Issue mozilla-mobile#25: Add yelp->ta verification fn.

This will allow devs to check out yelp places in different areas to see how
well TA matches.

scripts/prox_crosswalk.py

@@ -3,16 +3,22 @@
 
 Intended to be used from a REPL. Primary functions are:
 - yelp_ids_to_tripadvisor_ids
+  - (for debugging) verify_yelp_id_to_tripadvisor_ids
 - write_to_db
 
 TODO:
 - Handle hitting API limits
 - Handle network errors (e.g. TA API)
 - TEST ACCURACY
-- Make it easy to run (read place ids from file?)
+- Make it easy to run (read place ids from file? store intermediate xw into file? then try atomic update?)
+- Make this code re-usable and put into separate files.
+- How does this handle incorrect Yelp places?
+- Keeps throwing 504: Gateway Time-out from Yelp3Client (rate limit?).
 
 """
 from __future__ import print_function
+
+from app import util
 from app.constants import _tablePrefix
 from app.providers import tripadvisor as ta
 from app.providers import yelp
@@ -29,12 +35,16 @@
     'wikipedia',
 }
 
+# HACK: used to avoid duplicating Yelp place requests.
+_YELP_ID_TO_PLACE_CACHE = {}
+
 
 def _get_crosswalk_db(): return _firebase.database().child(_CROSSWALK_PATH)
 
 
 def _yelp_id_to_tripadvisor(yelp_id):
-    place = yelp.resolve_with_key(yelp_id)
+    place = _YELP_ID_TO_PLACE_CACHE.get(yelp_id, yelp.resolve_with_key(yelp_id))
+    _YELP_ID_TO_PLACE_CACHE[yelp_id] = place
     name = place['name']
     coord = place['coordinates']
     coord_tuple = (coord['latitude'], coord['longitude'])
@@ -130,3 +140,40 @@ def add_ids_to_dict(provider_key, yelp_to_other_id):
 
     for yelp_id, provider_map in crosswalk.iteritems():
         _write_crosswalk_to_db(yelp_id, provider_map)
+
+
+def verify_yelp_ids_to_tripadvisor_ids(yelp_ids):
+    """Returns the expanded output of 1) Yelp -> TA places that match to allow human verification that
+    they're the same places and 2) Yelp places that did not match a TA equivalent to allow humans to find
+    out why not!
+
+    It is recommended to `pprint` the results.
+
+    :return: {'not_missing_ta': {'yelp': <yelp-place-obj>,
+                                 'ta': [<ta-place-obj>, ...]},
+              'missing_ta': [<yelp-place-obj>, ...]}
+    """
+    tas = _yelp_ids_to_raw_tripadvisor(yelp_ids)
+    missing_out = []
+    not_missing_out = []
+    for yelp_id, ta in tas.iteritems():
+        yplace = _YELP_ID_TO_PLACE_CACHE.get(yelp_id, yelp.resolve_with_key(yelp_id))  # These should all be cached.
+        yout = {'name': yplace['name'],
+                'url': util.strip_url_params(yplace['url']),
+                'loc': ', '.join(yplace['location']['display_address'])}
+        if len(ta) < 1:
+            missing_out.append(yout)
+            continue
+
+        tout = []
+        for ta_place in ta:
+            tout.append({'name': ta_place['name'],
+                         'id': ta_place['location_id'],
+                         'distance': ta_place['distance'],
+                         'loc': ta_place['address_obj'].get('address_string', '')})
+
+        val = {'yelp': yout, 'ta': tout}
+        not_missing_out.append(val)
+
+    return {'not_missing_ta': not_missing_out,
+            'missing_ta': missing_out}