HomestuckDL

package com.pi.homestuck.pages;

import java.io.BufferedInputStream;
import java.io.BufferedReader;
import java.io.BufferedWriter;
import java.io.File;
import java.io.FileOutputStream;
import java.io.FileWriter;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.net.URL;
import java.net.URLConnection;
import java.util.ArrayList;
import java.util.List;

import javax.swing.JOptionPane;

public class ComicPage {
    public static boolean quiet = true;

    private final URL url;
    private String rawData;
    private String nextComicURL;
    private String nextComic;
    private List<String> dataParts = new ArrayList<String>();
    private String currentComic;
    private String nextComicReplace;

    public ComicPage(URL url) {
        this.url = url;
    }

    public void load() throws IOException {
        URLConnection conn = url.openConnection();
        InputStream in = conn.getInputStream();
        BufferedReader read = new BufferedReader(new InputStreamReader(in));
        rawData = null;
        StringBuilder build = new StringBuilder();
        boolean inComic = false;
        boolean inScript = false;
        while (true) {
            String line = read.readLine();
            if (line == null)
                break;
            if (line.toLowerCase().contains("<!--")) {
                if (line.toLowerCase().contains("begin comic content")) {
                    inComic = true;
                } else if (line.toLowerCase().contains("end comic content")) {
                    inComic = false;
                    rawData = build.toString();
                    break;
                }
            } else if (inComic) {
                if (line.contains("<script"))
                    inScript = true;
                if (!inScript && !line.contains("noscript>"))
                    build.append(line + "\n");
                if (line.contains("</script>"))
                    inScript = false;
            }
        }
        read.close();
        in.close();
    }

    public void parse() {
        parseCurrentComic();
        parseNextComic();

        dataParts.clear();
        parseLinks();
        parseSRC();
    }

    public void save(File root) throws IOException {
        File dataDir = new File(root, "data");
        if (!dataDir.exists())
            dataDir.mkdir();
        downloadAndInjectParts(dataDir, "data/");

        File dest = new File(root, getCurrentComic() + ".html");
        if (!dest.exists())
            dest.createNewFile();

        if (nextComic != null)
            rawData = rawData.replace(nextComicReplace, nextComic + ".html");

        BufferedWriter write = new BufferedWriter(new FileWriter(dest));
        write.write("<html><body>");
        write.write(rawData);
        write.write("</body></html>");
        write.close();
    }

    private void downloadAndInjectParts(File destDir, String pathPrefix)
            throws IOException {
        String rootPath = url.getProtocol() + "://" + url.getHost();
        String relPath = rootPath + url.getPath();
        for (String s : dataParts) {
            String url = "";
            if (s.startsWith("/")) {
                url = rootPath + s;
            } else if (s.startsWith("http://")) {
                url = s;
            } else {
                url = relPath + s;
            }

            URL rURL = new URL(url);
            String fPath = rURL.getPath();
            if (fPath.startsWith("/"))
                fPath = fPath.substring(1);
            File dest = new File(destDir, fPath);
            if (!dest.exists()) {
                if (fPath.contains("/")) {
                    File parent = new File(dest.getParent());
                    parent.mkdirs();
                }
                if (!quiet)
                    System.out.println("Downloading: " + rURL.toString()
                            + "\t=>\t" + fPath);
                try {
                    download(rURL, dest);
                } catch (IOException e) {
                    System.err.println("Error on " + rURL.toString());
                    continue;
                }
            } else {
                if (!quiet)
                    System.out.println("Skipping downloading: "
                            + rURL.toString() + "\t=>\t" + fPath);
            }
            rawData = rawData.replace(s, pathPrefix + fPath);
        }
    }

    private void parseCurrentComic() {
        String[] query = url.getQuery().split("&");
        currentComic = null;
        for (String s : query) {
            if (s.startsWith("p=")) {
                currentComic = s.substring(2);
                return;
            }
        }
    }

    private void parseNextComic() {
        int realIndex = rawData.indexOf("<font size=\"5\">&gt; ");
        nextComicURL = nextComicReplace = nextComic = null;
        if (realIndex != -1) {
            int href = rawData.indexOf("href=", realIndex);
            if (href != -1) {
                href += 5;
                if (rawData.charAt(href) == '"') {
                    int endQuote = rawData.indexOf('"', href + 1);
                    if (endQuote != -1) {
                        nextComicURL = nextComicReplace = rawData.substring(
                                href + 1, endQuote);
                        if (nextComicURL.startsWith("?")) {
                            nextComicURL = url.getProtocol() + "://"
                                    + url.getHost() + url.getPath()
                                    + nextComicURL;
                        }

                        String[] query = nextComicURL.split("&");
                        for (String s : query) {
                            if (s.startsWith("p=")) {
                                nextComic = s.substring(2);
                                break;
                            }
                        }

                        return;
                    }
                }
            }
        }
        System.err.println("Next Comic Parse Error! " + url.toExternalForm());
    }

    private void parseSRC() {
        int currentIndex = 0;
        while ((currentIndex = rawData.indexOf(" src=", currentIndex + 5)) != -1) {
            char quote = rawData.charAt(currentIndex + 5);
            if (quote == '\'' || quote == '"') {
                int quoteEnd = rawData.indexOf(quote, currentIndex + 6);
                if (quoteEnd != -1) {
                    String url = rawData.substring(currentIndex + 6, quoteEnd);
                    if (!dataParts.contains(url))
                        dataParts.add(url);
                }
            }
        }
    }

    private void parseLinks() {
        int currentIndex = 0;
        while ((currentIndex = rawData.indexOf(
                "http://www.mspaintadventures.com/", currentIndex + 30)) != -1) {
            int dblQuoteStart = rawData.lastIndexOf('"', currentIndex);
            int singleQuoteStart = rawData.lastIndexOf('\'', currentIndex);
            int quoteStart = -1;
            char quote;
            if (singleQuoteStart == -1 && dblQuoteStart == -1) {
                continue;
            }
            if (singleQuoteStart == -1) {
                quote = '"';
                quoteStart = singleQuoteStart;
            } else if (dblQuoteStart == -1) {
                quote = '\'';
                quoteStart = dblQuoteStart;
            } else if (dblQuoteStart < singleQuoteStart) {
                quoteStart = singleQuoteStart;
                quote = '\'';
            } else {
                quoteStart = dblQuoteStart;
                quote = '"';
            }
            if (quoteStart != -1) {
                int quoteEnd = rawData.indexOf(quote, currentIndex + 30);
                if (quoteEnd != -1) {
                    String url = rawData.substring(quoteStart + 1, quoteEnd);
                    if (!dataParts.contains(url))
                        dataParts.add(url);
                }
            }

        }
    }

    public String getNextComicURL() {
        return nextComicURL;
    }

    public String getNextComic() {
        return nextComic;
    }

    public String getCurrentComic() {
        return currentComic;
    }

    public static void main(String[] args) throws IOException {
        new File("Homestuck").mkdir();

        String origin = JOptionPane
                .showInputDialog("Starting comic?\nex: http://www.mspaintadventures.com/?s=6&p=001901");
        download(new URL(origin));
    }

    public static void download(URL url) throws IOException {
        ComicPage pg = new ComicPage(url);
        pg.load();
        pg.parse();

        String dest = "Homestuck/" + pg.getCurrentComic() + ".html";
        System.out.println("Downloading: " + pg.getCurrentComic() + "\t=>\t"
                + dest);

        pg.save(new File(dest).getParentFile());

        if (pg.getNextComic() != null)
            download(new URL(pg.getNextComicURL()));
    }

    public static void download(URL url, File dest) throws IOException {
        BufferedInputStream in = null;
        FileOutputStream fout = null;
        File f = new File(dest.getAbsolutePath() + ".part");
        f.createNewFile();
        try {
            in = new BufferedInputStream(url.openStream());
            fout = new FileOutputStream(f);
            byte data[] = new byte[1024];
            int count;
            while ((count = in.read(data, 0, 1024)) != -1) {
                fout.write(data, 0, count);
            }
        } finally {
            if (in != null)
                in.close();
            if (fout != null)
                fout.close();
        }
        dest.delete();
        f.renameTo(dest);
    }
}