[CAI-185] Chatbot/docker compose with Redis and DynamoDB for local development

apps/chatbot/.env.example

@@ -24,3 +24,5 @@ CHB_ENGINE_SIMILARITY_CUTOFF=...
 CHB_ENGINE_USE_ASYNC=...
 CHB_ENGINE_USE_STREAMING=...
 CHB_QUERY_TABLE_PREFIX=chatbot-local
+CHB_REDIS_INDEX_NAME=zero
+CHB_DYNAMODB_URL=http://locahost:8080

apps/chatbot/docker/app.local.Dockerfile

@@ -1,6 +1,10 @@
 FROM python:3.12.4-slim-bullseye
 ARG DEBIAN_FRONTEND=noninteractive
 
+RUN apt-get update && \
+ apt-get install -y \
+ curl
+
 ENV PYTHONPATH=/app
 
 RUN pip install --upgrade pip \
@@ -14,4 +18,4 @@ RUN poetry install
 
 COPY . .
 
-CMD ["fastapi", "dev", "src/app/main.py", "--port", "8080"]
+CMD ["fastapi", "dev", "src/app/main.py", "--port", "8080", "--host", "0.0.0.0"]

apps/chatbot/docker/compose.yaml

@@ -0,0 +1,56 @@
+services:
+ api:
+ build:
+ context: ..
+ dockerfile: docker/app.local.Dockerfile
+ ports:
+ - "8080:8080"
+ volumes:
+ - ..:/app
+ - ./files/.aws:/root/.aws
+ - ./build-devp/out:/app/build-devp/out
+ depends_on:
+ redis:
+ condition: service_started
+ dynamodb:
+ condition: service_started
+ networks:
+ - ntw
+
+ dynamodb:
+ image: amazon/dynamodb-local:2.5.2
+ environment:
+ - AWS_ACCESS_KEY_ID=dummy
+ - AWS_SECRET_ACCESS_KEY=dummy
+ - AWS_DEFAULT_REGION=local
+ ports:
+ - "8000:8000"
+ networks:
+ - ntw
+
+ redis:
+ image: redis/redis-stack:7.2.0-v13
+ ports:
+ - "6379:6379"
+ networks:
+ - ntw
+
+ create_index:
+ build:
+ context: ..
+ dockerfile: docker/app.local.Dockerfile
+ ports:
+ - "8080:8080"
+ volumes:
+ - ..:/app
+ - ./build-devp/out:/app/build-devp/out
+ command: "python src/modules/create_vector_index.py --params config/params.yaml"
+ tty: true
+ depends_on:
+ redis:
+ condition: service_started
+ networks:
+ - ntw
+
+networks:
+ ntw:

apps/chatbot/docker/docker-compose-up-api.sh

@@ -0,0 +1,2 @@
+#!/bin/bash
+docker compose -f docker/compose.yaml -p chatbot up api

apps/chatbot/docker/docker-run-create-index.sh

@@ -0,0 +1,2 @@
+#!/bin/bash
+docker compose -f docker/compose.yaml -p chatbot up create_index

apps/chatbot/docker/docker-run-local-bash.sh

@@ -0,0 +1,2 @@
+#!/bin/bash
+docker run -it --env-file ./.env fastapi-local bash

apps/chatbot/docker/files/.aws/config

@@ -0,0 +1,2 @@
+[profile default]
+region = eu-south-1

apps/chatbot/docker/files/.aws/credentials

@@ -0,0 +1,3 @@
+[default]
+aws_access_key_id = 123
+aws_secret_access_key = xyz

apps/chatbot/pyproject.toml

@@ -37,9 +37,10 @@ llama-index-llms-gemini = "^0.3.4"
 google-generativeai = "^0.5.2"
 llama-index-embeddings-gemini = "^0.2.0"
 llama-index-llms-bedrock-converse = "^0.3.0"
+chromedriver-py = "^129.0.6668.91"
 llama-index-postprocessor-presidio = "^0.2.0"
 
 
 [build-system]
 requires = ["poetry-core"]
-build-backend = "poetry.core.masonry.api"
+build-backend = "poetry.core.masonry.api"

apps/chatbot/src/app/main.py

@@ -18,30 +18,30 @@
 
 params = yaml.safe_load(open("config/params.yaml", "r"))
 prompts = yaml.safe_load(open("config/prompts.yaml", "r"))
-chatbot = Chatbot(params, prompts)
-
 AWS_DEFAULT_REGION = os.getenv('CHB_AWS_DEFAULT_REGION', os.getenv('AWS_DEFAULT_REGION', None))
 
+chatbot = Chatbot(params, prompts)
+
 
 class Query(BaseModel):
  question: str
  queriedAt: str | None = None
 
-if (os.getenv('environment', 'dev') == 'local'):
- profile_name='dummy'
- endpoint_url='http://localhost:8000'
- region_name = AWS_DEFAULT_REGION
-
 boto3_session = boto3.session.Session(
- profile_name = locals().get('profile_name', None),
- region_name=locals().get('region_name', None)
+ region_name=AWS_DEFAULT_REGION
 )
 
-dynamodb = boto3_session.resource( 
- 'dynamodb',
- endpoint_url=locals().get('endpoint_url', None),
- region_name=locals().get('region_name', None),
-)
+if (os.getenv('environment', 'dev') == 'local'):
+ dynamodb = boto3_session.resource( 
+ 'dynamodb',
+ endpoint_url=os.getenv('CHB_DYNAMODB_URL', 'http://localhost:8000'),
+ region_name=AWS_DEFAULT_REGION
+ )
+else:
+ dynamodb = boto3_session.resource( 
+ 'dynamodb',
+ region_name=AWS_DEFAULT_REGION
+ )
 
 table_queries = dynamodb.Table(
  f"{os.getenv('CHB_QUERY_TABLE_PREFIX', 'chatbot')}-queries"
@@ -160,12 +160,13 @@ async def sessions_fetching(
  raise HTTPException(status_code=422, detail=f"[sessions_fetching] userId: {userId}, error: {e}")
 
  # TODO: pagination
+ items = db_response.get('Items', [])
  result = {
- "items": db_response['Items'],
+ "items": items,
  "page": 1,
  "pages": 1,
- "size": len(db_response['Items']),
- "total": len(db_response['Items']),
+ "size": len(items),
+ "total": len(items),
  }
  return result
 
@@ -214,20 +215,26 @@ async def queries_fetching(
  sessionId = last_session_id(userId)
 
  try:
- # TODO: add userId filter
  db_response = table_queries.query(
- KeyConditionExpression=Key("sessionId").eq(sessionId)
+ KeyConditionExpression=Key("sessionId").eq(sessionId) &
+ Key("id").eq(userId)
  )
  except (BotoCoreError, ClientError) as e:
  raise HTTPException(status_code=422, detail=f"[queries_fetching] sessionId: {sessionId}, error: {e}")
 
- result = db_response['Items']
+ result = db_response.get('Items', [])
  return result
 
 
 def last_session_id(userId: str):
- # TODO: retrieve last user session
- return '1'
+ db_response = table_sessions.query(
+ IndexName='SessionsByCreatedAtIndex',
+ KeyConditionExpression=Key('userId').eq(userId),
+ ScanIndexForward=False,
+ Limit=1
+ )
+ items = db_response.get('Items', [])
+ return items[0] if items else None
 
 @app.patch("/queries/{id}")
 async def query_feedback (badAnswer: bool):

apps/chatbot/src/modules/utils.py

@@ -19,6 +19,7 @@ def get_ssm_parameter(name: str, default: str | None = None) -> str | None:
  :param default: The default value to return if the parameter is not found.
  :return: The value of the requested parameter.
  """
+
  ssm = boto3.client(
  "ssm",
  aws_access_key_id=AWS_ACCESS_KEY_ID,
@@ -37,4 +38,4 @@ def get_ssm_parameter(name: str, default: str | None = None) -> str | None:
  return default
 
  logging.debug(f"Parameter {name} retrieved from SSM")
- return response["Parameter"]["Value"]
+ return response["Parameter"]["Value"]

apps/chatbot/src/modules/vector_database.py

@@ -8,7 +8,10 @@
 import html2text
 from bs4 import BeautifulSoup
 from selenium import webdriver
+from selenium.webdriver.chrome.options import Options
+from selenium.webdriver.chrome.service import Service
 from typing import List, Tuple
+from chromedriver_py import binary_path
 
 from bs4 import BeautifulSoup
 from selenium import webdriver
@@ -39,7 +42,6 @@
 
 load_dotenv()
 
-
 PROVIDER = os.getenv("CHB_PROVIDER")
 assert PROVIDER in ["google", "aws"]
 
@@ -115,15 +117,18 @@ def filter_html_files(html_files: List[str]) -> List[str]:
  pattern = re.compile(r"/v\d{1,2}.")
  pattern2 = re.compile(r"/\d{1,2}.")
  filtered_files = [file for file in html_files if not pattern.search(file) and not pattern2.search(file)]
+ logging.info(f"[vector_database.py] filter_html_files len(filtered_files): {len(filtered_files)}")
  return filtered_files
 
 
 def get_html_files(root_folder: str) -> List[str]:
+ logging.info(f"[vector_database.py] get_html_files({root_folder})")
  html_files = []
  for root, _, files in os.walk(root_folder):
  for file in files:
  if file.endswith(".html"):
  html_files.append(os.path.join(root, file))
+ logging.info(f"[vector_database.py] get_html_files len(html_files): {len(html_files)}")
  return sorted(filter_html_files(html_files))
 
 
@@ -154,10 +159,14 @@ def create_documentation(
  if documentation_dir[-1] != "/":
  documentation_dir += "/"
 
- logging.info(f"Getting documentation from: {documentation_dir}")
+ logging.info(f"[vector_database.py] Getting documentation from: {documentation_dir}")
+ logging.info(f"[vector_database.py] create_documentation: DYNAMIC_HTML: {DYNAMIC_HTMLS}")
+ logging.info(f"[vector_database.py] create_documentation: documentation_dir: {documentation_dir}")
 
  html_files = get_html_files(documentation_dir)
+ logging.info(f"[vector_database.py] create_documentation: len(html_files): {len(html_files)}")
  dynamic_htmls = [os.path.join(documentation_dir, path) for path in DYNAMIC_HTMLS]
+ logging.info(f"[vector_database.py] create_documentation: len(dynamic_htmls): {len(dynamic_htmls)}")
  documents = []
  hash_table = {}
  empty_pages = []
@@ -166,17 +175,28 @@ def create_documentation(
 
  for file in tqdm.tqdm(html_files, total=len(html_files), desc="Extracting HTML"):
 
- if file in dynamic_htmls:
+# FIX: resolve webdriver.Chrome "self.assert_process_still_running" error in docker
+# if file in dynamic_htmls:
+ if 6 == 9:
  url = file.replace(documentation_dir, f"{website_url}/").replace(".html", "")
- driver = webdriver.Chrome()
+
+ # svc = webdriver.ChromeService(executable_path=binary_path)
+ service = Service(executable_path=binary_path)
+ options = webdriver.ChromeOptions()
+ options.add_argument('--headless=new')
+ options.add_argument('--no-sandbox')
+ options.add_argument('user-agent=fake-useragent')
+ driver = webdriver.Chrome(service=service, options=options)
+
+ logging.info(f"[vector_database.py] create_documentation: driver.get({url})")
  driver.get(url)
  time.sleep(5)
  title, text = html2markdown(driver.page_source)
  driver.quit()
  else:
  title, text = html2markdown(open(file))
 
- if text == None or text == "" or text == "None":
+ if text is None or text == "" or text == "None":
  # print(file)
  empty_pages.append(file)
 
@@ -220,7 +240,7 @@ def build_automerging_index_redis(
  chunk_overlap: int
  ) -> VectorStoreIndex:
 
- logging.info("Storing vector index and hash table on Redis..")
+ logging.info("[vector_database.py] Storing vector index and hash table on Redis..")
 
  Settings.llm = llm
  Settings.embed_model = embed_model
@@ -236,9 +256,9 @@ def build_automerging_index_redis(
  key=key,
  val=value
  )
- logging.info("Hash table is now on Redis.")
+ logging.info("[vector_database.py] Hash table is now on Redis.")
 
- logging.info("Creating index...")
+ logging.info("[vector_database.py] Creating index...")
  nodes = Settings.node_parser.get_nodes_from_documents(documents)
  leaf_nodes = get_leaf_nodes(nodes)
 
@@ -262,6 +282,7 @@ def build_automerging_index_redis(
  automerging_index.set_index_id(INDEX_ID)
  logging.info("Created vector index successfully and stored on Redis.")
 
+ automerging_index.set_index_id("1234")
  return automerging_index
 
 
@@ -285,7 +306,7 @@ def load_automerging_index_redis(
  schema=REDIS_SCHEMA
  )
 
- logging.info(f"Loading vector index from Redis...")
+ logging.info("[vector_database.py] Loading vector index from Redis...")
  storage_context = StorageContext.from_defaults(
  vector_store=redis_vector_store,
  docstore=REDIS_DOCSTORE,

apps/infrastructure/src/modules/chatbot/dynamodb.tf

@@ -41,5 +41,19 @@ module "dynamodb_chatbot_sessions" {
  name = "userId"
  type = "S"
  },
+ {
+ name = "createdAt"
+ type = "S"
+ },
+ ]
+
+ # LSI for query on created_at
+ local_secondary_indexes = [
+ {
+ name = "SessionsByCreatedAtIndex"
+ hash_key = "userId"
+ range_key = "createdAt"
+ projection_type = "ALL"
+ }
  ]
 }