import os, requests, time
YELP = os.environ["YELP_API_KEY"]
MV = os.environ["MAVERA_API_KEY"]
YELP_H = {"Authorization": f"Bearer {YELP}"}
MV_H = {"Authorization": f"Bearer {MV}", "Content-Type": "application/json"}
LOCATION = "Austin, TX"
CATEGORIES = [
"bubbletea", "acaibowls", "juicebars", "poke",
"korean", "ramen", "mediterranean", "vegan",
"coffee", "cocktailbars", "breweries", "wine_bars",
]
# 1. Search each category
category_data = []
for cat in CATEGORIES:
r = requests.get(f"https://api.yelp.com/v3/businesses/search",
headers=YELP_H,
params={"categories": cat, "location": LOCATION, "limit": 50, "sort_by": "review_count"})
if r.status_code == 429:
time.sleep(2)
continue
r.raise_for_status()
bizes = r.json().get("businesses", [])
total = r.json().get("total", 0)
if not bizes:
continue
avg_rating = sum(b.get("rating", 0) for b in bizes) / len(bizes)
avg_reviews = sum(b.get("review_count", 0) for b in bizes) / len(bizes)
high_rated = sum(1 for b in bizes if b.get("rating", 0) >= 4.5)
price_dist = {}
for b in bizes:
p = b.get("price", "N/A")
price_dist[p] = price_dist.get(p, 0) + 1
category_data.append({
"category": cat,
"total_in_area": total,
"sampled": len(bizes),
"avg_rating": round(avg_rating, 1),
"avg_review_count": round(avg_reviews),
"high_rated_pct": round(high_rated / len(bizes) * 100) if bizes else 0,
"price_distribution": price_dist,
"top_3": [b["name"] for b in sorted(bizes, key=lambda x: -x.get("review_count", 0))[:3]],
})
time.sleep(0.5)
# 2. Mave trend analysis
cat_block = "\n".join(
f"- {d['category']}: {d['total_in_area']} total, avg {d['avg_rating']}/5, "
f"avg {d['avg_review_count']} reviews, {d['high_rated_pct']}% rated 4.5+. "
f"Top: {', '.join(d['top_3'])}"
for d in sorted(category_data, key=lambda x: -x["total_in_area"])
)
analysis = requests.post("https://app.mavera.io/api/v1/mave/chat",
headers=MV_H,
json={"message": f"""Analyze category trends in {LOCATION} from Yelp data.
{cat_block}
Produce:
1. Emerging trends (categories with high ratings but low total count — growth signal)
2. Saturated categories (high count, declining quality — mature market)
3. Consumer behavior shifts (what do category patterns tell us about local preferences?)
4. Investment opportunities (where should a new business or product launch?)
5. Seasonal or demographic factors driving these trends"""}).json()
print(f"=== Category Trends in {LOCATION} ===")
print(analysis.get("content", "")[:1500])
