Merge pull request #693 from cyian-1756/e621UrlRegexFix

E621 ripper now accepts urls with order:Score at the end
This commit is contained in:
cyian-1756 2018-06-15 15:05:14 -04:00 committed by GitHub
commit fa3d149bd2
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -96,40 +96,40 @@ public class E621Ripper extends AbstractHTMLRipper{
private String getTerm(URL url) throws MalformedURLException{
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)
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());
if(m.matches())
if(m.matches()) {
LOGGER.info(m.group(2));
return m.group(2);
}
m = gidPatternPool.matcher(url.toExternalForm());
if(m.matches())
if(m.matches()) {
return m.group(2);
}
throw new MalformedURLException("Expected e621.net URL format: e621.net/post/index/1/searchterm - got "+url+" instead");
}
@Override
public String getGID(URL url) throws MalformedURLException {
try {
String prefix="";
if(url.getPath().startsWith("/pool/show/"))
prefix="pool_";
return Utils.filesystemSafe(prefix+new URI(getTerm(url)).getPath());
} catch (URISyntaxException ex) {
logger.error(ex);
String prefix="";
if (url.getPath().startsWith("/pool/show/")) {
prefix = "pool_";
}
throw new MalformedURLException("Expected e621.net URL format: e621.net/post/index/1/searchterm - got "+url+" instead");
return Utils.filesystemSafe(prefix+getTerm(url));
}
@Override
public URL sanitizeURL(URL url) throws MalformedURLException {
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());
if(m.matches())