Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- public class info {
- private static final String SAMPLE_CSV_FILE = "./sample.csv";
- public static void main(String[] args) throws InterruptedException, IOException {
- // this was testing from a page saved sourcefile
- //File input = new File("C:\\Users\\jonat\\Desktop\\haily0.html");
- //Document doc = Jsoup.parse(input , "UTF-8", "C:\\Users\\jonat\\Desktop\\info4.html");
- Document doc = Jsoup.parse("Insert current pagesource here");
- //System.out.println(bradon.text());
- Element[] person_block;
- Element[] email_block;
- Element[] phone_block;
- Element[] company_phone_block;
- Element[] company_name_block;
- Element[] company_detail_block;
- Element[] company_content_left;
- person_block = doc.getElementsByClass("person-row-container").toArray(new Element[doc.getElementsByClass("person-row-container").size()]);
- System.out.println(doc.getElementsByClass("contact-details-content").size());
- email_block = doc.getElementsByClass("email-link").toArray(new Element[doc.getElementsByClass("email-link").size()]);
- phone_block = doc.getElementsByClass("direct-phone").toArray(new Element[doc.getElementsByClass("direct-phone").size()]);
- company_phone_block = doc.getElementsByClass("hq-phone").toArray(new Element[doc.getElementsByClass("hq-phone").size()]);
- company_name_block = doc.getElementsByClass("company-name long-text").toArray(new Element[doc.getElementsByClass("company-name long-text").size()]);
- company_detail_block = doc.getElementsByClass("primaryCompany-container").toArray(new Element[doc.getElementsByClass("primaryCompany-container").size()]);
- company_content_left = doc.getElementsByClass("primary-content ng-star-inserted").toArray(new Element[doc.getElementsByClass("primary-content ng-star-inserted").size()]);
- //primary-content-left
- int lengthy = person_block.length;
- try (
- BufferedWriter writer = Files.newBufferedWriter(Paths.get(SAMPLE_CSV_FILE));
- CSVPrinter csvPrinter = new CSVPrinter(writer, CSVFormat.DEFAULT
- .withHeader("ID", "Full Name", "Title","Company", "Email", "Phone", "Company Phone", "Website","Address"));
- ) {
- int i = 0;
- while(i < person_block.length){
- String fullName = person_block[i].getElementsByClass("primary-name").text();
- String title = person_block[i].getElementsByClass("person-title").text();
- String email = email_block[i].getElementsByClass("email-link").text().replace("Email", "");
- String phone = phone_block[i].getElementsByClass("direct-phone").text().replace("- (Direct)","").replace("Phone (Direct)", "").replace("(Direct)", "");
- String hqPhone = company_phone_block[i].getElementsByClass("hq-phone").text().replace("- (HQ)","").replace("(HQ)","");
- String company_name = company_name_block[i].getElementsByClass("company-name long-text").text();
- String company_detail = company_detail_block[i].getElementsByClass("link").text();
- String location_detail = company_content_left[i].getElementsByClass("local-content location-content").text();
- //fix grabbing html
- csvPrinter.printRecord(i, fullName, title, company_name, email, phone, hqPhone, company_detail, location_detail);
- i++;
- }
- csvPrinter.flush();
- }
- }
Advertisement
Add Comment
Please, Sign In to add comment