public boolean populate(CrawlURI curi, HttpClient http, HttpMethod method, String payload) {
// http is not used.
// payload is not used.
boolean result = false;
Map formItems = null;
try {
formItems = getFormItems(curi);
} catch (AttributeNotFoundException e1) {
logger.severe("Failed get of form items for " + curi);
}
if (formItems == null || formItems.size() <= 0) {
try {
logger.severe("No form items for " + method.getURI());
} catch (URIException e) {
logger.severe("No form items and exception getting uri: " + e.getMessage());
}
return result;
}
NameValuePair[] data = new NameValuePair[formItems.size()];
int index = 0;
String key = null;
for (Iterator i = formItems.keySet().iterator(); i.hasNext(); ) {
key = (String) i.next();
data[index++] = new NameValuePair(key, (String) formItems.get(key));
}
if (method instanceof PostMethod) {
((PostMethod) method).setRequestBody(data);
result = true;
} else if (method instanceof GetMethod) {
// Append these values to the query string.
// Get current query string, then add data, then get it again
// only this time its our data only... then append.
HttpMethodBase hmb = (HttpMethodBase) method;
String currentQuery = hmb.getQueryString();
hmb.setQueryString(data);
String newQuery = hmb.getQueryString();
hmb.setQueryString(((currentQuery != null) ? currentQuery : "") + "&" + newQuery);
result = true;
} else {
logger.severe("Unknown method type: " + method);
}
return result;
}
/**
* @param since last modified time to use
* @param req
* @param url if null, ignored
* @param redirCount number of redirs we've done
*/
public static HttpData getDataOnce(
HttpServletRequest req,
HttpServletResponse res,
long since,
String surl,
int redirCount,
int timeout)
throws IOException, HttpException, DataSourceException, MalformedURLException {
HttpMethodBase request = null;
HostConfiguration hcfg = new HostConfiguration();
/*
[todo hqm 2006-02-01] Anyone know why this code was here? It is setting
the mime type to something which just confuses the DHTML parser.
if (res != null) {
res.setContentType("application/x-www-form-urlencoded;charset=UTF-8");
}
*/
try {
// TODO: [2002-01-09 bloch] cope with cache-control
// response headers (no-store, no-cache, must-revalidate,
// proxy-revalidate).
if (surl == null) {
surl = getURL(req);
}
if (surl == null || surl.equals("")) {
throw new MalformedURLException(
/* (non-Javadoc)
* @i18n.test
* @org-mes="url is empty or null"
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-312"));
}
String reqType = "";
String headers = "";
if (req != null) {
reqType = req.getParameter("reqtype");
headers = req.getParameter("headers");
}
boolean isPost = false;
mLogger.debug("reqtype = " + reqType);
if (reqType != null && reqType.equals("POST")) {
request = new LZPostMethod();
request.setRequestHeader("Content-Type", "application/x-www-form-urlencoded;charset=UTF-8");
isPost = true;
mLogger.debug("setting POST req method");
} else if (reqType != null && reqType.equals("PUT")) {
request = new LZPutMethod();
// todo [hqm 2007] treat PUT like POST?
isPost = true;
mLogger.debug("setting PUT req method");
} else if (reqType != null && reqType.equals("DELETE")) {
request = new LZDeleteMethod();
mLogger.debug("setting DELETE req method");
} else {
mLogger.debug("setting GET (default) req method");
request = new LZGetMethod();
}
request.setHttp11(mUseHttp11);
// Proxy the request headers
if (req != null) {
LZHttpUtils.proxyRequestHeaders(req, request);
}
// Set headers from query string
if (headers != null && headers.length() > 0) {
StringTokenizer st = new StringTokenizer(headers, "\n");
while (st.hasMoreTokens()) {
String h = st.nextToken();
int i = h.indexOf(":");
if (i > -1) {
String n = h.substring(0, i);
String v = h.substring(i + 2, h.length());
request.setRequestHeader(n, v);
mLogger.debug(
/* (non-Javadoc)
* @i18n.test
* @org-mes="setting header " + p[0] + "=" + p[1]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-359", new Object[] {n, v}));
}
}
}
mLogger.debug("Parsing url");
URI uri = LZHttpUtils.newURI(surl);
try {
hcfg.setHost(uri);
} catch (Exception e) {
throw new MalformedURLException(
/* (non-Javadoc)
* @i18n.test
* @org-mes="can't form uri from " + p[0]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-376", new Object[] {surl}));
}
// This gets us the url-encoded (escaped) path and query string
String path = uri.getEscapedPath();
String query = uri.getEscapedQuery();
mLogger.debug(
/* (non-Javadoc)
* @i18n.test
* @org-mes="encoded path: " + p[0]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-389", new Object[] {path}));
mLogger.debug(
/* (non-Javadoc)
* @i18n.test
* @org-mes="encoded query: " + p[0]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-397", new Object[] {query}));
// This call takes a decoded (unescaped) path
request.setPath(path);
boolean hasQuery = (query != null && query.length() > 0);
String rawcontent = null;
// Newer rawpost protocol puts lzpostbody as a separate
// top level query arg in the request.
rawcontent = req.getParameter("lzpostbody");
if (isPost) {
// Older rawpost protocol put the "lzpostbody" arg
// embedded in the "url" args's query args
if (rawcontent == null && hasQuery) {
rawcontent = findQueryArg("lzpostbody", query);
}
if (rawcontent != null) {
// Get the unescaped query string
((EntityEnclosingMethod) request).setRequestBody(rawcontent);
} else if (hasQuery) {
StringTokenizer st = new StringTokenizer(query, "&");
while (st.hasMoreTokens()) {
String it = st.nextToken();
int i = it.indexOf("=");
if (i > 0) {
String n = it.substring(0, i);
String v = it.substring(i + 1, it.length());
// POST encodes values during request
((PostMethod) request).addParameter(n, URLDecoder.decode(v, "UTF-8"));
} else {
mLogger.warn(
/* (non-Javadoc)
* @i18n.test
* @org-mes="ignoring bad token (missing '=' char) in query string: " + p[0]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-429", new Object[] {it}));
}
}
}
} else {
// This call takes an encoded (escaped) query string
request.setQueryString(query);
}
// Put in the If-Modified-Since headers
if (since != -1) {
String lms = LZHttpUtils.getDateString(since);
request.setRequestHeader(LZHttpUtils.IF_MODIFIED_SINCE, lms);
mLogger.debug(
/* (non-Javadoc)
* @i18n.test
* @org-mes="proxying lms: " + p[0]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-450", new Object[] {lms}));
}
mLogger.debug(
/* (non-Javadoc)
* @i18n.test
* @org-mes="setting up http client"
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-460"));
HttpClient htc = null;
if (mConnectionMgr != null) {
htc = new HttpClient(mConnectionMgr);
} else {
htc = new HttpClient();
}
htc.setHostConfiguration(hcfg);
// This is the data timeout
mLogger.debug(
/* (non-Javadoc)
* @i18n.test
* @org-mes="timeout set to " + p[0]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-478", new Object[] {new Integer(timeout)}));
htc.setTimeout(timeout);
// Set connection timeout the same
htc.setConnectionTimeout(mConnectionTimeout);
// Set timeout for getting a connection
htc.setHttpConnectionFactoryTimeout(mConnectionPoolTimeout);
// TODO: [2003-03-05 bloch] this should be more configurable (per app?)
if (!isPost) {
request.setFollowRedirects(mFollowRedirects > 0);
}
long t1 = System.currentTimeMillis();
mLogger.debug("starting remote request");
int rc = htc.executeMethod(hcfg, request);
String status = HttpStatus.getStatusText(rc);
if (status == null) {
status = "" + rc;
}
mLogger.debug(
/* (non-Javadoc)
* @i18n.test
* @org-mes="remote response status: " + p[0]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-504", new Object[] {status}));
HttpData data = null;
if (isRedirect(rc) && mFollowRedirects > redirCount) {
String loc = request.getResponseHeader("Location").toString();
String hostURI = loc.substring(loc.indexOf(": ") + 2, loc.length());
mLogger.info(
/* (non-Javadoc)
* @i18n.test
* @org-mes="Following URL from redirect: " + p[0]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-517", new Object[] {hostURI}));
long t2 = System.currentTimeMillis();
if (timeout > 0) {
timeout -= (t2 - t1);
if (timeout < 0) {
throw new InterruptedIOException(
/* (non-Javadoc)
* @i18n.test
* @org-mes=p[0] + " timed out after redirecting to " + p[1]
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(), "051018-529", new Object[] {surl, loc}));
}
}
data = getDataOnce(req, res, since, hostURI, redirCount++, timeout);
} else {
data = new HttpData(request, rc);
}
if (req != null && res != null) {
// proxy response headers
LZHttpUtils.proxyResponseHeaders(request, res, req.isSecure());
}
return data;
} catch (ConnectTimeoutException ce) {
// Transduce to an InterrupedIOException, since lps takes these to be timeouts.
if (request != null) {
request.releaseConnection();
}
throw new InterruptedIOException(
/* (non-Javadoc)
* @i18n.test
* @org-mes="connecting to " + p[0] + ":" + p[1] + " timed out beyond " + p[2] + " msecs."
*/
org.openlaszlo.i18n.LaszloMessages.getMessage(
HTTPDataSource.class.getName(),
"051018-557",
new Object[] {
hcfg.getHost(), new Integer(hcfg.getPort()), new Integer(mConnectionTimeout)
}));
} catch (HttpRecoverableException hre) {
if (request != null) {
request.releaseConnection();
}
throw hre;
} catch (HttpException e) {
if (request != null) {
request.releaseConnection();
}
throw e;
} catch (IOException ie) {
if (request != null) {
request.releaseConnection();
}
throw ie;
} catch (RuntimeException e) {
if (request != null) {
request.releaseConnection();
}
throw e;
}
}