Refactoring code, provide better abstraction and file structure

.idea/.gitignore

@@ -0,0 +1,8 @@
+# Default ignored files
+/shelf/
+/workspace.xml
+# Editor-based HTTP Client requests
+/httpRequests/
+# Datasource local storage ignored files
+/dataSources/
+/dataSources.local.xml

.idea/Financial_analysis_system_draft.iml

@@ -0,0 +1,14 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<module type="PYTHON_MODULE" version="4">
+  <component name="NewModuleRootManager">
+    <content url="file://$MODULE_DIR$">
+      <excludeFolder url="file://$MODULE_DIR$/.venv" />
+    </content>
+    <orderEntry type="jdk" jdkName="Python 3.9 (Financial_analysis_system_draft)" jdkType="Python SDK" />
+    <orderEntry type="sourceFolder" forTests="false" />
+  </component>
+  <component name="PyDocumentationSettings">
+    <option name="format" value="PLAIN" />
+    <option name="myDocStringFormat" value="Plain" />
+  </component>
+</module>

.idea/inspectionProfiles/profiles_settings.xml

@@ -0,0 +1,6 @@
+<component name="InspectionProjectProfileManager">
+  <settings>
+    <option name="USE_PROJECT_PROFILE" value="false" />
+    <version value="1.0" />
+  </settings>
+</component>

.idea/misc.xml

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="Black">
+    <option name="sdkName" value="Python 3.9 (Financial_analysis_system_draft)" />
+  </component>
+  <component name="ProjectRootManager" version="2" project-jdk-name="Python 3.9 (Financial_analysis_system_draft)" project-jdk-type="Python SDK" />
+</project>

.idea/modules.xml

@@ -0,0 +1,8 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="ProjectModuleManager">
+    <modules>
+      <module fileurl="file://$PROJECT_DIR$/.idea/Financial_analysis_system_draft.iml" filepath="$PROJECT_DIR$/.idea/Financial_analysis_system_draft.iml" />
+    </modules>
+  </component>
+</project>

.idea/vcs.xml

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="VcsDirectoryMappings">
+    <mapping directory="" vcs="Git" />
+  </component>
+</project>

agents/agent_builder.py

@@ -1,9 +1,6 @@
 from langchain.agents import AgentExecutor, create_tool_calling_agent
 from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
-from langchain_openai import ChatOpenAI
-from langgraph.graph import StateGraph, END
-from typing import TypedDict, List, Optional, Dict, Any
-from agents.tools import search_news_tool, fetch_earnings_tool, fetch_earnings_tool, fetch_market_snapshot_tool
+from tools.tools import search_news_tool, fetch_earnings_tool, fetch_market_snapshot_tool
 
 def build_agent(llm, tools, system_instructions: str) -> AgentExecutor:
     """Create a tool-calling agent with a focused system prompt."""

agents/build_graph.py

@@ -1,10 +1,14 @@
 import os
 from langgraph.graph import StateGraph, END
-from agents.agent_builder import make_news_agent, make_earnings_agent, make_market_agent, make_synthesizer
-from agents.graph_state import GraphState
+from agents.agent_builder import make_synthesizer
+from agents.earnings_agent.earnings_agent import create_earnings_agent
+from workflow.graph_state import GraphState
 from langchain_openai import ChatOpenAI
-from agents.nodes import supervisor_node, news_node, earnings_node, market_node, synth_node
-from agents.supervisor_router import AGENTS, supervisor_router
+
+from agents.market_agent.market_agent import create_market_agent
+from agents.news_agent.news_agent import create_news_agent
+from workflow.nodes.nodes import supervisor_node, news_node, earnings_node, market_node, synth_node, AGENTS
+
 
 def build_graph(llm_model_name: str = "gpt-4o-mini"):
 
@@ -13,9 +17,9 @@ def build_graph(llm_model_name: str = "gpt-4o-mini"):
     llm = ChatOpenAI(api_key=openai_api_key,model=llm_model_name, temperature=0)
 
     # --- Create specialized agents ---
-    news_agent = make_news_agent(llm)
-    earnings_agent = make_earnings_agent(llm)
-    market_agent = make_market_agent(llm)
+    news_agent = create_news_agent(llm)
+    earnings_agent = create_earnings_agent(llm)
+    market_agent = create_market_agent(llm)
 
     # --- Create synthesizer chain ---
     synthesizer = make_synthesizer(llm)

agents/earnings_agent/earnings_agent.py

@@ -0,0 +1,22 @@
+from langchain.agents import AgentExecutor, create_tool_calling_agent
+from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
+from agents.earnings_agent.tools import fetch_earnings_tool
+from pathlib import Path
+import yaml
+
+yaml_path = Path(__file__).parent / "prompts.yaml"
+with yaml_path.open() as f:
+    prompt_template = yaml.safe_load(f)
+
+def create_earnings_agent(model) -> AgentExecutor:
+
+    prompt = ChatPromptTemplate.from_messages(
+        [
+            ("system", prompt_template["system"]),
+            ("human", "{input}"),
+            MessagesPlaceholder("agent_scratchpad"),
+        ]
+    )
+    agent = create_tool_calling_agent(llm=model, tools=[fetch_earnings_tool], prompt=prompt)
+
+    return AgentExecutor(agent=agent, tools=[fetch_earnings_tool], verbose=False, handle_parsing_errors=True)

agents/earnings_agent/prompts.yaml

@@ -0,0 +1,4 @@
+system: |
+  You are an Earnings Analyst. Use the earnings tool to summarize the latest and upcoming 
+  earnings information (dates, surprises if available) and key line items. Provide a 
+  short view on momentum and watchouts. Output concise markdown.

agents/earnings_agent/tools.py

@@ -0,0 +1,54 @@
+from langchain.tools import tool
+import yfinance as yf
+import datetime as dt
+
+@tool("fetch_earnings", return_direct=False)
+def fetch_earnings_tool(ticker: str) -> str:
+    """
+    Fetch upcoming and recent earnings info via yfinance.
+    Returns a concise summary (dates + surprises if available).
+    """
+    tk = yf.Ticker(ticker)
+    lines = [f"EARNINGS SNAPSHOT for {ticker.upper()}"]
+
+    # Upcoming earnings (earnings_dates includes future dates)
+    try:
+        ed = tk.earnings_dates  # DataFrame if available
+        if ed is not None and not ed.empty:
+            # Take the next upcoming date and last reported
+            ed_sorted = ed.sort_index()
+            upcoming = ed_sorted[ed_sorted.index >= dt.datetime.now().date()]
+            last = ed_sorted[ed_sorted.index < dt.datetime.now().date()]
+            if not upcoming.empty:
+                lines.append(f"Upcoming: {upcoming.index[0].strftime('%Y-%m-%d')}")
+            if not last.empty:
+                # Try EPS surprise columns if present
+                row = last.iloc[-1]
+                surprise = None
+                for k in ["EPS Surprise %", "Surprise(%)", "epssurprisepct", "epssurprisepercent"]:
+                    if k in row and row[k] is not None:
+                        surprise = row[k]
+                        break
+                lines.append(
+                    f"Last reported: {last.index[-1].strftime('%Y-%m-%d')}"
+                    + (f", EPS surprise: {surprise}" if surprise is not None else "")
+                )
+        else:
+            lines.append("No earnings_dates available.")
+    except Exception as e:
+        lines.append(f"earnings_dates unavailable: {e}")
+
+    # Quarterly financials (very high-level)
+    try:
+        qf = tk.quarterly_financials
+        if qf is not None and not qf.empty:
+            cols = list(qf.columns)
+            if cols:
+                last_q = cols[0]
+                revenue = qf.loc["Total Revenue", last_q] if "Total Revenue" in qf.index else None
+                gross_profit = qf.loc["Gross Profit", last_q] if "Gross Profit" in qf.index else None
+                lines.append(f"Last quarter ({last_q.date()}): Revenue={revenue}, GrossProfit={gross_profit}")
+    except Exception:
+        pass
+
+    return "\n".join(lines)

agents/market_agent/market_agent.py

@@ -0,0 +1,23 @@
+from langchain.agents import AgentExecutor, create_tool_calling_agent
+from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
+from agents.market_agent.tools import fetch_market_snapshot_tool
+from pathlib import Path
+import yaml
+
+yaml_path = Path(__file__).parent / "prompts.yaml"
+with yaml_path.open() as f:
+    prompt_template = yaml.safe_load(f)
+
+
+def create_market_agent(model) -> AgentExecutor:
+
+    prompt = ChatPromptTemplate.from_messages(
+        [
+            ("system", prompt_template["system"]),
+            ("human", "{input}"),
+            MessagesPlaceholder("agent_scratchpad"),
+        ]
+    )
+    agent = create_tool_calling_agent(llm=model, tools=[fetch_market_snapshot_tool], prompt=prompt)
+
+    return AgentExecutor(agent=agent, tools=[fetch_market_snapshot_tool], verbose=False, handle_parsing_errors=True)

agents/market_agent/prompts.yaml

@@ -0,0 +1,4 @@
+system: |
+  You are a Market & Valuation Analyst. Use the market snapshot tool to extract current 
+  trading context and discuss short-term technicals/flow and high-level valuation notes. 
+  Output concise markdown.

agents/market_agent/tools.py

@@ -0,0 +1,33 @@
+from langchain.tools import tool
+import yfinance as yf
+import datetime as dt
+
+@tool("fetch_market_snapshot", return_direct=False)
+def fetch_market_snapshot_tool(ticker: str) -> str:
+    """
+    Pulls basic market snapshot with yfinance: price, change, volume, valuation.
+    Returns a compact textual snapshot.
+    """
+    tk = yf.Ticker(ticker)
+    info = {}
+    try:
+        price = tk.fast_info.last_price
+        prev_close = tk.fast_info.previous_close
+        change = None
+        if price is not None and prev_close:
+            change = (price - prev_close) / prev_close * 100
+        info.update({
+            "price": price, "prev_close": prev_close, "pct_change": change,
+            "market_cap": tk.fast_info.market_cap, "volume": tk.fast_info.last_volume,
+            "currency": tk.fast_info.currency
+        })
+    except Exception as e:
+        return f"Market snapshot failed: {e}"
+
+    lines = [f"MARKET SNAPSHOT for {ticker.upper()}"]
+    lines.append(f"Price: {info.get('price')} {info.get('currency')}")
+    if info.get("pct_change") is not None:
+        lines.append(f"Day change: {info['pct_change']:.2f}%")
+    lines.append(f"Market Cap: {info.get('market_cap')}")
+    lines.append(f"Volume: {info.get('volume')}")
+    return "\n".join(lines)

agents/news_agent/news_agent.py

@@ -0,0 +1,22 @@
+from langchain.agents import AgentExecutor, create_tool_calling_agent
+from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
+from agents.news_agent.tools import search_news_tool
+from pathlib import Path
+import yaml
+
+yaml_path = Path(__file__).parent / "prompts.yaml"
+with yaml_path.open() as f:
+    prompt_template = yaml.safe_load(f)
+
+def create_news_agent(model) -> AgentExecutor:
+
+    prompt = ChatPromptTemplate.from_messages(
+        [
+            ("system", prompt_template["system"]),
+            ("human", "{input}"),
+            MessagesPlaceholder("agent_scratchpad"),
+        ]
+    )
+    agent = create_tool_calling_agent(llm=model, tools=[search_news_tool], prompt=prompt)
+
+    return AgentExecutor(agent=agent, tools=[search_news_tool], verbose=False, handle_parsing_errors=True)

agents/news_agent/prompts.yaml

@@ -0,0 +1,4 @@
+system: |
+  You are a News Analyst. Use the search tool to gather 5-8 recent, credible items.
+  Synthesize themes, risks, catalysts, and sentiment for investors. Output a concise
+  markdown summary with bullet points and 1-2 short citations (URLs).

agents/news_agent/tools.py

@@ -0,0 +1,27 @@
+from langchain.tools import tool
+try:
+    from duckduckgo_search import DDGS
+except Exception:
+    DDGS = None
+
+@tool("search_news", return_direct=False)
+def search_news_tool(query: str, max_results: int = 5) -> str:
+    """
+    Search latest headlines & snippets relevant to a stock or topic.
+    Uses duckduckgo_search as a simple public news proxy.
+    Returns a concise, newline-separated list of 'title — url'.
+    """
+    if DDGS is None:
+        return ("duckduckgo_search not installed. "
+                "Install with `pip install duckduckgo-search` "
+                "or replace this tool with your news API.")
+    items = []
+    with DDGS() as ddgs:
+        for r in ddgs.news(query, timelimit="7d", max_results=max_results):
+            title = r.get("title", "")[:160]
+            url = r.get("url", "")
+            if title and url:
+                items.append(f"{title} — {url}")
+    if not items:
+        return "No recent news found."
+    return "\n".join(items)

agents/supervisor_router.py

@@ -1,7 +0,0 @@
-from agents.graph_state import GraphState
-
-AGENTS = ["news", "earnings", "market"]
-
-def supervisor_router(state: GraphState) -> str:
-    remaining = [a for a in AGENTS if a not in state.get("completed", [])]
-    return remaining[0] if remaining else "synth"

app.py

@@ -1,8 +1,8 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
-from agents.build_graph import build_graph
+from workflow.agents_workflow import build_agents_workflow
+from workflow.graph_state import GraphState
 
-app = build_graph(llm_model_name="gpt-4o-mini")
+app = build_agents_workflow(llm_model_name="gpt-4o-mini")
 
 def run_user_query(ticker, history):
     QUERY = f"Produce investor-ready insights for {ticker}."
@@ -29,4 +29,4 @@ with gr.Blocks() as demo:
     clear.click(lambda: None, None, chatbot, queue=False)
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch(share=True, show_api=False)

main.py

@@ -0,0 +1,32 @@
+from agents.build_graph import build_graph
+from workflow.agents_workflow import build_agents_workflow
+from workflow.graph_state import GraphState
+
+# Run locally
+
+app = build_agents_workflow(llm_model_name="gpt-4o-mini")
+
+def run_user_query(ticker):
+    QUERY = f"Produce investor-ready insights for {ticker}."
+    init_state: GraphState = {
+        "ticker": ticker,
+        "query": QUERY,
+        "news_summary": None,
+        "earnings_summary": None,
+        "market_summary": None,
+        "completed": [],
+        "final_recommendation": None,
+    }
+    final_state = app.invoke(init_state)
+
+    # Update the Gradio chat history
+
+    return final_state
+
+state = run_user_query("AAPL")
+
+print("\n" + "=" * 80)
+print(f"### NEWS SUMMARY\n{state['news_summary']}\n")
+print(f"### EARNINGS SUMMARY\n{state['earnings_summary']}\n")
+print(f"### MARKET SUMMARY\n{state['market_summary']}\n")
+print(f"### FINAL RECOMMENDATION\n{state['final_recommendation']}\n")

model/init_model.py

@@ -0,0 +1,8 @@
+import os
+from langchain_openai import ChatOpenAI
+
+def init_main_model(llm_model_name: str):
+    openai_api_key = os.getenv("openai_api_key")
+    llm = ChatOpenAI(api_key=openai_api_key, model=llm_model_name, temperature=0)
+
+    return llm

requirements.txt

@@ -2,4 +2,11 @@ langgraph==0.6.10
 langchain==0.3.27
 langchain_openai==0.3.35
 duckduckgo_search==8.1.1
-yfinance==0.2.66
+#yfinance==0.2.66
+websockets
+yfinance~=0.2.53
+duckduckgo-search~=8.1.1
+pyyaml~=6.0.3
+langchain-openai~=0.3.35
+gradio~=4.44.1
+huggingface-hub~=0.35.3

workflow/agents_workflow.py

@@ -0,0 +1,69 @@
+from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
+import yaml
+from langgraph.graph import StateGraph, END
+
+from agents.earnings_agent.earnings_agent import create_earnings_agent
+from agents.market_agent.market_agent import create_market_agent
+from agents.news_agent.news_agent import create_news_agent
+from model.init_model import init_main_model
+from workflow.graph_state import GraphState
+from workflow.nodes.nodes import news_node, earnings_node, market_node, synth_node, supervisor_node, AGENTS, supervisor_router
+from pathlib import Path
+
+yaml_path = Path(__file__).parent / "prompts.yaml"
+with yaml_path.open() as f:
+    prompt_template = yaml.safe_load(f)
+
+def make_synthesizer(model):
+    """Final writer to merge all agent outputs into actionable recommendations."""
+    template = ChatPromptTemplate.from_messages(
+        [
+            ("system", prompt_template["system"]),
+            ("human", prompt_template["human"])
+        ]
+    )
+    return template | model  # LC chain: Prompt -> LLM
+
+def build_agents_workflow(llm_model_name):
+    # --- Base LLM for agents & synthesizer, we can initiate different models for agents here ---
+    model = init_main_model(llm_model_name)
+
+    # --- Create specialized agents ---
+    news_agent = create_news_agent(model)
+    earnings_agent = create_earnings_agent(model)
+    market_agent = create_market_agent(model)
+
+    # --- Create synthesizer chain ---
+    synthesizer = make_synthesizer(model)
+
+    # --- LangGraph: wire nodes ---
+    g = StateGraph(GraphState)
+
+    # Bind node callables with their dependencies via closures
+    g.add_node("news", lambda s: news_node(s, news_agent))
+    g.add_node("earnings", lambda s: earnings_node(s, earnings_agent))
+    g.add_node("market", lambda s: market_node(s, market_agent))
+    g.add_node("synth", lambda s: synth_node(s, synthesizer))
+
+    # Supervisor node
+    g.add_node("supervisor", supervisor_node)
+    # Edges: start -> supervisor -> (news|earnings|market|synth) -> supervisor ... -> synth -> END
+    g.set_entry_point("supervisor")
+
+    for a in AGENTS:
+        g.add_edge(a, "supervisor")
+    g.add_edge("synth", END)
+
+    # Route decisions come from the router function (returns a string)
+    g.add_conditional_edges(
+        "supervisor",
+        supervisor_router,   # returns: "news" | "earnings" | "market" | "synth"
+        {
+            "news": "news",
+            "earnings": "earnings",
+            "market": "market",
+            "synth": "synth",
+        },
+    )
+
+    return g.compile()

{agents → workflow/nodes}/nodes.py

@@ -1,11 +1,18 @@
-from langchain.agents import AgentExecutor, create_tool_calling_agent
-from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
-from agents.graph_state import GraphState
+from langchain.agents import AgentExecutor
+from workflow.graph_state import GraphState
+from pathlib import Path
+import yaml
+
+yaml_path = Path(__file__).parent / "prompts.yaml"
+with yaml_path.open() as f:
+    prompt_template = yaml.safe_load(f)
+
+AGENTS = ["news", "earnings", "market"]
 
 def news_node(state: GraphState, agent: AgentExecutor) -> GraphState:
     ticker = state["ticker"]
-    q = f"Research recent news for {ticker}. Focus on price-moving catalysts."
-    res = agent.invoke({"input": q})
+    query = prompt_template['news_user_prompt'].format(ticker=ticker)
+    res = agent.invoke({"input": query})
     state["news_summary"] = res["output"]
     state["completed"] = list(set(state["completed"] + ["news"]))
     return state
@@ -13,8 +20,8 @@ def news_node(state: GraphState, agent: AgentExecutor) -> GraphState:
 
 def earnings_node(state: GraphState, agent: AgentExecutor) -> GraphState:
     ticker = state["ticker"]
-    q = f"Analyze earnings for {ticker}. Summarize last and upcoming earnings. Use the tool."
-    res = agent.invoke({"input": q})
+    query = prompt_template['earnings_user_prompt'].format(ticker=ticker)
+    res = agent.invoke({"input": query})
     state["earnings_summary"] = res["output"]
     state["completed"] = list(set(state["completed"] + ["earnings"]))
     return state
@@ -22,8 +29,8 @@ def earnings_node(state: GraphState, agent: AgentExecutor) -> GraphState:
 
 def market_node(state: GraphState, agent: AgentExecutor) -> GraphState:
     ticker = state["ticker"]
-    q = f"Provide a market snapshot for {ticker}. Use the tool."
-    res = agent.invoke({"input": q})
+    query = prompt_template['market_user_prompt'].format(ticker=ticker)
+    res = agent.invoke({"input": query})
     state["market_summary"] = res["output"]
     state["completed"] = list(set(state["completed"] + ["market"]))
     return state
@@ -45,8 +52,6 @@ def supervisor_node(state: GraphState) -> GraphState:
     # Do any bookkeeping here if needed; otherwise just pass state through
     return state
 
-AGENTS = ["news", "earnings", "market"]
-
 def supervisor_router(state: GraphState) -> str:
     remaining = [a for a in AGENTS if a not in state.get("completed", [])]
     return remaining[0] if remaining else "synth"

workflow/nodes/prompts.yaml

@@ -0,0 +1,3 @@
+news_user_prompt: Research recent news for {ticker}. Focus on price-moving catalysts.
+earnings_user_prompt: Analyze earnings for {ticker}. Summarize last and upcoming earnings. Use the tool.
+market_user_prompt: Provide a market snapshot for {ticker}. Use the tool.

workflow/prompts.yaml

@@ -0,0 +1,11 @@
+system: |
+  You are the Lead Portfolio Analyst. Merge inputs from News, Earnings, and Market agents. 
+  Produce a final, actionable recommendation block (Buy/Hold/Sell with confidence 0-1), 
+  key drivers (bull/bear), near-term catalysts, and 2-3 risks. Be concise and concrete.
+
+human: |
+  Ticker: {ticker}\n\n
+  ### News Summary\n{news_summary}\n\n
+  ### Earnings Summary\n{earnings_summary}\n\n
+  ### Market Summary\n{market_summary}\n\n
+  Write the final recommendation now.