SARS-CoV-2-Viz/parse.js

// SARS-CoV-2-Viz
// Animated COVID case count visualization
// Copyright 2022 Edward L. Platt <ed@elplatt.com>

// Convert csv to tsv
// Assumes quoted strings do not contain escaped quotes
async function csvToTSV(csv, onProgress) {

    onProgress("parsing CSV: 0 of ? parts");

    let quoteParts = csv.split('"');
    let tsv = "";
    const rowCount = quoteParts.length;
    let percent = 0;
    let partStack = quoteParts.reverse();
    
    return new Promise((resolve, reject) => {
	let index = 0;
	let processPart = () => {

	    while (true) {

		if (partStack.length == 0) {
		    resolve(tsv);
		    break;
		}
	    
		part = partStack.pop();
	    
		if (index % 2 == 0) {
		    // Not quoted
		    tsv += part.replace(/,/g, "\t");
		} else {
		    // Quoted
		    tsv += part;
		}
		
		index += 1;
		let newPercent = Math.floor(100 * index / rowCount);
		if (newPercent > percent) {
		    percent = newPercent;
		    onProgress(`parsing CSV: ${index} of ${rowCount} parts`);
		    setTimeout(processPart, 0);
		    break;
		}
	    }
	}
	setTimeout(processPart, 0);
    });
}

function tsvRowToJSON(keys, tsvRow) {
    let row = {};
    let rowParts = tsvRow.split("\t");
    for (const [index, key] of keys.entries()) {
	row[key] = rowParts[index];
    }
    return row;
}

function tsvToJSON(tsv, onProgress) {
    const lines = tsv.split(/\n/);
    let columnNames = [];
    let data = [];
    const count = lines.length;
    for (const [row, line] of lines.entries()) {
	if (row == 0) {
	    columnNames = line.split("\t").map((x) => x.trim());
	} else {
	    let dataRow = {};
	    for (const [col, cell] of line.split("\t").entries()) {
		dataRow[columnNames[col]] = cell.trim();
	    }
	    // Limit to US Counties
	    if (dataRow["iso3"] == "USA"
		&& dataRow["Admin2"] != ""
		&& dataRow["Admin2"] != "Unassigned"
		&& dataRow["FIPS"] != "") {
		data.push(dataRow);
	    }
	}
	let progress = count / row;
	onProgress(progress);
    }
    return data;
}