Select Distinct Rows from MySQL database through PHP/PDO [duplicate] - php

Question

This question already has answers here:
Closed 10 years ago.
Possible Duplicate:
How to prevent SQL injection in PHP?
Select distinct rows from MySQL Database
Hi I have a phonegap application that stores latitude, longitude, address and severity when a road surface deformation (such as pothole and speedbump) is detected.
So far saving these in the database through a PHP PDO is not a problem at all. The PDO is designed in a way that if the pothole being reported has already been reported 10 times (checks the database for any entries within a 15 meter range), then it would not be reported (i.e inserted in the database again). Also, loading the surface deformations is not a problem either, i am using the Haversine formula to do that, where I pass the latitude and longitude of the user and get the values within a certain distance.
$stmt = $dbh->prepare("
SELECT
lat, lng,
( 6378160 * acos( cos( radians(?) ) * cos( radians( lat ) ) * cos( radians( lng ) - radians(?) ) + sin( radians(?) ) * sin( radians( lat ) ) ) ) AS distance
FROM myTable
HAVING distance > 0
ORDER BY distance
LIMIT 0 , 30
");
The issue I have is that since the same pothole can be reported 10 times, I am ending up having the same pothole reported back to the application for charting on a map 10 times. What I need to do is, get the list of potholes that are within a certain distance from the user (done using the haversine formula), and then out of this list, filter the potholes so that I only get distinct potholes rather than the same pothole being returned back 10 times. Anyone has any idea how I can make such filtering? Can anyone tell me how is it posssible to do this in PHP/PDO or point me to some similar tutorial if available?
Here is what I need to do in brief: say I am near pothole A and Pothole B, and say I have 6 reports for pothole A, and 8 reports for pothole B (and so on) in the database. By using the haversine formula I get all the values of the reports for pothole A and pothole B (ie 14 results). What I need is rather I get the midpoint of the reports for pothole A and midpoint of the reports for Pothole B (using: http://www.geomidpoint.com/calculation.html) and return back 2 results (one for A and one for B) rather than 14 results.

Answer 1

You have to separate potholes and reports. First query what potholes are around GROUP BY lat, lng and then query one report for each pothole with LIMIT 1.
$rows = dbh_query("SELECT id FROM ... ");
foreach ($rows as $row)
{
dbh_query("SELECT ... WHERE id = :id LIMIT 1", array('id' => $row['id']));
}

Answer 2

I think you have 2 options to elaborate. Unfortunately both implies some complications.
1) Instead of writing multiple observations, you should always clusterize them at writing time. For example, if you support space granularity of 10 meters, then every time a new measurement arrives that is less then in 10 meters from existing record, you do not add a new pothole, but change average values (latitude, longitude, counter) in the nearest existing record. This way you'll end up with 2 records for potholes A and B for your example, so you can use DISTINCT query.
2) For every request, you can fetch all records in 15 meter range from existing table and create a temporary table on them for calculating a probability density function along the road axe. Again, this requires to choose a granularity which can be simulated as decimal accuracy in the ROUND function. For example, if you would have a stored function for calculating distance between current point and existing records, you could write:
INSERT INTO `temppdf` (dist, pothole_id)
SELECT FROM `maintable`
ROUND(distance(#current_lat, #current_lon, maintable.lat, maintable.lon), -1), pothole_id
WHERE distance(#current_lat, #current_lon, maintable.lat, maintable.lon) < 15;
Then you can query temppdf for rows with maximum counters for every pothole, something like this:
SELECT pothole, MAX(cnt) as `peak` FROM
(SELECT DISTINCT pothole, COUNT(dist) as cnt FROM `temppdf` GROUP BY pothole, dist) as `subq`
GROUP BY pothole;
The potholes with counters larger than a threshold are the result.