Homestuck

import java.io.File;
import java.io.FileOutputStream;
import java.io.IOException;
import java.util.Arrays;

import org.apache.commons.io.FileUtils;

import org.jsoup.*;
import org.jsoup.Connection.Response;
import org.jsoup.nodes.*;
import org.jsoup.select.*;

public class main {

    public static void main(String[] args) throws IOException {

    String username = System.getProperty("user.name");
    String outputFolder = "C:";

    //initial setup with user-editable variables
    if (System.getProperty("os.name").startsWith("Windows")) {
        // includes: Windows 2000,  Windows 95, Windows 98, Windows NT, Windows Vista, Windows XP
        outputFolder = ("C:\\Users\\" + username + "\\Homestuck\\");
    } else {
        outputFolder = ("/home/" + username + "/Homestuck/");
        System.out.println("UNIX/Linux system.");
    }

    //Override system DNS setting with Google free DNS server
    System.setProperty("sun.net.spi.nameservice.nameservers", "8.8.8.8");
    System.setProperty("sun.net.spi.nameservice.provider.1", "dns,sun");

    boolean success = (new File(outputFolder)).mkdirs();
    if (!success) {
        System.out.println("Directory creation failed or directory already exists.");
    }

    //String current_id = JOptionPane.showInputDialog("Enter the current comic id"),
    //7415
    String current_id = "007432", imgurl = null, imgurl2 = null, flash_object;

    //location of the image link in array
    int pic_id = 8, img_count = 0, first, last = 0;

    boolean multimage = false, flash = false, extras_downloaded = false, scratch = false, scratch_notify = false, sbahj = false, cascade=false, dota=false, isInSwf=false;

    //the homepage
    String pageurl = "http://www.mspaintadventures.com/?s=6&p=", id = "001901";

    // The page CSS is awful, and uses only tables. This grabs the images used to format the page
    while (extras_downloaded == false) {

        File q = new File(outputFolder + "v2_blankstrip.gif");
        File w = new File(outputFolder + "v2_blanksquare2.gif");
        File e = new File(outputFolder + "v2_blanksquare3.gif");
        File r = new File(outputFolder + "spacer.gif");
        File t = new File(outputFolder + "header_cascade.gif");

        if (q.isFile()) {
            System.out.println("1/4 spacers");
        } else {
        Response resultImageResponse = Jsoup.connect("http://www.mspaintadventures.com/images/v2_blankstrip.gif")
                .ignoreContentType(true).execute();
        FileOutputStream out = (new FileOutputStream(q));
        out.write(resultImageResponse.bodyAsBytes());           // resultImageResponse.body() is where the image's contents are.
        out.close(); }

        if (w.isFile()) {
            System.out.println("2/4 spacers");
        } else {
        Response resultImageResponse = Jsoup.connect("http://www.mspaintadventures.com/images/v2_blanksquare2.gif")
                .ignoreContentType(true).execute();
        FileOutputStream out = (new FileOutputStream(w));
        out.write(resultImageResponse.bodyAsBytes());           // resultImageResponse.body() is where the image's contents are.
        out.close();
        }

        if (e.isFile()) {
            System.out.println("3/4 spacers");
        } else {
        Response resultImageResponse = Jsoup.connect("http://www.mspaintadventures.com/images/v2_blanksquare3.gif")
                .ignoreContentType(true).execute();
        FileOutputStream out = (new FileOutputStream(e));
        out.write(resultImageResponse.bodyAsBytes());           // resultImageResponse.body() is where the image's contents are.
        out.close(); }

        if (r.isFile()) {
            System.out.println("4/4 spacers");
        } else {
        Response resultImageResponse = Jsoup.connect("http://www.mspaintadventures.com/images/spacer.gif")
                .ignoreContentType(true).execute();
        FileOutputStream out = (new FileOutputStream(r));
        out.write(resultImageResponse.bodyAsBytes());           // resultImageResponse.body() is where the image's contents are.
        out.close(); }

        if (t.isFile()) {
            System.out.println("extra heading for Cascade");
        } else {
        Response resultImageResponse = Jsoup.connect("http://www.mspaintadventures.com/images/header_cascade.gif")
                .ignoreContentType(true).execute();
        FileOutputStream out = (new FileOutputStream(t));
        out.write(resultImageResponse.bodyAsBytes());           // resultImageResponse.body() is where the image's contents are.
        out.close(); }

        System.out.println("Finished checking/collecting extra files.");
        extras_downloaded = true;
    }

    /*
     *
     *
     * This is the main download loop.
     *
     *
     */

    while (Integer.parseInt(id) < Integer.parseInt(current_id)) {

        File f = new File(outputFolder + id + ".html");
        File l = new File(outputFolder + id + ".gif");
        File u = new File(outputFolder + id + "_1.gif");
        File x = new File(outputFolder + id + ".swf");

        // Some pages are randomly skipped, and need to be manually added to this array.
        String[] dead_ids = { "004299", "004938", "004988" };

        if (Arrays.asList(dead_ids).contains(id)) {
            id = "00" + (Integer.parseInt(id) + 1);
        }

        //Avoid re-downloading files that already exist
        //This skips everything

        if(!f.isFile()&&(!l.isFile()||!u.isFile()||!x.isFile())) {
        Document doc = null;
        try {
            doc = Jsoup.connect(pageurl+id).get();
        } catch (IOException e) {
            e.printStackTrace();
        }

        System.out.println(pageurl+id);

        Elements imports = doc.select("[src]");

        // MSPA doesn't use divs or any css identifiers, so I put all the src elements into an array
        Object[] linklist;
        linklist = imports.toArray();

        //right now it's the eighth element, but this may change.
        //System.out.println((linklist[9]));

        /*
         * The website changes styles and decoration during a certain bit. There's an additional image header, meaning the comic id is one more
         */

        System.out.println("Comic ID: " + Integer.parseInt(id));

        if (Integer.parseInt(id)>=5664 && Integer.parseInt(id)<=5981)
        {
            if (scratch_notify == false) {
                System.out.println("Post-scratch mode is activated.");
            }
            imgurl = (linklist[pic_id+1]).toString();
            scratch = true;
            scratch_notify = true;
        }

        else if (Integer.parseInt(id)==5982)
        {
            //After this, the site gets even more messed up and requires operations for individual pages.
            sbahj = true;
            System.out.println("sweet bro & hella jeff");
            imgurl = (linklist[pic_id-2]).toString();
            System.out.println(linklist[pic_id-2].toString());

        }

        else if (Integer.parseInt(id)==5985)
        {
            System.out.println("LINKS. IN .SWF FILES.");
             isInSwf=true;
        }

        else if (Integer.parseInt(id)==6009)
        {
            cascade = true;
            imgurl = (linklist[pic_id+1]).toString();
        }

        else if (Integer.parseInt(id)==6715) {
            dota=true;
        }

        else
        {
            imgurl = (linklist[pic_id]).toString();
        }

        int first_quotes = imgurl.indexOf("\"");
        int last_quotes = imgurl.lastIndexOf("\"");

        imgurl = imgurl.substring((first_quotes+1), (last_quotes-1));

        if (imgurl.contains("AC_RunActiveContent.js") && !cascade) {

            System.out.println("Downloading flash content.");
            if (scratch==true) {
                imgurl = linklist[pic_id+2].toString();
            } else {
            imgurl = linklist[pic_id+1].toString();
            }
            //System.out.println(imgurl + "\n" + img_count);
            imgurl = imgurl.replace("<embed src=\"", "");
            first = imgurl.indexOf("http://www.mspaintadventures.com/storyfiles/hs2/");
            last = imgurl.indexOf("\"");
            //System.out.println(imgurl);
            imgurl = imgurl.substring(first,last);
            flash = true;
        }

        if ((imgurl.contains("_") && imgurl == "http://www.mspaintadventures.com/storyfiles/hs2/00898_1.gif") || multimage == true) {

            System.out.println("Multimage!");

            imgurl2 = linklist[(pic_id+1)].toString();
            first_quotes = imgurl2.indexOf("\"");
            last_quotes = imgurl2.lastIndexOf("\"");
            imgurl2 = imgurl2.substring((first_quotes+1), (last_quotes-1));

            //Open a URL Stream
            Response resultImageResponse = Jsoup.connect(imgurl2).ignoreContentType(true).execute();

            // output here
            FileOutputStream out = (new FileOutputStream(new java.io.File(outputFolder + id + "_2" + ".gif")));
            out.write(resultImageResponse.bodyAsBytes());           // resultImageResponse.body() is where the image's contents are.
            out.close();

            multimage = true;
        }

        imgurl = imgurl.replace("http://www.mspaintadventures.com/?s=6&p=", pageurl);
        System.out.println(imgurl);

        //you need to download the image before the html becuase you need to edit the path in the html to the
        //local path of the image

        //Open a URL Stream
        if (!cascade) {
        Response resultImageResponse = Jsoup.connect(imgurl).ignoreContentType(true).execute();

        // write the image

        if (flash == false) {
        FileOutputStream out = (new FileOutputStream(new java.io.File(outputFolder + id + ".gif")));
        out.write(resultImageResponse.bodyAsBytes());           // resultImageResponse.body() is where the image's contents are.
        out.close();
        }
        else {
        FileOutputStream out = (new FileOutputStream(new java.io.File(outputFolder + id + ".swf")));
        out.write(resultImageResponse.bodyAsBytes());           // resultImageResponse.body() is where the image's contents are.
        out.close();
        }

        }

        else if (cascade) {

        }

        imgurl = imgurl.replace(pageurl, "http://www.mspaintadventures.com/");

        //now save just the html

        String html = doc.html();

        //replace web link for image to local link
        html = html.replace(imgurl, outputFolder + id + ".gif");

        if (flash==true) {
            first = html.indexOf("<script language=\"javascript\">AC_FL_RunContent = 0;</script>");
            last = html.indexOf("</object>");

            flash_object = html.substring(first, last);

            html = html.replace(flash_object, ("<a href=\"" + outputFolder + id + ".swf" + "\">Click here for flash content</a>" ));
        }

        // replace web link to local link for the next comic
        if (!cascade)
        {
            html = html.replace(("?s=6&amp;p=" + "00" + (Integer.parseInt(id) + 1)), (outputFolder + ("00" + (Integer.parseInt(id) + 1)) + ".html"));
        }
        else if (cascade) {
            html = html.replace(html.substring(0, html.length()), ("<a href=\"" + outputFolder + ("00" + (Integer.parseInt(id) + 1)) + ".html\">Click here for the next page.</a>"));
        }


        if (multimage == true) {
            html = html.replace(imgurl2, outputFolder + id + "_2" + ".gif");
        }

        //remove junk html

        if (scratch && !sbahj) {

            first = html.indexOf("<!--  FULL LOGO HEADER  --> ");
            last = html.indexOf("</html>");

            html = html.replace(html.substring(first, last), "");

            first = html.indexOf("<td width=\"100%\"> <span style=\"font-size: 9px; line-height: normal\"><b> ");
            last = html.indexOf("</font></font></b></span>");

            html = html.replace("-->", "");

            html = html.replace(html.substring(first, last), "");

            //System.out.println(html);

        } else if (!sbahj && !cascade)
        //this is used for most things
        {
            html = html.substring(0, html.indexOf("<td background=\"images/bannerframe.png\" width=\"950\" height=\"110\" valign=\"middle\">"));

            first = html.indexOf("<!-- begin nav -->");
            last = html.indexOf("<!-- end nav -->");

            html = html.replace(html.substring(first, last), "");
        }
        else if (sbahj) {
            // the *one* sweet bro & hella jeff page
            html = html.replace(html.substring(html.indexOf("<td width=\"100%\"> <span style=\"font-size: 9px; line-height: normal\"><b>"),
                    html.indexOf("<font color=\"#bbbbbb\"> </font></font>")), "");

            html = html.substring(0, html.indexOf("<!--  FULL LOGO HEADER  -->"));
        }

        //For this comic, the link to the next comic is inside the .swf object.
        if (Integer.parseInt(id) == 5984) {
            html = html.replace(("<td width=\"100%\" bgcolor=\"#EEEEEE\">"),
                    ("<td width=\"100%\" bgcolor=\"#EEEEEE\">" + "<a href=\""+ outputFolder + "00" + (Integer.parseInt(id)+1) + ".html" + "\">[S] Attempt rare and highly dangerous 5x SHOWDOWN COMBO.</a>" ));
        }

        //fix pesterlogs
        //System.out.println(id);
        if (html.contains("<button")) {
        first = html.indexOf("<button type=\"button\" class=\"button\" onmouseover");


        if (html.contains("Hide Pesterlog")) {
        last = html.indexOf("Hide Pesterlog</button>");
        }
        else if (html.contains("Hide Spritelog")) {
            last = html.indexOf("Hide Spritelog</button>");
        }
        else if (html.contains("Hide Recap log")) {
            last = html.indexOf("Hide Recap log</button>");
        }
        else if (html.contains("Hide Journalog")) {
            last = html.indexOf("Hide Journalog</button>");
        }
        else if (html.contains("Hide Serious Business")) {
            last = html.indexOf("Hide Serious Business</button>");
        }
        else if (html.contains("Hide Dialoglog")) {
            last = html.indexOf("Hide Dialoglog</button>");
        }

        html = html.replace(html.substring(first, last+14), "");
        System.out.println("Button code detected and fixed.");
        }

        //fix links to image spacers
        html = html.replace("images/", "");

        FileUtils.writeStringToFile(f, html);

        id = "00" + (Integer.parseInt(id) + 1);

        img_count = (img_count+1);

        //avoid program crashes due to frequency of requests
        try {
            Thread.sleep(10);
        } catch (InterruptedException e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
        }

        multimage = false;
        flash = false;
        sbahj = false;
        cascade = false;
        isInSwf=false;
    }

    else {
        id = "00" + (Integer.parseInt(id) + 1);
    }

    }

    System.out.println("Done. Hussie's webmaster breathes a sigh of relief.");

    }

}