Support multi-page motherless albums

This commit is contained in:
4pr0n 2014-03-12 02:14:40 -07:00
parent e975ae6a5c
commit 1cdec765cc

View File

@ -47,27 +47,36 @@ public class MotherlessRipper extends AbstractRipper {
@Override @Override
public String getGID(URL url) throws MalformedURLException { public String getGID(URL url) throws MalformedURLException {
Pattern p = Pattern.compile("^https?://(www\\.)?motherless\\.com/G([A-Z0-9]{6,8}).*$"); Pattern p = Pattern.compile("^https?://(www\\.)?motherless\\.com/G([VI][A-F0-9]{6,8}).*$");
Matcher m = p.matcher(url.toExternalForm()); Matcher m = p.matcher(url.toExternalForm());
if (!m.matches()) { if (!m.matches()) {
throw new MalformedURLException("Expected URL format: http://motherless.com/GXXXXXXXX"); throw new MalformedURLException("Expected URL format: http://motherless.com/GIXXXXXXX, got: " + url);
} }
return m.group(m.groupCount()); return m.group(m.groupCount());
} }
@Override @Override
public void rip() throws IOException { public void rip() throws IOException {
int index = 0; int index = 0, page = 1;
logger.info(" Retrieving " + this.url.toExternalForm()); String nextURL = this.url.toExternalForm();
Document doc = Jsoup.connect(this.url.toExternalForm()) while (nextURL != null) {
.userAgent(USER_AGENT) logger.info(" Retrieving " + nextURL);
.get(); Document doc = Jsoup.connect(nextURL)
for (Element thumb : doc.select("div.thumb a.img-container")) { .userAgent(USER_AGENT)
URL url = new URL("http://" + DOMAIN + thumb.attr("href")); .get();
index += 1; for (Element thumb : doc.select("div.thumb a.img-container")) {
// Create thread for finding image at "url" page URL url = new URL("http://" + DOMAIN + thumb.attr("href"));
MotherlessImageThread mit = new MotherlessImageThread(url, index); index += 1;
motherlessThreadPool.addThread(mit); // Create thread for finding image at "url" page
MotherlessImageThread mit = new MotherlessImageThread(url, index);
motherlessThreadPool.addThread(mit);
}
// Next page
nextURL = null;
page++;
if (doc.html().contains("?page=" + page)) {
nextURL = this.url.toExternalForm() + "?page=" + page;
}
} }
motherlessThreadPool.waitForThreads(); motherlessThreadPool.waitForThreads();
waitForThreads(); waitForThreads();