index.js

#!/usr/bin/env node

var path = require("path"),
    fs = require("fs"),
    dsv = require("dsv"),
    optimist = require("optimist"),
    queue = require("queue-async"),
    topojson = require("topojson"),
    jsts = require("jsts"),
    d3 = require("d3");


var us = JSON.parse(fs.readFileSync("../us-counties.json", "utf-8"));

var top = {}, transform = us.transform;
var counties = d3.map();
var msas = d3.map();
var eas = d3.map();

for (var key in us.objects) {
    console.log(key);
    top[key] = topojson.feature(us, us.objects[key]);
}

console.log("top level objects: " + top);
console.log("counties: " + top.counties.type);

us.objects.counties.geometries.forEach(function(c) {
    counties.set(c.id, topojson.feature(us, c));
});

var reader = new jsts.io.GeoJSONReader();
var writer = new jsts.io.GeoJSONWriter();
top['msa'] = {type: "FeatureCollection", features: []};
top['ea'] = {type: "FeatureCollection", features: []};
fs.readFile("../msa.csv", "utf-8", function(error, text){
    var count = 0;
    dsv.csv.parse(text).map(function(row) {
        var msa_id = +row.regiongroup_id;
        var fips = row.region_code;
        if (!msas.has(msa_id)) {
            msas.set(msa_id, { id: msa_id, name: row.name_msa, counties: d3.set()});
        }
        msa = msas.get(msa_id);
        msa.counties.add(fips);
        count++;
    });

    var largest = 0;
    var i = 1;
    var errors = [];
    var invalid = 0;

    msas.forEach(function(_id ,msa) {
        console.log(i + "/" + msas.keys().length + " " + msa.name);
        msa.counties = msa.counties.values();
        if (msa.counties.length > largest) {
            largest = msa.counties.length;
        }
        var msa_geo;
        var geos = [];
        msa.counties.forEach(function(cid) {
            var c = counties.get(cid);
            if (!c) {
                console.log("    Could not file county: " + cid + " for msa: " + msa.id + " -- " + msa.name);
            } else {
                var geo = reader.read(c.geometry);
                geo.buffer(0);
                var isValidOp = new jsts.operation.valid.IsValidOp(geo);
                if (!isValidOp.isValid()) invalid++;
                geos.push(geo);
                try {
                    msa_geo = ( msa_geo ? msa_geo.union(geo) : geo);
                } catch (ex)  {
                    errors.push({msa: msa.name, id: msa.id, valid: (isValidOp.isValid() ? true: isValidOp.getValidationError()), error: ex, county: cid});
                }
            }
        });

        if (!msa_geo) console.log("not geo for msa: " + msa.name);
        else {
            msa.feature = {type: "Feature", id: msa.id, geometry: writer.write(msa_geo)};
            top['msa'].features.push(msa.feature);
        }
        i++;
    });
    console.log("invalid: " + invalid);
    console.log("errors: " + errors.length);
    console.log("mappings: " + count);
    console.log("msa count: " + msas.keys().length);
    console.log("largest msa: " + largest + " counties");

    fs.readFile("../ea.csv", "utf-8", function(error, text) {
        var count = 0;
        dsv.csv.parse(text).map(function(row) {
            //ea_code,eax,fips,county_name
            var ea_id = +row.ea_code;
            var fips = row.fips;
            if (!eas.has(ea_id)) {
                eas.set(ea_id, { id: ea_id, eax: row.eax, counties: d3.set()});
            }
            ea = eas.get(ea_id);
            ea.counties.add(fips);
            count++;
        });

        var i = 0;
        eas.forEach(function(_id, ea) {
            console.log(++i + "/" + eas.keys().length + " " + ea.id);
            ea.counties = ea.counties.values();
            var ea_geo;
            var geos = [];
            var j = 0;
            ea.counties.forEach(function(cid) {
                console.log("\t" + ++j + "/" + ea.counties.length + " counties");
                var c = counties.get(cid);
                if (!c) {
                    console.log("    Could not find county: " + cid + " for ea: " + ea.id + " -- " + ea.eax);
                } else {
                    var geo = reader.read(c.geometry);
                    geo.buffer(0);
                    var isValidOp = new jsts.operation.valid.IsValidOp(geo);
                    geos.push(geo);
                    try {
                        ea_geo = ( ea_geo ? ea_geo.union(geo) : geo);
                    } catch (ex)  {
                        errors.push({ea: ea.eax, id: ea.id, valid: (isValidOp.isValid() ? true: isValidOp.getValidationError()), error: ex, county: cid});
                    }
                }
            });
            if (!ea_geo) console.log("not geo for ea: " + ea.id);
            else {
                ea.feature = {type: "Feature", id: ea.id, geometry: writer.write(ea_geo)};
                top['ea'].features.push(ea.feature);
            }
        });

        output(top, 1, "output.json");
    });



});

function print_errors(errors) {
    if (errors.length > 0) {
        console.log("Errors");
        console.log("===========");
        errors.forEach(function(e) {
            console.log("    ID: " + e.id);
            console.log("  Name: " + e.msa);
            console.log("County: " + e.county);
            console.log(" Valid: " + e.valid);
            console.log(" Error: " + e.error);
            console.log("   GEO: " + JSON.stringify(e.geom));
            console.log("  MGEO: " + JSON.stringify(e.msag));
            console.log();
        });
    }
}


function output(objects, min_area, file) {
    var options = {
        "verbose": true,
        "quantization": 1e4,
        "coordinate-system": "cartesian",
        "id": function(d) { return d.id }
    };

    // Convert GeoJSON to TopoJSON.
    var object = topojson.topology(objects, options);

    // Simplify.
    topojson.simplify(object, {
        "verbose": true,
        "coordinate-system": options["coordinate-system"],
        "minimum-area": min_area
//        "retain-proportion": +argv["simplify-proportion"]
    });

    topojson.clockwise(object, {
        "verbose": true,
        "coordinate-system": options["coordinate-system"]
    });

    topojson.filter(object, {
        "verbose": true,
        "coordinate-system": options["coordinate-system"]
    });

    // Bind with external properties.
//    if (argv.e) topojson.bind(object, externalProperties);

    delete object.bbox;

    // Output JSON.
    var json = JSON.stringify(object);
    fs.writeFileSync(file, json, "utf8");
}

// Build same for EA
// Create data file that maps country names, containing ea, msa, state