Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -9,7 +9,7 @@ import os
|
|
| 9 |
import plotly.express as px
|
| 10 |
|
| 11 |
# --- Page Config ---
|
| 12 |
-
st.set_page_config(page_title="Global Issue Intelligence", page_icon="π‘", layout="wide")
|
| 13 |
|
| 14 |
# --- π API Key Load ---
|
| 15 |
api_key = os.environ.get("GROQ_API_KEY")
|
|
@@ -24,6 +24,9 @@ if not api_key:
|
|
| 24 |
# Groq ν΄λΌμ΄μΈνΈ μ΄κΈ°ν
|
| 25 |
client = Groq(api_key=api_key)
|
| 26 |
|
|
|
|
|
|
|
|
|
|
| 27 |
# --- Monitoring Nations ---
|
| 28 |
MONITOR_LIST = {
|
| 29 |
"South Korea": {"geo": "KR", "lang": "ko"}, "USA": {"geo": "US", "lang": "en"},
|
|
@@ -39,18 +42,18 @@ def fetch_google_news(query, geo="US", lang="en", limit=10):
|
|
| 39 |
feed = feedparser.parse(url)
|
| 40 |
return [{"title": e.title, "source": e.source.title if 'source' in e else "G-News"} for e in feed.entries[:limit]]
|
| 41 |
|
| 42 |
-
st.title("π‘ Omni-Watch: Global
|
| 43 |
-
st.caption("Strategic Intelligence for
|
| 44 |
st.markdown("---")
|
| 45 |
|
| 46 |
-
tab_search, tab_report = st.tabs(["π Global Issue Tracking
|
| 47 |
|
| 48 |
-
# --- [Tab 1: Global Issue Tracking
|
| 49 |
with tab_search:
|
| 50 |
-
st.subheader("Global Risk Propagation &
|
| 51 |
col_k1, col_k2 = st.columns([3, 1])
|
| 52 |
with col_k1:
|
| 53 |
-
search_keyword = st.text_input("Enter Issue Keyword
|
| 54 |
with col_k2:
|
| 55 |
st.write("")
|
| 56 |
run_global = st.button("πΊοΈ Global Deep Scan", type="primary", use_container_width=True)
|
|
@@ -62,116 +65,61 @@ with tab_search:
|
|
| 62 |
status_msg = st.empty()
|
| 63 |
|
| 64 |
for i, (name, conf) in enumerate(MONITOR_LIST.items()):
|
| 65 |
-
status_msg.text(f"Extracting
|
| 66 |
news = fetch_google_news(f"{name} {search_keyword}", geo=conf['geo'], lang=conf['lang'], limit=3)
|
| 67 |
|
| 68 |
if news:
|
| 69 |
titles = "\n".join([n['title'] for n in news])
|
| 70 |
all_captured_news.append(f"--- {name} News ---\n{titles}")
|
| 71 |
|
| 72 |
-
# μμ λ λΆλΆ: client.chat.completions.create
|
| 73 |
try:
|
| 74 |
res = client.chat.completions.create(
|
| 75 |
-
model=
|
| 76 |
-
messages=[{"role": "user", "content": f"
|
| 77 |
max_tokens=10
|
| 78 |
)
|
| 79 |
-
|
| 80 |
-
score = int(''.join(filter(str.isdigit, score_text)))
|
| 81 |
except: score = 50
|
| 82 |
results.append({"Country": name, "Risk": score})
|
| 83 |
progress.progress((i + 1) / len(MONITOR_LIST))
|
| 84 |
|
| 85 |
status_msg.empty()
|
| 86 |
df_global = pd.DataFrame(results).sort_values("Risk", ascending=False)
|
| 87 |
-
|
| 88 |
-
# μκ°ν
|
| 89 |
-
fig = px.bar(df_global, x="Country", y="Risk", color="Risk", color_continuous_scale="Reds", title=f"Global Risk Heatmap: {search_keyword}")
|
| 90 |
-
st.plotly_chart(fig, use_container_width=True)
|
| 91 |
|
| 92 |
-
# [Global Deep-Dive Report Section]
|
| 93 |
st.divider()
|
| 94 |
-
st.subheader(f"π§ Llama 3 Global Intelligence Report
|
| 95 |
-
|
| 96 |
-
|
| 97 |
|
| 98 |
-
with st.spinner("Synthesizing
|
| 99 |
combined_news = "\n\n".join(all_captured_news)
|
| 100 |
-
global_system_prompt = f"""
|
| 101 |
-
You are a Global Strategic Intelligence Officer. Analyze the propagation of '{search_keyword}' across multiple countries.
|
| 102 |
-
Your report must be long, professional, and insightful.
|
| 103 |
-
|
| 104 |
-
REQUIREMENTS (Output in English):
|
| 105 |
-
1. 'Global Narrative': Detailed summary of how this issue is trending (300+ words).
|
| 106 |
-
2. 'Contagion & Cross-Border Impact': How the issue flows between nations.
|
| 107 |
-
3. 'Social Vulnerability Map': Identify risks to minors, women, and ethnic groups globally.
|
| 108 |
-
4. 'Policy Recommendations': High-level moderation and guideline strategies.
|
| 109 |
-
"""
|
| 110 |
-
|
| 111 |
-
# μμ λ λΆλΆ: client.chat.completions.create
|
| 112 |
global_res = client.chat.completions.create(
|
| 113 |
-
model=
|
| 114 |
-
messages=[{"role": "system", "content":
|
| 115 |
-
max_tokens=3000
|
| 116 |
-
temperature=0.3
|
| 117 |
)
|
| 118 |
-
|
| 119 |
-
|
| 120 |
-
st.markdown(f"<div style='background-color: #e1f5fe; padding: 25px; border-radius: 10px; border-left: 5px solid #0288d1;'>{global_report}</div>", unsafe_allow_html=True)
|
| 121 |
|
| 122 |
-
# --- [Tab 2: National Deep-Dive
|
| 123 |
with tab_report:
|
| 124 |
col_d1, col_d2 = st.columns([1, 4])
|
| 125 |
with col_d1:
|
| 126 |
-
st.subheader("Target Selection")
|
| 127 |
target_country = st.selectbox("Market", list(MONITOR_LIST.keys()))
|
| 128 |
-
data_vol = st.slider("
|
| 129 |
run_deep = st.button("π Analyze Market", use_container_width=True)
|
| 130 |
|
| 131 |
if run_deep:
|
| 132 |
conf = MONITOR_LIST[target_country]
|
| 133 |
-
with st.status(f"Llama 3 Deep-Dive: {target_country}...", expanded=True):
|
| 134 |
-
|
| 135 |
-
|
| 136 |
-
feed_text = "\n".join([f"- {n['title']}" for n in g_news + l_news])
|
| 137 |
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
deep_system_prompt = f"""
|
| 141 |
-
Identify Community Guideline violations and risks for {target_country}.
|
| 142 |
-
Structure your report as a JSON object:
|
| 143 |
-
{{
|
| 144 |
-
"risk_score": 0-100,
|
| 145 |
-
"executive_report": "300+ word detailed English report",
|
| 146 |
-
"policy_impact": "Details on which guidelines are violated and why",
|
| 147 |
-
"vulnerability_analysis": "Risk assessment for minors, gender, and ethnicity",
|
| 148 |
-
"action_plan": "5 specific tactical steps"
|
| 149 |
-
}}
|
| 150 |
-
"""
|
| 151 |
-
|
| 152 |
-
# μμ λ λΆλΆ: client.chat.completions.create + response_format
|
| 153 |
deep_res = client.chat.completions.create(
|
| 154 |
-
model=
|
| 155 |
-
messages=[{"role": "system", "content":
|
| 156 |
response_format={"type": "json_object"}
|
| 157 |
)
|
| 158 |
data = json.loads(deep_res.choices[0].message.content)
|
| 159 |
|
| 160 |
-
|
| 161 |
-
st.
|
| 162 |
-
|
| 163 |
-
m1.metric("Risk Score", f"{data['risk_score']}/100")
|
| 164 |
-
|
| 165 |
-
st.subheader("π Executive Summary")
|
| 166 |
-
st.markdown(f"<div style='background-color: #f0f2f6; padding: 20px; border-radius: 10px;'>{data['executive_report']}</div>", unsafe_allow_html=True)
|
| 167 |
-
|
| 168 |
-
|
| 169 |
-
|
| 170 |
-
c_a, c_b = st.columns(2)
|
| 171 |
-
with c_a:
|
| 172 |
-
st.markdown("### π« Policy & Vulnerability")
|
| 173 |
-
st.write(data['policy_impact'])
|
| 174 |
-
st.warning(data['vulnerability_analysis'])
|
| 175 |
-
with c_b:
|
| 176 |
-
st.markdown("### π‘οΈ Tactical Action Plan")
|
| 177 |
-
st.success(data['action_plan'])
|
|
|
|
| 9 |
import plotly.express as px
|
| 10 |
|
| 11 |
# --- Page Config ---
|
| 12 |
+
st.set_page_config(page_title="Global Issue Intelligence Center", page_icon="π‘", layout="wide")
|
| 13 |
|
| 14 |
# --- π API Key Load ---
|
| 15 |
api_key = os.environ.get("GROQ_API_KEY")
|
|
|
|
| 24 |
# Groq ν΄λΌμ΄μΈνΈ μ΄κΈ°ν
|
| 25 |
client = Groq(api_key=api_key)
|
| 26 |
|
| 27 |
+
# μ΅μ λͺ¨λΈλͺ
μ μ (Llama 3.3 70B)
|
| 28 |
+
CURRENT_MODEL = "llama-3.3-70b-versatile"
|
| 29 |
+
|
| 30 |
# --- Monitoring Nations ---
|
| 31 |
MONITOR_LIST = {
|
| 32 |
"South Korea": {"geo": "KR", "lang": "ko"}, "USA": {"geo": "US", "lang": "en"},
|
|
|
|
| 42 |
feed = feedparser.parse(url)
|
| 43 |
return [{"title": e.title, "source": e.source.title if 'source' in e else "G-News"} for e in feed.entries[:limit]]
|
| 44 |
|
| 45 |
+
st.title("π‘ Omni-Watch: Global Intelligence (Llama 3.3)")
|
| 46 |
+
st.caption("Strategic Intelligence for Policy & Risk Management")
|
| 47 |
st.markdown("---")
|
| 48 |
|
| 49 |
+
tab_search, tab_report = st.tabs(["π Global Issue Tracking", "π National Deep-Dive"])
|
| 50 |
|
| 51 |
+
# --- [Tab 1: Global Issue Tracking] ---
|
| 52 |
with tab_search:
|
| 53 |
+
st.subheader("Global Risk Propagation & Deep Intelligence")
|
| 54 |
col_k1, col_k2 = st.columns([3, 1])
|
| 55 |
with col_k1:
|
| 56 |
+
search_keyword = st.text_input("Enter Issue Keyword", "Digital Misinformation")
|
| 57 |
with col_k2:
|
| 58 |
st.write("")
|
| 59 |
run_global = st.button("πΊοΈ Global Deep Scan", type="primary", use_container_width=True)
|
|
|
|
| 65 |
status_msg = st.empty()
|
| 66 |
|
| 67 |
for i, (name, conf) in enumerate(MONITOR_LIST.items()):
|
| 68 |
+
status_msg.text(f"Extracting Intel from {name}...")
|
| 69 |
news = fetch_google_news(f"{name} {search_keyword}", geo=conf['geo'], lang=conf['lang'], limit=3)
|
| 70 |
|
| 71 |
if news:
|
| 72 |
titles = "\n".join([n['title'] for n in news])
|
| 73 |
all_captured_news.append(f"--- {name} News ---\n{titles}")
|
| 74 |
|
|
|
|
| 75 |
try:
|
| 76 |
res = client.chat.completions.create(
|
| 77 |
+
model=CURRENT_MODEL,
|
| 78 |
+
messages=[{"role": "user", "content": f"Score 0-100 for '{search_keyword}' in {name}: {titles}. ONLY INTEGER."}],
|
| 79 |
max_tokens=10
|
| 80 |
)
|
| 81 |
+
score = int(''.join(filter(str.isdigit, res.choices[0].message.content)))
|
|
|
|
| 82 |
except: score = 50
|
| 83 |
results.append({"Country": name, "Risk": score})
|
| 84 |
progress.progress((i + 1) / len(MONITOR_LIST))
|
| 85 |
|
| 86 |
status_msg.empty()
|
| 87 |
df_global = pd.DataFrame(results).sort_values("Risk", ascending=False)
|
| 88 |
+
st.plotly_chart(px.bar(df_global, x="Country", y="Risk", color="Risk", color_continuous_scale="Reds"), use_container_width=True)
|
|
|
|
|
|
|
|
|
|
| 89 |
|
|
|
|
| 90 |
st.divider()
|
| 91 |
+
st.subheader(f"π§ Llama 3.3 Global Intelligence Report")
|
|
|
|
|
|
|
| 92 |
|
| 93 |
+
with st.spinner("Synthesizing Massive Data..."):
|
| 94 |
combined_news = "\n\n".join(all_captured_news)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 95 |
global_res = client.chat.completions.create(
|
| 96 |
+
model=CURRENT_MODEL,
|
| 97 |
+
messages=[{"role": "system", "content": "You are a Global Strategic Intelligence Officer. Analyze issue propagation and demographic risks in 500+ words."}, {"role": "user", "content": combined_news}],
|
| 98 |
+
max_tokens=3000
|
|
|
|
| 99 |
)
|
| 100 |
+
st.markdown(f"<div style='background-color: #e1f5fe; padding: 25px; border-radius: 10px; border-left: 5px solid #0288d1;'>{global_res.choices[0].message.content}</div>", unsafe_allow_html=True)
|
|
|
|
|
|
|
| 101 |
|
| 102 |
+
# --- [Tab 2: National Deep-Dive] ---
|
| 103 |
with tab_report:
|
| 104 |
col_d1, col_d2 = st.columns([1, 4])
|
| 105 |
with col_d1:
|
|
|
|
| 106 |
target_country = st.selectbox("Market", list(MONITOR_LIST.keys()))
|
| 107 |
+
data_vol = st.slider("Depth", 10, 60, 30)
|
| 108 |
run_deep = st.button("π Analyze Market", use_container_width=True)
|
| 109 |
|
| 110 |
if run_deep:
|
| 111 |
conf = MONITOR_LIST[target_country]
|
| 112 |
+
with st.status(f"Llama 3.3 Deep-Dive: {target_country}...", expanded=True):
|
| 113 |
+
news = fetch_google_news(f"{target_country} social controversy", limit=data_vol)
|
| 114 |
+
feed_text = "\n".join([f"- {n['title']}" for n in news])
|
|
|
|
| 115 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
deep_res = client.chat.completions.create(
|
| 117 |
+
model=CURRENT_MODEL,
|
| 118 |
+
messages=[{"role": "system", "content": "Analyze Community Guideline violations & demographics (Minors, Women, Race). Output JSON."}, {"role": "user", "content": feed_text}],
|
| 119 |
response_format={"type": "json_object"}
|
| 120 |
)
|
| 121 |
data = json.loads(deep_res.choices[0].message.content)
|
| 122 |
|
| 123 |
+
st.header(f"π Intelligence Report: {target_country}")
|
| 124 |
+
st.markdown(f"<div style='background-color: #f0f2f6; padding: 20px; border-radius: 10px;'>{data.get('executive_report', 'Processing...')}</div>", unsafe_allow_html=True)
|
| 125 |
+
st.warning(f"**Policy & Demographics Risk:** {data.get('vulnerability_analysis', 'N/A')}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|