Upload 2 files

App_Function_Libraries/Summarization/Local_Summarization_Lib.py

@@ -35,6 +35,25 @@ from App_Function_Libraries.Utils.Utils import load_and_log_configs, extract_tex
 logger = logging.getLogger()
 
 
+summarizer_prompt = """
+                    <s>You are a bulleted notes specialist. [INST]```When creating comprehensive bulleted notes, you should follow these guidelines: Use multiple headings based on the referenced topics, not categories like quotes or terms. Headings should be surrounded by bold formatting and not be listed as bullet points themselves. Leave no space between headings and their corresponding list items underneath. Important terms within the content should be emphasized by setting them in bold font. Any text that ends with a colon should also be bolded. Before submitting your response, review the instructions, and make any corrections necessary to adhered to the specified format. Do not reference these instructions within the notes.``` \nBased on the content between backticks create comprehensive bulleted notes.[/INST]
+                        **Bulleted Note Creation Guidelines**
+
+                        **Headings**:
+                        - Based on referenced topics, not categories like quotes or terms
+                        - Surrounded by **bold** formatting
+                        - Not listed as bullet points
+                        - No space between headings and list items underneath
+
+                        **Emphasis**:
+                        - **Important terms** set in bold font
+                        - **Text ending in a colon**: also bolded
+
+                        **Review**:
+                        - Ensure adherence to specified format
+                        - Do not reference these instructions in your response.</s>[INST] {{ .Prompt }} [/INST]
+                    """
+
 # FIXME - temp is not used
 def summarize_with_local_llm(input_data, custom_prompt_arg, temp, system_message=None):
     try:
@@ -108,7 +127,7 @@ def summarize_with_local_llm(input_data, custom_prompt_arg, temp, system_message
         return "Local LLM: Error occurred while processing summary"
 
 
-def summarize_with_llama(input_data, custom_prompt, api_url="http://127.0.0.1:8080/completion", api_key=None, temp=None, system_message=None):
+def summarize_with_llama(input_data, custom_prompt, api_key=None, temp=None, system_message=None, api_url="http://127.0.0.1:8080/completion",):
     try:
         logging.debug("Llama.cpp: Loading and validating configurations")
         loaded_config_data = load_and_log_configs()
@@ -138,12 +157,12 @@ def summarize_with_llama(input_data, custom_prompt, api_url="http://127.0.0.1:80
             logging.debug("Llama.cpp: Using provided string data for summarization")
             data = input_data
 
-        logging.debug(f"Llama.cpp: Loaded data: {data}")
-        logging.debug(f"Llama.cpp: Type of data: {type(data)}")
+        logging.debug(f"Llama Summarize: Loaded data: {data}")
+        logging.debug(f"Llama Summarize: Type of data: {type(data)}")
 
         if isinstance(data, dict) and 'summary' in data:
             # If the loaded data is a dictionary and already contains a summary, return it
-            logging.debug("Llama.cpp: Summary already exists in the loaded data")
+            logging.debug("Llama Summarize: Summary already exists in the loaded data")
             return data['summary']
 
         # If the loaded data is a list of segment dictionaries or a string, proceed with summarization
@@ -153,7 +172,7 @@ def summarize_with_llama(input_data, custom_prompt, api_url="http://127.0.0.1:80
         elif isinstance(data, str):
             text = data
         else:
-            raise ValueError("Llama.cpp: Invalid input data format")
+            raise ValueError("Llama Summarize: Invalid input data format")
 
         headers = {
             'accept': 'application/json',
@@ -162,13 +181,17 @@ def summarize_with_llama(input_data, custom_prompt, api_url="http://127.0.0.1:80
         if len(api_key) > 5:
             headers['Authorization'] = f'Bearer {api_key}'
 
-        llama_prompt = f"{custom_prompt} \n\n\n\n{text}"
         if system_message is None:
             system_message = "You are a helpful AI assistant."
-        logging.debug("llama: Prompt being sent is {llama_prompt}")
+        logging.debug(f":Llama Summarize: System Prompt being sent is {system_message}")
         if system_message is None:
             system_message = "You are a helpful AI assistant."
 
+        if custom_prompt is None:
+            llama_prompt = f"{summarizer_prompt}\n\n\n\n{text}"
+        else:
+            llama_prompt = f"{custom_prompt}\n\n\n\n{text}"
+
         data = {
             "messages": [
                 {"role": "system", "content": system_message},
@@ -201,7 +224,7 @@ def summarize_with_llama(input_data, custom_prompt, api_url="http://127.0.0.1:80
 
 
 # https://lite.koboldai.net/koboldcpp_api#/api%2Fv1/post_api_v1_generate
-def summarize_with_kobold(input_data, api_key, custom_prompt_input, kobold_api_ip="http://127.0.0.1:5001/api/v1/generate", temp=None, system_message=None):
+def summarize_with_kobold(input_data, api_key, custom_prompt_input,  system_message=None, temp=None, kobold_api_ip="http://127.0.0.1:5001/api/v1/generate"):
     logging.debug("Kobold: Summarization process starting...")
     try:
         logging.debug("Kobold: Loading and validating configurations")
@@ -253,9 +276,12 @@ def summarize_with_kobold(input_data, api_key, custom_prompt_input, kobold_api_i
             'accept': 'application/json',
             'content-type': 'application/json',
         }
+        if custom_prompt_input is None:
+            kobold_prompt = f"{summarizer_prompt}\n\n\n\n{text}"
+        else:
+            kobold_prompt = f"{custom_prompt_input}\n\n\n\n{text}"
 
-        kobold_prompt = f"{custom_prompt_input}\n\n\n\n{text}"
-        logging.debug("kobold: Prompt being sent is {kobold_prompt}")
+        logging.debug("Kobold summarization: Prompt being sent is {kobold_prompt}")
 
         # FIXME
         # Values literally c/p from the api docs....
@@ -269,12 +295,12 @@ def summarize_with_kobold(input_data, api_key, custom_prompt_input, kobold_api_i
             #"rep_penalty": 1.0,
         }
 
-        logging.debug("kobold: Submitting request to API endpoint")
-        print("kobold: Submitting request to API endpoint")
+        logging.debug("Kobold Summarization: Submitting request to API endpoint")
+        print("Kobold Summarization: Submitting request to API endpoint")
         kobold_api_ip = loaded_config_data['local_api_ip']['kobold']
         try:
             response = requests.post(kobold_api_ip, headers=headers, json=data)
-            logging.debug("kobold: API Response Status Code: %d", response.status_code)
+            logging.debug("Kobold Summarization: API Response Status Code: %d", response.status_code)
 
             if response.status_code == 200:
                 try:
@@ -303,7 +329,7 @@ def summarize_with_kobold(input_data, api_key, custom_prompt_input, kobold_api_i
 
 
 # https://github.com/oobabooga/text-generation-webui/wiki/12-%E2%80%90-OpenAI-API
-def summarize_with_oobabooga(input_data, api_key, custom_prompt, api_url="http://127.0.0.1:5000/v1/chat/completions", temp=None, system_message=None):
+def summarize_with_oobabooga(input_data, api_key, custom_prompt, system_message=None, temp=None, api_url="http://127.0.0.1:5000/v1/chat/completions"):
     logging.debug("Oobabooga: Summarization process starting...")
     try:
         logging.debug("Oobabooga: Loading and validating configurations")
@@ -356,9 +382,13 @@ def summarize_with_oobabooga(input_data, api_key, custom_prompt, api_url="http:/
             'content-type': 'application/json',
         }
 
-        # prompt_text = "I like to eat cake and bake cakes. I am a baker. I work in a French bakery baking cakes. It
-        # is a fun job. I have been baking cakes for ten years. I also bake lots of other baked goods, but cakes are
-        # my favorite." prompt_text += f"\n\n{text}"  # Uncomment this line if you want to include the text variable
+        if custom_prompt is None:
+            custom_prompt = f"{summarizer_prompt}\n\n\n\n{text}"
+        else:
+            custom_prompt = f"{custom_prompt}\n\n\n\n{text}"
+
+        logging.debug("Ooba Summarize: Prompt being sent is {kobold_prompt}")
+
         ooba_prompt = f"{text}" + f"\n\n\n\n{custom_prompt}"
         logging.debug("ooba: Prompt being sent is {ooba_prompt}")
 
@@ -392,8 +422,7 @@ def summarize_with_oobabooga(input_data, api_key, custom_prompt, api_url="http:/
         return f"ooba: Error occurred while processing summary with oobabooga: {str(e)}"
 
 
-
-def summarize_with_tabbyapi(input_data, custom_prompt_input, api_key=None, api_IP="http://127.0.0.1:5000/v1/chat/completions", temp=None, system_message=None):
+def summarize_with_tabbyapi(input_data, custom_prompt_input, system_message=None, api_key=None, temp=None, api_IP="http://127.0.0.1:5000/v1/chat/completions"):
     logging.debug("TabbyAPI: Summarization process starting...")
     try:
         logging.debug("TabbyAPI: Loading and validating configurations")
@@ -448,6 +477,11 @@ def summarize_with_tabbyapi(input_data, custom_prompt_input, api_key=None, api_I
         if system_message is None:
             system_message = "You are a helpful AI assistant."
 
+        if custom_prompt_input is None:
+            custom_prompt_input = f"{summarizer_prompt}\n\n\n\n{text}"
+        else:
+            custom_prompt_input = f"{custom_prompt_input}\n\n\n\n{text}"
+
         headers = {
             'Authorization': f'Bearer {api_key}',
             'Content-Type': 'application/json'
@@ -501,10 +535,10 @@ def summarize_with_vllm(
         input_data: Union[str, dict, list],
         custom_prompt_input: str,
         api_key: str = None,
-        vllm_api_url: str = "http://127.0.0.1:8000/v1/chat/completions",
         model: str = None,
         system_prompt: str = None,
-        temp: float = 0.7
+        temp: float = 0.7,
+        vllm_api_url: str = "http://127.0.0.1:8000/v1/chat/completions"
 ) -> str:
     logging.debug("vLLM: Summarization process starting...")
     try:
@@ -556,6 +590,11 @@ def summarize_with_vllm(
         if system_prompt is None:
             system_prompt = "You are a helpful AI assistant."
 
+        if custom_prompt_input is None:
+            custom_prompt_input = f"{summarizer_prompt}\n\n\n\n{text}"
+        else:
+            custom_prompt_input = f"{custom_prompt_input}\n\n\n\n{text}"
+
         model = model or loaded_config_data['models']['vllm']
         if system_prompt is None:
             system_prompt = "You are a helpful AI assistant."
@@ -602,7 +641,7 @@ def summarize_with_vllm(
 
 
 # FIXME - update to be a summarize request
-def summarize_with_ollama(input_data, custom_prompt, api_url="http://127.0.0.1:11434/api/generate", api_key=None, temp=None, system_message=None, model=None):
+def summarize_with_ollama(input_data, custom_prompt, api_key=None, temp=None, system_message=None, model=None, api_url="http://127.0.0.1:11434/api/generate",):
     try:
         logging.debug("ollama: Loading and validating configurations")
         loaded_config_data = load_and_log_configs()
@@ -651,6 +690,11 @@ def summarize_with_ollama(input_data, custom_prompt, api_url="http://127.0.0.1:1
         else:
             raise ValueError("Ollama: Invalid input data format")
 
+        if custom_prompt is None:
+            custom_prompt = f"{summarizer_prompt}\n\n\n\n{text}"
+        else:
+            custom_prompt = f"{custom_prompt}\n\n\n\n{text}"
+
         headers = {
             'accept': 'application/json',
             'content-type': 'application/json',
@@ -761,6 +805,11 @@ def summarize_with_custom_openai(api_key, input_data, custom_prompt_arg, temp=No
         logging.debug(f"Custom OpenAI API: Extracted text (first 500 chars): {text[:500]}...")
         logging.debug(f"v: Custom prompt: {custom_prompt_arg}")
 
+        if input_data is None:
+            input_data = f"{summarizer_prompt}\n\n\n\n{text}"
+        else:
+            input_data = f"{input_data}\n\n\n\n{text}"
+
         openai_model = loaded_config_data['models']['openai'] or "gpt-4o"
         logging.debug(f"Custom OpenAI API: Using model: {openai_model}")
 

App_Function_Libraries/Summarization/Summarization_General_Lib.py

@@ -30,7 +30,8 @@ from App_Function_Libraries.Chunk_Lib import semantic_chunking, rolling_summariz
     improved_chunking_process
 from App_Function_Libraries.Audio.Diarization_Lib import combine_transcription_and_diarization
 from App_Function_Libraries.Summarization.Local_Summarization_Lib import summarize_with_llama, summarize_with_kobold, \
-    summarize_with_oobabooga, summarize_with_tabbyapi, summarize_with_vllm, summarize_with_local_llm
+    summarize_with_oobabooga, summarize_with_tabbyapi, summarize_with_vllm, summarize_with_local_llm, \
+    summarize_with_ollama, summarize_with_custom_openai
 from App_Function_Libraries.DB.DB_Manager import add_media_to_database
 # Import Local
 from App_Function_Libraries.Utils.Utils import load_and_log_configs, load_comprehensive_config, sanitize_filename, \
@@ -1108,9 +1109,15 @@ def process_video_urls(url_list, num_speakers, whisper_model, custom_prompt_inpu
 
 
 def perform_transcription(video_path, offset, whisper_model, vad_filter, diarize=False):
+    temp_files = []
+    logging.info(f"Processing media: {video_path}")
     global segments_json_path
     audio_file_path = convert_to_wav(video_path, offset)
+    logging.debug(f"Converted audio file: {audio_file_path}")
+    temp_files.append(audio_file_path)
+    logging.debug("Replacing audio file with segments.json file")
     segments_json_path = audio_file_path.replace('.wav', '.segments.json')
+    temp_files.append(segments_json_path)
 
     if diarize:
         diarized_json_path = audio_file_path.replace('.wav', '.diarized.json')
@@ -1521,16 +1528,22 @@ def process_url(
                     summary = summarize_with_deepseek(api_key, chunk, custom_prompt_input, temp, system_message)
                 elif api_name == "OpenRouter":
                     summary = summarize_with_openrouter(api_key, chunk, custom_prompt_input, temp, system_message)
+                # Local LLM APIs
                 elif api_name == "Llama.cpp":
-                    summary = summarize_with_llama(chunk, custom_prompt_input, temp, system_message)
+                    summary = summarize_with_llama(chunk, custom_prompt_input, api_key, temp, system_message)
                 elif api_name == "Kobold":
-                    summary = summarize_with_kobold(chunk, custom_prompt_input, temp, system_message)
+                    summary = summarize_with_kobold(chunk, None, custom_prompt_input, system_message, temp)
                 elif api_name == "Ooba":
-                    summary = summarize_with_oobabooga(chunk, custom_prompt_input, temp, system_message)
+                    summary = summarize_with_oobabooga(chunk, None, custom_prompt_input, system_message, temp)
                 elif api_name == "Tabbyapi":
-                    summary = summarize_with_tabbyapi(chunk, custom_prompt_input, temp, system_message)
+                    summary = summarize_with_tabbyapi(chunk, custom_prompt_input, system_message, None, temp)
                 elif api_name == "VLLM":
-                    summary = summarize_with_vllm(chunk, custom_prompt_input, temp, system_message)
+                    summary = summarize_with_vllm(chunk, custom_prompt_input, None, None, system_message)
+                elif api_name == "Ollama":
+                    summary = summarize_with_ollama(chunk, custom_prompt_input, api_key, temp, system_message, None)
+                elif api_name == "custom_openai_api":
+                    summary = summarize_with_custom_openai(chunk, custom_prompt_input, api_key, temp=None, system_message=None)
+
                 summarized_chunk_transcriptions.append(summary)
 
         # Combine chunked transcriptions into a single file