import os
import json
from statistics import mean
from verified_rates import shop, HOTELS
from html import escape
import math
import time as time_module
import tempfile
from concurrent.futures import ThreadPoolExecutor, as_completed
from datetime import datetime, date, time, timedelta, timezone
from difflib import SequenceMatcher
from statistics import median
from zoneinfo import ZoneInfo
import gradio as gr
import pandas as pd
import requests
APP_NAME = "Hotel Rate Radar"
DEFAULT_HOTEL = "Surf City Inn & Suites"
DEFAULT_MARKET = "Santa Cruz, CA"
DEFAULT_LAT = 36.9741
DEFAULT_LON = -122.0308
DEFAULT_TARGET_CLASS = 2
PRIMARY_COMPETITOR = "Hotel Solares"
PREFERRED_COMPETITORS = [
"Hotel Solares",
"Riverside Inn & Suites Santa Cruz",
"Capri Motel",
"Aqua Inn",
]
TOTAL_ROOMS = 70
TZ = ZoneInfo("America/Los_Angeles")
# Strategic Santa Cruz road points. These are current flow samples, not future traffic forecasts.
TRAFFIC_POINTS = {
"Hwy 17 Summit": (37.1267, -121.9840),
"Hwy 17 Santa Cruz approach": (37.0038, -122.0308),
"Hwy 1 East": (36.9740, -121.9890),
"Hwy 1 West": (36.9800, -122.0550),
}
DEFAULT_OTA = pd.DataFrame(
[
["Hotel direct", None],
["Google Hotels", None],
["Booking.com", None],
["Expedia", None],
["Hotels.com", None],
["Agoda", None],
["Priceline", None],
],
columns=["Channel", "Base rate ($)"],
)
DEFAULT_COMPS = pd.DataFrame(
[
["Hotel Solares", None],
["Riverside Inn & Suites Santa Cruz", None],
["Capri Motel", None],
["Aqua Inn", None],
],
columns=["Comparable hotel", "Base rate ($)"],
)
ROOM_PROFILE = pd.DataFrame(
[
["1 Queen", 4, None, None, 0],
["King Kitchen (KK)", 4, None, None, 0],
["2 Queen Plus", 11, None, None, 0],
["2 Queen Kitchen (2QK)", 5, None, None, 0],
["King Queen Suite (KQS)", 33, None, None, 0],
["3 Queen", 13, None, None, 0],
],
columns=["Room type", "Total rooms", "Rooms left", "Current rate ($)", "Manual adjustment ($)"],
)
# Reference spacing learned from the long-running room-rate pattern the owner shared.
# These are relative offsets, not hard-coded prices for any date.
ROOM_RATE_OFFSETS = {
"1 Queen": 0,
"King Kitchen (KK)": 4,
"2 Queen Plus": 10,
"King Queen Suite (KQS)": 14,
"2 Queen Kitchen (2QK)": 24,
"3 Queen": 40,
}
RATE_CACHE = {}
RATE_CACHE_TTL_SECONDS = 60 * 45
TRAFFIC_CACHE = {"at": 0, "value": None}
TRAFFIC_CACHE_TTL_SECONDS = 60 * 10
WEATHER_CACHE = {"at": 0, "value": None}
WEATHER_CACHE_TTL_SECONDS = 60 * 20
def _safe_float(value, default=None):
try:
if value is None or (isinstance(value, float) and math.isnan(value)):
return default
if isinstance(value, str):
value = value.replace("$", "").replace(",", "").strip()
if not value:
return default
parsed = float(value)
return parsed if math.isfinite(parsed) else default
except (TypeError, ValueError):
return default
def _extract_rates(df, rate_col):
if df is None:
return []
if not isinstance(df, pd.DataFrame):
df = pd.DataFrame(df)
if rate_col not in df.columns:
return []
vals = []
for value in df[rate_col].tolist():
number = _safe_float(value)
if number is not None and 20 <= number <= 3000:
vals.append(number)
return vals
def _price_from_rate_object(rate_obj):
if not isinstance(rate_obj, dict):
return None
# Prefer pre-tax/base rate so comparisons are as apples-to-apples as possible.
for key in ("extracted_before_taxes_fees", "extracted_lowest"):
value = _safe_float(rate_obj.get(key))
if value is not None and 20 <= value <= 3000:
return value
return None
def _property_price(prop):
if not isinstance(prop, dict):
return None
rate = _price_from_rate_object(prop.get("rate_per_night"))
if rate is not None:
return rate
value = _safe_float(prop.get("extracted_price"))
if value is not None and 20 <= value <= 3000:
return value
return None
def _normalise_name(value):
return " ".join("".join(ch.lower() if ch.isalnum() else " " for ch in str(value or "")).split())
def _name_similarity(a, b):
a_n = _normalise_name(a)
b_n = _normalise_name(b)
if not a_n or not b_n:
return 0.0
if a_n in b_n or b_n in a_n:
return 0.95
return SequenceMatcher(None, a_n, b_n).ratio()
def _distance_miles(lat1, lon1, lat2, lon2):
try:
from math import radians, sin, cos, sqrt, atan2
r = 3958.7613
dlat = radians(float(lat2) - float(lat1))
dlon = radians(float(lon2) - float(lon1))
a = sin(dlat / 2) ** 2 + cos(radians(float(lat1))) * cos(radians(float(lat2))) * sin(dlon / 2) ** 2
return 2 * r * atan2(sqrt(a), sqrt(1 - a))
except Exception:
return None
def parse_target_date(value):
if isinstance(value, datetime):
return value.date()
if isinstance(value, date):
return value
if isinstance(value, (int, float)):
return datetime.fromtimestamp(value, TZ).date()
text = str(value or "").strip()
if not text:
raise ValueError("No date selected")
return pd.to_datetime(text).date()
def clamp(v, lo, hi):
return max(lo, min(hi, v))
def get_weather_map():
"""Live Open-Meteo daily forecast, up to its current forecast horizon."""
now = time_module.time()
if WEATHER_CACHE["value"] is not None and now - WEATHER_CACHE["at"] < WEATHER_CACHE_TTL_SECONDS:
return WEATHER_CACHE["value"]
try:
params = {
"latitude": DEFAULT_LAT,
"longitude": DEFAULT_LON,
"daily": "weather_code,temperature_2m_max,temperature_2m_min,precipitation_probability_max",
"temperature_unit": "fahrenheit",
"timezone": "America/Los_Angeles",
"forecast_days": 16,
}
response = requests.get("https://api.open-meteo.com/v1/forecast", params=params, timeout=10)
response.raise_for_status()
daily = response.json().get("daily", {})
out = {}
dates = daily.get("time", [])
for i, day in enumerate(dates):
out[day] = {
"max_f": (daily.get("temperature_2m_max") or [None] * len(dates))[i],
"min_f": (daily.get("temperature_2m_min") or [None] * len(dates))[i],
"precip_pct": (daily.get("precipitation_probability_max") or [None] * len(dates))[i],
"weather_code": (daily.get("weather_code") or [None] * len(dates))[i],
}
WEATHER_CACHE.update({"at": now, "value": out})
return out
except Exception:
return {}
def weather_adjustment(weather):
if not weather:
return 0.0, "No reliable live weather forecast available for this date"
high = _safe_float(weather.get("max_f"), 65)
rain = _safe_float(weather.get("precip_pct"), 0)
code = int(_safe_float(weather.get("weather_code"), 0))
adj = 0.0
if 68 <= high <= 82 and rain <= 20:
adj += 0.04
elif high >= 63 and rain <= 30:
adj += 0.02
if rain >= 70 or code in {65, 67, 82, 95, 96, 99}:
adj -= 0.05
elif rain >= 45:
adj -= 0.025
return adj, f"{high:.0f}°F high, {rain:.0f}% max precipitation"
def get_ticketmaster_events_range(start_date, end_date):
key = os.getenv("TICKETMASTER_API_KEY", "").strip()
if not key:
return {}, "Ticketmaster secret not configured"
try:
local_start = datetime.combine(start_date, time.min, TZ)
local_end = datetime.combine(end_date, time.max.replace(microsecond=0), TZ)
params = {
"apikey": key,
"latlong": f"{DEFAULT_LAT},{DEFAULT_LON}",
"radius": 25,
"unit": "miles",
"startDateTime": local_start.astimezone(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ"),
"endDateTime": local_end.astimezone(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ"),
"size": 200,
"sort": "date,asc",
}
response = requests.get("https://app.ticketmaster.com/discovery/v2/events.json", params=params, timeout=15)
response.raise_for_status()
events = response.json().get("_embedded", {}).get("events", [])
grouped = {}
for event in events:
day = event.get("dates", {}).get("start", {}).get("localDate")
if not day:
continue
venues = event.get("_embedded", {}).get("venues", [])
venue = venues[0].get("name", "") if venues else ""
classification = ""
classes = event.get("classifications", [])
if classes:
classification = classes[0].get("segment", {}).get("name", "")
grouped.setdefault(day, []).append({
"name": event.get("name", "Event"),
"venue": venue,
"classification": classification,
"end_time": event.get("dates", {}).get("end", {}).get("dateTime"),
})
return grouped, "Live Ticketmaster data"
except Exception as exc:
return {}, f"Ticketmaster unavailable: {type(exc).__name__}"
def event_adjustment(events, manual_event_level):
level_map = {
"None / normal day": 0.00,
"Small local activity": 0.02,
"Moderate event demand": 0.06,
"Large event / graduation / festival": 0.12,
"Exceptional sell-out demand": 0.18,
}
manual_adj = level_map.get(manual_event_level, 0.0)
if not events:
return manual_adj, []
score = 0.0
highlights = []
for event in events[:20]:
classification = (event.get("classification") or "").lower()
name = event.get("name", "Event")
venue = event.get("venue", "")
weight = 0.010
if any(k in classification for k in ["music", "sports", "arts"]):
weight = 0.017
text = f"{name}" + (f" — {venue}" if venue else "")
if any(k in text.lower() for k in ["boardwalk", "kaiser permanente arena", "uc santa cruz", "ucsc", "graduation", "commencement", "festival"]):
weight += 0.018
score += weight
highlights.append(text)
live_adj = min(score, 0.16)
return max(manual_adj, live_adj), highlights[:8]
def get_tomtom_traffic():
"""Live Santa Cruz traffic flow. Cached briefly to reduce quota usage."""
key = os.getenv("TOMTOM_API_KEY", "").strip()
if not key:
return None, [], "TomTom secret not configured"
now = time_module.time()
if TRAFFIC_CACHE["value"] is not None and now - TRAFFIC_CACHE["at"] < TRAFFIC_CACHE_TTL_SECONDS:
return TRAFFIC_CACHE["value"]
samples = []
details = []
closures = []
try:
endpoint = "https://api.tomtom.com/traffic/services/4/flowSegmentData/absolute/10/json"
for name, (lat, lon) in TRAFFIC_POINTS.items():
params = {"key": key, "point": f"{lat},{lon}", "unit": "mph", "thickness": 1}
response = requests.get(endpoint, params=params, timeout=8)
response.raise_for_status()
data = response.json().get("flowSegmentData", {})
current = _safe_float(data.get("currentSpeed"))
free = _safe_float(data.get("freeFlowSpeed"))
confidence = _safe_float(data.get("confidence"))
closed = bool(data.get("roadClosure", False))
if current is not None and free and free > 0:
congestion = 1.0 - max(0.0, min(1.0, current / free))
# Downweight low-confidence road samples.
quality = clamp(confidence if confidence is not None else 0.65, 0.25, 1.0)
samples.append((congestion, quality))
details.append(f"{name}: {current:.0f}/{free:.0f} mph" + (" • closure" if closed else ""))
if closed:
closures.append(name)
if not samples:
result = (None, details, "TomTom returned no usable flow segments")
else:
weighted = sum(c * q for c, q in samples) / sum(q for _, q in samples)
result = (weighted, details, "Live TomTom flow data" + (f" • closures: {', '.join(closures)}" if closures else ""))
TRAFFIC_CACHE.update({"at": now, "value": result})
return result
except Exception as exc:
return None, details, f"TomTom unavailable: {type(exc).__name__}"
def traffic_adjustment(auto_congestion, manual_level, target_date):
level_map = {
"Normal": 0.00,
"Busier than normal": 0.015,
"Heavy inbound traffic": 0.03,
"Severe inbound congestion": 0.045,
}
today = datetime.now(TZ).date()
# Current road speeds are live observations, not a forecast. Only let them change today's rate.
if target_date != today:
return 0.0, "Live traffic shown as current context only; no future-date rate impact"
if auto_congestion is None:
return level_map.get(manual_level, 0.0), f"Manual current-day traffic fallback: {manual_level}"
adj = min(max(auto_congestion - 0.10, 0) * 0.12, 0.04)
return adj, f"Sampled road speeds are {auto_congestion * 100:.0f}% below free-flow on average"
def _parse_google_hotel_sources(target_prop, reference_rate=None):
"""Return one lowest comparable base-room offer per booking source.
Google Hotels property details can expose many room/package prices for the same
provider. Those are useful for shopping, but they must not all be averaged as
if they were separate base-room quotes. We keep only the lowest offer per
provider and reject offers that are implausibly far from the property's lowest
comparable rate.
"""
rows = []
if not isinstance(target_prop, dict):
return rows
ref = _safe_float(reference_rate)
if ref is None:
ref = _property_price(target_prop)
for item in (target_prop.get("prices") or []) + (target_prop.get("featured_prices") or []):
if not isinstance(item, dict):
continue
source = item.get("source") or "Google Hotels source"
rate = _price_from_rate_object(item.get("rate_per_night"))
if rate is None:
continue
rows.append({"Channel": str(source).strip(), "Base rate ($)": round(rate, 2)})
property_rate = _property_price(target_prop)
if property_rate is not None:
rows.append({"Channel": "Google Hotels lowest", "Base rate ($)": round(property_rate, 2)})
if ref is None:
ref = property_rate
# One observation per provider: take that provider's lowest comparable quote.
dedup = {}
for row in rows:
source = str(row["Channel"]).strip()
price = _safe_float(row["Base rate ($)"])
if not source or price is None:
continue
key = _normalise_name(source)
if key not in dedup or price < dedup[key]["Base rate ($)"]:
dedup[key] = {"Channel": source, "Base rate ($)": round(price, 2)}
cleaned = list(dedup.values())
if ref is not None and 20 <= ref <= 3000:
# Base-room provider quotes should cluster fairly close to the property's
# lowest public price. Higher suites/packages are deliberately excluded.
low = max(20.0, ref * 0.70)
high = max(ref * 1.40, ref + 40.0)
filtered = [r for r in cleaned if low <= float(r["Base rate ($)"]) <= high]
# Never lose the property-level lowest quote because of a malformed source.
if filtered:
cleaned = filtered
priority = ["official", "surf city", "google hotels", "priceline", "kayak", "booking", "expedia", "hotels com", "agoda"]
def rank(row):
name = _normalise_name(row["Channel"])
for i, token in enumerate(priority):
if token in name:
return i
return len(priority)
return sorted(cleaned, key=lambda r: (rank(r), r["Base rate ($)"]))
def _is_primary_competitor(name):
return _name_similarity(name, PRIMARY_COMPETITOR) >= 0.62
def _hotel_result_from_json(data, hotel_name, max_comps=6):
properties = []
for prop in data.get("properties", []) or []:
if isinstance(prop, dict) and str(prop.get("type", "hotel")).lower() == "hotel":
properties.append(prop)
for ad in data.get("ads", []) or []:
if isinstance(ad, dict):
properties.append(ad)
priced = [prop for prop in properties if _property_price(prop) is not None]
target = None
target_score = 0.0
for prop in priced:
score = _name_similarity(hotel_name, prop.get("name", ""))
if score > target_score:
target, target_score = prop, score
if target_score < 0.55:
target = None
target_price = _property_price(target) if target else None
target_lat = DEFAULT_LAT
target_lon = DEFAULT_LON
if target:
coords = target.get("gps_coordinates", {}) or {}
target_lat = _safe_float(coords.get("latitude"), DEFAULT_LAT)
target_lon = _safe_float(coords.get("longitude"), DEFAULT_LON)
competitors = []
for prop in priced:
if target is not None and prop is target:
continue
name = str(prop.get("name", "Hotel"))
if target is not None and _name_similarity(hotel_name, name) >= 0.85:
continue
price = _property_price(prop)
if price is None:
continue
coords = prop.get("gps_coordinates", {}) or {}
lat = _safe_float(coords.get("latitude"))
lon = _safe_float(coords.get("longitude"))
distance = _distance_miles(target_lat, target_lon, lat, lon) if lat is not None and lon is not None else None
hotel_class = _safe_float(prop.get("hotel_class"))
is_primary = _is_primary_competitor(name)
preferred_match = max((_name_similarity(name, p) for p in PREFERRED_COMPETITORS), default=0.0)
# Prefer the owner's real comp set. If Google gives us too few of them, we
# may use another nearby budget peer, but not a luxury/property-package outlier.
if hotel_class is not None and hotel_class >= 4 and not is_primary:
continue
if preferred_match < 0.58 and not is_primary:
if distance is None or distance > 4.0:
continue
if hotel_class is not None and abs(hotel_class - DEFAULT_TARGET_CLASS) > 1:
continue
# Guard against a suite/package price being mistaken for a hotel's base rate.
if target_price is not None:
lo = max(20.0, target_price * 0.55)
hi = max(target_price * 1.65, target_price + 55.0)
if not (lo <= price <= hi):
continue
class_gap = abs(hotel_class - DEFAULT_TARGET_CLASS) if hotel_class is not None else 0.75
role = "Primary comp" if is_primary else ("Preferred comp" if preferred_match >= 0.58 else "Secondary 2-star/budget comp")
competitors.append({
"Comparable hotel": name,
"Base rate ($)": round(price, 2),
"Role": role,
"Distance (mi)": round(distance, 1) if distance is not None else None,
"Rating": _safe_float(prop.get("overall_rating")),
"Hotel class": int(hotel_class) if hotel_class is not None else None,
"_primary": 0 if is_primary else 1,
"_preferred": 0 if preferred_match >= 0.58 else 1,
"_distance": distance if distance is not None else 99,
"_class_gap": class_gap,
})
competitors.sort(key=lambda x: (x["_primary"], x["_preferred"], x["_class_gap"], x["_distance"]))
chosen = competitors[:max_comps]
for item in chosen:
for k in ("_primary", "_preferred", "_distance", "_class_gap"):
item.pop(k, None)
own_sources = _parse_google_hotel_sources(target, target_price)
return {
"target_found": target is not None,
"matched_name": target.get("name") if target else None,
"match_score": round(target_score, 3),
"target_lowest": target_price,
"property_token": target.get("property_token") if target else None,
"own_sources": own_sources,
"competitors": chosen,
"search_metadata": data.get("search_metadata", {}),
}
def get_live_google_hotel_rates(target_date, hotel_name, market, max_comps=4, use_cache=True, deep_sources=True, fresh=False):
result, status=shop(target_date, fresh=fresh)
status += "\n" + "\n".join(f"- **{a['Hotel']}**: {a['Status']}" for a in result['audit'])
return result,status
def get_live_google_hotel_rates_30(start_date, hotel_name, market, max_comps=4, fresh=False):
results={}; notes=[]
# Sequential calls bound provider pressure; explicit UI opt-in guards quota.
for i in range(30):
day=start_date+timedelta(days=i)
value,status=get_live_google_hotel_rates(day,hotel_name,market,fresh=fresh)
results[day.isoformat()]=value
notes.append(f"{day}: {sum(bool(a['Offers']) for a in value['audit'])}/5 hotels with matched offers" + "\n" + "\n".join(f"- {a['Hotel']}: {a['Status']}" for a in value['audit']))
return results,"\n".join(notes)
def _market_anchor(own_df, comp_df, owner_direct_rate):
own_rates=_extract_rates(own_df,"Base rate ($)")
comp_rates=_extract_rates(comp_df,"Base rate ($)")
own_avg=mean(own_rates) if own_rates else None
comp_avg=mean(comp_rates) if comp_rates else None
manual=_safe_float(owner_direct_rate)
manual=manual if manual is not None and manual > 0 else None
# Manual direct rate is an explicit fallback, never an observed market quote.
anchor=comp_avg if comp_avg is not None else own_avg if own_avg is not None else manual
label=f"Arithmetic mean of {len(comp_rates)}/4 competitor starting rates" if comp_rates else "Surf City advertised-rate average" if own_rates else "Owner-entered fallback (not observed)"
return anchor,own_avg,None,comp_avg,label,own_rates,comp_rates
def room_type_recommendations(base_bar, room_profile_df, occupancy, floor, ceiling, direct_discount, current_base_rate=None, late_high_traffic=False):
"""
Build the room ladder from the property's real pricing pattern.
The ladder uses relative spacing between room types rather than fixed percentages.
When the base rate rises sharply, the spacing widens as it historically does on
peak-demand dates. Inventory pressure is used only when Rooms left is actually
entered; blank inventory is never invented from overall occupancy.
"""
if room_profile_df is None:
room_profile_df = ROOM_PROFILE.copy()
if not isinstance(room_profile_df, pd.DataFrame):
room_profile_df = pd.DataFrame(room_profile_df)
if room_profile_df.empty:
room_profile_df = ROOM_PROFILE.copy()
# At normal rates, use the normal room offsets. As base price rises, widen the
# ladder; this mirrors the long-running pattern in the reference screenshots.
spread_factor = clamp(1.0 + max(float(base_bar) - 100.0, 0.0) / 70.0, 1.0, 3.5)
current_base = _safe_float(current_base_rate)
rows = []
for _, row in room_profile_df.iterrows():
room_type = str(row.get("Room type", "Room")).strip()
total = int(_safe_float(row.get("Total rooms"), 0) or 0)
if total <= 0:
continue
offset = float(ROOM_RATE_OFFSETS.get(room_type, 0))
manual_adj = _safe_float(row.get("Manual adjustment ($)"), 0.0)
# Owner can type an exact current room price. If blank, show a current-rate
# ladder derived from the owner's current base rate.
current_entered = _safe_float(row.get("Current rate ($)"))
if current_entered is not None:
current_rate = current_entered
current_note = "entered"
else:
current_rate = None
current_note = "not entered"
# Do not estimate room scarcity from the overall occupancy slider.
left_raw = _safe_float(row.get("Rooms left"))
if left_raw is None:
left = None
availability_adj = 0.0
availability_note = "Not entered"
else:
left = int(round(clamp(left_raw, 0, total)))
remaining_ratio = left / total if total else 1.0
if left <= 0:
availability_adj = 0.0
availability_note = "Sold out"
elif remaining_ratio <= 0.10:
availability_adj = 0.10
availability_note = f"{left}/{total} left"
elif remaining_ratio <= 0.25:
availability_adj = 0.06
availability_note = f"{left}/{total} left"
elif remaining_ratio <= 0.40:
availability_adj = 0.03
availability_note = f"{left}/{total} left"
elif remaining_ratio >= 0.80:
availability_adj = -0.02
availability_note = f"{left}/{total} left"
else:
availability_adj = 0.0
availability_note = f"{left}/{total} left"
if late_high_traffic and left is not None and 0 < left <= total * .25:
availability_adj = -0.03
availability_note = f"{left}/{total} left; late/high-traffic policy −3%"
if left_raw is not None and left <= 0:
rec_bar = None
direct = None
low = None
high = None
else:
reference_offsets = {
"1 Queen": (0,0,0), "King Kitchen (KK)": (4,30,6),
"2 Queen Plus": (10,50,60), "King Queen Suite (KQS)": (14,70,90),
"2 Queen Kitchen (2QK)": (24,76,106), "3 Queen": (40,100,140)}
a,b,c = reference_offsets.get(room_type,(offset,offset,offset))
base = float(base_bar)
reference_offset = a if base <= 85 else (a+(b-a)*min(1,(base-85)/114) if base <=199 else b+(c-b)*min(1,(base-199)/70))
ladder_rate = base + reference_offset + manual_adj
ladder_rate *= (1.0 + availability_adj)
rec_bar = round(clamp(ladder_rate, floor, ceiling))
direct = round(clamp(rec_bar * (1 - float(direct_discount) / 100.0), floor, ceiling))
low = round(clamp(rec_bar * 0.94, floor, ceiling))
high = round(clamp(rec_bar * 1.07, floor, ceiling))
rows.append({
"Room type": room_type,
"Inventory": total,
"Rooms left": "—" if left is None else left,
"Current rate": f"${current_rate:.0f}" if current_rate is not None else "—",
"Current basis": current_note,
"Reference offset": f"+${offset:.0f}",
"Availability impact": f"{availability_adj:+.0%}" if left_raw is not None and left > 0 else "—",
"Recommended BAR": f"${rec_bar}" if rec_bar is not None else "SOLD OUT",
"Direct rate": f"${direct}" if direct is not None else "—",
"Working range": f"${low}–${high}" if low is not None else "—",
"Availability": availability_note,
})
return pd.DataFrame(rows)
def occupancy_adjustment(occupancy):
if occupancy is None: return 0.0
occ = (float(occupancy) if occupancy is not None else 65.0)
if occ < 40:
return -0.10
if occ < 55:
return -0.06
if occ < 70:
return 0.00
if occ < 80:
return 0.05
if occ < 90:
return 0.10
if occ < 96:
return 0.17
return 0.24
def pickup_adjustment(pickup_points):
if pickup_points is None: return 0.0
p = (float(pickup_points) if pickup_points is not None else 0.0)
if p <= -5:
return -0.05
if p < 2:
return 0.00
if p < 6:
return 0.03
if p < 12:
return 0.06
return 0.10
def day_adjustment(target_date):
wd = target_date.weekday()
if wd == 4:
return 0.06
if wd == 5:
return 0.10
if wd == 6:
return 0.02
return 0.00
def lead_time_adjustment(target_date, occupancy):
today = datetime.now(TZ).date()
lead = (target_date - today).days
if occupancy is None: return 0.0, lead
occ = (float(occupancy) if occupancy is not None else 65.0)
if lead < 0:
return 0.0, lead
if lead <= 2:
if occ >= 90:
return 0.08, lead
if occ < 55:
return -0.10, lead
if lead <= 7 and occ >= 82:
return 0.05, lead
return 0.0, lead
def _build_rate_inputs(live_result, manual_ota_df, manual_comp_df, owner_direct_rate):
source = "Manual fallback"
live_own = list((live_result or {}).get("own_sources") or [])
live_comps = list((live_result or {}).get("competitors") or [])
if live_own:
own_df = pd.DataFrame(live_own)
source = "Automatic Google Hotels rate shop"
else:
own_df = manual_ota_df.copy() if isinstance(manual_ota_df, pd.DataFrame) else pd.DataFrame(manual_ota_df or [])
if live_comps:
comp_df = pd.DataFrame(live_comps)
source = "Automatic Google Hotels rate shop"
else:
comp_df = manual_comp_df.copy() if isinstance(manual_comp_df, pd.DataFrame) else pd.DataFrame(manual_comp_df or [])
if "Channel" not in own_df.columns:
own_df = pd.DataFrame(columns=["Channel", "Base rate ($)"])
if "Base rate ($)" not in own_df.columns:
own_df["Base rate ($)"] = None
if "Comparable hotel" not in comp_df.columns:
comp_df = pd.DataFrame(columns=["Comparable hotel", "Base rate ($)"])
if "Base rate ($)" not in comp_df.columns:
comp_df["Base rate ($)"] = None
own_rates = _extract_rates(own_df, "Base rate ($)")
comp_rates = _extract_rates(comp_df, "Base rate ($)")
return own_df, comp_df, own_rates, comp_rates, source
def rate_movement_html(current_rate, competitor_rate, recommended, low, high, title="Rate movement"):
current_rate = _safe_float(current_rate)
competitor_rate = _safe_float(competitor_rate)
recommended = _safe_float(recommended)
low = _safe_float(low)
high = _safe_float(high)
vals = [v for v in [current_rate, competitor_rate, recommended, low, high] if v is not None]
if not vals or recommended is None:
return ""
axis_min = min(vals)
axis_max = max(vals)
span = max(axis_max - axis_min, 10)
pad = max(5, span * 0.12)
axis_min -= pad
axis_max += pad
def pct(v):
if v is None:
return None
return clamp((float(v) - axis_min) / (axis_max - axis_min) * 100.0, 2.0, 98.0)
base = current_rate if current_rate is not None else competitor_rate
delta = recommended - base if base is not None else 0
if delta > 0.99:
arrow, action, cls = "↑", f"Raise ${delta:.0f}", "move-up"
elif delta < -0.99:
arrow, action, cls = "↓", f"Lower ${abs(delta):.0f}", "move-down"
else:
arrow, action, cls = "→", "Hold rate", "move-flat"
markers = []
if current_rate is not None:
markers.append(
f''
f'
Current rate ${current_rate:.0f}
'
f'
Comp avg ${competitor_rate:.0f}
'
f'
{arrow} ${recommended:.0f}{action}
'
f'
${low:.0f}Suggested range${high:.0f}
'
return (
f''
f'
{escape(title)}
'
f'{range_text}'
f'
'
f'
Daadu not happy 😅
'
return 'Daadu is happy 😄
'
def money_or_na(value):
return f"${value:.0f}" if value is not None else "—"
def hour_to_label(hour_value):
hour = int(_safe_float(hour_value, 15)) % 24
return datetime(2000, 1, 1, hour, 0).strftime("%I:00 %p")
def parse_hour_value(hour_value):
if isinstance(hour_value, (int, float)):
return int(hour_value) % 24
raw = str(hour_value or "").strip()
if raw.isdigit():
return int(raw) % 24
try:
return datetime.strptime(raw, "%I:%M %p").hour
except ValueError:
try:
return datetime.strptime(raw, "%H:%M").hour
except ValueError as exc:
raise ValueError("Select a valid pricing time") from exc
HOUR_CHOICES = [hour_to_label(h) for h in range(24)]
def hourly_context(target_date, mode, hour, restaurant_close, event_end, closure_discount, now=None, late_reduction=2.0):
now = now or datetime.now(TZ)
parsed_hour = parse_hour_value(hour)
selected = now if mode == "Live now" and target_date == now.date() else datetime.combine(
target_date, time(parsed_hour), TZ
)
live = mode == "Live now" and target_date == now.date()
note = f"Pricing time: {selected.strftime('%Y-%m-%d %I:%M %p %Z')}. "
note += "Current traffic may influence this rate. " if live else "Scenario: current traffic excluded; manual traffic estimate used. "
closed = []
# A clock time before 06:00 belongs to the end of this evening's trading window.
for label, value in (("Restaurants", restaurant_close), ("Local event", event_end)):
if value and str(value).strip():
try:
closing = time.fromisoformat(str(value).strip())
except ValueError as exc:
raise ValueError(f"{label}: enter closing time as HH:MM or leave blank") from exc
end = datetime.combine(target_date, closing, TZ)
if closing.hour < 6:
end += timedelta(days=1)
if selected >= end:
closed.append(label)
late_hours = max(0, selected.hour + selected.minute / 60 - 18)
late_discount = min(10.0, late_hours * max(0, float(late_reduction))) / 100
adjustment = -min(0.20, late_discount + (abs(float(closure_discount)) / 100 if closed else 0.0))
note += f"Late-day policy: {late_reduction:g}% per hour after 6 PM, capped at 10%; applied reduction {late_discount:.1%}. "
if closed:
note += f"Entered closing times passed: {', '.join(closed)}; manual adjustment {adjustment:+.1%}. "
else:
note += "No entered closing time has passed. "
note += "Opening hours are manual; unknown event end times are not guessed."
return {"selected": selected, "live": live, "adjustment": adjustment, "note": note}
def compute_night(
target_date,
live_rate_result,
manual_ota_df,
manual_comp_df,
owner_direct_rate,
occupancy,
pickup_points,
manual_event_level,
manual_traffic_level,
floor,
ceiling,
direct_discount,
weather_map,
events_by_date,
event_source,
traffic_tuple,
time_context=None,
):
own_df, comp_df, own_rates, comp_rates, rate_source = _build_rate_inputs(
live_rate_result, manual_ota_df, manual_comp_df, owner_direct_rate
)
entered_rate = _safe_float(owner_direct_rate)
owner_direct_rate = entered_rate if entered_rate is not None and entered_rate > 0 else None
reference_prices = dict(zip(
[(date(2026,9,20)+timedelta(days=i)).isoformat() for i in range(14)],
[85,85,85,85,85,199,269,85,85,85,85,85,199,229]))
reference_base = reference_prices.get(target_date.isoformat(),85)
reference_label = ("ASI screenshot reference · " + target_date.isoformat() + " · not live") if target_date.isoformat() in reference_prices else "ASI $85 weekday baseline · estimate, not a quote for this date"
anchor, own_median, primary_comp_rate, comp_benchmark, anchor_source, own_rates, comp_rates = _market_anchor(
own_df, comp_df, owner_direct_rate
)
reference_fallback = anchor is None
if reference_fallback:
anchor = reference_base
anchor_source = reference_label
rate_source = "Reference-based estimate; no current market quotes"
own_low = min(own_rates) if own_rates else None
weather = weather_map.get(target_date.isoformat())
weather_adj, weather_text = weather_adjustment(weather)
events = events_by_date.get(target_date.isoformat(), [])
if time_context:
active_events = []
for event in events:
end = event.get("end_time")
try:
ended = end and datetime.fromisoformat(end.replace("Z", "+00:00")).astimezone(TZ) <= time_context["selected"]
except (ValueError, TypeError):
ended = False
if not ended:
active_events.append(event)
events = active_events
event_adj, highlights = event_adjustment(events, manual_event_level)
auto_congestion, traffic_details, traffic_source = traffic_tuple
traffic_adj, traffic_text = traffic_adjustment(auto_congestion, manual_traffic_level, target_date)
if time_context and not time_context["live"]:
traffic_adj = {"Normal": 0, "Busier than normal": .015, "Heavy inbound traffic": .03, "Severe inbound congestion": .045}.get(manual_traffic_level, 0)
traffic_text = f"Manual scenario traffic estimate: {manual_traffic_level}; current road speeds excluded"
late_high_traffic = bool(time_context and time_context["selected"].hour >= 18 and traffic_adj >= .03)
if late_high_traffic:
traffic_adj = 0.0
traffic_text += "; late-day traffic uplift disabled; rooms with ≤25% available receive a 3% reduction"
time_adj = time_context["adjustment"] if time_context else 0.0
occ_adj = occupancy_adjustment(occupancy)
pick_adj = pickup_adjustment(pickup_points)
dow_adj = day_adjustment(target_date)
lead_adj, lead_days = lead_time_adjustment(target_date, occupancy)
total_adj = clamp(
occ_adj + pick_adj + dow_adj + lead_adj + event_adj + weather_adj + traffic_adj,
-0.20,
0.30,
)
raw_bar = anchor * (1 + total_adj)
# Because Surf City is positioned as a 2-star motel equal to Hotel Solares next door,
# keep the base room recommendation reasonably close to Solares unless occupancy/events are exceptional.
if primary_comp_rate is not None:
low_mult = 0.78 if (float(occupancy) if occupancy is not None else 65.0) < 55 else 0.84
high_mult = 1.28 if (float(occupancy) if occupancy is not None else 65.0) >= 90 or event_adj >= 0.12 else 1.16
raw_bar = clamp(raw_bar, primary_comp_rate * low_mult, primary_comp_rate * high_mult)
# Market guardrails prevent a noisy signal from sending the estimate far outside the observed comp set.
if comp_rates:
q25 = float(pd.Series(comp_rates).quantile(0.25))
q75 = float(pd.Series(comp_rates).quantile(0.75))
occ = (float(occupancy) if occupancy is not None else 65.0)
market_low = max(min(comp_rates) * 0.85, q25 * 0.88)
market_high_multiplier = 1.18 if occ >= 90 else 1.10
market_high = max(max(comp_rates), q75) * market_high_multiplier
raw_bar = clamp(raw_bar, market_low, market_high)
# Final sanity guardrail around the observable market. At ordinary occupancy,
# a recommendation cannot leap far above both the current rate and comparable
# public rates. Exceptional occupancy/event conditions get more headroom.
observed_refs = [v for v in [own_median, primary_comp_rate, comp_benchmark, _safe_float(owner_direct_rate)] if v is not None]
if observed_refs:
market_center = median(observed_refs)
if (float(occupancy) if occupancy is not None else 65.0) >= 90 or event_adj >= 0.12:
max_mult = 1.45
elif (float(occupancy) if occupancy is not None else 65.0) >= 80 or event_adj >= 0.06:
max_mult = 1.30
else:
max_mult = 1.20
max_allowed = max(market_center * max_mult, market_center + 20.0)
min_allowed = max(floor, market_center * 0.78)
raw_bar = clamp(raw_bar, min_allowed, max_allowed)
raw_bar *= (1.0 + time_adj) # Apply the time reduction once, after market guardrails.
recommended_bar = round(clamp(raw_bar, floor, ceiling))
direct_rate = round(clamp(recommended_bar * (1 - float(direct_discount) / 100.0), floor, ceiling))
safe_low = round(clamp(recommended_bar * 0.94, floor, ceiling))
safe_high = round(clamp(recommended_bar * 1.07, floor, ceiling))
parity = ((recommended_bar - own_low) / own_low) * 100 if own_low else None
live_rate_count = len(own_rates) + len(comp_rates)
live_source_points = 0
if live_rate_result:
live_source_points += 3
if weather:
live_source_points += 1
if event_source.startswith("Live"):
live_source_points += 1
if traffic_source.startswith("Live") and target_date == datetime.now(TZ).date():
live_source_points += 1
if live_rate_count >= 6 and live_source_points >= 4:
confidence = "High"
conf_score = 88
elif live_rate_count >= 4:
confidence = "Medium"
conf_score = 72
else:
confidence = "Low"
conf_score = 52
confidence = "Hotel/date/tax checked; room and cancellation unverified" if (live_rate_result or {}).get("own_sources") else "No matched Surf City price; inspect source status"
return {
"pricing_time": time_context["selected"].isoformat() if time_context else "",
"generated_at": datetime.now(TZ).isoformat(),
"time_note": time_context["note"] if time_context else "",
"closing_adjustment_pct": round(time_adj * 100, 1),
"hotel": DEFAULT_HOTEL,
"checkin_date": target_date.isoformat(),
"recommended_bar": recommended_bar,
"current_direct_rate": _safe_float(owner_direct_rate),
"reference_base": reference_base,
"reference_label": reference_label,
"reference_fallback": reference_fallback,
"source_audit_json": json.dumps((live_rate_result or {}).get("audit", [])),
"direct_rate": direct_rate,
"working_low": safe_low,
"working_high": safe_high,
"market_anchor": round(anchor, 2),
"hotel_channel_average": round(own_median, 2) if own_median is not None else None,
"competitor_average": round(comp_benchmark, 2) if comp_benchmark is not None else None,
"occupancy_pct": occupancy,
"pickup_points": pickup_points,
"event_adjustment_pct": round(event_adj * 100, 1),
"weather_adjustment_pct": round(weather_adj * 100, 1),
"traffic_adjustment_pct": round(traffic_adj * 100, 1),
"total_adjustment_pct": round(total_adj * 100, 1),
"confidence": confidence,
"evidence_note": "Heuristic, not a calibrated probability",
"rate_source": f"{rate_source}; {anchor_source}",
"primary_competitor": PRIMARY_COMPETITOR,
"primary_comp_rate": round(primary_comp_rate, 2) if primary_comp_rate is not None else None,
"weather_text": weather_text,
"event_source": event_source,
"event_highlights": highlights,
"traffic_text": traffic_text,
"traffic_source": traffic_source,
"traffic_details": traffic_details,
"own_df": own_df,
"comp_df": comp_df,
"parity": parity,
"lead_days": lead_days,
"occ_adj": occ_adj,
"pick_adj": pick_adj,
"dow_adj": dow_adj,
"lead_adj": lead_adj,
"event_adj": event_adj,
"weather_adj": weather_adj,
"traffic_adj": traffic_adj,
"late_high_traffic": late_high_traffic,
}
def quick_tomorrow():
return (datetime.now(TZ).date() + timedelta(days=1)).isoformat(), "Selected date"
def quick_next_30_days():
return (datetime.now(TZ).date() + timedelta(days=1)).isoformat(), "30-day forecast"
def quick_next_weekend():
today = datetime.now(TZ).date()
days_ahead = (4 - today.weekday()) % 7
if days_ahead == 0:
days_ahead = 7
return (today + timedelta(days=days_ahead)).isoformat(), "Selected date"
def _status_markdown(rate_status, event_status, traffic_status, weather_map):
stamp = datetime.now(TZ).strftime("%b %d, %Y • %I:%M %p PT")
serp_ok = "✅" if os.getenv("SERPAPI_API_KEY", "").strip() else "⚠️"
tm_ok = "✅" if os.getenv("TICKETMASTER_API_KEY", "").strip() else "⚠️"
tt_ok = "✅" if os.getenv("TOMTOM_API_KEY", "").strip() else "⚠️"
weather_ok = "✅" if weather_map else "⚠️"
return f"""
### Live data status
**Last refresh:** {stamp}
- {serp_ok} **Competitor rates:** {rate_status}
- {tt_ok} **Traffic:** {traffic_status}
- {tm_ok} **Events:** {event_status}
- {weather_ok} **Weather:** {'Open-Meteo live forecast loaded' if weather_map else 'Open-Meteo unavailable'}
"""
def analyze_selection(
view_mode,
checkin_value,
hotel_name,
market,
auto_rates,
max_comps,
owner_direct_rate,
manual_ota_df,
manual_comp_df,
room_profile_df,
occupancy,
pickup_points,
manual_event_level,
manual_traffic_level,
price_floor,
price_ceiling,
direct_discount,
time_mode="Selected time",
pricing_hour=15,
restaurant_close="",
event_end="",
closure_discount=0,
fresh_prices=False,
allow_bulk=False,
late_reduction=2.0,
):
if view_mode == "30-day forecast" and auto_rates and not allow_bulk:
return "Enable 30-day API queries in Sources & refresh first (up to 300 searches).
", pd.DataFrame(), pd.DataFrame(), "", pd.DataFrame(), "", None, "No pricing calls made."
try:
start_date = datetime.now(TZ).date() if time_mode == "Live now" else parse_target_date(checkin_value)
hourly_context(start_date, time_mode, pricing_hour, restaurant_close, event_end, closure_discount, late_reduction=late_reduction)
except Exception:
return "### ⚠️ Select a valid date/time and use HH:MM for closing times.", pd.DataFrame(), pd.DataFrame(), "", pd.DataFrame(), "", None, ""
hotel_name = (hotel_name or DEFAULT_HOTEL).strip()
market = (market or DEFAULT_MARKET).strip()
floor = _safe_float(price_floor, 59.0)
ceiling = _safe_float(price_ceiling, 399.0)
if floor >= ceiling:
return "### Minimum rate must be lower than maximum rate.", pd.DataFrame(), pd.DataFrame(), "", pd.DataFrame(), "", None, ""
if auto_rates and start_date < datetime.now(TZ).date():
return empty_dashboard("Choose today or a future stay date"), pd.DataFrame(), pd.DataFrame(), "", pd.DataFrame(), "", None, "Public rate searches cannot verify past stays."
if floor <= 0 or ceiling <= 0:
return empty_dashboard("Minimum and maximum rates must be positive"), pd.DataFrame(), pd.DataFrame(), "", pd.DataFrame(), "", None, "No pricing calls made."
weather_map = get_weather_map()
end_date = start_date if view_mode == "Selected date" else start_date + timedelta(days=29)
events_by_date, event_status = get_ticketmaster_events_range(start_date, end_date)
traffic_tuple = get_tomtom_traffic()
traffic_status = traffic_tuple[2]
live_map = {}
rate_status = "Automatic rate shopping disabled; manual fallback used"
if auto_rates:
if view_mode == "Selected date":
live, rate_status = get_live_google_hotel_rates(start_date, hotel_name, market, max_comps=int(max_comps), fresh=fresh_prices)
if live:
live_map[start_date.isoformat()] = live
else:
live_map, rate_status = get_live_google_hotel_rates_30(start_date, hotel_name, market, max_comps=int(max_comps), fresh=fresh_prices)
if view_mode == "Selected date":
night = compute_night(
start_date,
live_map.get(start_date.isoformat()),
manual_ota_df,
manual_comp_df,
owner_direct_rate,
occupancy,
pickup_points,
manual_event_level,
manual_traffic_level,
floor,
ceiling,
direct_discount,
weather_map,
events_by_date,
event_status,
traffic_tuple,
hourly_context(start_date, time_mode, pricing_hour, restaurant_close, event_end, closure_discount, late_reduction=late_reduction),
)
if not night:
return (
empty_dashboard("No current quote for this stay") + source_cards(live_map.get(start_date.isoformat())),
pd.DataFrame(), pd.DataFrame(), "", pd.DataFrame(), "", None,
_status_markdown(rate_status, event_status, traffic_status, weather_map),
)
own_df = night["own_df"].copy()
comp_df = night["comp_df"].copy()
for col in ["Role", "Distance (mi)", "Rating", "Hotel class"]:
if col not in comp_df.columns:
comp_df[col] = None
room_prices = room_type_recommendations(
night["recommended_bar"], room_profile_df, occupancy, floor, ceiling, direct_discount, owner_direct_rate, late_high_traffic=night["late_high_traffic"]
)
parity_text = ""
if night["parity"] is not None:
if night["parity"] > 3:
parity_text = f"The recommendation is **{night['parity']:.0f}% above** the lowest observed public rate for the hotel."
elif night["parity"] < -3:
parity_text = f"The recommendation is **{abs(night['parity']):.0f}% below** the lowest observed public rate for the hotel."
else:
parity_text = "The recommendation is close to the lowest observed public hotel rate."
movement = rate_movement_html(
night["current_direct_rate"],
night["competitor_average"],
night["recommended_bar"],
night["working_low"],
night["working_high"],
)
result_html = f"""
{start_date.strftime('%A, %B %d, %Y')}
{"Current direct rate (entered)" if night["current_direct_rate"] is not None else "ASI reference · not live"}{money_or_na(night["current_direct_rate"] if night["current_direct_rate"] is not None else night["reference_base"])}
{"Competitor average" if night["competitor_average"] is not None else "Pricing baseline · estimate"}{money_or_na(night["competitor_average"] if night["competitor_average"] is not None else night["market_anchor"])}
Suggested base rate · model${night['recommended_bar']}{daadu_note(night['recommended_bar'])}
{movement}
Working range: ${night['working_low']}–${night['working_high']} • Suggested direct rate: ${night['direct_rate']}
Observed source average: {money_or_na(night['hotel_channel_average'])} • Source checks: {night['confidence']}
"""
if night["reference_fallback"]:
result_html += 'Reference estimate' + escape(night["reference_label"]) + '. Live prices could not be verified. Time and demand rules adjust this reference; competitor prices have not been invented.
'
result_html += source_cards(live_map.get(start_date.isoformat()))
rows = [
["Market anchor", f"${night['market_anchor']:.0f}", "+0.0%", night["rate_source"]],
["Occupancy", f"${night['market_anchor']:.0f}", f"{night['occ_adj']:+.1%}", (f"Entered occupancy {float(occupancy):.0f}%" if occupancy is not None else "Not entered; neutral adjustment")],
["Booking pickup", f"${night['market_anchor']:.0f}", f"{night['pick_adj']:+.1%}", (f"Entered pickup {float(pickup_points):+.1f} pts" if pickup_points is not None else "Not entered; neutral adjustment")],
["Day of week", f"${night['market_anchor']:.0f}", f"{night['dow_adj']:+.1%}", start_date.strftime("%A")],
["Lead time", f"${night['market_anchor']:.0f}", f"{night['lead_adj']:+.1%}", f"{night['lead_days']} days before check-in"],
["Events", f"${night['market_anchor']:.0f}", f"{night['event_adj']:+.1%}", night["event_source"]],
["Weather", f"${night['market_anchor']:.0f}", f"{night['weather_adj']:+.1%}", night["weather_text"]],
["Closing-time adjustment", f"${night['market_anchor']:.0f}", f"{night['closing_adjustment_pct']:+.1f}%", night["time_note"]],
["Traffic", f"${night['market_anchor']:.0f}", f"{night['traffic_adj']:+.1%}", f"{night['traffic_text']}; {night['traffic_source']}"],
]
breakdown = pd.DataFrame(rows, columns=["Signal", "Anchor", "Rate impact", "Evidence"])
event_md = ""
if night["event_highlights"]:
event_md = "\n\n**Nearby events detected:**\n" + "\n".join(f"- {x}" for x in night["event_highlights"])
traffic_md = ""
if night["traffic_details"]:
traffic_md = "\n\n**Current traffic samples:** " + "; ".join(night["traffic_details"])
explanation = f"""
**Time note:** {night["time_note"]}
### Why this rate ${night['recommended_bar']}
The pricing anchor is **${night['market_anchor']:.0f}**. The competitor figure is an arithmetic mean of available hotel starting-rate offers, before tax. Missing hotels are excluded. Your current direct rate is **{money_or_na(night['current_direct_rate'])}**. Surf City's observed-source average is **{money_or_na(night['hotel_channel_average'])}**; the competitor average is **{money_or_na(night['competitor_average'])}**. Missing prices are excluded. Demand adjustment: **{night['total_adjustment_pct']:+.1f}%**. Advertised offers may have unverified room/cancellation details; review before changing prices.
The room-type table uses the property's fixed **70-room inventory** (4 1 Queen, 4 King Kitchen, 11 Two Queen Plus, 5 Two Queen Kitchen, 33 KQS, 13 Three Queen). Enter actual rooms-left counts when you know them; if a rooms-left cell is blank, the app estimates remaining inventory from the overall occupancy percentage.
{parity_text}
**Important:** this is a **best estimate, not an exact or guaranteed selling price**. Public OTA prices can differ by room type, cancellation rules, taxes, loyalty status and device/user targeting. Occupancy and booking pickup remain owner-entered because public sites do not know the hotel's true rooms-on-the-books.
{event_md}{traffic_md}
"""
export = pd.DataFrame([{k: v for k, v in night.items() if k not in {"own_df", "comp_df", "event_highlights", "traffic_details"}}])
tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".csv", prefix="hotel-rate-live-")
export.to_csv(tmp.name, index=False)
tmp.close()
status = _status_markdown(rate_status, event_status, traffic_status, weather_map)
return result_html, own_df, comp_df, compact_room_table(room_prices), breakdown, explanation, tmp.name, status
# 30-day forecast
rows = []
for offset in range(30):
target_date = start_date + timedelta(days=offset)
night = compute_night(
target_date,
live_map.get(target_date.isoformat()),
manual_ota_df,
manual_comp_df,
owner_direct_rate,
occupancy,
pickup_points,
manual_event_level,
manual_traffic_level,
floor,
ceiling,
direct_discount,
weather_map,
events_by_date,
event_status,
traffic_tuple,
hourly_context(target_date, time_mode, pricing_hour, restaurant_close, event_end, closure_discount, late_reduction=late_reduction),
)
if night:
rows.append(night)
if not rows:
return (
"### No current quotes · Check Source status or enter manual rates.",
pd.DataFrame(), pd.DataFrame(), "", pd.DataFrame(), "", None,
_status_markdown(rate_status, event_status, traffic_status, weather_map),
)
export_rows = []
for night in rows:
export_rows.append({k: v for k, v in night.items() if k not in {"own_df", "comp_df", "event_highlights", "traffic_details"}})
export = pd.DataFrame(export_rows)
display = pd.DataFrame({
"Date": pd.to_datetime(export["checkin_date"]).dt.strftime("%a %b %d"),
"Recommended BAR": export["recommended_bar"].map(lambda x: f"${x:.0f}"),
"Direct rate": export["direct_rate"].map(lambda x: f"${x:.0f}"),
"Working range": [f"${lo:.0f}–${hi:.0f}" for lo, hi in zip(export["working_low"], export["working_high"])],
"Market anchor": export["market_anchor"].map(lambda x: f"${x:.0f}"),
"Demand adj.": export["total_adjustment_pct"].map(lambda x: f"{x:+.1f}%"),
"Confidence": export["confidence"],
"Pricing time": export["pricing_time"],
"Time note": export["time_note"],
})
avg_bar = round(export["recommended_bar"].mean())
low_bar = int(export["recommended_bar"].min())
high_bar = int(export["recommended_bar"].max())
peak_idx = export["recommended_bar"].idxmax()
peak_date = pd.to_datetime(export.loc[peak_idx, "checkin_date"]).strftime("%A, %b %d")
peak_rate = int(export.loc[peak_idx, "recommended_bar"])
avg_current = _safe_float(owner_direct_rate)
avg_comp = float(export["competitor_average"].dropna().mean()) if export["competitor_average"].notna().any() else None
movement = rate_movement_html(avg_current, avg_comp, avg_bar, low_bar, high_bar, title="30-day average rate movement")
result_html = f"""
30-night live-market forecast • {start_date.strftime('%b %d')} – {(start_date + timedelta(days=29)).strftime('%b %d, %Y')}
AVERAGE RECOMMENDED BAR
${avg_bar}
{daadu_note(avg_bar)}
{movement}
Nightly recommendations: ${low_bar}–${high_bar} • Peak: {peak_date} ${peak_rate}
Each night is priced separately using the rate shop available for that date.
"""
explanation = f"""
### 30-day pricing outlook
Selected time: **{hour_to_label(parse_hour_value(pricing_hour))} America/Los_Angeles** (the first night uses current time in Live now mode).
Occupancy, pickup, manual rates and closing hours are shared scenario assumptions across these dates. Blank occupancy/pickup have no adjustment; manual prices are not live quotes. The room-type table previews the first night only.
The app attempted live Google Hotels rate shopping for each of the 30 nights, then combined the available public hotel/competitor rates with occupancy, pickup, events, weekday/weekend and available weather.
**API-quota note:** one uncached 30-day refresh can use up to **300 SerpApi searches**. Use the refresh only when you need a new market snapshot. The app caches each date for 60 minutes, and SerpApi also documents one-hour caching for identical searches.
**Traffic:** TomTom is truly live, so current road conditions are displayed but are **not used as if they were a forecast** for future nights. This prevents today's Hwy 17 congestion from incorrectly changing a rate three weeks from now.
**Weather:** Open-Meteo's standard live forecast reaches up to 16 days, so later nights receive no weather adjustment rather than an invented forecast.
"""
tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".csv", prefix="hotel-rate-30day-live-")
export.to_csv(tmp.name, index=False)
tmp.close()
# Show the start night's auto rate shop in the two live tables as a helpful preview.
first = rows[0]
own_df = first["own_df"].copy()
comp_df = first["comp_df"].copy()
room_prices = room_type_recommendations(
first["recommended_bar"], room_profile_df, occupancy, floor, ceiling, direct_discount, owner_direct_rate, late_high_traffic=first["late_high_traffic"]
)
status = _status_markdown(rate_status, event_status, traffic_status, weather_map)
return result_html, own_df, comp_df, compact_room_table(room_prices), display, explanation, tmp.name, status
def compact_room_table(frame):
if frame.empty:
return "No room recommendations available.
"
cols = ["Room type", "Inventory", "Rooms left", "Current rate", "Recommended BAR"]
headers = ["Room type", "Total rooms", "Rooms left", "Current (entered)", "Suggested (model)"]
head = "".join(f"{x} | " for x in headers)
rows = "".join(
"" + "".join(f"| {escape(str(row[c]))} | " for c in cols) + "
"
for _, row in frame.iterrows()
)
return f''
def empty_dashboard(message="Choose your stay. See the market clearly."):
return ('YOUR PRICING WORKSPACE
'
f'
{escape(message)}
Choose a date and refresh to compare public offers. '
'Your current rate and hotel occupancy stay under your control.
'
'
Current entered rate—
'
'
Observed market—
'
'
Suggested rate · model—
'
'
No sample prices. Every observation has a source and timestamp.
{escape(item["Hotel"])}'
f'{money_or_na(price)}{badge}'
f'{escape(stamp)}
')
return 'Public market snapshot Hotel starting rates · nightly · see tax basis
'+''.join(cards)+'
'
CUSTOM_CSS = """
.gradio-container {max-width:1180px!important;background:linear-gradient(180deg,#eef6ff 0%,#eaf3ff 100%)!important}
#hero{background:linear-gradient(135deg,#102a52 0%,#173c75 58%,#2460c7 100%);color:white;border-radius:14px;padding:22px 24px;margin-bottom:12px;box-shadow:0 8px 22px #254f8420;position:relative;overflow:hidden;min-height:120px;display:flex;flex-direction:column;justify-content:center}
#hero::after{content:"";position:absolute;inset:0;background:linear-gradient(120deg,rgba(255,255,255,.08),rgba(255,255,255,0) 52%);pointer-events:none}
#hero p{margin:6px 0 0;font-size:15px;line-height:1.45;color:#dce8ff!important;position:relative;z-index:1}
#hero .copyright{font-size:13px;color:#dce8ff!important;position:relative;z-index:1}
.result-card{background:rgba(247,251,255,.96);backdrop-filter:blur(2px);border:1px solid #cfdef1;border-radius:12px;padding:16px;margin:6px 0 16px;box-shadow:0 3px 12px #254f8410}.eyebrow{font-size:12px;color:#4e6582}.metric-row{display:grid;grid-template-columns:repeat(3,minmax(0,1fr));gap:12px;margin:12px 0}.metric-row>div{padding:14px 16px;background:#eaf3ff;border-radius:9px}.metric-row>div:last-child{background:linear-gradient(125deg,#1d55c6,#2879e8);box-shadow:0 6px 16px #2261c626}.metric-row span{display:block;font-size:12px;color:#4e6582}.metric-row strong{font-size:29px;color:#122b50;display:block;margin-top:5px}.metric-row>div:last-child span,.metric-row>div:last-child strong{color:white}.range,.confidence{font-size:12px;color:#4e6582;margin-top:6px}.big-price{font-size:30px}.price-label{font-size:12px}
button.primary{background:linear-gradient(110deg,#2059c5,#2879e8)!important;border-color:#2463db!important;color:white!important;border-radius:7px!important}footer{display:none!important}
#pricing-inputs{background:rgba(240,247,255,.95);backdrop-filter:blur(2px);border:1px solid #cfdef1;border-radius:10px;padding:14px;gap:10px!important}#pricing-inputs .block{box-shadow:none!important;border-radius:7px!important;min-width:0!important}#pricing-inputs input{min-height:38px!important;padding:7px 10px!important}#pricing-inputs label{font-size:12px!important}#pricing-inputs .form{min-width:0!important}.room-table{background:#f7fbff;border:1px solid #cfdef1;border-radius:10px;overflow:hidden}.room-table table{border-collapse:collapse;width:100%;table-layout:fixed;font-family:Inter,system-ui,sans-serif;font-size:13px;color:#122b50}.room-table th{background:#e2eeff;color:#4e6582;font-weight:600;text-align:right;padding:12px 10px}.room-table td{padding:13px 10px;border-bottom:1px solid #dae5f3;text-align:right;font-variant-numeric:tabular-nums;overflow-wrap:anywhere}.room-table tr:last-child td{border-bottom:0}.room-table td:first-child,.room-table th:first-child{text-align:left;width:38%}.room-table td:last-child{font-weight:650;color:#2059c5}.room-table tbody tr:nth-child(even){background:#eef5ff}
#hero h1{color:#fff!important;position:relative;z-index:1;font-size:40px;margin:0 0 4px}
.rate-movement{background:#edf5ff;border:1px solid #d4e3f6;border-radius:9px;padding:10px 12px 12px;margin:10px 0 8px}
.rr-title{font-size:12px;font-weight:700;color:#35577f;margin-bottom:4px}
.rr-range{display:flex;justify-content:space-between;font-size:10px;color:#6b82a0;margin:0 2px 5px}
.rr-track{position:relative;height:62px;background:linear-gradient(90deg,#dceaff,#f3f8ff,#dceaff);border-radius:8px;overflow:visible}
.rr-line{position:absolute;left:3%;right:3%;top:35px;height:7px;border-radius:999px;background:linear-gradient(90deg,#8fb7ef,#4f86dc);opacity:.5}
.rr-marker{position:absolute;top:0;transform:translateX(-50%);z-index:2}
.rr-dot,.rr-pin{position:absolute;top:31px;left:50%;transform:translateX(-50%);width:11px;height:11px;border-radius:50%;background:#123c79;border:2px solid #fff;box-shadow:0 2px 6px #244e8030}
.rr-label{position:absolute;top:44px;left:50%;transform:translateX(-50%);white-space:nowrap;font-size:9px;color:#4e6582}
.rr-label b{color:#122b50}
.rr-rec{z-index:4}
.rr-rec .rr-pin{background:#fff;border-color:#2059c5;width:13px;height:13px;top:30px}
.rr-callout{position:absolute;top:2px;left:50%;transform:translateX(-50%);white-space:nowrap;border-radius:7px;padding:4px 8px;font-size:11px;font-weight:750;box-shadow:0 3px 8px #254f8420}
.rr-callout span{font-size:14px;margin-right:3px}
.rr-callout small{display:block;font-size:8px;font-weight:650;line-height:1.1;text-align:center}
.move-up{background:#e4f8eb;color:#167245;border:1px solid #b8e6c8}
.move-down{background:#ffe9e9;color:#a62a2a;border:1px solid #f4c2c2}
.move-flat{background:#fff;color:#24508b;border:1px solid #cbdcf2}
.daadu-note{font-size:12px;font-weight:700;margin-top:5px;line-height:1.2}
.daadu-happy{color:#e9fff0}
.daadu-sad{color:#fff2c7}
.result-card>.daadu-note{display:inline-block;padding:4px 8px;border-radius:999px;background:#edf5ff;color:#24508b;border:1px solid #cbdcf2;margin:4px 0 8px}
.gradio-container input,.gradio-container textarea,.gradio-container select{background:#ffffff!important;color:#122b50!important;border:1px solid #cddcf1!important}
.gradio-container label,.gradio-container .wrap label,.gradio-container .block label{color:#24508b!important;font-weight:600!important}
.gradio-container [role="listbox"],.gradio-container [role="option"],.gradio-container .options,.gradio-container .option,.gradio-container ul[role="listbox"]{background:#ffffff!important;color:#122b50!important}
.gradio-container [role="option"]:hover,.gradio-container .option:hover,.gradio-container [aria-selected="true"]{background:#eaf3ff!important;color:#122b50!important}
.gradio-container button,.gradio-container .secondary{color:inherit}
@media(max-width:650px){.metric-row{grid-template-columns:1fr}.metric-row>div{display:flex;justify-content:space-between;align-items:center}.metric-row strong{font-size:24px;margin:0}.gradio-container{padding:12px!important}.room-table td,.room-table th{padding:10px 6px;font-size:12px}}
"""
CUSTOM_CSS += '\n.gradio-container{max-width:1280px!important;padding:28px!important;font-family:Inter,ui-sans-serif,system-ui,sans-serif!important;background:#edf4fc!important}\n#hero{border-radius:20px;padding:26px 30px;min-height:146px;background:radial-gradient(ellipse at 95% 15%,#327ad0 0%,transparent 48%),linear-gradient(120deg,#102744,#163e70);box-shadow:0 12px 32px #16355716;margin-bottom:16px}\n#hero h1{font-size:38px!important;letter-spacing:-1.7px;font-weight:750!important;margin:9px 0 2px}#hero h1 span{color:#72c9ff}\n.brand-kicker{font-size:10px;letter-spacing:2px;font-weight:650;color:#b3d0ef}.hero-bottom{display:flex;justify-content:space-between;align-items:center;gap:12px}.hero-bottom p{font-size:14px!important}.property-pill{background:#ffffff12;border:1px solid #ffffff29;border-radius:30px;padding:7px 12px;font-size:11px;color:#dfedff;white-space:nowrap}\n#hero .copyright{font-size:11px;color:#b9cfe8!important}\n#search-bar{gap:12px!important}#search-bar .block{border-radius:10px!important;box-shadow:none!important}#quick-actions{gap:8px!important}#quick-actions button{min-height:40px!important;font-size:12px!important;border-radius:9px!important}\n.search-context{display:flex;flex-wrap:wrap;gap:7px;align-items:center;margin:0 0 10px;color:#46607d}.search-context span{font-size:10px;font-weight:650;padding:4px 8px;border-radius:6px;background:#dfeaf7}.search-context small{font-size:11px;margin-left:4px}\n.result-card{border-radius:15px;padding:20px;background:#fafdff;border-color:#d6e2f1;box-shadow:0 4px 18px #18385805;margin:0 0 14px}.welcome h2{font-size:22px;letter-spacing:-.5px;color:#142d4b;margin:9px 0}.welcome p{font-size:13px;line-height:1.65;color:#5b6e85;max-width:650px}.eyebrow{font-size:10px;font-weight:700;letter-spacing:1.3px;color:#6c7e94}.metric-row{gap:12px}.metric-row>div{border:1px solid #dae6f4;background:#edf4fd;border-radius:11px;padding:16px}.metric-row strong{font-size:32px;font-weight:720;letter-spacing:-1px;font-variant-numeric:tabular-nums}.metric-row>div:last-child{border-color:#225fb0;background:linear-gradient(125deg,#164581,#2566bb)}\n.source-heading{display:flex;justify-content:space-between;font-size:12px;font-weight:650;color:#233e60;margin:8px 0}.source-heading span{font-weight:400;font-size:11px;color:#677b94}.source-grid{display:grid;grid-template-columns:repeat(5,minmax(0,1fr));gap:10px;margin-bottom:20px}.source-card{background:#f9fcff;border:1px solid #d7e3f1;border-radius:11px;padding:12px;display:flex;flex-direction:column;gap:8px}.source-name{font-size:11px;font-weight:600;line-height:1.4;min-height:31px;color:#45607e}.source-card strong{font-size:23px;color:#173957;font-variant-numeric:tabular-nums;letter-spacing:-.5px}.quote-badge{font-size:9px;font-weight:650;border-radius:5px;padding:4px 6px;align-self:flex-start}.observed{color:#14604c;background:#e1f3ec}.historical{color:#805b18;background:#fff1d7}.missing{color:#607189;background:#e9eef5}.source-card small{font-size:9px;line-height:1.5;color:#6c7e92;overflow-wrap:anywhere}\n#pricing-inputs{padding:16px;background:#e5effa;border-radius:13px;gap:12px!important}#pricing-inputs h3{font-size:16px!important;margin:0!important}#pricing-inputs p{font-size:11px!important;color:#62768c}#pricing-inputs .block{background:#f5f9fe!important}#pricing-inputs label{font-size:11px!important}\n.empty-panel{padding:28px 24px;min-height:180px;border:1px dashed #c4d5e9;background:#f4f9ff;border-radius:12px;margin:12px 0}.empty-panel h3{font-size:17px;color:#244567;margin:0 0 9px}.empty-panel p{font-size:13px;color:#5d718b}.empty-panel span{font-size:11px;color:#76869c}.room-table{margin-top:12px}.room-table th{font-size:10px;letter-spacing:.3px;padding:13px 10px}.room-table td{padding:15px 10px}.room-table tbody tr:hover{background:#e1effe}.room-table td:first-child,.room-table th:first-child{width:30%}\n.gradio-container button:focus-visible,.gradio-container input:focus-visible{outline:3px solid #68a9ed!important;outline-offset:3px}.gradio-container .tab-nav button{font-size:12px!important}.app-footer{border-top:1px solid #d6e2f0;margin-top:12px;padding:18px 0 0;display:flex;justify-content:space-between;color:#6b7f98;font-size:11px}\n@media(max-width:900px){.source-grid{grid-template-columns:repeat(3,minmax(0,1fr))}.gradio-container{padding:18px!important}}\n@media(max-width:650px){#hero{padding:22px;min-height:140px}#hero h1{font-size:32px!important}.hero-bottom{align-items:flex-start;flex-direction:column}.source-grid{grid-template-columns:repeat(2,minmax(0,1fr))}.metric-row{grid-template-columns:1fr!important}.source-heading,.app-footer{gap:8px;flex-direction:column}.source-name{min-height:0}.room-table{overflow-x:auto}.room-table table{min-width:490px}.gradio-container{padding:12px!important}.result-card{padding:16px}.brand-kicker{font-size:9px;letter-spacing:1.1px}}\n@media(prefers-reduced-motion:reduce){*{scroll-behavior:auto!important;transition:none!important}}\n'
with gr.Blocks(title=APP_NAME) as demo:
gr.HTML('SURF CITY INN & SUITES · SANTA CRUZ
Rate Radar.
© 2026 John DM. All rights reserved.
A clearer view of your next rate.
70 rooms · 6 room types1 night2 adultsHotel starting ratesUSD · before taxPublic offers are observations. Recommendations are estimates.
')
result_html=gr.HTML(value=empty_dashboard())
with gr.Row():
with gr.Column(scale=3,min_width=320):
with gr.Tabs():
with gr.Tab("Room prices"):
room_prices=gr.HTML(value='Your rooms, in one view
Run an analysis to see recommendations for all six room types.
Enter known room rates and rooms left under Room inventory.Refresh your recommendations
Select Analyze rates to apply your latest inputs.