code refactor part 1 complete need to test

.dockerignore

@@ -1,29 +1,29 @@
-#  Ignore node_modules
-node_modules
-
-# Ignore logs
-logs
-*.log
-
-# Ignore temporary files
-tmp
-*.tmp
-
-# Ignore build directories
-dist
-build
-
-# Ignore environment variables
-.env
-
-# Ignore Docker files
-Dockerfile
-docker-compose.yml
-
-# Ignore IDE specific files
-.vscode
-.idea
-
-# Ignore OS generated files
-.DS_Store
+#  Ignore node_modules
+node_modules
+
+# Ignore logs
+logs
+*.log
+
+# Ignore temporary files
+tmp
+*.tmp
+
+# Ignore build directories
+dist
+build
+
+# Ignore environment variables
+.env
+
+# Ignore Docker files
+Dockerfile
+docker-compose.yml
+
+# Ignore IDE specific files
+.vscode
+.idea
+
+# Ignore OS generated files
+.DS_Store
 Thumbs.db

.gitignore

@@ -1,135 +1,137 @@
-# Byte-compiled / optimized / DLL files
-__pycache__/
-*.py[cod]
-*$py.class
-
-# C extensions
-*.so
-
-# Distribution / packaging
-.Python
-build/
-develop-eggs/
-dist/
-downloads/
-eggs/
-.eggs/
-lib/
-lib64/
-parts/
-sdist/
-var/
-wheels/
-pip-wheel-metadata/
-share/python-wheels/
-*.egg-info/
-.installed.cfg
-*.egg
-MANIFEST
-
-# PyInstaller
-#  Usually these files are written by a python script from a template
-#  before PyInstaller builds the exe, so as to inject date/other infos into it.
-*.manifest
-*.spec
-
-# Installer logs
-pip-log.txt
-pip-delete-this-directory.txt
-
-# Unit test / coverage reports
-htmlcov/
-.tox/
-.nox/
-.coverage
-.coverage.*
-.cache
-nosetests.xml
-coverage.xml
-*.cover
-*.py,cover
-.hypothesis/
-.pytest_cache/
-
-# Translations
-*.mo
-*.pot
-
-# Django stuff:
-*.log
-local_settings.py
-db.sqlite3
-db.sqlite3-journal
-
-# Flask stuff:
-instance/
-.webassets-cache
-
-# Scrapy stuff:
-.scrapy
-
-# Sphinx documentation
-docs/_build/
-
-# PyBuilder
-target/
-
-# Jupyter Notebook
-.ipynb_checkpoints
-
-# IPython
-profile_default/
-ipython_config.py
-
-# pyenv
-.python-version
-
-# pipenv
-#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
-#   However, in case of collaboration, if having platform-specific dependencies or dependencies
-#   having no cross-platform support, pipenv may install dependencies that don't work, or not
-#   install all needed dependencies.
-#Pipfile.lock
-
-# PEP 582; used by e.g. github.com/David-OConnor/pyflow
-__pypackages__/
-
-# Celery stuff
-celerybeat-schedule
-celerybeat.pid
-
-# SageMath parsed files
-*.sage.py
-
-# Environments
-.env
-.venv
-env/
-venv/
-ENV/
-env.bak/
-venv.bak/
-
-# Spyder project settings
-.spyderproject
-.spyproject
-
-# Rope project settings
-.ropeproject
-
-# mkdocs documentation
-/site
-
-# mypy
-.mypy_cache/
-.dmypy.json
-dmypy.json
-
-# Pyre type checker
-.pyre/
-
-# testing files generated
-*.txt.json
-
-*.ipynb
-env
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# testing files generated
+*.txt.json
+
+*.ipynb
+env
+
+.reggpt/vectorstores/faiss_embeddings_2024/

CHANGELOG.txt

@@ -1,2 +1,3 @@
-2023-11-30 pipeline with only document retrieval
-2024-08-23 azure app serice , open ai 'gpt4o mini'
+2023-11-30 pipeline with only document retrieval
+2024-08-23 azure app serice , open ai 'gpt4o mini'
+2024-10-16 Code Refactor

Dockerfile

@@ -1,42 +1,42 @@
-# Step 1: Use Python 3.11.9 as required
-FROM python:3.11.9
-
-# Step 2: Set up environment variables and timezone configuration
-ENV TZ=Asia/Colombo
-RUN apt-get update && apt-get install -y libaio1 wget unzip tzdata \
-    && ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone
-
-# Step 4: Add a user for running the app (after installations)
-RUN useradd -m -u 1000 user
-
-# Step 5: Create the /app directory and set ownership to the new user
-RUN mkdir -p /app && chown -R user:user /app
-
-# Step 6: Switch to non-root user after the directory has the right permissions
-USER user
-ENV PATH="/home/user/.local/bin:$PATH"
-
-# Step 7: Set up the working directory for the app
-WORKDIR /app
-
-# Step 8: Copy the requirements file and install dependencies
-COPY --chown=user ./requirements.txt requirements.txt
-RUN pip install --no-cache-dir --upgrade -r requirements.txt
-
-# Step 9: Install pipenv and handle pipenv environment
-RUN pip install pipenv
-COPY --chown=user . /app
-RUN pipenv install
-
-# Step 10: Expose the necessary port (7860 for Hugging Face Spaces)
-EXPOSE 7860
-
-# Step 11: Set environment variables for the app
-ENV APP_HOST=0.0.0.0 
-ENV APP_PORT=7860
-
-# Step 12: Create logs directory (if necessary)
-RUN mkdir -p /app/logs
-
-# Step 13: Run the app using Uvicorn, listening on port 7860
+# Step 1: Use Python 3.11.9 as required
+FROM python:3.11.9
+
+# Step 2: Set up environment variables and timezone configuration
+ENV TZ=Asia/Colombo
+RUN apt-get update && apt-get install -y libaio1 wget unzip tzdata \
+    && ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone
+
+# Step 4: Add a user for running the app (after installations)
+RUN useradd -m -u 1000 user
+
+# Step 5: Create the /app directory and set ownership to the new user
+RUN mkdir -p /app && chown -R user:user /app
+
+# Step 6: Switch to non-root user after the directory has the right permissions
+USER user
+ENV PATH="/home/user/.local/bin:$PATH"
+
+# Step 7: Set up the working directory for the app
+WORKDIR /app
+
+# Step 8: Copy the requirements file and install dependencies
+COPY --chown=user ./requirements.txt requirements.txt
+RUN pip install --no-cache-dir --upgrade -r requirements.txt
+
+# Step 9: Install pipenv and handle pipenv environment
+RUN pip install pipenv
+COPY --chown=user . /app
+RUN pipenv install
+
+# Step 10: Expose the necessary port (7860 for Hugging Face Spaces)
+EXPOSE 7860
+
+# Step 11: Set environment variables for the app
+ENV APP_HOST=0.0.0.0 
+ENV APP_PORT=7860
+
+# Step 12: Create logs directory (if necessary)
+RUN mkdir -p /app/logs
+
+# Step 13: Run the app using Uvicorn, listening on port 7860
 CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

LICENSE

@@ -1,22 +1,22 @@
-License
-
-Copyright (2024-2025) AI Labs, IronOne Technologies, LLC
-All Rights Reserved
-
-This source code is protected under international copyright law.  All rights
-reserved and protected by the copyright holders.
-This file is confidential and only available to authorized individuals with the
-permission of the copyright holders.
-
-Permission is hereby granted, to  {User}. for testing and development purposes.
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+License
+
+Copyright (2024-2025) AI Labs, IronOne Technologies, LLC
+All Rights Reserved
+
+This source code is protected under international copyright law.  All rights
+reserved and protected by the copyright holders.
+This file is confidential and only available to authorized individuals with the
+permission of the copyright holders.
+
+Permission is hereby granted, to  {User}. for testing and development purposes.
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 SOFTWARE.

app.py

@@ -15,20 +15,16 @@
 """
 
 import os
-import time
-import sys
 import logging
-import datetime
 import uvicorn
 from dotenv import load_dotenv
 
-from fastapi import FastAPI, APIRouter, HTTPException, status
-from fastapi import HTTPException, status
+from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
 
-from schema import UserQuery, ResponseModel, Document, LoginRequest, UserModel
-from controller import get_QA_Answers, get_avaliable_models
+from reggpt.configs.api import API_TITLE, API_VERSION, API_DESCRIPTION 
 
+from reggpt.api.router import ChatAPI
 
 def filer():
     return "logs/log"
@@ -55,110 +51,11 @@ load_dotenv()
 host = os.environ.get("APP_HOST")
 port = int(os.environ.get("APP_PORT"))
 
-
-class ChatAPI:
-
-    def __init__(self):
-        self.router = APIRouter()
-        self.router.add_api_route("/api/v1/health", self.hello, methods=["GET"])
-        self.router.add_api_route("/api/v1/models", self.avaliable_models, methods=["GET"])
-        self.router.add_api_route(
-            "/api/v1/login", self.login, methods=["POST"], response_model=UserModel
-        )
-        self.router.add_api_route("/api/v1/chat", self.chat, methods=["POST"])
-
-    async def hello(self):
-        return "Hello there!"
-
-    async def avaliable_models(self):
-        logger.info("getting avaliable models")
-        models = get_avaliable_models()
-
-        if not models:
-            logger.exception("models not found")
-            raise HTTPException(
-                status_code=status.HTTP_404_NOT_FOUND, detail="models not found"
-            )
-
-        return models
-
-    async def login(self, login_request: LoginRequest):
-        logger.info(f"username password: {login_request} ")
-        # Dummy user data for demonstration (normally, you'd use a database)
-        dummy_users_db = {
-            "john_doe": {
-                "userId": 1,
-                "firstName": "John",
-                "lastName": "Doe",
-                "userName": "john_doe",
-                "password": "password",  # Normally, passwords would be hashed and stored securely
-                "token": "dummy_token_123",  # In a real scenario, this would be a JWT or another kind of token
-            }
-        }
-        # Fetch user by username
-        # user = dummy_users_db.get(login_request.username)
-        user = dummy_users_db.get("john_doe")
-        # Validate user credentials
-        if not user or user["password"] != login_request.password:
-            raise HTTPException(status_code=401, detail="Invalid username or password")
-
-        # Return the user model without the password
-        return UserModel(
-            userId=user["userId"],
-            firstName=user["firstName"],
-            lastName=user["lastName"],
-            userName=user["userName"],
-            token=user["token"],
-        )
-
-    async def chat(
-        self, userQuery: UserQuery
-    ):  #:UserQuery):# -> ResponseModel: #chat: QueryModel): # -> ResponseModel:
-        """Makes query to doc store via Langchain pipeline.
-
-        :param chat.: question, model, dataset location, history of the chat.
-        :type chat: QueryModel
-        """
-        logger.info(f"userQuery: {userQuery} ")
-
-        try:
-            start = time.time()
-            res = get_QA_Answers(userQuery)
-            logger.info(
-                f"--------------------------  answer: {res} -------------------------- "
-            )
-            # return res
-            end = time.time()
-            logger.info(
-                f"-------------------------- Server process (took {round(end - start, 2)} s.) \n: {res}"
-            )
-            print(
-                f" \n -------------------------- Server process (took {round(end - start, 2)} s.) -------------------------  \n"
-            )
-            # return res
-            return {
-                "user_input": userQuery.user_question,
-                "bot_response": res["bot_response"],
-                "format_data": res["format_data"],
-            }
-        
-
-        except HTTPException as e:
-            logger.exception(e)
-            raise e
-
-        except Exception as e:
-            logger.exception(e)
-            raise HTTPException(status_code=400, detail=f"Error : {e}")
-
-
 # initialize API
-app = FastAPI(title="Boardpac chatbot API")
+app = FastAPI(title=API_TITLE, version=API_VERSION, description=API_DESCRIPTION)
 api = ChatAPI()
 app.include_router(api.router)
 
-# origins = ['http://localhost:8000','http://192.168.10.100:8000']
-
 app.add_middleware(
     CORSMiddleware,
     allow_origins=["*"],  # origins,
@@ -169,9 +66,6 @@ app.add_middleware(
 
 if __name__ == "__main__":
 
-    host = "0.0.0.0"
-    port = 8000
-
     # config = uvicorn.Config("server:app",host=host, port=port, log_config= logging.basicConfig())
     config = uvicorn.Config("server:app", host=host, port=port)
     server = uvicorn.Server(config)

faiss_embeddings_2024/index.faiss

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:a3087f02172887cbf8bfb0fb3b371843548619c2a2873fdf4629339e2031a2c1
-size 10895405

faiss_embeddings_2024/index.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3da917c3758e2bfe0aedbd050199f4c80ec372d5b0349b49126b790fb1757db9
-size 3935715

prompts.py

@@ -1,123 +0,0 @@
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2024-2025) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 14/11/2023
- *  LastModifiedDate : 19/03/2024
- *************************************************************************/
-"""
-
-from langchain.prompts import PromptTemplate
-
-# multi query prompt
-MULTY_QUERY_PROMPT = PromptTemplate(
-    input_variables=["question"],
-    template="""You are an AI language model assistant. Your task is to generate three 
-    different versions of the given user question to retrieve relevant documents from a vector 
-    database. By generating multiple perspectives on the user question, your goal is to help
-    the user overcome some of the limitations of the distance-based similarity search. 
-    Provide these alternative questions separated by newlines.
-
-    Dont add anything extra before or after to the 3 questions. Just give 3 lines with 3 questions.
-    Just provide 3 lines having 3 questions only.
-    Answer should be in following format.
-
-    1. alternative question 1
-    2. alternative question 2
-    3. alternative question 3
-
-    Original question: {question}""",
-)
-
-#retrieval prompt
-B_INST, E_INST = "[INST]", "[/INST]"
-B_SYS, E_SYS = "<<SYS>>\n", "\n<</SYS>>\n\n"
-
-retrieval_qa_template = (
-"""<<SYS>>
-You are the AI assistant of company 'boardpac' which provide services to company board members related to banking and financial sector.
-
-please answer the question based on the chat history provided below.  Answer should be short and simple as possible and on to the point.
-<chat history>: {chat_history}
-
-If the  question is related to welcomes and greetings answer accordingly.
-
-Else If the question is related to Banking and Financial Services Sector like Banking & Financial regulations, legal framework, governance framework, compliance requirements as per Central Bank regulations.
-please answer the question based only on the information provided in following central bank documents published in various years.
-The published year is mentioned as the  metadata 'year' of each source document.
-Please notice that content of a one document of a past year can updated by a new document from a recent year.
-Always try to answer with latest information and mention the year which information extracted.
-If you dont know the answer say you dont know, dont try to makeup answers. Dont add any extra details that is not mentioned in the context.
-
-<</SYS>>
-
-[INST]
-<DOCUMENTS>
-{context}
-</DOCUMENTS>
-
-Question : {question}[/INST]"""
-)
-
-
-retrieval_qa_chain_prompt = PromptTemplate(
-    input_variables=["question", "context", "chat_history"], 
-    template=retrieval_qa_template
-)
-
-
-
-#document combine prompt
-document_combine_prompt = PromptTemplate(
-    input_variables=["source","year", "page","page_content"],
-    template= 
-    """<doc> source: {source}, year: {year}, page: {page}, page content: {page_content} </doc>"""
-)
-
-
-router_template_Mixtral_V0= """
-You are the AI assistant of company 'boardpac' which provide services to company board members related to banking and financial sector.
-
-If a user asks a question you have to classify it to following 3 types Relevant, Greeting, Other.
- 
-"Relevant”: If the question is related to Banking and Financial Services Sector like Banking & Financial regulations, legal framework, governance framework, compliance requirements as per Central Bank regulations.
-"Greeting”: If the question is a greeting like good morning, hi my name is., thank you or General Question ask about the AI assistance of a company boardpac.
-"Other”: If the question is not related to research papers.
- 
-Give the correct name of question type. If you are not sure return "Not Sure" instead.
- 
-Question : {question}
-"""
-router_prompt=PromptTemplate.from_template(router_template_Mixtral_V0)
-
-
-general_qa_template_Mixtral_V0= """
-You are the AI assistant of company 'boardpac' which provide services to company board members related to banking and financial sector.
-you can answer Banking and Financial Services Sector like Banking & Financial regulations, legal framework, governance framework, compliance requirements as per Central Bank regulations related question .
-
-Is the provided question below a greeting? First, evaluate whether the input resembles a typical greeting or not.
-
-Greetings are used to say 'hello' and 'how are you?' and to say 'goodbye' and 'nice speaking with you.' and 'hi, I'm (user's name).'
-Greetings are words used when we want to introduce ourselves to others and when we want to find out how someone is feeling.
-
-You can only reply to the user's greetings. 
-If the question is a greeting, reply accordingly as the AI assistant of company boardpac. 
-If the question is not related to greetings and research papers, say that it is out of your domain.
-If the question is not clear enough, ask for more details and don't try to make up answers.
-
-Answer should be polite, short, and simple.
-
-Additionally, it's important to note that this AI assistant has access to an internal collection of research papers, and answers can be provided using the information available in those CBSL Dataset.
-
-Question: {question}
-"""
-
-general_qa_chain_prompt = PromptTemplate.from_template(general_qa_template_Mixtral_V0)
-

reggpt/agent/agent.py

@@ -0,0 +1,33 @@
+import logging
+logger = logging.getLogger(__name__)
+from fastapi import HTTPException
+import time
+from reggpt.routers.qaPipeline import run_router_chain, chain_selector
+
+def run_agent(query):
+    try:
+        logger.info(f"run_agent : Question: {query}")
+        print(f"---------------- run_agent : Question: {query} ----------------")
+        # Get the answer from the chain
+        start = time.time()
+        chain_type = run_router_chain(query)
+        res = chain_selector(chain_type,query)
+        end = time.time()
+
+        # log the result
+        logger.error(f"---------------- Answer (took {round(end - start, 2)} s.) \n: {res}")
+        print(f" \n ---------------- Answer (took {round(end - start, 2)} s.): --------------  \n")
+
+        return res
+    
+    except HTTPException as e:
+            print('HTTPException')
+            print(e)
+            logger.exception(e)
+            raise e
+    
+    except Exception as e:
+        print('Exception')
+        print(e)
+        logger.exception(e)
+        raise e

server.py → reggpt/api/router.py

@@ -1,71 +1,28 @@
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2023-2025) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 14/11/2023
- *  LastModifiedDate : 15/10/2024
- *************************************************************************/
-"""
-
-import os
 import time
-import sys
-import logging
-import datetime
-import uvicorn
-from dotenv import load_dotenv
-
-from fastapi import FastAPI, APIRouter, HTTPException, status
-from fastapi import HTTPException, status
-from fastapi.middleware.cors import CORSMiddleware
-
-from schema import UserQuery, ResponseModel, Document, LoginRequest, UserModel
-from controller import get_QA_Answers, get_avaliable_models
 
 
-def filer():
-    return "logs/log"
-    # today = datetime.datetime.today()
-    # log_filename = f"logs/{today.year}-{today.month:02d}-{today.day:02d}.log"
-    # return log_filename
-
+from fastapi import  APIRouter, HTTPException, status
+from fastapi import HTTPException, status
 
-file_handler = logging.FileHandler(filer())
-# file_handler = logging.handlers.TimedRotatingFileHandler(filer(),when="D")
-file_handler.setLevel(logging.INFO)
+from reggpt.schemas.schema import UserQuery, LoginRequest, UserModel
+from reggpt.routers.controller import get_QA_Answers, get_avaliable_models
 
-logging.basicConfig(
-    level=logging.DEBUG,
-    format="%(asctime)s %(levelname)s (%(name)s) : %(message)s",
-    datefmt="%Y-%m-%d %H:%M:%S",
-    handlers=[file_handler],
-    force=True,
-)
+from reggpt.configs.api import API_ENDPOINT_LOGIN,API_ENDPOINT_CHAT, API_ENDPOINT_HEALTH, API_ENDPOINT_MODEL
+import logging
 
 logger = logging.getLogger(__name__)
 
-load_dotenv()
-host = os.environ.get("APP_HOST")
-port = int(os.environ.get("APP_PORT"))
-
 
 class ChatAPI:
 
     def __init__(self):
         self.router = APIRouter()
-        self.router.add_api_route("/api/v1/health", self.hello, methods=["GET"])
-        self.router.add_api_route("/api/v1/models", self.avaliable_models, methods=["GET"])
+        self.router.add_api_route(API_ENDPOINT_HEALTH, self.hello, methods=["GET"])
+        self.router.add_api_route(API_ENDPOINT_MODEL, self.avaliable_models, methods=["GET"])
         self.router.add_api_route(
-            "/api/v1/login", self.login, methods=["POST"], response_model=UserModel
+            API_ENDPOINT_LOGIN, self.login, methods=["POST"], response_model=UserModel
         )
-        self.router.add_api_route("/api/v1/chat", self.chat, methods=["POST"])
+        self.router.add_api_route(API_ENDPOINT_CHAT, self.chat, methods=["POST"])
 
     async def hello(self):
         return "Hello there!"
@@ -151,29 +108,3 @@ class ChatAPI:
             logger.exception(e)
             raise HTTPException(status_code=400, detail=f"Error : {e}")
 
-
-# initialize API
-app = FastAPI(title="Boardpac chatbot API")
-api = ChatAPI()
-app.include_router(api.router)
-
-# origins = ['http://localhost:8000','http://192.168.10.100:8000']
-
-app.add_middleware(
-    CORSMiddleware,
-    allow_origins=["*"],  # origins,
-    allow_credentials=True,
-    allow_methods=["*"],
-    allow_headers=["*"],
-)
-
-if __name__ == "__main__":
-
-    host = "0.0.0.0"
-    port = 8000
-
-    # config = uvicorn.Config("server:app",host=host, port=port, log_config= logging.basicConfig())
-    config = uvicorn.Config("server:app", host=host, port=port)
-    server = uvicorn.Server(config)
-    server.run()
-    # uvicorn.run(app)

{configs → reggpt/chains}/__init__.py

@@ -1,5 +1,5 @@
-# import os
-# import sys
-
-# if os.path.dirname(os.path.abspath(__file__)) not in sys.path:
-#     sys.path.append(os.path.dirname(os.path.abspath(__file__)))
+# import os
+# import sys
+
+# if os.path.dirname(os.path.abspath(__file__)) not in sys.path:
+#     sys.path.append(os.path.dirname(os.path.abspath(__file__)))

llmChain.py → reggpt/chains/llmChain.py

@@ -1,96 +1,99 @@
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 14/11/2023
- *  LastModifiedDate : 18/03/2024
- *************************************************************************/
- """
-
-import os
-import logging
-logger = logging.getLogger(__name__)
-from dotenv import load_dotenv
-
-load_dotenv()
-
-verbose = os.environ.get('VERBOSE')
-
-from llm import get_model
-from langchain.chains import  ConversationalRetrievalChain
-# from conversationBufferWindowMemory import ConversationBufferWindowMemory
-
-# from langchain.prompts import PromptTemplate
-from langchain.chains import LLMChain
-
-from prompts import retrieval_qa_chain_prompt, document_combine_prompt, general_qa_chain_prompt, router_prompt
-
-def get_qa_chain(model_type,retriever):
-    logger.info("creating qa_chain")
-    
-    try:
-        qa_llm = get_model(model_type)
-
-        qa_chain = ConversationalRetrievalChain.from_llm(
-            llm=qa_llm,
-            chain_type="stuff",
-            retriever = retriever, 
-            # retriever = self.retriever(search_kwargs={"k": target_source_chunks}
-            return_source_documents= True,
-            get_chat_history=lambda h : h,
-            combine_docs_chain_kwargs={
-                "prompt": retrieval_qa_chain_prompt,
-                "document_prompt": document_combine_prompt,
-            },
-            verbose=True,
-            # memory=memory,
-        )
-
-        logger.info("qa_chain created")
-        return qa_chain
-
-    except Exception as e:
-        msg=f"Error : {e}"
-        logger.exception(msg)
-        raise e
-
-
-def get_general_qa_chain(model_type):
-    logger.info("creating general_qa_chain")
-    
-    try:
-        general_qa_llm = get_model(model_type)
-        general_qa_chain = LLMChain(llm=general_qa_llm, prompt=general_qa_chain_prompt)
-
-        logger.info("general_qa_chain created")
-        return general_qa_chain
-
-    except Exception as e:
-        msg=f"Error : {e}"
-        logger.exception(msg)
-        raise e
-
-
-def get_router_chain(model_type):
-    logger.info("creating router_chain")
-    
-    try:
-        router_llm = get_model(model_type)
-        router_chain = LLMChain(llm=router_llm, prompt=router_prompt)
-
-        logger.info("router_chain created")
-        return router_chain
-
-    except Exception as e:
-        msg=f"Error : {e}"
-        logger.exception(msg)
-        raise e
-
-
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 14/11/2023
+ *  LastModifiedDate : 18/03/2024
+ *************************************************************************/
+ """
+
+import os
+import logging
+logger = logging.getLogger(__name__)
+from dotenv import load_dotenv
+
+load_dotenv()
+
+verbose = os.environ.get('VERBOSE')
+
+from reggpt.llms.llm import get_model
+from langchain.chains import  ConversationalRetrievalChain
+# from conversationBufferWindowMemory import ConversationBufferWindowMemory
+
+# from langchain.prompts import PromptTemplate
+from langchain.chains import LLMChain
+from reggpt.prompts import document_combine as document_combine_prompt
+from reggpt.prompts import retrieval as retrieval_qa_chain_prompt
+from reggpt.prompts import general as general_qa_chain_prompt
+from reggpt.prompts import router as router_prompt
+
+
+def get_qa_chain(model_type,retriever):
+    logger.info("creating qa_chain")
+    
+    try:
+        qa_llm = get_model(model_type)
+
+        qa_chain = ConversationalRetrievalChain.from_llm(
+            llm=qa_llm,
+            chain_type="stuff",
+            retriever = retriever, 
+            # retriever = self.retriever(search_kwargs={"k": target_source_chunks}
+            return_source_documents= True,
+            get_chat_history=lambda h : h,
+            combine_docs_chain_kwargs={
+                "prompt": retrieval_qa_chain_prompt,
+                "document_prompt": document_combine_prompt,
+            },
+            verbose=True,
+            # memory=memory,
+        )
+
+        logger.info("qa_chain created")
+        return qa_chain
+
+    except Exception as e:
+        msg=f"Error : {e}"
+        logger.exception(msg)
+        raise e
+
+
+def get_general_qa_chain(model_type):
+    logger.info("creating general_qa_chain")
+    
+    try:
+        general_qa_llm = get_model(model_type)
+        general_qa_chain = LLMChain(llm=general_qa_llm, prompt=general_qa_chain_prompt)
+
+        logger.info("general_qa_chain created")
+        return general_qa_chain
+
+    except Exception as e:
+        msg=f"Error : {e}"
+        logger.exception(msg)
+        raise e
+
+
+def get_router_chain(model_type):
+    logger.info("creating router_chain")
+    
+    try:
+        router_llm = get_model(model_type)
+        router_chain = LLMChain(llm=router_llm, prompt=router_prompt)
+
+        logger.info("router_chain created")
+        return router_chain
+
+    except Exception as e:
+        msg=f"Error : {e}"
+        logger.exception(msg)
+        raise e
+
+

{data → reggpt/configs}/__init__.py

@@ -1,5 +1,5 @@
-# import os
-# import sys
-
-# if os.path.dirname(os.path.abspath(__file__)) not in sys.path:
-#     sys.path.append(os.path.dirname(os.path.abspath(__file__)))
+# import os
+# import sys
+
+# if os.path.dirname(os.path.abspath(__file__)) not in sys.path:
+#     sys.path.append(os.path.dirname(os.path.abspath(__file__)))

reggpt/configs/api.py

@@ -0,0 +1,28 @@
+"""
+ /*************************************************************************
+ *
+ * CONFIDENTIAL
+ * __________________
+ *
+ *  Copyright (2024-2025) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Authors  : Hammaad Rizwan, Theekshana Samaradiwakara
+ *  Description : API Configurations
+ *  CreatedDate : 14/10/2024
+ *  LastModifiedDate : 15/10/2024
+ *************************************************************************/
+"""
+API_TITLE = "RegGPT Back End v1"
+API_VERSION = "0.1.0"
+API_DESCRIPTION = "API_DESC"
+
+API_ENDPOINT_PREFIX = "/api/v1"
+API_DOCS_URL = f"{API_ENDPOINT_PREFIX}/docs"
+API_REDOC_URL = f"{API_ENDPOINT_PREFIX}/redoc"
+API_OPENAPI_URL = f"{API_ENDPOINT_PREFIX}/openapi.json"
+
+API_ENDPOINT_HEALTH = f"{API_ENDPOINT_PREFIX}/health"
+API_ENDPOINT_CHAT = f"{API_ENDPOINT_PREFIX}/chat"
+API_ENDPOINT_MODEL = f"{API_ENDPOINT_PREFIX}/models"
+API_ENDPOINT_LOGIN = f"{API_ENDPOINT_PREFIX}/login"

config.py → reggpt/configs/config.py

@@ -1,36 +1,36 @@
-AVALIABLE_MODELS=[
-    {
-    "id":"gpt-4o-mini",
-    "model_name":"openai/gpt-4o-mini",
-    "description":"gpt-4o-mini model from openai"
-    }
-]
-
-MODELS={
-    "DEFAULT":"openai",
-    "gpt-4o-mini":"openai",
-    
-}
-
-DATASETS={
-    "DEFAULT":"faiss",
-    "a":"A",
-    "b":"B",
-    "c":"C"
-
-}
-
-MEMORY_WINDOW_K = 1
-
-QA_MODEL_TYPE = "openai"
-GENERAL_QA_MODEL_TYPE = "openai"
-ROUTER_MODEL_TYPE = "openai"
-Multi_Query_MODEL_TYPE = "openai"
-
-
-ANSWER_TYPES = [
-    "relevant",
-    "greeting",
-    "other",
-    "not sure",
+AVALIABLE_MODELS=[
+    {
+    "id":"gpt-4o-mini",
+    "model_name":"openai/gpt-4o-mini",
+    "description":"gpt-4o-mini model from openai"
+    }
+]
+
+MODELS={
+    "DEFAULT":"openai",
+    "gpt-4o-mini":"openai",
+    
+}
+
+DATASETS={
+    "DEFAULT":"faiss",
+    "a":"A",
+    "b":"B",
+    "c":"C"
+
+}
+
+MEMORY_WINDOW_K = 1
+
+QA_MODEL_TYPE = "openai"
+GENERAL_QA_MODEL_TYPE = "openai"
+ROUTER_MODEL_TYPE = "openai"
+Multi_Query_MODEL_TYPE = "openai"
+
+
+ANSWER_TYPES = [
+    "relevant",
+    "greeting",
+    "other",
+    "not sure",
 ]

{configs → reggpt/configs}/logger.py

@@ -1,40 +1,40 @@
-import logging
-import time
-# from functools import wraps
-
-logger = logging.getLogger(__name__)
-
-stream_handler = logging.StreamHandler()
-log_filename = "output.log"
-file_handler = logging.FileHandler(filename=log_filename)
-handlers = [stream_handler, file_handler]
-
-
-class TimeFilter(logging.Filter):
-    def filter(self, record):
-        return "Running" in record.getMessage()
-
-
-logger.addFilter(TimeFilter())
-
-# Configure the logging module
-logging.basicConfig(
-    level=logging.INFO,
-    format="%(name)s %(asctime)s - %(levelname)s - %(message)s",
-    handlers=handlers,
-)
-
-
-def time_logger(func):
-    """Decorator function to log time taken by any function."""
-
-    # @wraps(func)
-    def wrapper(*args, **kwargs):
-        start_time = time.time()  # Start time before function execution
-        result = func(*args, **kwargs)  # Function execution
-        end_time = time.time()  # End time after function execution
-        execution_time = end_time - start_time  # Calculate execution time
-        logger.info(f"Running {func.__name__}: --- {execution_time} seconds ---")
-        return result
-
+import logging
+import time
+# from functools import wraps
+
+logger = logging.getLogger(__name__)
+
+stream_handler = logging.StreamHandler()
+log_filename = "output.log"
+file_handler = logging.FileHandler(filename=log_filename)
+handlers = [stream_handler, file_handler]
+
+
+class TimeFilter(logging.Filter):
+    def filter(self, record):
+        return "Running" in record.getMessage()
+
+
+logger.addFilter(TimeFilter())
+
+# Configure the logging module
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(name)s %(asctime)s - %(levelname)s - %(message)s",
+    handlers=handlers,
+)
+
+
+def time_logger(func):
+    """Decorator function to log time taken by any function."""
+
+    # @wraps(func)
+    def wrapper(*args, **kwargs):
+        start_time = time.time()  # Start time before function execution
+        result = func(*args, **kwargs)  # Function execution
+        end_time = time.time()  # End time after function execution
+        execution_time = end_time - start_time  # Calculate execution time
+        logger.info(f"Running {func.__name__}: --- {execution_time} seconds ---")
+        return result
+
     return wrapper

qaPipeline.py → reggpt/controller/agent.py

@@ -1,150 +1,73 @@
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 14/11/2023
- *  LastModifiedDate : 18/03/2024
- *************************************************************************/
- """
-
-import os
-import time
-import logging
-logger = logging.getLogger(__name__)
-from dotenv import load_dotenv
-from fastapi import HTTPException
-from llmChain import get_qa_chain, get_general_qa_chain, get_router_chain
-from output_parser import general_qa_chain_output_parser, qa_chain_output_parser, out_of_domain_chain_parser
-
-from config import QA_MODEL_TYPE, GENERAL_QA_MODEL_TYPE, ROUTER_MODEL_TYPE, Multi_Query_MODEL_TYPE
-from retriever import load_faiss_retriever, load_ensemble_retriever, load_multi_query_retriever
-load_dotenv()
-
-verbose = os.environ.get('VERBOSE')
-
-qa_model_type=QA_MODEL_TYPE
-general_qa_model_type=GENERAL_QA_MODEL_TYPE
-router_model_type=ROUTER_MODEL_TYPE #"google/flan-t5-xxl"
-multi_query_model_type=Multi_Query_MODEL_TYPE #"google/flan-t5-xxl"
-# model_type="tiiuae/falcon-7b-instruct"
-
-# retriever=load_faiss_retriever()
-retriever=load_ensemble_retriever()
-# retriever=load_multi_query_retriever(multi_query_model_type)
-logger.info("retriever loaded:")
-
-qa_chain= get_qa_chain(qa_model_type,retriever)
-general_qa_chain= get_general_qa_chain(general_qa_model_type)
-router_chain= get_router_chain(router_model_type)
-
-def chain_selector(chain_type, query):
-    chain_type = chain_type.lower().strip()
-    logger.info(f"chain_selector : chain_type: {chain_type} Question: {query}")
-    if "greeting" in chain_type:
-        return run_general_qa_chain(query)
-    elif "other" in chain_type:
-        return run_out_of_domain_chain(query)
-    elif ("relevant" in chain_type) or ("not sure" in chain_type) :
-        return run_qa_chain(query)
-    else:
-        raise ValueError(
-            f"Received invalid type '{chain_type}'"
-        )
-    
-def run_agent(query):
-    try:
-        logger.info(f"run_agent : Question: {query}")
-        print(f"---------------- run_agent : Question: {query} ----------------")
-        # Get the answer from the chain
-        start = time.time()
-        chain_type = run_router_chain(query)
-        res = chain_selector(chain_type,query)
-        end = time.time()
-
-        # log the result
-        logger.error(f"---------------- Answer (took {round(end - start, 2)} s.) \n: {res}")
-        print(f" \n ---------------- Answer (took {round(end - start, 2)} s.): --------------  \n")
-
-        return res
-    
-    except HTTPException as e:
-            print('HTTPException eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee')
-            print(e)
-            logger.exception(e)
-            raise e
-    
-    except Exception as e:
-        print('Exception eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee')
-        print(e)
-        logger.exception(e)
-        raise e
-
-
-def run_router_chain(query):
-    try:
-        logger.info(f"run_router_chain : Question: {query}")
-        # Get the answer from the chain
-        start = time.time()
-        chain_type = router_chain.invoke(query)['text']
-        end = time.time()
-
-        # log the result
-        logger.info(f"Answer (took {round(end - start, 2)} s.) chain_type: {chain_type}")
-
-        return chain_type
-    
-    except Exception as e:
-        logger.exception(e)
-        raise e
-
-
-def run_qa_chain(query):
-    try:
-        logger.info(f"run_qa_chain : Question: {query}")
-        # Get the answer from the chain
-        start = time.time()
-        # res = qa_chain(query)
-        res = qa_chain.invoke({"question": query, "chat_history":""})
-        # res = response
-        # answer, docs = res['result'],res['source_documents']
-        end = time.time()
-
-        # log the result
-        logger.info(f"Answer (took {round(end - start, 2)} s.) \n: {res}")
-
-        return qa_chain_output_parser(res)
-    
-    except Exception as e:
-        logger.exception(e)
-        raise e
-
-
-def run_general_qa_chain(query):
-    try:
-        logger.info(f"run_general_qa_chain : Question: {query}")
-
-        # Get the answer from the chain
-        start = time.time()
-        res = general_qa_chain.invoke(query)
-        end = time.time()
-
-        # log the result
-       
-        logger.info(f"Answer (took {round(end - start, 2)} s.) \n: {res}")
-
-        return general_qa_chain_output_parser(res)
-
-    except Exception as e:
-        logger.exception(e)
-        raise e
-
-
-def run_out_of_domain_chain(query):
-    return out_of_domain_chain_parser(query)
+import os
+import time
+import logging
+logger = logging.getLogger(__name__)
+from dotenv import load_dotenv
+from fastapi import HTTPException
+from reggpt.chains.llmChain import get_qa_chain, get_general_qa_chain, get_router_chain
+from reggpt.output_parsers.output_parser import general_qa_chain_output_parser, qa_chain_output_parser, out_of_domain_chain_parser
+
+from reggpt.configs.config import QA_MODEL_TYPE, GENERAL_QA_MODEL_TYPE, ROUTER_MODEL_TYPE, Multi_Query_MODEL_TYPE
+from reggpt.utils.retriever import load_faiss_retriever, load_ensemble_retriever, load_multi_query_retriever
+load_dotenv()
+
+verbose = os.environ.get('VERBOSE')
+
+qa_model_type=QA_MODEL_TYPE
+general_qa_model_type=GENERAL_QA_MODEL_TYPE
+router_model_type=ROUTER_MODEL_TYPE #"google/flan-t5-xxl"
+multi_query_model_type=Multi_Query_MODEL_TYPE #"google/flan-t5-xxl"
+# model_type="tiiuae/falcon-7b-instruct"
+
+# retriever=load_faiss_retriever()
+retriever=load_ensemble_retriever()
+# retriever=load_multi_query_retriever(multi_query_model_type)
+logger.info("retriever loaded:")
+
+qa_chain= get_qa_chain(qa_model_type,retriever)
+general_qa_chain= get_general_qa_chain(general_qa_model_type)
+router_chain= get_router_chain(router_model_type)
+
+
+def chain_selector(chain_type, query):
+    chain_type = chain_type.lower().strip()
+    logger.info(f"chain_selector : chain_type: {chain_type} Question: {query}")
+    if "greeting" in chain_type:
+        return run_general_qa_chain(query)
+    elif "other" in chain_type:
+        return run_out_of_domain_chain(query)
+    elif ("relevant" in chain_type) or ("not sure" in chain_type) :
+        return run_qa_chain(query)
+    else:
+        raise ValueError(
+            f"Received invalid type '{chain_type}'"
+        )
+    
+def run_agent(query):
+    try:
+        logger.info(f"run_agent : Question: {query}")
+        print(f"---------------- run_agent : Question: {query} ----------------")
+        # Get the answer from the chain
+        start = time.time()
+        chain_type = run_router_chain(query)
+        res = chain_selector(chain_type,query)
+        end = time.time()
+
+        # log the result
+        logger.error(f"---------------- Answer (took {round(end - start, 2)} s.) \n: {res}")
+        print(f" \n ---------------- Answer (took {round(end - start, 2)} s.): --------------  \n")
+
+        return res
+    
+    except HTTPException as e:
+            print('HTTPException')
+            print(e)
+            logger.exception(e)
+            raise e
+    
+    except Exception as e:
+        print('Exception')
+        print(e)
+        logger.exception(e)
+        raise e
+

reggpt/controller/router.py

@@ -0,0 +1,62 @@
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 14/11/2023
+ *  LastModifiedDate : 18/03/2024
+ *************************************************************************/
+ """
+
+import os
+import time
+import logging
+logger = logging.getLogger(__name__)
+from dotenv import load_dotenv
+from fastapi import HTTPException
+from reggpt.chains.llmChain import get_qa_chain, get_general_qa_chain, get_router_chain
+from reggpt.output_parsers.output_parser import general_qa_chain_output_parser, qa_chain_output_parser, out_of_domain_chain_parser
+
+from reggpt.configs.config import QA_MODEL_TYPE, GENERAL_QA_MODEL_TYPE, ROUTER_MODEL_TYPE, Multi_Query_MODEL_TYPE
+from reggpt.utils.retriever import load_faiss_retriever, load_ensemble_retriever, load_multi_query_retriever
+load_dotenv()
+
+verbose = os.environ.get('VERBOSE')
+
+qa_model_type=QA_MODEL_TYPE
+general_qa_model_type=GENERAL_QA_MODEL_TYPE
+router_model_type=ROUTER_MODEL_TYPE #"google/flan-t5-xxl"
+multi_query_model_type=Multi_Query_MODEL_TYPE #"google/flan-t5-xxl"
+# model_type="tiiuae/falcon-7b-instruct"
+
+# retriever=load_faiss_retriever()
+retriever=load_ensemble_retriever()
+# retriever=load_multi_query_retriever(multi_query_model_type)
+logger.info("retriever loaded:")
+
+qa_chain= get_qa_chain(qa_model_type,retriever)
+general_qa_chain= get_general_qa_chain(general_qa_model_type)
+router_chain= get_router_chain(router_model_type)
+
+def run_router_chain(query):
+    try:
+        logger.info(f"run_router_chain : Question: {query}")
+        # Get the answer from the chain
+        start = time.time()
+        chain_type = router_chain.invoke(query)['text']
+        end = time.time()
+
+        # log the result
+        logger.info(f"Answer (took {round(end - start, 2)} s.) chain_type: {chain_type}")
+
+        return chain_type
+    
+    except Exception as e:
+        logger.exception(e)
+        raise e

reggpt/data/__init__.py

@@ -0,0 +1,5 @@
+# import os
+# import sys
+
+# if os.path.dirname(os.path.abspath(__file__)) not in sys.path:
+#     sys.path.append(os.path.dirname(os.path.abspath(__file__)))

llm.py → reggpt/llms/llm.py

@@ -1,47 +1,47 @@
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 14/11/2023
- *  LastModifiedDate : 18/03/2024
- *************************************************************************/
- """
-
-import os
-# import time
-import logging
-logger = logging.getLogger(__name__)
-from dotenv import load_dotenv
-
-# from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
-from langchain_openai import ChatOpenAI
-
-load_dotenv()
-
-openai_api_key = os.environ.get('OPENAI_API_KEY')
-# openai_api_key = "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"
-print(f"--- ---- ---- openai_api_key: {openai_api_key}")
-
-verbose = os.environ.get('VERBOSE')
-
-def get_model(model_type):
-   
-    match model_type:
-        case "openai":
-            llm = ChatOpenAI(model_name="gpt-4o-mini", temperature=0, openai_api_key=openai_api_key)
-        case _default:
-            # raise exception if model_type is not supported
-            msg=f"Model type '{model_type}' is not supported. Please choose a valid one"
-            logger.error(msg)
-            return Exception(msg)
-
-
-    logger.info(f"model_type: {model_type} loaded:")
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 14/11/2023
+ *  LastModifiedDate : 18/03/2024
+ *************************************************************************/
+ """
+
+import os
+# import time
+import logging
+logger = logging.getLogger(__name__)
+from dotenv import load_dotenv
+
+# from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
+from langchain_openai import ChatOpenAI
+
+load_dotenv()
+
+openai_api_key = os.environ.get('OPENAI_API_KEY')
+# openai_api_key = "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"
+print(f"--- ---- ---- openai_api_key: {openai_api_key}")
+
+verbose = os.environ.get('VERBOSE')
+
+def get_model(model_type):
+   
+    match model_type:
+        case "openai":
+            llm = ChatOpenAI(model_name="gpt-4o-mini", temperature=0, openai_api_key=openai_api_key)
+        case _default:
+            # raise exception if model_type is not supported
+            msg=f"Model type '{model_type}' is not supported. Please choose a valid one"
+            logger.error(msg)
+            return Exception(msg)
+
+
+    logger.info(f"model_type: {model_type} loaded:")
     return llm

conversationBufferWindowMemory.py → reggpt/memory/conversationBufferWindowMemory.py

@@ -1,134 +1,134 @@
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 14/11/2023
- *  LastModifiedDate : 18/11/2020
- *************************************************************************/
- """
-
-from abc import ABC
-from typing import Any, Dict, Optional, Tuple
-# import json
-
-from langchain.memory.chat_message_histories.in_memory import ChatMessageHistory
-from langchain.memory.utils import get_prompt_input_key
-from langchain.pydantic_v1 import Field
-from langchain.schema import BaseChatMessageHistory, BaseMemory
-
-from typing import List, Union
-
-# from langchain.memory.chat_memory import BaseChatMemory
-from langchain.schema.messages import BaseMessage, get_buffer_string
-
-
-class BaseChatMemory(BaseMemory, ABC):
-    """Abstract base class for chat memory."""
-
-    chat_memory: BaseChatMessageHistory = Field(default_factory=ChatMessageHistory)
-    output_key: Optional[str] = None
-    input_key: Optional[str] = None
-    return_messages: bool = False
-
-    def _get_input_output(
-        self, inputs: Dict[str, Any], outputs: Dict[str, str]
-    ) -> Tuple[str, str]:
-        
-
-        if self.input_key is None:
-            prompt_input_key = get_prompt_input_key(inputs, self.memory_variables)
-        else:
-            prompt_input_key = self.input_key
-
-        if self.output_key is None:
-            """
-            output for agent with LLM chain tool                     = {answer} 
-            output for agent with ConversationalRetrievalChain tool  = {'question', 'chat_history', 'answer','source_documents'}
-            """
-        
-            LLM_key = 'output'
-            Retrieval_key = 'answer'
-            if isinstance(outputs[LLM_key], dict):
-                Retrieval_dict = outputs[LLM_key]
-                if Retrieval_key in Retrieval_dict.keys():
-                    #output keys are 'answer' , 'source_documents'
-                    output = Retrieval_dict[Retrieval_key]
-                else:
-                    raise ValueError(f"output key: {LLM_key} not a valid dictionary")
-
-            else:
-                #otherwise output key will be 'output'
-                output_key = list(outputs.keys())[0]
-                output = outputs[output_key]
-
-            # if len(outputs) != 1:
-            #     raise ValueError(f"One output key expected, got {outputs.keys()}")
-                
-            
-        else:
-            output_key = self.output_key
-            output = outputs[output_key]
-
-        return inputs[prompt_input_key], output
-
-    def save_context(self, inputs: Dict[str, Any], outputs: Dict[str, str]) -> None:
-        """Save context from this conversation to buffer."""
-        input_str, output_str = self._get_input_output(inputs, outputs)
-        self.chat_memory.add_user_message(input_str)
-        self.chat_memory.add_ai_message(output_str)
-
-    def clear(self) -> None:
-        """Clear memory contents."""
-        self.chat_memory.clear()
-
-
-
-
-
-class ConversationBufferWindowMemory(BaseChatMemory):
-    """Buffer for storing conversation memory inside a limited size window."""
-
-    human_prefix: str = "Human"
-    ai_prefix: str = "AI"
-    memory_key: str = "history"  #: :meta private:
-    k: int = 5
-    """Number of messages to store in buffer."""
-
-    @property
-    def buffer(self) -> Union[str, List[BaseMessage]]:
-        """String buffer of memory."""
-        return self.buffer_as_messages if self.return_messages else self.buffer_as_str
-
-    @property
-    def buffer_as_str(self) -> str:
-        """Exposes the buffer as a string in case return_messages is True."""
-        messages = self.chat_memory.messages[-self.k * 2 :] if self.k > 0 else []
-        return get_buffer_string(
-            messages,
-            human_prefix=self.human_prefix,
-            ai_prefix=self.ai_prefix,
-        )
-
-    @property
-    def buffer_as_messages(self) -> List[BaseMessage]:
-        """Exposes the buffer as a list of messages in case return_messages is False."""
-        return self.chat_memory.messages[-self.k * 2 :] if self.k > 0 else []
-
-    @property
-    def memory_variables(self) -> List[str]:
-        """Will always return list of memory variables.
-
-        :meta private:
-        """
-        return [self.memory_key]
-
-    def load_memory_variables(self, inputs: Dict[str, Any]) -> Dict[str, Any]:
-        """Return history buffer."""
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 14/11/2023
+ *  LastModifiedDate : 18/11/2020
+ *************************************************************************/
+ """
+
+from abc import ABC
+from typing import Any, Dict, Optional, Tuple
+# import json
+
+from langchain.memory.chat_message_histories.in_memory import ChatMessageHistory
+from langchain.memory.utils import get_prompt_input_key
+from langchain.pydantic_v1 import Field
+from langchain.schema import BaseChatMessageHistory, BaseMemory
+
+from typing import List, Union
+
+# from langchain.memory.chat_memory import BaseChatMemory
+from langchain.schema.messages import BaseMessage, get_buffer_string
+
+
+class BaseChatMemory(BaseMemory, ABC):
+    """Abstract base class for chat memory."""
+
+    chat_memory: BaseChatMessageHistory = Field(default_factory=ChatMessageHistory)
+    output_key: Optional[str] = None
+    input_key: Optional[str] = None
+    return_messages: bool = False
+
+    def _get_input_output(
+        self, inputs: Dict[str, Any], outputs: Dict[str, str]
+    ) -> Tuple[str, str]:
+        
+
+        if self.input_key is None:
+            prompt_input_key = get_prompt_input_key(inputs, self.memory_variables)
+        else:
+            prompt_input_key = self.input_key
+
+        if self.output_key is None:
+            """
+            output for agent with LLM chain tool                     = {answer} 
+            output for agent with ConversationalRetrievalChain tool  = {'question', 'chat_history', 'answer','source_documents'}
+            """
+        
+            LLM_key = 'output'
+            Retrieval_key = 'answer'
+            if isinstance(outputs[LLM_key], dict):
+                Retrieval_dict = outputs[LLM_key]
+                if Retrieval_key in Retrieval_dict.keys():
+                    #output keys are 'answer' , 'source_documents'
+                    output = Retrieval_dict[Retrieval_key]
+                else:
+                    raise ValueError(f"output key: {LLM_key} not a valid dictionary")
+
+            else:
+                #otherwise output key will be 'output'
+                output_key = list(outputs.keys())[0]
+                output = outputs[output_key]
+
+            # if len(outputs) != 1:
+            #     raise ValueError(f"One output key expected, got {outputs.keys()}")
+                
+            
+        else:
+            output_key = self.output_key
+            output = outputs[output_key]
+
+        return inputs[prompt_input_key], output
+
+    def save_context(self, inputs: Dict[str, Any], outputs: Dict[str, str]) -> None:
+        """Save context from this conversation to buffer."""
+        input_str, output_str = self._get_input_output(inputs, outputs)
+        self.chat_memory.add_user_message(input_str)
+        self.chat_memory.add_ai_message(output_str)
+
+    def clear(self) -> None:
+        """Clear memory contents."""
+        self.chat_memory.clear()
+
+
+
+
+
+class ConversationBufferWindowMemory(BaseChatMemory):
+    """Buffer for storing conversation memory inside a limited size window."""
+
+    human_prefix: str = "Human"
+    ai_prefix: str = "AI"
+    memory_key: str = "history"  #: :meta private:
+    k: int = 5
+    """Number of messages to store in buffer."""
+
+    @property
+    def buffer(self) -> Union[str, List[BaseMessage]]:
+        """String buffer of memory."""
+        return self.buffer_as_messages if self.return_messages else self.buffer_as_str
+
+    @property
+    def buffer_as_str(self) -> str:
+        """Exposes the buffer as a string in case return_messages is True."""
+        messages = self.chat_memory.messages[-self.k * 2 :] if self.k > 0 else []
+        return get_buffer_string(
+            messages,
+            human_prefix=self.human_prefix,
+            ai_prefix=self.ai_prefix,
+        )
+
+    @property
+    def buffer_as_messages(self) -> List[BaseMessage]:
+        """Exposes the buffer as a list of messages in case return_messages is False."""
+        return self.chat_memory.messages[-self.k * 2 :] if self.k > 0 else []
+
+    @property
+    def memory_variables(self) -> List[str]:
+        """Will always return list of memory variables.
+
+        :meta private:
+        """
+        return [self.memory_key]
+
+    def load_memory_variables(self, inputs: Dict[str, Any]) -> Dict[str, Any]:
+        """Return history buffer."""
         return {self.memory_key: self.buffer}

reggpt/prompts/document_combine.py

@@ -0,0 +1,7 @@
+from langchain.prompts import PromptTemplate
+
+document_combine_prompt = PromptTemplate(
+    input_variables=["source","year", "page","page_content"],
+    template= 
+    """<doc> source: {source}, year: {year}, page: {page}, page content: {page_content} </doc>"""
+)

reggpt/prompts/general.py

@@ -0,0 +1,25 @@
+from langchain.prompts import PromptTemplate
+
+general_qa_template_Mixtral_V0= """
+You are the AI assistant of company 'boardpac' which provide services to company board members related to banking and financial sector.
+you can answer Banking and Financial Services Sector like Banking & Financial regulations, legal framework, governance framework, compliance requirements as per Central Bank regulations related question .
+
+Is the provided question below a greeting? First, evaluate whether the input resembles a typical greeting or not.
+
+Greetings are used to say 'hello' and 'how are you?' and to say 'goodbye' and 'nice speaking with you.' and 'hi, I'm (user's name).'
+Greetings are words used when we want to introduce ourselves to others and when we want to find out how someone is feeling.
+
+You can only reply to the user's greetings. 
+If the question is a greeting, reply accordingly as the AI assistant of company boardpac. 
+If the question is not related to greetings and research papers, say that it is out of your domain.
+If the question is not clear enough, ask for more details and don't try to make up answers.
+
+Answer should be polite, short, and simple.
+
+Additionally, it's important to note that this AI assistant has access to an internal collection of research papers, and answers can be provided using the information available in those CBSL Dataset.
+
+Question: {question}
+"""
+
+general_qa_chain_prompt = PromptTemplate.from_template(general_qa_template_Mixtral_V0)
+

reggpt/prompts/multi_query.py

@@ -0,0 +1,20 @@
+from langchain.prompts import PromptTemplate
+
+MULTY_QUERY_PROMPT = PromptTemplate(
+    input_variables=["question"],
+    template="""You are an AI language model assistant. Your task is to generate three 
+    different versions of the given user question to retrieve relevant documents from a vector 
+    database. By generating multiple perspectives on the user question, your goal is to help
+    the user overcome some of the limitations of the distance-based similarity search. 
+    Provide these alternative questions separated by newlines.
+
+    Dont add anything extra before or after to the 3 questions. Just give 3 lines with 3 questions.
+    Just provide 3 lines having 3 questions only.
+    Answer should be in following format.
+
+    1. alternative question 1
+    2. alternative question 2
+    3. alternative question 3
+
+    Original question: {question}""",
+)

reggpt/prompts/retrieval.py

@@ -0,0 +1,33 @@
+from langchain.prompts import PromptTemplate
+
+retrieval_qa_template = (
+"""<<SYS>>
+You are the AI assistant of company 'boardpac' which provide services to company board members related to banking and financial sector.
+
+please answer the question based on the chat history provided below.  Answer should be short and simple as possible and on to the point.
+<chat history>: {chat_history}
+
+If the  question is related to welcomes and greetings answer accordingly.
+
+Else If the question is related to Banking and Financial Services Sector like Banking & Financial regulations, legal framework, governance framework, compliance requirements as per Central Bank regulations.
+please answer the question based only on the information provided in following central bank documents published in various years.
+The published year is mentioned as the  metadata 'year' of each source document.
+Please notice that content of a one document of a past year can updated by a new document from a recent year.
+Always try to answer with latest information and mention the year which information extracted.
+If you dont know the answer say you dont know, dont try to makeup answers. Dont add any extra details that is not mentioned in the context.
+
+<</SYS>>
+
+[INST]
+<DOCUMENTS>
+{context}
+</DOCUMENTS>
+
+Question : {question}[/INST]"""
+)
+
+
+retrieval_qa_chain_prompt = PromptTemplate(
+    input_variables=["question", "context", "chat_history"], 
+    template=retrieval_qa_template
+)

reggpt/prompts/router.py

@@ -0,0 +1,17 @@
+from langchain.prompts import PromptTemplate
+
+router_template_Mixtral_V0= """
+You are the AI assistant of company 'boardpac' which provide services to company board members related to banking and financial sector.
+
+If a user asks a question you have to classify it to following 3 types Relevant, Greeting, Other.
+ 
+"Relevant”: If the question is related to Banking and Financial Services Sector like Banking & Financial regulations, legal framework, governance framework, compliance requirements as per Central Bank regulations.
+"Greeting”: If the question is a greeting like good morning, hi my name is., thank you or General Question ask about the AI assistance of a company boardpac.
+"Other”: If the question is not related to research papers.
+ 
+Give the correct name of question type. If you are not sure return "Not Sure" instead.
+ 
+Question : {question}
+"""
+router_prompt=PromptTemplate.from_template(router_template_Mixtral_V0)
+

ensemble_retriever.py → reggpt/retriever/ensemble_retriever.py

@@ -1,228 +1,228 @@
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 14/11/2023
- *  LastModifiedDate : 18/03/2024
- *************************************************************************/
- """
-
-"""
-Ensemble retriever that ensemble the results of 
-multiple retrievers by using weighted  Reciprocal Rank Fusion
-"""
-
-import os
-import sys
-
-from pathlib import Path
-Path(__file__).resolve().parent.parent
-
-if os.path.dirname(os.path.abspath(__file__)) not in sys.path:
-    sys.path.append(os.path.dirname(os.path.abspath(__file__)))
-
-
-import logging
-logger = logging.getLogger(__name__)
-from typing import Any, Dict, List
-
-from langchain.callbacks.manager import (
-    AsyncCallbackManagerForRetrieverRun,
-    CallbackManagerForRetrieverRun,
-)
-from langchain.pydantic_v1 import root_validator
-from langchain.schema import BaseRetriever, Document
-
-import numpy as np
-import pandas as pd
-
-
-class EnsembleRetriever(BaseRetriever):
-    """Retriever that ensembles the multiple retrievers.
-
-    It uses a rank fusion.
-
-    Args:
-        retrievers: A list of retrievers to ensemble.
-        weights: A list of weights corresponding to the retrievers. Defaults to equal
-            weighting for all retrievers.
-        c: A constant added to the rank, controlling the balance between the importance
-            of high-ranked items and the consideration given to lower-ranked items.
-            Default is 60.
-    """
-
-    retrievers: List[BaseRetriever]
-    weights: List[float]
-    c: int = 60
-    date_key: str = "year"
-    top_k: int = 4
-
-    @root_validator(pre=True,allow_reuse=True)
-    def set_weights(cls, values: Dict[str, Any]) -> Dict[str, Any]:
-        if not values.get("weights"):
-            n_retrievers = len(values["retrievers"])
-            values["weights"] = [1 / n_retrievers] * n_retrievers
-        return values
-
-    def _get_relevant_documents(
-        self,
-        query: str,
-        *,
-        run_manager: CallbackManagerForRetrieverRun,
-    ) -> List[Document]:
-        """
-        Get the relevant documents for a given query.
-
-        Args:
-            query: The query to search for.
-
-        Returns:
-            A list of reranked documents.
-        """
-
-        # Get fused result of the retrievers.
-        fused_documents = self.rank_fusion(query, run_manager)
-
-        # check for key exists
-        if fused_documents[0].metadata[self.date_key] != None:
-            doc_dates = pd.to_datetime(
-                [doc.metadata[self.date_key] for doc in fused_documents]
-            )
-            sorted_node_idxs = np.flip(doc_dates.argsort())
-            fused_documents = [fused_documents[idx] for idx in sorted_node_idxs]
-            logger.info('Ensemble Retriever Documents sorted by year')
-
-        # return fused_documents[:self.top_k]
-        return fused_documents
-
-    async def _aget_relevant_documents(
-        self,
-        query: str,
-        *,
-        run_manager: AsyncCallbackManagerForRetrieverRun,
-    ) -> List[Document]:
-        """
-        Asynchronously get the relevant documents for a given query.
-
-        Args:
-            query: The query to search for.
-
-        Returns:
-            A list of reranked documents.
-        """
-
-        # Get fused result of the retrievers.
-        fused_documents = await self.arank_fusion(query, run_manager)
-
-        return fused_documents
-
-    def rank_fusion(
-        self, query: str, run_manager: CallbackManagerForRetrieverRun
-    ) -> List[Document]:
-        """
-        Retrieve the results of the retrievers and use rank_fusion_func to get
-        the final result.
-
-        Args:
-            query: The query to search for.
-
-        Returns:
-            A list of reranked documents.
-        """
-
-        # Get the results of all retrievers.
-        retriever_docs = [
-            retriever.get_relevant_documents(
-                query, callbacks=run_manager.get_child(tag=f"retriever_{i+1}")
-            )
-            for i, retriever in enumerate(self.retrievers)
-        ]
-
-        # apply rank fusion
-        fused_documents = self.weighted_reciprocal_rank(retriever_docs)
-
-        return fused_documents
-
-    async def arank_fusion(
-        self, query: str, run_manager: AsyncCallbackManagerForRetrieverRun
-    ) -> List[Document]:
-        """
-        Asynchronously retrieve the results of the retrievers
-        and use rank_fusion_func to get the final result.
-
-        Args:
-            query: The query to search for.
-
-        Returns:
-            A list of reranked documents.
-        """
-
-        # Get the results of all retrievers.
-        retriever_docs = [
-            await retriever.aget_relevant_documents(
-                query, callbacks=run_manager.get_child(tag=f"retriever_{i+1}")
-            )
-            for i, retriever in enumerate(self.retrievers)
-        ]
-
-        # apply rank fusion
-        fused_documents = self.weighted_reciprocal_rank(retriever_docs)
-
-        return fused_documents
-
-    def weighted_reciprocal_rank(
-        self, doc_lists: List[List[Document]]
-    ) -> List[Document]:
-        """
-        Perform weighted Reciprocal Rank Fusion on multiple rank lists.
-        You can find more details about RRF here:
-        https://plg.uwaterloo.ca/~gvcormac/cormacksigir09-rrf.pdf
-
-        Args:
-            doc_lists: A list of rank lists, where each rank list contains unique items.
-
-        Returns:
-            list: The final aggregated list of items sorted by their weighted RRF
-                    scores in descending order.
-        """
-        if len(doc_lists) != len(self.weights):
-            raise ValueError(
-                "Number of rank lists must be equal to the number of weights."
-            )
-
-        # Create a union of all unique documents in the input doc_lists
-        all_documents = set()
-        for doc_list in doc_lists:
-            for doc in doc_list:
-                all_documents.add(doc.page_content)
-
-        # Initialize the RRF score dictionary for each document
-        rrf_score_dic = {doc: 0.0 for doc in all_documents}
-
-        # Calculate RRF scores for each document
-        for doc_list, weight in zip(doc_lists, self.weights):
-            for rank, doc in enumerate(doc_list, start=1):
-                rrf_score = weight * (1 / (rank + self.c))
-                rrf_score_dic[doc.page_content] += rrf_score
-
-        # Sort documents by their RRF scores in descending order
-        sorted_documents = sorted(
-            rrf_score_dic.keys(), key=lambda x: rrf_score_dic[x], reverse=True
-        )
-
-        # Map the sorted page_content back to the original document objects
-        page_content_to_doc_map = {
-            doc.page_content: doc for doc_list in doc_lists for doc in doc_list
-        }
-        sorted_docs = [
-            page_content_to_doc_map[page_content] for page_content in sorted_documents
-        ]
-
-        return sorted_docs
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 14/11/2023
+ *  LastModifiedDate : 18/03/2024
+ *************************************************************************/
+ """
+
+"""
+Ensemble retriever that ensemble the results of 
+multiple retrievers by using weighted  Reciprocal Rank Fusion
+"""
+
+import os
+import sys
+
+from pathlib import Path
+Path(__file__).resolve().parent.parent
+
+if os.path.dirname(os.path.abspath(__file__)) not in sys.path:
+    sys.path.append(os.path.dirname(os.path.abspath(__file__)))
+
+
+import logging
+logger = logging.getLogger(__name__)
+from typing import Any, Dict, List
+
+from langchain.callbacks.manager import (
+    AsyncCallbackManagerForRetrieverRun,
+    CallbackManagerForRetrieverRun,
+)
+from langchain.pydantic_v1 import root_validator
+from langchain.schema import BaseRetriever, Document
+
+import numpy as np
+import pandas as pd
+
+
+class EnsembleRetriever(BaseRetriever):
+    """Retriever that ensembles the multiple retrievers.
+
+    It uses a rank fusion.
+
+    Args:
+        retrievers: A list of retrievers to ensemble.
+        weights: A list of weights corresponding to the retrievers. Defaults to equal
+            weighting for all retrievers.
+        c: A constant added to the rank, controlling the balance between the importance
+            of high-ranked items and the consideration given to lower-ranked items.
+            Default is 60.
+    """
+
+    retrievers: List[BaseRetriever]
+    weights: List[float]
+    c: int = 60
+    date_key: str = "year"
+    top_k: int = 4
+
+    @root_validator(pre=True,allow_reuse=True)
+    def set_weights(cls, values: Dict[str, Any]) -> Dict[str, Any]:
+        if not values.get("weights"):
+            n_retrievers = len(values["retrievers"])
+            values["weights"] = [1 / n_retrievers] * n_retrievers
+        return values
+
+    def _get_relevant_documents(
+        self,
+        query: str,
+        *,
+        run_manager: CallbackManagerForRetrieverRun,
+    ) -> List[Document]:
+        """
+        Get the relevant documents for a given query.
+
+        Args:
+            query: The query to search for.
+
+        Returns:
+            A list of reranked documents.
+        """
+
+        # Get fused result of the retrievers.
+        fused_documents = self.rank_fusion(query, run_manager)
+
+        # check for key exists
+        if fused_documents[0].metadata[self.date_key] != None:
+            doc_dates = pd.to_datetime(
+                [doc.metadata[self.date_key] for doc in fused_documents]
+            )
+            sorted_node_idxs = np.flip(doc_dates.argsort())
+            fused_documents = [fused_documents[idx] for idx in sorted_node_idxs]
+            logger.info('Ensemble Retriever Documents sorted by year')
+
+        # return fused_documents[:self.top_k]
+        return fused_documents
+
+    async def _aget_relevant_documents(
+        self,
+        query: str,
+        *,
+        run_manager: AsyncCallbackManagerForRetrieverRun,
+    ) -> List[Document]:
+        """
+        Asynchronously get the relevant documents for a given query.
+
+        Args:
+            query: The query to search for.
+
+        Returns:
+            A list of reranked documents.
+        """
+
+        # Get fused result of the retrievers.
+        fused_documents = await self.arank_fusion(query, run_manager)
+
+        return fused_documents
+
+    def rank_fusion(
+        self, query: str, run_manager: CallbackManagerForRetrieverRun
+    ) -> List[Document]:
+        """
+        Retrieve the results of the retrievers and use rank_fusion_func to get
+        the final result.
+
+        Args:
+            query: The query to search for.
+
+        Returns:
+            A list of reranked documents.
+        """
+
+        # Get the results of all retrievers.
+        retriever_docs = [
+            retriever.get_relevant_documents(
+                query, callbacks=run_manager.get_child(tag=f"retriever_{i+1}")
+            )
+            for i, retriever in enumerate(self.retrievers)
+        ]
+
+        # apply rank fusion
+        fused_documents = self.weighted_reciprocal_rank(retriever_docs)
+
+        return fused_documents
+
+    async def arank_fusion(
+        self, query: str, run_manager: AsyncCallbackManagerForRetrieverRun
+    ) -> List[Document]:
+        """
+        Asynchronously retrieve the results of the retrievers
+        and use rank_fusion_func to get the final result.
+
+        Args:
+            query: The query to search for.
+
+        Returns:
+            A list of reranked documents.
+        """
+
+        # Get the results of all retrievers.
+        retriever_docs = [
+            await retriever.aget_relevant_documents(
+                query, callbacks=run_manager.get_child(tag=f"retriever_{i+1}")
+            )
+            for i, retriever in enumerate(self.retrievers)
+        ]
+
+        # apply rank fusion
+        fused_documents = self.weighted_reciprocal_rank(retriever_docs)
+
+        return fused_documents
+
+    def weighted_reciprocal_rank(
+        self, doc_lists: List[List[Document]]
+    ) -> List[Document]:
+        """
+        Perform weighted Reciprocal Rank Fusion on multiple rank lists.
+        You can find more details about RRF here:
+        https://plg.uwaterloo.ca/~gvcormac/cormacksigir09-rrf.pdf
+
+        Args:
+            doc_lists: A list of rank lists, where each rank list contains unique items.
+
+        Returns:
+            list: The final aggregated list of items sorted by their weighted RRF
+                    scores in descending order.
+        """
+        if len(doc_lists) != len(self.weights):
+            raise ValueError(
+                "Number of rank lists must be equal to the number of weights."
+            )
+
+        # Create a union of all unique documents in the input doc_lists
+        all_documents = set()
+        for doc_list in doc_lists:
+            for doc in doc_list:
+                all_documents.add(doc.page_content)
+
+        # Initialize the RRF score dictionary for each document
+        rrf_score_dic = {doc: 0.0 for doc in all_documents}
+
+        # Calculate RRF scores for each document
+        for doc_list, weight in zip(doc_lists, self.weights):
+            for rank, doc in enumerate(doc_list, start=1):
+                rrf_score = weight * (1 / (rank + self.c))
+                rrf_score_dic[doc.page_content] += rrf_score
+
+        # Sort documents by their RRF scores in descending order
+        sorted_documents = sorted(
+            rrf_score_dic.keys(), key=lambda x: rrf_score_dic[x], reverse=True
+        )
+
+        # Map the sorted page_content back to the original document objects
+        page_content_to_doc_map = {
+            doc.page_content: doc for doc_list in doc_lists for doc in doc_list
+        }
+        sorted_docs = [
+            page_content_to_doc_map[page_content] for page_content in sorted_documents
+        ]
+
+        return sorted_docs

multi_query_retriever.py → reggpt/retriever/multi_query_retriever.py

@@ -1,254 +1,254 @@
-
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 14/11/2023
- *  LastModifiedDate : 21/03/2024
- *************************************************************************/
-"""
-
-import asyncio
-import logging
-from typing import List, Optional, Sequence
-
-from langchain_core.callbacks import (
-    AsyncCallbackManagerForRetrieverRun,
-    CallbackManagerForRetrieverRun,
-)
-from langchain_core.documents import Document
-from langchain_core.language_models import BaseLanguageModel
-from langchain_core.output_parsers import BaseOutputParser
-from langchain_core.prompts.prompt import PromptTemplate
-from langchain_core.retrievers import BaseRetriever
-
-from langchain.chains.llm import LLMChain
-
-import numpy as np
-import pandas as pd
-
-logger = logging.getLogger(__name__)
-
-from prompts import MULTY_QUERY_PROMPT
-
-class LineListOutputParser(BaseOutputParser[List[str]]):
-    """Output parser for a list of lines."""
-
-    def parse(self, text: str) -> List[str]:
-        lines = text.strip().split("\n")
-        return lines
-
-
-# Default prompt
-# DEFAULT_QUERY_PROMPT = PromptTemplate(
-#     input_variables=["question"],
-#     template="""You are an AI language model assistant. Your task is 
-#     to generate 3 different versions of the given user 
-#     question to retrieve relevant documents from a vector  database. 
-#     By generating multiple perspectives on the user question, 
-#     your goal is to help the user overcome some of the limitations 
-#     of distance-based similarity search. Provide these alternative 
-#     questions separated by newlines. Original question: {question}""",
-# )
-
-
-
-
-def _unique_documents(documents: Sequence[Document]) -> List[Document]:
-    return [doc for i, doc in enumerate(documents) if doc not in documents[:i]]
-
-
-class MultiQueryRetriever(BaseRetriever):
-    """Given a query, use an LLM to write a set of queries.
-
-    Retrieve docs for each query. Return the unique union of all retrieved docs.
-    """
-
-    retriever: BaseRetriever
-    llm_chain: LLMChain
-    verbose: bool = True
-    parser_key: str = "lines"
-    """DEPRECATED. parser_key is no longer used and should not be specified."""
-    include_original: bool = False
-    """Whether to include the original query in the list of generated queries."""
-    date_key: str = "year"
-    top_k: int = 4
-
-    @classmethod
-    def from_llm(
-        cls,
-        retriever: BaseRetriever,
-        llm: BaseLanguageModel,
-        prompt: PromptTemplate = MULTY_QUERY_PROMPT,
-        parser_key: Optional[str] = None,
-        include_original: bool = False,
-    ) -> "MultiQueryRetriever":
-        """Initialize from llm using default template.
-
-        Args:
-            retriever: retriever to query documents from
-            llm: llm for query generation using DEFAULT_QUERY_PROMPT
-            include_original: Whether to include the original query in the list of
-                generated queries.
-
-        Returns:
-            MultiQueryRetriever
-        """
-        output_parser = LineListOutputParser()
-        llm_chain = LLMChain(llm=llm, prompt=prompt, output_parser=output_parser)
-        return cls(
-            retriever=retriever,
-            llm_chain=llm_chain,
-            include_original=include_original,
-        )
-
-    async def _aget_relevant_documents(
-        self,
-        query: str,
-        *,
-        run_manager: AsyncCallbackManagerForRetrieverRun,
-    ) -> List[Document]:
-        """Get relevant documents given a user query.
-
-        Args:
-            question: user query
-
-        Returns:
-            Unique union of relevant documents from all generated queries
-        """
-        queries = await self.agenerate_queries(query, run_manager)
-        if self.include_original:
-            queries.append(query)
-        documents = await self.aretrieve_documents(queries, run_manager)
-        return self.unique_union(documents)
-
-
-    async def agenerate_queries(
-        self, question: str, run_manager: AsyncCallbackManagerForRetrieverRun
-    ) -> List[str]:
-        """Generate queries based upon user input.
-
-        Args:
-            question: user query
-
-        Returns:
-            List of LLM generated queries that are similar to the user input
-        """
-        response = await self.llm_chain.ainvoke(
-            inputs={"question": question}, callbacks=run_manager.get_child()
-        )
-        lines = response["text"]
-        if self.verbose:
-            logger.info(f"Generated queries: {lines}")
-        return lines
-
-    async def aretrieve_documents(
-        self, queries: List[str], run_manager: AsyncCallbackManagerForRetrieverRun
-    ) -> List[Document]:
-        """Run all LLM generated queries.
-
-        Args:
-            queries: query list
-
-        Returns:
-            List of retrieved Documents
-        """
-        document_lists = await asyncio.gather(
-            *(
-                self.retriever.aget_relevant_documents(
-                    query, callbacks=run_manager.get_child()
-                )
-                for query in queries
-            )
-        )
-        return [doc for docs in document_lists for doc in docs]
-
-    def _get_relevant_documents(
-        self,
-        query: str,
-        *,
-        run_manager: CallbackManagerForRetrieverRun,
-    ) -> List[Document]:
-        """Get relevant documents given a user query.
-
-        Args:
-            question: user query
-
-        Returns:
-            Unique union of relevant documents from all generated queries
-        """
-        queries = self.generate_queries(query, run_manager)
-        if self.include_original:
-            queries.append(query)
-        documents = self.retrieve_documents(queries, run_manager)
-        fused_documents= self.unique_union(documents)
-        # check for key exists
-        if fused_documents[0].metadata[self.date_key] != None:
-            doc_dates = pd.to_datetime(
-                [doc.metadata[self.date_key] for doc in fused_documents]
-            )
-            sorted_node_idxs = np.flip(doc_dates.argsort())
-            fused_documents = [fused_documents[idx] for idx in sorted_node_idxs]
-            logger.info('Documents sorted by year')
-
-        return fused_documents[:self.top_k]
-    
-
-
-    def generate_queries(
-        self, question: str, run_manager: CallbackManagerForRetrieverRun
-    ) -> List[str]:
-        """Generate queries based upon user input.
-
-        Args:
-            question: user query
-
-        Returns:
-            List of LLM generated queries that are similar to the user input
-        """
-        response = self.llm_chain.invoke(
-            {"question": question}, callbacks=run_manager.get_child()
-        )
-        lines = response["text"]
-        if self.verbose:
-            logger.info(f"Generated queries: {lines}")
-        return lines
-
-    def retrieve_documents(
-        self, queries: List[str], run_manager: CallbackManagerForRetrieverRun
-    ) -> List[Document]:
-        """Run all LLM generated queries.
-
-        Args:
-            queries: query list
-
-        Returns:
-            List of retrieved Documents
-        """
-        documents = []
-        for query in queries:
-            logger.info(f"MQ Retriever question: {query}")
-            docs = self.retriever.get_relevant_documents(
-                query, callbacks=run_manager.get_child()
-            )
-            documents.extend(docs)
-        return documents
-
-    def unique_union(self, documents: List[Document]) -> List[Document]:
-        """Get unique Documents.
-
-        Args:
-            documents: List of retrieved Documents
-
-        Returns:
-            List of unique retrieved Documents
-        """
-        return _unique_documents(documents)
+
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 14/11/2023
+ *  LastModifiedDate : 21/03/2024
+ *************************************************************************/
+"""
+
+import asyncio
+import logging
+from typing import List, Optional, Sequence
+
+from langchain_core.callbacks import (
+    AsyncCallbackManagerForRetrieverRun,
+    CallbackManagerForRetrieverRun,
+)
+from langchain_core.documents import Document
+from langchain_core.language_models import BaseLanguageModel
+from langchain_core.output_parsers import BaseOutputParser
+from langchain_core.prompts.prompt import PromptTemplate
+from langchain_core.retrievers import BaseRetriever
+
+from langchain.chains.llm import LLMChain
+
+import numpy as np
+import pandas as pd
+
+logger = logging.getLogger(__name__)
+
+from reggpt.prompts.prompts import MULTY_QUERY_PROMPT
+
+class LineListOutputParser(BaseOutputParser[List[str]]):
+    """Output parser for a list of lines."""
+
+    def parse(self, text: str) -> List[str]:
+        lines = text.strip().split("\n")
+        return lines
+
+
+# Default prompt
+# DEFAULT_QUERY_PROMPT = PromptTemplate(
+#     input_variables=["question"],
+#     template="""You are an AI language model assistant. Your task is 
+#     to generate 3 different versions of the given user 
+#     question to retrieve relevant documents from a vector  database. 
+#     By generating multiple perspectives on the user question, 
+#     your goal is to help the user overcome some of the limitations 
+#     of distance-based similarity search. Provide these alternative 
+#     questions separated by newlines. Original question: {question}""",
+# )
+
+
+
+
+def _unique_documents(documents: Sequence[Document]) -> List[Document]:
+    return [doc for i, doc in enumerate(documents) if doc not in documents[:i]]
+
+
+class MultiQueryRetriever(BaseRetriever):
+    """Given a query, use an LLM to write a set of queries.
+
+    Retrieve docs for each query. Return the unique union of all retrieved docs.
+    """
+
+    retriever: BaseRetriever
+    llm_chain: LLMChain
+    verbose: bool = True
+    parser_key: str = "lines"
+    """DEPRECATED. parser_key is no longer used and should not be specified."""
+    include_original: bool = False
+    """Whether to include the original query in the list of generated queries."""
+    date_key: str = "year"
+    top_k: int = 4
+
+    @classmethod
+    def from_llm(
+        cls,
+        retriever: BaseRetriever,
+        llm: BaseLanguageModel,
+        prompt: PromptTemplate = MULTY_QUERY_PROMPT,
+        parser_key: Optional[str] = None,
+        include_original: bool = False,
+    ) -> "MultiQueryRetriever":
+        """Initialize from llm using default template.
+
+        Args:
+            retriever: retriever to query documents from
+            llm: llm for query generation using DEFAULT_QUERY_PROMPT
+            include_original: Whether to include the original query in the list of
+                generated queries.
+
+        Returns:
+            MultiQueryRetriever
+        """
+        output_parser = LineListOutputParser()
+        llm_chain = LLMChain(llm=llm, prompt=prompt, output_parser=output_parser)
+        return cls(
+            retriever=retriever,
+            llm_chain=llm_chain,
+            include_original=include_original,
+        )
+
+    async def _aget_relevant_documents(
+        self,
+        query: str,
+        *,
+        run_manager: AsyncCallbackManagerForRetrieverRun,
+    ) -> List[Document]:
+        """Get relevant documents given a user query.
+
+        Args:
+            question: user query
+
+        Returns:
+            Unique union of relevant documents from all generated queries
+        """
+        queries = await self.agenerate_queries(query, run_manager)
+        if self.include_original:
+            queries.append(query)
+        documents = await self.aretrieve_documents(queries, run_manager)
+        return self.unique_union(documents)
+
+
+    async def agenerate_queries(
+        self, question: str, run_manager: AsyncCallbackManagerForRetrieverRun
+    ) -> List[str]:
+        """Generate queries based upon user input.
+
+        Args:
+            question: user query
+
+        Returns:
+            List of LLM generated queries that are similar to the user input
+        """
+        response = await self.llm_chain.ainvoke(
+            inputs={"question": question}, callbacks=run_manager.get_child()
+        )
+        lines = response["text"]
+        if self.verbose:
+            logger.info(f"Generated queries: {lines}")
+        return lines
+
+    async def aretrieve_documents(
+        self, queries: List[str], run_manager: AsyncCallbackManagerForRetrieverRun
+    ) -> List[Document]:
+        """Run all LLM generated queries.
+
+        Args:
+            queries: query list
+
+        Returns:
+            List of retrieved Documents
+        """
+        document_lists = await asyncio.gather(
+            *(
+                self.retriever.aget_relevant_documents(
+                    query, callbacks=run_manager.get_child()
+                )
+                for query in queries
+            )
+        )
+        return [doc for docs in document_lists for doc in docs]
+
+    def _get_relevant_documents(
+        self,
+        query: str,
+        *,
+        run_manager: CallbackManagerForRetrieverRun,
+    ) -> List[Document]:
+        """Get relevant documents given a user query.
+
+        Args:
+            question: user query
+
+        Returns:
+            Unique union of relevant documents from all generated queries
+        """
+        queries = self.generate_queries(query, run_manager)
+        if self.include_original:
+            queries.append(query)
+        documents = self.retrieve_documents(queries, run_manager)
+        fused_documents= self.unique_union(documents)
+        # check for key exists
+        if fused_documents[0].metadata[self.date_key] != None:
+            doc_dates = pd.to_datetime(
+                [doc.metadata[self.date_key] for doc in fused_documents]
+            )
+            sorted_node_idxs = np.flip(doc_dates.argsort())
+            fused_documents = [fused_documents[idx] for idx in sorted_node_idxs]
+            logger.info('Documents sorted by year')
+
+        return fused_documents[:self.top_k]
+    
+
+
+    def generate_queries(
+        self, question: str, run_manager: CallbackManagerForRetrieverRun
+    ) -> List[str]:
+        """Generate queries based upon user input.
+
+        Args:
+            question: user query
+
+        Returns:
+            List of LLM generated queries that are similar to the user input
+        """
+        response = self.llm_chain.invoke(
+            {"question": question}, callbacks=run_manager.get_child()
+        )
+        lines = response["text"]
+        if self.verbose:
+            logger.info(f"Generated queries: {lines}")
+        return lines
+
+    def retrieve_documents(
+        self, queries: List[str], run_manager: CallbackManagerForRetrieverRun
+    ) -> List[Document]:
+        """Run all LLM generated queries.
+
+        Args:
+            queries: query list
+
+        Returns:
+            List of retrieved Documents
+        """
+        documents = []
+        for query in queries:
+            logger.info(f"MQ Retriever question: {query}")
+            docs = self.retriever.get_relevant_documents(
+                query, callbacks=run_manager.get_child()
+            )
+            documents.extend(docs)
+        return documents
+
+    def unique_union(self, documents: List[Document]) -> List[Document]:
+        """Get unique Documents.
+
+        Args:
+            documents: List of retrieved Documents
+
+        Returns:
+            List of unique retrieved Documents
+        """
+        return _unique_documents(documents)
     

controller.py → reggpt/routers/controller.py

@@ -16,13 +16,13 @@
 
 import logging
 logger = logging.getLogger(__name__)
-from config import AVALIABLE_MODELS , MEMORY_WINDOW_K
+from reggpt.configs.config import AVALIABLE_MODELS , MEMORY_WINDOW_K
 
 # from qaPipeline import QAPipeline
 # from qaPipeline_retriever_only import QAPipeline
 # qaPipeline = QAPipeline()
 
-from qaPipeline import run_agent
+from reggpt.routers.qaPipeline import run_agent
 
 def get_QA_Answers(userQuery):
     # model=userQuery.model

reggpt/routers/general.py

@@ -0,0 +1,49 @@
+
+import os
+import time
+import logging
+logger = logging.getLogger(__name__)
+from dotenv import load_dotenv
+from fastapi import HTTPException
+from reggpt.chains.llmChain import get_qa_chain, get_general_qa_chain, get_router_chain
+from reggpt.output_parsers.output_parser import general_qa_chain_output_parser, qa_chain_output_parser, out_of_domain_chain_parser
+
+from reggpt.configs.config import QA_MODEL_TYPE, GENERAL_QA_MODEL_TYPE, ROUTER_MODEL_TYPE, Multi_Query_MODEL_TYPE
+from reggpt.utils.retriever import load_faiss_retriever, load_ensemble_retriever, load_multi_query_retriever
+load_dotenv()
+
+verbose = os.environ.get('VERBOSE')
+
+qa_model_type=QA_MODEL_TYPE
+general_qa_model_type=GENERAL_QA_MODEL_TYPE
+router_model_type=ROUTER_MODEL_TYPE #"google/flan-t5-xxl"
+multi_query_model_type=Multi_Query_MODEL_TYPE #"google/flan-t5-xxl"
+# model_type="tiiuae/falcon-7b-instruct"
+
+# retriever=load_faiss_retriever()
+retriever=load_ensemble_retriever()
+# retriever=load_multi_query_retriever(multi_query_model_type)
+logger.info("retriever loaded:")
+
+qa_chain= get_qa_chain(qa_model_type,retriever)
+general_qa_chain= get_general_qa_chain(general_qa_model_type)
+router_chain= get_router_chain(router_model_type)
+
+def run_general_qa_chain(query):
+    try:
+        logger.info(f"run_general_qa_chain : Question: {query}")
+
+        # Get the answer from the chain
+        start = time.time()
+        res = general_qa_chain.invoke(query)
+        end = time.time()
+
+        # log the result
+       
+        logger.info(f"Answer (took {round(end - start, 2)} s.) \n: {res}")
+
+        return general_qa_chain_output_parser(res)
+
+    except Exception as e:
+        logger.exception(e)
+        raise e

reggpt/routers/out_of_domain.py

@@ -0,0 +1,31 @@
+import os
+import time
+import logging
+logger = logging.getLogger(__name__)
+from dotenv import load_dotenv
+from fastapi import HTTPException
+from reggpt.chains.llmChain import get_qa_chain, get_general_qa_chain, get_router_chain
+from reggpt.output_parsers.output_parser import general_qa_chain_output_parser, qa_chain_output_parser, out_of_domain_chain_parser
+
+from reggpt.configs.config import QA_MODEL_TYPE, GENERAL_QA_MODEL_TYPE, ROUTER_MODEL_TYPE, Multi_Query_MODEL_TYPE
+from reggpt.utils.retriever import load_faiss_retriever, load_ensemble_retriever, load_multi_query_retriever
+load_dotenv()
+
+verbose = os.environ.get('VERBOSE')
+
+qa_model_type=QA_MODEL_TYPE
+general_qa_model_type=GENERAL_QA_MODEL_TYPE
+router_model_type=ROUTER_MODEL_TYPE #"google/flan-t5-xxl"
+multi_query_model_type=Multi_Query_MODEL_TYPE #"google/flan-t5-xxl"
+# model_type="tiiuae/falcon-7b-instruct"
+
+# retriever=load_faiss_retriever()
+retriever=load_ensemble_retriever()
+# retriever=load_multi_query_retriever(multi_query_model_type)
+logger.info("retriever loaded:")
+
+qa_chain= get_qa_chain(qa_model_type,retriever)
+general_qa_chain= get_general_qa_chain(general_qa_model_type)
+router_chain= get_router_chain(router_model_type)
+def run_out_of_domain_chain(query):
+    return out_of_domain_chain_parser(query)

reggpt/routers/qa.py

@@ -0,0 +1,66 @@
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 14/11/2023
+ *  LastModifiedDate : 18/03/2024
+ *************************************************************************/
+ """
+
+import os
+import time
+import logging
+logger = logging.getLogger(__name__)
+from dotenv import load_dotenv
+from fastapi import HTTPException
+from reggpt.chains.llmChain import get_qa_chain, get_general_qa_chain, get_router_chain
+from reggpt.output_parsers.output_parser import general_qa_chain_output_parser, qa_chain_output_parser, out_of_domain_chain_parser
+
+from reggpt.configs.config import QA_MODEL_TYPE, GENERAL_QA_MODEL_TYPE, ROUTER_MODEL_TYPE, Multi_Query_MODEL_TYPE
+from reggpt.utils.retriever import load_faiss_retriever, load_ensemble_retriever, load_multi_query_retriever
+load_dotenv()
+
+verbose = os.environ.get('VERBOSE')
+
+qa_model_type=QA_MODEL_TYPE
+general_qa_model_type=GENERAL_QA_MODEL_TYPE
+router_model_type=ROUTER_MODEL_TYPE #"google/flan-t5-xxl"
+multi_query_model_type=Multi_Query_MODEL_TYPE #"google/flan-t5-xxl"
+# model_type="tiiuae/falcon-7b-instruct"
+
+# retriever=load_faiss_retriever()
+retriever=load_ensemble_retriever()
+# retriever=load_multi_query_retriever(multi_query_model_type)
+logger.info("retriever loaded:")
+
+qa_chain= get_qa_chain(qa_model_type,retriever)
+general_qa_chain= get_general_qa_chain(general_qa_model_type)
+router_chain= get_router_chain(router_model_type)
+
+def run_qa_chain(query):
+    try:
+        logger.info(f"run_qa_chain : Question: {query}")
+        # Get the answer from the chain
+        start = time.time()
+        # res = qa_chain(query)
+        res = qa_chain.invoke({"question": query, "chat_history":""})
+        # res = response
+        # answer, docs = res['result'],res['source_documents']
+        end = time.time()
+
+        # log the result
+        logger.info(f"Answer (took {round(end - start, 2)} s.) \n: {res}")
+
+        return qa_chain_output_parser(res)
+    
+    except Exception as e:
+        logger.exception(e)
+        raise e
+

reggpt/routers/qaPipeline.py

@@ -0,0 +1,45 @@
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 14/11/2023
+ *  LastModifiedDate : 18/03/2024
+ *************************************************************************/
+ """
+
+import os
+import time
+import logging
+logger = logging.getLogger(__name__)
+from dotenv import load_dotenv
+from fastapi import HTTPException
+from reggpt.chains.llmChain import get_qa_chain, get_general_qa_chain, get_router_chain
+from reggpt.output_parsers.output_parser import general_qa_chain_output_parser, qa_chain_output_parser, out_of_domain_chain_parser
+
+from reggpt.configs.config import QA_MODEL_TYPE, GENERAL_QA_MODEL_TYPE, ROUTER_MODEL_TYPE, Multi_Query_MODEL_TYPE
+from reggpt.utils.retriever import load_faiss_retriever, load_ensemble_retriever, load_multi_query_retriever
+load_dotenv()
+
+verbose = os.environ.get('VERBOSE')
+
+qa_model_type=QA_MODEL_TYPE
+general_qa_model_type=GENERAL_QA_MODEL_TYPE
+router_model_type=ROUTER_MODEL_TYPE #"google/flan-t5-xxl"
+multi_query_model_type=Multi_Query_MODEL_TYPE #"google/flan-t5-xxl"
+# model_type="tiiuae/falcon-7b-instruct"
+
+# retriever=load_faiss_retriever()
+retriever=load_ensemble_retriever()
+# retriever=load_multi_query_retriever(multi_query_model_type)
+logger.info("retriever loaded:")
+
+qa_chain= get_qa_chain(qa_model_type,retriever)
+general_qa_chain= get_general_qa_chain(general_qa_model_type)
+router_chain= get_router_chain(router_model_type)

retriever.py → reggpt/utils/retriever.py

@@ -1,137 +1,137 @@
-
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 19/03/2023
- *  LastModifiedDate : 19/03/2024
- *************************************************************************/
- """
-
-"""
-Ensemble retriever that ensemble the results of 
-multiple retrievers by using weighted  Reciprocal Rank Fusion
-"""
-
-import logging
-logger = logging.getLogger(__name__)
-
-from faissDb import load_FAISS_store
-
-from langchain_community.retrievers import BM25Retriever
-from langchain_community.document_loaders import PyPDFLoader
-from langchain_community.document_loaders import DirectoryLoader
-from langchain_text_splitters import RecursiveCharacterTextSplitter
-
-from langchain.schema import Document
-from typing import Iterable
-import json
-
-def save_docs_to_jsonl(array:Iterable[Document], file_path:str)->None:
-    with open(file_path, 'w') as jsonl_file:
-        for doc in array:
-            jsonl_file.write(doc.json() + '\n')
-
-def load_docs_from_jsonl(file_path)->Iterable[Document]:
-    array = []
-    with open(file_path, 'r') as jsonl_file:
-        for line in jsonl_file:
-            data = json.loads(line)
-            obj = Document(**data)
-            array.append(obj)
-    return array
-
-def split_documents():
-    chunk_size=2000
-    chunk_overlap=100
-
-    text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
-
-    years = [2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024]
-    docs_list=[]
-    splits_list=[]
-
-
-    for year in years:
-        data_path= f"data/CBSL/{year}"
-        logger.info(f"Loading year : {data_path}")
-
-        documents = DirectoryLoader(data_path,  loader_cls=PyPDFLoader).load()
-
-        for doc in documents:
-            doc.metadata['year']=year
-            logger.info(f"{doc.metadata['year']} : {doc.metadata['source']}" )
-            docs_list.append(doc)
-
-        texts = text_splitter.split_documents(documents)
-        for text in texts:
-            splits_list.append(text)
-
-    splitted_texts_file='data/splitted_texts.jsonl'
-    save_docs_to_jsonl(splits_list,splitted_texts_file)
-
-from ensemble_retriever import EnsembleRetriever
-from multi_query_retriever import MultiQueryRetriever
-
-def load_faiss_retriever():
-    try:
-        vectorstore=load_FAISS_store()
-        retriever = vectorstore.as_retriever(
-            # search_type="mmr",
-            search_kwargs={'k': 5, 'fetch_k': 10}
-            )
-        logger.info("FAISS Retriever loaded:")
-        return retriever
-    
-    except Exception as e:
-        logger.exception(e)
-        raise e
-    
-def load_ensemble_retriever():
-    try:
-        # splitted_texts_file=os.path.dirname(os.path.abspath(__file__).join('/data/splitted_texts.jsonl'))
-        splitted_texts_file='./data/splitted_texts.jsonl'
-        sementic_k = 4
-        bm25_k = 2
-        splits_list = load_docs_from_jsonl(splitted_texts_file)
-
-        bm25_retriever  = BM25Retriever.from_documents(splits_list)
-        bm25_retriever.k = bm25_k
-
-        faiss_vectorstore = load_FAISS_store()
-        faiss_retriever  = faiss_vectorstore.as_retriever(search_kwargs={'k': sementic_k,})
-        
-        ensemble_retriever = EnsembleRetriever(retrievers=[bm25_retriever, faiss_retriever], weights=[0.5, 0.5])
-        ensemble_retriever.top_k=4
-        
-        logger.info("EnsembleRetriever loaded:")
-        return ensemble_retriever
-    
-    except Exception as e:
-        logger.exception(e)
-        raise e
-
-from llm import get_model
-
-def load_multi_query_retriever(multi_query_model_type):
-    #multi query
-    try:
-        llm = get_model(multi_query_model_type)
-        ensembleRetriever = load_ensemble_retriever()
-        retriever  = MultiQueryRetriever.from_llm(
-            retriever=ensembleRetriever,
-            llm=llm
-        )
-        logger.info("MultiQueryRetriever loaded:")
-        return retriever
-
-    except Exception as e:
-        logger.exception(e)
+
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 19/03/2023
+ *  LastModifiedDate : 19/03/2024
+ *************************************************************************/
+ """
+
+"""
+Ensemble retriever that ensemble the results of 
+multiple retrievers by using weighted  Reciprocal Rank Fusion
+"""
+
+import logging
+logger = logging.getLogger(__name__)
+
+from reggpt.vectorstores.faissDb import load_FAISS_store
+
+from langchain_community.retrievers import BM25Retriever
+from langchain_community.document_loaders import PyPDFLoader
+from langchain_community.document_loaders import DirectoryLoader
+from langchain_text_splitters import RecursiveCharacterTextSplitter
+
+from langchain.schema import Document
+from typing import Iterable
+import json
+
+def save_docs_to_jsonl(array:Iterable[Document], file_path:str)->None:
+    with open(file_path, 'w') as jsonl_file:
+        for doc in array:
+            jsonl_file.write(doc.json() + '\n')
+
+def load_docs_from_jsonl(file_path)->Iterable[Document]:
+    array = []
+    with open(file_path, 'r') as jsonl_file:
+        for line in jsonl_file:
+            data = json.loads(line)
+            obj = Document(**data)
+            array.append(obj)
+    return array
+
+def split_documents():
+    chunk_size=2000
+    chunk_overlap=100
+
+    text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=chunk_overlap)
+
+    years = [2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024]
+    docs_list=[]
+    splits_list=[]
+
+
+    for year in years:
+        data_path= f"data/CBSL/{year}"
+        logger.info(f"Loading year : {data_path}")
+
+        documents = DirectoryLoader(data_path,  loader_cls=PyPDFLoader).load()
+
+        for doc in documents:
+            doc.metadata['year']=year
+            logger.info(f"{doc.metadata['year']} : {doc.metadata['source']}" )
+            docs_list.append(doc)
+
+        texts = text_splitter.split_documents(documents)
+        for text in texts:
+            splits_list.append(text)
+
+    splitted_texts_file='data/splitted_texts.jsonl'
+    save_docs_to_jsonl(splits_list,splitted_texts_file)
+
+from ensemble_retriever import EnsembleRetriever
+from multi_query_retriever import MultiQueryRetriever
+
+def load_faiss_retriever():
+    try:
+        vectorstore=load_FAISS_store()
+        retriever = vectorstore.as_retriever(
+            # search_type="mmr",
+            search_kwargs={'k': 5, 'fetch_k': 10}
+            )
+        logger.info("FAISS Retriever loaded:")
+        return retriever
+    
+    except Exception as e:
+        logger.exception(e)
+        raise e
+    
+def load_ensemble_retriever():
+    try:
+        # splitted_texts_file=os.path.dirname(os.path.abspath(__file__).join('/data/splitted_texts.jsonl'))
+        splitted_texts_file='./data/splitted_texts.jsonl'
+        sementic_k = 4
+        bm25_k = 2
+        splits_list = load_docs_from_jsonl(splitted_texts_file)
+
+        bm25_retriever  = BM25Retriever.from_documents(splits_list)
+        bm25_retriever.k = bm25_k
+
+        faiss_vectorstore = load_FAISS_store()
+        faiss_retriever  = faiss_vectorstore.as_retriever(search_kwargs={'k': sementic_k,})
+        
+        ensemble_retriever = EnsembleRetriever(retrievers=[bm25_retriever, faiss_retriever], weights=[0.5, 0.5])
+        ensemble_retriever.top_k=4
+        
+        logger.info("EnsembleRetriever loaded:")
+        return ensemble_retriever
+    
+    except Exception as e:
+        logger.exception(e)
+        raise e
+
+from reggpt.llms.llm import get_model
+
+def load_multi_query_retriever(multi_query_model_type):
+    #multi query
+    try:
+        llm = get_model(multi_query_model_type)
+        ensembleRetriever = load_ensemble_retriever()
+        retriever  = MultiQueryRetriever.from_llm(
+            retriever=ensembleRetriever,
+            llm=llm
+        )
+        logger.info("MultiQueryRetriever loaded:")
+        return retriever
+
+    except Exception as e:
+        logger.exception(e)
         raise e

{utils → reggpt/utils}/utils.py

@@ -1,40 +1,40 @@
-"""
-Python Backend API to chat with private data  
-15/11/2023
-Theekshana Samaradiwakara
-"""
-"""
- /*************************************************************************
- * 
- * CONFIDENTIAL
- * __________________
- * 
- *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
- *  All Rights Reserved
- *
- *  Author  : Theekshana Samaradiwakara
- *  Description :Python Backend API to chat with private data  
- *  CreatedDate : 15/11/2023
- *  LastModifiedDate : 10/12/2020
- *************************************************************************/
- """
-
-# from passlib.context import CryptContext
-# pwd_context = CryptContext(schemes=["bcrypt"], deprecated="auto")
-
-
-# def hash(password: str):
-#     return pwd_context.hash(password)
-
-
-# def verify(plain_password, hashed_password):
-#     return pwd_context.verify(plain_password, hashed_password)
-
-
-
-import re
-def is_valid_open_ai_api_key(secretKey):
-    if re.search("^sk-[a-zA-Z0-9]{32,}$", secretKey ): 
-        return True
-    else: return False
-
+"""
+Python Backend API to chat with private data  
+15/11/2023
+Theekshana Samaradiwakara
+"""
+"""
+ /*************************************************************************
+ * 
+ * CONFIDENTIAL
+ * __________________
+ * 
+ *  Copyright (2023-2024) AI Labs, IronOne Technologies, LLC
+ *  All Rights Reserved
+ *
+ *  Author  : Theekshana Samaradiwakara
+ *  Description :Python Backend API to chat with private data  
+ *  CreatedDate : 15/11/2023
+ *  LastModifiedDate : 10/12/2020
+ *************************************************************************/
+ """
+
+# from passlib.context import CryptContext
+# pwd_context = CryptContext(schemes=["bcrypt"], deprecated="auto")
+
+
+# def hash(password: str):
+#     return pwd_context.hash(password)
+
+
+# def verify(plain_password, hashed_password):
+#     return pwd_context.verify(plain_password, hashed_password)
+
+
+
+import re
+def is_valid_open_ai_api_key(secretKey):
+    if re.search("^sk-[a-zA-Z0-9]{32,}$", secretKey ): 
+        return True
+    else: return False
+