batch process for http request

[#1]

Hello,

I trying to write a script that sends an http request by putting together an endpoint.

Putting together the request itself is not an issue but in some cases, I need to request some info on more than a 1000 items (test_article) at once. 

I'm limited by the maximum length of the http request to requesting about a 1000 at a time.

Unfortunately, I have not been able to figure out how to put together a loop that would send requests for  batches of 1000 test_article and concatenate those as they are received.

Any insight would be greatly appreciated

[#2]

I had a question regarding background http requests that @Craige_Hales answered. 

https://community.jmp.com/t5/Uncharted/Can-JMP-run-JMP/ba-p/283748#U283748

I then took his code and modified it to do something like what you are looking for.  In my case I was querying over different dates but you can modify the code to loop over whatever you want.  You will probably need to make some adjustments depending on your specific use case.

Hopefully this helps.

url = "www....";

//Using this to generate different queries
query_keys = Associative Array();
query_keys["key1"] ="Bob";  //Or whatever your query parameters are

//I am looping over different dates
days_back = 20;
day_inc = 2;

start = Date Increment(today(),"day",-days_back);
end = Today();

cur_start = Date Increment(today(),"day",-day_inc);
cur_end = end;
index = 1;
//Here I am making files that contain the url and query key for each http request.  I will use these later.  This is only really needed if you intent to parallelize the http requset among multiple jmp instances.
//If you don't want to parallelize the queries you can skip the writing to disk and just make a list of your query keys and/or urls.
while(cur_end>start,
	query_key["startdate"] = char(Format(cur_start,"yyyy-mm-dd"));
	query_key["enddate"] = char(Format(cur_end,"yyyy-mm-dd"));
	Save Text File("$temp/url_file"||char(index)||".txt",url);
	Save Text File("$temp/query_keys_file"||char(index)||".txt",As JSON Expr(query_keys));
	index++;
	if(Date Increment(cur_start,"day",-day_inc)>start,
		cur_start = Date Increment(cur_start,"day",-day_inc),
		cur_start = start;
	);
	cur_end = Date Increment(cur_end,"day",-day_inc);
);


index--;
nscripts = index;

njmp = 1; //You can parallelize this as much as you want by launching more jmp instances.  Or if you want to 

// a place to keep up with running instances
rp = [=> ];
rpIndex=0;

For( i = 1, i <= njmp, i += 1, 
	// build a custom JSL for each JMP to execute
	workerFileName = Save Text File(
		"$temp\deleteMe_RunsWhenLaunchedBeCareful" || Char( i ) || ".jsl",
		"\[//!
		deletefile("$temp/sentinel.txt");
		for(i=1,i<=]\"||char(nscripts)||"\[,i++,
			if(Is File("$temp/url_file"||char(i)||".txt"),
				url = Load Text File( "$temp/url_file"||char(i)||".txt" )  ;
				query_key = Parse JSON( Load Text File( "$temp/query_keys_file"||char(i)||".txt" )  );
				Delete File("$temp/url_file"||char(i)||".txt");
				Delete File("$temp/query_keys_file"||char(i)||".txt");
//Using try so if one request fails I can keep moving through the requests
				Try(
					dig_http = New HTTP Request(URL(url),Method("Get"),Query String(query_key),Timeout(60*60*2))<<Send;
//My data is coming in as csv so I convert that to jmp.
					file_name = "$temp/data"||char(i)||".csv";
					Save Text File(file_name,dig_http);
					dt = Open(file_name);
					dt << Save("$temp/data"||char(i)||".jmp");
					Delete File("$temp/data"||char(i)||".csv");
					,
//Dummy file if try fails
					dt = New Table();
					dt << Save("$temp/data"||char(i)||".jmp");
				);
			);	
		);
		exit();
		]\"
	);
	// don't launch another copy until the previous copy is running.
	// JMP will become unhappy if the preferences file is busy in another
	// copy of JMP. The sentinel will be deleted when JMP starts
	savetextfile("$temp/sentinel.txt","");
	rp[rpIndex+=1] = Run Program( executable( "jmp" ), options( {workerFileName} ) );
	write("\!n",i," started...");
	// wait for the sentinel to vanish
	while(fileexists("$temp/sentinel.txt"),
		wait(1);// don't burn the CPU while waiting
	);
	write("ok");
);



// many queries are probably still queued, wait for them to finish
scripts_left = nscripts;
While( scripts_left,
	scripts_left = 0;
	For( i = 1, i <= nscripts, i++,
		If(Is File("$temp/url_file"||char(i)||".txt"),
			scripts_left += 1; // count the ones still alive
		)
	);
	write("\!n",scripts_left," http requests left");
	wait(1); // don't burn the CPU while waiting
	
);


Write( "\!ngathering results..." );
dt = New table("Final Table");
For( i = 1, i <= nscripts, i += 1, 
	filename="$temp/data"||char(i)||".jmp";
	// do something with the results
		// clean up the temp files
	tmp_dt = open(filename);
	dt << Concatenate(tmp_dt,Append to First Table);
	Close(tmp_dt);
	deletefile(filename);
	deletefile("$temp\deleteMe_RunsWhenLaunchedBeCareful" || Char( i ) || ".jsl");
	
);
Write( "\!ndone" );

[#3]

Thanks a lot! I'll try to adapt your approach to my issue.

[#4]

Here is an approach that I use. It uses a single instance of JMP. The code combines the HTTP results into a single JSON object which is later converted to a data table.

list = {};
For( i = 1, i <= number_of_iterations, i++, 
        request = HTTP Request(
            Url( url ),
            Method( "GET" ),
            Headers( request_headers ),
            QueryString( query ),
            Password( access_token ),
            Timeout( 360 )
        );
		
        data = request << Send;
        If( request << IsSuccess,
            response = Parse JSON( data );
            If( AndMZ( Is Associative Array( response ), Contains( response, "records" ), N Items( response["records"] ) > 0 ),
                list ||= response["records"]
            );
        ,
            Wait( 0.25 )
        );
);

[#5]

I realized I've left out the scripts I was starting from my initial inquiry. For the sake of clarity, here's the starting point script I would like to find a way to breakdown into batches when the number of items exceeds 1000.

Best

		Names Default To Here( 1 );

		dt = Data Table( "test_article" ); // This is a summary data table only contains 1 column of interest 'test_article' - I want to retrieve some info for each of the unique items in that column 
		//If the number of items exceeds 1000, I'll need to break down the query in batches of 1000 or less to construct the http request
		//Right if the list is 1000 or less, I'm golden, if it's greater, the request will fail
		dt:test_article << set data type( "Numeric" );
		dt << Transpose( columns( :test_article ), Output Table( "Transpose of test_article" ) );
		dt = Data Table( "Transpose of test_article" );
		Col_List = dt << Get Column Names( "String" );
		For( i = 1, i <= N Items( Col_List ), i++,
			If( Contains( Col_List[i], "Row" ),
				Column( Col_List[i] ) << Set Selected( 1 )
			)
		);
		dt << Combine Columns(
			delimiter( "," ),
			selected Columns only( 1 ),
			Selected Columns are Indicator Columns( 0 ),
			Column Name( "test_article" )
		);
		nc = N Cols( Data Table( "Transpose of test_article" ) );
		Show( nc );
		rowNum = 1;
		col = Column( Data Table( "Transpose of test_article" ), "test_article" );
		tableVal = col[rowNum];

				adr0 = "http:/xyz:1234/oligos/more-oligo-info-intended?ids=";
				address = adr0 || tableval;
		

			request = New HTTP Request( URL( address ), Method( "GET" ), Timeout( 240 ) );
			json1 = request << Send;
			dt = JSON To Data Table( json1 );
			Current Data Table() << set name( "test_article_info" );
			Data Table( "test_article" ) << close window;
			Data Table( "Transpose of test_article" ) << close window;
		// I need to build a for loop that will send 1 request after the next and progressively concatenate those file together

[#6]

Hi Ryan,

Thanks for the answer. I'll see if I can adapt your suggestion to the problem I've listed below.

Best,

Sebastien