new files

Dockerfile

@@ -0,0 +1,23 @@
+FROM python:3.11-slim
+
+# Set working directory (use a writable folder)
+WORKDIR /app
+
+# Copy and install dependencies
+COPY ./requirements.txt /app/requirements.txt
+RUN pip install --no-cache-dir --upgrade -r /app/requirements.txt
+
+# Copy source code
+COPY . /app
+
+# Ensure run.sh is executable
+RUN chmod +x /app/run.sh
+
+# Set environment variable for NLTK data
+ENV NLTK_DATA=/tmp/nltk_data
+
+# Expose the port for Chainlit
+EXPOSE 8000
+
+# Run the app
+CMD ["/app/run.sh"]

api.py

@@ -0,0 +1,31 @@
+from fastapi import FastAPI, Form
+import uvicorn
+
+from model import LogisticRegressionModel
+from helper import get_llm, classification_modeL_cache, llm_model_cache, prompt
+
+app = FastAPI()
+
+@app.get("/")
+async def root():
+    return {"message": "Sentiment Analysis API is running."}
+
+@app.post("/chat", response_model=str)
+async def chat_endpoint(message: str = Form(...)):
+    if "model" not in classification_modeL_cache:
+        classification_modeL_cache["model"] = LogisticRegressionModel()
+    
+    if "llm" not in llm_model_cache:
+        llm_model_cache["llm"] = get_llm()
+
+    prediction = classification_modeL_cache["model"].predict(message)
+    sentiment = "Positive" if prediction[0][1] > 0.5 else "Negative"
+
+    result = llm_model_cache["llm"].invoke(
+        prompt.format(text=message, prediction=sentiment))
+
+    if result:
+        return result.content
+
+if __name__ == "__main__":
+    uvicorn.run(app, host="127.0.0.1", port=7861)

download_data.py

@@ -0,0 +1,13 @@
+import nltk
+import os
+
+# Create a folder for NLTK data inside your repo
+NLTK_DATA_DIR = "/tmp/nltk_data"
+os.makedirs(NLTK_DATA_DIR, exist_ok=True)
+
+# Download datasets to that folder
+nltk.download("twitter_samples", download_dir=NLTK_DATA_DIR)
+nltk.download("stopwords", download_dir=NLTK_DATA_DIR)
+
+# Append the path so NLTK can find the datasets
+nltk.data.path.append(NLTK_DATA_DIR)

helper.py

@@ -0,0 +1,49 @@
+from dotenv import load_dotenv
+load_dotenv()
+import os
+from langchain.prompts import PromptTemplate
+from langchain_google_genai import ChatGoogleGenerativeAI
+
+classification_modeL_cache = {}
+
+llm_model_cache = {}
+
+def get_llm(model_name="gemini-2.5-flash-lite"): 
+    return ChatGoogleGenerativeAI(
+        model=model_name,
+        temperature=0.6,
+        top_p=0.8,
+        top_k=1,
+        max_tokens=None,
+        max_retries=3,
+        google_api_key=os.getenv("GOOGLE_API_KEY"),
+    )
+
+prompt = PromptTemplate(
+    input_variables=["text", "prediction"],
+    template="""
+    You are Sen, a sentiment analysis assistant.
+
+    TEXT:
+    {text}
+
+    PREDICTION:
+    {prediction}
+
+    Given the above text and sentiment prediction, please analyze the user's content.
+    """)
+
+eval_prompt = PromptTemplate(
+    input_variables=["text"],
+    template="""
+    You are agent who analyze the sentiment in Twitter's tweets.
+    LIST OF TWEETS:
+    {text}
+
+    INSTRUCTIONS
+    For each tweet in LIST OF TWEETS, analyze the sentiment only without any explanation, in the format: "positive" or "negative"
+
+    EXAMPLE:
+    ["positive", "negative", "positive", "negative", "positive"]
+    """
+)

requirements.txt

@@ -0,0 +1,11 @@
+pandas
+numpy
+scikit-learn
+nltk
+langchain
+langchain-core
+langchain-google-genai
+google-api-python-client
+fastapi
+uvicorn
+python-dotenv

run.sh

@@ -0,0 +1,2 @@
+#!/bin/bash
+chainlit run chainlit.py --host 0.0.0.0 --port 8000 --headless

sk_logreg.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c0a9258da8c61dc165d827a2037cc689d4bbc4b135b586eb8c628c3c8e2ba24
+size 863