Merge pull request #109 from FrancescoCaracciolo/master

src/constants.py

@@ -196,7 +196,7 @@
 Assistant: Hello, how can I assist you today?
 User: Can you help me?
 Assistant: Yes, of course, what do you need help with?""",
-    "get_suggestions_prompt": """Suggest a few questions that the user would ask and put them in a JSON array. You have to write ONLY the JSON array an nothing else""",
+    "get_suggestions_prompt": """Suggest a few questions that the user would ask and put them in a JSON array. You have to write ONLY the JSON array of strings an nothing else.""",
     "custom_prompt": "",
 
 }

src/extra.py

@@ -3,6 +3,7 @@
 import re, base64, io
 import os, sys
 import xml.dom.minidom, html
+import json
 
 class ReplaceHelper:
     DISTRO = None
@@ -263,6 +264,7 @@ def can_escape_sandbox() -> bool:
     except subprocess.CalledProcessError as _:
         return False
     return True
+
 def get_streaming_extra_setting():
             return {
                 "key": "streaming",
@@ -271,6 +273,7 @@ def get_streaming_extra_setting():
                 "type": "toggle",
                 "default": True
             }
+
 def override_prompts(override_setting, PROMPTS):
     prompt_list = {}
     for prompt in PROMPTS:
@@ -279,3 +282,77 @@ def override_prompts(override_setting, PROMPTS):
         else:
             prompt_list[prompt] = PROMPTS[prompt]
     return prompt_list
+
+
+def extract_json(input_string: str) -> str:
+    """Extract JSON string from input string
+
+    Args:
+        input_string (): The input string 
+
+    Returns:
+        str: The JSON string 
+    """
+    # Regular expression to find JSON objects or arrays
+    json_pattern = re.compile(r'\{.*?\}|\[.*?\]', re.DOTALL)
+
+    # Find all JSON-like substrings
+    matches = json_pattern.findall(input_string) 
+    # Parse each match and return the first valid JSON
+    for match in matches:
+        try:
+            json_data = json.loads(match)
+            return match
+        except json.JSONDecodeError:
+            continue
+    print("Wrong JSON", input_string)
+    return []
+
+
+def remove_markdown(text: str) -> str:
+    """
+    Remove markdown from text
+
+    Args:
+        text: The text to remove markdown from 
+
+    Returns:
+        str: The text without markdown 
+    """
+    # Remove headers
+    text = re.sub(r'^#{1,6}\s*', '', text, flags=re.MULTILINE)
+
+    # Remove emphasis (bold and italic)
+    text = re.sub(r'\*\*(.*?)\*\*', r'\1', text)  # Bold
+    text = re.sub(r'__(.*?)__', r'\1', text)          # Bold
+    text = re.sub(r'\*(.*?)\*', r'\1', text)        # Italic
+    text = re.sub(r'_(.*?)_', r'\1', text)            # Italic
+
+    # Remove inline code
+    text = re.sub(r'`([^`]*)`', r'\1', text)
+
+    # Remove code blocks
+    text = re.sub(r'```[\s\S]*?```', '', text)
+
+    # Remove links, keep the link text
+    text = re.sub(r'\[([^\]]+)\]\([^\)]+\)', r'\1', text)
+
+    # Remove images, keep the alt text
+    text = re.sub(r'!\[([^\]]*)\]\([^\)]+\)', r'\1', text)
+
+    # Remove strikethrough
+    text = re.sub(r'~~(.*?)~~', r'\1', text)
+
+    # Remove blockquotes
+    text = re.sub(r'^>\s*', '', text, flags=re.MULTILINE)
+
+    # Remove unordered list markers
+    text = re.sub(r'^\s*[-+*]\s+', '', text, flags=re.MULTILINE)
+
+    # Remove ordered list markers
+    text = re.sub(r'^\s*\d+\.\s+', '', text, flags=re.MULTILINE)
+
+    # Remove extra newlines
+    text = re.sub(r'\n{2,}', '\n', text)
+
+    return text.strip()

src/llm.py

@@ -4,7 +4,7 @@
 from typing import Callable, Any
 import json
 import base64
-from .extra import convert_history_openai, extract_image, find_module, get_streaming_extra_setting, install_module, open_website, get_image_path, get_spawn_command, quote_string
+from .extra import convert_history_openai, extract_image, extract_json, find_module, get_streaming_extra_setting, install_module, open_website, get_image_path, get_spawn_command, quote_string
 from .handler import Handler
 
 class LLMHandler(Handler):
@@ -130,7 +130,7 @@ def get_suggestions(self, request_prompt:str = "", amount:int=1) -> list[str]:
             history += message["User"] + ": " + message["Message"] + "\n"
         for i in range(0, amount):
             generated = self.generate_text(history + "\n\n" + request_prompt)
-            generated = generated.replace("```json", "").replace("```", "")
+            generated = extract_json(generated)
             try:
                 j = json.loads(generated)
             except Exception as _:
@@ -455,10 +455,36 @@ class GeminiHandler(LLMHandler):
     Official Google Gemini APIs, they support history and system prompts
     """
 
+    default_models = [("gemini-1.5-flash","gemini-1.5-flash"), ("gemini-1.5-flash-8b", "gemini-1.5-flash-8b") , ("gemini-1.0-pro", "gemini-1.0-pro"), ("gemini-1.5-pro","gemini-1.5-pro") ]
+
     def __init__(self, settings, path):
         super().__init__(settings, path)
         self.cache = {}
+        if self.get_setting("models", False) is None or len(self.get_setting("models", False)) == 0 or True:
+            self.models = self.default_models 
+            threading.Thread(target=self.get_models).start()
+        else:
+            self.models = json.loads(self.get_setting("models", False))
 
+    def get_models(self):
+        if self.is_installed():
+            try:
+                import google.generativeai as genai
+                api = self.get_setting("apikey", False)
+                if api is None:
+                    return
+                genai.configure(api_key=api)
+                models = genai.list_models()
+                result = tuple()
+                for model in models:
+                    if "generateContent" in model.supported_generation_methods:
+                        result += ((model.display_name, model.name,),)
+                self.models = result
+                self.set_setting("models", json.dumps(result))
+                self.settings_update()
+            except Exception as e:
+                print("Error getting " + self.key + " models: " + str(e))
+
     @staticmethod
     def get_extra_requirements() -> list:
         return ["google-generativeai"]
@@ -479,13 +505,21 @@ def get_extra_settings(self) -> list:
                 "type": "entry",
                 "default": ""
             },
+            {
+                "key": "refresh_models",
+                "title": _("Refresh Models"),
+                "description": _("Refresh models list"),
+                "type": "button",
+                "icon": "view-refresh-symbolic",
+                "callback": lambda button: self.get_models(),
+            },
             {
                 "key": "model",
                 "title": _("Model"),
                 "description": _("AI Model to use, available: gemini-1.5-pro, gemini-1.0-pro, gemini-1.5-flash"),
                 "type": "combo",
-                "default": "gemini-1.5-flash",
-                "values": [("gemini-1.5-flash-8b", "gemini-1.5-flash-8b"), ("gemini-1.5-flash","gemini-1.5-flash") , ("gemini-1.0-pro", "gemini-1.0-pro"), ("gemini-1.5-pro","gemini-1.5-pro") ]
+                "default": self.models[0][1],
+                "values": self.models, 
             },
             {
                 "key": "streaming",
@@ -1071,6 +1105,7 @@ def generate_text_stream(self, prompt: str, history: list[dict[str, str]] = [],
         from openai import OpenAI
         history.append({"User": "User", "Message": prompt})
         messages = self.convert_history(history, system_prompt)
+        print([message["role"] for message in messages])
         api = self.get_setting("api")
         if api == "":
             api = "nokey"
@@ -1116,7 +1151,7 @@ def get_extra_settings(self) -> list:
 
 class GroqHandler(OpenAIHandler):
     key = "groq"
-    default_models = (("llama-3.1-70B-versatile", "llama-3.1-70B-versatile" ), ) 
+    default_models = (("llama-3.3-70B-versatile", "llama-3.3-70B-versatile" ), ) 
     def supports_vision(self) -> bool:
         return "vision" in self.get_setting("model")
 

src/window.py

@@ -10,7 +10,7 @@
 from .constants import AVAILABLE_LLMS, AVAILABLE_PROMPTS, PROMPTS, AVAILABLE_TTS, AVAILABLE_STT
 from gi.repository import Gtk, Adw, Pango, Gio, Gdk, GObject, GLib, GdkPixbuf
 from .stt import AudioRecorder
-from .extra import get_spawn_command, install_module, markwon_to_pango, override_prompts, replace_variables
+from .extra import get_spawn_command, install_module, markwon_to_pango, override_prompts, remove_markdown, replace_variables
 import threading
 import posixpath
 import json
@@ -537,7 +537,7 @@ def handle_second_block_change(self,*a):
         elif type(header_widget) is Gtk.Box:
             self.explorer_panel_headerbox.append(self.headerbox)
 
-    def on_flap_button_toggled(self, toggle_button):
+    def on_flap_button_toggled(self, toggle_button): 
         self.flap_button_left.set_active(True)
         if self.main_program_block.get_name() == "visible":
             self.main_program_block.set_name("hide")
@@ -711,8 +711,8 @@ def generate_chat_name(self, button, multithreading=False):
             button.set_child(spinner)
             button.set_can_target(False)
             button.set_has_frame(True)
-            # TODO: take the history for the correct chat
-            self.model.set_history([], self)
+             
+            self.model.set_history([], self.get_history(self.chats[int(button.get_name())]["chat"]))
             name = self.model.generate_chat_name(self.prompts["generate_name_prompt"])
             if name != "Chat has been stopped":
                 self.chats[int(button.get_name())]["name"] = name
@@ -1248,10 +1248,12 @@ def wait_threads_sm():
         GLib.idle_add(self.scrolled_chat)
         self.save_chat()
 
-    def get_history(self) -> list[dict[str, str]]: 
+    def get_history(self, chat = None) -> list[dict[str, str]]: 
+        if chat is None:
+            chat = self.chat
         history = []
         count = self.memory
-        for msg in self.chat[:-1]:
+        for msg in chat[:-1]:
             if count == 0:
                 break
             if msg["User"] == "Console" and msg["Message"] == "None":
@@ -1281,7 +1283,7 @@ def send_message(self):
             self.curr_label = ""
             GLib.idle_add(self.create_streaming_message_label)
             self.streaming_lable = None
-            message_label = self.model.send_message_stream(self, self.chat[-1]["Message"], self.update_message)
+            message_label = self.model.send_message_stream(self, self.chat[-1]["Message"], self.update_message, [stream_number_variable])
             try:
                 self.streaming_box.get_parent().set_visible(False)
             except:
@@ -1295,6 +1297,7 @@ def send_message(self):
         tts_thread = None
         if self.tts_enabled: 
             message=re.sub(r"```.*?```", "", message_label, flags=re.DOTALL)
+            message = remove_markdown(message)
             if not(not message.strip() or message.isspace() or all(char == '\n' for char in message)):
                 tts_thread = threading.Thread(target=self.tts.play_audio, args=(message, ))
                 tts_thread.start()
@@ -1315,16 +1318,18 @@ def create_streaming_message_label(self):
         self.streaming_label = Gtk.TextView(wrap_mode=Gtk.WrapMode.WORD_CHAR, editable=False, hexpand=True)
         scrolled_window.add_css_class("scroll")
         self.streaming_label.add_css_class("scroll")
-        apply_css_to_widget(scrolled_window, ".scroll { background-color: rgba(0,0,0,0)}")
-        apply_css_to_widget(self.streaming_label, ".scroll { background-color: rgba(0,0,0,0)}")
+        apply_css_to_widget(scrolled_window, ".scroll { background-color: rgba(0,0,0,0);}")
+        apply_css_to_widget(self.streaming_label, ".scroll { background-color: rgba(0,0,0,0);}")
         scrolled_window.set_child(self.streaming_label)
         text_buffer = self.streaming_label.get_buffer()
         tag = text_buffer.create_tag("no-background", background_set=False, paragraph_background_set=False)
         text_buffer.apply_tag(tag, text_buffer.get_start_iter(), text_buffer.get_end_iter())
         self.streaming_box=self.add_message("Assistant", scrolled_window)
         self.streaming_box.set_overflow(Gtk.Overflow.VISIBLE)
 
-    def update_message(self, message):  
+    def update_message(self, message, stream_number_variable):  
+        if self.stream_number_variable != stream_number_variable:
+            return
         self.streamed_message = message
         if self.streaming_label is not None:
             added_message = message[len(self.curr_label):]