Update llama_index (#21)

* Update llama_index

* Fix trigger

* Update pipeline stuff

* Rebuild index and transparent model settings

* Compile requirements

* Use compiled reqs

* Modify dockerfile

.github/workflows/build_and_deploy.yaml

@@ -4,15 +4,15 @@ on:
   push:
     branches:
       - main
-    paths:
-      - "!jupyter/**"
-      - "!ui/**"
+    paths-ignore:
+      - "jupyter/**"
+      - "ui/**"
   pull_request:
     branches:
       - main
-    paths:
-      - "!jupyter/**"
-      - "!ui/**"
+    paths-ignore:
+      - "jupyter/**"
+      - "ui/**"
 
 env:
   GCLOUD_PROJECT_ID: ${{ secrets.GCLOUD_PROJECT_ID }}
@@ -26,44 +26,37 @@ jobs:
   build:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Set up Python
-        uses: actions/setup-python@v4
+        uses: actions/setup-python@v5
         with:
-          python-version: "3.10"
-      - uses: actions/cache@v3
+          python-version: "3.11"
+      - uses: actions/cache@v4
         with:
           path: ~/.cache/pip
           key: ${{ hashFiles('requirements.txt') }}
       - name: Run tests
         run: |
           python -m pip install --upgrade pip
           pip install ruff pytest
-          if [ -f requirements.txt ]; then pip install -r requirements.txt; fi
+          if [ -f requirements-all.txt ]; then pip install -r requirements-all.txt; fi
           pytest --junitxml=test_results.xml
         env:
           OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
       - id: "auth"
-        uses: "google-github-actions/auth@v1"
+        uses: "google-github-actions/auth@v2"
         with:
           credentials_json: "${{ secrets.GCLOUD_SA_KEY }}"
       - name: Setup Google Cloud CLI
         if: ${{ (github.ref == 'refs/heads/main') || (startsWith(github.ref, 'refs/tags/')) }}
-        uses: google-github-actions/setup-gcloud@v1
+        uses: google-github-actions/setup-gcloud@v2
         with:
-          version: ">= 363.0.0"
+          version: "latest"
       - name: Docker Login
-        if: ${{ (github.ref == 'refs/heads/main') || (startsWith(github.ref, 'refs/tags/')) }}
         run: |
           gcloud auth configure-docker europe-west3-docker.pkg.dev
-      - name: Build for verification
-        if: github.ref != 'refs/heads/main'
-        env:
-          GH_DEPLOY_USERNAME: ${{ env.GITHUB_REPOSITORY_OWNER }}
-          GH_DEPLOY_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: echo "implement me"
       - name: Build And Push Docker Image
-        uses: docker/build-push-action@v4
+        uses: docker/build-push-action@v5
         with:
           context: .
           push: ${{ (github.ref == 'refs/heads/main') || (startsWith(github.ref, 'refs/tags/')) }}
@@ -85,12 +78,6 @@ jobs:
           echo "Invoke endpoint:"
           gcloud run services list --platform managed --format json | jq -r \
             '.[] | select(.metadata.name == "google-cloud-run-maven-test") | .status.address.url'
-      - name: Release
-        uses: fnkr/github-action-ghr@v1
-        if: startsWith(github.ref, 'refs/tags/')
-        env:
-          GHR_PATH: build/distributions
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
       - name: Publish Test Results
         uses: EnricoMi/publish-unit-test-result-action@v2
         if: always()

Dockerfile

@@ -1,14 +1,14 @@
-FROM python:3.10-slim-bullseye
+FROM python:3.11-slim-bookworm
 
 LABEL org.opencontainers.image.source = "https://github.com/deepshore/knowledge-chatbot"
 
 WORKDIR /app
 
-COPY requirements.txt /app/
+COPY requirements-all.txt /app/
 COPY app/main.py /app/
 COPY storage /app/storage
 
 RUN python -m pip install --upgrade pip && \
-    pip install -r requirements.txt
+    pip install -r requirements-all.txt
 
 CMD uvicorn main:app --port 8080 --host 0.0.0.0

app/main.py

@@ -1,16 +1,24 @@
-from contextlib import asynccontextmanager
-from fastapi import FastAPI
+from fastapi import FastAPI, status
 import os
 import logging
 import sys
-from llama_index import StorageContext, load_index_from_storage
+from llama_index import StorageContext, ServiceContext, load_index_from_storage, set_global_service_context, VectorStoreIndex
+from llama_index.llms import OpenAI
+from llama_index.embeddings import OpenAIEmbedding
+from llama_index.embeddings.openai import OpenAIEmbeddingMode, OpenAIEmbeddingModelType
 from pydantic import BaseModel
 from typing import Union
-from llama_index.response.schema import Response
+from llama_hub.web.sitemap.base import SitemapReader
 
 logging.basicConfig(stream=sys.stdout, level=logging.INFO)
 logging.getLogger().addHandler(logging.StreamHandler(stream=sys.stdout))
 
+llm = OpenAI(model="gpt-3.5-turbo")
+embed_model = OpenAIEmbedding(mode=OpenAIEmbeddingMode.TEXT_SEARCH_MODE, model=OpenAIEmbeddingModelType.TEXT_EMBED_ADA_002)
+service_context = ServiceContext.from_defaults(llm=llm,
+                                               embed_model=embed_model)
+set_global_service_context(service_context=service_context)
+
 class DeepshoreChatRequest(BaseModel):
     question: str
     timestamp: Union[int, None] = None
@@ -25,26 +33,42 @@ class DeepshoreChatResponse(BaseModel):
 # todo: put this into lifespan startup and add async tests
 index = None
 
-persist_dir = './storage' if (os.environ.get('PERSIST_DIR') == None) else os.environ.get('PERSIST_DIR')
-storage_context = StorageContext.from_defaults(persist_dir=persist_dir)
-# load index
-index = load_index_from_storage(storage_context)
-print("PEEN")
+def refresh_index():
+    global index
+    print("refresh_index")
+    loader = SitemapReader(html_to_text=True)
+    documents = loader.load_data(sitemap_url='https://deepshore.de/sitemap.xml', filter='https://deepshore.de/knowledge')
+    index = VectorStoreIndex.from_documents(documents)
+    index.storage_context.persist()
+
+try:
+    persist_dir = './storage' if (os.environ.get('PERSIST_DIR') == None) else os.environ.get('PERSIST_DIR')
+    storage_context = StorageContext.from_defaults(persist_dir=persist_dir)
+    # load index
+    index = load_index_from_storage(storage_context)
+except Exception as err:
+    print(err)
+    refresh_index()
+
 if not index:
     raise Exception("Sorry, no numbers below zero")
 
 # app start
-
 app = FastAPI()
-
 app.name = "Deepshore Chatbot API"
 
 @app.get("/")
 async def read_root():
     return {"message": "Deepshore Chatbot API"}
 
+@app.put("/refresh-index", status_code=status.HTTP_204_NO_CONTENT)
+async def get_refresh_index():
+    global index
+    refresh_index()
+
 @app.post("/chatbot")
 async def ask_chatbot(request: DeepshoreChatRequest):
+    global index
     try:
         query_engine = index.as_query_engine()
         response = query_engine.query(request.question)

app/test_main.py

@@ -2,6 +2,7 @@
 from .main import app
 from .main import DeepshoreChatRequest
 import time
+import pytest
 
 client = TestClient(app)
 
@@ -10,6 +11,13 @@ def test_read_root():
     assert response.status_code == 200
     assert response.json() == {"message": "Deepshore Chatbot API"}
 
+@pytest.mark.skip(reason="this should be tested individually beacuse it really scrapes the site and generates the index")
+def test_get_refresh_index():
+    import nest_asyncio
+    nest_asyncio.apply()
+    response = client.put("/refresh-index")
+    assert response.status_code == 204
+
 def test_post_chatbot():
     ts = int(time.time())
     req = {"question": "Was ist k6.io?", "timestamp": ts}

docker-compose.yml

@@ -0,0 +1,7 @@
+version: '3'
+services:
+  app:
+    build: .
+    ports:
+      - "8080:8080"
+    env_file: .env