Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
import streamlit as st
|
| 2 |
import requests
|
| 3 |
from bs4 import BeautifulSoup
|
| 4 |
-
from langchain_core.messages import HumanMessage # แก้
|
| 5 |
from langchain_groq import ChatGroq
|
| 6 |
import json
|
| 7 |
import os
|
|
@@ -10,11 +10,10 @@ from transformers import pipeline
|
|
| 10 |
# --- 0. CONFIGURATION & INITIALIZATION ---
|
| 11 |
|
| 12 |
# การตั้งค่า Groq API Key
|
| 13 |
-
#
|
| 14 |
-
|
| 15 |
-
GROQ_API_KEY = os.getenv('GROQ_API_KEY')
|
| 16 |
if not GROQ_API_KEY:
|
| 17 |
-
st.error("GROQ_API_KEY is not set. Please configure it in your Space Secrets.")
|
| 18 |
st.stop()
|
| 19 |
|
| 20 |
# Initialize the LLM model
|
|
@@ -22,13 +21,10 @@ llm = ChatGroq(api_key=GROQ_API_KEY, model="llama-3.1-8b-instant")
|
|
| 22 |
|
| 23 |
# --- 1. SCRAPING FUNCTION (Yahoo Finance Only) ---
|
| 24 |
|
| 25 |
-
# ใช้โค้ด Yahoo Finance ที่เราแก้ไขล่าสุด
|
| 26 |
def extract_titles_and_summaries(company_name, num_articles=10):
|
| 27 |
"""ดึงหัวข้อและสรุปข่าวจาก Yahoo Finance หน้าหลัก"""
|
| 28 |
url = 'https://finance.yahoo.com/news/'
|
| 29 |
|
| 30 |
-
#st.info(f"-> Scraping Yahoo Finance at: {url}")
|
| 31 |
-
|
| 32 |
try:
|
| 33 |
# เพิ่ม User-Agent เพื่อหลีกเลี่ยงการถูกบล็อก
|
| 34 |
response = requests.get(url, headers={'User-Agent': 'Mozilla/5.0'})
|
|
@@ -72,7 +68,6 @@ def extract_titles_and_summaries(company_name, num_articles=10):
|
|
| 72 |
|
| 73 |
# --- 2. ANALYSIS FUNCTIONS ---
|
| 74 |
|
| 75 |
-
# Function to perform sentiment analysis
|
| 76 |
def perform_sentiment_analysis(news_data):
|
| 77 |
"""ใช้ Hugging Face Pipeline วิเคราะห์ Sentiment"""
|
| 78 |
# ใช้ device=-1 เพื่อให้ทำงานบน CPU/อัตโนมัติ
|
|
@@ -101,7 +96,6 @@ def perform_sentiment_analysis(news_data):
|
|
| 101 |
|
| 102 |
return news_data, sentiment_counts
|
| 103 |
|
| 104 |
-
# Function to extract topics
|
| 105 |
def extract_topics_with_hf(news_data):
|
| 106 |
"""ใช้ Hugging Face Pipeline สกัดหัวข้อ"""
|
| 107 |
structured_data = {
|
|
@@ -127,14 +121,12 @@ def extract_topics_with_hf(news_data):
|
|
| 127 |
|
| 128 |
return structured_data
|
| 129 |
|
| 130 |
-
# Function to extract JSON response from LLM output
|
| 131 |
def extract_json(response):
|
| 132 |
try:
|
| 133 |
return json.loads(response)
|
| 134 |
except json.JSONDecodeError:
|
| 135 |
return {}
|
| 136 |
|
| 137 |
-
# Function to generate a final sentiment summary using LLM
|
| 138 |
def generate_final_sentiment(news_data, sentiment_counts):
|
| 139 |
"""ใช้ LLM สรุปผลลัพธ์สุดท้าย"""
|
| 140 |
company_name = news_data["Company"]
|
|
@@ -151,12 +143,10 @@ def generate_final_sentiment(news_data, sentiment_counts):
|
|
| 151 |
|
| 152 |
Respond **ONLY** with a well-structured very concise and short paragraph in plain text, focusing on overall sentiment.
|
| 153 |
"""
|
| 154 |
-
# ใช้ HumanMessage จาก langchain_core.messages
|
| 155 |
response = llm.invoke([HumanMessage(content=prompt)], max_tokens=200)
|
| 156 |
final_sentiment = response if response else "Sentiment analysis summary not available."
|
| 157 |
return final_sentiment.content
|
| 158 |
|
| 159 |
-
# Function to compare articles based on common topics and sentiment variations
|
| 160 |
def compare_articles(news_data, sentiment_counts):
|
| 161 |
"""ใช้ LLM เปรียบเทียบและสรุปความแตกต่างของข่าว"""
|
| 162 |
articles = news_data.get("Articles", [])
|
|
@@ -216,7 +206,6 @@ def compare_articles(news_data, sentiment_counts):
|
|
| 216 |
|
| 217 |
# --- 3. STREAMLIT UI IMPLEMENTATION ---
|
| 218 |
|
| 219 |
-
# ฟังก์ชันแสดงผลลัพธ์ (คัดลอกมาจาก app.py เดิม)
|
| 220 |
def display_articles(articles):
|
| 221 |
for i, article in enumerate(articles, start=1):
|
| 222 |
st.markdown(f"##### **Article {i} ({article['Source']}): {article['Title']}**")
|
|
@@ -236,7 +225,6 @@ def display_coverage_differences(coverage_differences):
|
|
| 236 |
if coverage_differences:
|
| 237 |
st.markdown("#### **Coverage Differences:**")
|
| 238 |
for diff in coverage_differences:
|
| 239 |
-
# ปรับปรุงการแสดงผลเพื่อหลีกเลี่ยงการพึ่งพาคีย์ที่อาจไม่มี
|
| 240 |
comparison = diff.get('Comparison', 'No Comparison Detail')
|
| 241 |
impact = diff.get('Impact', 'No Impact Detail')
|
| 242 |
st.write(f"- **{comparison}:** {impact}")
|
|
@@ -248,7 +236,6 @@ def display_topic_overlap(topic_overlap):
|
|
| 248 |
for article, topics in topic_overlap.get("Unique Topics", {}).items():
|
| 249 |
st.write(f" - **{article}:** {', '.join(topics)}")
|
| 250 |
|
| 251 |
-
# ฟังก์ชันหลักในการประมวลผลทั้งหมด
|
| 252 |
def run_analysis(company_name):
|
| 253 |
# 1. ดึงข่าว
|
| 254 |
with st.spinner('1/4 Scraping news from Yahoo Finance...'):
|
|
@@ -298,8 +285,7 @@ def run_analysis(company_name):
|
|
| 298 |
st.markdown(data.get("Final Sentiment Analysis", "No sentiment analysis available."))
|
| 299 |
st.markdown("---")
|
| 300 |
|
| 301 |
-
|
| 302 |
-
st.json(final_report) # แสดงผล JSON ทั้งหมดให้ user เห็น
|
| 303 |
|
| 304 |
# --- MAIN STREAMLIT APP ---
|
| 305 |
|
|
|
|
| 1 |
import streamlit as st
|
| 2 |
import requests
|
| 3 |
from bs4 import BeautifulSoup
|
| 4 |
+
from langchain_core.messages import HumanMessage # ใช้ langchain_core เพื่อแก้ปัญหาการ Import
|
| 5 |
from langchain_groq import ChatGroq
|
| 6 |
import json
|
| 7 |
import os
|
|
|
|
| 10 |
# --- 0. CONFIGURATION & INITIALIZATION ---
|
| 11 |
|
| 12 |
# การตั้งค่า Groq API Key
|
| 13 |
+
# โค้ดจะดึงค่าจาก Secret ที่ชื่อ GROQ_API_KEY ใน Hugging Face Space
|
| 14 |
+
GROQ_API_KEY = os.getenv('GROQ_API_KEY')
|
|
|
|
| 15 |
if not GROQ_API_KEY:
|
| 16 |
+
st.error("GROQ_API_KEY is not set. Please configure it in your Space Secrets (Settings > Repository secrets).")
|
| 17 |
st.stop()
|
| 18 |
|
| 19 |
# Initialize the LLM model
|
|
|
|
| 21 |
|
| 22 |
# --- 1. SCRAPING FUNCTION (Yahoo Finance Only) ---
|
| 23 |
|
|
|
|
| 24 |
def extract_titles_and_summaries(company_name, num_articles=10):
|
| 25 |
"""ดึงหัวข้อและสรุปข่าวจาก Yahoo Finance หน้าหลัก"""
|
| 26 |
url = 'https://finance.yahoo.com/news/'
|
| 27 |
|
|
|
|
|
|
|
| 28 |
try:
|
| 29 |
# เพิ่ม User-Agent เพื่อหลีกเลี่ยงการถูกบล็อก
|
| 30 |
response = requests.get(url, headers={'User-Agent': 'Mozilla/5.0'})
|
|
|
|
| 68 |
|
| 69 |
# --- 2. ANALYSIS FUNCTIONS ---
|
| 70 |
|
|
|
|
| 71 |
def perform_sentiment_analysis(news_data):
|
| 72 |
"""ใช้ Hugging Face Pipeline วิเคราะห์ Sentiment"""
|
| 73 |
# ใช้ device=-1 เพื่อให้ทำงานบน CPU/อัตโนมัติ
|
|
|
|
| 96 |
|
| 97 |
return news_data, sentiment_counts
|
| 98 |
|
|
|
|
| 99 |
def extract_topics_with_hf(news_data):
|
| 100 |
"""ใช้ Hugging Face Pipeline สกัดหัวข้อ"""
|
| 101 |
structured_data = {
|
|
|
|
| 121 |
|
| 122 |
return structured_data
|
| 123 |
|
|
|
|
| 124 |
def extract_json(response):
|
| 125 |
try:
|
| 126 |
return json.loads(response)
|
| 127 |
except json.JSONDecodeError:
|
| 128 |
return {}
|
| 129 |
|
|
|
|
| 130 |
def generate_final_sentiment(news_data, sentiment_counts):
|
| 131 |
"""ใช้ LLM สรุปผลลัพธ์สุดท้าย"""
|
| 132 |
company_name = news_data["Company"]
|
|
|
|
| 143 |
|
| 144 |
Respond **ONLY** with a well-structured very concise and short paragraph in plain text, focusing on overall sentiment.
|
| 145 |
"""
|
|
|
|
| 146 |
response = llm.invoke([HumanMessage(content=prompt)], max_tokens=200)
|
| 147 |
final_sentiment = response if response else "Sentiment analysis summary not available."
|
| 148 |
return final_sentiment.content
|
| 149 |
|
|
|
|
| 150 |
def compare_articles(news_data, sentiment_counts):
|
| 151 |
"""ใช้ LLM เปรียบเทียบและสรุปความแตกต่างของข่าว"""
|
| 152 |
articles = news_data.get("Articles", [])
|
|
|
|
| 206 |
|
| 207 |
# --- 3. STREAMLIT UI IMPLEMENTATION ---
|
| 208 |
|
|
|
|
| 209 |
def display_articles(articles):
|
| 210 |
for i, article in enumerate(articles, start=1):
|
| 211 |
st.markdown(f"##### **Article {i} ({article['Source']}): {article['Title']}**")
|
|
|
|
| 225 |
if coverage_differences:
|
| 226 |
st.markdown("#### **Coverage Differences:**")
|
| 227 |
for diff in coverage_differences:
|
|
|
|
| 228 |
comparison = diff.get('Comparison', 'No Comparison Detail')
|
| 229 |
impact = diff.get('Impact', 'No Impact Detail')
|
| 230 |
st.write(f"- **{comparison}:** {impact}")
|
|
|
|
| 236 |
for article, topics in topic_overlap.get("Unique Topics", {}).items():
|
| 237 |
st.write(f" - **{article}:** {', '.join(topics)}")
|
| 238 |
|
|
|
|
| 239 |
def run_analysis(company_name):
|
| 240 |
# 1. ดึงข่าว
|
| 241 |
with st.spinner('1/4 Scraping news from Yahoo Finance...'):
|
|
|
|
| 285 |
st.markdown(data.get("Final Sentiment Analysis", "No sentiment analysis available."))
|
| 286 |
st.markdown("---")
|
| 287 |
|
| 288 |
+
st.json(final_report)
|
|
|
|
| 289 |
|
| 290 |
# --- MAIN STREAMLIT APP ---
|
| 291 |
|