boxtec
/
yacy_search_server
mirror of https://github.com/yacy/yacy_search_server
4
1
Fork 0
Code Issues Projects Releases Wiki Activity

showing /16 and /24 subnet option instead of 'bigrange' in network

Browse Source 
scanner
pull/1/head
Michael Christen 13 years ago
parent 3f1dc77743
commit 0d1042363c
2 changed files with 130 additions and 57 deletions
Show Stats Download Patch File Download Diff File

9
htroot/CrawlStartScanner_p.html
Unescape View File

@ -46,9 +46,16 @@
<input type="text" name="scanhosts" value="#[scanhosts]#" size="60" maxlength="400" /><br/><br/>
<input type="radio" name="source" id="sourcenet" value="intranet"#(intranet.checked)#:: checked="checked"#(/intranet.checked)# />Full Intranet Scan: #[intranethosts]#<br/>
timeout: <input type="text" name="timeout" value ="100" size="4"/> ms, <input type="checkbox" name="bigrange" />bigrange
#(intranetHint)#::<div class="info">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Do not use intranet scan results, you are not in an intranet environment!</div>#(/intranetHint)#
</dd>
<dt>Time-Out</dt>
<dd>
<input type="text" name="timeout" value ="100" size="4"/> ms
</dd>
<dt>Subnet</dt>
<dd>
<input type="radio" name="subnet" value="24" checked="checked"/>/24 <input type="radio" name="subnet" value="16"/>/16
</dd>
<dt>Scan Cache</dt>
<dd>
<input type="checkbox" name="accumulatescancache" id="accumulatescancache" checked="checked" />accumulate scan results with access type "granted" into scan cache (do not delete old scan result)

178
htroot/CrawlStartScanner_p.java
Unescape View File

@ -18,7 +18,6 @@
* If not, see <http://www.gnu.org/licenses/>.
*/
import java.net.InetAddress;
import java.net.MalformedURLException;
import java.util.ConcurrentModificationException;
@ -42,14 +41,18 @@ import de.anomic.data.WorkTables;
import de.anomic.server.serverObjects;
import de.anomic.server.serverSwitch;
public class CrawlStartScanner_p {
public class CrawlStartScanner_p
{
private final static int CONCURRENT_RUNNER = 100;
public static serverObjects respond(final RequestHeader header, final serverObjects post, final serverSwitch env) {
public static serverObjects respond(
final RequestHeader header,
final serverObjects post,
final serverSwitch env) {
final serverObjects prop = new serverObjects();
final Switchboard sb = (Switchboard)env;
final Switchboard sb = (Switchboard) env;
// clean up all search events
SearchEventCache.cleanupEvents(true);
@ -66,71 +69,112 @@ public class CrawlStartScanner_p {
final Set<InetAddress> ips = Domains.myIntranetIPs();
prop.put("intranethosts", ips.toString());
prop.put("intranetHint", sb.isIntranetMode() ? 0 : 1);
if (hosts.length() == 0) {
if ( hosts.length() == 0 ) {
InetAddress ip;
if (sb.isIntranetMode()) {
if (ips.size() > 0) ip = ips.iterator().next();
else ip = Domains.dnsResolve("192.168.0.1");
if ( sb.isIntranetMode() ) {
if ( ips.size() > 0 ) {
ip = ips.iterator().next();
} else {
ip = Domains.dnsResolve("192.168.0.1");
}
} else {
ip = Domains.myPublicLocalIP();
if (Domains.isThisHostIP(ip)) ip = sb.peers.mySeed().getInetAddress();
if ( Domains.isThisHostIP(ip) ) {
ip = sb.peers.mySeed().getInetAddress();
}
}
if ( ip != null ) {
hosts = ip.getHostAddress();
}
if (ip != null) hosts = ip.getHostAddress();
}
prop.put("scanhosts", hosts);
// parse post requests
if (post != null) {
if ( post != null ) {
int repeat_time = 0;
String repeat_unit = "seldays";
long validTime = 0;
// check scheduler
if (post.get("rescan", "").equals("scheduler")) {
if ( post.get("rescan", "").equals("scheduler") ) {
repeat_time = post.getInt("repeat_time", -1);
repeat_unit = post.get("repeat_unit", "selminutes"); // selminutes, selhours, seldays
if (repeat_unit.equals("selminutes")) validTime = repeat_time * 60 * 1000;
if (repeat_unit.equals("selhours")) validTime = repeat_time * 60 * 60 * 1000;
if (repeat_unit.equals("seldays")) validTime = repeat_time * 24 * 60 * 60 * 1000;
if ( repeat_unit.equals("selminutes") ) {
validTime = repeat_time * 60 * 1000;
}
if ( repeat_unit.equals("selhours") ) {
validTime = repeat_time * 60 * 60 * 1000;
}
if ( repeat_unit.equals("seldays") ) {
validTime = repeat_time * 24 * 60 * 60 * 1000;
}
}
final boolean bigrange = post.getBoolean("bigrange", false);
final boolean bigrange = post.get("subnet", "24").equals("16");
// case: an IP range was given; scan the range for services and display result
if (post.containsKey("scan") && "hosts".equals(post.get("source", ""))) {
if ( post.containsKey("scan") && "hosts".equals(post.get("source", "")) ) {
final Set<InetAddress> ia = new HashSet<InetAddress>();
for (String host : hosts.split(",")) {
if (host.startsWith("http://")) host = host.substring(7);
if (host.startsWith("https://")) host = host.substring(8);
if (host.startsWith("ftp://")) host = host.substring(6);
if (host.startsWith("smb://")) host = host.substring(6);
final int p = host.indexOf('/',0);
if (p >= 0) host = host.substring(0, p);
for ( String host : hosts.split(",") ) {
if ( host.startsWith("http://") ) {
host = host.substring(7);
}
if ( host.startsWith("https://") ) {
host = host.substring(8);
}
if ( host.startsWith("ftp://") ) {
host = host.substring(6);
}
if ( host.startsWith("smb://") ) {
host = host.substring(6);
}
final int p = host.indexOf('/', 0);
if ( p >= 0 ) {
host = host.substring(0, p);
}
ia.add(Domains.dnsResolve(host));
}
final Scanner scanner = new Scanner(ia, CONCURRENT_RUNNER, timeout);
if (post.get("scanftp", "").equals("on")) scanner.addFTP(bigrange);
if (post.get("scanhttp", "").equals("on")) scanner.addHTTP(bigrange);
if (post.get("scanhttps", "").equals("on")) scanner.addHTTPS(bigrange);
if (post.get("scansmb", "").equals("on")) scanner.addSMB(bigrange);
if ( post.get("scanftp", "").equals("on") ) {
scanner.addFTP(bigrange);
}
if ( post.get("scanhttp", "").equals("on") ) {
scanner.addHTTP(bigrange);
}
if ( post.get("scanhttps", "").equals("on") ) {
scanner.addHTTPS(bigrange);
}
if ( post.get("scansmb", "").equals("on") ) {
scanner.addSMB(bigrange);
}
scanner.start();
scanner.terminate();
if ("on".equals(post.get("accumulatescancache", "")) && !"scheduler".equals(post.get("rescan", ""))) {
if ( "on".equals(post.get("accumulatescancache", ""))
&& !"scheduler".equals(post.get("rescan", "")) ) {
Scanner.scancacheExtend(scanner, validTime);
} else {
Scanner.scancacheReplace(scanner, validTime);
}
}
if (post.containsKey("scan") && "intranet".equals(post.get("source", ""))) {
if ( post.containsKey("scan") && "intranet".equals(post.get("source", "")) ) {
final Scanner scanner = new Scanner(Domains.myIntranetIPs(), CONCURRENT_RUNNER, timeout);
if ("on".equals(post.get("scanftp", ""))) scanner.addFTP(bigrange);
if ("on".equals(post.get("scanhttp", ""))) scanner.addHTTP(bigrange);
if ("on".equals(post.get("scanhttps", ""))) scanner.addHTTPS(bigrange);
if ("on".equals(post.get("scansmb", ""))) scanner.addSMB(bigrange);
if ( "on".equals(post.get("scanftp", "")) ) {
scanner.addFTP(bigrange);
}
if ( "on".equals(post.get("scanhttp", "")) ) {
scanner.addHTTP(bigrange);
}
if ( "on".equals(post.get("scanhttps", "")) ) {
scanner.addHTTPS(bigrange);
}
if ( "on".equals(post.get("scansmb", "")) ) {
scanner.addSMB(bigrange);
}
scanner.start();
scanner.terminate();
if ("on".equals(post.get("accumulatescancache", "")) && !"scheduler".equals(post.get("rescan", ""))) {
if ( "on".equals(post.get("accumulatescancache", ""))
&& !"scheduler".equals(post.get("rescan", "")) ) {
Scanner.scancacheExtend(scanner, validTime);
} else {
Scanner.scancacheReplace(scanner, validTime);
@ -138,68 +182,90 @@ public class CrawlStartScanner_p {
}
// check crawl request
if (post.containsKey("crawl")) {
if ( post.containsKey("crawl") ) {
// make a pk/url mapping
final Iterator<Map.Entry<Scanner.Service, Scanner.Access>> se = Scanner.scancacheEntries();
final Map<byte[], DigestURI> pkmap = new TreeMap<byte[], DigestURI>(Base64Order.enhancedCoder);
while (se.hasNext()) {
final Map<byte[], DigestURI> pkmap =
new TreeMap<byte[], DigestURI>(Base64Order.enhancedCoder);
while ( se.hasNext() ) {
final Scanner.Service u = se.next().getKey();
DigestURI uu;
try {
uu = new DigestURI(u.url());
pkmap.put(uu.hash(), uu);
} catch (final MalformedURLException e) {
} catch ( final MalformedURLException e ) {
Log.logException(e);
}
}
// search for crawl start requests in this mapping
for (final Map.Entry<String, String> entry: post.entrySet()) {
if (entry.getValue().startsWith("mark_")) {
final byte [] pk = entry.getValue().substring(5).getBytes();
for ( final Map.Entry<String, String> entry : post.entrySet() ) {
if ( entry.getValue().startsWith("mark_") ) {
final byte[] pk = entry.getValue().substring(5).getBytes();
final DigestURI url = pkmap.get(pk);
if (url != null) {
String path = "/Crawler_p.html?createBookmark=off&xsstopw=off&crawlingDomMaxPages=10000&intention=&range=domain&indexMedia=on&recrawl=nodoubles&xdstopw=off&storeHTCache=on&sitemapURL=&repeat_time=7&crawlingQ=on&cachePolicy=iffresh&indexText=on&crawlingMode=url&mustnotmatch=&crawlingDomFilterDepth=1&crawlingDomFilterCheck=off&crawlingstart=Start%20New%20Crawl&xpstopw=off&repeat_unit=seldays&crawlingDepth=99";
if ( url != null ) {
String path =
"/Crawler_p.html?createBookmark=off&xsstopw=off&crawlingDomMaxPages=10000&intention=&range=domain&indexMedia=on&recrawl=nodoubles&xdstopw=off&storeHTCache=on&sitemapURL=&repeat_time=7&crawlingQ=on&cachePolicy=iffresh&indexText=on&crawlingMode=url&mustnotmatch=&crawlingDomFilterDepth=1&crawlingDomFilterCheck=off&crawlingstart=Start%20New%20Crawl&xpstopw=off&repeat_unit=seldays&crawlingDepth=99";
path += "&crawlingURL=" + url.toNormalform(true, false);
WorkTables.execAPICall("localhost", (int) sb.getConfigLong("port", 8090), sb.getConfig(SwitchboardConstants.ADMIN_ACCOUNT_B64MD5, ""), path, pk);
WorkTables.execAPICall(
"localhost",
(int) sb.getConfigLong("port", 8090),
sb.getConfig(SwitchboardConstants.ADMIN_ACCOUNT_B64MD5, ""),
path,
pk);
}
}
}
}
// check scheduler
if ("scheduler".equals(post.get("rescan", ""))) {
if ( "scheduler".equals(post.get("rescan", "")) ) {
// store this call as api call
if (repeat_time > 0) {
if ( repeat_time > 0 ) {
// store as scheduled api call
sb.tables.recordAPICall(post, "CrawlStartScanner_p.html", WorkTables.TABLE_API_TYPE_CRAWLER, "network scanner for hosts: " + hosts, repeat_time, repeat_unit.substring(3));
sb.tables.recordAPICall(
post,
"CrawlStartScanner_p.html",
WorkTables.TABLE_API_TYPE_CRAWLER,
"network scanner for hosts: " + hosts,
repeat_time,
repeat_unit.substring(3));
}
// execute the scan results
if (Scanner.scancacheSize() > 0) {
if ( Scanner.scancacheSize() > 0 ) {
// make a comment cache
final Map<byte[], String> apiCommentCache = WorkTables.commentCache(sb);
String urlString;
DigestURI u;
try {
final Iterator<Map.Entry<Scanner.Service, Scanner.Access>> se = Scanner.scancacheEntries();
final Iterator<Map.Entry<Scanner.Service, Scanner.Access>> se =
Scanner.scancacheEntries();
Map.Entry<Scanner.Service, Scanner.Access> host;
while (se.hasNext()) {
while ( se.hasNext() ) {
host = se.next();
try {
u = new DigestURI(host.getKey().url());
urlString = u.toNormalform(true, false);
if (host.getValue() == Access.granted && Scanner.inIndex(apiCommentCache, urlString) == null) {
String path = "/Crawler_p.html?createBookmark=off&xsstopw=off&crawlingDomMaxPages=10000&intention=&range=domain&indexMedia=on&recrawl=nodoubles&xdstopw=off&storeHTCache=on&sitemapURL=&repeat_time=7&crawlingQ=on&cachePolicy=iffresh&indexText=on&crawlingMode=url&mustnotmatch=&crawlingDomFilterDepth=1&crawlingDomFilterCheck=off&crawlingstart=Start%20New%20Crawl&xpstopw=off&repeat_unit=seldays&crawlingDepth=99";
if ( host.getValue() == Access.granted
&& Scanner.inIndex(apiCommentCache, urlString) == null ) {
String path =
"/Crawler_p.html?createBookmark=off&xsstopw=off&crawlingDomMaxPages=10000&intention=&range=domain&indexMedia=on&recrawl=nodoubles&xdstopw=off&storeHTCache=on&sitemapURL=&repeat_time=7&crawlingQ=on&cachePolicy=iffresh&indexText=on&crawlingMode=url&mustnotmatch=&crawlingDomFilterDepth=1&crawlingDomFilterCheck=off&crawlingstart=Start%20New%20Crawl&xpstopw=off&repeat_unit=seldays&crawlingDepth=99";
path += "&crawlingURL=" + urlString;
WorkTables.execAPICall("localhost", (int) sb.getConfigLong("port", 8090), sb.getConfig(SwitchboardConstants.ADMIN_ACCOUNT_B64MD5, ""), path, u.hash());
WorkTables.execAPICall(
"localhost",
(int) sb.getConfigLong("port", 8090),
sb.getConfig(SwitchboardConstants.ADMIN_ACCOUNT_B64MD5, ""),
path,
u.hash());
}
} catch (final MalformedURLException e) {
} catch ( final MalformedURLException e ) {
Log.logException(e);
}
}
} catch (final ConcurrentModificationException e) {}
} catch ( final ConcurrentModificationException e ) {
}
}
}

Write Preview
Loading…
Cancel
Save