@Override
public User getUser(Renderer renderer, String username, String password) throws IOException {
if (StringUtils.isEmpty(username) || StringUtils.isEmpty(password))
throw new AuthException("Username or password is empty");
if (StringUtils.isEmpty(renderer.getAuthServer()))
throw new AuthException("No auth server given, check the parameters of the renderer");
ActiveDirectory activeDirectory = null;
try {
String domain = renderer.getAuthDomain();
String authServer = renderer.getAuthServer();
User user = AuthUserCache.INSTANCE.get(username, domain);
if (user != null) return user;
NtlmPasswordAuthentication ntlmAuth = getNtlmAuth(renderer, username, password);
UniAddress dc = UniAddress.getByName(authServer, true);
SmbSession.logon(dc, ntlmAuth);
activeDirectory =
new ActiveDirectory(authServer, ntlmAuth.getUsername(), ntlmAuth.getPassword(), domain);
NamingEnumeration<SearchResult> result = activeDirectory.findUser(username);
Attributes attrs = ActiveDirectory.getAttributes(result);
if (attrs == null) throw new AuthException("No user found: " + username);
String userId = ActiveDirectory.getObjectSID(attrs);
List<ADGroup> groups = new ArrayList<ADGroup>();
activeDirectory.findUserGroups(attrs, groups);
String dnUser = ActiveDirectory.getStringAttribute(attrs, "DistinguishedName");
if (!StringUtils.isEmpty(dnUser)) activeDirectory.findUserGroup(dnUser, groups);
Logging.info("USER authenticated: " + user);
user =
new User(
userId.toLowerCase(),
username.toLowerCase(),
password,
ActiveDirectory.toArray(groups, "everyone"),
ActiveDirectory.getDisplayString(domain, username));
AuthUserCache.INSTANCE.add(username, domain, user);
return user;
} catch (SmbAuthException e) {
Logging.warn(e);
throw new AuthException("Authentication error (SmbAuthException) : " + e.getMessage());
} catch (UnknownHostException e) {
Logging.warn(e);
throw new AuthException("Authentication error (UnknownHostException) : " + e.getMessage());
} catch (NamingException e) {
Logging.warn(e);
throw new AuthException("LDAP error (NamingException) : " + e.getMessage());
} finally {
IOUtils.close(activeDirectory);
}
}
private void setTime(String t) {
if (t == null) return;
try {
time = contentLengthFormat.parse(t).intValue();
} catch (ParseException e) {
Logging.warn(e.getMessage());
}
}
private void setCrawlDate(String d) {
if (d == null) return;
try {
crawlDate = dateFormat.parse(d).getTime();
} catch (ParseException e) {
Logging.warn(e.getMessage());
}
}
@Override
public void release() {
lock.rl.lock();
try {
if (browserDriver == null) return;
browserDriver.close();
} catch (IOException e) {
Logging.warn(e);
} finally {
lock.rl.unlock();
}
}
private final void addDiscoverLink(
String href,
Origin origin,
String parentUrl,
int nextDepth,
URL currentURL,
UrlFilterItem[] urlFilterList,
List<LinkItem> newUrlList) {
if (href == null) return;
try {
URL url =
currentURL != null
? LinkUtils.getLink(currentURL, href, urlFilterList, false)
: LinkUtils.newEncodedURL(href);
if (exclusionMatcher != null) if (exclusionMatcher.matchPattern(url, null)) return;
if (inclusionMatcher != null) if (!inclusionMatcher.matchPattern(url, null)) return;
newUrlList.add(new LinkItem(url.toExternalForm(), origin, parentUrl, nextDepth));
} catch (MalformedURLException e) {
Logging.warn(href + " " + e.getMessage(), e);
} catch (URISyntaxException e) {
Logging.warn(href + " " + e.getMessage(), e);
}
}
/**
* Download the file and extract content informations
*
* @param httpDownloader
*/
public DownloadItem download(HttpDownloader httpDownloader) {
synchronized (this) {
InputStream is = null;
DownloadItem downloadItem = null;
try {
URL url = urlItem.getURL();
if (url == null) throw new MalformedURLException("Malformed URL: " + urlItem.getUrl());
// URL normalisation
URI uri = url.toURI();
url = uri.toURL();
credentialItem = credentialManager == null ? null : credentialManager.matchCredential(url);
String externalFormUrl = url.toExternalForm();
downloadItem = crawlCacheManager.loadCache(uri);
boolean fromCache = (downloadItem != null);
if (!fromCache) {
List<CookieItem> cookieList = cookieManager.getItems(externalFormUrl);
List<HeaderItem> headerList = headerManager.getItems(externalFormUrl);
downloadItem = httpDownloader.get(uri, credentialItem, headerList, cookieList);
} else if (Logging.isDebug) Logging.debug("Crawl cache deliver: " + uri);
urlItem.setContentDispositionFilename(downloadItem.getContentDispositionFilename());
urlItem.setContentBaseType(downloadItem.getContentBaseType());
urlItem.setContentTypeCharset(downloadItem.getContentTypeCharset());
urlItem.setContentEncoding(downloadItem.getContentEncoding());
urlItem.setContentLength(downloadItem.getContentLength());
urlItem.setLastModifiedDate(downloadItem.getLastModified());
urlItem.setFetchStatus(FetchStatus.FETCHED);
urlItem.setHeaders(downloadItem.getHeaders());
Integer code = downloadItem.getStatusCode();
if (code == null) throw new IOException("Http status is null");
urlItem.setResponseCode(code);
redirectUrlLocation = downloadItem.getRedirectLocation();
if (redirectUrlLocation != null)
urlItem.setRedirectionUrl(redirectUrlLocation.toURL().toExternalForm());
urlItem.setBacklinkCount(config.getUrlManager().countBackLinks(urlItem.getUrl()));
if (code >= 200 && code < 300) {
if (!fromCache) is = crawlCacheManager.storeCache(downloadItem);
else is = downloadItem.getContentInputStream();
parseContent(is);
} else if (code == 301) {
urlItem.setFetchStatus(FetchStatus.REDIR_PERM);
} else if (code > 301 && code < 400) {
urlItem.setFetchStatus(FetchStatus.REDIR_TEMP);
} else if (code >= 400 && code < 500) {
urlItem.setFetchStatus(FetchStatus.GONE);
} else if (code >= 500 && code < 600) {
urlItem.setFetchStatus(FetchStatus.HTTP_ERROR);
}
} catch (FileNotFoundException e) {
Logging.info("FileNotFound: " + urlItem.getUrl());
urlItem.setFetchStatus(FetchStatus.GONE);
setError("FileNotFound: " + urlItem.getUrl());
} catch (LimitException e) {
Logging.warn(e.toString() + " (" + urlItem.getUrl() + ")");
urlItem.setFetchStatus(FetchStatus.SIZE_EXCEED);
setError(e.getMessage());
} catch (InstantiationException e) {
Logging.error(e.getMessage(), e);
urlItem.setParserStatus(ParserStatus.PARSER_ERROR);
setError(e.getMessage());
} catch (IllegalAccessException e) {
Logging.error(e.getMessage(), e);
urlItem.setParserStatus(ParserStatus.PARSER_ERROR);
setError(e.getMessage());
} catch (ClassNotFoundException e) {
Logging.error(e.getMessage(), e);
urlItem.setParserStatus(ParserStatus.PARSER_ERROR);
setError(e.getMessage());
} catch (URISyntaxException e) {
Logging.warn(e.getMessage(), e);
urlItem.setFetchStatus(FetchStatus.URL_ERROR);
setError(e.getMessage());
} catch (MalformedURLException e) {
Logging.warn(e.getMessage(), e);
urlItem.setFetchStatus(FetchStatus.URL_ERROR);
setError(e.getMessage());
} catch (IOException e) {
Logging.error(e.getMessage(), e);
urlItem.setFetchStatus(FetchStatus.ERROR);
setError(e.getMessage());
} catch (IllegalArgumentException e) {
Logging.error(e.getMessage(), e);
urlItem.setFetchStatus(FetchStatus.ERROR);
setError(e.getMessage());
} catch (Exception e) {
Logging.error(e.getMessage(), e);
urlItem.setFetchStatus(FetchStatus.ERROR);
setError(e.getMessage());
} finally {
IOUtils.close(is);
}
return downloadItem;
}
}
