Reputation: 6010
I've a complex query with SQLite that seeks to grab a total from 3 related tables.
The following is an example of my tables:
----
cities
----
id, name
----
surveys
----
id, cities_id
----
population
----
id, survey_id
I need to find the total number of surveyed people for each city.
Here's the approach I've been taking using c#, but I believe there should be a straighter way to do this through SQLite. Plus, my brain is not solving my final step using c#.
//get a list of all cities
List<string> cities = new List<string>();
cmd.CommandText = "SELECT * from cities ";
cmd.ExecuteScalar();
SqlCeDataReader myReader = cmd.ExecuteReader();
while (myReader.Read())
{
list_cities.Add(myReader["id"].ToString().Trim());
}
//get survey ids related to these citities
List<list_key_val> list_survey_ids = new List<list_key_val>();
foreach (String city_id in cities)
{
cmd.CommandText = "SELECT * from surveys WHERE city_id = '" + city_id + "'";
cmd.ExecuteScalar();
SqlCeDataReader myReader = cmd.ExecuteReader();
while (myReader.Read())
{
list_survey_ids.Add(new list_key_val()
{
key = city_id,
val = myReader["id"].ToString().Trim()
});
}
}
//find total surveyed person count for each survey id.
List<list_key_val> list_surveyed_total = new List<list_key_val>();
foreach (String survey_id in list_survey_ids)
{
cmd.CommandText = "SELECT COUNT(*) from population WHERE survey_id = '" + survey_id + "'";
cmd.ExecuteScalar();
list_surveyed_total.Add(new list_key_val()
{
key = survey_id,
val = count = (int)cmd.ExecuteScalar();
}
}
//find surveyed count as city id -> survey id -> surveyed count
List<list_key_val> list_city_surveyed_total = new List<list_key_val>();
foreach (String city_id in list_cities)
{
//dar... complex nested key calculations here, possibly?
}
I understand that SQLite has some limitation when it comes to combining queries. Thank you for any and all help!
Upvotes: 0
Views: 939
Reputation: 31454
You should be able to pull all that information in one go, just alter your query:
var query =
"SELECT c.name, s.id, COUNT(*) " +
"FROM cities c " +
"JOIN surveys s ON (c.id = s.cities_id) " +
"JOIN population p ON (s.id = p.survey_id) " +
"GROUP BY c.name, s.id";
cmd.CommandText = query;
SqlCeDataReader myReader = cmd.ExecuteReader();
while (myReader.Read())
{
Console.WriteLine("City {0}, survey {1} = {2}", myReader[0], myReader[1],
myReader[2]);
}
Upvotes: 1
Reputation: 828
Hard to tell if I got it right but you should be able to perform the operation in a single query... something like:
select cities.name, surveys.id, count(*) total
from cities, surveys, population
where cities.id = surveys.cities_id and surveys.id=population.survey_id
group by cities.name, surveys.id
Chicago, 1, 4000 Chicago, 2, 3355 Denver, 1, 2580 etc....
Ok, here is a dump of a test I just did in sqlite
sqlite> create table cities (id integer primary key, name text);
sqlite> create table surveys (id integer primary key, cities_id integer);
sqlite> create table population (id integer primary key, survey_id integer);
sqlite> insert into cities values (1, 'Chicago');
sqlite> insert into cities values (2, 'Denver');
sqlite> insert into surveys values (1, 1);
sqlite> insert into surveys values (2, 1);
sqlite> insert into surveys values (3, 2);
sqlite> insert into population values (1, 1);
sqlite> insert into population values (2, 1);
sqlite> insert into population values (3, 1);
sqlite> insert into population values (4, 1);
sqlite> insert into population values (5, 2);
sqlite> insert into population values (6, 2);
sqlite> insert into population values (7, 2);
sqlite> insert into population values (8, 3);
sqlite> insert into population values (9, 3);
sqlite> select cities.name, surveys.id, count(*) total
...> from cities, surveys, population
...> where cities.id = surveys.cities_id and
...> surveys.id=population.survey_id
...> group by cities.name, surveys.id
...> ;
Chicago|1|4
Chicago|2|3
Denver|3|2
sqlite>
Upvotes: 1
Reputation: 5272
per city AND survey:
select count(c.id) as num, c.name, s.id as survey from cities c, sirveys s, population p
where
c.id = s.cities_id AND
s.id = p.survey_id
group by c.name, s.id
per city:
select count(c.id) as num, c.name, from cities c, sirveys s, population p
where
c.id = s.cities_id AND
s.id = p.survey_id
group by c.name
Upvotes: 1