Scrape(Sunbiz|Multi): Get Biz's By Zip - Snippet Editor

Save Snippet View Snippet

var data = "";
(async () => {
var delay = ms => new Promise(r => setTimeout(r, ms))
var downloadCSV = (csvString,zip) => {
 var d = document;
 var csvData = new Blob([csvString], { type: 'text/csv;charset=utf-8;'});
 var link = d.createElement('a');
 link.href = window.URL.createObjectURL(csvData);
 link.setAttribute('download', "sunbiz_"+zip+".csv");
 d.body.appendChild(link);
 link.click();
 d.body.removeChild(link);
},
getText = el => el.innerText ? el.innerText.replace(/[,\s\r\t]+/gim," ").trim() : "",
extractRow = (row, i) => {
  if(i == 0 && firstPageComplete){ return ""; }else{
    var rowMeta = { values: [] },
    cells = [...row.querySelectorAll("th,td")];
    if(cells.length > 0) rowMeta.values = rowMeta.values.concat(cells.map(getText));
    return rowMeta.values.join(",");
  }
},
extractTable = table => [...table.querySelectorAll("tr")].map(extractRow).filter(row => row.length > 0).join("\n");
var links = [`url,text`],
processPage = async (url,zip) => {
try{
  var res = await fetch(url);
}catch(e){
  await processPage(url,zip)
}
  try{
    var htmlString = await res.text();
  }catch(e){
    await processPage(url,zip)
  }
  parser = new DOMParser(),
  doc = parser.parseFromString(htmlString, "text/html");
  [...doc.querySelectorAll("td a")].forEach(a => {
    links.push(`${a.href},${a.innerText.replace(/,/gim,"")}`)
  });
 var nextLink = await doc.querySelector("a[title='Next List']");
 if(firstPageComplete) data += "\n";
 data += extractTable(doc.querySelector("table"));
 if(nextLink){
  firstPageComplete = true;
  await processPage(nextLink.href,zip);
 }else{
  console.log({data,links});
  downloadCSV(data,zip);
  downloadCSV(links.join("\n"),zip+"_links");
}};

var firstPageComplete = false,
zips = ["32256"];
for await(let zip of zips){
data = "";
try{
 var url = await `http://search.sunbiz.org/Inquiry/CorporationSearch/SearchResults?inquiryType=ZipCode&searchTerm=${zip}`;
 await processPage(url,zip);
}catch(e){console.log(e)}
}
})()