Add query parameter to filter results by name

[c-w]

What's this?

This pull request adds an additional query parameter called filter_name that lets users subsample the return values of the featureService to only those features whose name (partially) matches the provided string.

This is generally useful, e.g. imagine querying for all features in a bounding box or at a particular point but while already having an idea of the names of features that we want to retrieve. Additionally, the feature will be used in the Fortis dashboard for implementing auto-suggest in the place search pane.

Sample usage

Bounding box query without name filter: 32 results

Narrowing to just features matching "somali": 3 results

Try it live!

Implementation notes

We use the Postgres strpos function instead of a wildcard LIKE as it usually leads to better query plans for substring queries. For example, the below query plans show that strpos is a good 10% faster at execution and 40x faster at planning than the equivalent wildcard LIKE.

features=# explain analyze select name from features where strpos(lower(name),'aris')>0 limit 2;
                                                    QUERY PLAN
------------------------------------------------------------------------------------------------------------------
 Limit  (cost=0.00..1.20 rows=2 width=11) (actual time=0.044..3.088 rows=2 loops=1)
   ->  Seq Scan on features  (cost=0.00..94295.52 rows=157781 width=11) (actual time=0.043..3.087 rows=2 loops=1)
         Filter: (strpos(lower((name)::text), 'aris'::text) > 0)
         Rows Removed by Filter: 2538
 Planning time: 0.055 ms
 Execution time: 3.107 ms
(6 rows)

features=# explain analyze select name from features where lower(name) like '%aris%' limit 2;
                                                  QUERY PLAN
--------------------------------------------------------------------------------------------------------------
 Limit  (cost=0.00..4048.35 rows=2 width=11) (actual time=0.051..3.454 rows=2 loops=1)
   ->  Seq Scan on features  (cost=0.00..93112.16 rows=46 width=11) (actual time=0.051..3.454 rows=2 loops=1)
         Filter: (lower((name)::text) ~~ '%aris%'::text)
         Rows Removed by Filter: 2139
 Planning time: 0.260 ms
 Execution time: 3.472 ms
(6 rows)

[jcjimenez]

LGTM