from olca.tracing import TracingManager
-from olca.olcahelper import setup_required_directories, initialize_config_file
+from olca.olcahelper import setup_required_directories, initialize_config_file, prepare_input
 from prompts import SYSTEM_PROMPT_APPEND, HUMAN_APPEND_PROMPT
 
 #jgwill/olca1
@@ -128,18 +128,6 @@ def print_stream(stream):
         except Exception as e:
             print(s)
 
-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,system_instructions,user_input
-
 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]"

{olca-0.2.62 → olca-0.2.64}/olca/olcahelper.py

@@ -65,3 +65,13 @@ def initialize_config_file():
     except KeyboardInterrupt:
         print("\nConfiguration canceled by user.")
         exit(0)
+
+def prepare_input(user_input, system_instructions, append_prompt=True, human=False):
+    from olca.prompts import SYSTEM_PROMPT_APPEND, HUMAN_APPEND_PROMPT
+    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, system_instructions, user_input

{olca-0.2.62 → olca-0.2.64}/olca.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: olca
-Version: 0.2.62
+Version: 0.2.64
 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
@@ -375,7 +375,6 @@ oLCa is a Python package that provides a CLI tool for Experimenting Langchain wi
 
 ## Features
 
-
 ## Installation
 
 To install the package, you can use pip:
@@ -384,11 +383,31 @@ To install the package, you can use pip:
 pip install olca
 ```
 
+## Quick Start
+
+1. Install the package:
+   ```bash
+   pip install olca
+   ```
+2. Initialize configuration:
+   ```bash
+   olca init
+   ```
+3. Run the CLI with tracing:
+   ```bash
+   olca -T
+   ```
+
+## Environment Variables
+
+Set LANGFUSE_PUBLIC_KEY, LANGFUSE_SECRET_KEY, and LANGFUSE_HOST for tracing with Langfuse.  
+Set LANGCHAIN_API_KEY for LangSmith tracing.  
+Optionally, set OPENAI_API_KEY for OpenAI usage.  
+
 ## Usage
 
 ### CLI Tool
 
-
 #### Help
 
 To see the available commands and options, use the `--help` flag:
@@ -397,8 +416,6 @@ To see the available commands and options, use the `--help` flag:
 olca2 --help
 ```
 
-
-
 ## fusewill
 
 The `fusewill` command is a CLI tool that provides functionalities for interacting with Langfuse, including tracing, dataset management, and prompt operations.
@@ -407,12 +424,11 @@ The `fusewill` command is a CLI tool that provides functionalities for interacti
 
 To see the available commands and options for `fusewill`, use the `--help` flag:
 
-
 ----
+
 IMPORTED README from olca1
 ----
 
-
 ### Olca
 
 The olca.py script is designed to function as a command-line interface (CLI) agent. It performs various tasks based on given inputs and files present in the directory. The agent is capable of creating directories, producing reports, and writing instructions for self-learning. It operates within a GitHub repository environment and can commit and push changes if provided with an issue ID. The script ensures that it logs its internal actions and follows specific guidelines for handling tasks and reporting, without modifying certain configuration files or checking out branches unless explicitly instructed.

{olca-0.2.62 → olca-0.2.64}/olca.egg-info/SOURCES.txt

@@ -5,6 +5,7 @@ setup.py
 olca/__init__.py
 olca/fusewill_cli.py
 olca/fusewill_utils.py
+olca/oiv.py
 olca/olcacli.py
 olca/olcahelper.py
 olca/prompts.py

{olca-0.2.62 → olca-0.2.64}/olca.egg-info/entry_points.txt

@@ -1,4 +1,5 @@
 [console_scripts]
 fusewill = olca.fusewill_cli:main
+oiv = olca.oiv:main
 olca = olca.olcacli:main
 olca2 = olca.olcacli:main

{olca-0.2.62 → olca-0.2.64}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "olca"
-version = "0.2.62"
+version = "0.2.64"
 
 description = "A Python package for experimental usage of Langchain and Human-in-the-Loop"
 readme = "README.md"
@@ -45,3 +45,5 @@ classifiers = [
 olca2 = "olca.olcacli:main"
 olca = "olca.olcacli:main"
 fusewill = "olca.fusewill_cli:main"
+oiv = "olca.oiv:main"
+

{olca-0.2.62 → olca-0.2.64}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name='olca',
-    version = "0.2.62",
+    version = "0.2.64",
     author='Jean GUillaume ISabelle',
     author_email='jgi@jgwill.com',
     description='A Python package for experimenting with Langchain agent and interactivity in Terminal modalities.',
@@ -31,7 +31,8 @@ setup(
         'langchain-ollama',
         'langgraph',
         'llm',
-        'langgraph'
+        'langgraph',
+        'arxiv',
     ],
     entry_points={
         'console_scripts': [