import os, requests, time
from collections import defaultdict
YELP = os.environ["YELP_API_KEY"]
MV = os.environ["MAVERA_API_KEY"]
YELP_BASE = "https://api.yelp.com/v3"
YELP_H = {"Authorization": f"Bearer {YELP}"}
MV_H = {"Authorization": f"Bearer {MV}", "Content-Type": "application/json"}
# 1. Search businesses by category and location
CATEGORY = "coffee"
LOCATION = "Austin, TX"
businesses = []
for offset in range(0, 200, 50):
r = requests.get(f"{YELP_BASE}/businesses/search",
headers=YELP_H,
params={
"term": CATEGORY,
"location": LOCATION,
"limit": 50,
"offset": offset,
"sort_by": "review_count",
})
if r.status_code == 429:
time.sleep(2)
continue
r.raise_for_status()
batch = r.json().get("businesses", [])
if not batch:
break
businesses.extend(batch)
time.sleep(0.5)
print(f"Found {len(businesses)} {CATEGORY} businesses in {LOCATION}")
# 2. Aggregate market data
price_tiers = defaultdict(list)
categories_seen = defaultdict(int)
neighborhoods = defaultdict(int)
for biz in businesses:
price = biz.get("price", "N/A")
rating = biz.get("rating", 0)
reviews = biz.get("review_count", 0)
price_tiers[price].append({"rating": rating, "reviews": reviews, "name": biz["name"]})
for cat in biz.get("categories", []):
categories_seen[cat.get("alias", "")] += 1
loc = biz.get("location", {})
hood = loc.get("city", LOCATION)
neighborhoods[hood] += 1
# 3. Build analysis context
market_summary = []
for price, bizes in sorted(price_tiers.items()):
avg_rating = sum(b["rating"] for b in bizes) / len(bizes)
avg_reviews = sum(b["reviews"] for b in bizes) / len(bizes)
top = sorted(bizes, key=lambda x: -x["reviews"])[:3]
top_names = ", ".join(b["name"] for b in top)
market_summary.append(
f"- {price} tier: {len(bizes)} businesses, avg {avg_rating:.1f}/5, "
f"avg {avg_reviews:.0f} reviews. Leaders: {top_names}"
)
cat_summary = ", ".join(f"{k} ({v})" for k, v in
sorted(categories_seen.items(), key=lambda x: -x[1])[:10])
biz_block = "\n".join(
f"- {b['name']}: {b.get('rating',0)}/5, {b.get('review_count',0)} reviews, "
f"Price: {b.get('price','N/A')}, Categories: {', '.join(c.get('title','') for c in b.get('categories',[]))}"
for b in sorted(businesses, key=lambda x: -x.get("review_count", 0))[:15]
)
# 4. Mave competitive analysis
analysis = requests.post("https://app.mavera.io/api/v1/mave/chat",
headers=MV_H,
json={"message": f"""Analyze the competitive landscape for "{CATEGORY}" in {LOCATION}.
MARKET OVERVIEW:
{len(businesses)} businesses found
BY PRICE TIER:
{chr(10).join(market_summary)}
CATEGORY OVERLAP:
{cat_summary}
TOP 15 BY REVIEW VOLUME:
{biz_block}
Produce:
1. Market saturation assessment (oversaturated / balanced / underserved)
2. Competitive clusters (which businesses compete directly)
3. White space opportunities (underserved niches, price gaps, location gaps)
4. Positioning recommendations for a new entrant
5. Differentiator suggestions based on what existing players lack
6. Pricing strategy recommendation"""}).json()
print(f"\n=== {CATEGORY.title()} Market in {LOCATION} ===")
print(analysis.get("content", "")[:2000])
