package ca.ucalgary.services;
import ca.ucalgary.services.util.CGIUtils;
import ca.ucalgary.services.util.XHTMLForm;
import org.w3c.tidy.Tidy; //html parser and fixer
import org.w3c.dom.*;
import org.apache.commons.httpclient.*;
import org.apache.commons.httpclient.methods.*;
import org.apache.commons.httpclient.methods.multipart.*;
import org.apache.commons.httpclient.params.*;
import org.apache.commons.fileupload.FileItemIterator;
import org.apache.commons.fileupload.FileItemStream;
import org.apache.commons.fileupload.FileUploadBase.SizeLimitExceededException;
import org.apache.commons.fileupload.servlet.ServletFileUpload;
import org.apache.commons.io.IOUtils;
import javax.servlet.http.*;
import java.io.InputStream;
import java.io.PrintStream;
import java.net.URL;
import java.util.*;
import java.util.logging.*;
/**
* Provides an HTML interface for invoking CGI. The appearance and functionality of the class
* can be extended by registering a DataRecorder (e.g. Daggoo registers PBERecorder to get
* its functionality), please see the documentation at
* http://biomoby.open-bio.org/CVS_CONTENT/moby-live/Java/docs/soapServlet.html
*/
public class CGIServlet extends WrappingServlet{
// Passes along the list of hidden-type CGI input fields to the POST-based wrapping process
public static final String HIDDEN_PARAM = "_CGIServlet_hidden_params";
private static Logger logger = Logger.getLogger(CGIServlet.class.getName());
// Presents the CGI form as-is, but with javascript events and redirection of form submission
protected void writeServiceForm(HttpServletRequest request,
HttpServletResponse response,
URL url,
PrintStream out){
// Retrieve and instrument the CGI form
InputStream is = null;
try{
is = url.openStream();
} catch(Exception e){
out.print("Parsing Error\n"+
"\n" +
"
The URL specified (" + url + ") could not be opened
");
e.printStackTrace(out);
out.print("
\n");
return;
}
Tidy tidy = new Tidy();
tidy.setXHTML(true);
tidy.setQuiet(true);
tidy.setShowWarnings(false);
Document htmlDoc = tidy.parseDOM(is, null);
// set the href base in the html head so other parts of the page spec'ed as relative URLs show up alright
// (if not already set)
Element head = null;
NodeList heads = htmlDoc.getElementsByTagName("head");
if(heads == null || heads.getLength() == 0){
out.print("Parsing Error\n"+
"\n" +
"
The document at the specified URL (" + url + ") does not have " +
"an HTML head element
");
return;
}
head = (Element) heads.item(0);
NodeList base = head.getElementsByTagName("base");
if(base == null || base.getLength() == 0 ||
((Element) base.item(0)).getAttribute("href") == null ||
((Element) base.item(0)).getAttribute("href").length() == 0){ // could have just a tag
String baseString = null;
try{
baseString = (new URL(url, "")).toString();
} catch(Exception e){
out.print("Parsing Error\n"+
"\n" +
"
The URL to be shown (" + url + ") could not be reparsed in order to set the base href" +
"(bizzare, because we were able to load the URL contents in the first place...)
");
return;
}
Element baseEl = htmlDoc.createElement("base");
baseEl.setAttribute("href", baseString); //gets page's URL context
if(head.getFirstChild() != null){
// put as first child so CSS etc. declared in head resolve properly too
head.insertBefore(baseEl, head.getFirstChild());
}
else{
head.appendChild(baseEl); // insert into blank head
}
if(recorder != null){
for(Node newHeadNode: recorder.getHeadAsDOM(request, htmlDoc)){
head.appendChild(newHeadNode);
}
}
}
Element body = null;
NodeList bodies = htmlDoc.getElementsByTagName("body");
if(bodies == null || bodies.getLength() == 0){
out.print("Parsing Error\n"+
"\n" +
"
The document at the specified URL (" + url + ") does not have " +
"an HTML body element
");
return;
}
body = (Element) bodies.item(0);
if(recorder != null){
for(Attr newBodyAttr: recorder.getBodyAttrsAsDOM(request, htmlDoc)){
String existingEvent = body.getAttribute(newBodyAttr.getName());
if(existingEvent == null || existingEvent.trim().length() == 0){
body.setAttributeNode(newBodyAttr);
}
else{
// string existing and new JS (or CSS) statements together with a semi-colon
body.setAttribute(newBodyAttr.getName(), existingEvent+";"+newBodyAttr.getValue());
}
}
Node[] newBodyNodes = recorder.getBodyAsDOM(request, htmlDoc);
if(body.getFirstChild() == null){
body.appendChild(htmlDoc.createComment("blank document body"));
}
for(int i = newBodyNodes.length-1; i >= 0; i--){
// this way they end up in the DOM in the same order as in the array
body.insertBefore(newBodyNodes[i], body.getFirstChild());
}
}
// change the form submission target. Doing this after the event handler adding so we
// don't add events to our own hidden inputs
NodeList forms = htmlDoc.getElementsByTagName("form");
for(int i = 0; i < forms.getLength(); i++){
Element form = (Element) forms.item(i);
//resolve location of action relative to form's URL
URL origFormAction = null;
try{
origFormAction = new URL(url, form.getAttribute("action"));
} catch(Exception e){
logger.log(Level.WARNING, "Could not parse the form action URL ("+
form.getAttribute("action")+"), skipping the form:", e);
continue;
}
// add the javascript event handlers needed by the data recorder
MyNodeList inputs = new MyNodeList();
inputs.add(form.getElementsByTagName("input"));
inputs.add(form.getElementsByTagName("select"));
inputs.add(form.getElementsByTagName("textarea"));
if(recorder != null && recorder.getOnEventAsDOM(htmlDoc) != null){
List hiddenParamNames = new ArrayList();
for(int j = 0; j < inputs.getLength(); j++){
Attr eventAttr = recorder.getOnEventAsDOM(htmlDoc); //need new Node each time to not confuse doc
Element input = (Element) inputs.item(j);
if(input.getNodeName().equals("input") &&
"hidden".equals(input.getAttribute("type"))){
// note for later use by the wrapping system
hiddenParamNames.add(input.getAttribute("name"));
}
// only need to instrument interactive form fields: note this includes those hidden by CSS
else{
//See if there is already a Javascript event. If so append, if not, add one.
String existingEvent = input.getAttribute(eventAttr.getName());
if(existingEvent == null || existingEvent.trim().length() ==0){
// just set it, not cloberring anything
input.setAttributeNode(eventAttr);
}
else{
// separate JS statements with a semi-colon
input.setAttribute(eventAttr.getName(), existingEvent+";"+eventAttr.getValue());
}
}
}
form.appendChild(createHiddenParamsField(hiddenParamNames, htmlDoc));
}
// redirect form submission action to wrapping servlet
form.setAttribute("action", request.getRequestURL().toString());
String origFormMethod = form.getAttribute("method");
if(origFormMethod == null || origFormMethod.trim().length() == 0){
origFormMethod = "GET"; // from html4.01 spec
}
// Make all forms POST for purposes of wrapping
form.setAttribute("method", "POST");
String origFormEncodingType = form.getAttribute("enctype");
if(origFormEncodingType == null || origFormEncodingType.trim().length() == 0){
origFormEncodingType = XHTMLForm.URLENCODED; // from html4.01 spec
}
form.setAttribute("enctype", XHTMLForm.MULTIPART); //this'll handle file uploads too, safest bet
// Submit the original CGI settings as a hidden parameter so we can reconstruct the correct
// request to forward later.
Element hiddenSrcInput = htmlDoc.createElement("input");
hiddenSrcInput.setAttribute("type", "hidden");
hiddenSrcInput.setAttribute("name", SRC_PARAM);
hiddenSrcInput.setAttribute("value", url.toString());
form.appendChild(hiddenSrcInput);
// actionSpec encodes the invocation system for the CGI.
// The format is defined by ca.ucalgary.services.DaggooRegistration
String actionSpec = origFormAction.toString()+" "+origFormMethod+" "+origFormEncodingType;
Element hiddenSpecInput = htmlDoc.createElement("input");
hiddenSpecInput.setAttribute("type", "hidden");
hiddenSpecInput.setAttribute("name", SERVICE_SPEC_PARAM);
hiddenSpecInput.setAttribute("value", actionSpec);
form.appendChild(hiddenSpecInput);
// Add the recorder's onSubmit event
if(recorder != null && recorder.getOnSubmitAsDOM(htmlDoc) != null){
Attr submitEventAttr = recorder.getOnSubmitAsDOM(htmlDoc);
//See if there is already a Javascript event. If so append, if not, add one.
String existingEvent = form.getAttribute(submitEventAttr.getName());
if(existingEvent == null || existingEvent.trim().length() ==0){
// just set it, not cloberring anything
form.setAttributeNode(submitEventAttr);
}
else{
// separate JS statements with a semi-colon.
// Put our event first in case the existing one returns false and does AJAX-y stuff
form.setAttribute(submitEventAttr.getName(), submitEventAttr.getValue()+";"+existingEvent);
}
}
}
//todo: inputs.add(htmlDoc.getElementsByTagName("button"));
// print the instrumented doc to the client
tidy.pprint(htmlDoc, out);
}
// Info to pass along from CGI form creator to service wrapper about which fields aren't user editable
// The field that lists the hidden fields is itself a hidden field, just to make it interesting.
// Note to Bertrand Russell: this field created here is not included in the list of hidden fields.
private Element createHiddenParamsField(List hiddenFields, Document owner){
Element formInput = owner.createElement("input");
formInput.setAttribute("name", HIDDEN_PARAM);
formInput.setAttribute("type", "hidden");
formInput.setAttribute("value", XHTMLForm.join(" ", hiddenFields.toArray(new String[hiddenFields.size()])));
return formInput;
}
protected void callService(HttpServletRequest request,
HttpServletResponse response,
URL cgiFormURL,
PrintStream out){
if(!ServletFileUpload.isMultipartContent(request)){
out.print("Error\n"+
"\n"+
"The request was expected to be '"+
XHTMLForm.MULTIPART+"' encoded");
return;
}
String serviceSpec = null;
// list of params that are file uploads instead of regular form fields
ArrayList fileTypeParams = new ArrayList();
ArrayList hiddenParamsList = new ArrayList();
Map formDataInstanceMap = new HashMap();
try {
ServletFileUpload upload = new ServletFileUpload();
FileItemIterator iterator = upload.getItemIterator(request);
while (iterator.hasNext()) {
FileItemStream item = iterator.next();
InputStream in = item.openStream();
try{
if(item.getFieldName().equals(SERVICE_SPEC_PARAM)){
//do this because DataRecorder will expect it
serviceSpec = IOUtils.toString(in);
if(recorder != null){
recorder.setParameter(request.getSession(true), SERVICE_SPEC_PARAM, serviceSpec);
}
continue;
}
else if(item.getFieldName().equals(SRC_PARAM)){
//do this because DataRecorder will expect it
if(recorder != null){
recorder.setParameter(request.getSession(true), SRC_PARAM, IOUtils.toString(in));
}
continue;
}
// The wrapping form passes along the list of hidden input fields
// so that we can not show these later in the wrapping process also.
else if(item.getFieldName().equals(HIDDEN_PARAM)){
for(String hiddenParamName: IOUtils.toString(in).split(" ")){
hiddenParamsList.add(hiddenParamName);
}
continue;
}
if(!item.isFormField()){ //is a file upload
fileTypeParams.add(item.getFieldName());
}
formDataInstanceMap.put(item.getFieldName(), IOUtils.toByteArray(in));
} finally {
IOUtils.closeQuietly(in);
}
}
} catch (Exception e) {
out.print("Error\n"+
"\n"+
"Encountered an exception while parsing the submitted data:\n
");
e.printStackTrace(out);
out.print("
");
return;
}
if(serviceSpec == null || serviceSpec.trim().length() == 0){
out.print("Error\n"+
"\n"+
"No '"+SERVICE_SPEC_PARAM+"' parameter (specifying " +
"the action/method/enctype) was specified in the POST request");
return;
}
String[] serviceSpecs = serviceSpec.split(" ");
if(serviceSpecs.length != 3){
out.print("Error\n"+
"\n"+
"The '"+SERVICE_SPEC_PARAM+"' parameter (specifying " +
"the action/method/enctype) did not contain 3 space-separated values as expected, " +
"but rather '"+serviceSpecs+"'");
return;
}
if(recorder != null){ //PBE needs to know what the input looked like
recorder.setInputParams(request, formDataInstanceMap, hiddenParamsList);
}
HttpMethod method;
if("POST".equals(serviceSpecs[1].toUpperCase())){
method = new PostMethod(serviceSpecs[0]);
if(XHTMLForm.MULTIPART.toLowerCase().equals(serviceSpecs[2].toLowerCase())){
((PostMethod) method).setRequestEntity(CGIUtils.getMultipartRequest(formDataInstanceMap,
method.getParams(),
fileTypeParams));
}
else{
((PostMethod) method).setRequestBody(CGIUtils.getNameValuePairs(formDataInstanceMap));
}
try{logger.log(Level.INFO, serviceSpecs[1]+" request is " + method.getURI());}
catch(URIException urie){logger.log(Level.WARNING,
"Could not get URI of POST request (should be " +
serviceSpecs[0]+")",
urie);}
}
// If not POST, assume GET
else{
try{
method = new GetMethod(serviceSpecs[0]+"?"+CGIUtils.getURLQuery(formDataInstanceMap));
logger.log(Level.INFO, serviceSpecs[1]+" request is " + method.getURI());
} catch(URIException urie){
out.print("Error\n"+
"\n"+
"Encountered a problem encoding the CGI input data for GET:
");
urie.printStackTrace(out);
out.print("
");
return;
}
}
int statusCode;
byte[] responseBody;
// Send the request via HTTP
try {
HttpClient httpClient = new HttpClient();
// Execute the method
statusCode = httpClient.executeMethod(method);
// Read the response body
responseBody = method.getResponseBody();
} catch (HttpException he) {
logger.log(Level.SEVERE, "Fatal protocol violation: ", he.getMessage());
out.print("Error\n"+
"\n"+
"The CGI call encountered a fatal HTTP protocol error:
");
he.printStackTrace(out);
out.print("
");
return;
} catch (java.io.IOException ioe) {
logger.log(Level.SEVERE, "Fatal transport error: " + ioe.getMessage());
out.print("Error\n"+
"\n"+
"The CGI call encountered a fatal network I/O error:
");
ioe.printStackTrace(out);
out.print("
");
return;
} finally {
// Release the connection.
method.releaseConnection();
}
if (statusCode != HttpStatus.SC_OK) {
logger.log(Level.SEVERE, "Fatal CGI error: " + method.getStatusLine());
out.print("Error\n"+
"\n"+
"The CGI call failed:
"+
method.getStatusLine()+
"
");
return;
}
String contentType = method.getResponseHeader("Content-type").getValue();
response.setContentType(contentType);
// If the page is html, set the base href so the page is rendered properly
logger.log(Level.INFO, "The content type of the response is "+contentType);
String encoding = null;
if(contentType != null && contentType.indexOf("html") != -1){
logger.log(Level.INFO, "Setting the base href for the response for proper rendering");
encoding = response.getCharacterEncoding();
if(encoding == null){
encoding = java.nio.charset.Charset.defaultCharset().name();
}
try{
responseBody = (new String(responseBody, encoding))
.replaceFirst("<[Hh][Ee][Aa][Dd]>", "").getBytes();
} catch(java.io.UnsupportedEncodingException uee){
logger.log(Level.SEVERE, "Could not use character encoding ("+encoding+
") of response, therefore cannot set base href in HTML response...sending response as-is", uee);
}
}
if(recorder != null){
byte[] answer = null;
try{
answer = recorder.markupResponse(responseBody, contentType, encoding, request);
} catch(Exception e){
out.print("Error\n"+
"\n"+
"Exception in DataRecorder subsystem:\n