Added chatbot UI support using Gradio

.gitignore

@@ -159,4 +159,7 @@ cython_debug/
 #  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
-#.idea/
+#.idea/
+
+# ignore .DS_Store file on macOS
+.DS_Store

README.md

@@ -66,6 +66,23 @@ To run this sample code follow these steps:
 2. **Run the script** - run the `python main.py`
 3. **Command line argument** - Optionally you can add `--input "new input here"` to overide the default user input text.
 
+### Chatbot UI 
+To run the chatbot interface, follow these steps:
+
+1. Start the chatbot interface by running the following command:
+   ```bash
+   python chatbot_interface.py
+   ```
+   You should see output similar to:
+   ```
+   Running on local URL:  http://127.0.0.1:7860
+
+   To create a public link, set `share=True` in `launch()`.
+    ```
+2. Open the provided local URL in your web browser to start chatting.
+
+3. The UI will look like the following: ![Chatbot UI](images/chatbot_ui.png)
+
 ### Example
 
 **Default prompt**: "What is the current stock price of amazon stock in pounds?

chatbot_interface.py

@@ -0,0 +1,107 @@
+import sys
+from loguru import logger
+from pkg.ask import generate_text
+from tools.ticker import parse_and_run_get_stock_price
+from tools.currency import parse_and_run_convert_currency
+import argparse
+import json
+import gradio as gr
+
+
+def parse_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "input",
+        type=str,
+        nargs="?",
+        help="Input text for the LLM",
+        default="What is the current stock price of amazon stock in pounds?",
+    )
+    parser.add_argument(
+        "--log-level",
+        type=str,
+        default="INFO",
+        choices=["DEBUG", "INFO", "WARNING", "ERROR", "CRITICAL"],
+        help="Set the logging level",
+    )
+    return parser.parse_known_args()
+
+
+def build_message(input_text: str) -> list[dict]:
+    """
+    Build the message to send to the LLM from the input test
+    :param input_text: the input text to send to the LLM
+    :return: a list containing containing the role, content and user input as text
+    """
+    return [{"role": "user", "content": [{"text": input_text}]}]
+
+
+def chatbot(input_text: str) -> str:
+    logger.info("Starting")
+    msg = build_message(input_text)
+    logger.info(f"input text: {input_text}")
+
+    stop_reason: str = None
+    answer: str = None
+
+    # Run until the model end_turn
+    while stop_reason != "end_turn":
+        stop_reason, tools_requested, messages = generate_text(msg)
+        logger.debug(f"stop reason is {stop_reason}, continue work till final answer")
+
+        # Amazon Bedrock LLM ended turn and responded the final answer
+        if stop_reason == "end_turn":
+            logger.info("The question asked the LLM ended turn and this is the answer")
+            answers = messages.get("content", {})
+            # iterate over the returned answers from Amazon Bedrock LLM
+            answers_text = [a.get("text", "\n") for a in answers]
+            answer = "".join(answers_text)
+            break
+
+        if stop_reason == "tool_use":
+            # find from the content returned form tools_requested the tool to use
+            for content in tools_requested:
+                if "toolUse" in content:
+                    tool_use_id = content.get("toolUse", {}).get("toolUseId")
+                    tool_use_name = content.get("toolUse", {}).get("name")
+                    tool_use_input = content.get("toolUse", {}).get("input")
+                    logger.info(
+                        f"tool use id is {tool_use_id}, tool use name is {tool_use_name}"
+                    )
+                    # stock price tool
+                    if tool_use_name == "get_stock_price":
+                        message = parse_and_run_get_stock_price(
+                            tool_use_id, tool_use_input
+                        )
+                        messages.append(message)
+
+                    # currency conversion tool
+                    if tool_use_name == "convert_currency":
+                        message = parse_and_run_convert_currency(
+                            tool_use_id, tool_use_input
+                        )
+                        messages.append(message)
+
+                # See the messages appended that are being built for the LLM, this will allow the Bedrock LLM to provide the final answer.
+                logger.debug(f"messages:\n{json.dumps(messages)}")
+
+        else:
+            # Stop reasons can be: 'end_turn'|'tool_use'|'max_tokens'|'stop_sequence'|'guardrail_intervened'|'content_filtered'
+            # This code sample only covers end_turn, and tool_use, you may need to implement additional code to cover all the rest of the responses.
+            logger.warning(
+                f"llm didn't end_turn, or asked to use a tool, he asked to {stop_reason}"
+            )
+            return "An error occurred."
+
+    # Printing the final response from the model
+    logger.info(answer)
+    return answer
+
+
+if __name__ == "__main__":
+    args, _ = parse_args()
+    logger.remove()  # Remove the default logger
+    logger.add(sys.stderr, level=args.log_level.upper())
+
+    iface = gr.Interface(fn=chatbot, inputs="text", outputs="text", title="Chatbot")
+    iface.launch()

requirements.txt

@@ -1,4 +1,5 @@
 boto3>=1.34.141
 loguru>=0.7.2
 yfinance==0.2.40
-currencyconverter==0.17.25
+currencyconverter==0.17.25
+gradio==4.44.1