In a prior post, I link to some code that outputs a path file.  I’ve done something a tad different because I needed some select link analysis and reading the path file in Cube was taking far too long to do it the normal way.

So, I took that program on Github and extended it to perform a selected link:

	//Check for specific nodes
	bool output=false;
	for(int pn=1;pn<prrn-1;pn++){
	if(prd.nodes[pn]==9798 && prd.nodes[pn+1]==9750)
	output=true;
	}
	if(output){
	of.open(argv[2],iostream::trunc);
	of << prd.I << ",";
	for(int pn=0;pn<prrn;pn++){
	of << prd.nodes[pn];
	if(prd.nodes[pn]!=prd.J)
	of <<",";
	}
	of << endl;
	of.close();
	}

view raw selectLink hosted with by GitHub

And this outputs a few GB of paths in CSV format.  I went from 42 GB of paths in the AM to 3.4 GB of CSV paths.  Still not good enough. The next thing I did was use GAWK to get just the Origin and Destination

gawk 'BEGIN {FS=","};{print $1,",",$NF}' amPaths.csv >amOD.CSV

view raw gawk_getfirstlast hosted with by GitHub

This returns a CSV file of just the origin and destination (which can be linked to the vehicle trip matrix).

Part 2 will discuss how to link to a vehicle trip matrix and if this approach actually works!

Tags: awk, csv files, cube, gawk, path files, select link

Posted on Wednesday, January 30th, 2013 By Andrew Rohne Comments are closed.