JSON into CSV : most efficient for large files?

[ChenXiu]

If a json decoded array is huge (1,000,000+ lines) what is the fastest, least CPU intensive way to get it into a CSV?
I only want the ["content"] portion into the .csv file:

// $result = json_decode($jsonFile); print_r($result);
stdClass Object
(
    [content] => Array
        (
            [0] => stdClass Object
                (
                    [id] => 1234
                    [day] => monday
                    [location] => USA
                )
            [1] => stdClass Object
                (
                    [id] => 5678
                    [day] => thursday
                    [location] => EU
                )
        )
    [searchCriteria] => stdClass Object
        (
            [from] => monday
            [toDate] => friday
        )
    [size] => 2
    [number] => 0
)

What is the best way to loop to only get the "id" "day" and "location" from the ["content"] portion of the array in to the .csv file?

$fp = fopen( 'csv_file.csv' , 'w' );
foreach( $result as $i ) { <------- this part.
fputcsv( $fp , $i );
}
fclose($fp);

Thank you.

[Barand]

Decode to an array. fputcsv() doesn't play nice with objects.

Then

$fp = fopen( 'csv_file.csv' , 'w' );
foreach( $result['content'] as $i ) {
    fputcsv( $fp , $i );
}
fclose($fp);

 

[ChenXiu]

7 hours ago, Barand said:

fputcsv() doesn't play nice with objects

Thank you. I've done a lot of benchmarkes and indeed, the fastest speed I could reach was with exactly what you are suggested (using json_decode($result,true) and then looping through $data["content"] ). I thought there might be a way to save 500ms by "capturing" the "content" portion of the json string (using a string function while json is still a string) because string functions always are faster than looping. But maybe not.... I think monkeying with the json with string functions would open a pandoras box of errors 😃

 

[Barand]

Consider 3 things

• The amount of money saved by shaving a few millisecs off the execution time
• The cost of your time spent trying to shave those few millisecs off the execution time.
• Is it worth it?