Merge pull request #389 from rmusser01/dev

Potential fix for #388

App_Function_Libraries/Gradio_UI/Audio_ingestion_tab.py

@@ -106,7 +106,7 @@ def update_prompts(preset_name):
                     inputs=preset_prompt,
                     outputs=[custom_prompt_input, system_prompt_input]
                 )
-
+                global_api_endpoints
                 api_name_input = gr.Dropdown(
                     choices=[None, "Local-LLM", "OpenAI", "Anthropic", "Cohere", "Groq", "DeepSeek", "Mistral", "OpenRouter",
                              "Llama.cpp", "Kobold", "Ooba", "Tabbyapi", "VLLM","ollama", "HuggingFace", "Custom-OpenAI-API"],

App_Function_Libraries/Gradio_UI/Llamafile_tab.py

@@ -19,6 +19,9 @@
 #
 # Functions:
 
+BASE_DIR = os.path.dirname(os.path.abspath(__file__))
+MODELS_DIR = os.path.join(BASE_DIR, "Models")
+
 def create_chat_with_llamafile_tab():
     # Function to update model path based on selection
     def on_local_model_change(selected_model: str, search_directory: str) -> str:
@@ -35,8 +38,13 @@ def update_dropdowns(search_directory: str) -> Tuple[dict, str]:
             logging.debug(f"Directory does not exist: {search_directory}")  # Debug print for non-existing directory
             return gr.update(choices=[], value=None), "Directory does not exist."
 
-        logging.debug(f"Directory exists: {search_directory}, scanning for files...")  # Confirm directory exists
-        model_files = get_gguf_llamafile_files(search_directory)
+        try:
+            logging.debug(f"Directory exists: {search_directory}, scanning for files...")  # Confirm directory exists
+            model_files = get_gguf_llamafile_files(search_directory)
+            logging.debug("Completed scanning for model files.")
+        except Exception as e:
+            logging.error(f"Error scanning directory: {e}")
+            return gr.update(choices=[], value=None), f"Error scanning directory: {e}"
 
         if not model_files:
             logging.debug(f"No model files found in {search_directory}")  # Debug print for no files found
@@ -117,15 +125,22 @@ def download_preset_model(selected_model: str) -> Tuple[str, str]:
 
                 # Option 1: Select from Local Filesystem
                 with gr.Row():
-                    search_directory = gr.Textbox(label="Model Directory",
-                                                  placeholder="Enter directory path(currently '.\Models')",
-                                                  value=".\Models",
-                                                  interactive=True)
+                    search_directory = gr.Textbox(
+                        label="Model Directory",
+                        placeholder="Enter directory path (currently './Models')",
+                        value=MODELS_DIR,
+                        interactive=True
+                    )
 
                 # Initial population of local models
-                initial_dropdown_update, _ = update_dropdowns(".\Models")
+                initial_dropdown_update, _ = update_dropdowns(MODELS_DIR)
+                logging.debug(f"Scanning directory: {MODELS_DIR}")
                 refresh_button = gr.Button("Refresh Models")
-                local_model_dropdown = gr.Dropdown(label="Select Model from Directory", choices=[])
+                local_model_dropdown = gr.Dropdown(
+                    label="Select Model from Directory",
+                    choices=initial_dropdown_update["choices"],
+                    value=None
+                )
                 # Display selected model path
                 model_value = gr.Textbox(label="Selected Model File Path", value="", interactive=False)
 

App_Function_Libraries/Gradio_UI/Website_scraping_tab.py

@@ -527,7 +527,6 @@ async def scrape_and_summarize_wrapper(
                         return convert_json_to_markdown(json.dumps({"error": f"Invalid JSON format for custom cookies: {e}"}))
 
                 if scrape_method == "Individual URLs":
-                    # FIXME modify scrape_and_summarize_multiple to accept custom_cookies
                     result = await scrape_and_summarize_multiple(url_input, custom_prompt, api_name, api_key, keywords,
                                                                  custom_titles, system_prompt, summarize_checkbox, custom_cookies=custom_cookies_list)
                 elif scrape_method == "Sitemap":

App_Function_Libraries/Local_LLM/Local_LLM_Inference_Engine_Lib.py

@@ -14,14 +14,13 @@
 ####################
 # Import necessary libraries
 #import atexit
-import glob
 import logging
 import os
 import re
 import signal
 import subprocess
 import sys
-import time
+from pathlib import Path
 from typing import List, Optional
 #
 # Import 3rd-pary Libraries
@@ -157,21 +156,25 @@ def get_gguf_llamafile_files(directory: str) -> List[str]:
     """
     logging.debug(f"Scanning directory: {directory}")  # Debug print for directory
 
-    # Print all files in the directory for debugging
-    all_files = os.listdir(directory)
-    logging.debug(f"All files in directory: {all_files}")
-
-    pattern_gguf = os.path.join(directory, "*.gguf")
-    pattern_llamafile = os.path.join(directory, "*.llamafile")
+    try:
+        dir_path = Path(directory)
+        all_files = list(dir_path.iterdir())
+        logging.debug(f"All files in directory: {[str(f) for f in all_files]}")
+    except Exception as e:
+        logging.error(f"Failed to list files in directory {directory}: {e}")
+        return []
 
-    gguf_files = glob.glob(pattern_gguf)
-    llamafile_files = glob.glob(pattern_llamafile)
+    try:
+        gguf_files = list(dir_path.glob("*.gguf"))
+        llamafile_files = list(dir_path.glob("*.llamafile"))
 
-    # Debug: Print the files found
-    logging.debug(f"Found .gguf files: {gguf_files}")
-    logging.debug(f"Found .llamafile files: {llamafile_files}")
+        logging.debug(f"Found .gguf files: {[str(f) for f in gguf_files]}")
+        logging.debug(f"Found .llamafile files: {[str(f) for f in llamafile_files]}")
 
-    return [os.path.basename(f) for f in gguf_files + llamafile_files]
+        return [f.name for f in gguf_files + llamafile_files]
+    except Exception as e:
+        logging.error(f"Error during glob operations in directory {directory}: {e}")
+        return []
 
 
 # Initialize process with type annotation

App_Function_Libraries/Local_LLM/Local_LLM_ollama.py

@@ -5,92 +5,197 @@
 import psutil
 import os
 import signal
-
+import logging
+import threading
+import shutil
+
+# Configure Logging
+# logging.basicConfig(
+#     level=logging.DEBUG,  # Set to DEBUG to capture all levels of logs
+#     format='%(asctime)s - %(levelname)s - %(message)s',
+#     handlers=[
+#         logging.FileHandler("app.log"),
+#         logging.StreamHandler()
+#     ]
+# )
+
+def is_ollama_installed():
+    """
+    Checks if the 'ollama' executable is available in the system's PATH.
+    Returns True if installed, False otherwise.
+    """
+    return shutil.which('ollama') is not None
 
 def get_ollama_models():
+    """
+    Retrieves available Ollama models by executing 'ollama list'.
+    Returns a list of model names or an empty list if an error occurs.
+    """
     try:
-        result = subprocess.run(['ollama', 'list'], capture_output=True, text=True, check=True)
+        result = subprocess.run(['ollama', 'list'], capture_output=True, text=True, check=True, timeout=10)
         models = result.stdout.strip().split('\n')[1:]  # Skip header
-        return [model.split()[0] for model in models]
-    except subprocess.CalledProcessError:
+        model_names = [model.split()[0] for model in models if model.strip()]
+        logging.debug(f"Available Ollama models: {model_names}")
+        return model_names
+    except FileNotFoundError:
+        logging.error("Ollama executable not found. Please ensure Ollama is installed and in your PATH.")
+        return []
+    except subprocess.TimeoutExpired:
+        logging.error("Ollama 'list' command timed out.")
+        return []
+    except subprocess.CalledProcessError as e:
+        logging.error(f"Error executing Ollama 'list': {e}")
+        return []
+    except Exception as e:
+        logging.error(f"Unexpected error in get_ollama_models: {e}")
         return []
-
 
 def pull_ollama_model(model_name):
+    """
+    Pulls the specified Ollama model if Ollama is installed.
+    """
+    if not is_ollama_installed():
+        logging.error("Ollama is not installed.")
+        return "Failed to pull model: Ollama is not installed or not in your PATH."
+
     try:
-        subprocess.run(['ollama', 'pull', model_name], check=True)
+        subprocess.run(['ollama', 'pull', model_name], check=True, timeout=300)  # Adjust timeout as needed
+        logging.info(f"Successfully pulled model: {model_name}")
         return f"Successfully pulled model: {model_name}"
+    except subprocess.TimeoutExpired:
+        logging.error(f"Pulling model '{model_name}' timed out.")
+        return f"Failed to pull model '{model_name}': Operation timed out."
     except subprocess.CalledProcessError as e:
-        return f"Failed to pull model: {e}"
-
+        logging.error(f"Failed to pull model '{model_name}': {e}")
+        return f"Failed to pull model '{model_name}': {e}"
+    except FileNotFoundError:
+        logging.error("Ollama executable not found. Please ensure Ollama is installed and in your PATH.")
+        return "Failed to pull model: Ollama executable not found."
+    except Exception as e:
+        logging.error(f"Unexpected error in pull_ollama_model: {e}")
+        return f"Failed to pull model '{model_name}': {e}"
 
 def serve_ollama_model(model_name, port):
+    """
+    Serves the specified Ollama model on the given port if Ollama is installed.
+    """
+    if not is_ollama_installed():
+        logging.error("Ollama is not installed.")
+        return "Error: Ollama is not installed or not in your PATH."
+
     try:
         # Check if a server is already running on the specified port
         for conn in psutil.net_connections():
             if conn.laddr.port == int(port):
-                return f"Port {port} is already in use. Please choose a different port."
+                logging.warning(f"Port {port} is already in use.")
+                return f"Error: Port {port} is already in use. Please choose a different port."
 
         # Start the Ollama server
-        port = str(port)
-        os.environ["OLLAMA_HOST"] = port
-        cmd = f"ollama serve"
-        process = subprocess.Popen(cmd, shell=True)
-        return f"Started Ollama server for model {model_name} on port {port}. Process ID: {process.pid}"
+        cmd = ['ollama', 'serve', model_name, '--port', str(port)]
+        process = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+        logging.info(f"Started Ollama server for model '{model_name}' on port {port}. PID: {process.pid}")
+        return f"Started Ollama server for model '{model_name}' on port {port}. Process ID: {process.pid}"
+    except FileNotFoundError:
+        logging.error("Ollama executable not found.")
+        return "Error: Ollama executable not found. Please ensure Ollama is installed and in your PATH."
     except Exception as e:
+        logging.error(f"Error starting Ollama server: {e}")
         return f"Error starting Ollama server: {e}"
 
-
 def stop_ollama_server(pid):
+    """
+    Stops the Ollama server with the specified process ID if Ollama is installed.
+    """
+    if not is_ollama_installed():
+        logging.error("Ollama is not installed.")
+        return "Error: Ollama is not installed or not in your PATH."
+
     try:
         if platform.system() == "Windows":
-            os.system(f"taskkill /F /PID {pid}")
-            return f"Stopped Ollama server with PID {pid}"
-        elif platform.system() == "Linux":
-            os.system(f"kill {pid}")
-            return f"Stopped Ollama server with PID {pid}"
-        elif platform.system() == "Darwin":
-            os.system("""osascript -e 'tell app "Ollama" to quit'""")
-            return f"(Hopefully) Stopped Ollama server using osascript..."
+            subprocess.run(['taskkill', '/F', '/PID', str(pid)], check=True)
+        elif platform.system() in ["Linux", "Darwin"]:
+            os.kill(pid, signal.SIGTERM)
+        logging.info(f"Stopped Ollama server with PID {pid}")
+        return f"Stopped Ollama server with PID {pid}"
     except ProcessLookupError:
+        logging.warning(f"No process found with PID {pid}")
         return f"No process found with PID {pid}"
     except Exception as e:
+        logging.error(f"Error stopping Ollama server: {e}")
         return f"Error stopping Ollama server: {e}"
 
-
 def create_ollama_tab():
+    """
+    Creates the Ollama Model Serving tab in the Gradio interface with lazy loading.
+    """
+    ollama_installed = is_ollama_installed()
+
     with gr.Tab("Ollama Model Serving"):
+        if not ollama_installed:
+            gr.Markdown(
+                "# Ollama Model Serving\n\n"
+                "**Ollama is not installed or not found in your PATH. Please install Ollama to use this feature.**"
+            )
+            return  # Exit early, no need to add further components
+
         gr.Markdown("# Ollama Model Serving")
 
         with gr.Row():
-            model_list = gr.Dropdown(label="Available Models", choices=get_ollama_models())
+            # Initialize Dropdowns with placeholders
+            model_list = gr.Dropdown(
+                label="Available Models",
+                choices=["Click 'Refresh Model List' to load models"],
+                value="Click 'Refresh Model List' to load models"
+            )
             refresh_button = gr.Button("Refresh Model List")
 
         with gr.Row():
-            new_model_name = gr.Textbox(label="Model to Pull")
+            new_model_name = gr.Textbox(label="Model to Pull", placeholder="Enter model name")
             pull_button = gr.Button("Pull Model")
 
         pull_output = gr.Textbox(label="Pull Status")
 
         with gr.Row():
-            # FIXME - Update to update config.txt file
-            serve_model = gr.Dropdown(label="Model to Serve", choices=get_ollama_models())
+            serve_model = gr.Dropdown(
+                label="Model to Serve",
+                choices=["Click 'Refresh Model List' to load models"],
+                value="Click 'Refresh Model List' to load models"
+            )
             port = gr.Number(label="Port", value=11434, precision=0)
             serve_button = gr.Button("Start Server")
 
         serve_output = gr.Textbox(label="Server Status")
 
         with gr.Row():
-            pid = gr.Number(label="Server Process ID", precision=0)
+            pid = gr.Number(label="Server Process ID (Enter the PID to stop)", precision=0)
             stop_button = gr.Button("Stop Server")
 
         stop_output = gr.Textbox(label="Stop Status")
 
         def update_model_lists():
+            """
+            Retrieves the list of available Ollama models and updates the dropdowns.
+            """
             models = get_ollama_models()
-            return gr.update(choices=models), gr.update(choices=models)
-
-        refresh_button.click(update_model_lists, outputs=[model_list, serve_model])
-        pull_button.click(pull_ollama_model, inputs=[new_model_name], outputs=[pull_output])
-        serve_button.click(serve_ollama_model, inputs=[serve_model, port], outputs=[serve_output])
-        stop_button.click(stop_ollama_server, inputs=[pid], outputs=[stop_output])
+            if models:
+                return gr.update(choices=models, value=models[0]), gr.update(choices=models, value=models[0])
+            else:
+                return gr.update(choices=["No models found"], value="No models found"), gr.update(choices=["No models found"], value="No models found")
+
+        def async_update_model_lists():
+            """
+            Asynchronously updates the model lists to prevent blocking.
+            """
+            def task():
+                choices1, choices2 = update_model_lists()
+                model_list.update(choices=choices1['choices'], value=choices1.get('value'))
+                serve_model.update(choices=choices2['choices'], value=choices2.get('value'))
+            threading.Thread(target=task).start()
+
+        # Bind the refresh button to the asynchronous update function
+        refresh_button.click(fn=async_update_model_lists, inputs=[], outputs=[])
+
+        # Bind the pull, serve, and stop buttons to their respective functions
+        pull_button.click(fn=pull_ollama_model, inputs=[new_model_name], outputs=[pull_output])
+        serve_button.click(fn=serve_ollama_model, inputs=[serve_model, port], outputs=[serve_output])
+        stop_button.click(fn=stop_ollama_server, inputs=[pid], outputs=[stop_output])

App_Function_Libraries/Utils/Utils.py

@@ -254,6 +254,8 @@ def load_and_log_configs():
         logging.debug(f"Loaded Tabby API IP: {tabby_api_IP}")
         logging.debug(f"Loaded VLLM API URL: {vllm_api_url}")
 
+        # Retrieve default API choices from the configuration file
+        default_api = config.get('API', 'default_api', fallback='openai')
 
         # Retrieve output paths from the configuration file
         output_path = config.get('Paths', 'output_path', fallback='results')
@@ -340,7 +342,8 @@ def load_and_log_configs():
                 'embedding_api_key': embedding_api_key,
                 'chunk_size': chunk_size,
                 'overlap': overlap
-            }
+            },
+            'default_api': default_api
         }
 
     except Exception as e: