mirror/ripme
1
0
Fork 0
mirror of https://github.com/RipMeApp/ripme.git synced 2025-08-26 15:24:51 +02:00
Code Issues Releases Wiki Activity

Fixed E621 ripper not ripping.

Browse Source
This commit is contained in:
Tushar
2019-01-25 12:24:51 +05:30
parent e94ef1b368
commit e8b1c0ec84
2 changed files with 145 additions and 107 deletions
Download Patch File Download Diff File Expand all files Collapse all files

92
src/main/java/com/rarchives/ripme/ripper/rippers/E621Ripper.java
View File

@@ -6,8 +6,6 @@ import com.rarchives.ripme.utils.Http;
import com.rarchives.ripme.utils.Utils; import com.rarchives.ripme.utils.Utils;
import java.io.IOException; import java.io.IOException;
import java.net.MalformedURLException; import java.net.MalformedURLException;
import java.net.URI;
import java.net.URISyntaxException;
import java.net.URL; import java.net.URL;
import java.util.ArrayList; import java.util.ArrayList;
import java.util.List; import java.util.List;
@@ -54,27 +52,6 @@ public class E621Ripper extends AbstractHTMLRipper{
return Http.url("https://e621.net/post/index/1/" + getTerm(url)).get(); return Http.url("https://e621.net/post/index/1/" + getTerm(url)).get();
} }
private String getFullSizedImage(String url) {
try {
Document page = Http.url("https://e621.net" + url).get();
Elements video = page.select("video > source");
Elements flash = page.select("embed");
Elements image = page.select("a#highres");
if (video.size() > 0) {
return video.attr("src");
} else if (flash.size() > 0) {
return flash.attr("src");
} else if (image.size() > 0) {
return image.attr("href");
} else {
throw new IOException();
}
} catch (IOException e) {
logger.error("Unable to get full sized image from " + url);
return null;
}
}
@Override @Override
public List<String> getURLsFromPage(Document page) { public List<String> getURLsFromPage(Document page) {
Elements elements = page.select("div > span.thumb > a"); Elements elements = page.select("div > span.thumb > a");
@@ -82,10 +59,7 @@ public class E621Ripper extends AbstractHTMLRipper{
for (Element e : elements) { for (Element e : elements) {
if (!e.attr("href").isEmpty()) { if (!e.attr("href").isEmpty()) {
String fullSizedImage = getFullSizedImage(e.attr("href")); res.add(e.attr("abs:href"));
if (fullSizedImage != null && !fullSizedImage.equals("")) {
res.add(getFullSizedImage(e.attr("href")));
}
} }
} }
@@ -94,23 +68,26 @@ public class E621Ripper extends AbstractHTMLRipper{
@Override @Override
public Document getNextPage(Document page) throws IOException { public Document getNextPage(Document page) throws IOException {
if (page.select("a.next_page") != null) { if (!page.select("a.next_page").isEmpty()) {
return Http.url("https://e621.net" + page.select("a.next_page").attr("href")).get(); return Http.url(page.select("a.next_page").attr("abs:href")).get();
} else { } else {
throw new IOException("No more pages"); throw new IOException("No more pages.");
} }
} }
@Override @Override
public void downloadURL(final URL url, int index) { public void downloadURL(final URL url, int index) {
addURLToDownload(url, getPrefix(index)); // addURLToDownload(url, getPrefix(index));
e621ThreadPool.addThread(new E621FileThread(url, getPrefix(index)));
} }
private String getTerm(URL url) throws MalformedURLException { private String getTerm(URL url) throws MalformedURLException {
if (gidPattern == null) if (gidPattern == null)
gidPattern=Pattern.compile("^https?://(www\\.)?e621\\.net/post/index/[^/]+/([a-zA-Z0-9$_.+!*'():,%\\-]+)(/.*)?(#.*)?$"); gidPattern = Pattern.compile(
"^https?://(www\\.)?e621\\.net/post/index/[^/]+/([a-zA-Z0-9$_.+!*'():,%\\-]+)(/.*)?(#.*)?$");
if (gidPatternPool == null) if (gidPatternPool == null)
gidPatternPool=Pattern.compile("^https?://(www\\.)?e621\\.net/pool/show/([a-zA-Z0-9$_.+!*'(),%:\\-]+)(\\?.*)?(/.*)?(#.*)?$"); gidPatternPool = Pattern.compile(
"^https?://(www\\.)?e621\\.net/pool/show/([a-zA-Z0-9$_.+!*'(),%:\\-]+)(\\?.*)?(/.*)?(#.*)?$");
Matcher m = gidPattern.matcher(url.toExternalForm()); Matcher m = gidPattern.matcher(url.toExternalForm());
if (m.matches()) { if (m.matches()) {
@@ -123,25 +100,24 @@ public class E621Ripper extends AbstractHTMLRipper{
return m.group(2); return m.group(2);
} }
throw new MalformedURLException("Expected e621.net URL format: e621.net/post/index/1/searchterm - got "+url+" instead"); throw new MalformedURLException(
"Expected e621.net URL format: e621.net/post/index/1/searchterm - got " + url + " instead");
} }
@Override @Override
public String getGID(URL url) throws MalformedURLException { public String getGID(URL url) throws MalformedURLException {
String prefix = ""; String prefix = "";
if (url.getPath().startsWith("/pool/show/")) { if (url.getPath().startsWith("/pool/show/")) {
prefix = "pool_"; prefix = "pool_";
} }
return Utils.filesystemSafe(prefix + getTerm(url)); return Utils.filesystemSafe(prefix + getTerm(url));
} }
@Override @Override
public URL sanitizeURL(URL url) throws MalformedURLException { public URL sanitizeURL(URL url) throws MalformedURLException {
if (gidPattern2 == null) if (gidPattern2 == null)
gidPattern2=Pattern.compile("^https?://(www\\.)?e621\\.net/post/search\\?tags=([a-zA-Z0-9$_.+!*'():,%-]+)(/.*)?(#.*)?$"); gidPattern2 = Pattern.compile(
"^https?://(www\\.)?e621\\.net/post/search\\?tags=([a-zA-Z0-9$_.+!*'():,%-]+)(/.*)?(#.*)?$");
Matcher m = gidPattern2.matcher(url.toExternalForm()); Matcher m = gidPattern2.matcher(url.toExternalForm());
if (m.matches()) if (m.matches())
@@ -150,4 +126,44 @@ public class E621Ripper extends AbstractHTMLRipper{
return url; return url;
} }
public class E621FileThread extends Thread {
private URL url;
private String index;
public E621FileThread(URL url, String index) {
this.url = url;
this.index = index;
}
@Override
public void run() {
try {
String fullSizedImage = getFullSizedImage(url);
if (fullSizedImage != null && !fullSizedImage.equals("")) {
addURLToDownload(new URL(fullSizedImage), index);
}
} catch (IOException e) {
logger.error("Unable to get full sized image from " + url);
}
}
private String getFullSizedImage(URL imageURL) throws IOException {
Document page = Http.url(imageURL).retries(3).get();
Elements video = page.select("video > source");
Elements flash = page.select("embed");
Elements image = page.select("a#highres");
if (video.size() > 0) {
return video.attr("src");
} else if (flash.size() > 0) {
return flash.attr("src");
} else if (image.size() > 0) {
return image.attr("href");
} else {
throw new IOException();
}
}
}
} }

22
src/test/java/com/rarchives/ripme/tst/ripper/rippers/E621RipperTest.java
View File

@@ -10,4 +10,26 @@ public class E621RipperTest extends RippersTest {
E621Ripper ripper = new E621Ripper(new URL("https://e621.net/post/index/1/beach")); E621Ripper ripper = new E621Ripper(new URL("https://e621.net/post/index/1/beach"));
testRipper(ripper); testRipper(ripper);
} }
public void testFlashOrWebm() throws IOException {
E621Ripper ripper = new E621Ripper(new URL("https://e621.net/post/index/1/gif"));
testRipper(ripper);
}
public void testGetNextPage() throws IOException {
E621Ripper nextPageRipper = new E621Ripper(new URL("https://e621.net/post/index/1/cosmicminerals"));
try {
nextPageRipper.getNextPage(nextPageRipper.getFirstPage());
assert (true);
} catch (IOException e) {
throw e;
}
E621Ripper noNextPageRipper = new E621Ripper(new URL("https://e621.net/post/index/1/cosmicminerals"));
try {
noNextPageRipper.getNextPage(noNextPageRipper.getFirstPage());
} catch (IOException e) {
assertEquals(e.getMessage(), "No more pages.");
}
}
} }