"""
DEWA PR Tracker — Streamlit Dashboard
Run: streamlit run dashboard.py
"""
import math
import re as _re
from datetime import datetime, timedelta
import pandas as pd
import plotly.express as px
import plotly.graph_objects as go
import streamlit as st
import database as db
from config import (
COST_PER_PODCAST_OLD_AED,
NOTEBOOKLM_MONTHLY_USD,
NOTEBOOKLM_START_DATE,
USD_TO_AED,
DASHBOARD_TITLE,
PAGE_SIZE,
)
# ── Page setup ────────────────────────────────────────────────────────────────
st.set_page_config(
page_title=DASHBOARD_TITLE,
page_icon="📰",
layout="wide",
initial_sidebar_state="collapsed", # collapsed by default on mobile
)
# ── SVG helpers ───────────────────────────────────────────────────────────────
def _svg(paths, size=16, color="#0077b6"):
return (
f''
)
# Icon paths
_P_DOC = ''
_P_MIC = ''
_P_COIN = ''
_P_BAR = ''
_P_LIST = ''
_P_CAL = ''
_P_LINK = ''
_P_CLK = ''
_P_WAVE = ''
_P_SYNC = ''
def section_title(icon_path, text, icon_color="#0077b6"):
st.markdown(
f''
f'{_svg(icon_path, size=18, color=icon_color)}{text}
',
unsafe_allow_html=True,
)
# ── CSS — light theme + mobile ────────────────────────────────────────────────
st.markdown("""
""", unsafe_allow_html=True)
# ── Data loading (cached 5 min) ───────────────────────────────────────────────
@st.cache_data(ttl=300)
def load_data():
db.init_db()
prs = [dict(r) for r in db.get_all_press_releases()]
pods = [dict(r) for r in db.get_all_podcasts()]
roi_rows = [dict(r) for r in db.get_roi_monthly_since(NOTEBOOKLM_START_DATE)]
stats = db.get_stats()
log = [dict(r) for r in db.get_scrape_log()]
return prs, pods, roi_rows, stats, log
prs, pods, roi_rows, stats, scrape_log = load_data()
df_pr = pd.DataFrame(prs)
df_pod = pd.DataFrame(pods)
if not df_pod.empty and "title" in df_pod.columns:
df_pod["language"] = df_pod["title"].apply(
lambda t: "ar" if _re.search(r"[\u0600-\u06FF]", str(t)) else "en"
)
if not df_pr.empty:
df_pr["date"] = pd.to_datetime(df_pr["date"], errors="coerce")
df_pr["week"] = df_pr["date"].dt.to_period("W").astype(str)
df_pr["month"] = df_pr["date"].dt.to_period("M").astype(str)
if not df_pod.empty:
df_pod["date"] = pd.to_datetime(df_pod["date"], errors="coerce")
# ── ROI pre-calculations ──────────────────────────────────────────────────────
NOTEBOOKLM_AED = NOTEBOOKLM_MONTHLY_USD * USD_TO_AED
NLM_START = pd.Timestamp(NOTEBOOKLM_START_DATE)
df_roi = pd.DataFrame(roi_rows) if roi_rows else pd.DataFrame(
columns=["month","pr_count","en_episodes","ar_episodes","total_episodes"]
)
if not df_roi.empty:
df_roi["old_cost"] = df_roi["pr_count"] * COST_PER_PODCAST_OLD_AED
df_roi["new_cost"] = NOTEBOOKLM_AED
df_roi["savings"] = df_roi["old_cost"] - df_roi["new_cost"]
total_prs_nlm = int(df_roi["pr_count"].sum())
total_old_cost = float(df_roi["old_cost"].sum())
total_new_cost = float(df_roi["new_cost"].sum())
total_savings = float(df_roi["savings"].sum())
roi_pct = (total_savings / total_old_cost * 100) if total_old_cost > 0 else 0
else:
total_prs_nlm = total_old_cost = total_new_cost = total_savings = roi_pct = 0
# ── Sidebar ───────────────────────────────────────────────────────────────────
with st.sidebar:
st.markdown(
'DEWA
'
'Dubai Electricity & Water Authority
',
unsafe_allow_html=True,
)
st.divider()
if not df_pr.empty and df_pr["date"].notna().any():
min_date = df_pr["date"].min().date()
max_date = df_pr["date"].max().date()
else:
min_date = datetime(2019, 1, 1).date()
max_date = datetime.today().date()
date_from, date_to = st.date_input(
"Date range",
value=(NLM_START.date(), max_date),
min_value=min_date, max_value=max_date,
)
categories = ["All"]
if not df_pr.empty and "category" in df_pr.columns:
cats = df_pr["category"].dropna().unique().tolist()
categories += sorted(cats)
selected_cat = st.selectbox("Category", categories)
has_podcast_filter = st.checkbox("Show only PRs with podcast", value=False)
st.divider()
if st.button("Refresh data", use_container_width=True):
st.cache_data.clear()
st.rerun()
if st.button("Run scraper now", use_container_width=True):
with st.spinner("Scraping DEWA website…"):
try:
from scraper import run_scrape
prs_n, pods_n = run_scrape()
st.success(f"Done — {prs_n} PRs, {pods_n} podcasts saved.")
st.cache_data.clear()
st.rerun()
except Exception as e:
st.error(f"Scraper error: {e}")
st.divider()
last_scraped = stats.get('last_scraped','Never') or 'Never'
st.caption(f"Last scraped: {last_scraped}")
# ── Live countdown to next 08:00 Dubai time ───────────────────────────────
st.markdown("""
Next auto-refresh
Daily at 08:00 Dubai time
""", unsafe_allow_html=True)
# ── Apply filters ─────────────────────────────────────────────────────────────
filtered = df_pr.copy()
if not filtered.empty:
if filtered["date"].notna().any():
filtered = filtered[
(filtered["date"].dt.date >= date_from) &
(filtered["date"].dt.date <= date_to)
]
if selected_cat != "All":
filtered = filtered[filtered["category"] == selected_cat]
if has_podcast_filter:
filtered = filtered[filtered["podcast_count"] > 0]
# ═══════════════════════════════════════════════════════════════════════════════
# HEADER
# ═══════════════════════════════════════════════════════════════════════════════
st.markdown(
f''
f'{_svg(_P_DOC, size=30, color="#0077b6")}'
f'
{DASHBOARD_TITLE}
'
f''
f''
f'NotebookLM adoption since {NOTEBOOKLM_START_DATE} · '
f'{datetime.today().strftime("%d %B %Y")}
',
unsafe_allow_html=True,
)
st.markdown("
", unsafe_allow_html=True)
# ═══════════════════════════════════════════════════════════════════════════════
# KPI CARDS
# ═══════════════════════════════════════════════════════════════════════════════
today = datetime.today().date()
week_start = today - timedelta(days=today.weekday())
month_start = today.replace(day=1)
last_month = (month_start - timedelta(days=1)).replace(day=1)
if not filtered.empty and filtered["date"].notna().any():
prs_today = len(filtered[filtered["date"].dt.date == today])
prs_week = len(filtered[filtered["date"].dt.date >= week_start])
prs_month = len(filtered[filtered["date"].dt.date >= month_start])
prs_last_month = len(filtered[
(filtered["date"].dt.date >= last_month) &
(filtered["date"].dt.date < month_start)
])
else:
prs_today = prs_week = prs_month = prs_last_month = 0
prs_with_pod = len(filtered[filtered["podcast_count"] > 0]) if not filtered.empty else 0
coverage_pct = (prs_with_pod / len(filtered) * 100) if len(filtered) > 0 else 0
month_delta = prs_month - prs_last_month
d_cls = "kpi-delta-pos" if month_delta >= 0 else "kpi-delta-neg"
d_sign = "+" if month_delta >= 0 else ""
def kpi_card(color, icon_path, label, value, sub="", icon_color=None):
clr = icon_color or {"blue":"#0077b6","green":"#16a34a","amber":"#d97706","purple":"#7c3aed","rose":"#e11d48"}[color]
ico = _svg(icon_path, size=13, color=clr)
return (
f''
f'
{ico}{label}
'
f'
{value}
'
f'{"
" + sub + "
" if sub else ""}'
f'
'
f'{kpi_card("blue", _P_DOC, "Total Press Releases", f"{len(filtered):,}", f"since {date_from.strftime(chr(37)+chr(100)+chr(32)+chr(37)+chr(98)+chr(32)+chr(37)+chr(89))}")}'
f'{kpi_card("green", _P_MIC, "Podcast Coverage", f"{coverage_pct:.1f}%", f"{prs_with_pod} of {len(filtered)} PRs", "#16a34a")}'
f'{kpi_card("amber", _P_CAL, "This Month", f"{prs_month:,}", f"{d_sign}{month_delta} vs last month", "#d97706")}'
f'{kpi_card("purple",_P_WAVE, "This Week", f"{prs_week:,}", f"since {week_start.strftime(chr(37)+chr(100)+chr(32)+chr(37)+chr(98))}", "#7c3aed")}'
f'{kpi_card("rose", _P_CLK, "Today", f"{prs_today:,}", today.strftime("%d %B %Y"), "#e11d48")}'
f'
',
unsafe_allow_html=True,
)
st.markdown("
", unsafe_allow_html=True)
st.divider()
# ═══════════════════════════════════════════════════════════════════════════════
# ROI — NotebookLM
# ═══════════════════════════════════════════════════════════════════════════════
section_title(_P_COIN, "NotebookLM ROI — Since 23 May 2025", "#16a34a")
months_active = len(df_roi)
st.markdown(
f''
f'
'
f'
{_svg(_P_CLK, 13, "#e11d48")}Traditional Cost
'
f'
AED {total_old_cost:,.0f}
'
f'
{total_prs_nlm} PRs × AED {COST_PER_PODCAST_OLD_AED:,}/podcast
'
f'
'
f'
'
f'
{_svg(_P_MIC, 13, "#16a34a")}NotebookLM Cost
'
f'
AED {total_new_cost:,.0f}
'
f'
{months_active} months × AED {NOTEBOOKLM_AED:.0f}/mo (${NOTEBOOKLM_MONTHLY_USD})
'
f'
'
f'
'
f'
{_svg(_P_COIN, 13, "#d97706")}Total Savings
'
f'
AED {total_savings:,.0f}
'
f'
Since {NOTEBOOKLM_START_DATE}
'
f'
'
f'
'
f'
{_svg(_P_BAR, 13, "#0077b6")}Return on Investment
'
f'
{roi_pct:.1f}%
'
f'
Savings ÷ traditional cost
'
f'
'
f'
'
'AED 20,000 per PR podcast production (EN + AR = one production unit). '
'NotebookLM billed as flat $20/month.
',
unsafe_allow_html=True,
)
rows_html = ""
for _, r in df_roi.iterrows():
rows_html += (
f''
f'| {r["month"]} | '
f'{int(r["pr_count"])} | '
f'{int(r["en_episodes"])} | '
f'{int(r["ar_episodes"])} | '
f'AED {r["old_cost"]:,.0f} | '
f'AED {r["new_cost"]:,.0f} | '
f'AED {r["savings"]:,.0f} | '
f'
'
)
rows_html += (
f''
f'| TOTAL | '
f'{total_prs_nlm} | '
f'{int(df_roi["en_episodes"].sum())} | '
f'{int(df_roi["ar_episodes"].sum())} | '
f'AED {total_old_cost:,.0f} | '
f'AED {total_new_cost:,.0f} | '
f'AED {total_savings:,.0f} | '
f'
'
)
st.markdown(
f''
f'
'
f'| Month | PRs | EN | AR | '
f'Old Cost | New Cost | Savings | '
f'
{rows_html}
'
f'
'
f'
'
f'
{title}'
f'
'
f'{date_str}{pod_badge}'
f'
'
f'{"
" + desc_short + "
" if desc_short else ""}'
f'
'
'DEWA PR Tracker · Playwright · SQLite · Streamlit
',
unsafe_allow_html=True,
)