Implement sjoin within with contains (geopandas#575)

We switch the sides of the join and use the complementary predicate.
This seems to be significantly faster in the common case.

Author: mrocklin

geopandas/tools/sjoin.py

@@ -206,11 +206,23 @@ def sjoin(left_df, right_df, op='intersects', how='inner',
         raise ValueError("'right_df' should be GeoDataFrame, got {}".format(
                          type(right_df)))
 
-    allowed_hows = ['left', 'right', 'inner']
+    allowed_hows = ('left', 'right', 'inner')
     if how not in allowed_hows:
         raise ValueError("`how` was \"%s\" but is expected to be in %s" %
                          (how, allowed_hows))
 
+    original_op = op
+    original_how = how
+    if op == "within":
+        # within implemented as the inverse of contains; swap names
+        # This is done for efficiency reasons
+        op = 'contains'
+        left_df, right_df = right_df, left_df
+        if how == 'left':
+            how = 'right'
+        elif how == 'right':
+            how = 'left'
+
     if left_df.crs != right_df.crs:
         warn(
             ('CRS of frames being joined does not match!'
@@ -240,6 +252,11 @@ def sjoin(left_df, right_df, op='intersects', how='inner',
     left = left_df.take(left_indices)
     right = right_df.take(right_indices)
 
+    if original_op == 'within':  # switch back
+        left, right = right, left
+        n_left, n_right = n_right, n_left
+        how = original_how
+
     if how in ('inner', 'left'):
         del right[right._geometry_column_name]
         index = left.index