/**
* 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;
}
}
