Completely reformatted PahealRipper; add missing braces.

This commit is contained in:
MetaPrime 2017-05-09 15:36:37 -07:00
parent d2a4412a85
commit 56093cc1ad

View File

@ -21,18 +21,14 @@ import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;
/**
*
* @author
*/
public class PahealRipper extends AbstractHTMLRipper{
private static Map<String,String> cookies=null;
private static Pattern gidPattern=null;
public class PahealRipper extends AbstractHTMLRipper {
private static Map<String, String> cookies = null;
private static Pattern gidPattern = null;
private static Map<String, String> getCookies() {
if(cookies==null){
cookies=new HashMap<String, String>(1);
cookies.put("ui-tnc-agreed","true");
if (cookies == null) {
cookies = new HashMap<String, String>(1);
cookies.put("ui-tnc-agreed", "true");
}
return cookies;
}
@ -53,26 +49,28 @@ public class PahealRipper extends AbstractHTMLRipper{
@Override
public Document getFirstPage() throws IOException {
return Http.url("http://rule34.paheal.net/post/list/"+getTerm(url)+"/1").cookies(getCookies()).get();
return Http.url("http://rule34.paheal.net/post/list/" + getTerm(url) + "/1").cookies(getCookies()).get();
}
@Override
public Document getNextPage(Document page) throws IOException {
for(Element e:page.select("#paginator a")){
if(e.text().toLowerCase().equals("next"))
for (Element e : page.select("#paginator a")) {
if (e.text().toLowerCase().equals("next")) {
return Http.url(e.absUrl("href")).cookies(getCookies()).get();
}
}
return null;
}
@Override
public List<String> getURLsFromPage(Document page) {
Elements elements=page.select(".shm-thumb.thumb>a").not(".shm-thumb-link");
List<String> res=new ArrayList<String>(elements.size());
Elements elements = page.select(".shm-thumb.thumb>a").not(".shm-thumb-link");
List<String> res = new ArrayList<String>(elements.size());
for(Element e:elements)
for (Element e : elements) {
res.add(e.absUrl("href"));
}
return res;
}
@ -80,16 +78,20 @@ public class PahealRipper extends AbstractHTMLRipper{
@Override
public void downloadURL(URL url, int index) {
try {
String name=url.getPath();
String ext=".png";
String name = url.getPath();
String ext = ".png";
name=name.substring(name.lastIndexOf('/')+1);
if(name.indexOf('.')>=0){
ext=name.substring(name.lastIndexOf('.'));
name=name.substring(0,name.length()-ext.length());
name = name.substring(name.lastIndexOf('/') + 1);
if (name.indexOf('.') >= 0) {
ext = name.substring(name.lastIndexOf('.'));
name = name.substring(0, name.length() - ext.length());
}
addURLToDownload(url,new File(workingDir.getCanonicalPath()+File.separator+Utils.filesystemSafe(new URI(name).getPath())+ext));
File outFile = new File(workingDir.getCanonicalPath()
+ File.separator
+ Utils.filesystemSafe(new URI(name).getPath())
+ ext);
addURLToDownload(url, outFile);
} catch (IOException ex) {
Logger.getLogger(PahealRipper.class.getName()).log(Level.SEVERE, null, ex);
} catch (URISyntaxException ex) {
@ -97,15 +99,17 @@ public class PahealRipper extends AbstractHTMLRipper{
}
}
private String getTerm(URL url) throws MalformedURLException{
if(gidPattern==null)
gidPattern=Pattern.compile("^https?://(www\\.)?rule34\\.paheal\\.net/post/list/([a-zA-Z0-9$_.+!*'(),%-]+)(/.*)?(#.*)?$");
private String getTerm(URL url) throws MalformedURLException {
if (gidPattern == null) {
gidPattern = Pattern.compile("^https?://(www\\.)?rule34\\.paheal\\.net/post/list/([a-zA-Z0-9$_.+!*'(),%-]+)(/.*)?(#.*)?$");
}
Matcher m = gidPattern.matcher(url.toExternalForm());
if(m.matches())
if (m.matches()) {
return m.group(2);
}
throw new MalformedURLException("Expected paheal.net URL format: rule34.paheal.net/post/list/searchterm - got "+url+" instead");
throw new MalformedURLException("Expected paheal.net URL format: rule34.paheal.net/post/list/searchterm - got " + url + " instead");
}
@Override
@ -116,7 +120,6 @@ public class PahealRipper extends AbstractHTMLRipper{
Logger.getLogger(PahealRipper.class.getName()).log(Level.SEVERE, null, ex);
}
throw new MalformedURLException("Expected paheal.net URL format: rule34.paheal.net/post/list/searchterm - got "+url+" instead");
throw new MalformedURLException("Expected paheal.net URL format: rule34.paheal.net/post/list/searchterm - got " + url + " instead");
}
}