simplify variable setup, add graph rag step (#24)

* add graph rag step

* add setup script

* tide up notes

* add image

Author: qkfang

docs/01-Lab-Setup.mdx

@@ -117,6 +117,7 @@ https://azure.microsoft.com/en-au/pricing/member-offers/credit-for-visual-studio
 | AZURE_OPENAI_API_DEPLOYMENT_NAME            | completions                                       |
 | AZURE_OPENAI_API_EMBEDDINGS_DEPLOYMENT_NAME | embeddings                                        |
 | AZURE_OPENAI_API_VERSION                    | 2023-09-01-preview                                |
+| GRAPH_RAG_API                               | https://arg-syd-aiapp1day-ca--miqityv.niceisland-66754352.eastus.azurecontainerapps.io |
 
 ### Azure Cosmos DB for MongoDB
 
@@ -141,34 +142,58 @@ Let us test the connection to Cosmos DB instance and OpenAI service.
 
 1.  Open VS Code and navigate to the `aiapp1day` repository folder. If you are using Codespaces, you should already be in the repository folder.
 
-2.  Switch to VS Code `Search` tab, replace `<MONGODB_Name>` placeholder with your own database name. Search for `<MONGODB_Name>` and replace with `aiapp1day_xxxxx_xx` (for example: aiapp1day_daniel_66). Then click `Replace All` icon, there will be multiple items to be replaced. The placeholder string will be replaced in the whole repo so that you have a dedicated database on the shared Cosmos DB instance.
+2.  Navigate to the setup folder `~/labs/00-Setup/` within the repository. Go to `Terminal` window in VS Code.
+    ```bash
+    cd labs/00-Setup/
+    ```
 
-    :::info
-    Make sure the `<` and `>` included in search & replace.
+3.  Open `setup.sh`, let's update `<MONGODB_Name>` setting with your own database name. Replace `add_value` with `aiapp1day_xxxxx_xx` (for example: aiapp1day_daniel_66).
+
+    ![edit setting](images/edit-settings.png)
 
+    :::info
     if you are deploying your own Azure resources, please update above shared keys and connection string with your own.
     :::
 
-    ![alt text](images/setup.png)
+4.  Update `<MONGODB_CONNECTION_STRING>` with `mongodb+srv://aiapp1dayadmin:Aiapp1daypassword123@arg-syd-aiapp1day-mongo.mongocluster.cosmos.azure.com/?tls=true&authMechanism=SCRAM-SHA-256&retrywrites=false&maxIdleTimeMS=120000`
+
+5.  Update `<AZURE_OPENAI_API_INSTANCE_NAME>` with `arg-syd-aiapp1day-openai`
+
+6.  Update `<AZURE_OPENAI_API_KEY>` with `0f73b2e1cba543ce8c9518712a5b1efc`
 
-3.  Replace `<MONGODB_CONNECTION_STRING>` with `mongodb+srv://aiapp1dayadmin:Aiapp1daypassword123@arg-syd-aiapp1day-mongo.mongocluster.cosmos.azure.com/?tls=true&authMechanism=SCRAM-SHA-256&retrywrites=false&maxIdleTimeMS=120000`
+7.  Update `<GRAPH_RAG_API>` with `https://arg-syd-aiapp1day-ca--miqityv.niceisland-66754352.eastus.azurecontainerapps.io`
 
-4.  Replace `<AZURE_OPENAI_API_INSTANCE_NAME>` with `arg-syd-aiapp1day-openai`
+8.  Go to terminal windows and run below bash command to replace all variables in the repo.
 
-5.  Replace `<AZURE_OPENAI_API_KEY>` with `0f73b2e1cba543ce8c9518712a5b1efc`
+    ```bash
+    chmod +x setup.sh
+    ./setup.sh
+    ```
+
+    :::info
+    If you have issue running above script, you can also carry out variable replacement manually. 
+    
+    Switch to VS Code `Search` tab, replace `<MONGODB_Name>` placeholder with your own database name. 
+    
+    Search for `<MONGODB_Name>` and replace with `aiapp1day_xxxxx_xx` (for example: aiapp1day_daniel_66). 
+    
+    Then click `Replace All` icon, there will be multiple items to be replaced. The placeholder string will be replaced in the whole repo so that you have a dedicated database on the shared Cosmos DB instance.
+
+    ![alt text](images/setup.png)
+    :::
 
-6.  Navigate to the setup test folder `~/labs/00-Setup/` within the repository.
+9.  Install node packages for test program.
 
     ```bash
     npm install
     ```
 
-7.  Test Connection to CosMos Db and OpenAI service. Run the following command in the terminal window to execute the script:
+10.  Test Connection to CosMos Db and OpenAI service. Run the following command in the terminal window to execute the script:
 
     ```bash
     node test.js
     ```
 
-8.  If you see below message in the console, you are ready to go! otherwise, please `YELL for help`.
+11.  If you see below message in the console, you are ready to go! otherwise, please `YELL for help`.
 
     ![alt text](images/setup-test.png)

docs/02-Part-2-Building-Chatbot/4-Using-Langchain.md

@@ -343,9 +343,27 @@ In this scenario, an agent will be equipped with two tools, one that uses a retr
        },
      });
 
+    const productFeedbackTool = new DynamicTool({
+        name: "product_feedback_lookup_tool",
+        description: `Searches Contoso Bike Store feedback based on the question.
+                    Returns the product name, sku and feedback in text format.
+                    If the product is not found, returns null.`,
+        func: async (input) => {
+            const url = `${graphRagAPI}/query/global`;
+            const requestBody = {
+                index_name: "bike",
+                query: `whats the top 2 products based on:  ${input}?`,
+                community_level: 1
+            };
+            
+            const response = await axios.post(url, requestBody, {});
+            return response.data.result;
+        },
+    });
+
      // Generate OpenAI function metadata to provide to the LLM
      // The LLM will use this metadata to decide which tool to use based on the description.
-     const tools = [productsRetrieverTool, productLookupTool];
+     const tools = [productsRetrieverTool, productLookupTool, productFeedbackTool];
      const modelWithFunctions = chatModel.bind({
        functions: tools.map((tool) => convertToOpenAIFunction(tool)),
      });
@@ -440,7 +458,7 @@ In this scenario, an agent will be equipped with two tools, one that uses a retr
 Please think about the difference between `langchain-rag.js` and `langchain-agent.js`. Which one do you think is better?
 :::
 
-12. Experiment with additional questions of your own.
+12. Experiment with additional questions of your own. For example, change the question in the `executeAgent` function to `What's the top 2 Tire based on feedback?`. It will use `productFeedbackTool` to invoike API call to Graph RAG endpoint.
 
 ## Summary
 

docs/images/edit-settings.png

@@ -0,0 +1,37 @@
+#!/bin/bash
+# chmod +x setup.sh
+# ./setup.sh
+
+# Define the search and replace pairs
+declare -A replacements=(
+    ["<MONGODB_Name>"]="add_value"
+    ["<MONGODB_CONNECTION_STRING>"]="add_value"
+    ["<AZURE_OPENAI_API_INSTANCE_NAME>"]="add_value"
+    ["<AZURE_OPENAI_API_KEY>"]="add_value"
+    ["<GRAPH_RAG_API>"]="add_value"
+    # see examples below
+    # ["<MONGODB_Name>"]="aiapp1day_daniel_66"
+    # ["<MONGODB_CONNECTION_STRING>"]="mongodb+srv://aiapp1dayadmin:Aiapp1daypassword123@arg-syd-aiapp1day-mongo.mongocluster.cosmos.azure.com/?tls=true&authMechanism=SCRAM-SHA-256&retrywrites=false&maxIdleTimeMS=120000"
+    # ["<AZURE_OPENAI_API_INSTANCE_NAME>"]="arg-syd-aiapp1day-openai"
+    # ["<AZURE_OPENAI_API_KEY>"]="0f73b2e1cba543ce8c9518712a5b1efc"
+    # ["<GRAPH_RAG_API>"]="https://arg-syd-aiapp1day-ca--miqityv.niceisland-66754352.eastus.azurecontainerapps.io"
+)
+
+# Check if parent directory exists
+if [ ! -d "../" ]; then
+    echo "Parent directory not found!"
+    exit 1
+fi
+
+echo "Searching for .env and .js files..."
+
+# Find .env and .js files in parent directory
+find /workspaces/aiapp1day/ -type f \( -name "*.env" -o -name "*.js" \) -not -path "*/node_modules/*" | while read -r file; do
+    echo "Processing: $file"
+    for search in "${!replacements[@]}"; do
+        replace="${replacements[$search]}"
+        sed -i "s|${search}|${replace}|g" "$file"
+    done
+done
+
+echo "Replacements complete!"