A yelp Search Request only returns min(n, 20) responses for a certain geo-query. This script will recursively Quadrisect the box until there is 20 in each box. It will also print the json output of the response

yelp_location_crawler.py

from __future__ import print_function
from Queue import Queue

import json

def subdivide(posn1, posn2):
    """
    Accepts two Tuple2 and returns a generator that produces the four new boxes.

    Args:
        posn1 - south-west coord
        posn2 - north-east coord
    Yields:
        posn - SW, NE tuples
    """
    mid = lambda (a, b): (a + b) / 2.
    a, b = posn1
    c, d = posn2
    bd, ac = mid(b, d), mid(a, c)
    # top_left
    yield ((a,  bd), (ac,  d))
    yield ((ac, bd), (c,   d))
    yield ((a,   b), (ac, bd))
    yield ((ac,  b), (c,  bd))

def recurse_divide(posn1, posn2, n, queue):
    """
    Populates lst with 4^n subdivisions of a box defined by posn1, posn2

    Args:
        posn1 - south-west coord
        posn2 - north-east coord
        n - number of new quadrant divisions
        queue - container for new divisions
    """
    for (posn11, posn22) in subdivide(posn1, posn2):
        recurse_divide(posn11, posn22, n - 1, queue) if n - 1 \
            else queue.put((posn11, posn22))
    return None

def crawl(queue):
    """
    Crawl a queue of box coordinates

    Args:
        queue - Queue()
    """
    temp = "http://api.yelp.com/v2/search?bounds={},{}"
    while not queue.empty():
        p1, p2 = queue.get()
        url = temp.format(*p1) + "|" + "{},{}".format(*p2)
        content = yelp_search.get(url).content
        response = json.loads(content)
        try:
            total = response[u'total']
        except:
            total = 0
        if total > 20 and total != cur:
            recurse_divide(p1, p2, 1, queue)
        if total > 0:
            yield response

# Initializing Emily's Jobs


# Toronto
posn1 = (43.601887,-79.409602)
posn2 = (43.693345,-79.350158)

# Create YelpSession
yelp_search = OAuth1Session(...)

# Initalize Queue and fill with 16 subdivisions
box_queue = Queue()
recurse_divide(posn1, posn2, 2, box_queue)

# Crawl!!!!
for response in crawl(box_queue):
    json_str = json.dumps(response)
    print(json_str)

rekt'd yelp.