lastseen feature now works with instagram

This commit is contained in:
cyian-1756 2018-03-09 13:36:53 -05:00
parent a54b71065c
commit 4af4691893

View File

@ -24,6 +24,7 @@ import com.rarchives.ripme.utils.Utils;
public class InstagramRipper extends AbstractHTMLRipper {
String nextPageID = "";
private String userID;
@ -198,7 +199,6 @@ public class InstagramRipper extends AbstractHTMLRipper {
@Override
public List<String> getURLsFromPage(Document doc) {
String nextPageID = "";
List<String> imageURLs = new ArrayList<>();
JSONObject json = new JSONObject();
try {
@ -261,33 +261,7 @@ public class InstagramRipper extends AbstractHTMLRipper {
break;
}
}
// Rip the next page
if (!nextPageID.equals("") && !isThisATest()) {
if (url.toExternalForm().contains("/tags/")) {
try {
// Sleep for a while to avoid a ban
sleep(2500);
if (url.toExternalForm().substring(url.toExternalForm().length() - 1).equals("/")) {
getURLsFromPage(Http.url(url.toExternalForm() + "?max_id=" + nextPageID).get());
} else {
getURLsFromPage(Http.url(url.toExternalForm() + "/?max_id=" + nextPageID).get());
}
} catch (IOException e) {
return imageURLs;
}
}
try {
// Sleep for a while to avoid a ban
sleep(2500);
getURLsFromPage(Http.url("https://www.instagram.com/" + userID + "/?max_id=" + nextPageID).get());
} catch (IOException e) {
return imageURLs;
}
} else {
logger.warn("Can't get net page");
}
} else { // We're ripping from a single page
logger.info("Ripping from single page");
imageURLs = getPostsFromSinglePage(doc);
@ -296,6 +270,37 @@ public class InstagramRipper extends AbstractHTMLRipper {
return imageURLs;
}
@Override
public Document getNextPage(Document doc) {
if (!nextPageID.equals("") && !isThisATest()) {
if (url.toExternalForm().contains("/tags/")) {
try {
// Sleep for a while to avoid a ban
sleep(2500);
if (url.toExternalForm().substring(url.toExternalForm().length() - 1).equals("/")) {
return Http.url(url.toExternalForm() + "?max_id=" + nextPageID).get();
} else {
return Http.url(url.toExternalForm() + "/?max_id=" + nextPageID).get();
}
} catch (IOException e) {
return null;
}
}
try {
// Sleep for a while to avoid a ban
sleep(2500);
return Http.url("https://www.instagram.com/" + userID + "/?max_id=" + nextPageID).get();
} catch (IOException e) {
return null;
}
} else {
logger.warn("Can't get net page");
}
return null;
}
@Override
public void downloadURL(URL url, int index) {
addURLToDownload(url);