VkXml.java

package com.statusParser.parsers;

import java.io.IOException;

import com.statusParser.Configuring;
import com.statusParser.Mysql;

import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

import java.sql.SQLException;
import java.util.Properties;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

public class VkXml {
    private static final String USER_AGENT = "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/28.0.1500.95 Safari/537.36";
    private static final int POSTS_ON_ROTATION = 100;
    private static boolean mysqlOn = false;
    private static Mysql mysql;

    private int parseOwnerId(String groupName) {
        String regex = "\\d+";                                                                                          //Только цифры
        Pattern p = Pattern.compile(regex);
        Matcher m = p.matcher(groupName);
        if (m.matches()) {
            return Integer.parseInt(groupName);
        } else try {
            Document xmlDocScreenName = Jsoup.connect("https://api.vk.com/method/utils.resolveScreenName.xml?screen_name=" + groupName)
                    .userAgent(USER_AGENT)
                    .get();

            Element ownerIdElement = xmlDocScreenName.select("object_id").first();
            return Integer.parseInt(ownerIdElement.text());
        } catch (IOException e) {
            e.printStackTrace();
        }
        return -1;
    }

    public void parse(String group) {
        int ownerId = parseOwnerId(group);
        Configuring conf = new Configuring();
        Properties prop = conf.load();
        if (prop.getProperty("mysql") != null) {
            if (prop.getProperty("mysql").equals("on")) {
                mysqlOn = true;
                mysql = new Mysql();
            }
        }

        try {
            Document xmlDocCountPosts = Jsoup.connect("https://api.vk.com/method/wall.get.xml?owner_id=-" + ownerId + "&filte=owner&offset=0&count=1")
                    .userAgent(USER_AGENT)
                    .get();

            Element counterPostsElement = xmlDocCountPosts.select("count").first();
            int countPosts = Integer.parseInt(counterPostsElement.text());

            int countRotations = (countPosts + (POSTS_ON_ROTATION - 1)) / POSTS_ON_ROTATION;

            for (int i = 0; i < countRotations; i++) {
                String[][] parsedData = partParse(ownerId, i * POSTS_ON_ROTATION);
                //System.out.println("Progress..." + (i * POSTS_ON_ROTATION * 100) / countPosts);
                if (mysqlOn) {
                    for (String post[] : parsedData) {
                        //postIdOk, statusOk, imgOk, likesOk
                        mysql.sentData(Integer.parseInt(post[0]), post[1], post[2], Integer.parseInt(post[3]));         //FIXME!!! Null pointer exception when running in terminal
                    }
                }
            }
        } catch (IOException e) {
            e.printStackTrace();
        }
    }

    private String[][] partParse(int group, int offset) {
        int iter = 0;
        String[][] parsedData =new String[POSTS_ON_ROTATION][4];
        try {
            Document xmlDoc = Jsoup.connect("https://api.vk.com/method/wall.get.xml?owner_id=-" + group + "&filte=owner&offset=" + offset + "&count=" + POSTS_ON_ROTATION)
                    .userAgent(USER_AGENT)
                    .get();

            Elements posts = xmlDoc.select("post");
            for (Element post : posts) {

                //postid parsing
                Element postid = post.select("id").first();
                parsedData[iter][0] = postid.text();

                //text parsing
                Element text = post.select("text").first();
                String status = text.text();
                if (!status.equals("")) {
                    if (status.length() <= 500) {
                        System.out.println(iter + ": " + status);
                        parsedData[iter][1] = status;                                                                   /*FIXME!!! Incorrect string value with emoticons*/
                    } else {
                        System.out.println(iter + ": Too long post");
                    }
                } else {
                    System.out.println(iter + ": No text in post");
                }

                //image parsing
                if (post.select("attachment > photo > src_big").first() != null) {
                    Element img = post.select("attachment > photo > src_big").first();
                    parsedData[iter][2] = img.text();
                }

                //likes parsing
                Element likes = post.select("likes").first();
                parsedData[iter][3] = likes.text();

                iter++;
            }
        } catch (IOException e) {
            e.printStackTrace();
        }
        return (parsedData);
    }

}