Merge branch 'main' into feature/bump-dialog-lib

docker-compose.dev-container.yml

@@ -39,3 +39,21 @@ services:
       - PGPASSWORD=talkdai
     env_file:
       - .env
+  openwebui:
+    image: ghcr.io/open-webui/open-webui:main
+    ports:
+      - '3000:8080'
+    environment:
+      - OPENAI_API_KEYS=FAKE-KEY;
+      - OPENAI_API_BASE_URLS=http://dialog:8000/openai;
+      - ENABLE_OPENAI_API=true
+    volumes:
+      - open-webui:/app/backend/data
+    depends_on:
+      db:
+        condition: service_healthy
+      dialog:
+        condition: service_started
+
+volumes:
+  open-webui:

docs/quick-start.md

@@ -98,7 +98,7 @@ See [our documentation](https://dialog.talkd.ai/settings#csv-knowledge-base) for
 
 #### `.toml` prompt configuration
 
-The `[prompt.header]`, `[prompt.suggested]`, and `[fallback.prompt]` fields are mandatory fields used for processing the conversation and connecting to the LLM.
+The `[prompt.header]`, `[prompt.history_header]`, `[prompt.suggested]`, and `[fallback.prompt]` fields are mandatory fields used for processing the conversation and connecting to the LLM.
 
 The `[prompt.fallback]` field is used when the LLM does not find a compatible embedding in the database; that is, the `[prompt.header]` **is ignored** and the `[prompt.fallback]` is used. Without it, there could be hallucinations about possible answers to questions outside the scope of the embeddings.
 
@@ -117,6 +117,8 @@ qualified service to high-end customers. Be brief in your answers, without being
 and objective in your responses. Never say that you are a model (AI), always answer as Avelino.
 Be polite and friendly!"""
 
+history_header = """This is the history of the conversation with the user:"""
+
 suggested = "Here is some possible content
 that could help the user in a better way."
 

docs/settings.md

@@ -28,7 +28,7 @@ Here is a brief explanation of the environment variables:
  - `DIALOG_LOADCSV_CLEARDB`: if set to `true`, the script `load_csv.py` will delete all previously imported vectors and reimport everything again.
  - `COSINE_SIMILARITY_THRESHOLD`: the cosine similarity threshold used to filter the results from the database's similarity query. The default is `0.5`.
  - `PLUGINS`: the path to the plugins that will be loaded into the application comma-separated. An example is: `dialog-whatsapp,plugins.my-custom-plugin`.
-
+ - `OPENWEB_UI_SESSION`: the session ID used to connect to the OpenWeb UI API. By default, it is `dialog-openweb-ui` and it can be changed.
 
 ## CSV format
 

sample_data/prompt.toml

@@ -16,6 +16,10 @@ suggested = """
 Here are some contents that you need to use in order to help you with the user's question:
 """
 
+history_header = """
+This is the history of the conversation:
+"""
+
 [fallback]
 prompt = """
 Answer the following message to the user.

src/dialog/llm/agents/default.py

@@ -17,6 +17,7 @@
 from dialog.settings import Settings
 from dialog.db import get_session
 
+
 class DialogLLM(AbstractRAG):
     def __init__(self, *args, **kwargs):
         kwargs["dbsession"] = next(get_session())
@@ -47,24 +48,35 @@ def generate_prompt(self, text):
             self.config.get("fallback").get("prompt") # maintaining compatibility with the previous configuration
         header = prompt_config.get("header")
         suggested = prompt_config.get("suggested")
+        history_header = prompt_config.get("history_header")
         messages = []
+
+        messages.append(SystemMessagePromptTemplate.from_template(header))
         if len(self.relevant_contents) > 0:
             context = "Context: \n".join(
                 [f"{c.question}\n{c.content}\n" for c in self.relevant_contents]
             )
-            messages.append(SystemMessagePromptTemplate.from_template(header))
-            messages.append(SystemMessagePromptTemplate.from_template(
-                f"{suggested}. {context}"))
             messages.append(
-                MessagesPlaceholder(
-                    variable_name="chat_history", optional=True))
-            messages.append(
-                HumanMessagePromptTemplate.from_template("{user_message}"))
+                SystemMessagePromptTemplate.from_template(f"{suggested}. {context}")
+            )
         else:
             messages.append(
-                SystemMessagePromptTemplate.from_template(fallback))
-            messages.append(
-                HumanMessagePromptTemplate.from_template("{user_message}"))
+                SystemMessagePromptTemplate.from_template(fallback)
+            )
+
+        messages.append(
+            SystemMessagePromptTemplate.from_template(history_header)
+        )
+
+        messages.append(
+            MessagesPlaceholder(
+                variable_name="chat_history"
+            )
+        )
+
+        messages.append(
+            HumanMessagePromptTemplate.from_template("{user_message}")
+        )
         self.prompt = ChatPromptTemplate.from_messages(messages)
 
         if Settings().VERBOSE_LLM:

src/dialog/routers/openai.py

@@ -1,7 +1,8 @@
 # *-* coding: utf-8 *-*
-from uuid import uuid4
-import datetime
+import os
 import logging
+import datetime
+from uuid import uuid4
 
 from dialog.db import engine, get_session
 from dialog_lib.db.models import Chat as ChatEntity, ChatMessages
@@ -44,11 +45,20 @@ async def ask_question_to_llm(message: OpenAIChat, session: Session = Depends(ge
     """
     This posts a message to the LLM and returns the response in the OpenAI format.
     """
+
     start_time = datetime.datetime.now()
-    new_chat = ChatEntity(
-        session_id = f"openai-{str(uuid4())}",
-    )
-    session.add(new_chat)
+    chat_entity = session.query(ChatEntity).filter(ChatEntity.session_id == Settings().OPENWEB_UI_SESSION).first()
+
+    if not chat_entity:
+        logging.info("Creating new chat entity")
+        new_chat = ChatEntity(
+            session_id = Settings().OPENWEB_UI_SESSION,
+        )
+        session.add(new_chat)
+        session.flush()
+    else:
+        logging.info("Using old chat entity")
+        new_chat = chat_entity
 
     non_empty_messages = []
 

src/dialog/settings.py

@@ -1,12 +1,15 @@
 import tomllib
 import logging
+from uuid import uuid4
 from pathlib import Path
 from decouple import Csv, Config
 
 config = Config(".env")
 
 logger = logging.getLogger(__name__)
 
+openweb_ui_session_id_fallback = uuid4()
+
 class Settings:
 
     @property
@@ -113,4 +116,8 @@ def CORS_ALLOW_METHODS(self):
 
     @property
     def CORS_ALLOW_HEADERS(self):
-        return config.get("CORS_ALLOW_HEADERS", cast=Csv(), default="*")
+        return config.get("CORS_ALLOW_HEADERS", cast=Csv(), default="*")
+
+    @property
+    def OPENWEB_UI_SESSION(self):
+        return config.get("OPENWEB_UI_SESSION", f"dialog-openweb-ui")