small changes

.env-example

@@ -1,18 +1,10 @@
-JUPYTER_PORT=8181
+APP_PORT=8181
 
 LANGCHAIN_ENDPOINT="https://api.smith.langchain.com"
-LANGCHAIN_API_KEY=...
+LANGCHAIN_API_KEY=ls__...
 LANGCHAIN_TRACING_V2="true"
-LANGCHAIN_PROJECT="AI_GIS_language"
+LANGCHAIN_PROJECT="streamlit_chatbot"
 
-ANYSCALE_API_KEY=...
+ANYSCALE_API_KEY=secret_...
 OPENAI_API_KEY=sk-...
-ANTHROPIC_API_KEY=...
-
-#https://github.com/streamlit/app-starter-kit
-#https://github.com/langchain-ai/streamlit-agent
-#https://github.com/hwchase17/langchain-streamlit-template
-#https://github.com/dataprofessor/langchain-ask-the-doc
-#https://github.com/dataprofessor/langchain-ask-the-data
-#https://github.com/dataprofessor/langchain-blog-outline-generator
-#https://github.com/dataprofessor/langchain-text-summarization
+ANTHROPIC_API_KEY="sk-..."

.github/pull_request_template.md

@@ -1,4 +1,4 @@
-Thank you for contributing to `AI_chatbot`.
+Thank you for contributing!
 Before submitting this PR, please make sure:
 
 - [ ] Your code builds clean without any errors or warnings

.idea/.name

@@ -0,0 +1 @@
+langchain-streamlit-demo

.idea/{AI_chatbot.iml → langchain-streamlit-demo.iml}

@@ -2,7 +2,7 @@
 <module type="PYTHON_MODULE" version="4">
   <component name="NewModuleRootManager">
     <content url="file://$MODULE_DIR$" />
-    <orderEntry type="inheritedJdk" />
+    <orderEntry type="jdk" jdkName="Remote Python 3.11.4 Docker (&lt;none&gt;:&lt;none&gt;) (5)" jdkType="Python SDK" />
     <orderEntry type="sourceFolder" forTests="false" />
   </component>
 </module>

.idea/misc.xml

@@ -1,4 +1,4 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <project version="4">
-  <component name="ProjectRootManager" version="2" project-jdk-name="arcgis-pc (2)" project-jdk-type="Python SDK" />
+  <component name="ProjectRootManager" version="2" project-jdk-name="Remote Python 3.11.4 Docker (&lt;none&gt;:&lt;none&gt;) (5)" project-jdk-type="Python SDK" />
 </project>

docker-compose.yml

@@ -9,6 +9,12 @@ services:
     ports:
       - "8000:8000"
     volumes:
-        - .:/home/appuser/app:rw
+      - .:/home/appuser/app:rw
     working_dir: /home/appuser/app/
-    entrypoint: ["python", "-m", "streamlit", "run", "${APP}", "--server.port", "8000", "--server.enableCORS", "false", "--server.address", "0.0.0.0"]
+    command: [
+      "python", "-m",
+      "streamlit", "run",
+      "/home/appuser/app/langchain-streamlit-demo/app.py",
+      "--server.port", "8000",
+      "--server.address", "0.0.0.0"
+    ]

{AI_chatbot → langchain-streamlit-demo}/app.py

@@ -1,42 +1,23 @@
-from datetime import datetime
-
 import streamlit as st
-from langchain import LLMChain
-from langchain.callbacks.base import BaseCallbackHandler
 from langchain.callbacks.tracers.langchain import wait_for_all_tracers
 from langchain.callbacks.tracers.run_collector import RunCollectorCallbackHandler
-from langchain.chat_models import ChatOpenAI
-from langchain.memory import StreamlitChatMessageHistory, ConversationBufferMemory
-from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
 from langchain.schema.runnable import RunnableConfig
-from langsmith import Client
-from streamlit_feedback import streamlit_feedback
+
+
+from llm_stuff import (
+    _DEFAULT_SYSTEM_PROMPT,
+    get_memory,
+    get_llm_chain,
+    StreamHandler,
+    feedback_component,
+    get_langsmith_client,
+)
 
 st.set_page_config(
     page_title="Chat LangSmith",
     page_icon="🦜",
 )
 
-
-def get_llm_chain(system_prompt: str, memory: ConversationBufferMemory) -> LLMChain:
-    """Return a basic LLMChain with memory."""
-    prompt = ChatPromptTemplate.from_messages(
-        [
-            (
-                "system",
-                system_prompt + "\nIt's currently {time}.",
-            ),
-            MessagesPlaceholder(variable_name="chat_history"),
-            ("human", "{input}"),
-        ],
-    ).partial(time=lambda: str(datetime.now()))
-    llm = ChatOpenAI(temperature=0.7, streaming=True)
-    return LLMChain(prompt=prompt, llm=llm, memory=memory)
-
-
-client = Client()
-
-
 # "# Chat🦜🛠️"
 # Initialize State
 if "trace_link" not in st.session_state:
@@ -49,21 +30,25 @@ st.sidebar.markdown(
 """,
 )
 
-_DEFAULT_SYSTEM_PROMPT = "You are a helpful chatbot."
-
-system_prompt = st.sidebar.text_area(
-    "Custom Instructions",
-    _DEFAULT_SYSTEM_PROMPT,
-    help="Custom instructions to provide the language model to determine style, personality, etc.",
-)
-system_prompt = system_prompt.strip().replace("{", "{{").replace("}", "}}")
-memory = ConversationBufferMemory(
-    chat_memory=StreamlitChatMessageHistory(key="langchain_messages"),
-    return_messages=True,
-    memory_key="chat_history",
+system_prompt = (
+    st.sidebar.text_area(
+        "Custom Instructions",
+        _DEFAULT_SYSTEM_PROMPT,
+        help="Custom instructions to provide the language model to determine style, personality, etc.",
+    )
+    .strip()
+    .replace("{", "{{")
+    .replace("}", "}}")
 )
 
-chain = get_llm_chain(system_prompt, memory)
+memory = get_memory()
+
+chain = get_llm_chain(memory, system_prompt)
+
+client = get_langsmith_client()
+
+run_collector = RunCollectorCallbackHandler()
+
 
 if st.sidebar.button("Clear message history"):
     print("Clearing message history")
@@ -96,19 +81,6 @@ if st.session_state.trace_link:
     )
 
 
-class StreamHandler(BaseCallbackHandler):
-    def __init__(self, container, initial_text=""):
-        self.container = container
-        self.text = initial_text
-
-    def on_llm_new_token(self, token: str, **kwargs) -> None:
-        self.text += token
-        self.container.markdown(self.text)
-
-
-run_collector = RunCollectorCallbackHandler()
-
-
 def _reset_feedback():
     st.session_state.feedback_update = None
     st.session_state.feedback = None
@@ -135,59 +107,6 @@ if prompt := st.chat_input(placeholder="Ask me a question!"):
         url = client.read_run(run.id).url
         st.session_state.trace_link = url
 
-# Simple feedback section
-# Optionally add a thumbs up/down button for feedback
+
 if st.session_state.get("run_id"):
-    # feedback = streamlit_feedback(
-    #     feedback_type="thumbs",
-    #     key=f"feedback_{st.session_state.run_id}",
-    # )
-    # scores = {"👍": 1, "👎": 0}
-    scores = {"😀": 1, "🙂": 0.75, "😐": 0.5, "🙁": 0.25, "😞": 0}
-    feedback = streamlit_feedback(
-        feedback_type="faces",
-        optional_text_label="[Optional] Please provide an explanation",
-        key=f"feedback_{st.session_state.run_id}",
-    )
-    if feedback:
-        score = scores[feedback["score"]]
-        feedback = client.create_feedback(
-            st.session_state.run_id,
-            feedback["type"],
-            score=score,
-            comment=feedback.get("text", None),
-        )
-        st.session_state.feedback = {"feedback_id": str(feedback.id), "score": score}
-        st.toast("Feedback recorded!", icon="📝")
-
-
-# # Prompt for more information, if feedback was submitted
-# if st.session_state.get("feedback"):
-#     feedback = st.session_state.get("feedback")
-#     feedback_id = feedback["feedback_id"]
-#     score = feedback["score"]
-#     if score == 0:
-#         if correction := st.text_input(
-#             label="What would the correct or preferred response have been?",
-#             key=f"correction_{feedback_id}",
-#         ):
-#             st.session_state.feedback_update = {
-#                 "correction": {"desired": correction},
-#                 "feedback_id": feedback_id,
-#             }
-#     elif score == 1:
-#         if comment := st.text_input(
-#             label="Anything else you'd like to add about this response?",
-#             key=f"comment_{feedback_id}",
-#         ):
-#             st.session_state.feedback_update = {
-#                 "comment": comment,
-#                 "feedback_id": feedback_id,
-#             }
-# # Update the feedback if additional information was provided
-# if st.session_state.get("feedback_update"):
-#     feedback_update = st.session_state.get("feedback_update")
-#     feedback_id = feedback_update.pop("feedback_id")
-#     client.update_feedback(feedback_id, **feedback_update)
-#     # Clear the comment or correction box
-#     _reset_feedback()
+    feedback_component(client)

langchain-streamlit-demo/llm_stuff.py

@@ -0,0 +1,71 @@
+from datetime import datetime
+
+import streamlit as st
+from langchain import LLMChain
+from langchain.callbacks.base import BaseCallbackHandler
+from langchain.chat_models import ChatOpenAI
+from langchain.memory import ConversationBufferMemory, StreamlitChatMessageHistory
+from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
+from streamlit_feedback import streamlit_feedback
+from langsmith.client import Client
+
+_DEFAULT_SYSTEM_PROMPT = "You are a helpful chatbot."
+
+
+def get_langsmith_client():
+    return Client()
+
+
+def get_memory() -> ConversationBufferMemory:
+    return ConversationBufferMemory(
+        chat_memory=StreamlitChatMessageHistory(key="langchain_messages"),
+        return_messages=True,
+        memory_key="chat_history",
+    )
+
+
+def get_llm_chain(
+    memory: ConversationBufferMemory,
+    system_prompt: str = _DEFAULT_SYSTEM_PROMPT,
+) -> LLMChain:
+    """Return a basic LLMChain with memory."""
+    prompt = ChatPromptTemplate.from_messages(
+        [
+            (
+                "system",
+                system_prompt + "\nIt's currently {time}.",
+            ),
+            MessagesPlaceholder(variable_name="chat_history"),
+            ("human", "{input}"),
+        ],
+    ).partial(time=lambda: str(datetime.now()))
+    llm = ChatOpenAI(temperature=0.7, streaming=True)
+    return LLMChain(prompt=prompt, llm=llm, memory=memory or get_memory())
+
+
+class StreamHandler(BaseCallbackHandler):
+    def __init__(self, container, initial_text=""):
+        self.container = container
+        self.text = initial_text
+
+    def on_llm_new_token(self, token: str, **kwargs) -> None:
+        self.text += token
+        self.container.markdown(self.text)
+
+
+def feedback_component(client):
+    scores = {"😀": 1, "🙂": 0.75, "😐": 0.5, "🙁": 0.25, "😞": 0}
+    if feedback := streamlit_feedback(
+        feedback_type="faces",
+        optional_text_label="[Optional] Please provide an explanation",
+        key=f"feedback_{st.session_state.run_id}",
+    ):
+        score = scores[feedback["score"]]
+        feedback = client.create_feedback(
+            st.session_state.run_id,
+            feedback["type"],
+            score=score,
+            comment=feedback.get("text", None),
+        )
+        st.session_state.feedback = {"feedback_id": str(feedback.id), "score": score}
+        st.toast("Feedback recorded!", icon="📝")