EVE Technology Lab

1)
from my point of view it is useless to use a temporary table .
you can get the lowest order directly in your for loop .

2)
file_get_contents() is a blocking call.
if you have tons of request to do why not use threads ?

Thanks for the help, have managed to double the speed and eliminate the excessive temporary database writes and reads.

@Dragonaire, many thanks for taking the time to work through and mark up all those edits!
Have already used some of the suggested edits from your tidy up pass, and will incorporate several more when I get to a final version.

@Princess
1 - Yup - temporary table being useless was always my thinking, I was just having a stack of trouble working out how Not to need it.
Still haven't managed to get the lowest order within that same loop. But, after a lot of false starts, did manage a workaround that while probably still horribly inefficient is much better than what I had.
- found I couldn't work out how to either exclude other stations, or find the lowest price (much, much pain trying to understand just how a multidimensional array actually works ). Instead was able to identify if an array value was at the correct station, and then push just the price out into a new array, and from there I was able to sort that array and then use the first value.

2 - I have no idea how to get at that json data without using file_get_contents - will be my google project for tomorrow.
Equally so, no idea how to use multiple threads - will try and work on that after replacing file_get

For those playing along at home, shrug, maybe this is useful to someone else who wants to build something similar who coincidentally is just as unknowledge'd as I -> Here's how it currently all looks: http://pastebin.com/dWExLjwC
(Not suggesting that anyone actually uses this; Plenty to be done to it still... but... well, it works as it currently is if your standards aren't exceptionally high ...)

Thanks once again for the edits, notes and the more or less complete re-write to make it all better.
Performance is vastly improved from my initial version, and I was thinking I would need to spend some weekend time sorting out how and what to do if values didn't exist or weren't returned, so thanks also for including all of the error handling.

A random good Samaritan (who I presume wishes to remain anonymous by virtue of his contact) left me a mail with just a link (and nothing more) to a decade old blogpost about using curl to fetch data, from which I was able to get a working function that I've included as get_data - Some testing has shown that using this method rather than the previous file_get_contents method saves a fraction of a second per query, so that's why that is there now. (Whilst also ticking off Princess Honneamise suggestion to not use file_get_contents)

The crest/json/'array within an array of arrays' format broke your $lowest, min bit - Wasn't correctly identifying the locationID and thus was storing the INT_MAX for everything. Managed to find the problem;
$item['location'] is itself an array, but moving over to $item['location']['id'] actually gets to the station to then match back to locationID, so that's why that change is now in there.

Overall is taking 35 seconds to complete a run through with 20 mostly random items in the itemlist table - so working on understanding and implementing some form of threading will be my next challenge.

Current version updated here:
http://pastebin.com/aXj0kMA7