|
|
@ -62,37 +62,37 @@ public class QuickCrawlLink_p {
|
|
|
|
final serverObjects prop = new serverObjects();
|
|
|
|
final serverObjects prop = new serverObjects();
|
|
|
|
final Switchboard sb = (Switchboard) env;
|
|
|
|
final Switchboard sb = (Switchboard) env;
|
|
|
|
|
|
|
|
|
|
|
|
// get segment
|
|
|
|
int port = sb.getConfigInt("port", 8090);
|
|
|
|
Segment indexSegment = sb.index;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
if (post == null) {
|
|
|
|
|
|
|
|
// send back usage example
|
|
|
|
|
|
|
|
prop.put("mode", "0");
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// get the http host header
|
|
|
|
// get the http host header
|
|
|
|
|
|
|
|
if (header.containsKey(HeaderFramework.HOST)) {
|
|
|
|
final String hostSocket = header.get(HeaderFramework.HOST);
|
|
|
|
final String hostSocket = header.get(HeaderFramework.HOST);
|
|
|
|
|
|
|
|
final int pos = hostSocket.indexOf(':', 0);
|
|
|
|
//String host = hostSocket;
|
|
|
|
|
|
|
|
int port = 80;
|
|
|
|
|
|
|
|
final int pos = hostSocket.indexOf(':',0);
|
|
|
|
|
|
|
|
if (pos != -1) {
|
|
|
|
if (pos != -1) {
|
|
|
|
port = NumberTools.parseIntDecSubstring(hostSocket, pos + 1);
|
|
|
|
port = NumberTools.parseIntDecSubstring(hostSocket, pos + 1);
|
|
|
|
//host = hostSocket.substring(0, pos);
|
|
|
|
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
prop.put("mode_host", Domains.LOCALHOST);
|
|
|
|
prop.put("mode_host", Domains.LOCALHOST);
|
|
|
|
prop.put("mode_port", port);
|
|
|
|
prop.put("mode_port", port);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
if (post == null) {
|
|
|
|
|
|
|
|
// send back usage example
|
|
|
|
|
|
|
|
prop.put("mode", "0");
|
|
|
|
return prop;
|
|
|
|
return prop;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
prop.put("mode", "1");
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// get the URL
|
|
|
|
// get the URL
|
|
|
|
String crawlingStart = post.get("url",null);
|
|
|
|
String crawlingStart = post.get("url",null);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
if (crawlingStart != null) {
|
|
|
|
|
|
|
|
prop.put("mode", "1");
|
|
|
|
crawlingStart = UTF8.decodeURL(crawlingStart);
|
|
|
|
crawlingStart = UTF8.decodeURL(crawlingStart);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// get segment
|
|
|
|
|
|
|
|
Segment indexSegment = sb.index;
|
|
|
|
|
|
|
|
|
|
|
|
// get the browser title
|
|
|
|
// get the browser title
|
|
|
|
final String title = post.get("title",null);
|
|
|
|
final String title = post.get("title", null);
|
|
|
|
|
|
|
|
|
|
|
|
// get other parameters if set
|
|
|
|
// get other parameters if set
|
|
|
|
final String crawlingMustMatch = post.get("mustmatch", CrawlProfile.MATCH_ALL_STRING);
|
|
|
|
final String crawlingMustMatch = post.get("mustmatch", CrawlProfile.MATCH_ALL_STRING);
|
|
|
@ -111,7 +111,6 @@ public class QuickCrawlLink_p {
|
|
|
|
prop.put("mode_url", (crawlingStart == null) ? "unknown" : crawlingStart);
|
|
|
|
prop.put("mode_url", (crawlingStart == null) ? "unknown" : crawlingStart);
|
|
|
|
prop.putHTML("mode_title", (title == null) ? "unknown" : title);
|
|
|
|
prop.putHTML("mode_title", (title == null) ? "unknown" : title);
|
|
|
|
|
|
|
|
|
|
|
|
if (crawlingStart != null) {
|
|
|
|
|
|
|
|
crawlingStart = crawlingStart.trim();
|
|
|
|
crawlingStart = crawlingStart.trim();
|
|
|
|
try {crawlingStart = new DigestURL(crawlingStart).toNormalform(true);} catch (final MalformedURLException e1) {}
|
|
|
|
try {crawlingStart = new DigestURL(crawlingStart).toNormalform(true);} catch (final MalformedURLException e1) {}
|
|
|
|
|
|
|
|
|
|
|
|