Added support for video downloading
This commit is contained in:
parent
7e890bce2f
commit
2eadcd675e
@ -43,6 +43,9 @@ public class XhamsterRipper extends AbstractHTMLRipper {
|
|||||||
URLToReturn = URLToReturn.replaceAll("\\w\\w.xhamster.com", "xhamster.com");
|
URLToReturn = URLToReturn.replaceAll("\\w\\w.xhamster.com", "xhamster.com");
|
||||||
URL san_url = new URL(URLToReturn.replaceAll("xhamster.com", "m.xhamster.com"));
|
URL san_url = new URL(URLToReturn.replaceAll("xhamster.com", "m.xhamster.com"));
|
||||||
LOGGER.info("sanitized URL is " + san_url.toExternalForm());
|
LOGGER.info("sanitized URL is " + san_url.toExternalForm());
|
||||||
|
if (isVideoUrl(url)) {
|
||||||
|
return url;
|
||||||
|
}
|
||||||
return san_url;
|
return san_url;
|
||||||
}
|
}
|
||||||
|
|
||||||
@ -58,7 +61,13 @@ public class XhamsterRipper extends AbstractHTMLRipper {
|
|||||||
if (m.matches()) {
|
if (m.matches()) {
|
||||||
return "user_" + m.group(1);
|
return "user_" + m.group(1);
|
||||||
}
|
}
|
||||||
throw new MalformedURLException(
|
p = Pattern.compile("^https?://.*xhamster\\.com/(movies|videos)/(.*)$");
|
||||||
|
m = p.matcher(url.toExternalForm());
|
||||||
|
if (m.matches()) {
|
||||||
|
return m.group(2);
|
||||||
|
}
|
||||||
|
|
||||||
|
throw new MalformedURLException(
|
||||||
"Expected xhamster.com gallery formats: "
|
"Expected xhamster.com gallery formats: "
|
||||||
+ "xhamster.com/photos/gallery/xxxxx-#####"
|
+ "xhamster.com/photos/gallery/xxxxx-#####"
|
||||||
+ " Got: " + url);
|
+ " Got: " + url);
|
||||||
@ -108,9 +117,20 @@ public class XhamsterRipper extends AbstractHTMLRipper {
|
|||||||
if (m.matches()) {
|
if (m.matches()) {
|
||||||
return true;
|
return true;
|
||||||
}
|
}
|
||||||
|
p = Pattern.compile("^https?://.*xhamster\\.com/(movies|videos)/.*$");
|
||||||
|
m = p.matcher(url.toExternalForm());
|
||||||
|
if (m.matches()) {
|
||||||
|
return true;
|
||||||
|
}
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
private boolean isVideoUrl(URL url) {
|
||||||
|
Pattern p = Pattern.compile("^https?://.*xhamster\\.com/(movies|videos)/.*$");
|
||||||
|
Matcher m = p.matcher(url.toExternalForm());
|
||||||
|
return m.matches();
|
||||||
|
}
|
||||||
|
|
||||||
@Override
|
@Override
|
||||||
public Document getNextPage(Document doc) throws IOException {
|
public Document getNextPage(Document doc) throws IOException {
|
||||||
if (doc.select("a[data-page=next]").first() != null) {
|
if (doc.select("a[data-page=next]").first() != null) {
|
||||||
@ -125,19 +145,23 @@ public class XhamsterRipper extends AbstractHTMLRipper {
|
|||||||
@Override
|
@Override
|
||||||
public List<String> getURLsFromPage(Document doc) {
|
public List<String> getURLsFromPage(Document doc) {
|
||||||
List<String> result = new ArrayList<>();
|
List<String> result = new ArrayList<>();
|
||||||
for (Element thumb : doc.select("div.picture_view > div.pictures_block > div.items > div.item-container > a > div.thumb_container > div.img > img")) {
|
if (!isVideoUrl(url)) {
|
||||||
String image = thumb.attr("src");
|
for (Element thumb : doc.select("div.picture_view > div.pictures_block > div.items > div.item-container > a > div.thumb_container > div.img > img")) {
|
||||||
// replace thumbnail urls with the urls to the full sized images
|
String image = thumb.attr("src");
|
||||||
image = image.replaceAll(
|
// replace thumbnail urls with the urls to the full sized images
|
||||||
"https://upt.xhcdn\\.",
|
image = image.replaceAll(
|
||||||
"http://up.xhamster.");
|
"https://upt.xhcdn\\.",
|
||||||
image = image.replaceAll("ept\\.xhcdn", "ep.xhamster");
|
"http://up.xhamster.");
|
||||||
image = image.replaceAll(
|
image = image.replaceAll("ept\\.xhcdn", "ep.xhamster");
|
||||||
"_160\\.",
|
image = image.replaceAll(
|
||||||
"_1000.");
|
"_160\\.",
|
||||||
// Xhamster has bad cert management and uses invalid certs for some cdns, so we change all our requests to http
|
"_1000.");
|
||||||
image = image.replaceAll("https", "http");
|
// Xhamster has bad cert management and uses invalid certs for some cdns, so we change all our requests to http
|
||||||
result.add(image);
|
image = image.replaceAll("https", "http");
|
||||||
|
result.add(image);
|
||||||
|
}
|
||||||
|
} else {
|
||||||
|
result.add(doc.select("div.player-container > a").attr("href"));
|
||||||
}
|
}
|
||||||
return result;
|
return result;
|
||||||
}
|
}
|
||||||
|
Loading…
Reference in New Issue
Block a user