Upload community_contributions/chrys/app.py with huggingface_hub

community_contributions/chrys/app.py

@@ -0,0 +1,159 @@
+"""Gradio UI for ARIA: agent-colored log stream and results table."""
+import html
+import logging
+import queue
+import sys
+import os
+import time
+
+# Ensure chrys is on path
+_this_dir = os.path.dirname(os.path.abspath(__file__))
+if _this_dir not in sys.path:
+    sys.path.insert(0, _this_dir)
+
+import gradio as gr
+from dotenv import load_dotenv
+load_dotenv()
+
+from orchestrator import run_pipeline
+from models import DecisionRecord
+
+# Agent name + color for UI (logger name -> (display name, hex color))
+AGENT_COLORS = {
+    "aria.data_fetcher": ("DATA_FETCHER", "#3498db"),
+    "aria.tech_analyst": ("TECH_ANALYST", "#27ae60"),
+    "aria.sentiment_agent": ("SENTIMENT_AGENT", "#f39c12"),
+    "aria.decision_agent": ("DECISION_AGENT", "#9b59b6"),
+    "aria.notifier": ("NOTIFIER", "#1abc9c"),
+}
+DEFAULT_COLOR = "#bdc3c7"
+MAX_LOG_LINES = 150
+
+
+class AgentQueueHandler(logging.Handler):
+    """Put (logger_name, formatted_message) into queue for colored UI."""
+    def __init__(self, log_queue: queue.Queue):
+        super().__init__()
+        self.log_queue = log_queue
+
+    def emit(self, record: logging.LogRecord):
+        try:
+            msg = self.format(record)
+            self.log_queue.put((record.name, msg))
+        except Exception:
+            self.handleError(record)
+
+
+def agent_line_to_html(logger_name: str, message: str) -> str:
+    label, color = AGENT_COLORS.get(logger_name, ("ARIA", DEFAULT_COLOR))
+    safe_msg = html.escape(message)
+    return f'<span style="color:{color};font-weight:600">[{label}]</span> {safe_msg}'
+
+
+def html_for_log(log_lines: list) -> str:
+    recent = log_lines[-MAX_LOG_LINES:]
+    content = "<br>".join(recent)
+    return f"""
+    <div style="height:420px;overflow-y:auto;border:1px solid #444;background:#1e1e2e;padding:12px;font-family:monospace;font-size:13px;color:#cdd6f4;">
+    {content}
+    </div>
+    """
+
+
+def setup_aria_logging(log_queue: queue.Queue) -> None:
+    root = logging.getLogger()
+    handler = AgentQueueHandler(log_queue)
+    handler.setFormatter(logging.Formatter("[%(asctime)s] %(message)s", datefmt="%H:%M:%S"))
+    root.addHandler(handler)
+    root.setLevel(logging.INFO)
+    for name in AGENT_COLORS:
+        logging.getLogger(name).setLevel(logging.INFO)
+
+
+def table_for(records: list[DecisionRecord]) -> list[list]:
+    if not records:
+        return []
+    return [
+        [r.asset, str(r.tech_score), r.sentiment, f"{r.final_score:.1f}", r.decision, r.skip_reason or ""]
+        for r in records
+    ]
+
+
+def run_pipeline_with_logging(log_queue: queue.Queue, result_queue: queue.Queue) -> None:
+    try:
+        records, _ = run_pipeline()
+        result_queue.put(records)
+    except Exception as e:
+        logging.exception("Pipeline failed")
+        result_queue.put([])
+
+
+def run_clicked(log_state):
+    log_queue = queue.Queue()
+    result_queue = queue.Queue()
+    root = logging.getLogger()
+    for h in root.handlers[:]:
+        if isinstance(h, AgentQueueHandler):
+            root.removeHandler(h)
+    setup_aria_logging(log_queue)
+    log_lines = []
+    thread = __import__("threading").Thread(
+        target=run_pipeline_with_logging,
+        args=(log_queue, result_queue),
+        daemon=True,
+    )
+    thread.start()
+    table = []
+    while True:
+        while True:
+            try:
+                logger_name, msg = log_queue.get_nowait()
+                log_lines.append(agent_line_to_html(logger_name, msg))
+            except queue.Empty:
+                break
+        try:
+            records = result_queue.get_nowait()
+            table = table_for(records)
+            break
+        except queue.Empty:
+            pass
+        yield log_lines, html_for_log(log_lines), table
+        time.sleep(0.08)
+        if not thread.is_alive():
+            try:
+                records = result_queue.get_nowait()
+                table = table_for(records)
+            except queue.Empty:
+                table = []
+            break
+    yield log_lines, html_for_log(log_lines), table
+
+
+def build_ui():
+    with gr.Blocks(title="ARIA — Market Intelligence", theme=gr.themes.Soft(), css="""
+        .log-panel { font-family: ui-monospace, monospace; }
+    """) as ui:
+        gr.Markdown("# ARIA — Automated Real-time Investment Alert Agent")
+        gr.Markdown("Multi-agent pipeline: Data → Tech Analysis → Sentiment → Decision → Pushover alerts.")
+        log_state = gr.State([])
+        with gr.Row():
+            run_btn = gr.Button("Run pipeline", variant="primary")
+        with gr.Row():
+            log_html = gr.HTML(value=html_for_log([]), label="Agent log")
+        with gr.Row():
+            results_df = gr.Dataframe(
+                headers=["Asset", "Tech score", "Sentiment", "Final score", "Decision", "Skip reason"],
+                datatype=["str", "str", "str", "str", "str", "str"],
+                label="Last run results",
+            )
+        run_btn.click(
+            fn=run_clicked,
+            inputs=[log_state],
+            outputs=[log_state, log_html, results_df],
+        )
+    return ui
+
+
+if __name__ == "__main__":
+    ui = build_ui()
+    ui.launch(share=False, inbrowser=True)