Create app.py

app.py

@@ -0,0 +1,125 @@
+import os
+import streamlit as st
+import PyPDF2
+import matplotlib.pyplot as plt
+from io import BytesIO
+from llama_index.core import Settings, VectorStoreIndex, SimpleDirectoryReader
+from llama_index.embeddings.fastembed import FastEmbedEmbedding
+from llama_index.llms.gemini import Gemini
+import re
+from crewai import Agent, Task, Crew, Process
+import json
+
+# Configure Google Gemini
+Settings.embed_model = FastEmbedEmbedding(model_name="BAAI/bge-small-en-v1.5")
+Settings.llm = Gemini(api_key=os.getenv("GOOGLE_API_KEY"), temperature=0.5, model_name="models/gemini-pro")
+
+class FinAnalyst:
+    def __init__(self):
+        self.configure_agents()
+
+    def configure_agents(self):
+        self.document_processor = Agent(
+            role='Document Processor',
+            goal='Process and extract text from financial documents',
+            backstory='Expert in handling various document formats and extracting relevant information',
+            allow_delegation=False
+        )
+
+        self.data_extractor = Agent(
+            role='Data Extractor',
+            goal='Extract key financial data from processed documents',
+            backstory='Specialist in identifying and parsing financial information from text',
+            allow_delegation=False
+        )
+
+        self.financial_analyst = Agent(
+            role='Financial Analyst',
+            goal='Analyze financial data and provide insightful summaries',
+            backstory='Experienced financial expert with deep knowledge of Fortune 500 companies',
+            allow_delegation=False
+        )
+
+        self.data_visualizer = Agent(
+            role='Data Visualizer',
+            goal='Create visual representations of financial data',
+            backstory='Expert in data visualization techniques and financial charting',
+            allow_delegation=False
+        )
+
+    def write_to_file(self, content, filename="./files/uploaded.pdf"):
+        os.makedirs(os.path.dirname(filename), exist_ok=True)
+        with open(filename, "wb") as f:
+            f.write(content)
+
+    def process_document(self, file_content):
+        task = Task(
+            description="Process the uploaded financial document and extract its text content",
+            agent=self.document_processor
+        )
+        return task.execute(file_content)
+
+    def extract_financial_data(self, document_text):
+        task = Task(
+            description="Extract key financial data from the document text. Focus on revenue figures and corresponding dates. Return the data as a JSON string with 'Revenue' and 'Date' lists.",
+            agent=self.data_extractor
+        )
+        return task.execute(document_text)
+
+    def analyze_financials(self, financial_data, query):
+        task = Task(
+            description=f"Analyze the financial data and answer the query: {query}. Provide a comprehensive analysis covering revenue trends, key metrics, major events, period comparisons, future outlook, and potential risks/opportunities.",
+            agent=self.financial_analyst
+        )
+        return task.execute(financial_data)
+
+    def visualize_data(self, financial_data):
+        task = Task(
+            description="Create a revenue comparison graph based on the financial data. Return the plot as a base64 encoded string.",
+            agent=self.data_visualizer
+        )
+        return task.execute(financial_data)
+
+    def run(self):
+        st.title("FinAnalyst: Fortune 500 Financial Document Analyzer")
+        st.write("Upload a financial document, ask questions, and get detailed analysis!")
+
+        uploaded_file = st.file_uploader("Choose a financial document file", type=["pdf"])
+
+        if uploaded_file is not None:
+            file_content = uploaded_file.getvalue()
+            self.write_to_file(file_content)
+
+            st.write("Analyzing financial document...")
+
+            document_text = self.process_document(file_content)
+            financial_data = self.extract_financial_data(document_text)
+
+            # Parse the JSON string to a Python dictionary
+            financial_dict = json.loads(financial_data)
+
+            query = st.text_input("Enter your financial analysis query (e.g., 'What are the revenue trends?')", "")
+
+            if query:
+                analysis = self.analyze_financials(financial_data, query)
+                st.write("## Financial Analysis Result")
+                st.write(analysis)
+
+            st.write("## Revenue Comparison")
+            if financial_dict["Revenue"] and financial_dict["Date"]:
+                fig, ax = plt.subplots(figsize=(10, 6))
+                ax.plot(financial_dict["Date"], financial_dict["Revenue"], marker="o", linestyle="-", color="b", label="Revenue")
+                ax.set_title("Revenue Comparison")
+                ax.set_xlabel("Date")
+                ax.set_ylabel("Revenue (in millions)")
+                ax.grid(True)
+                ax.legend()
+                plt.xticks(rotation=45, ha="right")
+                plt.tight_layout()
+                st.pyplot(fig)
+            else:
+                st.write("No revenue data found for comparison.")
+
+if __name__ == "__main__":
+    fin_analyst = FinAnalyst()
+    fin_analyst.run()